1 @c Copyright (C) 1988,1989,1992,1993,1994,1995,1996,1997,1998,1999,2000,2001,
2 @c 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009
3 @c Free Software Foundation, Inc.
4 @c This is part of the GCC manual.
5 @c For copying conditions, see the file gcc.texi.
8 @chapter Target Description Macros and Functions
9 @cindex machine description macros
10 @cindex target description macros
11 @cindex macros, target description
12 @cindex @file{tm.h} macros
14 In addition to the file @file{@var{machine}.md}, a machine description
15 includes a C header file conventionally given the name
16 @file{@var{machine}.h} and a C source file named @file{@var{machine}.c}.
17 The header file defines numerous macros that convey the information
18 about the target machine that does not fit into the scheme of the
19 @file{.md} file. The file @file{tm.h} should be a link to
20 @file{@var{machine}.h}. The header file @file{config.h} includes
21 @file{tm.h} and most compiler source files include @file{config.h}. The
22 source file defines a variable @code{targetm}, which is a structure
23 containing pointers to functions and data relating to the target
24 machine. @file{@var{machine}.c} should also contain their definitions,
25 if they are not defined elsewhere in GCC, and other functions called
26 through the macros defined in the @file{.h} file.
29 * Target Structure:: The @code{targetm} variable.
30 * Driver:: Controlling how the driver runs the compilation passes.
31 * Run-time Target:: Defining @samp{-m} options like @option{-m68000} and @option{-m68020}.
32 * Per-Function Data:: Defining data structures for per-function information.
33 * Storage Layout:: Defining sizes and alignments of data.
34 * Type Layout:: Defining sizes and properties of basic user data types.
35 * Registers:: Naming and describing the hardware registers.
36 * Register Classes:: Defining the classes of hardware registers.
37 * Old Constraints:: The old way to define machine-specific constraints.
38 * Stack and Calling:: Defining which way the stack grows and by how much.
39 * Varargs:: Defining the varargs macros.
40 * Trampolines:: Code set up at run time to enter a nested function.
41 * Library Calls:: Controlling how library routines are implicitly called.
42 * Addressing Modes:: Defining addressing modes valid for memory operands.
43 * Anchored Addresses:: Defining how @option{-fsection-anchors} should work.
44 * Condition Code:: Defining how insns update the condition code.
45 * Costs:: Defining relative costs of different operations.
46 * Scheduling:: Adjusting the behavior of the instruction scheduler.
47 * Sections:: Dividing storage into text, data, and other sections.
48 * PIC:: Macros for position independent code.
49 * Assembler Format:: Defining how to write insns and pseudo-ops to output.
50 * Debugging Info:: Defining the format of debugging output.
51 * Floating Point:: Handling floating point for cross-compilers.
52 * Mode Switching:: Insertion of mode-switching instructions.
53 * Target Attributes:: Defining target-specific uses of @code{__attribute__}.
54 * Emulated TLS:: Emulated TLS support.
55 * MIPS Coprocessors:: MIPS coprocessor support and how to customize it.
56 * PCH Target:: Validity checking for precompiled headers.
57 * C++ ABI:: Controlling C++ ABI changes.
58 * Named Address Spaces:: Adding support for named address spaces
59 * Misc:: Everything else.
62 @node Target Structure
63 @section The Global @code{targetm} Variable
65 @cindex target functions
67 @deftypevar {struct gcc_target} targetm
68 The target @file{.c} file must define the global @code{targetm} variable
69 which contains pointers to functions and data relating to the target
70 machine. The variable is declared in @file{target.h};
71 @file{target-def.h} defines the macro @code{TARGET_INITIALIZER} which is
72 used to initialize the variable, and macros for the default initializers
73 for elements of the structure. The @file{.c} file should override those
74 macros for which the default definition is inappropriate. For example:
77 #include "target-def.h"
79 /* @r{Initialize the GCC target structure.} */
81 #undef TARGET_COMP_TYPE_ATTRIBUTES
82 #define TARGET_COMP_TYPE_ATTRIBUTES @var{machine}_comp_type_attributes
84 struct gcc_target targetm = TARGET_INITIALIZER;
88 Where a macro should be defined in the @file{.c} file in this manner to
89 form part of the @code{targetm} structure, it is documented below as a
90 ``Target Hook'' with a prototype. Many macros will change in future
91 from being defined in the @file{.h} file to being part of the
92 @code{targetm} structure.
95 @section Controlling the Compilation Driver, @file{gcc}
97 @cindex controlling the compilation driver
99 @c prevent bad page break with this line
100 You can control the compilation driver.
102 @defmac SWITCH_TAKES_ARG (@var{char})
103 A C expression which determines whether the option @option{-@var{char}}
104 takes arguments. The value should be the number of arguments that
105 option takes--zero, for many options.
107 By default, this macro is defined as
108 @code{DEFAULT_SWITCH_TAKES_ARG}, which handles the standard options
109 properly. You need not define @code{SWITCH_TAKES_ARG} unless you
110 wish to add additional options which take arguments. Any redefinition
111 should call @code{DEFAULT_SWITCH_TAKES_ARG} and then check for
115 @defmac WORD_SWITCH_TAKES_ARG (@var{name})
116 A C expression which determines whether the option @option{-@var{name}}
117 takes arguments. The value should be the number of arguments that
118 option takes--zero, for many options. This macro rather than
119 @code{SWITCH_TAKES_ARG} is used for multi-character option names.
121 By default, this macro is defined as
122 @code{DEFAULT_WORD_SWITCH_TAKES_ARG}, which handles the standard options
123 properly. You need not define @code{WORD_SWITCH_TAKES_ARG} unless you
124 wish to add additional options which take arguments. Any redefinition
125 should call @code{DEFAULT_WORD_SWITCH_TAKES_ARG} and then check for
129 @defmac SWITCH_CURTAILS_COMPILATION (@var{char})
130 A C expression which determines whether the option @option{-@var{char}}
131 stops compilation before the generation of an executable. The value is
132 boolean, nonzero if the option does stop an executable from being
133 generated, zero otherwise.
135 By default, this macro is defined as
136 @code{DEFAULT_SWITCH_CURTAILS_COMPILATION}, which handles the standard
137 options properly. You need not define
138 @code{SWITCH_CURTAILS_COMPILATION} unless you wish to add additional
139 options which affect the generation of an executable. Any redefinition
140 should call @code{DEFAULT_SWITCH_CURTAILS_COMPILATION} and then check
141 for additional options.
144 @defmac SWITCHES_NEED_SPACES
145 A string-valued C expression which enumerates the options for which
146 the linker needs a space between the option and its argument.
148 If this macro is not defined, the default value is @code{""}.
151 @defmac TARGET_OPTION_TRANSLATE_TABLE
152 If defined, a list of pairs of strings, the first of which is a
153 potential command line target to the @file{gcc} driver program, and the
154 second of which is a space-separated (tabs and other whitespace are not
155 supported) list of options with which to replace the first option. The
156 target defining this list is responsible for assuring that the results
157 are valid. Replacement options may not be the @code{--opt} style, they
158 must be the @code{-opt} style. It is the intention of this macro to
159 provide a mechanism for substitution that affects the multilibs chosen,
160 such as one option that enables many options, some of which select
161 multilibs. Example nonsensical definition, where @option{-malt-abi},
162 @option{-EB}, and @option{-mspoo} cause different multilibs to be chosen:
165 #define TARGET_OPTION_TRANSLATE_TABLE \
166 @{ "-fast", "-march=fast-foo -malt-abi -I/usr/fast-foo" @}, \
167 @{ "-compat", "-EB -malign=4 -mspoo" @}
171 @defmac DRIVER_SELF_SPECS
172 A list of specs for the driver itself. It should be a suitable
173 initializer for an array of strings, with no surrounding braces.
175 The driver applies these specs to its own command line between loading
176 default @file{specs} files (but not command-line specified ones) and
177 choosing the multilib directory or running any subcommands. It
178 applies them in the order given, so each spec can depend on the
179 options added by earlier ones. It is also possible to remove options
180 using @samp{%<@var{option}} in the usual way.
182 This macro can be useful when a port has several interdependent target
183 options. It provides a way of standardizing the command line so
184 that the other specs are easier to write.
186 Do not define this macro if it does not need to do anything.
189 @defmac OPTION_DEFAULT_SPECS
190 A list of specs used to support configure-time default options (i.e.@:
191 @option{--with} options) in the driver. It should be a suitable initializer
192 for an array of structures, each containing two strings, without the
193 outermost pair of surrounding braces.
195 The first item in the pair is the name of the default. This must match
196 the code in @file{config.gcc} for the target. The second item is a spec
197 to apply if a default with this name was specified. The string
198 @samp{%(VALUE)} in the spec will be replaced by the value of the default
199 everywhere it occurs.
201 The driver will apply these specs to its own command line between loading
202 default @file{specs} files and processing @code{DRIVER_SELF_SPECS}, using
203 the same mechanism as @code{DRIVER_SELF_SPECS}.
205 Do not define this macro if it does not need to do anything.
209 A C string constant that tells the GCC driver program options to
210 pass to CPP@. It can also specify how to translate options you
211 give to GCC into options for GCC to pass to the CPP@.
213 Do not define this macro if it does not need to do anything.
216 @defmac CPLUSPLUS_CPP_SPEC
217 This macro is just like @code{CPP_SPEC}, but is used for C++, rather
218 than C@. If you do not define this macro, then the value of
219 @code{CPP_SPEC} (if any) will be used instead.
223 A C string constant that tells the GCC driver program options to
224 pass to @code{cc1}, @code{cc1plus}, @code{f771}, and the other language
226 It can also specify how to translate options you give to GCC into options
227 for GCC to pass to front ends.
229 Do not define this macro if it does not need to do anything.
233 A C string constant that tells the GCC driver program options to
234 pass to @code{cc1plus}. It can also specify how to translate options you
235 give to GCC into options for GCC to pass to the @code{cc1plus}.
237 Do not define this macro if it does not need to do anything.
238 Note that everything defined in CC1_SPEC is already passed to
239 @code{cc1plus} so there is no need to duplicate the contents of
240 CC1_SPEC in CC1PLUS_SPEC@.
244 A C string constant that tells the GCC driver program options to
245 pass to the assembler. It can also specify how to translate options
246 you give to GCC into options for GCC to pass to the assembler.
247 See the file @file{sun3.h} for an example of this.
249 Do not define this macro if it does not need to do anything.
252 @defmac ASM_FINAL_SPEC
253 A C string constant that tells the GCC driver program how to
254 run any programs which cleanup after the normal assembler.
255 Normally, this is not needed. See the file @file{mips.h} for
258 Do not define this macro if it does not need to do anything.
261 @defmac AS_NEEDS_DASH_FOR_PIPED_INPUT
262 Define this macro, with no value, if the driver should give the assembler
263 an argument consisting of a single dash, @option{-}, to instruct it to
264 read from its standard input (which will be a pipe connected to the
265 output of the compiler proper). This argument is given after any
266 @option{-o} option specifying the name of the output file.
268 If you do not define this macro, the assembler is assumed to read its
269 standard input if given no non-option arguments. If your assembler
270 cannot read standard input at all, use a @samp{%@{pipe:%e@}} construct;
271 see @file{mips.h} for instance.
275 A C string constant that tells the GCC driver program options to
276 pass to the linker. It can also specify how to translate options you
277 give to GCC into options for GCC to pass to the linker.
279 Do not define this macro if it does not need to do anything.
283 Another C string constant used much like @code{LINK_SPEC}. The difference
284 between the two is that @code{LIB_SPEC} is used at the end of the
285 command given to the linker.
287 If this macro is not defined, a default is provided that
288 loads the standard C library from the usual place. See @file{gcc.c}.
292 Another C string constant that tells the GCC driver program
293 how and when to place a reference to @file{libgcc.a} into the
294 linker command line. This constant is placed both before and after
295 the value of @code{LIB_SPEC}.
297 If this macro is not defined, the GCC driver provides a default that
298 passes the string @option{-lgcc} to the linker.
301 @defmac REAL_LIBGCC_SPEC
302 By default, if @code{ENABLE_SHARED_LIBGCC} is defined, the
303 @code{LIBGCC_SPEC} is not directly used by the driver program but is
304 instead modified to refer to different versions of @file{libgcc.a}
305 depending on the values of the command line flags @option{-static},
306 @option{-shared}, @option{-static-libgcc}, and @option{-shared-libgcc}. On
307 targets where these modifications are inappropriate, define
308 @code{REAL_LIBGCC_SPEC} instead. @code{REAL_LIBGCC_SPEC} tells the
309 driver how to place a reference to @file{libgcc} on the link command
310 line, but, unlike @code{LIBGCC_SPEC}, it is used unmodified.
313 @defmac USE_LD_AS_NEEDED
314 A macro that controls the modifications to @code{LIBGCC_SPEC}
315 mentioned in @code{REAL_LIBGCC_SPEC}. If nonzero, a spec will be
316 generated that uses --as-needed and the shared libgcc in place of the
317 static exception handler library, when linking without any of
318 @code{-static}, @code{-static-libgcc}, or @code{-shared-libgcc}.
322 If defined, this C string constant is added to @code{LINK_SPEC}.
323 When @code{USE_LD_AS_NEEDED} is zero or undefined, it also affects
324 the modifications to @code{LIBGCC_SPEC} mentioned in
325 @code{REAL_LIBGCC_SPEC}.
328 @defmac STARTFILE_SPEC
329 Another C string constant used much like @code{LINK_SPEC}. The
330 difference between the two is that @code{STARTFILE_SPEC} is used at
331 the very beginning of the command given to the linker.
333 If this macro is not defined, a default is provided that loads the
334 standard C startup file from the usual place. See @file{gcc.c}.
338 Another C string constant used much like @code{LINK_SPEC}. The
339 difference between the two is that @code{ENDFILE_SPEC} is used at
340 the very end of the command given to the linker.
342 Do not define this macro if it does not need to do anything.
345 @defmac THREAD_MODEL_SPEC
346 GCC @code{-v} will print the thread model GCC was configured to use.
347 However, this doesn't work on platforms that are multilibbed on thread
348 models, such as AIX 4.3. On such platforms, define
349 @code{THREAD_MODEL_SPEC} such that it evaluates to a string without
350 blanks that names one of the recognized thread models. @code{%*}, the
351 default value of this macro, will expand to the value of
352 @code{thread_file} set in @file{config.gcc}.
355 @defmac SYSROOT_SUFFIX_SPEC
356 Define this macro to add a suffix to the target sysroot when GCC is
357 configured with a sysroot. This will cause GCC to search for usr/lib,
358 et al, within sysroot+suffix.
361 @defmac SYSROOT_HEADERS_SUFFIX_SPEC
362 Define this macro to add a headers_suffix to the target sysroot when
363 GCC is configured with a sysroot. This will cause GCC to pass the
364 updated sysroot+headers_suffix to CPP, causing it to search for
365 usr/include, et al, within sysroot+headers_suffix.
369 Define this macro to provide additional specifications to put in the
370 @file{specs} file that can be used in various specifications like
373 The definition should be an initializer for an array of structures,
374 containing a string constant, that defines the specification name, and a
375 string constant that provides the specification.
377 Do not define this macro if it does not need to do anything.
379 @code{EXTRA_SPECS} is useful when an architecture contains several
380 related targets, which have various @code{@dots{}_SPECS} which are similar
381 to each other, and the maintainer would like one central place to keep
384 For example, the PowerPC System V.4 targets use @code{EXTRA_SPECS} to
385 define either @code{_CALL_SYSV} when the System V calling sequence is
386 used or @code{_CALL_AIX} when the older AIX-based calling sequence is
389 The @file{config/rs6000/rs6000.h} target file defines:
392 #define EXTRA_SPECS \
393 @{ "cpp_sysv_default", CPP_SYSV_DEFAULT @},
395 #define CPP_SYS_DEFAULT ""
398 The @file{config/rs6000/sysv.h} target file defines:
402 "%@{posix: -D_POSIX_SOURCE @} \
403 %@{mcall-sysv: -D_CALL_SYSV @} \
404 %@{!mcall-sysv: %(cpp_sysv_default) @} \
405 %@{msoft-float: -D_SOFT_FLOAT@} %@{mcpu=403: -D_SOFT_FLOAT@}"
407 #undef CPP_SYSV_DEFAULT
408 #define CPP_SYSV_DEFAULT "-D_CALL_SYSV"
411 while the @file{config/rs6000/eabiaix.h} target file defines
412 @code{CPP_SYSV_DEFAULT} as:
415 #undef CPP_SYSV_DEFAULT
416 #define CPP_SYSV_DEFAULT "-D_CALL_AIX"
420 @defmac LINK_LIBGCC_SPECIAL_1
421 Define this macro if the driver program should find the library
422 @file{libgcc.a}. If you do not define this macro, the driver program will pass
423 the argument @option{-lgcc} to tell the linker to do the search.
426 @defmac LINK_GCC_C_SEQUENCE_SPEC
427 The sequence in which libgcc and libc are specified to the linker.
428 By default this is @code{%G %L %G}.
431 @defmac LINK_COMMAND_SPEC
432 A C string constant giving the complete command line need to execute the
433 linker. When you do this, you will need to update your port each time a
434 change is made to the link command line within @file{gcc.c}. Therefore,
435 define this macro only if you need to completely redefine the command
436 line for invoking the linker and there is no other way to accomplish
437 the effect you need. Overriding this macro may be avoidable by overriding
438 @code{LINK_GCC_C_SEQUENCE_SPEC} instead.
441 @defmac LINK_ELIMINATE_DUPLICATE_LDIRECTORIES
442 A nonzero value causes @command{collect2} to remove duplicate @option{-L@var{directory}} search
443 directories from linking commands. Do not give it a nonzero value if
444 removing duplicate search directories changes the linker's semantics.
447 @defmac MULTILIB_DEFAULTS
448 Define this macro as a C expression for the initializer of an array of
449 string to tell the driver program which options are defaults for this
450 target and thus do not need to be handled specially when using
451 @code{MULTILIB_OPTIONS}.
453 Do not define this macro if @code{MULTILIB_OPTIONS} is not defined in
454 the target makefile fragment or if none of the options listed in
455 @code{MULTILIB_OPTIONS} are set by default.
456 @xref{Target Fragment}.
459 @defmac RELATIVE_PREFIX_NOT_LINKDIR
460 Define this macro to tell @command{gcc} that it should only translate
461 a @option{-B} prefix into a @option{-L} linker option if the prefix
462 indicates an absolute file name.
465 @defmac MD_EXEC_PREFIX
466 If defined, this macro is an additional prefix to try after
467 @code{STANDARD_EXEC_PREFIX}. @code{MD_EXEC_PREFIX} is not searched
468 when the @option{-b} option is used, or the compiler is built as a cross
469 compiler. If you define @code{MD_EXEC_PREFIX}, then be sure to add it
470 to the list of directories used to find the assembler in @file{configure.in}.
473 @defmac STANDARD_STARTFILE_PREFIX
474 Define this macro as a C string constant if you wish to override the
475 standard choice of @code{libdir} as the default prefix to
476 try when searching for startup files such as @file{crt0.o}.
477 @code{STANDARD_STARTFILE_PREFIX} is not searched when the compiler
478 is built as a cross compiler.
481 @defmac STANDARD_STARTFILE_PREFIX_1
482 Define this macro as a C string constant if you wish to override the
483 standard choice of @code{/lib} as a prefix to try after the default prefix
484 when searching for startup files such as @file{crt0.o}.
485 @code{STANDARD_STARTFILE_PREFIX_1} is not searched when the compiler
486 is built as a cross compiler.
489 @defmac STANDARD_STARTFILE_PREFIX_2
490 Define this macro as a C string constant if you wish to override the
491 standard choice of @code{/lib} as yet another prefix to try after the
492 default prefix when searching for startup files such as @file{crt0.o}.
493 @code{STANDARD_STARTFILE_PREFIX_2} is not searched when the compiler
494 is built as a cross compiler.
497 @defmac MD_STARTFILE_PREFIX
498 If defined, this macro supplies an additional prefix to try after the
499 standard prefixes. @code{MD_EXEC_PREFIX} is not searched when the
500 @option{-b} option is used, or when the compiler is built as a cross
504 @defmac MD_STARTFILE_PREFIX_1
505 If defined, this macro supplies yet another prefix to try after the
506 standard prefixes. It is not searched when the @option{-b} option is
507 used, or when the compiler is built as a cross compiler.
510 @defmac INIT_ENVIRONMENT
511 Define this macro as a C string constant if you wish to set environment
512 variables for programs called by the driver, such as the assembler and
513 loader. The driver passes the value of this macro to @code{putenv} to
514 initialize the necessary environment variables.
517 @defmac LOCAL_INCLUDE_DIR
518 Define this macro as a C string constant if you wish to override the
519 standard choice of @file{/usr/local/include} as the default prefix to
520 try when searching for local header files. @code{LOCAL_INCLUDE_DIR}
521 comes before @code{SYSTEM_INCLUDE_DIR} in the search order.
523 Cross compilers do not search either @file{/usr/local/include} or its
527 @defmac MODIFY_TARGET_NAME
528 Define this macro if you wish to define command-line switches that
529 modify the default target name.
531 For each switch, you can include a string to be appended to the first
532 part of the configuration name or a string to be deleted from the
533 configuration name, if present. The definition should be an initializer
534 for an array of structures. Each array element should have three
535 elements: the switch name (a string constant, including the initial
536 dash), one of the enumeration codes @code{ADD} or @code{DELETE} to
537 indicate whether the string should be inserted or deleted, and the string
538 to be inserted or deleted (a string constant).
540 For example, on a machine where @samp{64} at the end of the
541 configuration name denotes a 64-bit target and you want the @option{-32}
542 and @option{-64} switches to select between 32- and 64-bit targets, you would
546 #define MODIFY_TARGET_NAME \
547 @{ @{ "-32", DELETE, "64"@}, \
548 @{"-64", ADD, "64"@}@}
552 @defmac SYSTEM_INCLUDE_DIR
553 Define this macro as a C string constant if you wish to specify a
554 system-specific directory to search for header files before the standard
555 directory. @code{SYSTEM_INCLUDE_DIR} comes before
556 @code{STANDARD_INCLUDE_DIR} in the search order.
558 Cross compilers do not use this macro and do not search the directory
562 @defmac STANDARD_INCLUDE_DIR
563 Define this macro as a C string constant if you wish to override the
564 standard choice of @file{/usr/include} as the default prefix to
565 try when searching for header files.
567 Cross compilers ignore this macro and do not search either
568 @file{/usr/include} or its replacement.
571 @defmac STANDARD_INCLUDE_COMPONENT
572 The ``component'' corresponding to @code{STANDARD_INCLUDE_DIR}.
573 See @code{INCLUDE_DEFAULTS}, below, for the description of components.
574 If you do not define this macro, no component is used.
577 @defmac INCLUDE_DEFAULTS
578 Define this macro if you wish to override the entire default search path
579 for include files. For a native compiler, the default search path
580 usually consists of @code{GCC_INCLUDE_DIR}, @code{LOCAL_INCLUDE_DIR},
581 @code{SYSTEM_INCLUDE_DIR}, @code{GPLUSPLUS_INCLUDE_DIR}, and
582 @code{STANDARD_INCLUDE_DIR}. In addition, @code{GPLUSPLUS_INCLUDE_DIR}
583 and @code{GCC_INCLUDE_DIR} are defined automatically by @file{Makefile},
584 and specify private search areas for GCC@. The directory
585 @code{GPLUSPLUS_INCLUDE_DIR} is used only for C++ programs.
587 The definition should be an initializer for an array of structures.
588 Each array element should have four elements: the directory name (a
589 string constant), the component name (also a string constant), a flag
590 for C++-only directories,
591 and a flag showing that the includes in the directory don't need to be
592 wrapped in @code{extern @samp{C}} when compiling C++. Mark the end of
593 the array with a null element.
595 The component name denotes what GNU package the include file is part of,
596 if any, in all uppercase letters. For example, it might be @samp{GCC}
597 or @samp{BINUTILS}. If the package is part of a vendor-supplied
598 operating system, code the component name as @samp{0}.
600 For example, here is the definition used for VAX/VMS:
603 #define INCLUDE_DEFAULTS \
605 @{ "GNU_GXX_INCLUDE:", "G++", 1, 1@}, \
606 @{ "GNU_CC_INCLUDE:", "GCC", 0, 0@}, \
607 @{ "SYS$SYSROOT:[SYSLIB.]", 0, 0, 0@}, \
614 Here is the order of prefixes tried for exec files:
618 Any prefixes specified by the user with @option{-B}.
621 The environment variable @code{GCC_EXEC_PREFIX} or, if @code{GCC_EXEC_PREFIX}
622 is not set and the compiler has not been installed in the configure-time
623 @var{prefix}, the location in which the compiler has actually been installed.
626 The directories specified by the environment variable @code{COMPILER_PATH}.
629 The macro @code{STANDARD_EXEC_PREFIX}, if the compiler has been installed
630 in the configured-time @var{prefix}.
633 The location @file{/usr/libexec/gcc/}, but only if this is a native compiler.
636 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
639 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
643 Here is the order of prefixes tried for startfiles:
647 Any prefixes specified by the user with @option{-B}.
650 The environment variable @code{GCC_EXEC_PREFIX} or its automatically determined
651 value based on the installed toolchain location.
654 The directories specified by the environment variable @code{LIBRARY_PATH}
655 (or port-specific name; native only, cross compilers do not use this).
658 The macro @code{STANDARD_EXEC_PREFIX}, but only if the toolchain is installed
659 in the configured @var{prefix} or this is a native compiler.
662 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
665 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
669 The macro @code{MD_STARTFILE_PREFIX}, if defined, but only if this is a
670 native compiler, or we have a target system root.
673 The macro @code{MD_STARTFILE_PREFIX_1}, if defined, but only if this is a
674 native compiler, or we have a target system root.
677 The macro @code{STANDARD_STARTFILE_PREFIX}, with any sysroot modifications.
678 If this path is relative it will be prefixed by @code{GCC_EXEC_PREFIX} and
679 the machine suffix or @code{STANDARD_EXEC_PREFIX} and the machine suffix.
682 The macro @code{STANDARD_STARTFILE_PREFIX_1}, but only if this is a native
683 compiler, or we have a target system root. The default for this macro is
687 The macro @code{STANDARD_STARTFILE_PREFIX_2}, but only if this is a native
688 compiler, or we have a target system root. The default for this macro is
692 @node Run-time Target
693 @section Run-time Target Specification
694 @cindex run-time target specification
695 @cindex predefined macros
696 @cindex target specifications
698 @c prevent bad page break with this line
699 Here are run-time target specifications.
701 @defmac TARGET_CPU_CPP_BUILTINS ()
702 This function-like macro expands to a block of code that defines
703 built-in preprocessor macros and assertions for the target CPU, using
704 the functions @code{builtin_define}, @code{builtin_define_std} and
705 @code{builtin_assert}. When the front end
706 calls this macro it provides a trailing semicolon, and since it has
707 finished command line option processing your code can use those
710 @code{builtin_assert} takes a string in the form you pass to the
711 command-line option @option{-A}, such as @code{cpu=mips}, and creates
712 the assertion. @code{builtin_define} takes a string in the form
713 accepted by option @option{-D} and unconditionally defines the macro.
715 @code{builtin_define_std} takes a string representing the name of an
716 object-like macro. If it doesn't lie in the user's namespace,
717 @code{builtin_define_std} defines it unconditionally. Otherwise, it
718 defines a version with two leading underscores, and another version
719 with two leading and trailing underscores, and defines the original
720 only if an ISO standard was not requested on the command line. For
721 example, passing @code{unix} defines @code{__unix}, @code{__unix__}
722 and possibly @code{unix}; passing @code{_mips} defines @code{__mips},
723 @code{__mips__} and possibly @code{_mips}, and passing @code{_ABI64}
724 defines only @code{_ABI64}.
726 You can also test for the C dialect being compiled. The variable
727 @code{c_language} is set to one of @code{clk_c}, @code{clk_cplusplus}
728 or @code{clk_objective_c}. Note that if we are preprocessing
729 assembler, this variable will be @code{clk_c} but the function-like
730 macro @code{preprocessing_asm_p()} will return true, so you might want
731 to check for that first. If you need to check for strict ANSI, the
732 variable @code{flag_iso} can be used. The function-like macro
733 @code{preprocessing_trad_p()} can be used to check for traditional
737 @defmac TARGET_OS_CPP_BUILTINS ()
738 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
739 and is used for the target operating system instead.
742 @defmac TARGET_OBJFMT_CPP_BUILTINS ()
743 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
744 and is used for the target object format. @file{elfos.h} uses this
745 macro to define @code{__ELF__}, so you probably do not need to define
749 @deftypevar {extern int} target_flags
750 This variable is declared in @file{options.h}, which is included before
751 any target-specific headers.
754 @deftypevr {Target Hook} int TARGET_DEFAULT_TARGET_FLAGS
755 This variable specifies the initial value of @code{target_flags}.
756 Its default setting is 0.
759 @cindex optional hardware or system features
760 @cindex features, optional, in system conventions
762 @deftypefn {Target Hook} bool TARGET_HANDLE_OPTION (size_t @var{code}, const char *@var{arg}, int @var{value})
763 This hook is called whenever the user specifies one of the
764 target-specific options described by the @file{.opt} definition files
765 (@pxref{Options}). It has the opportunity to do some option-specific
766 processing and should return true if the option is valid. The default
767 definition does nothing but return true.
769 @var{code} specifies the @code{OPT_@var{name}} enumeration value
770 associated with the selected option; @var{name} is just a rendering of
771 the option name in which non-alphanumeric characters are replaced by
772 underscores. @var{arg} specifies the string argument and is null if
773 no argument was given. If the option is flagged as a @code{UInteger}
774 (@pxref{Option properties}), @var{value} is the numeric value of the
775 argument. Otherwise @var{value} is 1 if the positive form of the
776 option was used and 0 if the ``no-'' form was.
779 @deftypefn {Target Hook} bool TARGET_HANDLE_C_OPTION (size_t @var{code}, const char *@var{arg}, int @var{value})
780 This target hook is called whenever the user specifies one of the
781 target-specific C language family options described by the @file{.opt}
782 definition files(@pxref{Options}). It has the opportunity to do some
783 option-specific processing and should return true if the option is
784 valid. The default definition does nothing but return false.
786 In general, you should use @code{TARGET_HANDLE_OPTION} to handle
787 options. However, if processing an option requires routines that are
788 only available in the C (and related language) front ends, then you
789 should use @code{TARGET_HANDLE_C_OPTION} instead.
792 @defmac TARGET_VERSION
793 This macro is a C statement to print on @code{stderr} a string
794 describing the particular machine description choice. Every machine
795 description should define @code{TARGET_VERSION}. For example:
799 #define TARGET_VERSION \
800 fprintf (stderr, " (68k, Motorola syntax)");
802 #define TARGET_VERSION \
803 fprintf (stderr, " (68k, MIT syntax)");
808 @defmac OVERRIDE_OPTIONS
809 Sometimes certain combinations of command options do not make sense on
810 a particular target machine. You can define a macro
811 @code{OVERRIDE_OPTIONS} to take account of this. This macro, if
812 defined, is executed once just after all the command options have been
815 Don't use this macro to turn on various extra optimizations for
816 @option{-O}. That is what @code{OPTIMIZATION_OPTIONS} is for.
819 @defmac C_COMMON_OVERRIDE_OPTIONS
820 This is similar to @code{OVERRIDE_OPTIONS} but is only used in the C
821 language frontends (C, Objective-C, C++, Objective-C++) and so can be
822 used to alter option flag variables which only exist in those
826 @defmac OPTIMIZATION_OPTIONS (@var{level}, @var{size})
827 Some machines may desire to change what optimizations are performed for
828 various optimization levels. This macro, if defined, is executed once
829 just after the optimization level is determined and before the remainder
830 of the command options have been parsed. Values set in this macro are
831 used as the default values for the other command line options.
833 @var{level} is the optimization level specified; 2 if @option{-O2} is
834 specified, 1 if @option{-O} is specified, and 0 if neither is specified.
836 @var{size} is nonzero if @option{-Os} is specified and zero otherwise.
838 This macro is run once at program startup and when the optimization
839 options are changed via @code{#pragma GCC optimize} or by using the
840 @code{optimize} attribute.
842 @strong{Do not examine @code{write_symbols} in
843 this macro!} The debugging options are not supposed to alter the
847 @deftypefn {Target Hook} bool TARGET_HELP (void)
848 This hook is called in response to the user invoking
849 @option{--target-help} on the command line. It gives the target a
850 chance to display extra information on the target specific command
851 line options found in its @file{.opt} file.
854 @defmac CAN_DEBUG_WITHOUT_FP
855 Define this macro if debugging can be performed even without a frame
856 pointer. If this macro is defined, GCC will turn on the
857 @option{-fomit-frame-pointer} option whenever @option{-O} is specified.
860 @node Per-Function Data
861 @section Defining data structures for per-function information.
862 @cindex per-function data
863 @cindex data structures
865 If the target needs to store information on a per-function basis, GCC
866 provides a macro and a couple of variables to allow this. Note, just
867 using statics to store the information is a bad idea, since GCC supports
868 nested functions, so you can be halfway through encoding one function
869 when another one comes along.
871 GCC defines a data structure called @code{struct function} which
872 contains all of the data specific to an individual function. This
873 structure contains a field called @code{machine} whose type is
874 @code{struct machine_function *}, which can be used by targets to point
875 to their own specific data.
877 If a target needs per-function specific data it should define the type
878 @code{struct machine_function} and also the macro @code{INIT_EXPANDERS}.
879 This macro should be used to initialize the function pointer
880 @code{init_machine_status}. This pointer is explained below.
882 One typical use of per-function, target specific data is to create an
883 RTX to hold the register containing the function's return address. This
884 RTX can then be used to implement the @code{__builtin_return_address}
885 function, for level 0.
887 Note---earlier implementations of GCC used a single data area to hold
888 all of the per-function information. Thus when processing of a nested
889 function began the old per-function data had to be pushed onto a
890 stack, and when the processing was finished, it had to be popped off the
891 stack. GCC used to provide function pointers called
892 @code{save_machine_status} and @code{restore_machine_status} to handle
893 the saving and restoring of the target specific information. Since the
894 single data area approach is no longer used, these pointers are no
897 @defmac INIT_EXPANDERS
898 Macro called to initialize any target specific information. This macro
899 is called once per function, before generation of any RTL has begun.
900 The intention of this macro is to allow the initialization of the
901 function pointer @code{init_machine_status}.
904 @deftypevar {void (*)(struct function *)} init_machine_status
905 If this function pointer is non-@code{NULL} it will be called once per
906 function, before function compilation starts, in order to allow the
907 target to perform any target specific initialization of the
908 @code{struct function} structure. It is intended that this would be
909 used to initialize the @code{machine} of that structure.
911 @code{struct machine_function} structures are expected to be freed by GC@.
912 Generally, any memory that they reference must be allocated by using
913 @code{ggc_alloc}, including the structure itself.
917 @section Storage Layout
918 @cindex storage layout
920 Note that the definitions of the macros in this table which are sizes or
921 alignments measured in bits do not need to be constant. They can be C
922 expressions that refer to static variables, such as the @code{target_flags}.
923 @xref{Run-time Target}.
925 @defmac BITS_BIG_ENDIAN
926 Define this macro to have the value 1 if the most significant bit in a
927 byte has the lowest number; otherwise define it to have the value zero.
928 This means that bit-field instructions count from the most significant
929 bit. If the machine has no bit-field instructions, then this must still
930 be defined, but it doesn't matter which value it is defined to. This
931 macro need not be a constant.
933 This macro does not affect the way structure fields are packed into
934 bytes or words; that is controlled by @code{BYTES_BIG_ENDIAN}.
937 @defmac BYTES_BIG_ENDIAN
938 Define this macro to have the value 1 if the most significant byte in a
939 word has the lowest number. This macro need not be a constant.
942 @defmac WORDS_BIG_ENDIAN
943 Define this macro to have the value 1 if, in a multiword object, the
944 most significant word has the lowest number. This applies to both
945 memory locations and registers; GCC fundamentally assumes that the
946 order of words in memory is the same as the order in registers. This
947 macro need not be a constant.
950 @defmac LIBGCC2_WORDS_BIG_ENDIAN
951 Define this macro if @code{WORDS_BIG_ENDIAN} is not constant. This must be a
952 constant value with the same meaning as @code{WORDS_BIG_ENDIAN}, which will be
953 used only when compiling @file{libgcc2.c}. Typically the value will be set
954 based on preprocessor defines.
957 @defmac FLOAT_WORDS_BIG_ENDIAN
958 Define this macro to have the value 1 if @code{DFmode}, @code{XFmode} or
959 @code{TFmode} floating point numbers are stored in memory with the word
960 containing the sign bit at the lowest address; otherwise define it to
961 have the value 0. This macro need not be a constant.
963 You need not define this macro if the ordering is the same as for
967 @defmac BITS_PER_UNIT
968 Define this macro to be the number of bits in an addressable storage
969 unit (byte). If you do not define this macro the default is 8.
972 @defmac BITS_PER_WORD
973 Number of bits in a word. If you do not define this macro, the default
974 is @code{BITS_PER_UNIT * UNITS_PER_WORD}.
977 @defmac MAX_BITS_PER_WORD
978 Maximum number of bits in a word. If this is undefined, the default is
979 @code{BITS_PER_WORD}. Otherwise, it is the constant value that is the
980 largest value that @code{BITS_PER_WORD} can have at run-time.
983 @defmac UNITS_PER_WORD
984 Number of storage units in a word; normally the size of a general-purpose
985 register, a power of two from 1 or 8.
988 @defmac MIN_UNITS_PER_WORD
989 Minimum number of units in a word. If this is undefined, the default is
990 @code{UNITS_PER_WORD}. Otherwise, it is the constant value that is the
991 smallest value that @code{UNITS_PER_WORD} can have at run-time.
994 @defmac UNITS_PER_SIMD_WORD (@var{mode})
995 Number of units in the vectors that the vectorizer can produce for
996 scalar mode @var{mode}. The default is equal to @code{UNITS_PER_WORD},
997 because the vectorizer can do some transformations even in absence of
998 specialized @acronym{SIMD} hardware.
1001 @defmac POINTER_SIZE
1002 Width of a pointer, in bits. You must specify a value no wider than the
1003 width of @code{Pmode}. If it is not equal to the width of @code{Pmode},
1004 you must define @code{POINTERS_EXTEND_UNSIGNED}. If you do not specify
1005 a value the default is @code{BITS_PER_WORD}.
1008 @defmac POINTERS_EXTEND_UNSIGNED
1009 A C expression that determines how pointers should be extended from
1010 @code{ptr_mode} to either @code{Pmode} or @code{word_mode}. It is
1011 greater than zero if pointers should be zero-extended, zero if they
1012 should be sign-extended, and negative if some other sort of conversion
1013 is needed. In the last case, the extension is done by the target's
1014 @code{ptr_extend} instruction.
1016 You need not define this macro if the @code{ptr_mode}, @code{Pmode}
1017 and @code{word_mode} are all the same width.
1020 @defmac PROMOTE_MODE (@var{m}, @var{unsignedp}, @var{type})
1021 A macro to update @var{m} and @var{unsignedp} when an object whose type
1022 is @var{type} and which has the specified mode and signedness is to be
1023 stored in a register. This macro is only called when @var{type} is a
1026 On most RISC machines, which only have operations that operate on a full
1027 register, define this macro to set @var{m} to @code{word_mode} if
1028 @var{m} is an integer mode narrower than @code{BITS_PER_WORD}. In most
1029 cases, only integer modes should be widened because wider-precision
1030 floating-point operations are usually more expensive than their narrower
1033 For most machines, the macro definition does not change @var{unsignedp}.
1034 However, some machines, have instructions that preferentially handle
1035 either signed or unsigned quantities of certain modes. For example, on
1036 the DEC Alpha, 32-bit loads from memory and 32-bit add instructions
1037 sign-extend the result to 64 bits. On such machines, set
1038 @var{unsignedp} according to which kind of extension is more efficient.
1040 Do not define this macro if it would never modify @var{m}.
1043 @deftypefn {Target Hook} enum machine_mode TARGET_PROMOTE_FUNCTION_MODE (tree @var{type}, enum machine_mode @var{mode}, int *@var{punsignedp}, tree @var{funtype}, int @var{for_return})
1044 Like @code{PROMOTE_MODE}, but it is applied to outgoing function arguments or
1045 function return values. The target hook should return the new mode
1046 and possibly change @code{*@var{punsignedp}} if the promotion should
1047 change signedness. This function is called only for scalar @emph{or
1050 @var{for_return} allows to distinguish the promotion of arguments and
1051 return values. If it is @code{1}, a return value is being promoted and
1052 @code{TARGET_FUNCTION_VALUE} must perform the same promotions done here.
1053 If it is @code{2}, the returned mode should be that of the register in
1054 which an incoming parameter is copied, or the outgoing result is computed;
1055 then the hook should return the same mode as @code{promote_mode}, though
1056 the signedness may be different.
1058 The default is to not promote arguments and return values. You can
1059 also define the hook to @code{default_promote_function_mode_always_promote}
1060 if you would like to apply the same rules given by @code{PROMOTE_MODE}.
1063 @defmac PARM_BOUNDARY
1064 Normal alignment required for function parameters on the stack, in
1065 bits. All stack parameters receive at least this much alignment
1066 regardless of data type. On most machines, this is the same as the
1070 @defmac STACK_BOUNDARY
1071 Define this macro to the minimum alignment enforced by hardware for the
1072 stack pointer on this machine. The definition is a C expression for the
1073 desired alignment (measured in bits). This value is used as a default
1074 if @code{PREFERRED_STACK_BOUNDARY} is not defined. On most machines,
1075 this should be the same as @code{PARM_BOUNDARY}.
1078 @defmac PREFERRED_STACK_BOUNDARY
1079 Define this macro if you wish to preserve a certain alignment for the
1080 stack pointer, greater than what the hardware enforces. The definition
1081 is a C expression for the desired alignment (measured in bits). This
1082 macro must evaluate to a value equal to or larger than
1083 @code{STACK_BOUNDARY}.
1086 @defmac INCOMING_STACK_BOUNDARY
1087 Define this macro if the incoming stack boundary may be different
1088 from @code{PREFERRED_STACK_BOUNDARY}. This macro must evaluate
1089 to a value equal to or larger than @code{STACK_BOUNDARY}.
1092 @defmac FUNCTION_BOUNDARY
1093 Alignment required for a function entry point, in bits.
1096 @defmac BIGGEST_ALIGNMENT
1097 Biggest alignment that any data type can require on this machine, in
1098 bits. Note that this is not the biggest alignment that is supported,
1099 just the biggest alignment that, when violated, may cause a fault.
1102 @defmac MALLOC_ABI_ALIGNMENT
1103 Alignment, in bits, a C conformant malloc implementation has to
1104 provide. If not defined, the default value is @code{BITS_PER_WORD}.
1107 @defmac ATTRIBUTE_ALIGNED_VALUE
1108 Alignment used by the @code{__attribute__ ((aligned))} construct. If
1109 not defined, the default value is @code{BIGGEST_ALIGNMENT}.
1112 @defmac MINIMUM_ATOMIC_ALIGNMENT
1113 If defined, the smallest alignment, in bits, that can be given to an
1114 object that can be referenced in one operation, without disturbing any
1115 nearby object. Normally, this is @code{BITS_PER_UNIT}, but may be larger
1116 on machines that don't have byte or half-word store operations.
1119 @defmac BIGGEST_FIELD_ALIGNMENT
1120 Biggest alignment that any structure or union field can require on this
1121 machine, in bits. If defined, this overrides @code{BIGGEST_ALIGNMENT} for
1122 structure and union fields only, unless the field alignment has been set
1123 by the @code{__attribute__ ((aligned (@var{n})))} construct.
1126 @defmac ADJUST_FIELD_ALIGN (@var{field}, @var{computed})
1127 An expression for the alignment of a structure field @var{field} if the
1128 alignment computed in the usual way (including applying of
1129 @code{BIGGEST_ALIGNMENT} and @code{BIGGEST_FIELD_ALIGNMENT} to the
1130 alignment) is @var{computed}. It overrides alignment only if the
1131 field alignment has not been set by the
1132 @code{__attribute__ ((aligned (@var{n})))} construct.
1135 @defmac MAX_STACK_ALIGNMENT
1136 Biggest stack alignment guaranteed by the backend. Use this macro
1137 to specify the maximum alignment of a variable on stack.
1139 If not defined, the default value is @code{STACK_BOUNDARY}.
1141 @c FIXME: The default should be @code{PREFERRED_STACK_BOUNDARY}.
1142 @c But the fix for PR 32893 indicates that we can only guarantee
1143 @c maximum stack alignment on stack up to @code{STACK_BOUNDARY}, not
1144 @c @code{PREFERRED_STACK_BOUNDARY}, if stack alignment isn't supported.
1147 @defmac MAX_OFILE_ALIGNMENT
1148 Biggest alignment supported by the object file format of this machine.
1149 Use this macro to limit the alignment which can be specified using the
1150 @code{__attribute__ ((aligned (@var{n})))} construct. If not defined,
1151 the default value is @code{BIGGEST_ALIGNMENT}.
1153 On systems that use ELF, the default (in @file{config/elfos.h}) is
1154 the largest supported 32-bit ELF section alignment representable on
1155 a 32-bit host e.g. @samp{(((unsigned HOST_WIDEST_INT) 1 << 28) * 8)}.
1156 On 32-bit ELF the largest supported section alignment in bits is
1157 @samp{(0x80000000 * 8)}, but this is not representable on 32-bit hosts.
1160 @defmac DATA_ALIGNMENT (@var{type}, @var{basic-align})
1161 If defined, a C expression to compute the alignment for a variable in
1162 the static store. @var{type} is the data type, and @var{basic-align} is
1163 the alignment that the object would ordinarily have. The value of this
1164 macro is used instead of that alignment to align the object.
1166 If this macro is not defined, then @var{basic-align} is used.
1169 One use of this macro is to increase alignment of medium-size data to
1170 make it all fit in fewer cache lines. Another is to cause character
1171 arrays to be word-aligned so that @code{strcpy} calls that copy
1172 constants to character arrays can be done inline.
1175 @defmac CONSTANT_ALIGNMENT (@var{constant}, @var{basic-align})
1176 If defined, a C expression to compute the alignment given to a constant
1177 that is being placed in memory. @var{constant} is the constant and
1178 @var{basic-align} is the alignment that the object would ordinarily
1179 have. The value of this macro is used instead of that alignment to
1182 If this macro is not defined, then @var{basic-align} is used.
1184 The typical use of this macro is to increase alignment for string
1185 constants to be word aligned so that @code{strcpy} calls that copy
1186 constants can be done inline.
1189 @defmac LOCAL_ALIGNMENT (@var{type}, @var{basic-align})
1190 If defined, a C expression to compute the alignment for a variable in
1191 the local store. @var{type} is the data type, and @var{basic-align} is
1192 the alignment that the object would ordinarily have. The value of this
1193 macro is used instead of that alignment to align the object.
1195 If this macro is not defined, then @var{basic-align} is used.
1197 One use of this macro is to increase alignment of medium-size data to
1198 make it all fit in fewer cache lines.
1201 @defmac STACK_SLOT_ALIGNMENT (@var{type}, @var{mode}, @var{basic-align})
1202 If defined, a C expression to compute the alignment for stack slot.
1203 @var{type} is the data type, @var{mode} is the widest mode available,
1204 and @var{basic-align} is the alignment that the slot would ordinarily
1205 have. The value of this macro is used instead of that alignment to
1208 If this macro is not defined, then @var{basic-align} is used when
1209 @var{type} is @code{NULL}. Otherwise, @code{LOCAL_ALIGNMENT} will
1212 This macro is to set alignment of stack slot to the maximum alignment
1213 of all possible modes which the slot may have.
1216 @defmac LOCAL_DECL_ALIGNMENT (@var{decl})
1217 If defined, a C expression to compute the alignment for a local
1218 variable @var{decl}.
1220 If this macro is not defined, then
1221 @code{LOCAL_ALIGNMENT (TREE_TYPE (@var{decl}), DECL_ALIGN (@var{decl}))}
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 MINIMUM_ALIGNMENT (@var{exp}, @var{mode}, @var{align})
1229 If defined, a C expression to compute the minimum required alignment
1230 for dynamic stack realignment purposes for @var{exp} (a type or decl),
1231 @var{mode}, assuming normal alignment @var{align}.
1233 If this macro is not defined, then @var{align} will be used.
1236 @defmac EMPTY_FIELD_BOUNDARY
1237 Alignment in bits to be given to a structure bit-field that follows an
1238 empty field such as @code{int : 0;}.
1240 If @code{PCC_BITFIELD_TYPE_MATTERS} is true, it overrides this macro.
1243 @defmac STRUCTURE_SIZE_BOUNDARY
1244 Number of bits which any structure or union's size must be a multiple of.
1245 Each structure or union's size is rounded up to a multiple of this.
1247 If you do not define this macro, the default is the same as
1248 @code{BITS_PER_UNIT}.
1251 @defmac STRICT_ALIGNMENT
1252 Define this macro to be the value 1 if instructions will fail to work
1253 if given data not on the nominal alignment. If instructions will merely
1254 go slower in that case, define this macro as 0.
1257 @defmac PCC_BITFIELD_TYPE_MATTERS
1258 Define this if you wish to imitate the way many other C compilers handle
1259 alignment of bit-fields and the structures that contain them.
1261 The behavior is that the type written for a named bit-field (@code{int},
1262 @code{short}, or other integer type) imposes an alignment for the entire
1263 structure, as if the structure really did contain an ordinary field of
1264 that type. In addition, the bit-field is placed within the structure so
1265 that it would fit within such a field, not crossing a boundary for it.
1267 Thus, on most machines, a named bit-field whose type is written as
1268 @code{int} would not cross a four-byte boundary, and would force
1269 four-byte alignment for the whole structure. (The alignment used may
1270 not be four bytes; it is controlled by the other alignment parameters.)
1272 An unnamed bit-field will not affect the alignment of the containing
1275 If the macro is defined, its definition should be a C expression;
1276 a nonzero value for the expression enables this behavior.
1278 Note that if this macro is not defined, or its value is zero, some
1279 bit-fields may cross more than one alignment boundary. The compiler can
1280 support such references if there are @samp{insv}, @samp{extv}, and
1281 @samp{extzv} insns that can directly reference memory.
1283 The other known way of making bit-fields work is to define
1284 @code{STRUCTURE_SIZE_BOUNDARY} as large as @code{BIGGEST_ALIGNMENT}.
1285 Then every structure can be accessed with fullwords.
1287 Unless the machine has bit-field instructions or you define
1288 @code{STRUCTURE_SIZE_BOUNDARY} that way, you must define
1289 @code{PCC_BITFIELD_TYPE_MATTERS} to have a nonzero value.
1291 If your aim is to make GCC use the same conventions for laying out
1292 bit-fields as are used by another compiler, here is how to investigate
1293 what the other compiler does. Compile and run this program:
1312 printf ("Size of foo1 is %d\n",
1313 sizeof (struct foo1));
1314 printf ("Size of foo2 is %d\n",
1315 sizeof (struct foo2));
1320 If this prints 2 and 5, then the compiler's behavior is what you would
1321 get from @code{PCC_BITFIELD_TYPE_MATTERS}.
1324 @defmac BITFIELD_NBYTES_LIMITED
1325 Like @code{PCC_BITFIELD_TYPE_MATTERS} except that its effect is limited
1326 to aligning a bit-field within the structure.
1329 @deftypefn {Target Hook} bool TARGET_ALIGN_ANON_BITFIELD (void)
1330 When @code{PCC_BITFIELD_TYPE_MATTERS} is true this hook will determine
1331 whether unnamed bitfields affect the alignment of the containing
1332 structure. The hook should return true if the structure should inherit
1333 the alignment requirements of an unnamed bitfield's type.
1336 @deftypefn {Target Hook} bool TARGET_NARROW_VOLATILE_BITFIELD (void)
1337 This target hook should return @code{true} if accesses to volatile bitfields
1338 should use the narrowest mode possible. It should return @code{false} if
1339 these accesses should use the bitfield container type.
1341 The default is @code{!TARGET_STRICT_ALIGN}.
1344 @defmac MEMBER_TYPE_FORCES_BLK (@var{field}, @var{mode})
1345 Return 1 if a structure or array containing @var{field} should be accessed using
1348 If @var{field} is the only field in the structure, @var{mode} is its
1349 mode, otherwise @var{mode} is VOIDmode. @var{mode} is provided in the
1350 case where structures of one field would require the structure's mode to
1351 retain the field's mode.
1353 Normally, this is not needed.
1356 @defmac ROUND_TYPE_ALIGN (@var{type}, @var{computed}, @var{specified})
1357 Define this macro as an expression for the alignment of a type (given
1358 by @var{type} as a tree node) if the alignment computed in the usual
1359 way is @var{computed} and the alignment explicitly specified was
1362 The default is to use @var{specified} if it is larger; otherwise, use
1363 the smaller of @var{computed} and @code{BIGGEST_ALIGNMENT}
1366 @defmac MAX_FIXED_MODE_SIZE
1367 An integer expression for the size in bits of the largest integer
1368 machine mode that should actually be used. All integer machine modes of
1369 this size or smaller can be used for structures and unions with the
1370 appropriate sizes. If this macro is undefined, @code{GET_MODE_BITSIZE
1371 (DImode)} is assumed.
1374 @defmac STACK_SAVEAREA_MODE (@var{save_level})
1375 If defined, an expression of type @code{enum machine_mode} that
1376 specifies the mode of the save area operand of a
1377 @code{save_stack_@var{level}} named pattern (@pxref{Standard Names}).
1378 @var{save_level} is one of @code{SAVE_BLOCK}, @code{SAVE_FUNCTION}, or
1379 @code{SAVE_NONLOCAL} and selects which of the three named patterns is
1380 having its mode specified.
1382 You need not define this macro if it always returns @code{Pmode}. You
1383 would most commonly define this macro if the
1384 @code{save_stack_@var{level}} patterns need to support both a 32- and a
1388 @defmac STACK_SIZE_MODE
1389 If defined, an expression of type @code{enum machine_mode} that
1390 specifies the mode of the size increment operand of an
1391 @code{allocate_stack} named pattern (@pxref{Standard Names}).
1393 You need not define this macro if it always returns @code{word_mode}.
1394 You would most commonly define this macro if the @code{allocate_stack}
1395 pattern needs to support both a 32- and a 64-bit mode.
1398 @deftypefn {Target Hook} {enum machine_mode} TARGET_LIBGCC_CMP_RETURN_MODE ()
1399 This target hook should return the mode to be used for the return value
1400 of compare instructions expanded to libgcc calls. If not defined
1401 @code{word_mode} is returned which is the right choice for a majority of
1405 @deftypefn {Target Hook} {enum machine_mode} TARGET_LIBGCC_SHIFT_COUNT_MODE ()
1406 This target hook should return the mode to be used for the shift count operand
1407 of shift instructions expanded to libgcc calls. If not defined
1408 @code{word_mode} is returned which is the right choice for a majority of
1412 @defmac ROUND_TOWARDS_ZERO
1413 If defined, this macro should be true if the prevailing rounding
1414 mode is towards zero.
1416 Defining this macro only affects the way @file{libgcc.a} emulates
1417 floating-point arithmetic.
1419 Not defining this macro is equivalent to returning zero.
1422 @defmac LARGEST_EXPONENT_IS_NORMAL (@var{size})
1423 This macro should return true if floats with @var{size}
1424 bits do not have a NaN or infinity representation, but use the largest
1425 exponent for normal numbers instead.
1427 Defining this macro only affects the way @file{libgcc.a} emulates
1428 floating-point arithmetic.
1430 The default definition of this macro returns false for all sizes.
1433 @deftypefn {Target Hook} bool TARGET_MS_BITFIELD_LAYOUT_P (tree @var{record_type})
1434 This target hook returns @code{true} if bit-fields in the given
1435 @var{record_type} are to be laid out following the rules of Microsoft
1436 Visual C/C++, namely: (i) a bit-field won't share the same storage
1437 unit with the previous bit-field if their underlying types have
1438 different sizes, and the bit-field will be aligned to the highest
1439 alignment of the underlying types of itself and of the previous
1440 bit-field; (ii) a zero-sized bit-field will affect the alignment of
1441 the whole enclosing structure, even if it is unnamed; except that
1442 (iii) a zero-sized bit-field will be disregarded unless it follows
1443 another bit-field of nonzero size. If this hook returns @code{true},
1444 other macros that control bit-field layout are ignored.
1446 When a bit-field is inserted into a packed record, the whole size
1447 of the underlying type is used by one or more same-size adjacent
1448 bit-fields (that is, if its long:3, 32 bits is used in the record,
1449 and any additional adjacent long bit-fields are packed into the same
1450 chunk of 32 bits. However, if the size changes, a new field of that
1451 size is allocated). In an unpacked record, this is the same as using
1452 alignment, but not equivalent when packing.
1454 If both MS bit-fields and @samp{__attribute__((packed))} are used,
1455 the latter will take precedence. If @samp{__attribute__((packed))} is
1456 used on a single field when MS bit-fields are in use, it will take
1457 precedence for that field, but the alignment of the rest of the structure
1458 may affect its placement.
1461 @deftypefn {Target Hook} {bool} TARGET_DECIMAL_FLOAT_SUPPORTED_P (void)
1462 Returns true if the target supports decimal floating point.
1465 @deftypefn {Target Hook} {bool} TARGET_FIXED_POINT_SUPPORTED_P (void)
1466 Returns true if the target supports fixed-point arithmetic.
1469 @deftypefn {Target Hook} void TARGET_EXPAND_TO_RTL_HOOK (void)
1470 This hook is called just before expansion into rtl, allowing the target
1471 to perform additional initializations or analysis before the expansion.
1472 For example, the rs6000 port uses it to allocate a scratch stack slot
1473 for use in copying SDmode values between memory and floating point
1474 registers whenever the function being expanded has any SDmode
1478 @deftypefn {Target Hook} void TARGET_INSTANTIATE_DECLS (void)
1479 This hook allows the backend to perform additional instantiations on rtl
1480 that are not actually in any insns yet, but will be later.
1483 @deftypefn {Target Hook} {const char *} TARGET_MANGLE_TYPE (tree @var{type})
1484 If your target defines any fundamental types, or any types your target
1485 uses should be mangled differently from the default, define this hook
1486 to return the appropriate encoding for these types as part of a C++
1487 mangled name. The @var{type} argument is the tree structure representing
1488 the type to be mangled. The hook may be applied to trees which are
1489 not target-specific fundamental types; it should return @code{NULL}
1490 for all such types, as well as arguments it does not recognize. If the
1491 return value is not @code{NULL}, it must point to a statically-allocated
1494 Target-specific fundamental types might be new fundamental types or
1495 qualified versions of ordinary fundamental types. Encode new
1496 fundamental types as @samp{@w{u @var{n} @var{name}}}, where @var{name}
1497 is the name used for the type in source code, and @var{n} is the
1498 length of @var{name} in decimal. Encode qualified versions of
1499 ordinary types as @samp{@w{U @var{n} @var{name} @var{code}}}, where
1500 @var{name} is the name used for the type qualifier in source code,
1501 @var{n} is the length of @var{name} as above, and @var{code} is the
1502 code used to represent the unqualified version of this type. (See
1503 @code{write_builtin_type} in @file{cp/mangle.c} for the list of
1504 codes.) In both cases the spaces are for clarity; do not include any
1505 spaces in your string.
1507 This hook is applied to types prior to typedef resolution. If the mangled
1508 name for a particular type depends only on that type's main variant, you
1509 can perform typedef resolution yourself using @code{TYPE_MAIN_VARIANT}
1512 The default version of this hook always returns @code{NULL}, which is
1513 appropriate for a target that does not define any new fundamental
1518 @section Layout of Source Language Data Types
1520 These macros define the sizes and other characteristics of the standard
1521 basic data types used in programs being compiled. Unlike the macros in
1522 the previous section, these apply to specific features of C and related
1523 languages, rather than to fundamental aspects of storage layout.
1525 @defmac INT_TYPE_SIZE
1526 A C expression for the size in bits of the type @code{int} on the
1527 target machine. If you don't define this, the default is one word.
1530 @defmac SHORT_TYPE_SIZE
1531 A C expression for the size in bits of the type @code{short} on the
1532 target machine. If you don't define this, the default is half a word.
1533 (If this would be less than one storage unit, it is rounded up to one
1537 @defmac LONG_TYPE_SIZE
1538 A C expression for the size in bits of the type @code{long} on the
1539 target machine. If you don't define this, the default is one word.
1542 @defmac ADA_LONG_TYPE_SIZE
1543 On some machines, the size used for the Ada equivalent of the type
1544 @code{long} by a native Ada compiler differs from that used by C@. In
1545 that situation, define this macro to be a C expression to be used for
1546 the size of that type. If you don't define this, the default is the
1547 value of @code{LONG_TYPE_SIZE}.
1550 @defmac LONG_LONG_TYPE_SIZE
1551 A C expression for the size in bits of the type @code{long long} on the
1552 target machine. If you don't define this, the default is two
1553 words. If you want to support GNU Ada on your machine, the value of this
1554 macro must be at least 64.
1557 @defmac CHAR_TYPE_SIZE
1558 A C expression for the size in bits of the type @code{char} on the
1559 target machine. If you don't define this, the default is
1560 @code{BITS_PER_UNIT}.
1563 @defmac BOOL_TYPE_SIZE
1564 A C expression for the size in bits of the C++ type @code{bool} and
1565 C99 type @code{_Bool} on the target machine. If you don't define
1566 this, and you probably shouldn't, the default is @code{CHAR_TYPE_SIZE}.
1569 @defmac FLOAT_TYPE_SIZE
1570 A C expression for the size in bits of the type @code{float} on the
1571 target machine. If you don't define this, the default is one word.
1574 @defmac DOUBLE_TYPE_SIZE
1575 A C expression for the size in bits of the type @code{double} on the
1576 target machine. If you don't define this, the default is two
1580 @defmac LONG_DOUBLE_TYPE_SIZE
1581 A C expression for the size in bits of the type @code{long double} on
1582 the target machine. If you don't define this, the default is two
1586 @defmac SHORT_FRACT_TYPE_SIZE
1587 A C expression for the size in bits of the type @code{short _Fract} on
1588 the target machine. If you don't define this, the default is
1589 @code{BITS_PER_UNIT}.
1592 @defmac FRACT_TYPE_SIZE
1593 A C expression for the size in bits of the type @code{_Fract} on
1594 the target machine. If you don't define this, the default is
1595 @code{BITS_PER_UNIT * 2}.
1598 @defmac LONG_FRACT_TYPE_SIZE
1599 A C expression for the size in bits of the type @code{long _Fract} on
1600 the target machine. If you don't define this, the default is
1601 @code{BITS_PER_UNIT * 4}.
1604 @defmac LONG_LONG_FRACT_TYPE_SIZE
1605 A C expression for the size in bits of the type @code{long long _Fract} on
1606 the target machine. If you don't define this, the default is
1607 @code{BITS_PER_UNIT * 8}.
1610 @defmac SHORT_ACCUM_TYPE_SIZE
1611 A C expression for the size in bits of the type @code{short _Accum} on
1612 the target machine. If you don't define this, the default is
1613 @code{BITS_PER_UNIT * 2}.
1616 @defmac ACCUM_TYPE_SIZE
1617 A C expression for the size in bits of the type @code{_Accum} on
1618 the target machine. If you don't define this, the default is
1619 @code{BITS_PER_UNIT * 4}.
1622 @defmac LONG_ACCUM_TYPE_SIZE
1623 A C expression for the size in bits of the type @code{long _Accum} on
1624 the target machine. If you don't define this, the default is
1625 @code{BITS_PER_UNIT * 8}.
1628 @defmac LONG_LONG_ACCUM_TYPE_SIZE
1629 A C expression for the size in bits of the type @code{long long _Accum} on
1630 the target machine. If you don't define this, the default is
1631 @code{BITS_PER_UNIT * 16}.
1634 @defmac LIBGCC2_LONG_DOUBLE_TYPE_SIZE
1635 Define this macro if @code{LONG_DOUBLE_TYPE_SIZE} is not constant or
1636 if you want routines in @file{libgcc2.a} for a size other than
1637 @code{LONG_DOUBLE_TYPE_SIZE}. If you don't define this, the
1638 default is @code{LONG_DOUBLE_TYPE_SIZE}.
1641 @defmac LIBGCC2_HAS_DF_MODE
1642 Define this macro if neither @code{LIBGCC2_DOUBLE_TYPE_SIZE} nor
1643 @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is
1644 @code{DFmode} but you want @code{DFmode} routines in @file{libgcc2.a}
1645 anyway. If you don't define this and either @code{LIBGCC2_DOUBLE_TYPE_SIZE}
1646 or @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is 64 then the default is 1,
1650 @defmac LIBGCC2_HAS_XF_MODE
1651 Define this macro if @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is not
1652 @code{XFmode} but you want @code{XFmode} routines in @file{libgcc2.a}
1653 anyway. If you don't define this and @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE}
1654 is 80 then the default is 1, otherwise it is 0.
1657 @defmac LIBGCC2_HAS_TF_MODE
1658 Define this macro if @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is not
1659 @code{TFmode} but you want @code{TFmode} routines in @file{libgcc2.a}
1660 anyway. If you don't define this and @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE}
1661 is 128 then the default is 1, otherwise it is 0.
1668 Define these macros to be the size in bits of the mantissa of
1669 @code{SFmode}, @code{DFmode}, @code{XFmode} and @code{TFmode} values,
1670 if the defaults in @file{libgcc2.h} are inappropriate. By default,
1671 @code{FLT_MANT_DIG} is used for @code{SF_SIZE}, @code{LDBL_MANT_DIG}
1672 for @code{XF_SIZE} and @code{TF_SIZE}, and @code{DBL_MANT_DIG} or
1673 @code{LDBL_MANT_DIG} for @code{DF_SIZE} according to whether
1674 @code{LIBGCC2_DOUBLE_TYPE_SIZE} or
1675 @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is 64.
1678 @defmac TARGET_FLT_EVAL_METHOD
1679 A C expression for the value for @code{FLT_EVAL_METHOD} in @file{float.h},
1680 assuming, if applicable, that the floating-point control word is in its
1681 default state. If you do not define this macro the value of
1682 @code{FLT_EVAL_METHOD} will be zero.
1685 @defmac WIDEST_HARDWARE_FP_SIZE
1686 A C expression for the size in bits of the widest floating-point format
1687 supported by the hardware. If you define this macro, you must specify a
1688 value less than or equal to the value of @code{LONG_DOUBLE_TYPE_SIZE}.
1689 If you do not define this macro, the value of @code{LONG_DOUBLE_TYPE_SIZE}
1693 @defmac DEFAULT_SIGNED_CHAR
1694 An expression whose value is 1 or 0, according to whether the type
1695 @code{char} should be signed or unsigned by default. The user can
1696 always override this default with the options @option{-fsigned-char}
1697 and @option{-funsigned-char}.
1700 @deftypefn {Target Hook} bool TARGET_DEFAULT_SHORT_ENUMS (void)
1701 This target hook should return true if the compiler should give an
1702 @code{enum} type only as many bytes as it takes to represent the range
1703 of possible values of that type. It should return false if all
1704 @code{enum} types should be allocated like @code{int}.
1706 The default is to return false.
1710 A C expression for a string describing the name of the data type to use
1711 for size values. The typedef name @code{size_t} is defined using the
1712 contents of the string.
1714 The string can contain more than one keyword. If so, separate them with
1715 spaces, and write first any length keyword, then @code{unsigned} if
1716 appropriate, and finally @code{int}. The string must exactly match one
1717 of the data type names defined in the function
1718 @code{init_decl_processing} in the file @file{c-decl.c}. You may not
1719 omit @code{int} or change the order---that would cause the compiler to
1722 If you don't define this macro, the default is @code{"long unsigned
1726 @defmac PTRDIFF_TYPE
1727 A C expression for a string describing the name of the data type to use
1728 for the result of subtracting two pointers. The typedef name
1729 @code{ptrdiff_t} is defined using the contents of the string. See
1730 @code{SIZE_TYPE} above for more information.
1732 If you don't define this macro, the default is @code{"long int"}.
1736 A C expression for a string describing the name of the data type to use
1737 for wide characters. The typedef name @code{wchar_t} is defined using
1738 the contents of the string. See @code{SIZE_TYPE} above for more
1741 If you don't define this macro, the default is @code{"int"}.
1744 @defmac WCHAR_TYPE_SIZE
1745 A C expression for the size in bits of the data type for wide
1746 characters. This is used in @code{cpp}, which cannot make use of
1751 A C expression for a string describing the name of the data type to
1752 use for wide characters passed to @code{printf} and returned from
1753 @code{getwc}. The typedef name @code{wint_t} is defined using the
1754 contents of the string. See @code{SIZE_TYPE} above for more
1757 If you don't define this macro, the default is @code{"unsigned int"}.
1761 A C expression for a string describing the name of the data type that
1762 can represent any value of any standard or extended signed integer type.
1763 The typedef name @code{intmax_t} is defined using the contents of the
1764 string. See @code{SIZE_TYPE} above for more information.
1766 If you don't define this macro, the default is the first of
1767 @code{"int"}, @code{"long int"}, or @code{"long long int"} that has as
1768 much precision as @code{long long int}.
1771 @defmac UINTMAX_TYPE
1772 A C expression for a string describing the name of the data type that
1773 can represent any value of any standard or extended unsigned integer
1774 type. The typedef name @code{uintmax_t} is defined using the contents
1775 of the string. See @code{SIZE_TYPE} above for more information.
1777 If you don't define this macro, the default is the first of
1778 @code{"unsigned int"}, @code{"long unsigned int"}, or @code{"long long
1779 unsigned int"} that has as much precision as @code{long long unsigned
1783 @defmac SIG_ATOMIC_TYPE
1789 @defmacx UINT16_TYPE
1790 @defmacx UINT32_TYPE
1791 @defmacx UINT64_TYPE
1792 @defmacx INT_LEAST8_TYPE
1793 @defmacx INT_LEAST16_TYPE
1794 @defmacx INT_LEAST32_TYPE
1795 @defmacx INT_LEAST64_TYPE
1796 @defmacx UINT_LEAST8_TYPE
1797 @defmacx UINT_LEAST16_TYPE
1798 @defmacx UINT_LEAST32_TYPE
1799 @defmacx UINT_LEAST64_TYPE
1800 @defmacx INT_FAST8_TYPE
1801 @defmacx INT_FAST16_TYPE
1802 @defmacx INT_FAST32_TYPE
1803 @defmacx INT_FAST64_TYPE
1804 @defmacx UINT_FAST8_TYPE
1805 @defmacx UINT_FAST16_TYPE
1806 @defmacx UINT_FAST32_TYPE
1807 @defmacx UINT_FAST64_TYPE
1808 @defmacx INTPTR_TYPE
1809 @defmacx UINTPTR_TYPE
1810 C expressions for the standard types @code{sig_atomic_t},
1811 @code{int8_t}, @code{int16_t}, @code{int32_t}, @code{int64_t},
1812 @code{uint8_t}, @code{uint16_t}, @code{uint32_t}, @code{uint64_t},
1813 @code{int_least8_t}, @code{int_least16_t}, @code{int_least32_t},
1814 @code{int_least64_t}, @code{uint_least8_t}, @code{uint_least16_t},
1815 @code{uint_least32_t}, @code{uint_least64_t}, @code{int_fast8_t},
1816 @code{int_fast16_t}, @code{int_fast32_t}, @code{int_fast64_t},
1817 @code{uint_fast8_t}, @code{uint_fast16_t}, @code{uint_fast32_t},
1818 @code{uint_fast64_t}, @code{intptr_t}, and @code{uintptr_t}. See
1819 @code{SIZE_TYPE} above for more information.
1821 If any of these macros evaluates to a null pointer, the corresponding
1822 type is not supported; if GCC is configured to provide
1823 @code{<stdint.h>} in such a case, the header provided may not conform
1824 to C99, depending on the type in question. The defaults for all of
1825 these macros are null pointers.
1828 @defmac TARGET_PTRMEMFUNC_VBIT_LOCATION
1829 The C++ compiler represents a pointer-to-member-function with a struct
1836 ptrdiff_t vtable_index;
1843 The C++ compiler must use one bit to indicate whether the function that
1844 will be called through a pointer-to-member-function is virtual.
1845 Normally, we assume that the low-order bit of a function pointer must
1846 always be zero. Then, by ensuring that the vtable_index is odd, we can
1847 distinguish which variant of the union is in use. But, on some
1848 platforms function pointers can be odd, and so this doesn't work. In
1849 that case, we use the low-order bit of the @code{delta} field, and shift
1850 the remainder of the @code{delta} field to the left.
1852 GCC will automatically make the right selection about where to store
1853 this bit using the @code{FUNCTION_BOUNDARY} setting for your platform.
1854 However, some platforms such as ARM/Thumb have @code{FUNCTION_BOUNDARY}
1855 set such that functions always start at even addresses, but the lowest
1856 bit of pointers to functions indicate whether the function at that
1857 address is in ARM or Thumb mode. If this is the case of your
1858 architecture, you should define this macro to
1859 @code{ptrmemfunc_vbit_in_delta}.
1861 In general, you should not have to define this macro. On architectures
1862 in which function addresses are always even, according to
1863 @code{FUNCTION_BOUNDARY}, GCC will automatically define this macro to
1864 @code{ptrmemfunc_vbit_in_pfn}.
1867 @defmac TARGET_VTABLE_USES_DESCRIPTORS
1868 Normally, the C++ compiler uses function pointers in vtables. This
1869 macro allows the target to change to use ``function descriptors''
1870 instead. Function descriptors are found on targets for whom a
1871 function pointer is actually a small data structure. Normally the
1872 data structure consists of the actual code address plus a data
1873 pointer to which the function's data is relative.
1875 If vtables are used, the value of this macro should be the number
1876 of words that the function descriptor occupies.
1879 @defmac TARGET_VTABLE_ENTRY_ALIGN
1880 By default, the vtable entries are void pointers, the so the alignment
1881 is the same as pointer alignment. The value of this macro specifies
1882 the alignment of the vtable entry in bits. It should be defined only
1883 when special alignment is necessary. */
1886 @defmac TARGET_VTABLE_DATA_ENTRY_DISTANCE
1887 There are a few non-descriptor entries in the vtable at offsets below
1888 zero. If these entries must be padded (say, to preserve the alignment
1889 specified by @code{TARGET_VTABLE_ENTRY_ALIGN}), set this to the number
1890 of words in each data entry.
1894 @section Register Usage
1895 @cindex register usage
1897 This section explains how to describe what registers the target machine
1898 has, and how (in general) they can be used.
1900 The description of which registers a specific instruction can use is
1901 done with register classes; see @ref{Register Classes}. For information
1902 on using registers to access a stack frame, see @ref{Frame Registers}.
1903 For passing values in registers, see @ref{Register Arguments}.
1904 For returning values in registers, see @ref{Scalar Return}.
1907 * Register Basics:: Number and kinds of registers.
1908 * Allocation Order:: Order in which registers are allocated.
1909 * Values in Registers:: What kinds of values each reg can hold.
1910 * Leaf Functions:: Renumbering registers for leaf functions.
1911 * Stack Registers:: Handling a register stack such as 80387.
1914 @node Register Basics
1915 @subsection Basic Characteristics of Registers
1917 @c prevent bad page break with this line
1918 Registers have various characteristics.
1920 @defmac FIRST_PSEUDO_REGISTER
1921 Number of hardware registers known to the compiler. They receive
1922 numbers 0 through @code{FIRST_PSEUDO_REGISTER-1}; thus, the first
1923 pseudo register's number really is assigned the number
1924 @code{FIRST_PSEUDO_REGISTER}.
1927 @defmac FIXED_REGISTERS
1928 @cindex fixed register
1929 An initializer that says which registers are used for fixed purposes
1930 all throughout the compiled code and are therefore not available for
1931 general allocation. These would include the stack pointer, the frame
1932 pointer (except on machines where that can be used as a general
1933 register when no frame pointer is needed), the program counter on
1934 machines where that is considered one of the addressable registers,
1935 and any other numbered register with a standard use.
1937 This information is expressed as a sequence of numbers, separated by
1938 commas and surrounded by braces. The @var{n}th number is 1 if
1939 register @var{n} is fixed, 0 otherwise.
1941 The table initialized from this macro, and the table initialized by
1942 the following one, may be overridden at run time either automatically,
1943 by the actions of the macro @code{CONDITIONAL_REGISTER_USAGE}, or by
1944 the user with the command options @option{-ffixed-@var{reg}},
1945 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}.
1948 @defmac CALL_USED_REGISTERS
1949 @cindex call-used register
1950 @cindex call-clobbered register
1951 @cindex call-saved register
1952 Like @code{FIXED_REGISTERS} but has 1 for each register that is
1953 clobbered (in general) by function calls as well as for fixed
1954 registers. This macro therefore identifies the registers that are not
1955 available for general allocation of values that must live across
1958 If a register has 0 in @code{CALL_USED_REGISTERS}, the compiler
1959 automatically saves it on function entry and restores it on function
1960 exit, if the register is used within the function.
1963 @defmac CALL_REALLY_USED_REGISTERS
1964 @cindex call-used register
1965 @cindex call-clobbered register
1966 @cindex call-saved register
1967 Like @code{CALL_USED_REGISTERS} except this macro doesn't require
1968 that the entire set of @code{FIXED_REGISTERS} be included.
1969 (@code{CALL_USED_REGISTERS} must be a superset of @code{FIXED_REGISTERS}).
1970 This macro is optional. If not specified, it defaults to the value
1971 of @code{CALL_USED_REGISTERS}.
1974 @defmac HARD_REGNO_CALL_PART_CLOBBERED (@var{regno}, @var{mode})
1975 @cindex call-used register
1976 @cindex call-clobbered register
1977 @cindex call-saved register
1978 A C expression that is nonzero if it is not permissible to store a
1979 value of mode @var{mode} in hard register number @var{regno} across a
1980 call without some part of it being clobbered. For most machines this
1981 macro need not be defined. It is only required for machines that do not
1982 preserve the entire contents of a register across a call.
1986 @findex call_used_regs
1989 @findex reg_class_contents
1990 @defmac CONDITIONAL_REGISTER_USAGE
1991 Zero or more C statements that may conditionally modify five variables
1992 @code{fixed_regs}, @code{call_used_regs}, @code{global_regs},
1993 @code{reg_names}, and @code{reg_class_contents}, to take into account
1994 any dependence of these register sets on target flags. The first three
1995 of these are of type @code{char []} (interpreted as Boolean vectors).
1996 @code{global_regs} is a @code{const char *[]}, and
1997 @code{reg_class_contents} is a @code{HARD_REG_SET}. Before the macro is
1998 called, @code{fixed_regs}, @code{call_used_regs},
1999 @code{reg_class_contents}, and @code{reg_names} have been initialized
2000 from @code{FIXED_REGISTERS}, @code{CALL_USED_REGISTERS},
2001 @code{REG_CLASS_CONTENTS}, and @code{REGISTER_NAMES}, respectively.
2002 @code{global_regs} has been cleared, and any @option{-ffixed-@var{reg}},
2003 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}
2004 command options have been applied.
2006 You need not define this macro if it has no work to do.
2008 @cindex disabling certain registers
2009 @cindex controlling register usage
2010 If the usage of an entire class of registers depends on the target
2011 flags, you may indicate this to GCC by using this macro to modify
2012 @code{fixed_regs} and @code{call_used_regs} to 1 for each of the
2013 registers in the classes which should not be used by GCC@. Also define
2014 the macro @code{REG_CLASS_FROM_LETTER} / @code{REG_CLASS_FROM_CONSTRAINT}
2015 to return @code{NO_REGS} if it
2016 is called with a letter for a class that shouldn't be used.
2018 (However, if this class is not included in @code{GENERAL_REGS} and all
2019 of the insn patterns whose constraints permit this class are
2020 controlled by target switches, then GCC will automatically avoid using
2021 these registers when the target switches are opposed to them.)
2024 @defmac INCOMING_REGNO (@var{out})
2025 Define this macro if the target machine has register windows. This C
2026 expression returns the register number as seen by the called function
2027 corresponding to the register number @var{out} as seen by the calling
2028 function. Return @var{out} if register number @var{out} is not an
2032 @defmac OUTGOING_REGNO (@var{in})
2033 Define this macro if the target machine has register windows. This C
2034 expression returns the register number as seen by the calling function
2035 corresponding to the register number @var{in} as seen by the called
2036 function. Return @var{in} if register number @var{in} is not an inbound
2040 @defmac LOCAL_REGNO (@var{regno})
2041 Define this macro if the target machine has register windows. This C
2042 expression returns true if the register is call-saved but is in the
2043 register window. Unlike most call-saved registers, such registers
2044 need not be explicitly restored on function exit or during non-local
2049 If the program counter has a register number, define this as that
2050 register number. Otherwise, do not define it.
2053 @node Allocation Order
2054 @subsection Order of Allocation of Registers
2055 @cindex order of register allocation
2056 @cindex register allocation order
2058 @c prevent bad page break with this line
2059 Registers are allocated in order.
2061 @defmac REG_ALLOC_ORDER
2062 If defined, an initializer for a vector of integers, containing the
2063 numbers of hard registers in the order in which GCC should prefer
2064 to use them (from most preferred to least).
2066 If this macro is not defined, registers are used lowest numbered first
2067 (all else being equal).
2069 One use of this macro is on machines where the highest numbered
2070 registers must always be saved and the save-multiple-registers
2071 instruction supports only sequences of consecutive registers. On such
2072 machines, define @code{REG_ALLOC_ORDER} to be an initializer that lists
2073 the highest numbered allocable register first.
2076 @defmac ORDER_REGS_FOR_LOCAL_ALLOC
2077 A C statement (sans semicolon) to choose the order in which to allocate
2078 hard registers for pseudo-registers local to a basic block.
2080 Store the desired register order in the array @code{reg_alloc_order}.
2081 Element 0 should be the register to allocate first; element 1, the next
2082 register; and so on.
2084 The macro body should not assume anything about the contents of
2085 @code{reg_alloc_order} before execution of the macro.
2087 On most machines, it is not necessary to define this macro.
2090 @defmac IRA_HARD_REGNO_ADD_COST_MULTIPLIER (@var{regno})
2091 In some case register allocation order is not enough for the
2092 Integrated Register Allocator (@acronym{IRA}) to generate a good code.
2093 If this macro is defined, it should return a floating point value
2094 based on @var{regno}. The cost of using @var{regno} for a pseudo will
2095 be increased by approximately the pseudo's usage frequency times the
2096 value returned by this macro. Not defining this macro is equivalent
2097 to having it always return @code{0.0}.
2099 On most machines, it is not necessary to define this macro.
2102 @node Values in Registers
2103 @subsection How Values Fit in Registers
2105 This section discusses the macros that describe which kinds of values
2106 (specifically, which machine modes) each register can hold, and how many
2107 consecutive registers are needed for a given mode.
2109 @defmac HARD_REGNO_NREGS (@var{regno}, @var{mode})
2110 A C expression for the number of consecutive hard registers, starting
2111 at register number @var{regno}, required to hold a value of mode
2112 @var{mode}. This macro must never return zero, even if a register
2113 cannot hold the requested mode - indicate that with HARD_REGNO_MODE_OK
2114 and/or CANNOT_CHANGE_MODE_CLASS instead.
2116 On a machine where all registers are exactly one word, a suitable
2117 definition of this macro is
2120 #define HARD_REGNO_NREGS(REGNO, MODE) \
2121 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) \
2126 @defmac HARD_REGNO_NREGS_HAS_PADDING (@var{regno}, @var{mode})
2127 A C expression that is nonzero if a value of mode @var{mode}, stored
2128 in memory, ends with padding that causes it to take up more space than
2129 in registers starting at register number @var{regno} (as determined by
2130 multiplying GCC's notion of the size of the register when containing
2131 this mode by the number of registers returned by
2132 @code{HARD_REGNO_NREGS}). By default this is zero.
2134 For example, if a floating-point value is stored in three 32-bit
2135 registers but takes up 128 bits in memory, then this would be
2138 This macros only needs to be defined if there are cases where
2139 @code{subreg_get_info}
2140 would otherwise wrongly determine that a @code{subreg} can be
2141 represented by an offset to the register number, when in fact such a
2142 @code{subreg} would contain some of the padding not stored in
2143 registers and so not be representable.
2146 @defmac HARD_REGNO_NREGS_WITH_PADDING (@var{regno}, @var{mode})
2147 For values of @var{regno} and @var{mode} for which
2148 @code{HARD_REGNO_NREGS_HAS_PADDING} returns nonzero, a C expression
2149 returning the greater number of registers required to hold the value
2150 including any padding. In the example above, the value would be four.
2153 @defmac REGMODE_NATURAL_SIZE (@var{mode})
2154 Define this macro if the natural size of registers that hold values
2155 of mode @var{mode} is not the word size. It is a C expression that
2156 should give the natural size in bytes for the specified mode. It is
2157 used by the register allocator to try to optimize its results. This
2158 happens for example on SPARC 64-bit where the natural size of
2159 floating-point registers is still 32-bit.
2162 @defmac HARD_REGNO_MODE_OK (@var{regno}, @var{mode})
2163 A C expression that is nonzero if it is permissible to store a value
2164 of mode @var{mode} in hard register number @var{regno} (or in several
2165 registers starting with that one). For a machine where all registers
2166 are equivalent, a suitable definition is
2169 #define HARD_REGNO_MODE_OK(REGNO, MODE) 1
2172 You need not include code to check for the numbers of fixed registers,
2173 because the allocation mechanism considers them to be always occupied.
2175 @cindex register pairs
2176 On some machines, double-precision values must be kept in even/odd
2177 register pairs. You can implement that by defining this macro to reject
2178 odd register numbers for such modes.
2180 The minimum requirement for a mode to be OK in a register is that the
2181 @samp{mov@var{mode}} instruction pattern support moves between the
2182 register and other hard register in the same class and that moving a
2183 value into the register and back out not alter it.
2185 Since the same instruction used to move @code{word_mode} will work for
2186 all narrower integer modes, it is not necessary on any machine for
2187 @code{HARD_REGNO_MODE_OK} to distinguish between these modes, provided
2188 you define patterns @samp{movhi}, etc., to take advantage of this. This
2189 is useful because of the interaction between @code{HARD_REGNO_MODE_OK}
2190 and @code{MODES_TIEABLE_P}; it is very desirable for all integer modes
2193 Many machines have special registers for floating point arithmetic.
2194 Often people assume that floating point machine modes are allowed only
2195 in floating point registers. This is not true. Any registers that
2196 can hold integers can safely @emph{hold} a floating point machine
2197 mode, whether or not floating arithmetic can be done on it in those
2198 registers. Integer move instructions can be used to move the values.
2200 On some machines, though, the converse is true: fixed-point machine
2201 modes may not go in floating registers. This is true if the floating
2202 registers normalize any value stored in them, because storing a
2203 non-floating value there would garble it. In this case,
2204 @code{HARD_REGNO_MODE_OK} should reject fixed-point machine modes in
2205 floating registers. But if the floating registers do not automatically
2206 normalize, if you can store any bit pattern in one and retrieve it
2207 unchanged without a trap, then any machine mode may go in a floating
2208 register, so you can define this macro to say so.
2210 The primary significance of special floating registers is rather that
2211 they are the registers acceptable in floating point arithmetic
2212 instructions. However, this is of no concern to
2213 @code{HARD_REGNO_MODE_OK}. You handle it by writing the proper
2214 constraints for those instructions.
2216 On some machines, the floating registers are especially slow to access,
2217 so that it is better to store a value in a stack frame than in such a
2218 register if floating point arithmetic is not being done. As long as the
2219 floating registers are not in class @code{GENERAL_REGS}, they will not
2220 be used unless some pattern's constraint asks for one.
2223 @defmac HARD_REGNO_RENAME_OK (@var{from}, @var{to})
2224 A C expression that is nonzero if it is OK to rename a hard register
2225 @var{from} to another hard register @var{to}.
2227 One common use of this macro is to prevent renaming of a register to
2228 another register that is not saved by a prologue in an interrupt
2231 The default is always nonzero.
2234 @defmac MODES_TIEABLE_P (@var{mode1}, @var{mode2})
2235 A C expression that is nonzero if a value of mode
2236 @var{mode1} is accessible in mode @var{mode2} without copying.
2238 If @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode1})} and
2239 @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode2})} are always the same for
2240 any @var{r}, then @code{MODES_TIEABLE_P (@var{mode1}, @var{mode2})}
2241 should be nonzero. If they differ for any @var{r}, you should define
2242 this macro to return zero unless some other mechanism ensures the
2243 accessibility of the value in a narrower mode.
2245 You should define this macro to return nonzero in as many cases as
2246 possible since doing so will allow GCC to perform better register
2250 @deftypefn {Target Hook} bool TARGET_HARD_REGNO_SCRATCH_OK (unsigned int @var{regno})
2251 This target hook should return @code{true} if it is OK to use a hard register
2252 @var{regno} as scratch reg in peephole2.
2254 One common use of this macro is to prevent using of a register that
2255 is not saved by a prologue in an interrupt handler.
2257 The default version of this hook always returns @code{true}.
2260 @defmac AVOID_CCMODE_COPIES
2261 Define this macro if the compiler should avoid copies to/from @code{CCmode}
2262 registers. You should only define this macro if support for copying to/from
2263 @code{CCmode} is incomplete.
2266 @node Leaf Functions
2267 @subsection Handling Leaf Functions
2269 @cindex leaf functions
2270 @cindex functions, leaf
2271 On some machines, a leaf function (i.e., one which makes no calls) can run
2272 more efficiently if it does not make its own register window. Often this
2273 means it is required to receive its arguments in the registers where they
2274 are passed by the caller, instead of the registers where they would
2277 The special treatment for leaf functions generally applies only when
2278 other conditions are met; for example, often they may use only those
2279 registers for its own variables and temporaries. We use the term ``leaf
2280 function'' to mean a function that is suitable for this special
2281 handling, so that functions with no calls are not necessarily ``leaf
2284 GCC assigns register numbers before it knows whether the function is
2285 suitable for leaf function treatment. So it needs to renumber the
2286 registers in order to output a leaf function. The following macros
2289 @defmac LEAF_REGISTERS
2290 Name of a char vector, indexed by hard register number, which
2291 contains 1 for a register that is allowable in a candidate for leaf
2294 If leaf function treatment involves renumbering the registers, then the
2295 registers marked here should be the ones before renumbering---those that
2296 GCC would ordinarily allocate. The registers which will actually be
2297 used in the assembler code, after renumbering, should not be marked with 1
2300 Define this macro only if the target machine offers a way to optimize
2301 the treatment of leaf functions.
2304 @defmac LEAF_REG_REMAP (@var{regno})
2305 A C expression whose value is the register number to which @var{regno}
2306 should be renumbered, when a function is treated as a leaf function.
2308 If @var{regno} is a register number which should not appear in a leaf
2309 function before renumbering, then the expression should yield @minus{}1, which
2310 will cause the compiler to abort.
2312 Define this macro only if the target machine offers a way to optimize the
2313 treatment of leaf functions, and registers need to be renumbered to do
2317 @findex current_function_is_leaf
2318 @findex current_function_uses_only_leaf_regs
2319 @code{TARGET_ASM_FUNCTION_PROLOGUE} and
2320 @code{TARGET_ASM_FUNCTION_EPILOGUE} must usually treat leaf functions
2321 specially. They can test the C variable @code{current_function_is_leaf}
2322 which is nonzero for leaf functions. @code{current_function_is_leaf} is
2323 set prior to local register allocation and is valid for the remaining
2324 compiler passes. They can also test the C variable
2325 @code{current_function_uses_only_leaf_regs} which is nonzero for leaf
2326 functions which only use leaf registers.
2327 @code{current_function_uses_only_leaf_regs} is valid after all passes
2328 that modify the instructions have been run and is only useful if
2329 @code{LEAF_REGISTERS} is defined.
2330 @c changed this to fix overfull. ALSO: why the "it" at the beginning
2331 @c of the next paragraph?! --mew 2feb93
2333 @node Stack Registers
2334 @subsection Registers That Form a Stack
2336 There are special features to handle computers where some of the
2337 ``registers'' form a stack. Stack registers are normally written by
2338 pushing onto the stack, and are numbered relative to the top of the
2341 Currently, GCC can only handle one group of stack-like registers, and
2342 they must be consecutively numbered. Furthermore, the existing
2343 support for stack-like registers is specific to the 80387 floating
2344 point coprocessor. If you have a new architecture that uses
2345 stack-like registers, you will need to do substantial work on
2346 @file{reg-stack.c} and write your machine description to cooperate
2347 with it, as well as defining these macros.
2350 Define this if the machine has any stack-like registers.
2353 @defmac STACK_REG_COVER_CLASS
2354 This is a cover class containing the stack registers. Define this if
2355 the machine has any stack-like registers.
2358 @defmac FIRST_STACK_REG
2359 The number of the first stack-like register. This one is the top
2363 @defmac LAST_STACK_REG
2364 The number of the last stack-like register. This one is the bottom of
2368 @node Register Classes
2369 @section Register Classes
2370 @cindex register class definitions
2371 @cindex class definitions, register
2373 On many machines, the numbered registers are not all equivalent.
2374 For example, certain registers may not be allowed for indexed addressing;
2375 certain registers may not be allowed in some instructions. These machine
2376 restrictions are described to the compiler using @dfn{register classes}.
2378 You define a number of register classes, giving each one a name and saying
2379 which of the registers belong to it. Then you can specify register classes
2380 that are allowed as operands to particular instruction patterns.
2384 In general, each register will belong to several classes. In fact, one
2385 class must be named @code{ALL_REGS} and contain all the registers. Another
2386 class must be named @code{NO_REGS} and contain no registers. Often the
2387 union of two classes will be another class; however, this is not required.
2389 @findex GENERAL_REGS
2390 One of the classes must be named @code{GENERAL_REGS}. There is nothing
2391 terribly special about the name, but the operand constraint letters
2392 @samp{r} and @samp{g} specify this class. If @code{GENERAL_REGS} is
2393 the same as @code{ALL_REGS}, just define it as a macro which expands
2396 Order the classes so that if class @var{x} is contained in class @var{y}
2397 then @var{x} has a lower class number than @var{y}.
2399 The way classes other than @code{GENERAL_REGS} are specified in operand
2400 constraints is through machine-dependent operand constraint letters.
2401 You can define such letters to correspond to various classes, then use
2402 them in operand constraints.
2404 You should define a class for the union of two classes whenever some
2405 instruction allows both classes. For example, if an instruction allows
2406 either a floating point (coprocessor) register or a general register for a
2407 certain operand, you should define a class @code{FLOAT_OR_GENERAL_REGS}
2408 which includes both of them. Otherwise you will get suboptimal code.
2410 You must also specify certain redundant information about the register
2411 classes: for each class, which classes contain it and which ones are
2412 contained in it; for each pair of classes, the largest class contained
2415 When a value occupying several consecutive registers is expected in a
2416 certain class, all the registers used must belong to that class.
2417 Therefore, register classes cannot be used to enforce a requirement for
2418 a register pair to start with an even-numbered register. The way to
2419 specify this requirement is with @code{HARD_REGNO_MODE_OK}.
2421 Register classes used for input-operands of bitwise-and or shift
2422 instructions have a special requirement: each such class must have, for
2423 each fixed-point machine mode, a subclass whose registers can transfer that
2424 mode to or from memory. For example, on some machines, the operations for
2425 single-byte values (@code{QImode}) are limited to certain registers. When
2426 this is so, each register class that is used in a bitwise-and or shift
2427 instruction must have a subclass consisting of registers from which
2428 single-byte values can be loaded or stored. This is so that
2429 @code{PREFERRED_RELOAD_CLASS} can always have a possible value to return.
2431 @deftp {Data type} {enum reg_class}
2432 An enumerated type that must be defined with all the register class names
2433 as enumerated values. @code{NO_REGS} must be first. @code{ALL_REGS}
2434 must be the last register class, followed by one more enumerated value,
2435 @code{LIM_REG_CLASSES}, which is not a register class but rather
2436 tells how many classes there are.
2438 Each register class has a number, which is the value of casting
2439 the class name to type @code{int}. The number serves as an index
2440 in many of the tables described below.
2443 @defmac N_REG_CLASSES
2444 The number of distinct register classes, defined as follows:
2447 #define N_REG_CLASSES (int) LIM_REG_CLASSES
2451 @defmac REG_CLASS_NAMES
2452 An initializer containing the names of the register classes as C string
2453 constants. These names are used in writing some of the debugging dumps.
2456 @defmac REG_CLASS_CONTENTS
2457 An initializer containing the contents of the register classes, as integers
2458 which are bit masks. The @var{n}th integer specifies the contents of class
2459 @var{n}. The way the integer @var{mask} is interpreted is that
2460 register @var{r} is in the class if @code{@var{mask} & (1 << @var{r})} is 1.
2462 When the machine has more than 32 registers, an integer does not suffice.
2463 Then the integers are replaced by sub-initializers, braced groupings containing
2464 several integers. Each sub-initializer must be suitable as an initializer
2465 for the type @code{HARD_REG_SET} which is defined in @file{hard-reg-set.h}.
2466 In this situation, the first integer in each sub-initializer corresponds to
2467 registers 0 through 31, the second integer to registers 32 through 63, and
2471 @defmac REGNO_REG_CLASS (@var{regno})
2472 A C expression whose value is a register class containing hard register
2473 @var{regno}. In general there is more than one such class; choose a class
2474 which is @dfn{minimal}, meaning that no smaller class also contains the
2478 @defmac BASE_REG_CLASS
2479 A macro whose definition is the name of the class to which a valid
2480 base register must belong. A base register is one used in an address
2481 which is the register value plus a displacement.
2484 @defmac MODE_BASE_REG_CLASS (@var{mode})
2485 This is a variation of the @code{BASE_REG_CLASS} macro which allows
2486 the selection of a base register in a mode dependent manner. If
2487 @var{mode} is VOIDmode then it should return the same value as
2488 @code{BASE_REG_CLASS}.
2491 @defmac MODE_BASE_REG_REG_CLASS (@var{mode})
2492 A C expression whose value is the register class to which a valid
2493 base register must belong in order to be used in a base plus index
2494 register address. You should define this macro if base plus index
2495 addresses have different requirements than other base register uses.
2498 @defmac MODE_CODE_BASE_REG_CLASS (@var{mode}, @var{outer_code}, @var{index_code})
2499 A C expression whose value is the register class to which a valid
2500 base register must belong. @var{outer_code} and @var{index_code} define the
2501 context in which the base register occurs. @var{outer_code} is the code of
2502 the immediately enclosing expression (@code{MEM} for the top level of an
2503 address, @code{ADDRESS} for something that occurs in an
2504 @code{address_operand}). @var{index_code} is the code of the corresponding
2505 index expression if @var{outer_code} is @code{PLUS}; @code{SCRATCH} otherwise.
2508 @defmac INDEX_REG_CLASS
2509 A macro whose definition is the name of the class to which a valid
2510 index register must belong. An index register is one used in an
2511 address where its value is either multiplied by a scale factor or
2512 added to another register (as well as added to a displacement).
2515 @defmac REGNO_OK_FOR_BASE_P (@var{num})
2516 A C expression which is nonzero if register number @var{num} is
2517 suitable for use as a base register in operand addresses.
2518 Like @code{TARGET_LEGITIMATE_ADDRESS_P}, this macro should also
2519 define a strict and a non-strict variant. Both variants behave
2520 the same for hard register; for pseudos, the strict variant will
2521 pass only those that have been allocated to a valid hard registers,
2522 while the non-strict variant will pass all pseudos.
2524 @findex REG_OK_STRICT
2525 Compiler source files that want to use the strict variant of this and
2526 other macros define the macro @code{REG_OK_STRICT}. You should use an
2527 @code{#ifdef REG_OK_STRICT} conditional to define the strict variant in
2528 that case and the non-strict variant otherwise.
2531 @defmac REGNO_MODE_OK_FOR_BASE_P (@var{num}, @var{mode})
2532 A C expression that is just like @code{REGNO_OK_FOR_BASE_P}, except that
2533 that expression may examine the mode of the memory reference in
2534 @var{mode}. You should define this macro if the mode of the memory
2535 reference affects whether a register may be used as a base register. If
2536 you define this macro, the compiler will use it instead of
2537 @code{REGNO_OK_FOR_BASE_P}. The mode may be @code{VOIDmode} for
2538 addresses that appear outside a @code{MEM}, i.e., as an
2539 @code{address_operand}.
2541 This macro also has strict and non-strict variants.
2544 @defmac REGNO_MODE_OK_FOR_REG_BASE_P (@var{num}, @var{mode})
2545 A C expression which is nonzero if register number @var{num} is suitable for
2546 use as a base register in base plus index operand addresses, accessing
2547 memory in mode @var{mode}. It may be either a suitable hard register or a
2548 pseudo register that has been allocated such a hard register. You should
2549 define this macro if base plus index addresses have different requirements
2550 than other base register uses.
2552 Use of this macro is deprecated; please use the more general
2553 @code{REGNO_MODE_CODE_OK_FOR_BASE_P}.
2555 This macro also has strict and non-strict variants.
2558 @defmac REGNO_MODE_CODE_OK_FOR_BASE_P (@var{num}, @var{mode}, @var{outer_code}, @var{index_code})
2559 A C expression that is just like @code{REGNO_MODE_OK_FOR_BASE_P}, except
2560 that that expression may examine the context in which the register
2561 appears in the memory reference. @var{outer_code} is the code of the
2562 immediately enclosing expression (@code{MEM} if at the top level of the
2563 address, @code{ADDRESS} for something that occurs in an
2564 @code{address_operand}). @var{index_code} is the code of the
2565 corresponding index expression if @var{outer_code} is @code{PLUS};
2566 @code{SCRATCH} otherwise. The mode may be @code{VOIDmode} for addresses
2567 that appear outside a @code{MEM}, i.e., as an @code{address_operand}.
2569 This macro also has strict and non-strict variants.
2572 @defmac REGNO_OK_FOR_INDEX_P (@var{num})
2573 A C expression which is nonzero if register number @var{num} is
2574 suitable for use as an index register in operand addresses. It may be
2575 either a suitable hard register or a pseudo register that has been
2576 allocated such a hard register.
2578 The difference between an index register and a base register is that
2579 the index register may be scaled. If an address involves the sum of
2580 two registers, neither one of them scaled, then either one may be
2581 labeled the ``base'' and the other the ``index''; but whichever
2582 labeling is used must fit the machine's constraints of which registers
2583 may serve in each capacity. The compiler will try both labelings,
2584 looking for one that is valid, and will reload one or both registers
2585 only if neither labeling works.
2587 This macro also has strict and non-strict variants.
2590 @defmac PREFERRED_RELOAD_CLASS (@var{x}, @var{class})
2591 A C expression that places additional restrictions on the register class
2592 to use when it is necessary to copy value @var{x} into a register in class
2593 @var{class}. The value is a register class; perhaps @var{class}, or perhaps
2594 another, smaller class. On many machines, the following definition is
2598 #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS
2601 Sometimes returning a more restrictive class makes better code. For
2602 example, on the 68000, when @var{x} is an integer constant that is in range
2603 for a @samp{moveq} instruction, the value of this macro is always
2604 @code{DATA_REGS} as long as @var{class} includes the data registers.
2605 Requiring a data register guarantees that a @samp{moveq} will be used.
2607 One case where @code{PREFERRED_RELOAD_CLASS} must not return
2608 @var{class} is if @var{x} is a legitimate constant which cannot be
2609 loaded into some register class. By returning @code{NO_REGS} you can
2610 force @var{x} into a memory location. For example, rs6000 can load
2611 immediate values into general-purpose registers, but does not have an
2612 instruction for loading an immediate value into a floating-point
2613 register, so @code{PREFERRED_RELOAD_CLASS} returns @code{NO_REGS} when
2614 @var{x} is a floating-point constant. If the constant can't be loaded
2615 into any kind of register, code generation will be better if
2616 @code{LEGITIMATE_CONSTANT_P} makes the constant illegitimate instead
2617 of using @code{PREFERRED_RELOAD_CLASS}.
2619 If an insn has pseudos in it after register allocation, reload will go
2620 through the alternatives and call repeatedly @code{PREFERRED_RELOAD_CLASS}
2621 to find the best one. Returning @code{NO_REGS}, in this case, makes
2622 reload add a @code{!} in front of the constraint: the x86 back-end uses
2623 this feature to discourage usage of 387 registers when math is done in
2624 the SSE registers (and vice versa).
2627 @defmac PREFERRED_OUTPUT_RELOAD_CLASS (@var{x}, @var{class})
2628 Like @code{PREFERRED_RELOAD_CLASS}, but for output reloads instead of
2629 input reloads. If you don't define this macro, the default is to use
2630 @var{class}, unchanged.
2632 You can also use @code{PREFERRED_OUTPUT_RELOAD_CLASS} to discourage
2633 reload from using some alternatives, like @code{PREFERRED_RELOAD_CLASS}.
2636 @defmac LIMIT_RELOAD_CLASS (@var{mode}, @var{class})
2637 A C expression that places additional restrictions on the register class
2638 to use when it is necessary to be able to hold a value of mode
2639 @var{mode} in a reload register for which class @var{class} would
2642 Unlike @code{PREFERRED_RELOAD_CLASS}, this macro should be used when
2643 there are certain modes that simply can't go in certain reload classes.
2645 The value is a register class; perhaps @var{class}, or perhaps another,
2648 Don't define this macro unless the target machine has limitations which
2649 require the macro to do something nontrivial.
2652 @deftypefn {Target Hook} {enum reg_class} TARGET_SECONDARY_RELOAD (bool @var{in_p}, rtx @var{x}, enum reg_class @var{reload_class}, enum machine_mode @var{reload_mode}, secondary_reload_info *@var{sri})
2653 Many machines have some registers that cannot be copied directly to or
2654 from memory or even from other types of registers. An example is the
2655 @samp{MQ} register, which on most machines, can only be copied to or
2656 from general registers, but not memory. Below, we shall be using the
2657 term 'intermediate register' when a move operation cannot be performed
2658 directly, but has to be done by copying the source into the intermediate
2659 register first, and then copying the intermediate register to the
2660 destination. An intermediate register always has the same mode as
2661 source and destination. Since it holds the actual value being copied,
2662 reload might apply optimizations to re-use an intermediate register
2663 and eliding the copy from the source when it can determine that the
2664 intermediate register still holds the required value.
2666 Another kind of secondary reload is required on some machines which
2667 allow copying all registers to and from memory, but require a scratch
2668 register for stores to some memory locations (e.g., those with symbolic
2669 address on the RT, and those with certain symbolic address on the SPARC
2670 when compiling PIC)@. Scratch registers need not have the same mode
2671 as the value being copied, and usually hold a different value than
2672 that being copied. Special patterns in the md file are needed to
2673 describe how the copy is performed with the help of the scratch register;
2674 these patterns also describe the number, register class(es) and mode(s)
2675 of the scratch register(s).
2677 In some cases, both an intermediate and a scratch register are required.
2679 For input reloads, this target hook is called with nonzero @var{in_p},
2680 and @var{x} is an rtx that needs to be copied to a register of class
2681 @var{reload_class} in @var{reload_mode}. For output reloads, this target
2682 hook is called with zero @var{in_p}, and a register of class @var{reload_class}
2683 needs to be copied to rtx @var{x} in @var{reload_mode}.
2685 If copying a register of @var{reload_class} from/to @var{x} requires
2686 an intermediate register, the hook @code{secondary_reload} should
2687 return the register class required for this intermediate register.
2688 If no intermediate register is required, it should return NO_REGS.
2689 If more than one intermediate register is required, describe the one
2690 that is closest in the copy chain to the reload register.
2692 If scratch registers are needed, you also have to describe how to
2693 perform the copy from/to the reload register to/from this
2694 closest intermediate register. Or if no intermediate register is
2695 required, but still a scratch register is needed, describe the
2696 copy from/to the reload register to/from the reload operand @var{x}.
2698 You do this by setting @code{sri->icode} to the instruction code of a pattern
2699 in the md file which performs the move. Operands 0 and 1 are the output
2700 and input of this copy, respectively. Operands from operand 2 onward are
2701 for scratch operands. These scratch operands must have a mode, and a
2702 single-register-class
2703 @c [later: or memory]
2706 When an intermediate register is used, the @code{secondary_reload}
2707 hook will be called again to determine how to copy the intermediate
2708 register to/from the reload operand @var{x}, so your hook must also
2709 have code to handle the register class of the intermediate operand.
2711 @c [For later: maybe we'll allow multi-alternative reload patterns -
2712 @c the port maintainer could name a mov<mode> pattern that has clobbers -
2713 @c and match the constraints of input and output to determine the required
2714 @c alternative. A restriction would be that constraints used to match
2715 @c against reloads registers would have to be written as register class
2716 @c constraints, or we need a new target macro / hook that tells us if an
2717 @c arbitrary constraint can match an unknown register of a given class.
2718 @c Such a macro / hook would also be useful in other places.]
2721 @var{x} might be a pseudo-register or a @code{subreg} of a
2722 pseudo-register, which could either be in a hard register or in memory.
2723 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2724 in memory and the hard register number if it is in a register.
2726 Scratch operands in memory (constraint @code{"=m"} / @code{"=&m"}) are
2727 currently not supported. For the time being, you will have to continue
2728 to use @code{SECONDARY_MEMORY_NEEDED} for that purpose.
2730 @code{copy_cost} also uses this target hook to find out how values are
2731 copied. If you want it to include some extra cost for the need to allocate
2732 (a) scratch register(s), set @code{sri->extra_cost} to the additional cost.
2733 Or if two dependent moves are supposed to have a lower cost than the sum
2734 of the individual moves due to expected fortuitous scheduling and/or special
2735 forwarding logic, you can set @code{sri->extra_cost} to a negative amount.
2738 @defmac SECONDARY_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2739 @defmacx SECONDARY_INPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2740 @defmacx SECONDARY_OUTPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2741 These macros are obsolete, new ports should use the target hook
2742 @code{TARGET_SECONDARY_RELOAD} instead.
2744 These are obsolete macros, replaced by the @code{TARGET_SECONDARY_RELOAD}
2745 target hook. Older ports still define these macros to indicate to the
2746 reload phase that it may
2747 need to allocate at least one register for a reload in addition to the
2748 register to contain the data. Specifically, if copying @var{x} to a
2749 register @var{class} in @var{mode} requires an intermediate register,
2750 you were supposed to define @code{SECONDARY_INPUT_RELOAD_CLASS} to return the
2751 largest register class all of whose registers can be used as
2752 intermediate registers or scratch registers.
2754 If copying a register @var{class} in @var{mode} to @var{x} requires an
2755 intermediate or scratch register, @code{SECONDARY_OUTPUT_RELOAD_CLASS}
2756 was supposed to be defined be defined to return the largest register
2757 class required. If the
2758 requirements for input and output reloads were the same, the macro
2759 @code{SECONDARY_RELOAD_CLASS} should have been used instead of defining both
2762 The values returned by these macros are often @code{GENERAL_REGS}.
2763 Return @code{NO_REGS} if no spare register is needed; i.e., if @var{x}
2764 can be directly copied to or from a register of @var{class} in
2765 @var{mode} without requiring a scratch register. Do not define this
2766 macro if it would always return @code{NO_REGS}.
2768 If a scratch register is required (either with or without an
2769 intermediate register), you were supposed to define patterns for
2770 @samp{reload_in@var{m}} or @samp{reload_out@var{m}}, as required
2771 (@pxref{Standard Names}. These patterns, which were normally
2772 implemented with a @code{define_expand}, should be similar to the
2773 @samp{mov@var{m}} patterns, except that operand 2 is the scratch
2776 These patterns need constraints for the reload register and scratch
2778 contain a single register class. If the original reload register (whose
2779 class is @var{class}) can meet the constraint given in the pattern, the
2780 value returned by these macros is used for the class of the scratch
2781 register. Otherwise, two additional reload registers are required.
2782 Their classes are obtained from the constraints in the insn pattern.
2784 @var{x} might be a pseudo-register or a @code{subreg} of a
2785 pseudo-register, which could either be in a hard register or in memory.
2786 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2787 in memory and the hard register number if it is in a register.
2789 These macros should not be used in the case where a particular class of
2790 registers can only be copied to memory and not to another class of
2791 registers. In that case, secondary reload registers are not needed and
2792 would not be helpful. Instead, a stack location must be used to perform
2793 the copy and the @code{mov@var{m}} pattern should use memory as an
2794 intermediate storage. This case often occurs between floating-point and
2798 @defmac SECONDARY_MEMORY_NEEDED (@var{class1}, @var{class2}, @var{m})
2799 Certain machines have the property that some registers cannot be copied
2800 to some other registers without using memory. Define this macro on
2801 those machines to be a C expression that is nonzero if objects of mode
2802 @var{m} in registers of @var{class1} can only be copied to registers of
2803 class @var{class2} by storing a register of @var{class1} into memory
2804 and loading that memory location into a register of @var{class2}.
2806 Do not define this macro if its value would always be zero.
2809 @defmac SECONDARY_MEMORY_NEEDED_RTX (@var{mode})
2810 Normally when @code{SECONDARY_MEMORY_NEEDED} is defined, the compiler
2811 allocates a stack slot for a memory location needed for register copies.
2812 If this macro is defined, the compiler instead uses the memory location
2813 defined by this macro.
2815 Do not define this macro if you do not define
2816 @code{SECONDARY_MEMORY_NEEDED}.
2819 @defmac SECONDARY_MEMORY_NEEDED_MODE (@var{mode})
2820 When the compiler needs a secondary memory location to copy between two
2821 registers of mode @var{mode}, it normally allocates sufficient memory to
2822 hold a quantity of @code{BITS_PER_WORD} bits and performs the store and
2823 load operations in a mode that many bits wide and whose class is the
2824 same as that of @var{mode}.
2826 This is right thing to do on most machines because it ensures that all
2827 bits of the register are copied and prevents accesses to the registers
2828 in a narrower mode, which some machines prohibit for floating-point
2831 However, this default behavior is not correct on some machines, such as
2832 the DEC Alpha, that store short integers in floating-point registers
2833 differently than in integer registers. On those machines, the default
2834 widening will not work correctly and you must define this macro to
2835 suppress that widening in some cases. See the file @file{alpha.h} for
2838 Do not define this macro if you do not define
2839 @code{SECONDARY_MEMORY_NEEDED} or if widening @var{mode} to a mode that
2840 is @code{BITS_PER_WORD} bits wide is correct for your machine.
2843 @defmac SMALL_REGISTER_CLASSES
2844 On some machines, it is risky to let hard registers live across arbitrary
2845 insns. Typically, these machines have instructions that require values
2846 to be in specific registers (like an accumulator), and reload will fail
2847 if the required hard register is used for another purpose across such an
2850 Define @code{SMALL_REGISTER_CLASSES} to be an expression with a nonzero
2851 value on these machines. When this macro has a nonzero value, the
2852 compiler will try to minimize the lifetime of hard registers.
2854 It is always safe to define this macro with a nonzero value, but if you
2855 unnecessarily define it, you will reduce the amount of optimizations
2856 that can be performed in some cases. If you do not define this macro
2857 with a nonzero value when it is required, the compiler will run out of
2858 spill registers and print a fatal error message. For most machines, you
2859 should not define this macro at all.
2862 @defmac CLASS_LIKELY_SPILLED_P (@var{class})
2863 A C expression whose value is nonzero if pseudos that have been assigned
2864 to registers of class @var{class} would likely be spilled because
2865 registers of @var{class} are needed for spill registers.
2867 The default value of this macro returns 1 if @var{class} has exactly one
2868 register and zero otherwise. On most machines, this default should be
2869 used. Only define this macro to some other expression if pseudos
2870 allocated by @file{local-alloc.c} end up in memory because their hard
2871 registers were needed for spill registers. If this macro returns nonzero
2872 for those classes, those pseudos will only be allocated by
2873 @file{global.c}, which knows how to reallocate the pseudo to another
2874 register. If there would not be another register available for
2875 reallocation, you should not change the definition of this macro since
2876 the only effect of such a definition would be to slow down register
2880 @defmac CLASS_MAX_NREGS (@var{class}, @var{mode})
2881 A C expression for the maximum number of consecutive registers
2882 of class @var{class} needed to hold a value of mode @var{mode}.
2884 This is closely related to the macro @code{HARD_REGNO_NREGS}. In fact,
2885 the value of the macro @code{CLASS_MAX_NREGS (@var{class}, @var{mode})}
2886 should be the maximum value of @code{HARD_REGNO_NREGS (@var{regno},
2887 @var{mode})} for all @var{regno} values in the class @var{class}.
2889 This macro helps control the handling of multiple-word values
2893 @defmac CANNOT_CHANGE_MODE_CLASS (@var{from}, @var{to}, @var{class})
2894 If defined, a C expression that returns nonzero for a @var{class} for which
2895 a change from mode @var{from} to mode @var{to} is invalid.
2897 For the example, loading 32-bit integer or floating-point objects into
2898 floating-point registers on the Alpha extends them to 64 bits.
2899 Therefore loading a 64-bit object and then storing it as a 32-bit object
2900 does not store the low-order 32 bits, as would be the case for a normal
2901 register. Therefore, @file{alpha.h} defines @code{CANNOT_CHANGE_MODE_CLASS}
2905 #define CANNOT_CHANGE_MODE_CLASS(FROM, TO, CLASS) \
2906 (GET_MODE_SIZE (FROM) != GET_MODE_SIZE (TO) \
2907 ? reg_classes_intersect_p (FLOAT_REGS, (CLASS)) : 0)
2911 @deftypefn {Target Hook} {const enum reg_class *} TARGET_IRA_COVER_CLASSES ()
2912 Return an array of cover classes for the Integrated Register Allocator
2913 (@acronym{IRA}). Cover classes are a set of non-intersecting register
2914 classes covering all hard registers used for register allocation
2915 purposes. If a move between two registers in the same cover class is
2916 possible, it should be cheaper than a load or store of the registers.
2917 The array is terminated by a @code{LIM_REG_CLASSES} element.
2919 The order of cover classes in the array is important. If two classes
2920 have the same cost of usage for a pseudo, the class occurred first in
2921 the array is chosen for the pseudo.
2923 This hook is called once at compiler startup, after the command-line
2924 options have been processed. It is then re-examined by every call to
2925 @code{target_reinit}.
2927 The default implementation returns @code{IRA_COVER_CLASSES}, if defined,
2928 otherwise there is no default implementation. You must define either this
2929 macro or @code{IRA_COVER_CLASSES} in order to use the integrated register
2930 allocator with Chaitin-Briggs coloring. If the macro is not defined,
2931 the only available coloring algorithm is Chow's priority coloring.
2934 @defmac IRA_COVER_CLASSES
2935 See the documentation for @code{TARGET_IRA_COVER_CLASSES}.
2938 @node Old Constraints
2939 @section Obsolete Macros for Defining Constraints
2940 @cindex defining constraints, obsolete method
2941 @cindex constraints, defining, obsolete method
2943 Machine-specific constraints can be defined with these macros instead
2944 of the machine description constructs described in @ref{Define
2945 Constraints}. This mechanism is obsolete. New ports should not use
2946 it; old ports should convert to the new mechanism.
2948 @defmac CONSTRAINT_LEN (@var{char}, @var{str})
2949 For the constraint at the start of @var{str}, which starts with the letter
2950 @var{c}, return the length. This allows you to have register class /
2951 constant / extra constraints that are longer than a single letter;
2952 you don't need to define this macro if you can do with single-letter
2953 constraints only. The definition of this macro should use
2954 DEFAULT_CONSTRAINT_LEN for all the characters that you don't want
2955 to handle specially.
2956 There are some sanity checks in genoutput.c that check the constraint lengths
2957 for the md file, so you can also use this macro to help you while you are
2958 transitioning from a byzantine single-letter-constraint scheme: when you
2959 return a negative length for a constraint you want to re-use, genoutput
2960 will complain about every instance where it is used in the md file.
2963 @defmac REG_CLASS_FROM_LETTER (@var{char})
2964 A C expression which defines the machine-dependent operand constraint
2965 letters for register classes. If @var{char} is such a letter, the
2966 value should be the register class corresponding to it. Otherwise,
2967 the value should be @code{NO_REGS}. The register letter @samp{r},
2968 corresponding to class @code{GENERAL_REGS}, will not be passed
2969 to this macro; you do not need to handle it.
2972 @defmac REG_CLASS_FROM_CONSTRAINT (@var{char}, @var{str})
2973 Like @code{REG_CLASS_FROM_LETTER}, but you also get the constraint string
2974 passed in @var{str}, so that you can use suffixes to distinguish between
2978 @defmac CONST_OK_FOR_LETTER_P (@var{value}, @var{c})
2979 A C expression that defines the machine-dependent operand constraint
2980 letters (@samp{I}, @samp{J}, @samp{K}, @dots{} @samp{P}) that specify
2981 particular ranges of integer values. If @var{c} is one of those
2982 letters, the expression should check that @var{value}, an integer, is in
2983 the appropriate range and return 1 if so, 0 otherwise. If @var{c} is
2984 not one of those letters, the value should be 0 regardless of
2988 @defmac CONST_OK_FOR_CONSTRAINT_P (@var{value}, @var{c}, @var{str})
2989 Like @code{CONST_OK_FOR_LETTER_P}, but you also get the constraint
2990 string passed in @var{str}, so that you can use suffixes to distinguish
2991 between different variants.
2994 @defmac CONST_DOUBLE_OK_FOR_LETTER_P (@var{value}, @var{c})
2995 A C expression that defines the machine-dependent operand constraint
2996 letters that specify particular ranges of @code{const_double} values
2997 (@samp{G} or @samp{H}).
2999 If @var{c} is one of those letters, the expression should check that
3000 @var{value}, an RTX of code @code{const_double}, is in the appropriate
3001 range and return 1 if so, 0 otherwise. If @var{c} is not one of those
3002 letters, the value should be 0 regardless of @var{value}.
3004 @code{const_double} is used for all floating-point constants and for
3005 @code{DImode} fixed-point constants. A given letter can accept either
3006 or both kinds of values. It can use @code{GET_MODE} to distinguish
3007 between these kinds.
3010 @defmac CONST_DOUBLE_OK_FOR_CONSTRAINT_P (@var{value}, @var{c}, @var{str})
3011 Like @code{CONST_DOUBLE_OK_FOR_LETTER_P}, but you also get the constraint
3012 string passed in @var{str}, so that you can use suffixes to distinguish
3013 between different variants.
3016 @defmac EXTRA_CONSTRAINT (@var{value}, @var{c})
3017 A C expression that defines the optional machine-dependent constraint
3018 letters that can be used to segregate specific types of operands, usually
3019 memory references, for the target machine. Any letter that is not
3020 elsewhere defined and not matched by @code{REG_CLASS_FROM_LETTER} /
3021 @code{REG_CLASS_FROM_CONSTRAINT}
3022 may be used. Normally this macro will not be defined.
3024 If it is required for a particular target machine, it should return 1
3025 if @var{value} corresponds to the operand type represented by the
3026 constraint letter @var{c}. If @var{c} is not defined as an extra
3027 constraint, the value returned should be 0 regardless of @var{value}.
3029 For example, on the ROMP, load instructions cannot have their output
3030 in r0 if the memory reference contains a symbolic address. Constraint
3031 letter @samp{Q} is defined as representing a memory address that does
3032 @emph{not} contain a symbolic address. An alternative is specified with
3033 a @samp{Q} constraint on the input and @samp{r} on the output. The next
3034 alternative specifies @samp{m} on the input and a register class that
3035 does not include r0 on the output.
3038 @defmac EXTRA_CONSTRAINT_STR (@var{value}, @var{c}, @var{str})
3039 Like @code{EXTRA_CONSTRAINT}, but you also get the constraint string passed
3040 in @var{str}, so that you can use suffixes to distinguish between different
3044 @defmac EXTRA_MEMORY_CONSTRAINT (@var{c}, @var{str})
3045 A C expression that defines the optional machine-dependent constraint
3046 letters, amongst those accepted by @code{EXTRA_CONSTRAINT}, that should
3047 be treated like memory constraints by the reload pass.
3049 It should return 1 if the operand type represented by the constraint
3050 at the start of @var{str}, the first letter of which is the letter @var{c},
3051 comprises a subset of all memory references including
3052 all those whose address is simply a base register. This allows the reload
3053 pass to reload an operand, if it does not directly correspond to the operand
3054 type of @var{c}, by copying its address into a base register.
3056 For example, on the S/390, some instructions do not accept arbitrary
3057 memory references, but only those that do not make use of an index
3058 register. The constraint letter @samp{Q} is defined via
3059 @code{EXTRA_CONSTRAINT} as representing a memory address of this type.
3060 If the letter @samp{Q} is marked as @code{EXTRA_MEMORY_CONSTRAINT},
3061 a @samp{Q} constraint can handle any memory operand, because the
3062 reload pass knows it can be reloaded by copying the memory address
3063 into a base register if required. This is analogous to the way
3064 an @samp{o} constraint can handle any memory operand.
3067 @defmac EXTRA_ADDRESS_CONSTRAINT (@var{c}, @var{str})
3068 A C expression that defines the optional machine-dependent constraint
3069 letters, amongst those accepted by @code{EXTRA_CONSTRAINT} /
3070 @code{EXTRA_CONSTRAINT_STR}, that should
3071 be treated like address constraints by the reload pass.
3073 It should return 1 if the operand type represented by the constraint
3074 at the start of @var{str}, which starts with the letter @var{c}, comprises
3075 a subset of all memory addresses including
3076 all those that consist of just a base register. This allows the reload
3077 pass to reload an operand, if it does not directly correspond to the operand
3078 type of @var{str}, by copying it into a base register.
3080 Any constraint marked as @code{EXTRA_ADDRESS_CONSTRAINT} can only
3081 be used with the @code{address_operand} predicate. It is treated
3082 analogously to the @samp{p} constraint.
3085 @node Stack and Calling
3086 @section Stack Layout and Calling Conventions
3087 @cindex calling conventions
3089 @c prevent bad page break with this line
3090 This describes the stack layout and calling conventions.
3094 * Exception Handling::
3099 * Register Arguments::
3101 * Aggregate Return::
3106 * Stack Smashing Protection::
3110 @subsection Basic Stack Layout
3111 @cindex stack frame layout
3112 @cindex frame layout
3114 @c prevent bad page break with this line
3115 Here is the basic stack layout.
3117 @defmac STACK_GROWS_DOWNWARD
3118 Define this macro if pushing a word onto the stack moves the stack
3119 pointer to a smaller address.
3121 When we say, ``define this macro if @dots{}'', it means that the
3122 compiler checks this macro only with @code{#ifdef} so the precise
3123 definition used does not matter.
3126 @defmac STACK_PUSH_CODE
3127 This macro defines the operation used when something is pushed
3128 on the stack. In RTL, a push operation will be
3129 @code{(set (mem (STACK_PUSH_CODE (reg sp))) @dots{})}
3131 The choices are @code{PRE_DEC}, @code{POST_DEC}, @code{PRE_INC},
3132 and @code{POST_INC}. Which of these is correct depends on
3133 the stack direction and on whether the stack pointer points
3134 to the last item on the stack or whether it points to the
3135 space for the next item on the stack.
3137 The default is @code{PRE_DEC} when @code{STACK_GROWS_DOWNWARD} is
3138 defined, which is almost always right, and @code{PRE_INC} otherwise,
3139 which is often wrong.
3142 @defmac FRAME_GROWS_DOWNWARD
3143 Define this macro to nonzero value if the addresses of local variable slots
3144 are at negative offsets from the frame pointer.
3147 @defmac ARGS_GROW_DOWNWARD
3148 Define this macro if successive arguments to a function occupy decreasing
3149 addresses on the stack.
3152 @defmac STARTING_FRAME_OFFSET
3153 Offset from the frame pointer to the first local variable slot to be allocated.
3155 If @code{FRAME_GROWS_DOWNWARD}, find the next slot's offset by
3156 subtracting the first slot's length from @code{STARTING_FRAME_OFFSET}.
3157 Otherwise, it is found by adding the length of the first slot to the
3158 value @code{STARTING_FRAME_OFFSET}.
3159 @c i'm not sure if the above is still correct.. had to change it to get
3160 @c rid of an overfull. --mew 2feb93
3163 @defmac STACK_ALIGNMENT_NEEDED
3164 Define to zero to disable final alignment of the stack during reload.
3165 The nonzero default for this macro is suitable for most ports.
3167 On ports where @code{STARTING_FRAME_OFFSET} is nonzero or where there
3168 is a register save block following the local block that doesn't require
3169 alignment to @code{STACK_BOUNDARY}, it may be beneficial to disable
3170 stack alignment and do it in the backend.
3173 @defmac STACK_POINTER_OFFSET
3174 Offset from the stack pointer register to the first location at which
3175 outgoing arguments are placed. If not specified, the default value of
3176 zero is used. This is the proper value for most machines.
3178 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
3179 the first location at which outgoing arguments are placed.
3182 @defmac FIRST_PARM_OFFSET (@var{fundecl})
3183 Offset from the argument pointer register to the first argument's
3184 address. On some machines it may depend on the data type of the
3187 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
3188 the first argument's address.
3191 @defmac STACK_DYNAMIC_OFFSET (@var{fundecl})
3192 Offset from the stack pointer register to an item dynamically allocated
3193 on the stack, e.g., by @code{alloca}.
3195 The default value for this macro is @code{STACK_POINTER_OFFSET} plus the
3196 length of the outgoing arguments. The default is correct for most
3197 machines. See @file{function.c} for details.
3200 @defmac INITIAL_FRAME_ADDRESS_RTX
3201 A C expression whose value is RTL representing the address of the initial
3202 stack frame. This address is passed to @code{RETURN_ADDR_RTX} and
3203 @code{DYNAMIC_CHAIN_ADDRESS}. If you don't define this macro, a reasonable
3204 default value will be used. Define this macro in order to make frame pointer
3205 elimination work in the presence of @code{__builtin_frame_address (count)} and
3206 @code{__builtin_return_address (count)} for @code{count} not equal to zero.
3209 @defmac DYNAMIC_CHAIN_ADDRESS (@var{frameaddr})
3210 A C expression whose value is RTL representing the address in a stack
3211 frame where the pointer to the caller's frame is stored. Assume that
3212 @var{frameaddr} is an RTL expression for the address of the stack frame
3215 If you don't define this macro, the default is to return the value
3216 of @var{frameaddr}---that is, the stack frame address is also the
3217 address of the stack word that points to the previous frame.
3220 @defmac SETUP_FRAME_ADDRESSES
3221 If defined, a C expression that produces the machine-specific code to
3222 setup the stack so that arbitrary frames can be accessed. For example,
3223 on the SPARC, we must flush all of the register windows to the stack
3224 before we can access arbitrary stack frames. You will seldom need to
3228 @deftypefn {Target Hook} bool TARGET_BUILTIN_SETJMP_FRAME_VALUE ()
3229 This target hook should return an rtx that is used to store
3230 the address of the current frame into the built in @code{setjmp} buffer.
3231 The default value, @code{virtual_stack_vars_rtx}, is correct for most
3232 machines. One reason you may need to define this target hook is if
3233 @code{hard_frame_pointer_rtx} is the appropriate value on your machine.
3236 @defmac FRAME_ADDR_RTX (@var{frameaddr})
3237 A C expression whose value is RTL representing the value of the frame
3238 address for the current frame. @var{frameaddr} is the frame pointer
3239 of the current frame. This is used for __builtin_frame_address.
3240 You need only define this macro if the frame address is not the same
3241 as the frame pointer. Most machines do not need to define it.
3244 @defmac RETURN_ADDR_RTX (@var{count}, @var{frameaddr})
3245 A C expression whose value is RTL representing the value of the return
3246 address for the frame @var{count} steps up from the current frame, after
3247 the prologue. @var{frameaddr} is the frame pointer of the @var{count}
3248 frame, or the frame pointer of the @var{count} @minus{} 1 frame if
3249 @code{RETURN_ADDR_IN_PREVIOUS_FRAME} is defined.
3251 The value of the expression must always be the correct address when
3252 @var{count} is zero, but may be @code{NULL_RTX} if there is no way to
3253 determine the return address of other frames.
3256 @defmac RETURN_ADDR_IN_PREVIOUS_FRAME
3257 Define this if the return address of a particular stack frame is accessed
3258 from the frame pointer of the previous stack frame.
3261 @defmac INCOMING_RETURN_ADDR_RTX
3262 A C expression whose value is RTL representing the location of the
3263 incoming return address at the beginning of any function, before the
3264 prologue. This RTL is either a @code{REG}, indicating that the return
3265 value is saved in @samp{REG}, or a @code{MEM} representing a location in
3268 You only need to define this macro if you want to support call frame
3269 debugging information like that provided by DWARF 2.
3271 If this RTL is a @code{REG}, you should also define
3272 @code{DWARF_FRAME_RETURN_COLUMN} to @code{DWARF_FRAME_REGNUM (REGNO)}.
3275 @defmac DWARF_ALT_FRAME_RETURN_COLUMN
3276 A C expression whose value is an integer giving a DWARF 2 column
3277 number that may be used as an alternative return column. The column
3278 must not correspond to any gcc hard register (that is, it must not
3279 be in the range of @code{DWARF_FRAME_REGNUM}).
3281 This macro can be useful if @code{DWARF_FRAME_RETURN_COLUMN} is set to a
3282 general register, but an alternative column needs to be used for signal
3283 frames. Some targets have also used different frame return columns
3287 @defmac DWARF_ZERO_REG
3288 A C expression whose value is an integer giving a DWARF 2 register
3289 number that is considered to always have the value zero. This should
3290 only be defined if the target has an architected zero register, and
3291 someone decided it was a good idea to use that register number to
3292 terminate the stack backtrace. New ports should avoid this.
3295 @deftypefn {Target Hook} void TARGET_DWARF_HANDLE_FRAME_UNSPEC (const char *@var{label}, rtx @var{pattern}, int @var{index})
3296 This target hook allows the backend to emit frame-related insns that
3297 contain UNSPECs or UNSPEC_VOLATILEs. The DWARF 2 call frame debugging
3298 info engine will invoke it on insns of the form
3300 (set (reg) (unspec [@dots{}] UNSPEC_INDEX))
3304 (set (reg) (unspec_volatile [@dots{}] UNSPECV_INDEX)).
3306 to let the backend emit the call frame instructions. @var{label} is
3307 the CFI label attached to the insn, @var{pattern} is the pattern of
3308 the insn and @var{index} is @code{UNSPEC_INDEX} or @code{UNSPECV_INDEX}.
3311 @defmac INCOMING_FRAME_SP_OFFSET
3312 A C expression whose value is an integer giving the offset, in bytes,
3313 from the value of the stack pointer register to the top of the stack
3314 frame at the beginning of any function, before the prologue. The top of
3315 the frame is defined to be the value of the stack pointer in the
3316 previous frame, just before the call instruction.
3318 You only need to define this macro if you want to support call frame
3319 debugging information like that provided by DWARF 2.
3322 @defmac ARG_POINTER_CFA_OFFSET (@var{fundecl})
3323 A C expression whose value is an integer giving the offset, in bytes,
3324 from the argument pointer to the canonical frame address (cfa). The
3325 final value should coincide with that calculated by
3326 @code{INCOMING_FRAME_SP_OFFSET}. Which is unfortunately not usable
3327 during virtual register instantiation.
3329 The default value for this macro is @code{FIRST_PARM_OFFSET (fundecl)},
3330 which is correct for most machines; in general, the arguments are found
3331 immediately before the stack frame. Note that this is not the case on
3332 some targets that save registers into the caller's frame, such as SPARC
3333 and rs6000, and so such targets need to define this macro.
3335 You only need to define this macro if the default is incorrect, and you
3336 want to support call frame debugging information like that provided by
3340 @defmac FRAME_POINTER_CFA_OFFSET (@var{fundecl})
3341 If defined, a C expression whose value is an integer giving the offset
3342 in bytes from the frame pointer to the canonical frame address (cfa).
3343 The final value should coincide with that calculated by
3344 @code{INCOMING_FRAME_SP_OFFSET}.
3346 Normally the CFA is calculated as an offset from the argument pointer,
3347 via @code{ARG_POINTER_CFA_OFFSET}, but if the argument pointer is
3348 variable due to the ABI, this may not be possible. If this macro is
3349 defined, it implies that the virtual register instantiation should be
3350 based on the frame pointer instead of the argument pointer. Only one
3351 of @code{FRAME_POINTER_CFA_OFFSET} and @code{ARG_POINTER_CFA_OFFSET}
3355 @defmac CFA_FRAME_BASE_OFFSET (@var{fundecl})
3356 If defined, a C expression whose value is an integer giving the offset
3357 in bytes from the canonical frame address (cfa) to the frame base used
3358 in DWARF 2 debug information. The default is zero. A different value
3359 may reduce the size of debug information on some ports.
3362 @node Exception Handling
3363 @subsection Exception Handling Support
3364 @cindex exception handling
3366 @defmac EH_RETURN_DATA_REGNO (@var{N})
3367 A C expression whose value is the @var{N}th register number used for
3368 data by exception handlers, or @code{INVALID_REGNUM} if fewer than
3369 @var{N} registers are usable.
3371 The exception handling library routines communicate with the exception
3372 handlers via a set of agreed upon registers. Ideally these registers
3373 should be call-clobbered; it is possible to use call-saved registers,
3374 but may negatively impact code size. The target must support at least
3375 2 data registers, but should define 4 if there are enough free registers.
3377 You must define this macro if you want to support call frame exception
3378 handling like that provided by DWARF 2.
3381 @defmac EH_RETURN_STACKADJ_RTX
3382 A C expression whose value is RTL representing a location in which
3383 to store a stack adjustment to be applied before function return.
3384 This is used to unwind the stack to an exception handler's call frame.
3385 It will be assigned zero on code paths that return normally.
3387 Typically this is a call-clobbered hard register that is otherwise
3388 untouched by the epilogue, but could also be a stack slot.
3390 Do not define this macro if the stack pointer is saved and restored
3391 by the regular prolog and epilog code in the call frame itself; in
3392 this case, the exception handling library routines will update the
3393 stack location to be restored in place. Otherwise, you must define
3394 this macro if you want to support call frame exception handling like
3395 that provided by DWARF 2.
3398 @defmac EH_RETURN_HANDLER_RTX
3399 A C expression whose value is RTL representing a location in which
3400 to store the address of an exception handler to which we should
3401 return. It will not be assigned on code paths that return normally.
3403 Typically this is the location in the call frame at which the normal
3404 return address is stored. For targets that return by popping an
3405 address off the stack, this might be a memory address just below
3406 the @emph{target} call frame rather than inside the current call
3407 frame. If defined, @code{EH_RETURN_STACKADJ_RTX} will have already
3408 been assigned, so it may be used to calculate the location of the
3411 Some targets have more complex requirements than storing to an
3412 address calculable during initial code generation. In that case
3413 the @code{eh_return} instruction pattern should be used instead.
3415 If you want to support call frame exception handling, you must
3416 define either this macro or the @code{eh_return} instruction pattern.
3419 @defmac RETURN_ADDR_OFFSET
3420 If defined, an integer-valued C expression for which rtl will be generated
3421 to add it to the exception handler address before it is searched in the
3422 exception handling tables, and to subtract it again from the address before
3423 using it to return to the exception handler.
3426 @defmac ASM_PREFERRED_EH_DATA_FORMAT (@var{code}, @var{global})
3427 This macro chooses the encoding of pointers embedded in the exception
3428 handling sections. If at all possible, this should be defined such
3429 that the exception handling section will not require dynamic relocations,
3430 and so may be read-only.
3432 @var{code} is 0 for data, 1 for code labels, 2 for function pointers.
3433 @var{global} is true if the symbol may be affected by dynamic relocations.
3434 The macro should return a combination of the @code{DW_EH_PE_*} defines
3435 as found in @file{dwarf2.h}.
3437 If this macro is not defined, pointers will not be encoded but
3438 represented directly.
3441 @defmac ASM_MAYBE_OUTPUT_ENCODED_ADDR_RTX (@var{file}, @var{encoding}, @var{size}, @var{addr}, @var{done})
3442 This macro allows the target to emit whatever special magic is required
3443 to represent the encoding chosen by @code{ASM_PREFERRED_EH_DATA_FORMAT}.
3444 Generic code takes care of pc-relative and indirect encodings; this must
3445 be defined if the target uses text-relative or data-relative encodings.
3447 This is a C statement that branches to @var{done} if the format was
3448 handled. @var{encoding} is the format chosen, @var{size} is the number
3449 of bytes that the format occupies, @var{addr} is the @code{SYMBOL_REF}
3453 @defmac MD_UNWIND_SUPPORT
3454 A string specifying a file to be #include'd in unwind-dw2.c. The file
3455 so included typically defines @code{MD_FALLBACK_FRAME_STATE_FOR}.
3458 @defmac MD_FALLBACK_FRAME_STATE_FOR (@var{context}, @var{fs})
3459 This macro allows the target to add CPU and operating system specific
3460 code to the call-frame unwinder for use when there is no unwind data
3461 available. The most common reason to implement this macro is to unwind
3462 through signal frames.
3464 This macro is called from @code{uw_frame_state_for} in
3465 @file{unwind-dw2.c}, @file{unwind-dw2-xtensa.c} and
3466 @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context};
3467 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{context->ra}
3468 for the address of the code being executed and @code{context->cfa} for
3469 the stack pointer value. If the frame can be decoded, the register
3470 save addresses should be updated in @var{fs} and the macro should
3471 evaluate to @code{_URC_NO_REASON}. If the frame cannot be decoded,
3472 the macro should evaluate to @code{_URC_END_OF_STACK}.
3474 For proper signal handling in Java this macro is accompanied by
3475 @code{MAKE_THROW_FRAME}, defined in @file{libjava/include/*-signal.h} headers.
3478 @defmac MD_HANDLE_UNWABI (@var{context}, @var{fs})
3479 This macro allows the target to add operating system specific code to the
3480 call-frame unwinder to handle the IA-64 @code{.unwabi} unwinding directive,
3481 usually used for signal or interrupt frames.
3483 This macro is called from @code{uw_update_context} in @file{unwind-ia64.c}.
3484 @var{context} is an @code{_Unwind_Context};
3485 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{fs->unwabi}
3486 for the abi and context in the @code{.unwabi} directive. If the
3487 @code{.unwabi} directive can be handled, the register save addresses should
3488 be updated in @var{fs}.
3491 @defmac TARGET_USES_WEAK_UNWIND_INFO
3492 A C expression that evaluates to true if the target requires unwind
3493 info to be given comdat linkage. Define it to be @code{1} if comdat
3494 linkage is necessary. The default is @code{0}.
3497 @node Stack Checking
3498 @subsection Specifying How Stack Checking is Done
3500 GCC will check that stack references are within the boundaries of the
3501 stack, if the option @option{-fstack-check} is specified, in one of
3506 If the value of the @code{STACK_CHECK_BUILTIN} macro is nonzero, GCC
3507 will assume that you have arranged for full stack checking to be done
3508 at appropriate places in the configuration files. GCC will not do
3509 other special processing.
3512 If @code{STACK_CHECK_BUILTIN} is zero and the value of the
3513 @code{STACK_CHECK_STATIC_BUILTIN} macro is nonzero, GCC will assume
3514 that you have arranged for static stack checking (checking of the
3515 static stack frame of functions) to be done at appropriate places
3516 in the configuration files. GCC will only emit code to do dynamic
3517 stack checking (checking on dynamic stack allocations) using the third
3521 If neither of the above are true, GCC will generate code to periodically
3522 ``probe'' the stack pointer using the values of the macros defined below.
3525 If neither STACK_CHECK_BUILTIN nor STACK_CHECK_STATIC_BUILTIN is defined,
3526 GCC will change its allocation strategy for large objects if the option
3527 @option{-fstack-check} is specified: they will always be allocated
3528 dynamically if their size exceeds @code{STACK_CHECK_MAX_VAR_SIZE} bytes.
3530 @defmac STACK_CHECK_BUILTIN
3531 A nonzero value if stack checking is done by the configuration files in a
3532 machine-dependent manner. You should define this macro if stack checking
3533 is required by the ABI of your machine or if you would like to do stack
3534 checking in some more efficient way than the generic approach. The default
3535 value of this macro is zero.
3538 @defmac STACK_CHECK_STATIC_BUILTIN
3539 A nonzero value if static stack checking is done by the configuration files
3540 in a machine-dependent manner. You should define this macro if you would
3541 like to do static stack checking in some more efficient way than the generic
3542 approach. The default value of this macro is zero.
3545 @defmac STACK_CHECK_PROBE_INTERVAL
3546 An integer representing the interval at which GCC must generate stack
3547 probe instructions. You will normally define this macro to be no larger
3548 than the size of the ``guard pages'' at the end of a stack area. The
3549 default value of 4096 is suitable for most systems.
3552 @defmac STACK_CHECK_PROBE_LOAD
3553 An integer which is nonzero if GCC should perform the stack probe
3554 as a load instruction and zero if GCC should use a store instruction.
3555 The default is zero, which is the most efficient choice on most systems.
3558 @defmac STACK_CHECK_PROTECT
3559 The number of bytes of stack needed to recover from a stack overflow,
3560 for languages where such a recovery is supported. The default value of
3561 75 words should be adequate for most machines.
3564 The following macros are relevant only if neither STACK_CHECK_BUILTIN
3565 nor STACK_CHECK_STATIC_BUILTIN is defined; you can omit them altogether
3566 in the opposite case.
3568 @defmac STACK_CHECK_MAX_FRAME_SIZE
3569 The maximum size of a stack frame, in bytes. GCC will generate probe
3570 instructions in non-leaf functions to ensure at least this many bytes of
3571 stack are available. If a stack frame is larger than this size, stack
3572 checking will not be reliable and GCC will issue a warning. The
3573 default is chosen so that GCC only generates one instruction on most
3574 systems. You should normally not change the default value of this macro.
3577 @defmac STACK_CHECK_FIXED_FRAME_SIZE
3578 GCC uses this value to generate the above warning message. It
3579 represents the amount of fixed frame used by a function, not including
3580 space for any callee-saved registers, temporaries and user variables.
3581 You need only specify an upper bound for this amount and will normally
3582 use the default of four words.
3585 @defmac STACK_CHECK_MAX_VAR_SIZE
3586 The maximum size, in bytes, of an object that GCC will place in the
3587 fixed area of the stack frame when the user specifies
3588 @option{-fstack-check}.
3589 GCC computed the default from the values of the above macros and you will
3590 normally not need to override that default.
3594 @node Frame Registers
3595 @subsection Registers That Address the Stack Frame
3597 @c prevent bad page break with this line
3598 This discusses registers that address the stack frame.
3600 @defmac STACK_POINTER_REGNUM
3601 The register number of the stack pointer register, which must also be a
3602 fixed register according to @code{FIXED_REGISTERS}. On most machines,
3603 the hardware determines which register this is.
3606 @defmac FRAME_POINTER_REGNUM
3607 The register number of the frame pointer register, which is used to
3608 access automatic variables in the stack frame. On some machines, the
3609 hardware determines which register this is. On other machines, you can
3610 choose any register you wish for this purpose.
3613 @defmac HARD_FRAME_POINTER_REGNUM
3614 On some machines the offset between the frame pointer and starting
3615 offset of the automatic variables is not known until after register
3616 allocation has been done (for example, because the saved registers are
3617 between these two locations). On those machines, define
3618 @code{FRAME_POINTER_REGNUM} the number of a special, fixed register to
3619 be used internally until the offset is known, and define
3620 @code{HARD_FRAME_POINTER_REGNUM} to be the actual hard register number
3621 used for the frame pointer.
3623 You should define this macro only in the very rare circumstances when it
3624 is not possible to calculate the offset between the frame pointer and
3625 the automatic variables until after register allocation has been
3626 completed. When this macro is defined, you must also indicate in your
3627 definition of @code{ELIMINABLE_REGS} how to eliminate
3628 @code{FRAME_POINTER_REGNUM} into either @code{HARD_FRAME_POINTER_REGNUM}
3629 or @code{STACK_POINTER_REGNUM}.
3631 Do not define this macro if it would be the same as
3632 @code{FRAME_POINTER_REGNUM}.
3635 @defmac ARG_POINTER_REGNUM
3636 The register number of the arg pointer register, which is used to access
3637 the function's argument list. On some machines, this is the same as the
3638 frame pointer register. On some machines, the hardware determines which
3639 register this is. On other machines, you can choose any register you
3640 wish for this purpose. If this is not the same register as the frame
3641 pointer register, then you must mark it as a fixed register according to
3642 @code{FIXED_REGISTERS}, or arrange to be able to eliminate it
3643 (@pxref{Elimination}).
3646 @defmac RETURN_ADDRESS_POINTER_REGNUM
3647 The register number of the return address pointer register, which is used to
3648 access the current function's return address from the stack. On some
3649 machines, the return address is not at a fixed offset from the frame
3650 pointer or stack pointer or argument pointer. This register can be defined
3651 to point to the return address on the stack, and then be converted by
3652 @code{ELIMINABLE_REGS} into either the frame pointer or stack pointer.
3654 Do not define this macro unless there is no other way to get the return
3655 address from the stack.
3658 @defmac STATIC_CHAIN_REGNUM
3659 @defmacx STATIC_CHAIN_INCOMING_REGNUM
3660 Register numbers used for passing a function's static chain pointer. If
3661 register windows are used, the register number as seen by the called
3662 function is @code{STATIC_CHAIN_INCOMING_REGNUM}, while the register
3663 number as seen by the calling function is @code{STATIC_CHAIN_REGNUM}. If
3664 these registers are the same, @code{STATIC_CHAIN_INCOMING_REGNUM} need
3667 The static chain register need not be a fixed register.
3669 If the static chain is passed in memory, these macros should not be
3670 defined; instead, the @code{TARGET_STATIC_CHAIN} hook should be used.
3673 @deftypefn {Target Hook} rtx TARGET_STATIC_CHAIN (const_tree @var{fndecl}, bool @var{incoming_p})
3674 This hook replaces the use of @code{STATIC_CHAIN_REGNUM} et al for
3675 targets that may use different static chain locations for different
3676 nested functions. This may be required if the target has function
3677 attributes that affect the calling conventions of the function and
3678 those calling conventions use different static chain locations.
3680 The default version of this hook uses @code{STATIC_CHAIN_REGNUM} et al.
3682 If the static chain is passed in memory, this hook should be used to
3683 provide rtx giving @code{mem} expressions that denote where they are stored.
3684 Often the @code{mem} expression as seen by the caller will be at an offset
3685 from the stack pointer and the @code{mem} expression as seen by the callee
3686 will be at an offset from the frame pointer.
3687 @findex stack_pointer_rtx
3688 @findex frame_pointer_rtx
3689 @findex arg_pointer_rtx
3690 The variables @code{stack_pointer_rtx}, @code{frame_pointer_rtx}, and
3691 @code{arg_pointer_rtx} will have been initialized and should be used
3692 to refer to those items.
3695 @defmac DWARF_FRAME_REGISTERS
3696 This macro specifies the maximum number of hard registers that can be
3697 saved in a call frame. This is used to size data structures used in
3698 DWARF2 exception handling.
3700 Prior to GCC 3.0, this macro was needed in order to establish a stable
3701 exception handling ABI in the face of adding new hard registers for ISA
3702 extensions. In GCC 3.0 and later, the EH ABI is insulated from changes
3703 in the number of hard registers. Nevertheless, this macro can still be
3704 used to reduce the runtime memory requirements of the exception handling
3705 routines, which can be substantial if the ISA contains a lot of
3706 registers that are not call-saved.
3708 If this macro is not defined, it defaults to
3709 @code{FIRST_PSEUDO_REGISTER}.
3712 @defmac PRE_GCC3_DWARF_FRAME_REGISTERS
3714 This macro is similar to @code{DWARF_FRAME_REGISTERS}, but is provided
3715 for backward compatibility in pre GCC 3.0 compiled code.
3717 If this macro is not defined, it defaults to
3718 @code{DWARF_FRAME_REGISTERS}.
3721 @defmac DWARF_REG_TO_UNWIND_COLUMN (@var{regno})
3723 Define this macro if the target's representation for dwarf registers
3724 is different than the internal representation for unwind column.
3725 Given a dwarf register, this macro should return the internal unwind
3726 column number to use instead.
3728 See the PowerPC's SPE target for an example.
3731 @defmac DWARF_FRAME_REGNUM (@var{regno})
3733 Define this macro if the target's representation for dwarf registers
3734 used in .eh_frame or .debug_frame is different from that used in other
3735 debug info sections. Given a GCC hard register number, this macro
3736 should return the .eh_frame register number. The default is
3737 @code{DBX_REGISTER_NUMBER (@var{regno})}.
3741 @defmac DWARF2_FRAME_REG_OUT (@var{regno}, @var{for_eh})
3743 Define this macro to map register numbers held in the call frame info
3744 that GCC has collected using @code{DWARF_FRAME_REGNUM} to those that
3745 should be output in .debug_frame (@code{@var{for_eh}} is zero) and
3746 .eh_frame (@code{@var{for_eh}} is nonzero). The default is to
3747 return @code{@var{regno}}.
3752 @subsection Eliminating Frame Pointer and Arg Pointer
3754 @c prevent bad page break with this line
3755 This is about eliminating the frame pointer and arg pointer.
3757 @deftypefn {Target Hook} bool TARGET_FRAME_POINTER_REQUIRED (void)
3758 This target hook should return @code{true} if a function must have and use
3759 a frame pointer. This target hook is called in the reload pass. If its return
3760 value is @code{true} the function will have a frame pointer.
3762 This target hook can in principle examine the current function and decide
3763 according to the facts, but on most machines the constant @code{false} or the
3764 constant @code{true} suffices. Use @code{false} when the machine allows code
3765 to be generated with no frame pointer, and doing so saves some time or space.
3766 Use @code{true} when there is no possible advantage to avoiding a frame
3769 In certain cases, the compiler does not know how to produce valid code
3770 without a frame pointer. The compiler recognizes those cases and
3771 automatically gives the function a frame pointer regardless of what
3772 @code{TARGET_FRAME_POINTER_REQUIRED} returns. You don't need to worry about
3775 In a function that does not require a frame pointer, the frame pointer
3776 register can be allocated for ordinary usage, unless you mark it as a
3777 fixed register. See @code{FIXED_REGISTERS} for more information.
3779 Default return value is @code{false}.
3782 @findex get_frame_size
3783 @defmac INITIAL_FRAME_POINTER_OFFSET (@var{depth-var})
3784 A C statement to store in the variable @var{depth-var} the difference
3785 between the frame pointer and the stack pointer values immediately after
3786 the function prologue. The value would be computed from information
3787 such as the result of @code{get_frame_size ()} and the tables of
3788 registers @code{regs_ever_live} and @code{call_used_regs}.
3790 If @code{ELIMINABLE_REGS} is defined, this macro will be not be used and
3791 need not be defined. Otherwise, it must be defined even if
3792 @code{TARGET_FRAME_POINTER_REQUIRED} always returns true; in that
3793 case, you may set @var{depth-var} to anything.
3796 @defmac ELIMINABLE_REGS
3797 If defined, this macro specifies a table of register pairs used to
3798 eliminate unneeded registers that point into the stack frame. If it is not
3799 defined, the only elimination attempted by the compiler is to replace
3800 references to the frame pointer with references to the stack pointer.
3802 The definition of this macro is a list of structure initializations, each
3803 of which specifies an original and replacement register.
3805 On some machines, the position of the argument pointer is not known until
3806 the compilation is completed. In such a case, a separate hard register
3807 must be used for the argument pointer. This register can be eliminated by
3808 replacing it with either the frame pointer or the argument pointer,
3809 depending on whether or not the frame pointer has been eliminated.
3811 In this case, you might specify:
3813 #define ELIMINABLE_REGS \
3814 @{@{ARG_POINTER_REGNUM, STACK_POINTER_REGNUM@}, \
3815 @{ARG_POINTER_REGNUM, FRAME_POINTER_REGNUM@}, \
3816 @{FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM@}@}
3819 Note that the elimination of the argument pointer with the stack pointer is
3820 specified first since that is the preferred elimination.
3823 @deftypefn {Target Hook} bool TARGET_CAN_ELIMINATE (const int @var{from-reg}, const int @var{to-reg})
3824 This target hook should returns @code{true} if the compiler is allowed to
3825 try to replace register number @var{from-reg} with register number
3826 @var{to-reg}. This target hook need only be defined if @code{ELIMINABLE_REGS}
3827 is defined, and will usually be @code{true}, since most of the cases
3828 preventing register elimination are things that the compiler already
3831 Default return value is @code{true}.
3834 @defmac INITIAL_ELIMINATION_OFFSET (@var{from-reg}, @var{to-reg}, @var{offset-var})
3835 This macro is similar to @code{INITIAL_FRAME_POINTER_OFFSET}. It
3836 specifies the initial difference between the specified pair of
3837 registers. This macro must be defined if @code{ELIMINABLE_REGS} is
3841 @node Stack Arguments
3842 @subsection Passing Function Arguments on the Stack
3843 @cindex arguments on stack
3844 @cindex stack arguments
3846 The macros in this section control how arguments are passed
3847 on the stack. See the following section for other macros that
3848 control passing certain arguments in registers.
3850 @deftypefn {Target Hook} bool TARGET_PROMOTE_PROTOTYPES (tree @var{fntype})
3851 This target hook returns @code{true} if an argument declared in a
3852 prototype as an integral type smaller than @code{int} should actually be
3853 passed as an @code{int}. In addition to avoiding errors in certain
3854 cases of mismatch, it also makes for better code on certain machines.
3855 The default is to not promote prototypes.
3859 A C expression. If nonzero, push insns will be used to pass
3861 If the target machine does not have a push instruction, set it to zero.
3862 That directs GCC to use an alternate strategy: to
3863 allocate the entire argument block and then store the arguments into
3864 it. When @code{PUSH_ARGS} is nonzero, @code{PUSH_ROUNDING} must be defined too.
3867 @defmac PUSH_ARGS_REVERSED
3868 A C expression. If nonzero, function arguments will be evaluated from
3869 last to first, rather than from first to last. If this macro is not
3870 defined, it defaults to @code{PUSH_ARGS} on targets where the stack
3871 and args grow in opposite directions, and 0 otherwise.
3874 @defmac PUSH_ROUNDING (@var{npushed})
3875 A C expression that is the number of bytes actually pushed onto the
3876 stack when an instruction attempts to push @var{npushed} bytes.
3878 On some machines, the definition
3881 #define PUSH_ROUNDING(BYTES) (BYTES)
3885 will suffice. But on other machines, instructions that appear
3886 to push one byte actually push two bytes in an attempt to maintain
3887 alignment. Then the definition should be
3890 #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1)
3894 @findex current_function_outgoing_args_size
3895 @defmac ACCUMULATE_OUTGOING_ARGS
3896 A C expression. If nonzero, the maximum amount of space required for outgoing arguments
3897 will be computed and placed into the variable
3898 @code{current_function_outgoing_args_size}. No space will be pushed
3899 onto the stack for each call; instead, the function prologue should
3900 increase the stack frame size by this amount.
3902 Setting both @code{PUSH_ARGS} and @code{ACCUMULATE_OUTGOING_ARGS}
3906 @defmac REG_PARM_STACK_SPACE (@var{fndecl})
3907 Define this macro if functions should assume that stack space has been
3908 allocated for arguments even when their values are passed in
3911 The value of this macro is the size, in bytes, of the area reserved for
3912 arguments passed in registers for the function represented by @var{fndecl},
3913 which can be zero if GCC is calling a library function.
3914 The argument @var{fndecl} can be the FUNCTION_DECL, or the type itself
3917 This space can be allocated by the caller, or be a part of the
3918 machine-dependent stack frame: @code{OUTGOING_REG_PARM_STACK_SPACE} says
3921 @c above is overfull. not sure what to do. --mew 5feb93 did
3922 @c something, not sure if it looks good. --mew 10feb93
3924 @defmac OUTGOING_REG_PARM_STACK_SPACE (@var{fntype})
3925 Define this to a nonzero value if it is the responsibility of the
3926 caller to allocate the area reserved for arguments passed in registers
3927 when calling a function of @var{fntype}. @var{fntype} may be NULL
3928 if the function called is a library function.
3930 If @code{ACCUMULATE_OUTGOING_ARGS} is defined, this macro controls
3931 whether the space for these arguments counts in the value of
3932 @code{current_function_outgoing_args_size}.
3935 @defmac STACK_PARMS_IN_REG_PARM_AREA
3936 Define this macro if @code{REG_PARM_STACK_SPACE} is defined, but the
3937 stack parameters don't skip the area specified by it.
3938 @c i changed this, makes more sens and it should have taken care of the
3939 @c overfull.. not as specific, tho. --mew 5feb93
3941 Normally, when a parameter is not passed in registers, it is placed on the
3942 stack beyond the @code{REG_PARM_STACK_SPACE} area. Defining this macro
3943 suppresses this behavior and causes the parameter to be passed on the
3944 stack in its natural location.
3947 @defmac RETURN_POPS_ARGS (@var{fundecl}, @var{funtype}, @var{stack-size})
3948 A C expression that should indicate the number of bytes of its own
3949 arguments that a function pops on returning, or 0 if the
3950 function pops no arguments and the caller must therefore pop them all
3951 after the function returns.
3953 @var{fundecl} is a C variable whose value is a tree node that describes
3954 the function in question. Normally it is a node of type
3955 @code{FUNCTION_DECL} that describes the declaration of the function.
3956 From this you can obtain the @code{DECL_ATTRIBUTES} of the function.
3958 @var{funtype} is a C variable whose value is a tree node that
3959 describes the function in question. Normally it is a node of type
3960 @code{FUNCTION_TYPE} that describes the data type of the function.
3961 From this it is possible to obtain the data types of the value and
3962 arguments (if known).
3964 When a call to a library function is being considered, @var{fundecl}
3965 will contain an identifier node for the library function. Thus, if
3966 you need to distinguish among various library functions, you can do so
3967 by their names. Note that ``library function'' in this context means
3968 a function used to perform arithmetic, whose name is known specially
3969 in the compiler and was not mentioned in the C code being compiled.
3971 @var{stack-size} is the number of bytes of arguments passed on the
3972 stack. If a variable number of bytes is passed, it is zero, and
3973 argument popping will always be the responsibility of the calling function.
3975 On the VAX, all functions always pop their arguments, so the definition
3976 of this macro is @var{stack-size}. On the 68000, using the standard
3977 calling convention, no functions pop their arguments, so the value of
3978 the macro is always 0 in this case. But an alternative calling
3979 convention is available in which functions that take a fixed number of
3980 arguments pop them but other functions (such as @code{printf}) pop
3981 nothing (the caller pops all). When this convention is in use,
3982 @var{funtype} is examined to determine whether a function takes a fixed
3983 number of arguments.
3986 @defmac CALL_POPS_ARGS (@var{cum})
3987 A C expression that should indicate the number of bytes a call sequence
3988 pops off the stack. It is added to the value of @code{RETURN_POPS_ARGS}
3989 when compiling a function call.
3991 @var{cum} is the variable in which all arguments to the called function
3992 have been accumulated.
3994 On certain architectures, such as the SH5, a call trampoline is used
3995 that pops certain registers off the stack, depending on the arguments
3996 that have been passed to the function. Since this is a property of the
3997 call site, not of the called function, @code{RETURN_POPS_ARGS} is not
4001 @node Register Arguments
4002 @subsection Passing Arguments in Registers
4003 @cindex arguments in registers
4004 @cindex registers arguments
4006 This section describes the macros which let you control how various
4007 types of arguments are passed in registers or how they are arranged in
4010 @defmac FUNCTION_ARG (@var{cum}, @var{mode}, @var{type}, @var{named})
4011 A C expression that controls whether a function argument is passed
4012 in a register, and which register.
4014 The arguments are @var{cum}, which summarizes all the previous
4015 arguments; @var{mode}, the machine mode of the argument; @var{type},
4016 the data type of the argument as a tree node or 0 if that is not known
4017 (which happens for C support library functions); and @var{named},
4018 which is 1 for an ordinary argument and 0 for nameless arguments that
4019 correspond to @samp{@dots{}} in the called function's prototype.
4020 @var{type} can be an incomplete type if a syntax error has previously
4023 The value of the expression is usually either a @code{reg} RTX for the
4024 hard register in which to pass the argument, or zero to pass the
4025 argument on the stack.
4027 For machines like the VAX and 68000, where normally all arguments are
4028 pushed, zero suffices as a definition.
4030 The value of the expression can also be a @code{parallel} RTX@. This is
4031 used when an argument is passed in multiple locations. The mode of the
4032 @code{parallel} should be the mode of the entire argument. The
4033 @code{parallel} holds any number of @code{expr_list} pairs; each one
4034 describes where part of the argument is passed. In each
4035 @code{expr_list} the first operand must be a @code{reg} RTX for the hard
4036 register in which to pass this part of the argument, and the mode of the
4037 register RTX indicates how large this part of the argument is. The
4038 second operand of the @code{expr_list} is a @code{const_int} which gives
4039 the offset in bytes into the entire argument of where this part starts.
4040 As a special exception the first @code{expr_list} in the @code{parallel}
4041 RTX may have a first operand of zero. This indicates that the entire
4042 argument is also stored on the stack.
4044 The last time this macro is called, it is called with @code{MODE ==
4045 VOIDmode}, and its result is passed to the @code{call} or @code{call_value}
4046 pattern as operands 2 and 3 respectively.
4048 @cindex @file{stdarg.h} and register arguments
4049 The usual way to make the ISO library @file{stdarg.h} work on a machine
4050 where some arguments are usually passed in registers, is to cause
4051 nameless arguments to be passed on the stack instead. This is done
4052 by making @code{FUNCTION_ARG} return 0 whenever @var{named} is 0.
4054 @cindex @code{TARGET_MUST_PASS_IN_STACK}, and @code{FUNCTION_ARG}
4055 @cindex @code{REG_PARM_STACK_SPACE}, and @code{FUNCTION_ARG}
4056 You may use the hook @code{targetm.calls.must_pass_in_stack}
4057 in the definition of this macro to determine if this argument is of a
4058 type that must be passed in the stack. If @code{REG_PARM_STACK_SPACE}
4059 is not defined and @code{FUNCTION_ARG} returns nonzero for such an
4060 argument, the compiler will abort. If @code{REG_PARM_STACK_SPACE} is
4061 defined, the argument will be computed in the stack and then loaded into
4065 @deftypefn {Target Hook} bool TARGET_MUST_PASS_IN_STACK (enum machine_mode @var{mode}, tree @var{type})
4066 This target hook should return @code{true} if we should not pass @var{type}
4067 solely in registers. The file @file{expr.h} defines a
4068 definition that is usually appropriate, refer to @file{expr.h} for additional
4072 @defmac FUNCTION_INCOMING_ARG (@var{cum}, @var{mode}, @var{type}, @var{named})
4073 Define this macro if the target machine has ``register windows'', so
4074 that the register in which a function sees an arguments is not
4075 necessarily the same as the one in which the caller passed the
4078 For such machines, @code{FUNCTION_ARG} computes the register in which
4079 the caller passes the value, and @code{FUNCTION_INCOMING_ARG} should
4080 be defined in a similar fashion to tell the function being called
4081 where the arguments will arrive.
4083 If @code{FUNCTION_INCOMING_ARG} is not defined, @code{FUNCTION_ARG}
4084 serves both purposes.
4087 @deftypefn {Target Hook} int TARGET_ARG_PARTIAL_BYTES (CUMULATIVE_ARGS *@var{cum}, enum machine_mode @var{mode}, tree @var{type}, bool @var{named})
4088 This target hook returns the number of bytes at the beginning of an
4089 argument that must be put in registers. The value must be zero for
4090 arguments that are passed entirely in registers or that are entirely
4091 pushed on the stack.
4093 On some machines, certain arguments must be passed partially in
4094 registers and partially in memory. On these machines, typically the
4095 first few words of arguments are passed in registers, and the rest
4096 on the stack. If a multi-word argument (a @code{double} or a
4097 structure) crosses that boundary, its first few words must be passed
4098 in registers and the rest must be pushed. This macro tells the
4099 compiler when this occurs, and how many bytes should go in registers.
4101 @code{FUNCTION_ARG} for these arguments should return the first
4102 register to be used by the caller for this argument; likewise
4103 @code{FUNCTION_INCOMING_ARG}, for the called function.
4106 @deftypefn {Target Hook} bool TARGET_PASS_BY_REFERENCE (CUMULATIVE_ARGS *@var{cum}, enum machine_mode @var{mode}, tree @var{type}, bool @var{named})
4107 This target hook should return @code{true} if an argument at the
4108 position indicated by @var{cum} should be passed by reference. This
4109 predicate is queried after target independent reasons for being
4110 passed by reference, such as @code{TREE_ADDRESSABLE (type)}.
4112 If the hook returns true, a copy of that argument is made in memory and a
4113 pointer to the argument is passed instead of the argument itself.
4114 The pointer is passed in whatever way is appropriate for passing a pointer
4118 @deftypefn {Target Hook} bool TARGET_CALLEE_COPIES (CUMULATIVE_ARGS *@var{cum}, enum machine_mode @var{mode}, tree @var{type}, bool @var{named})
4119 The function argument described by the parameters to this hook is
4120 known to be passed by reference. The hook should return true if the
4121 function argument should be copied by the callee instead of copied
4124 For any argument for which the hook returns true, if it can be
4125 determined that the argument is not modified, then a copy need
4128 The default version of this hook always returns false.
4131 @defmac CUMULATIVE_ARGS
4132 A C type for declaring a variable that is used as the first argument of
4133 @code{FUNCTION_ARG} and other related values. For some target machines,
4134 the type @code{int} suffices and can hold the number of bytes of
4137 There is no need to record in @code{CUMULATIVE_ARGS} anything about the
4138 arguments that have been passed on the stack. The compiler has other
4139 variables to keep track of that. For target machines on which all
4140 arguments are passed on the stack, there is no need to store anything in
4141 @code{CUMULATIVE_ARGS}; however, the data structure must exist and
4142 should not be empty, so use @code{int}.
4145 @defmac OVERRIDE_ABI_FORMAT (@var{fndecl})
4146 If defined, this macro is called before generating any code for a
4147 function, but after the @var{cfun} descriptor for the function has been
4148 created. The back end may use this macro to update @var{cfun} to
4149 reflect an ABI other than that which would normally be used by default.
4150 If the compiler is generating code for a compiler-generated function,
4151 @var{fndecl} may be @code{NULL}.
4154 @defmac INIT_CUMULATIVE_ARGS (@var{cum}, @var{fntype}, @var{libname}, @var{fndecl}, @var{n_named_args})
4155 A C statement (sans semicolon) for initializing the variable
4156 @var{cum} for the state at the beginning of the argument list. The
4157 variable has type @code{CUMULATIVE_ARGS}. The value of @var{fntype}
4158 is the tree node for the data type of the function which will receive
4159 the args, or 0 if the args are to a compiler support library function.
4160 For direct calls that are not libcalls, @var{fndecl} contain the
4161 declaration node of the function. @var{fndecl} is also set when
4162 @code{INIT_CUMULATIVE_ARGS} is used to find arguments for the function
4163 being compiled. @var{n_named_args} is set to the number of named
4164 arguments, including a structure return address if it is passed as a
4165 parameter, when making a call. When processing incoming arguments,
4166 @var{n_named_args} is set to @minus{}1.
4168 When processing a call to a compiler support library function,
4169 @var{libname} identifies which one. It is a @code{symbol_ref} rtx which
4170 contains the name of the function, as a string. @var{libname} is 0 when
4171 an ordinary C function call is being processed. Thus, each time this
4172 macro is called, either @var{libname} or @var{fntype} is nonzero, but
4173 never both of them at once.
4176 @defmac INIT_CUMULATIVE_LIBCALL_ARGS (@var{cum}, @var{mode}, @var{libname})
4177 Like @code{INIT_CUMULATIVE_ARGS} but only used for outgoing libcalls,
4178 it gets a @code{MODE} argument instead of @var{fntype}, that would be
4179 @code{NULL}. @var{indirect} would always be zero, too. If this macro
4180 is not defined, @code{INIT_CUMULATIVE_ARGS (cum, NULL_RTX, libname,
4181 0)} is used instead.
4184 @defmac INIT_CUMULATIVE_INCOMING_ARGS (@var{cum}, @var{fntype}, @var{libname})
4185 Like @code{INIT_CUMULATIVE_ARGS} but overrides it for the purposes of
4186 finding the arguments for the function being compiled. If this macro is
4187 undefined, @code{INIT_CUMULATIVE_ARGS} is used instead.
4189 The value passed for @var{libname} is always 0, since library routines
4190 with special calling conventions are never compiled with GCC@. The
4191 argument @var{libname} exists for symmetry with
4192 @code{INIT_CUMULATIVE_ARGS}.
4193 @c could use "this macro" in place of @code{INIT_CUMULATIVE_ARGS}, maybe.
4194 @c --mew 5feb93 i switched the order of the sentences. --mew 10feb93
4197 @defmac FUNCTION_ARG_ADVANCE (@var{cum}, @var{mode}, @var{type}, @var{named})
4198 A C statement (sans semicolon) to update the summarizer variable
4199 @var{cum} to advance past an argument in the argument list. The
4200 values @var{mode}, @var{type} and @var{named} describe that argument.
4201 Once this is done, the variable @var{cum} is suitable for analyzing
4202 the @emph{following} argument with @code{FUNCTION_ARG}, etc.
4204 This macro need not do anything if the argument in question was passed
4205 on the stack. The compiler knows how to track the amount of stack space
4206 used for arguments without any special help.
4209 @defmac FUNCTION_ARG_OFFSET (@var{mode}, @var{type})
4210 If defined, a C expression that is the number of bytes to add to the
4211 offset of the argument passed in memory. This is needed for the SPU,
4212 which passes @code{char} and @code{short} arguments in the preferred
4213 slot that is in the middle of the quad word instead of starting at the
4217 @defmac FUNCTION_ARG_PADDING (@var{mode}, @var{type})
4218 If defined, a C expression which determines whether, and in which direction,
4219 to pad out an argument with extra space. The value should be of type
4220 @code{enum direction}: either @code{upward} to pad above the argument,
4221 @code{downward} to pad below, or @code{none} to inhibit padding.
4223 The @emph{amount} of padding is always just enough to reach the next
4224 multiple of @code{FUNCTION_ARG_BOUNDARY}; this macro does not control
4227 This macro has a default definition which is right for most systems.
4228 For little-endian machines, the default is to pad upward. For
4229 big-endian machines, the default is to pad downward for an argument of
4230 constant size shorter than an @code{int}, and upward otherwise.
4233 @defmac PAD_VARARGS_DOWN
4234 If defined, a C expression which determines whether the default
4235 implementation of va_arg will attempt to pad down before reading the
4236 next argument, if that argument is smaller than its aligned space as
4237 controlled by @code{PARM_BOUNDARY}. If this macro is not defined, all such
4238 arguments are padded down if @code{BYTES_BIG_ENDIAN} is true.
4241 @defmac BLOCK_REG_PADDING (@var{mode}, @var{type}, @var{first})
4242 Specify padding for the last element of a block move between registers and
4243 memory. @var{first} is nonzero if this is the only element. Defining this
4244 macro allows better control of register function parameters on big-endian
4245 machines, without using @code{PARALLEL} rtl. In particular,
4246 @code{MUST_PASS_IN_STACK} need not test padding and mode of types in
4247 registers, as there is no longer a "wrong" part of a register; For example,
4248 a three byte aggregate may be passed in the high part of a register if so
4252 @defmac FUNCTION_ARG_BOUNDARY (@var{mode}, @var{type})
4253 If defined, a C expression that gives the alignment boundary, in bits,
4254 of an argument with the specified mode and type. If it is not defined,
4255 @code{PARM_BOUNDARY} is used for all arguments.
4258 @defmac FUNCTION_ARG_REGNO_P (@var{regno})
4259 A C expression that is nonzero if @var{regno} is the number of a hard
4260 register in which function arguments are sometimes passed. This does
4261 @emph{not} include implicit arguments such as the static chain and
4262 the structure-value address. On many machines, no registers can be
4263 used for this purpose since all function arguments are pushed on the
4267 @deftypefn {Target Hook} bool TARGET_SPLIT_COMPLEX_ARG (tree @var{type})
4268 This hook should return true if parameter of type @var{type} are passed
4269 as two scalar parameters. By default, GCC will attempt to pack complex
4270 arguments into the target's word size. Some ABIs require complex arguments
4271 to be split and treated as their individual components. For example, on
4272 AIX64, complex floats should be passed in a pair of floating point
4273 registers, even though a complex float would fit in one 64-bit floating
4276 The default value of this hook is @code{NULL}, which is treated as always
4280 @deftypefn {Target Hook} tree TARGET_BUILD_BUILTIN_VA_LIST (void)
4281 This hook returns a type node for @code{va_list} for the target.
4282 The default version of the hook returns @code{void*}.
4285 @deftypefn {Target Hook} tree TARGET_FN_ABI_VA_LIST (tree @var{fndecl})
4286 This hook returns the va_list type of the calling convention specified by
4288 The default version of this hook returns @code{va_list_type_node}.
4291 @deftypefn {Target Hook} tree TARGET_CANONICAL_VA_LIST_TYPE (tree @var{type})
4292 This hook returns the va_list type of the calling convention specified by the
4293 type of @var{type}. If @var{type} is not a valid va_list type, it returns
4297 @deftypefn {Target Hook} tree TARGET_GIMPLIFY_VA_ARG_EXPR (tree @var{valist}, tree @var{type}, tree *@var{pre_p}, tree *@var{post_p})
4298 This hook performs target-specific gimplification of
4299 @code{VA_ARG_EXPR}. The first two parameters correspond to the
4300 arguments to @code{va_arg}; the latter two are as in
4301 @code{gimplify.c:gimplify_expr}.
4304 @deftypefn {Target Hook} bool TARGET_VALID_POINTER_MODE (enum machine_mode @var{mode})
4305 Define this to return nonzero if the port can handle pointers
4306 with machine mode @var{mode}. The default version of this
4307 hook returns true for both @code{ptr_mode} and @code{Pmode}.
4310 @deftypefn {Target Hook} bool TARGET_SCALAR_MODE_SUPPORTED_P (enum machine_mode @var{mode})
4311 Define this to return nonzero if the port is prepared to handle
4312 insns involving scalar mode @var{mode}. For a scalar mode to be
4313 considered supported, all the basic arithmetic and comparisons
4316 The default version of this hook returns true for any mode
4317 required to handle the basic C types (as defined by the port).
4318 Included here are the double-word arithmetic supported by the
4319 code in @file{optabs.c}.
4322 @deftypefn {Target Hook} bool TARGET_VECTOR_MODE_SUPPORTED_P (enum machine_mode @var{mode})
4323 Define this to return nonzero if the port is prepared to handle
4324 insns involving vector mode @var{mode}. At the very least, it
4325 must have move patterns for this mode.
4329 @subsection How Scalar Function Values Are Returned
4330 @cindex return values in registers
4331 @cindex values, returned by functions
4332 @cindex scalars, returned as values
4334 This section discusses the macros that control returning scalars as
4335 values---values that can fit in registers.
4337 @deftypefn {Target Hook} rtx TARGET_FUNCTION_VALUE (tree @var{ret_type}, tree @var{fn_decl_or_type}, bool @var{outgoing})
4339 Define this to return an RTX representing the place where a function
4340 returns or receives a value of data type @var{ret_type}, a tree node
4341 representing a data type. @var{fn_decl_or_type} is a tree node
4342 representing @code{FUNCTION_DECL} or @code{FUNCTION_TYPE} of a
4343 function being called. If @var{outgoing} is false, the hook should
4344 compute the register in which the caller will see the return value.
4345 Otherwise, the hook should return an RTX representing the place where
4346 a function returns a value.
4348 On many machines, only @code{TYPE_MODE (@var{ret_type})} is relevant.
4349 (Actually, on most machines, scalar values are returned in the same
4350 place regardless of mode.) The value of the expression is usually a
4351 @code{reg} RTX for the hard register where the return value is stored.
4352 The value can also be a @code{parallel} RTX, if the return value is in
4353 multiple places. See @code{FUNCTION_ARG} for an explanation of the
4354 @code{parallel} form. Note that the callee will populate every
4355 location specified in the @code{parallel}, but if the first element of
4356 the @code{parallel} contains the whole return value, callers will use
4357 that element as the canonical location and ignore the others. The m68k
4358 port uses this type of @code{parallel} to return pointers in both
4359 @samp{%a0} (the canonical location) and @samp{%d0}.
4361 If @code{TARGET_PROMOTE_FUNCTION_RETURN} returns true, you must apply
4362 the same promotion rules specified in @code{PROMOTE_MODE} if
4363 @var{valtype} is a scalar type.
4365 If the precise function being called is known, @var{func} is a tree
4366 node (@code{FUNCTION_DECL}) for it; otherwise, @var{func} is a null
4367 pointer. This makes it possible to use a different value-returning
4368 convention for specific functions when all their calls are
4371 Some target machines have ``register windows'' so that the register in
4372 which a function returns its value is not the same as the one in which
4373 the caller sees the value. For such machines, you should return
4374 different RTX depending on @var{outgoing}.
4376 @code{TARGET_FUNCTION_VALUE} is not used for return values with
4377 aggregate data types, because these are returned in another way. See
4378 @code{TARGET_STRUCT_VALUE_RTX} and related macros, below.
4381 @defmac FUNCTION_VALUE (@var{valtype}, @var{func})
4382 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE} for
4383 a new target instead.
4386 @defmac FUNCTION_OUTGOING_VALUE (@var{valtype}, @var{func})
4387 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE} for
4388 a new target instead.
4391 @defmac LIBCALL_VALUE (@var{mode})
4392 A C expression to create an RTX representing the place where a library
4393 function returns a value of mode @var{mode}.
4395 Note that ``library function'' in this context means a compiler
4396 support routine, used to perform arithmetic, whose name is known
4397 specially by the compiler and was not mentioned in the C code being
4401 @deftypefn {Target Hook} rtx TARGET_LIBCALL_VALUE (enum machine_mode
4402 @var{mode}, const_rtx @var{fun})
4403 Define this hook if the back-end needs to know the name of the libcall
4404 function in order to determine where the result should be returned.
4406 The mode of the result is given by @var{mode} and the name of the called
4407 library function is given by @var{fun}. The hook should return an RTX
4408 representing the place where the library function result will be returned.
4410 If this hook is not defined, then LIBCALL_VALUE will be used.
4413 @defmac FUNCTION_VALUE_REGNO_P (@var{regno})
4414 A C expression that is nonzero if @var{regno} is the number of a hard
4415 register in which the values of called function may come back.
4417 A register whose use for returning values is limited to serving as the
4418 second of a pair (for a value of type @code{double}, say) need not be
4419 recognized by this macro. So for most machines, this definition
4423 #define FUNCTION_VALUE_REGNO_P(N) ((N) == 0)
4426 If the machine has register windows, so that the caller and the called
4427 function use different registers for the return value, this macro
4428 should recognize only the caller's register numbers.
4431 @defmac TARGET_ENUM_VA_LIST (@var{idx}, @var{pname}, @var{ptype})
4432 This target macro is used in function @code{c_common_nodes_and_builtins}
4433 to iterate through the target specific builtin types for va_list. The
4434 variable @var{idx} is used as iterator. @var{pname} has to be a pointer
4435 to a @code{const char *} and @var{ptype} a pointer to a @code{tree} typed
4437 The arguments @var{pname} and @var{ptype} are used to store the result of
4438 this macro and are set to the name of the va_list builtin type and its
4440 If the return value of this macro is zero, then there is no more element.
4441 Otherwise the @var{IDX} should be increased for the next call of this
4442 macro to iterate through all types.
4445 @defmac APPLY_RESULT_SIZE
4446 Define this macro if @samp{untyped_call} and @samp{untyped_return}
4447 need more space than is implied by @code{FUNCTION_VALUE_REGNO_P} for
4448 saving and restoring an arbitrary return value.
4451 @deftypefn {Target Hook} bool TARGET_RETURN_IN_MSB (tree @var{type})
4452 This hook should return true if values of type @var{type} are returned
4453 at the most significant end of a register (in other words, if they are
4454 padded at the least significant end). You can assume that @var{type}
4455 is returned in a register; the caller is required to check this.
4457 Note that the register provided by @code{TARGET_FUNCTION_VALUE} must
4458 be able to hold the complete return value. For example, if a 1-, 2-
4459 or 3-byte structure is returned at the most significant end of a
4460 4-byte register, @code{TARGET_FUNCTION_VALUE} should provide an
4464 @node Aggregate Return
4465 @subsection How Large Values Are Returned
4466 @cindex aggregates as return values
4467 @cindex large return values
4468 @cindex returning aggregate values
4469 @cindex structure value address
4471 When a function value's mode is @code{BLKmode} (and in some other
4472 cases), the value is not returned according to
4473 @code{TARGET_FUNCTION_VALUE} (@pxref{Scalar Return}). Instead, the
4474 caller passes the address of a block of memory in which the value
4475 should be stored. This address is called the @dfn{structure value
4478 This section describes how to control returning structure values in
4481 @deftypefn {Target Hook} bool TARGET_RETURN_IN_MEMORY (tree @var{type}, tree @var{fntype})
4482 This target hook should return a nonzero value to say to return the
4483 function value in memory, just as large structures are always returned.
4484 Here @var{type} will be the data type of the value, and @var{fntype}
4485 will be the type of the function doing the returning, or @code{NULL} for
4488 Note that values of mode @code{BLKmode} must be explicitly handled
4489 by this function. Also, the option @option{-fpcc-struct-return}
4490 takes effect regardless of this macro. On most systems, it is
4491 possible to leave the hook undefined; this causes a default
4492 definition to be used, whose value is the constant 1 for @code{BLKmode}
4493 values, and 0 otherwise.
4495 Do not use this hook to indicate that structures and unions should always
4496 be returned in memory. You should instead use @code{DEFAULT_PCC_STRUCT_RETURN}
4500 @defmac DEFAULT_PCC_STRUCT_RETURN
4501 Define this macro to be 1 if all structure and union return values must be
4502 in memory. Since this results in slower code, this should be defined
4503 only if needed for compatibility with other compilers or with an ABI@.
4504 If you define this macro to be 0, then the conventions used for structure
4505 and union return values are decided by the @code{TARGET_RETURN_IN_MEMORY}
4508 If not defined, this defaults to the value 1.
4511 @deftypefn {Target Hook} rtx TARGET_STRUCT_VALUE_RTX (tree @var{fndecl}, int @var{incoming})
4512 This target hook should return the location of the structure value
4513 address (normally a @code{mem} or @code{reg}), or 0 if the address is
4514 passed as an ``invisible'' first argument. Note that @var{fndecl} may
4515 be @code{NULL}, for libcalls. You do not need to define this target
4516 hook if the address is always passed as an ``invisible'' first
4519 On some architectures the place where the structure value address
4520 is found by the called function is not the same place that the
4521 caller put it. This can be due to register windows, or it could
4522 be because the function prologue moves it to a different place.
4523 @var{incoming} is @code{1} or @code{2} when the location is needed in
4524 the context of the called function, and @code{0} in the context of
4527 If @var{incoming} is nonzero and the address is to be found on the
4528 stack, return a @code{mem} which refers to the frame pointer. If
4529 @var{incoming} is @code{2}, the result is being used to fetch the
4530 structure value address at the beginning of a function. If you need
4531 to emit adjusting code, you should do it at this point.
4534 @defmac PCC_STATIC_STRUCT_RETURN
4535 Define this macro if the usual system convention on the target machine
4536 for returning structures and unions is for the called function to return
4537 the address of a static variable containing the value.
4539 Do not define this if the usual system convention is for the caller to
4540 pass an address to the subroutine.
4542 This macro has effect in @option{-fpcc-struct-return} mode, but it does
4543 nothing when you use @option{-freg-struct-return} mode.
4547 @subsection Caller-Saves Register Allocation
4549 If you enable it, GCC can save registers around function calls. This
4550 makes it possible to use call-clobbered registers to hold variables that
4551 must live across calls.
4553 @defmac CALLER_SAVE_PROFITABLE (@var{refs}, @var{calls})
4554 A C expression to determine whether it is worthwhile to consider placing
4555 a pseudo-register in a call-clobbered hard register and saving and
4556 restoring it around each function call. The expression should be 1 when
4557 this is worth doing, and 0 otherwise.
4559 If you don't define this macro, a default is used which is good on most
4560 machines: @code{4 * @var{calls} < @var{refs}}.
4563 @defmac HARD_REGNO_CALLER_SAVE_MODE (@var{regno}, @var{nregs})
4564 A C expression specifying which mode is required for saving @var{nregs}
4565 of a pseudo-register in call-clobbered hard register @var{regno}. If
4566 @var{regno} is unsuitable for caller save, @code{VOIDmode} should be
4567 returned. For most machines this macro need not be defined since GCC
4568 will select the smallest suitable mode.
4571 @node Function Entry
4572 @subsection Function Entry and Exit
4573 @cindex function entry and exit
4577 This section describes the macros that output function entry
4578 (@dfn{prologue}) and exit (@dfn{epilogue}) code.
4580 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_PROLOGUE (FILE *@var{file}, HOST_WIDE_INT @var{size})
4581 If defined, a function that outputs the assembler code for entry to a
4582 function. The prologue is responsible for setting up the stack frame,
4583 initializing the frame pointer register, saving registers that must be
4584 saved, and allocating @var{size} additional bytes of storage for the
4585 local variables. @var{size} is an integer. @var{file} is a stdio
4586 stream to which the assembler code should be output.
4588 The label for the beginning of the function need not be output by this
4589 macro. That has already been done when the macro is run.
4591 @findex regs_ever_live
4592 To determine which registers to save, the macro can refer to the array
4593 @code{regs_ever_live}: element @var{r} is nonzero if hard register
4594 @var{r} is used anywhere within the function. This implies the function
4595 prologue should save register @var{r}, provided it is not one of the
4596 call-used registers. (@code{TARGET_ASM_FUNCTION_EPILOGUE} must likewise use
4597 @code{regs_ever_live}.)
4599 On machines that have ``register windows'', the function entry code does
4600 not save on the stack the registers that are in the windows, even if
4601 they are supposed to be preserved by function calls; instead it takes
4602 appropriate steps to ``push'' the register stack, if any non-call-used
4603 registers are used in the function.
4605 @findex frame_pointer_needed
4606 On machines where functions may or may not have frame-pointers, the
4607 function entry code must vary accordingly; it must set up the frame
4608 pointer if one is wanted, and not otherwise. To determine whether a
4609 frame pointer is in wanted, the macro can refer to the variable
4610 @code{frame_pointer_needed}. The variable's value will be 1 at run
4611 time in a function that needs a frame pointer. @xref{Elimination}.
4613 The function entry code is responsible for allocating any stack space
4614 required for the function. This stack space consists of the regions
4615 listed below. In most cases, these regions are allocated in the
4616 order listed, with the last listed region closest to the top of the
4617 stack (the lowest address if @code{STACK_GROWS_DOWNWARD} is defined, and
4618 the highest address if it is not defined). You can use a different order
4619 for a machine if doing so is more convenient or required for
4620 compatibility reasons. Except in cases where required by standard
4621 or by a debugger, there is no reason why the stack layout used by GCC
4622 need agree with that used by other compilers for a machine.
4625 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_END_PROLOGUE (FILE *@var{file})
4626 If defined, a function that outputs assembler code at the end of a
4627 prologue. This should be used when the function prologue is being
4628 emitted as RTL, and you have some extra assembler that needs to be
4629 emitted. @xref{prologue instruction pattern}.
4632 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_BEGIN_EPILOGUE (FILE *@var{file})
4633 If defined, a function that outputs assembler code at the start of an
4634 epilogue. This should be used when the function epilogue is being
4635 emitted as RTL, and you have some extra assembler that needs to be
4636 emitted. @xref{epilogue instruction pattern}.
4639 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_EPILOGUE (FILE *@var{file}, HOST_WIDE_INT @var{size})
4640 If defined, a function that outputs the assembler code for exit from a
4641 function. The epilogue is responsible for restoring the saved
4642 registers and stack pointer to their values when the function was
4643 called, and returning control to the caller. This macro takes the
4644 same arguments as the macro @code{TARGET_ASM_FUNCTION_PROLOGUE}, and the
4645 registers to restore are determined from @code{regs_ever_live} and
4646 @code{CALL_USED_REGISTERS} in the same way.
4648 On some machines, there is a single instruction that does all the work
4649 of returning from the function. On these machines, give that
4650 instruction the name @samp{return} and do not define the macro
4651 @code{TARGET_ASM_FUNCTION_EPILOGUE} at all.
4653 Do not define a pattern named @samp{return} if you want the
4654 @code{TARGET_ASM_FUNCTION_EPILOGUE} to be used. If you want the target
4655 switches to control whether return instructions or epilogues are used,
4656 define a @samp{return} pattern with a validity condition that tests the
4657 target switches appropriately. If the @samp{return} pattern's validity
4658 condition is false, epilogues will be used.
4660 On machines where functions may or may not have frame-pointers, the
4661 function exit code must vary accordingly. Sometimes the code for these
4662 two cases is completely different. To determine whether a frame pointer
4663 is wanted, the macro can refer to the variable
4664 @code{frame_pointer_needed}. The variable's value will be 1 when compiling
4665 a function that needs a frame pointer.
4667 Normally, @code{TARGET_ASM_FUNCTION_PROLOGUE} and
4668 @code{TARGET_ASM_FUNCTION_EPILOGUE} must treat leaf functions specially.
4669 The C variable @code{current_function_is_leaf} is nonzero for such a
4670 function. @xref{Leaf Functions}.
4672 On some machines, some functions pop their arguments on exit while
4673 others leave that for the caller to do. For example, the 68020 when
4674 given @option{-mrtd} pops arguments in functions that take a fixed
4675 number of arguments.
4677 @findex current_function_pops_args
4678 Your definition of the macro @code{RETURN_POPS_ARGS} decides which
4679 functions pop their own arguments. @code{TARGET_ASM_FUNCTION_EPILOGUE}
4680 needs to know what was decided. The variable that is called
4681 @code{current_function_pops_args} is the number of bytes of its
4682 arguments that a function should pop. @xref{Scalar Return}.
4683 @c what is the "its arguments" in the above sentence referring to, pray
4684 @c tell? --mew 5feb93
4689 @findex current_function_pretend_args_size
4690 A region of @code{current_function_pretend_args_size} bytes of
4691 uninitialized space just underneath the first argument arriving on the
4692 stack. (This may not be at the very start of the allocated stack region
4693 if the calling sequence has pushed anything else since pushing the stack
4694 arguments. But usually, on such machines, nothing else has been pushed
4695 yet, because the function prologue itself does all the pushing.) This
4696 region is used on machines where an argument may be passed partly in
4697 registers and partly in memory, and, in some cases to support the
4698 features in @code{<stdarg.h>}.
4701 An area of memory used to save certain registers used by the function.
4702 The size of this area, which may also include space for such things as
4703 the return address and pointers to previous stack frames, is
4704 machine-specific and usually depends on which registers have been used
4705 in the function. Machines with register windows often do not require
4709 A region of at least @var{size} bytes, possibly rounded up to an allocation
4710 boundary, to contain the local variables of the function. On some machines,
4711 this region and the save area may occur in the opposite order, with the
4712 save area closer to the top of the stack.
4715 @cindex @code{ACCUMULATE_OUTGOING_ARGS} and stack frames
4716 Optionally, when @code{ACCUMULATE_OUTGOING_ARGS} is defined, a region of
4717 @code{current_function_outgoing_args_size} bytes to be used for outgoing
4718 argument lists of the function. @xref{Stack Arguments}.
4721 @defmac EXIT_IGNORE_STACK
4722 Define this macro as a C expression that is nonzero if the return
4723 instruction or the function epilogue ignores the value of the stack
4724 pointer; in other words, if it is safe to delete an instruction to
4725 adjust the stack pointer before a return from the function. The
4728 Note that this macro's value is relevant only for functions for which
4729 frame pointers are maintained. It is never safe to delete a final
4730 stack adjustment in a function that has no frame pointer, and the
4731 compiler knows this regardless of @code{EXIT_IGNORE_STACK}.
4734 @defmac EPILOGUE_USES (@var{regno})
4735 Define this macro as a C expression that is nonzero for registers that are
4736 used by the epilogue or the @samp{return} pattern. The stack and frame
4737 pointer registers are already assumed to be used as needed.
4740 @defmac EH_USES (@var{regno})
4741 Define this macro as a C expression that is nonzero for registers that are
4742 used by the exception handling mechanism, and so should be considered live
4743 on entry to an exception edge.
4746 @defmac DELAY_SLOTS_FOR_EPILOGUE
4747 Define this macro if the function epilogue contains delay slots to which
4748 instructions from the rest of the function can be ``moved''. The
4749 definition should be a C expression whose value is an integer
4750 representing the number of delay slots there.
4753 @defmac ELIGIBLE_FOR_EPILOGUE_DELAY (@var{insn}, @var{n})
4754 A C expression that returns 1 if @var{insn} can be placed in delay
4755 slot number @var{n} of the epilogue.
4757 The argument @var{n} is an integer which identifies the delay slot now
4758 being considered (since different slots may have different rules of
4759 eligibility). It is never negative and is always less than the number
4760 of epilogue delay slots (what @code{DELAY_SLOTS_FOR_EPILOGUE} returns).
4761 If you reject a particular insn for a given delay slot, in principle, it
4762 may be reconsidered for a subsequent delay slot. Also, other insns may
4763 (at least in principle) be considered for the so far unfilled delay
4766 @findex current_function_epilogue_delay_list
4767 @findex final_scan_insn
4768 The insns accepted to fill the epilogue delay slots are put in an RTL
4769 list made with @code{insn_list} objects, stored in the variable
4770 @code{current_function_epilogue_delay_list}. The insn for the first
4771 delay slot comes first in the list. Your definition of the macro
4772 @code{TARGET_ASM_FUNCTION_EPILOGUE} should fill the delay slots by
4773 outputting the insns in this list, usually by calling
4774 @code{final_scan_insn}.
4776 You need not define this macro if you did not define
4777 @code{DELAY_SLOTS_FOR_EPILOGUE}.
4780 @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_MI_THUNK (FILE *@var{file}, tree @var{thunk_fndecl}, HOST_WIDE_INT @var{delta}, HOST_WIDE_INT @var{vcall_offset}, tree @var{function})
4781 A function that outputs the assembler code for a thunk
4782 function, used to implement C++ virtual function calls with multiple
4783 inheritance. The thunk acts as a wrapper around a virtual function,
4784 adjusting the implicit object parameter before handing control off to
4787 First, emit code to add the integer @var{delta} to the location that
4788 contains the incoming first argument. Assume that this argument
4789 contains a pointer, and is the one used to pass the @code{this} pointer
4790 in C++. This is the incoming argument @emph{before} the function prologue,
4791 e.g.@: @samp{%o0} on a sparc. The addition must preserve the values of
4792 all other incoming arguments.
4794 Then, if @var{vcall_offset} is nonzero, an additional adjustment should be
4795 made after adding @code{delta}. In particular, if @var{p} is the
4796 adjusted pointer, the following adjustment should be made:
4799 p += (*((ptrdiff_t **)p))[vcall_offset/sizeof(ptrdiff_t)]
4802 After the additions, emit code to jump to @var{function}, which is a
4803 @code{FUNCTION_DECL}. This is a direct pure jump, not a call, and does
4804 not touch the return address. Hence returning from @var{FUNCTION} will
4805 return to whoever called the current @samp{thunk}.
4807 The effect must be as if @var{function} had been called directly with
4808 the adjusted first argument. This macro is responsible for emitting all
4809 of the code for a thunk function; @code{TARGET_ASM_FUNCTION_PROLOGUE}
4810 and @code{TARGET_ASM_FUNCTION_EPILOGUE} are not invoked.
4812 The @var{thunk_fndecl} is redundant. (@var{delta} and @var{function}
4813 have already been extracted from it.) It might possibly be useful on
4814 some targets, but probably not.
4816 If you do not define this macro, the target-independent code in the C++
4817 front end will generate a less efficient heavyweight thunk that calls
4818 @var{function} instead of jumping to it. The generic approach does
4819 not support varargs.
4822 @deftypefn {Target Hook} bool TARGET_ASM_CAN_OUTPUT_MI_THUNK (tree @var{thunk_fndecl}, HOST_WIDE_INT @var{delta}, HOST_WIDE_INT @var{vcall_offset}, tree @var{function})
4823 A function that returns true if TARGET_ASM_OUTPUT_MI_THUNK would be able
4824 to output the assembler code for the thunk function specified by the
4825 arguments it is passed, and false otherwise. In the latter case, the
4826 generic approach will be used by the C++ front end, with the limitations
4831 @subsection Generating Code for Profiling
4832 @cindex profiling, code generation
4834 These macros will help you generate code for profiling.
4836 @defmac FUNCTION_PROFILER (@var{file}, @var{labelno})
4837 A C statement or compound statement to output to @var{file} some
4838 assembler code to call the profiling subroutine @code{mcount}.
4841 The details of how @code{mcount} expects to be called are determined by
4842 your operating system environment, not by GCC@. To figure them out,
4843 compile a small program for profiling using the system's installed C
4844 compiler and look at the assembler code that results.
4846 Older implementations of @code{mcount} expect the address of a counter
4847 variable to be loaded into some register. The name of this variable is
4848 @samp{LP} followed by the number @var{labelno}, so you would generate
4849 the name using @samp{LP%d} in a @code{fprintf}.
4852 @defmac PROFILE_HOOK
4853 A C statement or compound statement to output to @var{file} some assembly
4854 code to call the profiling subroutine @code{mcount} even the target does
4855 not support profiling.
4858 @defmac NO_PROFILE_COUNTERS
4859 Define this macro to be an expression with a nonzero value if the
4860 @code{mcount} subroutine on your system does not need a counter variable
4861 allocated for each function. This is true for almost all modern
4862 implementations. If you define this macro, you must not use the
4863 @var{labelno} argument to @code{FUNCTION_PROFILER}.
4866 @defmac PROFILE_BEFORE_PROLOGUE
4867 Define this macro if the code for function profiling should come before
4868 the function prologue. Normally, the profiling code comes after.
4872 @subsection Permitting tail calls
4875 @deftypefn {Target Hook} bool TARGET_FUNCTION_OK_FOR_SIBCALL (tree @var{decl}, tree @var{exp})
4876 True if it is ok to do sibling call optimization for the specified
4877 call expression @var{exp}. @var{decl} will be the called function,
4878 or @code{NULL} if this is an indirect call.
4880 It is not uncommon for limitations of calling conventions to prevent
4881 tail calls to functions outside the current unit of translation, or
4882 during PIC compilation. The hook is used to enforce these restrictions,
4883 as the @code{sibcall} md pattern can not fail, or fall over to a
4884 ``normal'' call. The criteria for successful sibling call optimization
4885 may vary greatly between different architectures.
4888 @deftypefn {Target Hook} void TARGET_EXTRA_LIVE_ON_ENTRY (bitmap *@var{regs})
4889 Add any hard registers to @var{regs} that are live on entry to the
4890 function. This hook only needs to be defined to provide registers that
4891 cannot be found by examination of FUNCTION_ARG_REGNO_P, the callee saved
4892 registers, STATIC_CHAIN_INCOMING_REGNUM, STATIC_CHAIN_REGNUM,
4893 TARGET_STRUCT_VALUE_RTX, FRAME_POINTER_REGNUM, EH_USES,
4894 FRAME_POINTER_REGNUM, ARG_POINTER_REGNUM, and the PIC_OFFSET_TABLE_REGNUM.
4897 @node Stack Smashing Protection
4898 @subsection Stack smashing protection
4899 @cindex stack smashing protection
4901 @deftypefn {Target Hook} tree TARGET_STACK_PROTECT_GUARD (void)
4902 This hook returns a @code{DECL} node for the external variable to use
4903 for the stack protection guard. This variable is initialized by the
4904 runtime to some random value and is used to initialize the guard value
4905 that is placed at the top of the local stack frame. The type of this
4906 variable must be @code{ptr_type_node}.
4908 The default version of this hook creates a variable called
4909 @samp{__stack_chk_guard}, which is normally defined in @file{libgcc2.c}.
4912 @deftypefn {Target Hook} tree TARGET_STACK_PROTECT_FAIL (void)
4913 This hook returns a tree expression that alerts the runtime that the
4914 stack protect guard variable has been modified. This expression should
4915 involve a call to a @code{noreturn} function.
4917 The default version of this hook invokes a function called
4918 @samp{__stack_chk_fail}, taking no arguments. This function is
4919 normally defined in @file{libgcc2.c}.
4923 @section Implementing the Varargs Macros
4924 @cindex varargs implementation
4926 GCC comes with an implementation of @code{<varargs.h>} and
4927 @code{<stdarg.h>} that work without change on machines that pass arguments
4928 on the stack. Other machines require their own implementations of
4929 varargs, and the two machine independent header files must have
4930 conditionals to include it.
4932 ISO @code{<stdarg.h>} differs from traditional @code{<varargs.h>} mainly in
4933 the calling convention for @code{va_start}. The traditional
4934 implementation takes just one argument, which is the variable in which
4935 to store the argument pointer. The ISO implementation of
4936 @code{va_start} takes an additional second argument. The user is
4937 supposed to write the last named argument of the function here.
4939 However, @code{va_start} should not use this argument. The way to find
4940 the end of the named arguments is with the built-in functions described
4943 @defmac __builtin_saveregs ()
4944 Use this built-in function to save the argument registers in memory so
4945 that the varargs mechanism can access them. Both ISO and traditional
4946 versions of @code{va_start} must use @code{__builtin_saveregs}, unless
4947 you use @code{TARGET_SETUP_INCOMING_VARARGS} (see below) instead.
4949 On some machines, @code{__builtin_saveregs} is open-coded under the
4950 control of the target hook @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. On
4951 other machines, it calls a routine written in assembler language,
4952 found in @file{libgcc2.c}.
4954 Code generated for the call to @code{__builtin_saveregs} appears at the
4955 beginning of the function, as opposed to where the call to
4956 @code{__builtin_saveregs} is written, regardless of what the code is.
4957 This is because the registers must be saved before the function starts
4958 to use them for its own purposes.
4959 @c i rewrote the first sentence above to fix an overfull hbox. --mew
4963 @defmac __builtin_args_info (@var{category})
4964 Use this built-in function to find the first anonymous arguments in
4967 In general, a machine may have several categories of registers used for
4968 arguments, each for a particular category of data types. (For example,
4969 on some machines, floating-point registers are used for floating-point
4970 arguments while other arguments are passed in the general registers.)
4971 To make non-varargs functions use the proper calling convention, you
4972 have defined the @code{CUMULATIVE_ARGS} data type to record how many
4973 registers in each category have been used so far
4975 @code{__builtin_args_info} accesses the same data structure of type
4976 @code{CUMULATIVE_ARGS} after the ordinary argument layout is finished
4977 with it, with @var{category} specifying which word to access. Thus, the
4978 value indicates the first unused register in a given category.
4980 Normally, you would use @code{__builtin_args_info} in the implementation
4981 of @code{va_start}, accessing each category just once and storing the
4982 value in the @code{va_list} object. This is because @code{va_list} will
4983 have to update the values, and there is no way to alter the
4984 values accessed by @code{__builtin_args_info}.
4987 @defmac __builtin_next_arg (@var{lastarg})
4988 This is the equivalent of @code{__builtin_args_info}, for stack
4989 arguments. It returns the address of the first anonymous stack
4990 argument, as type @code{void *}. If @code{ARGS_GROW_DOWNWARD}, it
4991 returns the address of the location above the first anonymous stack
4992 argument. Use it in @code{va_start} to initialize the pointer for
4993 fetching arguments from the stack. Also use it in @code{va_start} to
4994 verify that the second parameter @var{lastarg} is the last named argument
4995 of the current function.
4998 @defmac __builtin_classify_type (@var{object})
4999 Since each machine has its own conventions for which data types are
5000 passed in which kind of register, your implementation of @code{va_arg}
5001 has to embody these conventions. The easiest way to categorize the
5002 specified data type is to use @code{__builtin_classify_type} together
5003 with @code{sizeof} and @code{__alignof__}.
5005 @code{__builtin_classify_type} ignores the value of @var{object},
5006 considering only its data type. It returns an integer describing what
5007 kind of type that is---integer, floating, pointer, structure, and so on.
5009 The file @file{typeclass.h} defines an enumeration that you can use to
5010 interpret the values of @code{__builtin_classify_type}.
5013 These machine description macros help implement varargs:
5015 @deftypefn {Target Hook} rtx TARGET_EXPAND_BUILTIN_SAVEREGS (void)
5016 If defined, this hook produces the machine-specific code for a call to
5017 @code{__builtin_saveregs}. This code will be moved to the very
5018 beginning of the function, before any parameter access are made. The
5019 return value of this function should be an RTX that contains the value
5020 to use as the return of @code{__builtin_saveregs}.
5023 @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})
5024 This target hook offers an alternative to using
5025 @code{__builtin_saveregs} and defining the hook
5026 @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. Use it to store the anonymous
5027 register arguments into the stack so that all the arguments appear to
5028 have been passed consecutively on the stack. Once this is done, you can
5029 use the standard implementation of varargs that works for machines that
5030 pass all their arguments on the stack.
5032 The argument @var{args_so_far} points to the @code{CUMULATIVE_ARGS} data
5033 structure, containing the values that are obtained after processing the
5034 named arguments. The arguments @var{mode} and @var{type} describe the
5035 last named argument---its machine mode and its data type as a tree node.
5037 The target hook should do two things: first, push onto the stack all the
5038 argument registers @emph{not} used for the named arguments, and second,
5039 store the size of the data thus pushed into the @code{int}-valued
5040 variable pointed to by @var{pretend_args_size}. The value that you
5041 store here will serve as additional offset for setting up the stack
5044 Because you must generate code to push the anonymous arguments at
5045 compile time without knowing their data types,
5046 @code{TARGET_SETUP_INCOMING_VARARGS} is only useful on machines that
5047 have just a single category of argument register and use it uniformly
5050 If the argument @var{second_time} is nonzero, it means that the
5051 arguments of the function are being analyzed for the second time. This
5052 happens for an inline function, which is not actually compiled until the
5053 end of the source file. The hook @code{TARGET_SETUP_INCOMING_VARARGS} should
5054 not generate any instructions in this case.
5057 @deftypefn {Target Hook} bool TARGET_STRICT_ARGUMENT_NAMING (CUMULATIVE_ARGS *@var{ca})
5058 Define this hook to return @code{true} if the location where a function
5059 argument is passed depends on whether or not it is a named argument.
5061 This hook controls how the @var{named} argument to @code{FUNCTION_ARG}
5062 is set for varargs and stdarg functions. If this hook returns
5063 @code{true}, the @var{named} argument is always true for named
5064 arguments, and false for unnamed arguments. If it returns @code{false},
5065 but @code{TARGET_PRETEND_OUTGOING_VARARGS_NAMED} returns @code{true},
5066 then all arguments are treated as named. Otherwise, all named arguments
5067 except the last are treated as named.
5069 You need not define this hook if it always returns zero.
5072 @deftypefn {Target Hook} bool TARGET_PRETEND_OUTGOING_VARARGS_NAMED
5073 If you need to conditionally change ABIs so that one works with
5074 @code{TARGET_SETUP_INCOMING_VARARGS}, but the other works like neither
5075 @code{TARGET_SETUP_INCOMING_VARARGS} nor @code{TARGET_STRICT_ARGUMENT_NAMING} was
5076 defined, then define this hook to return @code{true} if
5077 @code{TARGET_SETUP_INCOMING_VARARGS} is used, @code{false} otherwise.
5078 Otherwise, you should not define this hook.
5082 @section Trampolines for Nested Functions
5083 @cindex trampolines for nested functions
5084 @cindex nested functions, trampolines for
5086 A @dfn{trampoline} is a small piece of code that is created at run time
5087 when the address of a nested function is taken. It normally resides on
5088 the stack, in the stack frame of the containing function. These macros
5089 tell GCC how to generate code to allocate and initialize a
5092 The instructions in the trampoline must do two things: load a constant
5093 address into the static chain register, and jump to the real address of
5094 the nested function. On CISC machines such as the m68k, this requires
5095 two instructions, a move immediate and a jump. Then the two addresses
5096 exist in the trampoline as word-long immediate operands. On RISC
5097 machines, it is often necessary to load each address into a register in
5098 two parts. Then pieces of each address form separate immediate
5101 The code generated to initialize the trampoline must store the variable
5102 parts---the static chain value and the function address---into the
5103 immediate operands of the instructions. On a CISC machine, this is
5104 simply a matter of copying each address to a memory reference at the
5105 proper offset from the start of the trampoline. On a RISC machine, it
5106 may be necessary to take out pieces of the address and store them
5109 @deftypefn {Target Hook} void TARGET_ASM_TRAMPOLINE_TEMPLATE (FILE *@var{f})
5110 This hook is called by @code{assemble_trampoline_template} to output,
5111 on the stream @var{f}, assembler code for a block of data that contains
5112 the constant parts of a trampoline. This code should not include a
5113 label---the label is taken care of automatically.
5115 If you do not define this hook, it means no template is needed
5116 for the target. Do not define this hook on systems where the block move
5117 code to copy the trampoline into place would be larger than the code
5118 to generate it on the spot.
5121 @defmac TRAMPOLINE_SECTION
5122 Return the section into which the trampoline template is to be placed
5123 (@pxref{Sections}). The default value is @code{readonly_data_section}.
5126 @defmac TRAMPOLINE_SIZE
5127 A C expression for the size in bytes of the trampoline, as an integer.
5130 @defmac TRAMPOLINE_ALIGNMENT
5131 Alignment required for trampolines, in bits.
5133 If you don't define this macro, the value of @code{FUNCTION_ALIGNMENT}
5134 is used for aligning trampolines.
5137 @deftypefn {Target Hook} void TARGET_TRAMPOLINE_INIT (rtx @var{m_tramp}, tree @var{fndecl}, rtx @var{static_chain})
5138 This hook is called to initialize a trampoline.
5139 @var{m_tramp} is an RTX for the memory block for the trampoline; @var{fndecl}
5140 is the @code{FUNCTION_DECL} for the nested function; @var{static_chain} is an
5141 RTX for the static chain value that should be passed to the function
5144 If the target defines @code{TARGET_ASM_TRAMPOLINE_TEMPLATE}, then the
5145 first thing this hook should do is emit a block move into @var{m_tramp}
5146 from the memory block returned by @code{assemble_trampoline_template}.
5147 Note that the block move need only cover the constant parts of the
5148 trampoline. If the target isolates the variable parts of the trampoline
5149 to the end, not all @code{TRAMPOLINE_SIZE} bytes need be copied.
5151 If the target requires any other actions, such as flushing caches or
5152 enabling stack execution, these actions should be performed after
5153 initializing the trampoline proper.
5156 @deftypefn {Target Hook} rtx TARGET_TRAMPOLINE_ADJUST_ADDRESS (rtx @var{addr})
5157 This hook should perform any machine-specific adjustment in
5158 the address of the trampoline. Its argument contains the address of the
5159 memory block that was passed to @code{TARGET_TRAMPOLINE_INIT}. In case
5160 the address to be used for a function call should be different from the
5161 address at which the template was stored, the different address should
5162 be returned; otherwise @var{addr} should be returned unchanged.
5163 If this hook is not defined, @var{addr} will be used for function calls.
5166 Implementing trampolines is difficult on many machines because they have
5167 separate instruction and data caches. Writing into a stack location
5168 fails to clear the memory in the instruction cache, so when the program
5169 jumps to that location, it executes the old contents.
5171 Here are two possible solutions. One is to clear the relevant parts of
5172 the instruction cache whenever a trampoline is set up. The other is to
5173 make all trampolines identical, by having them jump to a standard
5174 subroutine. The former technique makes trampoline execution faster; the
5175 latter makes initialization faster.
5177 To clear the instruction cache when a trampoline is initialized, define
5178 the following macro.
5180 @defmac CLEAR_INSN_CACHE (@var{beg}, @var{end})
5181 If defined, expands to a C expression clearing the @emph{instruction
5182 cache} in the specified interval. The definition of this macro would
5183 typically be a series of @code{asm} statements. Both @var{beg} and
5184 @var{end} are both pointer expressions.
5187 The operating system may also require the stack to be made executable
5188 before calling the trampoline. To implement this requirement, define
5189 the following macro.
5191 @defmac ENABLE_EXECUTE_STACK
5192 Define this macro if certain operations must be performed before executing
5193 code located on the stack. The macro should expand to a series of C
5194 file-scope constructs (e.g.@: functions) and provide a unique entry point
5195 named @code{__enable_execute_stack}. The target is responsible for
5196 emitting calls to the entry point in the code, for example from the
5197 @code{TARGET_TRAMPOLINE_INIT} hook.
5200 To use a standard subroutine, define the following macro. In addition,
5201 you must make sure that the instructions in a trampoline fill an entire
5202 cache line with identical instructions, or else ensure that the
5203 beginning of the trampoline code is always aligned at the same point in
5204 its cache line. Look in @file{m68k.h} as a guide.
5206 @defmac TRANSFER_FROM_TRAMPOLINE
5207 Define this macro if trampolines need a special subroutine to do their
5208 work. The macro should expand to a series of @code{asm} statements
5209 which will be compiled with GCC@. They go in a library function named
5210 @code{__transfer_from_trampoline}.
5212 If you need to avoid executing the ordinary prologue code of a compiled
5213 C function when you jump to the subroutine, you can do so by placing a
5214 special label of your own in the assembler code. Use one @code{asm}
5215 statement to generate an assembler label, and another to make the label
5216 global. Then trampolines can use that label to jump directly to your
5217 special assembler code.
5221 @section Implicit Calls to Library Routines
5222 @cindex library subroutine names
5223 @cindex @file{libgcc.a}
5225 @c prevent bad page break with this line
5226 Here is an explanation of implicit calls to library routines.
5228 @defmac DECLARE_LIBRARY_RENAMES
5229 This macro, if defined, should expand to a piece of C code that will get
5230 expanded when compiling functions for libgcc.a. It can be used to
5231 provide alternate names for GCC's internal library functions if there
5232 are ABI-mandated names that the compiler should provide.
5235 @findex init_one_libfunc
5236 @findex set_optab_libfunc
5237 @deftypefn {Target Hook} void TARGET_INIT_LIBFUNCS (void)
5238 This hook should declare additional library routines or rename
5239 existing ones, using the functions @code{set_optab_libfunc} and
5240 @code{init_one_libfunc} defined in @file{optabs.c}.
5241 @code{init_optabs} calls this macro after initializing all the normal
5244 The default is to do nothing. Most ports don't need to define this hook.
5247 @defmac FLOAT_LIB_COMPARE_RETURNS_BOOL (@var{mode}, @var{comparison})
5248 This macro should return @code{true} if the library routine that
5249 implements the floating point comparison operator @var{comparison} in
5250 mode @var{mode} will return a boolean, and @var{false} if it will
5253 GCC's own floating point libraries return tristates from the
5254 comparison operators, so the default returns false always. Most ports
5255 don't need to define this macro.
5258 @defmac TARGET_LIB_INT_CMP_BIASED
5259 This macro should evaluate to @code{true} if the integer comparison
5260 functions (like @code{__cmpdi2}) return 0 to indicate that the first
5261 operand is smaller than the second, 1 to indicate that they are equal,
5262 and 2 to indicate that the first operand is greater than the second.
5263 If this macro evaluates to @code{false} the comparison functions return
5264 @minus{}1, 0, and 1 instead of 0, 1, and 2. If the target uses the routines
5265 in @file{libgcc.a}, you do not need to define this macro.
5268 @cindex US Software GOFAST, floating point emulation library
5269 @cindex floating point emulation library, US Software GOFAST
5270 @cindex GOFAST, floating point emulation library
5271 @findex gofast_maybe_init_libfuncs
5272 @defmac US_SOFTWARE_GOFAST
5273 Define this macro if your system C library uses the US Software GOFAST
5274 library to provide floating point emulation.
5276 In addition to defining this macro, your architecture must set
5277 @code{TARGET_INIT_LIBFUNCS} to @code{gofast_maybe_init_libfuncs}, or
5278 else call that function from its version of that hook. It is defined
5279 in @file{config/gofast.h}, which must be included by your
5280 architecture's @file{@var{cpu}.c} file. See @file{sparc/sparc.c} for
5283 If this macro is defined, the
5284 @code{TARGET_FLOAT_LIB_COMPARE_RETURNS_BOOL} target hook must return
5285 false for @code{SFmode} and @code{DFmode} comparisons.
5288 @cindex @code{EDOM}, implicit usage
5291 The value of @code{EDOM} on the target machine, as a C integer constant
5292 expression. If you don't define this macro, GCC does not attempt to
5293 deposit the value of @code{EDOM} into @code{errno} directly. Look in
5294 @file{/usr/include/errno.h} to find the value of @code{EDOM} on your
5297 If you do not define @code{TARGET_EDOM}, then compiled code reports
5298 domain errors by calling the library function and letting it report the
5299 error. If mathematical functions on your system use @code{matherr} when
5300 there is an error, then you should leave @code{TARGET_EDOM} undefined so
5301 that @code{matherr} is used normally.
5304 @cindex @code{errno}, implicit usage
5305 @defmac GEN_ERRNO_RTX
5306 Define this macro as a C expression to create an rtl expression that
5307 refers to the global ``variable'' @code{errno}. (On certain systems,
5308 @code{errno} may not actually be a variable.) If you don't define this
5309 macro, a reasonable default is used.
5312 @cindex C99 math functions, implicit usage
5313 @defmac TARGET_C99_FUNCTIONS
5314 When this macro is nonzero, GCC will implicitly optimize @code{sin} calls into
5315 @code{sinf} and similarly for other functions defined by C99 standard. The
5316 default is zero because a number of existing systems lack support for these
5317 functions in their runtime so this macro needs to be redefined to one on
5318 systems that do support the C99 runtime.
5321 @cindex sincos math function, implicit usage
5322 @defmac TARGET_HAS_SINCOS
5323 When this macro is nonzero, GCC will implicitly optimize calls to @code{sin}
5324 and @code{cos} with the same argument to a call to @code{sincos}. The
5325 default is zero. The target has to provide the following functions:
5327 void sincos(double x, double *sin, double *cos);
5328 void sincosf(float x, float *sin, float *cos);
5329 void sincosl(long double x, long double *sin, long double *cos);
5333 @defmac NEXT_OBJC_RUNTIME
5334 Define this macro to generate code for Objective-C message sending using
5335 the calling convention of the NeXT system. This calling convention
5336 involves passing the object, the selector and the method arguments all
5337 at once to the method-lookup library function.
5339 The default calling convention passes just the object and the selector
5340 to the lookup function, which returns a pointer to the method.
5343 @node Addressing Modes
5344 @section Addressing Modes
5345 @cindex addressing modes
5347 @c prevent bad page break with this line
5348 This is about addressing modes.
5350 @defmac HAVE_PRE_INCREMENT
5351 @defmacx HAVE_PRE_DECREMENT
5352 @defmacx HAVE_POST_INCREMENT
5353 @defmacx HAVE_POST_DECREMENT
5354 A C expression that is nonzero if the machine supports pre-increment,
5355 pre-decrement, post-increment, or post-decrement addressing respectively.
5358 @defmac HAVE_PRE_MODIFY_DISP
5359 @defmacx HAVE_POST_MODIFY_DISP
5360 A C expression that is nonzero if the machine supports pre- or
5361 post-address side-effect generation involving constants other than
5362 the size of the memory operand.
5365 @defmac HAVE_PRE_MODIFY_REG
5366 @defmacx HAVE_POST_MODIFY_REG
5367 A C expression that is nonzero if the machine supports pre- or
5368 post-address side-effect generation involving a register displacement.
5371 @defmac CONSTANT_ADDRESS_P (@var{x})
5372 A C expression that is 1 if the RTX @var{x} is a constant which
5373 is a valid address. On most machines the default definition of
5374 @code{(CONSTANT_P (@var{x}) && GET_CODE (@var{x}) != CONST_DOUBLE)}
5375 is acceptable, but a few machines are more restrictive as to which
5376 constant addresses are supported.
5379 @defmac CONSTANT_P (@var{x})
5380 @code{CONSTANT_P}, which is defined by target-independent code,
5381 accepts integer-values expressions whose values are not explicitly
5382 known, such as @code{symbol_ref}, @code{label_ref}, and @code{high}
5383 expressions and @code{const} arithmetic expressions, in addition to
5384 @code{const_int} and @code{const_double} expressions.
5387 @defmac MAX_REGS_PER_ADDRESS
5388 A number, the maximum number of registers that can appear in a valid
5389 memory address. Note that it is up to you to specify a value equal to
5390 the maximum number that @code{TARGET_LEGITIMATE_ADDRESS_P} would ever
5394 @deftypefn {Target Hook} TARGET_LEGITIMATE_ADDRESS_P (enum machine_mode @var{mode}, rtx @var{x}, bool @var{strict})
5395 A function that returns whether @var{x} (an RTX) is a legitimate memory
5396 address on the target machine for a memory operand of mode @var{mode}.
5398 Legitimate addresses are defined in two variants: a strict variant and a
5399 non-strict one. The @code{strict} parameter chooses which variant is
5400 desired by the caller.
5402 The strict variant is used in the reload pass. It must be defined so
5403 that any pseudo-register that has not been allocated a hard register is
5404 considered a memory reference. This is because in contexts where some
5405 kind of register is required, a pseudo-register with no hard register
5406 must be rejected. For non-hard registers, the strict variant should look
5407 up the @code{reg_renumber} array; it should then proceed using the hard
5408 register number in the array, or treat the pseudo as a memory reference
5409 if the array holds @code{-1}.
5411 The non-strict variant is used in other passes. It must be defined to
5412 accept all pseudo-registers in every context where some kind of
5413 register is required.
5415 Normally, constant addresses which are the sum of a @code{symbol_ref}
5416 and an integer are stored inside a @code{const} RTX to mark them as
5417 constant. Therefore, there is no need to recognize such sums
5418 specifically as legitimate addresses. Normally you would simply
5419 recognize any @code{const} as legitimate.
5421 Usually @code{PRINT_OPERAND_ADDRESS} is not prepared to handle constant
5422 sums that are not marked with @code{const}. It assumes that a naked
5423 @code{plus} indicates indexing. If so, then you @emph{must} reject such
5424 naked constant sums as illegitimate addresses, so that none of them will
5425 be given to @code{PRINT_OPERAND_ADDRESS}.
5427 @cindex @code{TARGET_ENCODE_SECTION_INFO} and address validation
5428 On some machines, whether a symbolic address is legitimate depends on
5429 the section that the address refers to. On these machines, define the
5430 target hook @code{TARGET_ENCODE_SECTION_INFO} to store the information
5431 into the @code{symbol_ref}, and then check for it here. When you see a
5432 @code{const}, you will have to look inside it to find the
5433 @code{symbol_ref} in order to determine the section. @xref{Assembler
5436 @cindex @code{GO_IF_LEGITIMATE_ADDRESS}
5437 Some ports are still using a deprecated legacy substitute for
5438 this hook, the @code{GO_IF_LEGITIMATE_ADDRESS} macro. This macro
5442 #define GO_IF_LEGITIMATE_ADDRESS (@var{mode}, @var{x}, @var{label})
5446 and should @code{goto @var{label}} if the address @var{x} is a valid
5447 address on the target machine for a memory operand of mode @var{mode}.
5448 Whether the strict or non-strict variants are desired is defined by
5449 the @code{REG_OK_STRICT} macro introduced earlier in this section.
5450 Using the hook is usually simpler because it limits the number of
5451 files that are recompiled when changes are made.
5454 @defmac TARGET_MEM_CONSTRAINT
5455 A single character to be used instead of the default @code{'m'}
5456 character for general memory addresses. This defines the constraint
5457 letter which matches the memory addresses accepted by
5458 @code{TARGET_LEGITIMATE_ADDRESS_P}. Define this macro if you want to
5459 support new address formats in your back end without changing the
5460 semantics of the @code{'m'} constraint. This is necessary in order to
5461 preserve functionality of inline assembly constructs using the
5462 @code{'m'} constraint.
5465 @defmac FIND_BASE_TERM (@var{x})
5466 A C expression to determine the base term of address @var{x},
5467 or to provide a simplified version of @var{x} from which @file{alias.c}
5468 can easily find the base term. This macro is used in only two places:
5469 @code{find_base_value} and @code{find_base_term} in @file{alias.c}.
5471 It is always safe for this macro to not be defined. It exists so
5472 that alias analysis can understand machine-dependent addresses.
5474 The typical use of this macro is to handle addresses containing
5475 a label_ref or symbol_ref within an UNSPEC@.
5478 @deftypefn {Target Hook} rtx TARGET_LEGITIMIZE_ADDRESS (rtx @var{x}, rtx @var{oldx}, enum machine_mode @var{mode})
5479 This hook is given an invalid memory address @var{x} for an
5480 operand of mode @var{mode} and should try to return a valid memory
5483 @findex break_out_memory_refs
5484 @var{x} will always be the result of a call to @code{break_out_memory_refs},
5485 and @var{oldx} will be the operand that was given to that function to produce
5488 The code of the hook should not alter the substructure of
5489 @var{x}. If it transforms @var{x} into a more legitimate form, it
5490 should return the new @var{x}.
5492 It is not necessary for this hook to come up with a legitimate address.
5493 The compiler has standard ways of doing so in all cases. In fact, it
5494 is safe to omit this hook or make it return @var{x} if it cannot find
5495 a valid way to legitimize the address. But often a machine-dependent
5496 strategy can generate better code.
5499 @defmac LEGITIMIZE_RELOAD_ADDRESS (@var{x}, @var{mode}, @var{opnum}, @var{type}, @var{ind_levels}, @var{win})
5500 A C compound statement that attempts to replace @var{x}, which is an address
5501 that needs reloading, with a valid memory address for an operand of mode
5502 @var{mode}. @var{win} will be a C statement label elsewhere in the code.
5503 It is not necessary to define this macro, but it might be useful for
5504 performance reasons.
5506 For example, on the i386, it is sometimes possible to use a single
5507 reload register instead of two by reloading a sum of two pseudo
5508 registers into a register. On the other hand, for number of RISC
5509 processors offsets are limited so that often an intermediate address
5510 needs to be generated in order to address a stack slot. By defining
5511 @code{LEGITIMIZE_RELOAD_ADDRESS} appropriately, the intermediate addresses
5512 generated for adjacent some stack slots can be made identical, and thus
5515 @emph{Note}: This macro should be used with caution. It is necessary
5516 to know something of how reload works in order to effectively use this,
5517 and it is quite easy to produce macros that build in too much knowledge
5518 of reload internals.
5520 @emph{Note}: This macro must be able to reload an address created by a
5521 previous invocation of this macro. If it fails to handle such addresses
5522 then the compiler may generate incorrect code or abort.
5525 The macro definition should use @code{push_reload} to indicate parts that
5526 need reloading; @var{opnum}, @var{type} and @var{ind_levels} are usually
5527 suitable to be passed unaltered to @code{push_reload}.
5529 The code generated by this macro must not alter the substructure of
5530 @var{x}. If it transforms @var{x} into a more legitimate form, it
5531 should assign @var{x} (which will always be a C variable) a new value.
5532 This also applies to parts that you change indirectly by calling
5535 @findex strict_memory_address_p
5536 The macro definition may use @code{strict_memory_address_p} to test if
5537 the address has become legitimate.
5540 If you want to change only a part of @var{x}, one standard way of doing
5541 this is to use @code{copy_rtx}. Note, however, that it unshares only a
5542 single level of rtl. Thus, if the part to be changed is not at the
5543 top level, you'll need to replace first the top level.
5544 It is not necessary for this macro to come up with a legitimate
5545 address; but often a machine-dependent strategy can generate better code.
5548 @defmac GO_IF_MODE_DEPENDENT_ADDRESS (@var{addr}, @var{label})
5549 A C statement or compound statement with a conditional @code{goto
5550 @var{label};} executed if memory address @var{x} (an RTX) can have
5551 different meanings depending on the machine mode of the memory
5552 reference it is used for or if the address is valid for some modes
5555 Autoincrement and autodecrement addresses typically have mode-dependent
5556 effects because the amount of the increment or decrement is the size
5557 of the operand being addressed. Some machines have other mode-dependent
5558 addresses. Many RISC machines have no mode-dependent addresses.
5560 You may assume that @var{addr} is a valid address for the machine.
5563 @defmac LEGITIMATE_CONSTANT_P (@var{x})
5564 A C expression that is nonzero if @var{x} is a legitimate constant for
5565 an immediate operand on the target machine. You can assume that
5566 @var{x} satisfies @code{CONSTANT_P}, so you need not check this. In fact,
5567 @samp{1} is a suitable definition for this macro on machines where
5568 anything @code{CONSTANT_P} is valid.
5571 @deftypefn {Target Hook} rtx TARGET_DELEGITIMIZE_ADDRESS (rtx @var{x})
5572 This hook is used to undo the possibly obfuscating effects of the
5573 @code{LEGITIMIZE_ADDRESS} and @code{LEGITIMIZE_RELOAD_ADDRESS} target
5574 macros. Some backend implementations of these macros wrap symbol
5575 references inside an @code{UNSPEC} rtx to represent PIC or similar
5576 addressing modes. This target hook allows GCC's optimizers to understand
5577 the semantics of these opaque @code{UNSPEC}s by converting them back
5578 into their original form.
5581 @deftypefn {Target Hook} bool TARGET_CANNOT_FORCE_CONST_MEM (rtx @var{x})
5582 This hook should return true if @var{x} is of a form that cannot (or
5583 should not) be spilled to the constant pool. The default version of
5584 this hook returns false.
5586 The primary reason to define this hook is to prevent reload from
5587 deciding that a non-legitimate constant would be better reloaded
5588 from the constant pool instead of spilling and reloading a register
5589 holding the constant. This restriction is often true of addresses
5590 of TLS symbols for various targets.
5593 @deftypefn {Target Hook} bool TARGET_USE_BLOCKS_FOR_CONSTANT_P (enum machine_mode @var{mode}, rtx @var{x})
5594 This hook should return true if pool entries for constant @var{x} can
5595 be placed in an @code{object_block} structure. @var{mode} is the mode
5598 The default version returns false for all constants.
5601 @deftypefn {Target Hook} tree TARGET_BUILTIN_RECIPROCAL (enum tree_code @var{fn}, bool @var{tm_fn}, bool @var{sqrt})
5602 This hook should return the DECL of a function that implements reciprocal of
5603 the builtin function with builtin function code @var{fn}, or
5604 @code{NULL_TREE} if such a function is not available. @var{tm_fn} is true
5605 when @var{fn} is a code of a machine-dependent builtin function. When
5606 @var{sqrt} is true, additional optimizations that apply only to the reciprocal
5607 of a square root function are performed, and only reciprocals of @code{sqrt}
5611 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_MASK_FOR_LOAD (void)
5612 This hook should return the DECL of a function @var{f} that given an
5613 address @var{addr} as an argument returns a mask @var{m} that can be
5614 used to extract from two vectors the relevant data that resides in
5615 @var{addr} in case @var{addr} is not properly aligned.
5617 The autovectorizer, when vectorizing a load operation from an address
5618 @var{addr} that may be unaligned, will generate two vector loads from
5619 the two aligned addresses around @var{addr}. It then generates a
5620 @code{REALIGN_LOAD} operation to extract the relevant data from the
5621 two loaded vectors. The first two arguments to @code{REALIGN_LOAD},
5622 @var{v1} and @var{v2}, are the two vectors, each of size @var{VS}, and
5623 the third argument, @var{OFF}, defines how the data will be extracted
5624 from these two vectors: if @var{OFF} is 0, then the returned vector is
5625 @var{v2}; otherwise, the returned vector is composed from the last
5626 @var{VS}-@var{OFF} elements of @var{v1} concatenated to the first
5627 @var{OFF} elements of @var{v2}.
5629 If this hook is defined, the autovectorizer will generate a call
5630 to @var{f} (using the DECL tree that this hook returns) and will
5631 use the return value of @var{f} as the argument @var{OFF} to
5632 @code{REALIGN_LOAD}. Therefore, the mask @var{m} returned by @var{f}
5633 should comply with the semantics expected by @code{REALIGN_LOAD}
5635 If this hook is not defined, then @var{addr} will be used as
5636 the argument @var{OFF} to @code{REALIGN_LOAD}, in which case the low
5637 log2(@var{VS}) @minus{} 1 bits of @var{addr} will be considered.
5640 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_EVEN (tree @var{x})
5641 This hook should return the DECL of a function @var{f} that implements
5642 widening multiplication of the even elements of two input vectors of type @var{x}.
5644 If this hook is defined, the autovectorizer will use it along with the
5645 @code{TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_ODD} target hook when vectorizing
5646 widening multiplication in cases that the order of the results does not have to be
5647 preserved (e.g.@: used only by a reduction computation). Otherwise, the
5648 @code{widen_mult_hi/lo} idioms will be used.
5651 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_ODD (tree @var{x})
5652 This hook should return the DECL of a function @var{f} that implements
5653 widening multiplication of the odd elements of two input vectors of type @var{x}.
5655 If this hook is defined, the autovectorizer will use it along with the
5656 @code{TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_EVEN} target hook when vectorizing
5657 widening multiplication in cases that the order of the results does not have to be
5658 preserved (e.g.@: used only by a reduction computation). Otherwise, the
5659 @code{widen_mult_hi/lo} idioms will be used.
5662 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_CONVERSION (enum tree_code @var{code}, tree @var{type})
5663 This hook should return the DECL of a function that implements conversion of the
5664 input vector of type @var{type}.
5665 If @var{type} is an integral type, the result of the conversion is a vector of
5666 floating-point type of the same size.
5667 If @var{type} is a floating-point type, the result of the conversion is a vector
5668 of integral type of the same size.
5669 @var{code} specifies how the conversion is to be applied
5670 (truncation, rounding, etc.).
5672 If this hook is defined, the autovectorizer will use the
5673 @code{TARGET_VECTORIZE_BUILTIN_CONVERSION} target hook when vectorizing
5674 conversion. Otherwise, it will return @code{NULL_TREE}.
5677 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION (enum built_in_function @var{code}, tree @var{vec_type_out}, tree @var{vec_type_in})
5678 This hook should return the decl of a function that implements the vectorized
5679 variant of the builtin function with builtin function code @var{code} or
5680 @code{NULL_TREE} if such a function is not available. The return type of
5681 the vectorized function shall be of vector type @var{vec_type_out} and the
5682 argument types should be @var{vec_type_in}.
5685 @deftypefn {Target Hook} bool TARGET_SUPPORT_VECTOR_MISALIGNMENT (enum machine_mode @var{mode}, tree @var{type}, int @var{misalignment}, bool @var{is_packed})
5686 This hook should return true if the target supports misaligned vector
5687 store/load of a specific factor denoted in the @var{misalignment}
5688 parameter. The vector store/load should be of machine mode @var{mode} and
5689 the elements in the vectors should be of type @var{type}. @var{is_packed}
5690 parameter is true if the memory access is defined in a packed struct.
5693 @node Anchored Addresses
5694 @section Anchored Addresses
5695 @cindex anchored addresses
5696 @cindex @option{-fsection-anchors}
5698 GCC usually addresses every static object as a separate entity.
5699 For example, if we have:
5703 int foo (void) @{ return a + b + c; @}
5706 the code for @code{foo} will usually calculate three separate symbolic
5707 addresses: those of @code{a}, @code{b} and @code{c}. On some targets,
5708 it would be better to calculate just one symbolic address and access
5709 the three variables relative to it. The equivalent pseudocode would
5715 register int *xr = &x;
5716 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
5720 (which isn't valid C). We refer to shared addresses like @code{x} as
5721 ``section anchors''. Their use is controlled by @option{-fsection-anchors}.
5723 The hooks below describe the target properties that GCC needs to know
5724 in order to make effective use of section anchors. It won't use
5725 section anchors at all unless either @code{TARGET_MIN_ANCHOR_OFFSET}
5726 or @code{TARGET_MAX_ANCHOR_OFFSET} is set to a nonzero value.
5728 @deftypevr {Target Hook} HOST_WIDE_INT TARGET_MIN_ANCHOR_OFFSET
5729 The minimum offset that should be applied to a section anchor.
5730 On most targets, it should be the smallest offset that can be
5731 applied to a base register while still giving a legitimate address
5732 for every mode. The default value is 0.
5735 @deftypevr {Target Hook} HOST_WIDE_INT TARGET_MAX_ANCHOR_OFFSET
5736 Like @code{TARGET_MIN_ANCHOR_OFFSET}, but the maximum (inclusive)
5737 offset that should be applied to section anchors. The default
5741 @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_ANCHOR (rtx @var{x})
5742 Write the assembly code to define section anchor @var{x}, which is a
5743 @code{SYMBOL_REF} for which @samp{SYMBOL_REF_ANCHOR_P (@var{x})} is true.
5744 The hook is called with the assembly output position set to the beginning
5745 of @code{SYMBOL_REF_BLOCK (@var{x})}.
5747 If @code{ASM_OUTPUT_DEF} is available, the hook's default definition uses
5748 it to define the symbol as @samp{. + SYMBOL_REF_BLOCK_OFFSET (@var{x})}.
5749 If @code{ASM_OUTPUT_DEF} is not available, the hook's default definition
5750 is @code{NULL}, which disables the use of section anchors altogether.
5753 @deftypefn {Target Hook} bool TARGET_USE_ANCHORS_FOR_SYMBOL_P (rtx @var{x})
5754 Return true if GCC should attempt to use anchors to access @code{SYMBOL_REF}
5755 @var{x}. You can assume @samp{SYMBOL_REF_HAS_BLOCK_INFO_P (@var{x})} and
5756 @samp{!SYMBOL_REF_ANCHOR_P (@var{x})}.
5758 The default version is correct for most targets, but you might need to
5759 intercept this hook to handle things like target-specific attributes
5760 or target-specific sections.
5763 @node Condition Code
5764 @section Condition Code Status
5765 @cindex condition code status
5767 The macros in this section can be split in two families, according to the
5768 two ways of representing condition codes in GCC.
5770 The first representation is the so called @code{(cc0)} representation
5771 (@pxref{Jump Patterns}), where all instructions can have an implicit
5772 clobber of the condition codes. The second is the condition code
5773 register representation, which provides better schedulability for
5774 architectures that do have a condition code register, but on which
5775 most instructions do not affect it. The latter category includes
5778 The implicit clobbering poses a strong restriction on the placement of
5779 the definition and use of the condition code, which need to be in adjacent
5780 insns for machines using @code{(cc0)}. This can prevent important
5781 optimizations on some machines. For example, on the IBM RS/6000, there
5782 is a delay for taken branches unless the condition code register is set
5783 three instructions earlier than the conditional branch. The instruction
5784 scheduler cannot perform this optimization if it is not permitted to
5785 separate the definition and use of the condition code register.
5787 For this reason, it is possible and suggested to use a register to
5788 represent the condition code for new ports. If there is a specific
5789 condition code register in the machine, use a hard register. If the
5790 condition code or comparison result can be placed in any general register,
5791 or if there are multiple condition registers, use a pseudo register.
5792 Registers used to store the condition code value will usually have a mode
5793 that is in class @code{MODE_CC}.
5795 Alternatively, you can use @code{BImode} if the comparison operator is
5796 specified already in the compare instruction. In this case, you are not
5797 interested in most macros in this section.
5800 * CC0 Condition Codes:: Old style representation of condition codes.
5801 * MODE_CC Condition Codes:: Modern representation of condition codes.
5802 * Cond. Exec. Macros:: Macros to control conditional execution.
5805 @node CC0 Condition Codes
5806 @subsection Representation of condition codes using @code{(cc0)}
5810 The file @file{conditions.h} defines a variable @code{cc_status} to
5811 describe how the condition code was computed (in case the interpretation of
5812 the condition code depends on the instruction that it was set by). This
5813 variable contains the RTL expressions on which the condition code is
5814 currently based, and several standard flags.
5816 Sometimes additional machine-specific flags must be defined in the machine
5817 description header file. It can also add additional machine-specific
5818 information by defining @code{CC_STATUS_MDEP}.
5820 @defmac CC_STATUS_MDEP
5821 C code for a data type which is used for declaring the @code{mdep}
5822 component of @code{cc_status}. It defaults to @code{int}.
5824 This macro is not used on machines that do not use @code{cc0}.
5827 @defmac CC_STATUS_MDEP_INIT
5828 A C expression to initialize the @code{mdep} field to ``empty''.
5829 The default definition does nothing, since most machines don't use
5830 the field anyway. If you want to use the field, you should probably
5831 define this macro to initialize it.
5833 This macro is not used on machines that do not use @code{cc0}.
5836 @defmac NOTICE_UPDATE_CC (@var{exp}, @var{insn})
5837 A C compound statement to set the components of @code{cc_status}
5838 appropriately for an insn @var{insn} whose body is @var{exp}. It is
5839 this macro's responsibility to recognize insns that set the condition
5840 code as a byproduct of other activity as well as those that explicitly
5843 This macro is not used on machines that do not use @code{cc0}.
5845 If there are insns that do not set the condition code but do alter
5846 other machine registers, this macro must check to see whether they
5847 invalidate the expressions that the condition code is recorded as
5848 reflecting. For example, on the 68000, insns that store in address
5849 registers do not set the condition code, which means that usually
5850 @code{NOTICE_UPDATE_CC} can leave @code{cc_status} unaltered for such
5851 insns. But suppose that the previous insn set the condition code
5852 based on location @samp{a4@@(102)} and the current insn stores a new
5853 value in @samp{a4}. Although the condition code is not changed by
5854 this, it will no longer be true that it reflects the contents of
5855 @samp{a4@@(102)}. Therefore, @code{NOTICE_UPDATE_CC} must alter
5856 @code{cc_status} in this case to say that nothing is known about the
5857 condition code value.
5859 The definition of @code{NOTICE_UPDATE_CC} must be prepared to deal
5860 with the results of peephole optimization: insns whose patterns are
5861 @code{parallel} RTXs containing various @code{reg}, @code{mem} or
5862 constants which are just the operands. The RTL structure of these
5863 insns is not sufficient to indicate what the insns actually do. What
5864 @code{NOTICE_UPDATE_CC} should do when it sees one is just to run
5865 @code{CC_STATUS_INIT}.
5867 A possible definition of @code{NOTICE_UPDATE_CC} is to call a function
5868 that looks at an attribute (@pxref{Insn Attributes}) named, for example,
5869 @samp{cc}. This avoids having detailed information about patterns in
5870 two places, the @file{md} file and in @code{NOTICE_UPDATE_CC}.
5873 @node MODE_CC Condition Codes
5874 @subsection Representation of condition codes using registers
5878 @defmac SELECT_CC_MODE (@var{op}, @var{x}, @var{y})
5879 On many machines, the condition code may be produced by other instructions
5880 than compares, for example the branch can use directly the condition
5881 code set by a subtract instruction. However, on some machines
5882 when the condition code is set this way some bits (such as the overflow
5883 bit) are not set in the same way as a test instruction, so that a different
5884 branch instruction must be used for some conditional branches. When
5885 this happens, use the machine mode of the condition code register to
5886 record different formats of the condition code register. Modes can
5887 also be used to record which compare instruction (e.g. a signed or an
5888 unsigned comparison) produced the condition codes.
5890 If other modes than @code{CCmode} are required, add them to
5891 @file{@var{machine}-modes.def} and define @code{SELECT_CC_MODE} to choose
5892 a mode given an operand of a compare. This is needed because the modes
5893 have to be chosen not only during RTL generation but also, for example,
5894 by instruction combination. The result of @code{SELECT_CC_MODE} should
5895 be consistent with the mode used in the patterns; for example to support
5896 the case of the add on the SPARC discussed above, we have the pattern
5900 [(set (reg:CC_NOOV 0)
5902 (plus:SI (match_operand:SI 0 "register_operand" "%r")
5903 (match_operand:SI 1 "arith_operand" "rI"))
5910 together with a @code{SELECT_CC_MODE} that returns @code{CC_NOOVmode}
5911 for comparisons whose argument is a @code{plus}:
5914 #define SELECT_CC_MODE(OP,X,Y) \
5915 (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \
5916 ? ((OP == EQ || OP == NE) ? CCFPmode : CCFPEmode) \
5917 : ((GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \
5918 || GET_CODE (X) == NEG) \
5919 ? CC_NOOVmode : CCmode))
5922 Another reason to use modes is to retain information on which operands
5923 were used by the comparison; see @code{REVERSIBLE_CC_MODE} later in
5926 You should define this macro if and only if you define extra CC modes
5927 in @file{@var{machine}-modes.def}.
5930 @defmac CANONICALIZE_COMPARISON (@var{code}, @var{op0}, @var{op1})
5931 On some machines not all possible comparisons are defined, but you can
5932 convert an invalid comparison into a valid one. For example, the Alpha
5933 does not have a @code{GT} comparison, but you can use an @code{LT}
5934 comparison instead and swap the order of the operands.
5936 On such machines, define this macro to be a C statement to do any
5937 required conversions. @var{code} is the initial comparison code
5938 and @var{op0} and @var{op1} are the left and right operands of the
5939 comparison, respectively. You should modify @var{code}, @var{op0}, and
5940 @var{op1} as required.
5942 GCC will not assume that the comparison resulting from this macro is
5943 valid but will see if the resulting insn matches a pattern in the
5946 You need not define this macro if it would never change the comparison
5950 @defmac REVERSIBLE_CC_MODE (@var{mode})
5951 A C expression whose value is one if it is always safe to reverse a
5952 comparison whose mode is @var{mode}. If @code{SELECT_CC_MODE}
5953 can ever return @var{mode} for a floating-point inequality comparison,
5954 then @code{REVERSIBLE_CC_MODE (@var{mode})} must be zero.
5956 You need not define this macro if it would always returns zero or if the
5957 floating-point format is anything other than @code{IEEE_FLOAT_FORMAT}.
5958 For example, here is the definition used on the SPARC, where floating-point
5959 inequality comparisons are always given @code{CCFPEmode}:
5962 #define REVERSIBLE_CC_MODE(MODE) ((MODE) != CCFPEmode)
5966 @defmac REVERSE_CONDITION (@var{code}, @var{mode})
5967 A C expression whose value is reversed condition code of the @var{code} for
5968 comparison done in CC_MODE @var{mode}. The macro is used only in case
5969 @code{REVERSIBLE_CC_MODE (@var{mode})} is nonzero. Define this macro in case
5970 machine has some non-standard way how to reverse certain conditionals. For
5971 instance in case all floating point conditions are non-trapping, compiler may
5972 freely convert unordered compares to ordered one. Then definition may look
5976 #define REVERSE_CONDITION(CODE, MODE) \
5977 ((MODE) != CCFPmode ? reverse_condition (CODE) \
5978 : reverse_condition_maybe_unordered (CODE))
5982 @deftypefn {Target Hook} bool TARGET_FIXED_CONDITION_CODE_REGS (unsigned int *, unsigned int *)
5983 On targets which do not use @code{(cc0)}, and which use a hard
5984 register rather than a pseudo-register to hold condition codes, the
5985 regular CSE passes are often not able to identify cases in which the
5986 hard register is set to a common value. Use this hook to enable a
5987 small pass which optimizes such cases. This hook should return true
5988 to enable this pass, and it should set the integers to which its
5989 arguments point to the hard register numbers used for condition codes.
5990 When there is only one such register, as is true on most systems, the
5991 integer pointed to by the second argument should be set to
5992 @code{INVALID_REGNUM}.
5994 The default version of this hook returns false.
5997 @deftypefn {Target Hook} enum machine_mode TARGET_CC_MODES_COMPATIBLE (enum machine_mode, enum machine_mode)
5998 On targets which use multiple condition code modes in class
5999 @code{MODE_CC}, it is sometimes the case that a comparison can be
6000 validly done in more than one mode. On such a system, define this
6001 target hook to take two mode arguments and to return a mode in which
6002 both comparisons may be validly done. If there is no such mode,
6003 return @code{VOIDmode}.
6005 The default version of this hook checks whether the modes are the
6006 same. If they are, it returns that mode. If they are different, it
6007 returns @code{VOIDmode}.
6010 @node Cond. Exec. Macros
6011 @subsection Macros to control conditional execution
6012 @findex conditional execution
6015 There is one macro that may need to be defined for targets
6016 supporting conditional execution, independent of how they
6017 represent conditional branches.
6019 @defmac REVERSE_CONDEXEC_PREDICATES_P (@var{op1}, @var{op2})
6020 A C expression that returns true if the conditional execution predicate
6021 @var{op1}, a comparison operation, is the inverse of @var{op2} and vice
6022 versa. Define this to return 0 if the target has conditional execution
6023 predicates that cannot be reversed safely. There is no need to validate
6024 that the arguments of op1 and op2 are the same, this is done separately.
6025 If no expansion is specified, this macro is defined as follows:
6028 #define REVERSE_CONDEXEC_PREDICATES_P (x, y) \
6029 (GET_CODE ((x)) == reversed_comparison_code ((y), NULL))
6034 @section Describing Relative Costs of Operations
6035 @cindex costs of instructions
6036 @cindex relative costs
6037 @cindex speed of instructions
6039 These macros let you describe the relative speed of various operations
6040 on the target machine.
6042 @defmac REGISTER_MOVE_COST (@var{mode}, @var{from}, @var{to})
6043 A C expression for the cost of moving data of mode @var{mode} from a
6044 register in class @var{from} to one in class @var{to}. The classes are
6045 expressed using the enumeration values such as @code{GENERAL_REGS}. A
6046 value of 2 is the default; other values are interpreted relative to
6049 It is not required that the cost always equal 2 when @var{from} is the
6050 same as @var{to}; on some machines it is expensive to move between
6051 registers if they are not general registers.
6053 If reload sees an insn consisting of a single @code{set} between two
6054 hard registers, and if @code{REGISTER_MOVE_COST} applied to their
6055 classes returns a value of 2, reload does not check to ensure that the
6056 constraints of the insn are met. Setting a cost of other than 2 will
6057 allow reload to verify that the constraints are met. You should do this
6058 if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
6061 @defmac MEMORY_MOVE_COST (@var{mode}, @var{class}, @var{in})
6062 A C expression for the cost of moving data of mode @var{mode} between a
6063 register of class @var{class} and memory; @var{in} is zero if the value
6064 is to be written to memory, nonzero if it is to be read in. This cost
6065 is relative to those in @code{REGISTER_MOVE_COST}. If moving between
6066 registers and memory is more expensive than between two registers, you
6067 should define this macro to express the relative cost.
6069 If you do not define this macro, GCC uses a default cost of 4 plus
6070 the cost of copying via a secondary reload register, if one is
6071 needed. If your machine requires a secondary reload register to copy
6072 between memory and a register of @var{class} but the reload mechanism is
6073 more complex than copying via an intermediate, define this macro to
6074 reflect the actual cost of the move.
6076 GCC defines the function @code{memory_move_secondary_cost} if
6077 secondary reloads are needed. It computes the costs due to copying via
6078 a secondary register. If your machine copies from memory using a
6079 secondary register in the conventional way but the default base value of
6080 4 is not correct for your machine, define this macro to add some other
6081 value to the result of that function. The arguments to that function
6082 are the same as to this macro.
6085 @defmac BRANCH_COST (@var{speed_p}, @var{predictable_p})
6086 A C expression for the cost of a branch instruction. A value of 1 is the
6087 default; other values are interpreted relative to that. Parameter @var{speed_p}
6088 is true when the branch in question should be optimized for speed. When
6089 it is false, @code{BRANCH_COST} should be returning value optimal for code size
6090 rather then performance considerations. @var{predictable_p} is true for well
6091 predictable branches. On many architectures the @code{BRANCH_COST} can be
6095 Here are additional macros which do not specify precise relative costs,
6096 but only that certain actions are more expensive than GCC would
6099 @defmac SLOW_BYTE_ACCESS
6100 Define this macro as a C expression which is nonzero if accessing less
6101 than a word of memory (i.e.@: a @code{char} or a @code{short}) is no
6102 faster than accessing a word of memory, i.e., if such access
6103 require more than one instruction or if there is no difference in cost
6104 between byte and (aligned) word loads.
6106 When this macro is not defined, the compiler will access a field by
6107 finding the smallest containing object; when it is defined, a fullword
6108 load will be used if alignment permits. Unless bytes accesses are
6109 faster than word accesses, using word accesses is preferable since it
6110 may eliminate subsequent memory access if subsequent accesses occur to
6111 other fields in the same word of the structure, but to different bytes.
6114 @defmac SLOW_UNALIGNED_ACCESS (@var{mode}, @var{alignment})
6115 Define this macro to be the value 1 if memory accesses described by the
6116 @var{mode} and @var{alignment} parameters have a cost many times greater
6117 than aligned accesses, for example if they are emulated in a trap
6120 When this macro is nonzero, the compiler will act as if
6121 @code{STRICT_ALIGNMENT} were nonzero when generating code for block
6122 moves. This can cause significantly more instructions to be produced.
6123 Therefore, do not set this macro nonzero if unaligned accesses only add a
6124 cycle or two to the time for a memory access.
6126 If the value of this macro is always zero, it need not be defined. If
6127 this macro is defined, it should produce a nonzero value when
6128 @code{STRICT_ALIGNMENT} is nonzero.
6131 @defmac MOVE_RATIO (@var{speed})
6132 The threshold of number of scalar memory-to-memory move insns, @emph{below}
6133 which a sequence of insns should be generated instead of a
6134 string move insn or a library call. Increasing the value will always
6135 make code faster, but eventually incurs high cost in increased code size.
6137 Note that on machines where the corresponding move insn is a
6138 @code{define_expand} that emits a sequence of insns, this macro counts
6139 the number of such sequences.
6141 The parameter @var{speed} is true if the code is currently being
6142 optimized for speed rather than size.
6144 If you don't define this, a reasonable default is used.
6147 @defmac MOVE_BY_PIECES_P (@var{size}, @var{alignment})
6148 A C expression used to determine whether @code{move_by_pieces} will be used to
6149 copy a chunk of memory, or whether some other block move mechanism
6150 will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6151 than @code{MOVE_RATIO}.
6154 @defmac MOVE_MAX_PIECES
6155 A C expression used by @code{move_by_pieces} to determine the largest unit
6156 a load or store used to copy memory is. Defaults to @code{MOVE_MAX}.
6159 @defmac CLEAR_RATIO (@var{speed})
6160 The threshold of number of scalar move insns, @emph{below} which a sequence
6161 of insns should be generated to clear memory instead of a string clear insn
6162 or a library call. Increasing the value will always make code faster, but
6163 eventually incurs high cost in increased code size.
6165 The parameter @var{speed} is true if the code is currently being
6166 optimized for speed rather than size.
6168 If you don't define this, a reasonable default is used.
6171 @defmac CLEAR_BY_PIECES_P (@var{size}, @var{alignment})
6172 A C expression used to determine whether @code{clear_by_pieces} will be used
6173 to clear a chunk of memory, or whether some other block clear mechanism
6174 will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6175 than @code{CLEAR_RATIO}.
6178 @defmac SET_RATIO (@var{speed})
6179 The threshold of number of scalar move insns, @emph{below} which a sequence
6180 of insns should be generated to set memory to a constant value, instead of
6181 a block set insn or a library call.
6182 Increasing the value will always make code faster, but
6183 eventually incurs high cost in increased code size.
6185 The parameter @var{speed} is true if the code is currently being
6186 optimized for speed rather than size.
6188 If you don't define this, it defaults to the value of @code{MOVE_RATIO}.
6191 @defmac SET_BY_PIECES_P (@var{size}, @var{alignment})
6192 A C expression used to determine whether @code{store_by_pieces} will be
6193 used to set a chunk of memory to a constant value, or whether some
6194 other mechanism will be used. Used by @code{__builtin_memset} when
6195 storing values other than constant zero.
6196 Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6197 than @code{SET_RATIO}.
6200 @defmac STORE_BY_PIECES_P (@var{size}, @var{alignment})
6201 A C expression used to determine whether @code{store_by_pieces} will be
6202 used to set a chunk of memory to a constant string value, or whether some
6203 other mechanism will be used. Used by @code{__builtin_strcpy} when
6204 called with a constant source string.
6205 Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6206 than @code{MOVE_RATIO}.
6209 @defmac USE_LOAD_POST_INCREMENT (@var{mode})
6210 A C expression used to determine whether a load postincrement is a good
6211 thing to use for a given mode. Defaults to the value of
6212 @code{HAVE_POST_INCREMENT}.
6215 @defmac USE_LOAD_POST_DECREMENT (@var{mode})
6216 A C expression used to determine whether a load postdecrement is a good
6217 thing to use for a given mode. Defaults to the value of
6218 @code{HAVE_POST_DECREMENT}.
6221 @defmac USE_LOAD_PRE_INCREMENT (@var{mode})
6222 A C expression used to determine whether a load preincrement is a good
6223 thing to use for a given mode. Defaults to the value of
6224 @code{HAVE_PRE_INCREMENT}.
6227 @defmac USE_LOAD_PRE_DECREMENT (@var{mode})
6228 A C expression used to determine whether a load predecrement is a good
6229 thing to use for a given mode. Defaults to the value of
6230 @code{HAVE_PRE_DECREMENT}.
6233 @defmac USE_STORE_POST_INCREMENT (@var{mode})
6234 A C expression used to determine whether a store postincrement is a good
6235 thing to use for a given mode. Defaults to the value of
6236 @code{HAVE_POST_INCREMENT}.
6239 @defmac USE_STORE_POST_DECREMENT (@var{mode})
6240 A C expression used to determine whether a store postdecrement is a good
6241 thing to use for a given mode. Defaults to the value of
6242 @code{HAVE_POST_DECREMENT}.
6245 @defmac USE_STORE_PRE_INCREMENT (@var{mode})
6246 This macro is used to determine whether a store preincrement is a good
6247 thing to use for a given mode. Defaults to the value of
6248 @code{HAVE_PRE_INCREMENT}.
6251 @defmac USE_STORE_PRE_DECREMENT (@var{mode})
6252 This macro is used to determine whether a store predecrement is a good
6253 thing to use for a given mode. Defaults to the value of
6254 @code{HAVE_PRE_DECREMENT}.
6257 @defmac NO_FUNCTION_CSE
6258 Define this macro if it is as good or better to call a constant
6259 function address than to call an address kept in a register.
6262 @defmac RANGE_TEST_NON_SHORT_CIRCUIT
6263 Define this macro if a non-short-circuit operation produced by
6264 @samp{fold_range_test ()} is optimal. This macro defaults to true if
6265 @code{BRANCH_COST} is greater than or equal to the value 2.
6268 @deftypefn {Target Hook} bool TARGET_RTX_COSTS (rtx @var{x}, int @var{code}, int @var{outer_code}, int *@var{total}, bool @var{speed})
6269 This target hook describes the relative costs of RTL expressions.
6271 The cost may depend on the precise form of the expression, which is
6272 available for examination in @var{x}, and the rtx code of the expression
6273 in which it is contained, found in @var{outer_code}. @var{code} is the
6274 expression code---redundant, since it can be obtained with
6275 @code{GET_CODE (@var{x})}.
6277 In implementing this hook, you can use the construct
6278 @code{COSTS_N_INSNS (@var{n})} to specify a cost equal to @var{n} fast
6281 On entry to the hook, @code{*@var{total}} contains a default estimate
6282 for the cost of the expression. The hook should modify this value as
6283 necessary. Traditionally, the default costs are @code{COSTS_N_INSNS (5)}
6284 for multiplications, @code{COSTS_N_INSNS (7)} for division and modulus
6285 operations, and @code{COSTS_N_INSNS (1)} for all other operations.
6287 When optimizing for code size, i.e.@: when @code{speed} is
6288 false, this target hook should be used to estimate the relative
6289 size cost of an expression, again relative to @code{COSTS_N_INSNS}.
6291 The hook returns true when all subexpressions of @var{x} have been
6292 processed, and false when @code{rtx_cost} should recurse.
6295 @deftypefn {Target Hook} int TARGET_ADDRESS_COST (rtx @var{address}, bool @var{speed})
6296 This hook computes the cost of an addressing mode that contains
6297 @var{address}. If not defined, the cost is computed from
6298 the @var{address} expression and the @code{TARGET_RTX_COST} hook.
6300 For most CISC machines, the default cost is a good approximation of the
6301 true cost of the addressing mode. However, on RISC machines, all
6302 instructions normally have the same length and execution time. Hence
6303 all addresses will have equal costs.
6305 In cases where more than one form of an address is known, the form with
6306 the lowest cost will be used. If multiple forms have the same, lowest,
6307 cost, the one that is the most complex will be used.
6309 For example, suppose an address that is equal to the sum of a register
6310 and a constant is used twice in the same basic block. When this macro
6311 is not defined, the address will be computed in a register and memory
6312 references will be indirect through that register. On machines where
6313 the cost of the addressing mode containing the sum is no higher than
6314 that of a simple indirect reference, this will produce an additional
6315 instruction and possibly require an additional register. Proper
6316 specification of this macro eliminates this overhead for such machines.
6318 This hook is never called with an invalid address.
6320 On machines where an address involving more than one register is as
6321 cheap as an address computation involving only one register, defining
6322 @code{TARGET_ADDRESS_COST} to reflect this can cause two registers to
6323 be live over a region of code where only one would have been if
6324 @code{TARGET_ADDRESS_COST} were not defined in that manner. This effect
6325 should be considered in the definition of this macro. Equivalent costs
6326 should probably only be given to addresses with different numbers of
6327 registers on machines with lots of registers.
6331 @section Adjusting the Instruction Scheduler
6333 The instruction scheduler may need a fair amount of machine-specific
6334 adjustment in order to produce good code. GCC provides several target
6335 hooks for this purpose. It is usually enough to define just a few of
6336 them: try the first ones in this list first.
6338 @deftypefn {Target Hook} int TARGET_SCHED_ISSUE_RATE (void)
6339 This hook returns the maximum number of instructions that can ever
6340 issue at the same time on the target machine. The default is one.
6341 Although the insn scheduler can define itself the possibility of issue
6342 an insn on the same cycle, the value can serve as an additional
6343 constraint to issue insns on the same simulated processor cycle (see
6344 hooks @samp{TARGET_SCHED_REORDER} and @samp{TARGET_SCHED_REORDER2}).
6345 This value must be constant over the entire compilation. If you need
6346 it to vary depending on what the instructions are, you must use
6347 @samp{TARGET_SCHED_VARIABLE_ISSUE}.
6350 @deftypefn {Target Hook} int TARGET_SCHED_VARIABLE_ISSUE (FILE *@var{file}, int @var{verbose}, rtx @var{insn}, int @var{more})
6351 This hook is executed by the scheduler after it has scheduled an insn
6352 from the ready list. It should return the number of insns which can
6353 still be issued in the current cycle. The default is
6354 @samp{@w{@var{more} - 1}} for insns other than @code{CLOBBER} and
6355 @code{USE}, which normally are not counted against the issue rate.
6356 You should define this hook if some insns take more machine resources
6357 than others, so that fewer insns can follow them in the same cycle.
6358 @var{file} is either a null pointer, or a stdio stream to write any
6359 debug output to. @var{verbose} is the verbose level provided by
6360 @option{-fsched-verbose-@var{n}}. @var{insn} is the instruction that
6364 @deftypefn {Target Hook} int TARGET_SCHED_ADJUST_COST (rtx @var{insn}, rtx @var{link}, rtx @var{dep_insn}, int @var{cost})
6365 This function corrects the value of @var{cost} based on the
6366 relationship between @var{insn} and @var{dep_insn} through the
6367 dependence @var{link}. It should return the new value. The default
6368 is to make no adjustment to @var{cost}. This can be used for example
6369 to specify to the scheduler using the traditional pipeline description
6370 that an output- or anti-dependence does not incur the same cost as a
6371 data-dependence. If the scheduler using the automaton based pipeline
6372 description, the cost of anti-dependence is zero and the cost of
6373 output-dependence is maximum of one and the difference of latency
6374 times of the first and the second insns. If these values are not
6375 acceptable, you could use the hook to modify them too. See also
6376 @pxref{Processor pipeline description}.
6379 @deftypefn {Target Hook} int TARGET_SCHED_ADJUST_PRIORITY (rtx @var{insn}, int @var{priority})
6380 This hook adjusts the integer scheduling priority @var{priority} of
6381 @var{insn}. It should return the new priority. Increase the priority to
6382 execute @var{insn} earlier, reduce the priority to execute @var{insn}
6383 later. Do not define this hook if you do not need to adjust the
6384 scheduling priorities of insns.
6387 @deftypefn {Target Hook} int TARGET_SCHED_REORDER (FILE *@var{file}, int @var{verbose}, rtx *@var{ready}, int *@var{n_readyp}, int @var{clock})
6388 This hook is executed by the scheduler after it has scheduled the ready
6389 list, to allow the machine description to reorder it (for example to
6390 combine two small instructions together on @samp{VLIW} machines).
6391 @var{file} is either a null pointer, or a stdio stream to write any
6392 debug output to. @var{verbose} is the verbose level provided by
6393 @option{-fsched-verbose-@var{n}}. @var{ready} is a pointer to the ready
6394 list of instructions that are ready to be scheduled. @var{n_readyp} is
6395 a pointer to the number of elements in the ready list. The scheduler
6396 reads the ready list in reverse order, starting with
6397 @var{ready}[@var{*n_readyp} @minus{} 1] and going to @var{ready}[0]. @var{clock}
6398 is the timer tick of the scheduler. You may modify the ready list and
6399 the number of ready insns. The return value is the number of insns that
6400 can issue this cycle; normally this is just @code{issue_rate}. See also
6401 @samp{TARGET_SCHED_REORDER2}.
6404 @deftypefn {Target Hook} int TARGET_SCHED_REORDER2 (FILE *@var{file}, int @var{verbose}, rtx *@var{ready}, int *@var{n_ready}, @var{clock})
6405 Like @samp{TARGET_SCHED_REORDER}, but called at a different time. That
6406 function is called whenever the scheduler starts a new cycle. This one
6407 is called once per iteration over a cycle, immediately after
6408 @samp{TARGET_SCHED_VARIABLE_ISSUE}; it can reorder the ready list and
6409 return the number of insns to be scheduled in the same cycle. Defining
6410 this hook can be useful if there are frequent situations where
6411 scheduling one insn causes other insns to become ready in the same
6412 cycle. These other insns can then be taken into account properly.
6415 @deftypefn {Target Hook} void TARGET_SCHED_DEPENDENCIES_EVALUATION_HOOK (rtx @var{head}, rtx @var{tail})
6416 This hook is called after evaluation forward dependencies of insns in
6417 chain given by two parameter values (@var{head} and @var{tail}
6418 correspondingly) but before insns scheduling of the insn chain. For
6419 example, it can be used for better insn classification if it requires
6420 analysis of dependencies. This hook can use backward and forward
6421 dependencies of the insn scheduler because they are already
6425 @deftypefn {Target Hook} void TARGET_SCHED_INIT (FILE *@var{file}, int @var{verbose}, int @var{max_ready})
6426 This hook is executed by the scheduler at the beginning of each block of
6427 instructions that are to be scheduled. @var{file} is either a null
6428 pointer, or a stdio stream to write any debug output to. @var{verbose}
6429 is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6430 @var{max_ready} is the maximum number of insns in the current scheduling
6431 region that can be live at the same time. This can be used to allocate
6432 scratch space if it is needed, e.g.@: by @samp{TARGET_SCHED_REORDER}.
6435 @deftypefn {Target Hook} void TARGET_SCHED_FINISH (FILE *@var{file}, int @var{verbose})
6436 This hook is executed by the scheduler at the end of each block of
6437 instructions that are to be scheduled. It can be used to perform
6438 cleanup of any actions done by the other scheduling hooks. @var{file}
6439 is either a null pointer, or a stdio stream to write any debug output
6440 to. @var{verbose} is the verbose level provided by
6441 @option{-fsched-verbose-@var{n}}.
6444 @deftypefn {Target Hook} void TARGET_SCHED_INIT_GLOBAL (FILE *@var{file}, int @var{verbose}, int @var{old_max_uid})
6445 This hook is executed by the scheduler after function level initializations.
6446 @var{file} is either a null pointer, or a stdio stream to write any debug output to.
6447 @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6448 @var{old_max_uid} is the maximum insn uid when scheduling begins.
6451 @deftypefn {Target Hook} void TARGET_SCHED_FINISH_GLOBAL (FILE *@var{file}, int @var{verbose})
6452 This is the cleanup hook corresponding to @code{TARGET_SCHED_INIT_GLOBAL}.
6453 @var{file} is either a null pointer, or a stdio stream to write any debug output to.
6454 @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6457 @deftypefn {Target Hook} int TARGET_SCHED_DFA_PRE_CYCLE_INSN (void)
6458 The hook returns an RTL insn. The automaton state used in the
6459 pipeline hazard recognizer is changed as if the insn were scheduled
6460 when the new simulated processor cycle starts. Usage of the hook may
6461 simplify the automaton pipeline description for some @acronym{VLIW}
6462 processors. If the hook is defined, it is used only for the automaton
6463 based pipeline description. The default is not to change the state
6464 when the new simulated processor cycle starts.
6467 @deftypefn {Target Hook} void TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN (void)
6468 The hook can be used to initialize data used by the previous hook.
6471 @deftypefn {Target Hook} int TARGET_SCHED_DFA_POST_CYCLE_INSN (void)
6472 The hook is analogous to @samp{TARGET_SCHED_DFA_PRE_CYCLE_INSN} but used
6473 to changed the state as if the insn were scheduled when the new
6474 simulated processor cycle finishes.
6477 @deftypefn {Target Hook} void TARGET_SCHED_INIT_DFA_POST_CYCLE_INSN (void)
6478 The hook is analogous to @samp{TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN} but
6479 used to initialize data used by the previous hook.
6482 @deftypefn {Target Hook} void TARGET_SCHED_DFA_PRE_CYCLE_ADVANCE (void)
6483 The hook to notify target that the current simulated cycle is about to finish.
6484 The hook is analogous to @samp{TARGET_SCHED_DFA_PRE_CYCLE_INSN} but used
6485 to change the state in more complicated situations - e.g., when advancing
6486 state on a single insn is not enough.
6489 @deftypefn {Target Hook} void TARGET_SCHED_DFA_POST_CYCLE_ADVANCE (void)
6490 The hook to notify target that new simulated cycle has just started.
6491 The hook is analogous to @samp{TARGET_SCHED_DFA_POST_CYCLE_INSN} but used
6492 to change the state in more complicated situations - e.g., when advancing
6493 state on a single insn is not enough.
6496 @deftypefn {Target Hook} int TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD (void)
6497 This hook controls better choosing an insn from the ready insn queue
6498 for the @acronym{DFA}-based insn scheduler. Usually the scheduler
6499 chooses the first insn from the queue. If the hook returns a positive
6500 value, an additional scheduler code tries all permutations of
6501 @samp{TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD ()}
6502 subsequent ready insns to choose an insn whose issue will result in
6503 maximal number of issued insns on the same cycle. For the
6504 @acronym{VLIW} processor, the code could actually solve the problem of
6505 packing simple insns into the @acronym{VLIW} insn. Of course, if the
6506 rules of @acronym{VLIW} packing are described in the automaton.
6508 This code also could be used for superscalar @acronym{RISC}
6509 processors. Let us consider a superscalar @acronym{RISC} processor
6510 with 3 pipelines. Some insns can be executed in pipelines @var{A} or
6511 @var{B}, some insns can be executed only in pipelines @var{B} or
6512 @var{C}, and one insn can be executed in pipeline @var{B}. The
6513 processor may issue the 1st insn into @var{A} and the 2nd one into
6514 @var{B}. In this case, the 3rd insn will wait for freeing @var{B}
6515 until the next cycle. If the scheduler issues the 3rd insn the first,
6516 the processor could issue all 3 insns per cycle.
6518 Actually this code demonstrates advantages of the automaton based
6519 pipeline hazard recognizer. We try quickly and easy many insn
6520 schedules to choose the best one.
6522 The default is no multipass scheduling.
6525 @deftypefn {Target Hook} int TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD (rtx)
6527 This hook controls what insns from the ready insn queue will be
6528 considered for the multipass insn scheduling. If the hook returns
6529 zero for insn passed as the parameter, the insn will be not chosen to
6532 The default is that any ready insns can be chosen to be issued.
6535 @deftypefn {Target Hook} int TARGET_SCHED_DFA_NEW_CYCLE (FILE *, int, rtx, int, int, int *)
6537 This hook is called by the insn scheduler before issuing insn passed
6538 as the third parameter on given cycle. If the hook returns nonzero,
6539 the insn is not issued on given processors cycle. Instead of that,
6540 the processor cycle is advanced. If the value passed through the last
6541 parameter is zero, the insn ready queue is not sorted on the new cycle
6542 start as usually. The first parameter passes file for debugging
6543 output. The second one passes the scheduler verbose level of the
6544 debugging output. The forth and the fifth parameter values are
6545 correspondingly processor cycle on which the previous insn has been
6546 issued and the current processor cycle.
6549 @deftypefn {Target Hook} bool TARGET_SCHED_IS_COSTLY_DEPENDENCE (struct dep_def *@var{_dep}, int @var{cost}, int @var{distance})
6550 This hook is used to define which dependences are considered costly by
6551 the target, so costly that it is not advisable to schedule the insns that
6552 are involved in the dependence too close to one another. The parameters
6553 to this hook are as follows: The first parameter @var{_dep} is the dependence
6554 being evaluated. The second parameter @var{cost} is the cost of the
6555 dependence, and the third
6556 parameter @var{distance} is the distance in cycles between the two insns.
6557 The hook returns @code{true} if considering the distance between the two
6558 insns the dependence between them is considered costly by the target,
6559 and @code{false} otherwise.
6561 Defining this hook can be useful in multiple-issue out-of-order machines,
6562 where (a) it's practically hopeless to predict the actual data/resource
6563 delays, however: (b) there's a better chance to predict the actual grouping
6564 that will be formed, and (c) correctly emulating the grouping can be very
6565 important. In such targets one may want to allow issuing dependent insns
6566 closer to one another---i.e., closer than the dependence distance; however,
6567 not in cases of "costly dependences", which this hooks allows to define.
6570 @deftypefn {Target Hook} void TARGET_SCHED_H_I_D_EXTENDED (void)
6571 This hook is called by the insn scheduler after emitting a new instruction to
6572 the instruction stream. The hook notifies a target backend to extend its
6573 per instruction data structures.
6576 @deftypefn {Target Hook} void * TARGET_SCHED_ALLOC_SCHED_CONTEXT (void)
6577 Return a pointer to a store large enough to hold target scheduling context.
6580 @deftypefn {Target Hook} void TARGET_SCHED_INIT_SCHED_CONTEXT (void *@var{tc}, bool @var{clean_p})
6581 Initialize store pointed to by @var{tc} to hold target scheduling context.
6582 It @var{clean_p} is true then initialize @var{tc} as if scheduler is at the
6583 beginning of the block. Otherwise, make a copy of the current context in
6587 @deftypefn {Target Hook} void TARGET_SCHED_SET_SCHED_CONTEXT (void *@var{tc})
6588 Copy target scheduling context pointer to by @var{tc} to the current context.
6591 @deftypefn {Target Hook} void TARGET_SCHED_CLEAR_SCHED_CONTEXT (void *@var{tc})
6592 Deallocate internal data in target scheduling context pointed to by @var{tc}.
6595 @deftypefn {Target Hook} void TARGET_SCHED_FREE_SCHED_CONTEXT (void *@var{tc})
6596 Deallocate a store for target scheduling context pointed to by @var{tc}.
6599 @deftypefn {Target Hook} void * TARGET_SCHED_ALLOC_SCHED_CONTEXT (void)
6600 Return a pointer to a store large enough to hold target scheduling context.
6603 @deftypefn {Target Hook} void TARGET_SCHED_INIT_SCHED_CONTEXT (void *@var{tc}, bool @var{clean_p})
6604 Initialize store pointed to by @var{tc} to hold target scheduling context.
6605 It @var{clean_p} is true then initialize @var{tc} as if scheduler is at the
6606 beginning of the block. Otherwise, make a copy of the current context in
6610 @deftypefn {Target Hook} void TARGET_SCHED_SET_SCHED_CONTEXT (void *@var{tc})
6611 Copy target scheduling context pointer to by @var{tc} to the current context.
6614 @deftypefn {Target Hook} void TARGET_SCHED_CLEAR_SCHED_CONTEXT (void *@var{tc})
6615 Deallocate internal data in target scheduling context pointed to by @var{tc}.
6618 @deftypefn {Target Hook} void TARGET_SCHED_FREE_SCHED_CONTEXT (void *@var{tc})
6619 Deallocate a store for target scheduling context pointed to by @var{tc}.
6622 @deftypefn {Target Hook} int TARGET_SCHED_SPECULATE_INSN (rtx @var{insn}, int @var{request}, rtx *@var{new_pat})
6623 This hook is called by the insn scheduler when @var{insn} has only
6624 speculative dependencies and therefore can be scheduled speculatively.
6625 The hook is used to check if the pattern of @var{insn} has a speculative
6626 version and, in case of successful check, to generate that speculative
6627 pattern. The hook should return 1, if the instruction has a speculative form,
6628 or @minus{}1, if it doesn't. @var{request} describes the type of requested
6629 speculation. If the return value equals 1 then @var{new_pat} is assigned
6630 the generated speculative pattern.
6633 @deftypefn {Target Hook} int TARGET_SCHED_NEEDS_BLOCK_P (rtx @var{insn})
6634 This hook is called by the insn scheduler during generation of recovery code
6635 for @var{insn}. It should return nonzero, if the corresponding check
6636 instruction should branch to recovery code, or zero otherwise.
6639 @deftypefn {Target Hook} rtx TARGET_SCHED_GEN_CHECK (rtx @var{insn}, rtx @var{label}, int @var{mutate_p})
6640 This hook is called by the insn scheduler to generate a pattern for recovery
6641 check instruction. If @var{mutate_p} is zero, then @var{insn} is a
6642 speculative instruction for which the check should be generated.
6643 @var{label} is either a label of a basic block, where recovery code should
6644 be emitted, or a null pointer, when requested check doesn't branch to
6645 recovery code (a simple check). If @var{mutate_p} is nonzero, then
6646 a pattern for a branchy check corresponding to a simple check denoted by
6647 @var{insn} should be generated. In this case @var{label} can't be null.
6650 @deftypefn {Target Hook} int TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD_SPEC (rtx @var{insn})
6651 This hook is used as a workaround for
6652 @samp{TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD} not being
6653 called on the first instruction of the ready list. The hook is used to
6654 discard speculative instruction that stand first in the ready list from
6655 being scheduled on the current cycle. For non-speculative instructions,
6656 the hook should always return nonzero. For example, in the ia64 backend
6657 the hook is used to cancel data speculative insns when the ALAT table
6661 @deftypefn {Target Hook} void TARGET_SCHED_SET_SCHED_FLAGS (unsigned int *@var{flags}, spec_info_t @var{spec_info})
6662 This hook is used by the insn scheduler to find out what features should be
6663 enabled/used. @var{flags} initially may have either the SCHED_RGN or SCHED_EBB
6664 bit set. This denotes the scheduler pass for which the data should be
6665 provided. The target backend should modify @var{flags} by modifying
6666 the bits corresponding to the following features: USE_DEPS_LIST, USE_GLAT,
6667 DETACH_LIFE_INFO, and DO_SPECULATION@. For the DO_SPECULATION feature
6668 an additional structure @var{spec_info} should be filled by the target.
6669 The structure describes speculation types that can be used in the scheduler.
6672 @deftypefn {Target Hook} int TARGET_SCHED_SMS_RES_MII (struct ddg *@var{g})
6673 This hook is called by the swing modulo scheduler to calculate a
6674 resource-based lower bound which is based on the resources available in
6675 the machine and the resources required by each instruction. The target
6676 backend can use @var{g} to calculate such bound. A very simple lower
6677 bound will be used in case this hook is not implemented: the total number
6678 of instructions divided by the issue rate.
6682 @section Dividing the Output into Sections (Texts, Data, @dots{})
6683 @c the above section title is WAY too long. maybe cut the part between
6684 @c the (...)? --mew 10feb93
6686 An object file is divided into sections containing different types of
6687 data. In the most common case, there are three sections: the @dfn{text
6688 section}, which holds instructions and read-only data; the @dfn{data
6689 section}, which holds initialized writable data; and the @dfn{bss
6690 section}, which holds uninitialized data. Some systems have other kinds
6693 @file{varasm.c} provides several well-known sections, such as
6694 @code{text_section}, @code{data_section} and @code{bss_section}.
6695 The normal way of controlling a @code{@var{foo}_section} variable
6696 is to define the associated @code{@var{FOO}_SECTION_ASM_OP} macro,
6697 as described below. The macros are only read once, when @file{varasm.c}
6698 initializes itself, so their values must be run-time constants.
6699 They may however depend on command-line flags.
6701 @emph{Note:} Some run-time files, such @file{crtstuff.c}, also make
6702 use of the @code{@var{FOO}_SECTION_ASM_OP} macros, and expect them
6703 to be string literals.
6705 Some assemblers require a different string to be written every time a
6706 section is selected. If your assembler falls into this category, you
6707 should define the @code{TARGET_ASM_INIT_SECTIONS} hook and use
6708 @code{get_unnamed_section} to set up the sections.
6710 You must always create a @code{text_section}, either by defining
6711 @code{TEXT_SECTION_ASM_OP} or by initializing @code{text_section}
6712 in @code{TARGET_ASM_INIT_SECTIONS}. The same is true of
6713 @code{data_section} and @code{DATA_SECTION_ASM_OP}. If you do not
6714 create a distinct @code{readonly_data_section}, the default is to
6715 reuse @code{text_section}.
6717 All the other @file{varasm.c} sections are optional, and are null
6718 if the target does not provide them.
6720 @defmac TEXT_SECTION_ASM_OP
6721 A C expression whose value is a string, including spacing, containing the
6722 assembler operation that should precede instructions and read-only data.
6723 Normally @code{"\t.text"} is right.
6726 @defmac HOT_TEXT_SECTION_NAME
6727 If defined, a C string constant for the name of the section containing most
6728 frequently executed functions of the program. If not defined, GCC will provide
6729 a default definition if the target supports named sections.
6732 @defmac UNLIKELY_EXECUTED_TEXT_SECTION_NAME
6733 If defined, a C string constant for the name of the section containing unlikely
6734 executed functions in the program.
6737 @defmac DATA_SECTION_ASM_OP
6738 A C expression whose value is a string, including spacing, containing the
6739 assembler operation to identify the following data as writable initialized
6740 data. Normally @code{"\t.data"} is right.
6743 @defmac SDATA_SECTION_ASM_OP
6744 If defined, a C expression whose value is a string, including spacing,
6745 containing the assembler operation to identify the following data as
6746 initialized, writable small data.
6749 @defmac READONLY_DATA_SECTION_ASM_OP
6750 A C expression whose value is a string, including spacing, containing the
6751 assembler operation to identify the following data as read-only initialized
6755 @defmac BSS_SECTION_ASM_OP
6756 If defined, a C expression whose value is a string, including spacing,
6757 containing the assembler operation to identify the following data as
6758 uninitialized global data. If not defined, and neither
6759 @code{ASM_OUTPUT_BSS} nor @code{ASM_OUTPUT_ALIGNED_BSS} are defined,
6760 uninitialized global data will be output in the data section if
6761 @option{-fno-common} is passed, otherwise @code{ASM_OUTPUT_COMMON} will be
6765 @defmac SBSS_SECTION_ASM_OP
6766 If defined, a C expression whose value is a string, including spacing,
6767 containing the assembler operation to identify the following data as
6768 uninitialized, writable small data.
6771 @defmac INIT_SECTION_ASM_OP
6772 If defined, a C expression whose value is a string, including spacing,
6773 containing the assembler operation to identify the following data as
6774 initialization code. If not defined, GCC will assume such a section does
6775 not exist. This section has no corresponding @code{init_section}
6776 variable; it is used entirely in runtime code.
6779 @defmac FINI_SECTION_ASM_OP
6780 If defined, a C expression whose value is a string, including spacing,
6781 containing the assembler operation to identify the following data as
6782 finalization code. If not defined, GCC will assume such a section does
6783 not exist. This section has no corresponding @code{fini_section}
6784 variable; it is used entirely in runtime code.
6787 @defmac INIT_ARRAY_SECTION_ASM_OP
6788 If defined, a C expression whose value is a string, including spacing,
6789 containing the assembler operation to identify the following data as
6790 part of the @code{.init_array} (or equivalent) section. If not
6791 defined, GCC will assume such a section does not exist. Do not define
6792 both this macro and @code{INIT_SECTION_ASM_OP}.
6795 @defmac FINI_ARRAY_SECTION_ASM_OP
6796 If defined, a C expression whose value is a string, including spacing,
6797 containing the assembler operation to identify the following data as
6798 part of the @code{.fini_array} (or equivalent) section. If not
6799 defined, GCC will assume such a section does not exist. Do not define
6800 both this macro and @code{FINI_SECTION_ASM_OP}.
6803 @defmac CRT_CALL_STATIC_FUNCTION (@var{section_op}, @var{function})
6804 If defined, an ASM statement that switches to a different section
6805 via @var{section_op}, calls @var{function}, and switches back to
6806 the text section. This is used in @file{crtstuff.c} if
6807 @code{INIT_SECTION_ASM_OP} or @code{FINI_SECTION_ASM_OP} to calls
6808 to initialization and finalization functions from the init and fini
6809 sections. By default, this macro uses a simple function call. Some
6810 ports need hand-crafted assembly code to avoid dependencies on
6811 registers initialized in the function prologue or to ensure that
6812 constant pools don't end up too far way in the text section.
6815 @defmac TARGET_LIBGCC_SDATA_SECTION
6816 If defined, a string which names the section into which small
6817 variables defined in crtstuff and libgcc should go. This is useful
6818 when the target has options for optimizing access to small data, and
6819 you want the crtstuff and libgcc routines to be conservative in what
6820 they expect of your application yet liberal in what your application
6821 expects. For example, for targets with a @code{.sdata} section (like
6822 MIPS), you could compile crtstuff with @code{-G 0} so that it doesn't
6823 require small data support from your application, but use this macro
6824 to put small data into @code{.sdata} so that your application can
6825 access these variables whether it uses small data or not.
6828 @defmac FORCE_CODE_SECTION_ALIGN
6829 If defined, an ASM statement that aligns a code section to some
6830 arbitrary boundary. This is used to force all fragments of the
6831 @code{.init} and @code{.fini} sections to have to same alignment
6832 and thus prevent the linker from having to add any padding.
6835 @defmac JUMP_TABLES_IN_TEXT_SECTION
6836 Define this macro to be an expression with a nonzero value if jump
6837 tables (for @code{tablejump} insns) should be output in the text
6838 section, along with the assembler instructions. Otherwise, the
6839 readonly data section is used.
6841 This macro is irrelevant if there is no separate readonly data section.
6844 @deftypefn {Target Hook} void TARGET_ASM_INIT_SECTIONS (void)
6845 Define this hook if you need to do something special to set up the
6846 @file{varasm.c} sections, or if your target has some special sections
6847 of its own that you need to create.
6849 GCC calls this hook after processing the command line, but before writing
6850 any assembly code, and before calling any of the section-returning hooks
6854 @deftypefn {Target Hook} TARGET_ASM_RELOC_RW_MASK (void)
6855 Return a mask describing how relocations should be treated when
6856 selecting sections. Bit 1 should be set if global relocations
6857 should be placed in a read-write section; bit 0 should be set if
6858 local relocations should be placed in a read-write section.
6860 The default version of this function returns 3 when @option{-fpic}
6861 is in effect, and 0 otherwise. The hook is typically redefined
6862 when the target cannot support (some kinds of) dynamic relocations
6863 in read-only sections even in executables.
6866 @deftypefn {Target Hook} {section *} TARGET_ASM_SELECT_SECTION (tree @var{exp}, int @var{reloc}, unsigned HOST_WIDE_INT @var{align})
6867 Return the section into which @var{exp} should be placed. You can
6868 assume that @var{exp} is either a @code{VAR_DECL} node or a constant of
6869 some sort. @var{reloc} indicates whether the initial value of @var{exp}
6870 requires link-time relocations. Bit 0 is set when variable contains
6871 local relocations only, while bit 1 is set for global relocations.
6872 @var{align} is the constant alignment in bits.
6874 The default version of this function takes care of putting read-only
6875 variables in @code{readonly_data_section}.
6877 See also @var{USE_SELECT_SECTION_FOR_FUNCTIONS}.
6880 @defmac USE_SELECT_SECTION_FOR_FUNCTIONS
6881 Define this macro if you wish TARGET_ASM_SELECT_SECTION to be called
6882 for @code{FUNCTION_DECL}s as well as for variables and constants.
6884 In the case of a @code{FUNCTION_DECL}, @var{reloc} will be zero if the
6885 function has been determined to be likely to be called, and nonzero if
6886 it is unlikely to be called.
6889 @deftypefn {Target Hook} void TARGET_ASM_UNIQUE_SECTION (tree @var{decl}, int @var{reloc})
6890 Build up a unique section name, expressed as a @code{STRING_CST} node,
6891 and assign it to @samp{DECL_SECTION_NAME (@var{decl})}.
6892 As with @code{TARGET_ASM_SELECT_SECTION}, @var{reloc} indicates whether
6893 the initial value of @var{exp} requires link-time relocations.
6895 The default version of this function appends the symbol name to the
6896 ELF section name that would normally be used for the symbol. For
6897 example, the function @code{foo} would be placed in @code{.text.foo}.
6898 Whatever the actual target object format, this is often good enough.
6901 @deftypefn {Target Hook} {section *} TARGET_ASM_FUNCTION_RODATA_SECTION (tree @var{decl})
6902 Return the readonly data section associated with
6903 @samp{DECL_SECTION_NAME (@var{decl})}.
6904 The default version of this function selects @code{.gnu.linkonce.r.name} if
6905 the function's section is @code{.gnu.linkonce.t.name}, @code{.rodata.name}
6906 if function is in @code{.text.name}, and the normal readonly-data section
6910 @deftypefn {Target Hook} {section *} TARGET_ASM_SELECT_RTX_SECTION (enum machine_mode @var{mode}, rtx @var{x}, unsigned HOST_WIDE_INT @var{align})
6911 Return the section into which a constant @var{x}, of mode @var{mode},
6912 should be placed. You can assume that @var{x} is some kind of
6913 constant in RTL@. The argument @var{mode} is redundant except in the
6914 case of a @code{const_int} rtx. @var{align} is the constant alignment
6917 The default version of this function takes care of putting symbolic
6918 constants in @code{flag_pic} mode in @code{data_section} and everything
6919 else in @code{readonly_data_section}.
6922 @deftypefn {Target Hook} void TARGET_MANGLE_DECL_ASSEMBLER_NAME (tree @var{decl}, tree @var{id})
6923 Define this hook if you need to postprocess the assembler name generated
6924 by target-independent code. The @var{id} provided to this hook will be
6925 the computed name (e.g., the macro @code{DECL_NAME} of the @var{decl} in C,
6926 or the mangled name of the @var{decl} in C++). The return value of the
6927 hook is an @code{IDENTIFIER_NODE} for the appropriate mangled name on
6928 your target system. The default implementation of this hook just
6929 returns the @var{id} provided.
6932 @deftypefn {Target Hook} void TARGET_ENCODE_SECTION_INFO (tree @var{decl}, rtx @var{rtl}, int @var{new_decl_p})
6933 Define this hook if references to a symbol or a constant must be
6934 treated differently depending on something about the variable or
6935 function named by the symbol (such as what section it is in).
6937 The hook is executed immediately after rtl has been created for
6938 @var{decl}, which may be a variable or function declaration or
6939 an entry in the constant pool. In either case, @var{rtl} is the
6940 rtl in question. Do @emph{not} use @code{DECL_RTL (@var{decl})}
6941 in this hook; that field may not have been initialized yet.
6943 In the case of a constant, it is safe to assume that the rtl is
6944 a @code{mem} whose address is a @code{symbol_ref}. Most decls
6945 will also have this form, but that is not guaranteed. Global
6946 register variables, for instance, will have a @code{reg} for their
6947 rtl. (Normally the right thing to do with such unusual rtl is
6950 The @var{new_decl_p} argument will be true if this is the first time
6951 that @code{TARGET_ENCODE_SECTION_INFO} has been invoked on this decl. It will
6952 be false for subsequent invocations, which will happen for duplicate
6953 declarations. Whether or not anything must be done for the duplicate
6954 declaration depends on whether the hook examines @code{DECL_ATTRIBUTES}.
6955 @var{new_decl_p} is always true when the hook is called for a constant.
6957 @cindex @code{SYMBOL_REF_FLAG}, in @code{TARGET_ENCODE_SECTION_INFO}
6958 The usual thing for this hook to do is to record flags in the
6959 @code{symbol_ref}, using @code{SYMBOL_REF_FLAG} or @code{SYMBOL_REF_FLAGS}.
6960 Historically, the name string was modified if it was necessary to
6961 encode more than one bit of information, but this practice is now
6962 discouraged; use @code{SYMBOL_REF_FLAGS}.
6964 The default definition of this hook, @code{default_encode_section_info}
6965 in @file{varasm.c}, sets a number of commonly-useful bits in
6966 @code{SYMBOL_REF_FLAGS}. Check whether the default does what you need
6967 before overriding it.
6970 @deftypefn {Target Hook} const char *TARGET_STRIP_NAME_ENCODING (const char *name)
6971 Decode @var{name} and return the real name part, sans
6972 the characters that @code{TARGET_ENCODE_SECTION_INFO}
6976 @deftypefn {Target Hook} bool TARGET_IN_SMALL_DATA_P (tree @var{exp})
6977 Returns true if @var{exp} should be placed into a ``small data'' section.
6978 The default version of this hook always returns false.
6981 @deftypevr {Target Hook} bool TARGET_HAVE_SRODATA_SECTION
6982 Contains the value true if the target places read-only
6983 ``small data'' into a separate section. The default value is false.
6986 @deftypefn {Target Hook} bool TARGET_BINDS_LOCAL_P (tree @var{exp})
6987 Returns true if @var{exp} names an object for which name resolution
6988 rules must resolve to the current ``module'' (dynamic shared library
6989 or executable image).
6991 The default version of this hook implements the name resolution rules
6992 for ELF, which has a looser model of global name binding than other
6993 currently supported object file formats.
6996 @deftypevr {Target Hook} bool TARGET_HAVE_TLS
6997 Contains the value true if the target supports thread-local storage.
6998 The default value is false.
7003 @section Position Independent Code
7004 @cindex position independent code
7007 This section describes macros that help implement generation of position
7008 independent code. Simply defining these macros is not enough to
7009 generate valid PIC; you must also add support to the hook
7010 @code{TARGET_LEGITIMATE_ADDRESS_P} and to the macro
7011 @code{PRINT_OPERAND_ADDRESS}, as well as @code{LEGITIMIZE_ADDRESS}. You
7012 must modify the definition of @samp{movsi} to do something appropriate
7013 when the source operand contains a symbolic address. You may also
7014 need to alter the handling of switch statements so that they use
7016 @c i rearranged the order of the macros above to try to force one of
7017 @c them to the next line, to eliminate an overfull hbox. --mew 10feb93
7019 @defmac PIC_OFFSET_TABLE_REGNUM
7020 The register number of the register used to address a table of static
7021 data addresses in memory. In some cases this register is defined by a
7022 processor's ``application binary interface'' (ABI)@. When this macro
7023 is defined, RTL is generated for this register once, as with the stack
7024 pointer and frame pointer registers. If this macro is not defined, it
7025 is up to the machine-dependent files to allocate such a register (if
7026 necessary). Note that this register must be fixed when in use (e.g.@:
7027 when @code{flag_pic} is true).
7030 @defmac PIC_OFFSET_TABLE_REG_CALL_CLOBBERED
7031 Define this macro if the register defined by
7032 @code{PIC_OFFSET_TABLE_REGNUM} is clobbered by calls. Do not define
7033 this macro if @code{PIC_OFFSET_TABLE_REGNUM} is not defined.
7036 @defmac LEGITIMATE_PIC_OPERAND_P (@var{x})
7037 A C expression that is nonzero if @var{x} is a legitimate immediate
7038 operand on the target machine when generating position independent code.
7039 You can assume that @var{x} satisfies @code{CONSTANT_P}, so you need not
7040 check this. You can also assume @var{flag_pic} is true, so you need not
7041 check it either. You need not define this macro if all constants
7042 (including @code{SYMBOL_REF}) can be immediate operands when generating
7043 position independent code.
7046 @node Assembler Format
7047 @section Defining the Output Assembler Language
7049 This section describes macros whose principal purpose is to describe how
7050 to write instructions in assembler language---rather than what the
7054 * File Framework:: Structural information for the assembler file.
7055 * Data Output:: Output of constants (numbers, strings, addresses).
7056 * Uninitialized Data:: Output of uninitialized variables.
7057 * Label Output:: Output and generation of labels.
7058 * Initialization:: General principles of initialization
7059 and termination routines.
7060 * Macros for Initialization::
7061 Specific macros that control the handling of
7062 initialization and termination routines.
7063 * Instruction Output:: Output of actual instructions.
7064 * Dispatch Tables:: Output of jump tables.
7065 * Exception Region Output:: Output of exception region code.
7066 * Alignment Output:: Pseudo ops for alignment and skipping data.
7069 @node File Framework
7070 @subsection The Overall Framework of an Assembler File
7071 @cindex assembler format
7072 @cindex output of assembler code
7074 @c prevent bad page break with this line
7075 This describes the overall framework of an assembly file.
7077 @deftypefn {Target Hook} void TARGET_ASM_FILE_START ()
7078 @findex default_file_start
7079 Output to @code{asm_out_file} any text which the assembler expects to
7080 find at the beginning of a file. The default behavior is controlled
7081 by two flags, documented below. Unless your target's assembler is
7082 quite unusual, if you override the default, you should call
7083 @code{default_file_start} at some point in your target hook. This
7084 lets other target files rely on these variables.
7087 @deftypevr {Target Hook} bool TARGET_ASM_FILE_START_APP_OFF
7088 If this flag is true, the text of the macro @code{ASM_APP_OFF} will be
7089 printed as the very first line in the assembly file, unless
7090 @option{-fverbose-asm} is in effect. (If that macro has been defined
7091 to the empty string, this variable has no effect.) With the normal
7092 definition of @code{ASM_APP_OFF}, the effect is to notify the GNU
7093 assembler that it need not bother stripping comments or extra
7094 whitespace from its input. This allows it to work a bit faster.
7096 The default is false. You should not set it to true unless you have
7097 verified that your port does not generate any extra whitespace or
7098 comments that will cause GAS to issue errors in NO_APP mode.
7101 @deftypevr {Target Hook} bool TARGET_ASM_FILE_START_FILE_DIRECTIVE
7102 If this flag is true, @code{output_file_directive} will be called
7103 for the primary source file, immediately after printing
7104 @code{ASM_APP_OFF} (if that is enabled). Most ELF assemblers expect
7105 this to be done. The default is false.
7108 @deftypefn {Target Hook} void TARGET_ASM_FILE_END ()
7109 Output to @code{asm_out_file} any text which the assembler expects
7110 to find at the end of a file. The default is to output nothing.
7113 @deftypefun void file_end_indicate_exec_stack ()
7114 Some systems use a common convention, the @samp{.note.GNU-stack}
7115 special section, to indicate whether or not an object file relies on
7116 the stack being executable. If your system uses this convention, you
7117 should define @code{TARGET_ASM_FILE_END} to this function. If you
7118 need to do other things in that hook, have your hook function call
7122 @defmac ASM_COMMENT_START
7123 A C string constant describing how to begin a comment in the target
7124 assembler language. The compiler assumes that the comment will end at
7125 the end of the line.
7129 A C string constant for text to be output before each @code{asm}
7130 statement or group of consecutive ones. Normally this is
7131 @code{"#APP"}, which is a comment that has no effect on most
7132 assemblers but tells the GNU assembler that it must check the lines
7133 that follow for all valid assembler constructs.
7137 A C string constant for text to be output after each @code{asm}
7138 statement or group of consecutive ones. Normally this is
7139 @code{"#NO_APP"}, which tells the GNU assembler to resume making the
7140 time-saving assumptions that are valid for ordinary compiler output.
7143 @defmac ASM_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name})
7144 A C statement to output COFF information or DWARF debugging information
7145 which indicates that filename @var{name} is the current source file to
7146 the stdio stream @var{stream}.
7148 This macro need not be defined if the standard form of output
7149 for the file format in use is appropriate.
7152 @defmac OUTPUT_QUOTED_STRING (@var{stream}, @var{string})
7153 A C statement to output the string @var{string} to the stdio stream
7154 @var{stream}. If you do not call the function @code{output_quoted_string}
7155 in your config files, GCC will only call it to output filenames to
7156 the assembler source. So you can use it to canonicalize the format
7157 of the filename using this macro.
7160 @defmac ASM_OUTPUT_IDENT (@var{stream}, @var{string})
7161 A C statement to output something to the assembler file to handle a
7162 @samp{#ident} directive containing the text @var{string}. If this
7163 macro is not defined, nothing is output for a @samp{#ident} directive.
7166 @deftypefn {Target Hook} void TARGET_ASM_NAMED_SECTION (const char *@var{name}, unsigned int @var{flags}, unsigned int @var{align})
7167 Output assembly directives to switch to section @var{name}. The section
7168 should have attributes as specified by @var{flags}, which is a bit mask
7169 of the @code{SECTION_*} flags defined in @file{output.h}. If @var{align}
7170 is nonzero, it contains an alignment in bytes to be used for the section,
7171 otherwise some target default should be used. Only targets that must
7172 specify an alignment within the section directive need pay attention to
7173 @var{align} -- we will still use @code{ASM_OUTPUT_ALIGN}.
7176 @deftypefn {Target Hook} bool TARGET_HAVE_NAMED_SECTIONS
7177 This flag is true if the target supports @code{TARGET_ASM_NAMED_SECTION}.
7180 @anchor{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}
7181 @deftypefn {Target Hook} bool TARGET_HAVE_SWITCHABLE_BSS_SECTIONS
7182 This flag is true if we can create zeroed data by switching to a BSS
7183 section and then using @code{ASM_OUTPUT_SKIP} to allocate the space.
7184 This is true on most ELF targets.
7187 @deftypefn {Target Hook} {unsigned int} TARGET_SECTION_TYPE_FLAGS (tree @var{decl}, const char *@var{name}, int @var{reloc})
7188 Choose a set of section attributes for use by @code{TARGET_ASM_NAMED_SECTION}
7189 based on a variable or function decl, a section name, and whether or not the
7190 declaration's initializer may contain runtime relocations. @var{decl} may be
7191 null, in which case read-write data should be assumed.
7193 The default version of this function handles choosing code vs data,
7194 read-only vs read-write data, and @code{flag_pic}. You should only
7195 need to override this if your target has special flags that might be
7196 set via @code{__attribute__}.
7199 @deftypefn {Target Hook} {int} TARGET_ASM_RECORD_GCC_SWITCHES (print_switch_type @var{type}, const char * @var{text})
7200 Provides the target with the ability to record the gcc command line
7201 switches that have been passed to the compiler, and options that are
7202 enabled. The @var{type} argument specifies what is being recorded.
7203 It can take the following values:
7206 @item SWITCH_TYPE_PASSED
7207 @var{text} is a command line switch that has been set by the user.
7209 @item SWITCH_TYPE_ENABLED
7210 @var{text} is an option which has been enabled. This might be as a
7211 direct result of a command line switch, or because it is enabled by
7212 default or because it has been enabled as a side effect of a different
7213 command line switch. For example, the @option{-O2} switch enables
7214 various different individual optimization passes.
7216 @item SWITCH_TYPE_DESCRIPTIVE
7217 @var{text} is either NULL or some descriptive text which should be
7218 ignored. If @var{text} is NULL then it is being used to warn the
7219 target hook that either recording is starting or ending. The first
7220 time @var{type} is SWITCH_TYPE_DESCRIPTIVE and @var{text} is NULL, the
7221 warning is for start up and the second time the warning is for
7222 wind down. This feature is to allow the target hook to make any
7223 necessary preparations before it starts to record switches and to
7224 perform any necessary tidying up after it has finished recording
7227 @item SWITCH_TYPE_LINE_START
7228 This option can be ignored by this target hook.
7230 @item SWITCH_TYPE_LINE_END
7231 This option can be ignored by this target hook.
7234 The hook's return value must be zero. Other return values may be
7235 supported in the future.
7237 By default this hook is set to NULL, but an example implementation is
7238 provided for ELF based targets. Called @var{elf_record_gcc_switches},
7239 it records the switches as ASCII text inside a new, string mergeable
7240 section in the assembler output file. The name of the new section is
7241 provided by the @code{TARGET_ASM_RECORD_GCC_SWITCHES_SECTION} target
7245 @deftypefn {Target Hook} {const char *} TARGET_ASM_RECORD_GCC_SWITCHES_SECTION
7246 This is the name of the section that will be created by the example
7247 ELF implementation of the @code{TARGET_ASM_RECORD_GCC_SWITCHES} target
7253 @subsection Output of Data
7256 @deftypevr {Target Hook} {const char *} TARGET_ASM_BYTE_OP
7257 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_HI_OP
7258 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_SI_OP
7259 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_DI_OP
7260 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_TI_OP
7261 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_HI_OP
7262 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_SI_OP
7263 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_DI_OP
7264 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_TI_OP
7265 These hooks specify assembly directives for creating certain kinds
7266 of integer object. The @code{TARGET_ASM_BYTE_OP} directive creates a
7267 byte-sized object, the @code{TARGET_ASM_ALIGNED_HI_OP} one creates an
7268 aligned two-byte object, and so on. Any of the hooks may be
7269 @code{NULL}, indicating that no suitable directive is available.
7271 The compiler will print these strings at the start of a new line,
7272 followed immediately by the object's initial value. In most cases,
7273 the string should contain a tab, a pseudo-op, and then another tab.
7276 @deftypefn {Target Hook} bool TARGET_ASM_INTEGER (rtx @var{x}, unsigned int @var{size}, int @var{aligned_p})
7277 The @code{assemble_integer} function uses this hook to output an
7278 integer object. @var{x} is the object's value, @var{size} is its size
7279 in bytes and @var{aligned_p} indicates whether it is aligned. The
7280 function should return @code{true} if it was able to output the
7281 object. If it returns false, @code{assemble_integer} will try to
7282 split the object into smaller parts.
7284 The default implementation of this hook will use the
7285 @code{TARGET_ASM_BYTE_OP} family of strings, returning @code{false}
7286 when the relevant string is @code{NULL}.
7289 @defmac OUTPUT_ADDR_CONST_EXTRA (@var{stream}, @var{x}, @var{fail})
7290 A C statement to recognize @var{rtx} patterns that
7291 @code{output_addr_const} can't deal with, and output assembly code to
7292 @var{stream} corresponding to the pattern @var{x}. This may be used to
7293 allow machine-dependent @code{UNSPEC}s to appear within constants.
7295 If @code{OUTPUT_ADDR_CONST_EXTRA} fails to recognize a pattern, it must
7296 @code{goto fail}, so that a standard error message is printed. If it
7297 prints an error message itself, by calling, for example,
7298 @code{output_operand_lossage}, it may just complete normally.
7301 @defmac ASM_OUTPUT_ASCII (@var{stream}, @var{ptr}, @var{len})
7302 A C statement to output to the stdio stream @var{stream} an assembler
7303 instruction to assemble a string constant containing the @var{len}
7304 bytes at @var{ptr}. @var{ptr} will be a C expression of type
7305 @code{char *} and @var{len} a C expression of type @code{int}.
7307 If the assembler has a @code{.ascii} pseudo-op as found in the
7308 Berkeley Unix assembler, do not define the macro
7309 @code{ASM_OUTPUT_ASCII}.
7312 @defmac ASM_OUTPUT_FDESC (@var{stream}, @var{decl}, @var{n})
7313 A C statement to output word @var{n} of a function descriptor for
7314 @var{decl}. This must be defined if @code{TARGET_VTABLE_USES_DESCRIPTORS}
7315 is defined, and is otherwise unused.
7318 @defmac CONSTANT_POOL_BEFORE_FUNCTION
7319 You may define this macro as a C expression. You should define the
7320 expression to have a nonzero value if GCC should output the constant
7321 pool for a function before the code for the function, or a zero value if
7322 GCC should output the constant pool after the function. If you do
7323 not define this macro, the usual case, GCC will output the constant
7324 pool before the function.
7327 @defmac ASM_OUTPUT_POOL_PROLOGUE (@var{file}, @var{funname}, @var{fundecl}, @var{size})
7328 A C statement to output assembler commands to define the start of the
7329 constant pool for a function. @var{funname} is a string giving
7330 the name of the function. Should the return type of the function
7331 be required, it can be obtained via @var{fundecl}. @var{size}
7332 is the size, in bytes, of the constant pool that will be written
7333 immediately after this call.
7335 If no constant-pool prefix is required, the usual case, this macro need
7339 @defmac ASM_OUTPUT_SPECIAL_POOL_ENTRY (@var{file}, @var{x}, @var{mode}, @var{align}, @var{labelno}, @var{jumpto})
7340 A C statement (with or without semicolon) to output a constant in the
7341 constant pool, if it needs special treatment. (This macro need not do
7342 anything for RTL expressions that can be output normally.)
7344 The argument @var{file} is the standard I/O stream to output the
7345 assembler code on. @var{x} is the RTL expression for the constant to
7346 output, and @var{mode} is the machine mode (in case @var{x} is a
7347 @samp{const_int}). @var{align} is the required alignment for the value
7348 @var{x}; you should output an assembler directive to force this much
7351 The argument @var{labelno} is a number to use in an internal label for
7352 the address of this pool entry. The definition of this macro is
7353 responsible for outputting the label definition at the proper place.
7354 Here is how to do this:
7357 @code{(*targetm.asm_out.internal_label)} (@var{file}, "LC", @var{labelno});
7360 When you output a pool entry specially, you should end with a
7361 @code{goto} to the label @var{jumpto}. This will prevent the same pool
7362 entry from being output a second time in the usual manner.
7364 You need not define this macro if it would do nothing.
7367 @defmac ASM_OUTPUT_POOL_EPILOGUE (@var{file} @var{funname} @var{fundecl} @var{size})
7368 A C statement to output assembler commands to at the end of the constant
7369 pool for a function. @var{funname} is a string giving the name of the
7370 function. Should the return type of the function be required, you can
7371 obtain it via @var{fundecl}. @var{size} is the size, in bytes, of the
7372 constant pool that GCC wrote immediately before this call.
7374 If no constant-pool epilogue is required, the usual case, you need not
7378 @defmac IS_ASM_LOGICAL_LINE_SEPARATOR (@var{C}, @var{STR})
7379 Define this macro as a C expression which is nonzero if @var{C} is
7380 used as a logical line separator by the assembler. @var{STR} points
7381 to the position in the string where @var{C} was found; this can be used if
7382 a line separator uses multiple characters.
7384 If you do not define this macro, the default is that only
7385 the character @samp{;} is treated as a logical line separator.
7388 @deftypevr {Target Hook} {const char *} TARGET_ASM_OPEN_PAREN
7389 @deftypevrx {Target Hook} {const char *} TARGET_ASM_CLOSE_PAREN
7390 These target hooks are C string constants, describing the syntax in the
7391 assembler for grouping arithmetic expressions. If not overridden, they
7392 default to normal parentheses, which is correct for most assemblers.
7395 These macros are provided by @file{real.h} for writing the definitions
7396 of @code{ASM_OUTPUT_DOUBLE} and the like:
7398 @defmac REAL_VALUE_TO_TARGET_SINGLE (@var{x}, @var{l})
7399 @defmacx REAL_VALUE_TO_TARGET_DOUBLE (@var{x}, @var{l})
7400 @defmacx REAL_VALUE_TO_TARGET_LONG_DOUBLE (@var{x}, @var{l})
7401 @defmacx REAL_VALUE_TO_TARGET_DECIMAL32 (@var{x}, @var{l})
7402 @defmacx REAL_VALUE_TO_TARGET_DECIMAL64 (@var{x}, @var{l})
7403 @defmacx REAL_VALUE_TO_TARGET_DECIMAL128 (@var{x}, @var{l})
7404 These translate @var{x}, of type @code{REAL_VALUE_TYPE}, to the
7405 target's floating point representation, and store its bit pattern in
7406 the variable @var{l}. For @code{REAL_VALUE_TO_TARGET_SINGLE} and
7407 @code{REAL_VALUE_TO_TARGET_DECIMAL32}, this variable should be a
7408 simple @code{long int}. For the others, it should be an array of
7409 @code{long int}. The number of elements in this array is determined
7410 by the size of the desired target floating point data type: 32 bits of
7411 it go in each @code{long int} array element. Each array element holds
7412 32 bits of the result, even if @code{long int} is wider than 32 bits
7413 on the host machine.
7415 The array element values are designed so that you can print them out
7416 using @code{fprintf} in the order they should appear in the target
7420 @node Uninitialized Data
7421 @subsection Output of Uninitialized Variables
7423 Each of the macros in this section is used to do the whole job of
7424 outputting a single uninitialized variable.
7426 @defmac ASM_OUTPUT_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
7427 A C statement (sans semicolon) to output to the stdio stream
7428 @var{stream} the assembler definition of a common-label named
7429 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7430 is the size rounded up to whatever alignment the caller wants. It is
7431 possible that @var{size} may be zero, for instance if a struct with no
7432 other member than a zero-length array is defined. In this case, the
7433 backend must output a symbol definition that allocates at least one
7434 byte, both so that the address of the resulting object does not compare
7435 equal to any other, and because some object formats cannot even express
7436 the concept of a zero-sized common symbol, as that is how they represent
7437 an ordinary undefined external.
7439 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7440 output the name itself; before and after that, output the additional
7441 assembler syntax for defining the name, and a newline.
7443 This macro controls how the assembler definitions of uninitialized
7444 common global variables are output.
7447 @defmac ASM_OUTPUT_ALIGNED_COMMON (@var{stream}, @var{name}, @var{size}, @var{alignment})
7448 Like @code{ASM_OUTPUT_COMMON} except takes the required alignment as a
7449 separate, explicit argument. If you define this macro, it is used in
7450 place of @code{ASM_OUTPUT_COMMON}, and gives you more flexibility in
7451 handling the required alignment of the variable. The alignment is specified
7452 as the number of bits.
7455 @defmac ASM_OUTPUT_ALIGNED_DECL_COMMON (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7456 Like @code{ASM_OUTPUT_ALIGNED_COMMON} except that @var{decl} of the
7457 variable to be output, if there is one, or @code{NULL_TREE} if there
7458 is no corresponding variable. If you define this macro, GCC will use it
7459 in place of both @code{ASM_OUTPUT_COMMON} and
7460 @code{ASM_OUTPUT_ALIGNED_COMMON}. Define this macro when you need to see
7461 the variable's decl in order to chose what to output.
7464 @defmac ASM_OUTPUT_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{rounded})
7465 A C statement (sans semicolon) to output to the stdio stream
7466 @var{stream} the assembler definition of uninitialized global @var{decl} named
7467 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7468 is the size rounded up to whatever alignment the caller wants.
7470 Try to use function @code{asm_output_bss} defined in @file{varasm.c} when
7471 defining this macro. If unable, use the expression
7472 @code{assemble_name (@var{stream}, @var{name})} to output the name itself;
7473 before and after that, output the additional assembler syntax for defining
7474 the name, and a newline.
7476 There are two ways of handling global BSS@. One is to define either
7477 this macro or its aligned counterpart, @code{ASM_OUTPUT_ALIGNED_BSS}.
7478 The other is to have @code{TARGET_ASM_SELECT_SECTION} return a
7479 switchable BSS section (@pxref{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}).
7480 You do not need to do both.
7482 Some languages do not have @code{common} data, and require a
7483 non-common form of global BSS in order to handle uninitialized globals
7484 efficiently. C++ is one example of this. However, if the target does
7485 not support global BSS, the front end may choose to make globals
7486 common in order to save space in the object file.
7489 @defmac ASM_OUTPUT_ALIGNED_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7490 Like @code{ASM_OUTPUT_BSS} except takes the required alignment as a
7491 separate, explicit argument. If you define this macro, it is used in
7492 place of @code{ASM_OUTPUT_BSS}, and gives you more flexibility in
7493 handling the required alignment of the variable. The alignment is specified
7494 as the number of bits.
7496 Try to use function @code{asm_output_aligned_bss} defined in file
7497 @file{varasm.c} when defining this macro.
7500 @defmac ASM_OUTPUT_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded})
7501 A C statement (sans semicolon) to output to the stdio stream
7502 @var{stream} the assembler definition of a local-common-label named
7503 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7504 is the size rounded up to whatever alignment the caller wants.
7506 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7507 output the name itself; before and after that, output the additional
7508 assembler syntax for defining the name, and a newline.
7510 This macro controls how the assembler definitions of uninitialized
7511 static variables are output.
7514 @defmac ASM_OUTPUT_ALIGNED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{alignment})
7515 Like @code{ASM_OUTPUT_LOCAL} except takes the required alignment as a
7516 separate, explicit argument. If you define this macro, it is used in
7517 place of @code{ASM_OUTPUT_LOCAL}, and gives you more flexibility in
7518 handling the required alignment of the variable. The alignment is specified
7519 as the number of bits.
7522 @defmac ASM_OUTPUT_ALIGNED_DECL_LOCAL (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7523 Like @code{ASM_OUTPUT_ALIGNED_DECL} except that @var{decl} of the
7524 variable to be output, if there is one, or @code{NULL_TREE} if there
7525 is no corresponding variable. If you define this macro, GCC will use it
7526 in place of both @code{ASM_OUTPUT_DECL} and
7527 @code{ASM_OUTPUT_ALIGNED_DECL}. Define this macro when you need to see
7528 the variable's decl in order to chose what to output.
7532 @subsection Output and Generation of Labels
7534 @c prevent bad page break with this line
7535 This is about outputting labels.
7537 @findex assemble_name
7538 @defmac ASM_OUTPUT_LABEL (@var{stream}, @var{name})
7539 A C statement (sans semicolon) to output to the stdio stream
7540 @var{stream} the assembler definition of a label named @var{name}.
7541 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7542 output the name itself; before and after that, output the additional
7543 assembler syntax for defining the name, and a newline. A default
7544 definition of this macro is provided which is correct for most systems.
7547 @findex assemble_name_raw
7548 @defmac ASM_OUTPUT_INTERNAL_LABEL (@var{stream}, @var{name})
7549 Identical to @code{ASM_OUTPUT_LABEL}, except that @var{name} is known
7550 to refer to a compiler-generated label. The default definition uses
7551 @code{assemble_name_raw}, which is like @code{assemble_name} except
7552 that it is more efficient.
7556 A C string containing the appropriate assembler directive to specify the
7557 size of a symbol, without any arguments. On systems that use ELF, the
7558 default (in @file{config/elfos.h}) is @samp{"\t.size\t"}; on other
7559 systems, the default is not to define this macro.
7561 Define this macro only if it is correct to use the default definitions
7562 of @code{ASM_OUTPUT_SIZE_DIRECTIVE} and @code{ASM_OUTPUT_MEASURED_SIZE}
7563 for your system. If you need your own custom definitions of those
7564 macros, or if you do not need explicit symbol sizes at all, do not
7568 @defmac ASM_OUTPUT_SIZE_DIRECTIVE (@var{stream}, @var{name}, @var{size})
7569 A C statement (sans semicolon) to output to the stdio stream
7570 @var{stream} a directive telling the assembler that the size of the
7571 symbol @var{name} is @var{size}. @var{size} is a @code{HOST_WIDE_INT}.
7572 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
7576 @defmac ASM_OUTPUT_MEASURED_SIZE (@var{stream}, @var{name})
7577 A C statement (sans semicolon) to output to the stdio stream
7578 @var{stream} a directive telling the assembler to calculate the size of
7579 the symbol @var{name} by subtracting its address from the current
7582 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
7583 provided. The default assumes that the assembler recognizes a special
7584 @samp{.} symbol as referring to the current address, and can calculate
7585 the difference between this and another symbol. If your assembler does
7586 not recognize @samp{.} or cannot do calculations with it, you will need
7587 to redefine @code{ASM_OUTPUT_MEASURED_SIZE} to use some other technique.
7591 A C string containing the appropriate assembler directive to specify the
7592 type of a symbol, without any arguments. On systems that use ELF, the
7593 default (in @file{config/elfos.h}) is @samp{"\t.type\t"}; on other
7594 systems, the default is not to define this macro.
7596 Define this macro only if it is correct to use the default definition of
7597 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
7598 custom definition of this macro, or if you do not need explicit symbol
7599 types at all, do not define this macro.
7602 @defmac TYPE_OPERAND_FMT
7603 A C string which specifies (using @code{printf} syntax) the format of
7604 the second operand to @code{TYPE_ASM_OP}. On systems that use ELF, the
7605 default (in @file{config/elfos.h}) is @samp{"@@%s"}; on other systems,
7606 the default is not to define this macro.
7608 Define this macro only if it is correct to use the default definition of
7609 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
7610 custom definition of this macro, or if you do not need explicit symbol
7611 types at all, do not define this macro.
7614 @defmac ASM_OUTPUT_TYPE_DIRECTIVE (@var{stream}, @var{type})
7615 A C statement (sans semicolon) to output to the stdio stream
7616 @var{stream} a directive telling the assembler that the type of the
7617 symbol @var{name} is @var{type}. @var{type} is a C string; currently,
7618 that string is always either @samp{"function"} or @samp{"object"}, but
7619 you should not count on this.
7621 If you define @code{TYPE_ASM_OP} and @code{TYPE_OPERAND_FMT}, a default
7622 definition of this macro is provided.
7625 @defmac ASM_DECLARE_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl})
7626 A C statement (sans semicolon) to output to the stdio stream
7627 @var{stream} any text necessary for declaring the name @var{name} of a
7628 function which is being defined. This macro is responsible for
7629 outputting the label definition (perhaps using
7630 @code{ASM_OUTPUT_LABEL}). The argument @var{decl} is the
7631 @code{FUNCTION_DECL} tree node representing the function.
7633 If this macro is not defined, then the function name is defined in the
7634 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
7636 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition
7640 @defmac ASM_DECLARE_FUNCTION_SIZE (@var{stream}, @var{name}, @var{decl})
7641 A C statement (sans semicolon) to output to the stdio stream
7642 @var{stream} any text necessary for declaring the size of a function
7643 which is being defined. The argument @var{name} is the name of the
7644 function. The argument @var{decl} is the @code{FUNCTION_DECL} tree node
7645 representing the function.
7647 If this macro is not defined, then the function size is not defined.
7649 You may wish to use @code{ASM_OUTPUT_MEASURED_SIZE} in the definition
7653 @defmac ASM_DECLARE_OBJECT_NAME (@var{stream}, @var{name}, @var{decl})
7654 A C statement (sans semicolon) to output to the stdio stream
7655 @var{stream} any text necessary for declaring the name @var{name} of an
7656 initialized variable which is being defined. This macro must output the
7657 label definition (perhaps using @code{ASM_OUTPUT_LABEL}). The argument
7658 @var{decl} is the @code{VAR_DECL} tree node representing the variable.
7660 If this macro is not defined, then the variable name is defined in the
7661 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
7663 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} and/or
7664 @code{ASM_OUTPUT_SIZE_DIRECTIVE} in the definition of this macro.
7667 @defmac ASM_DECLARE_CONSTANT_NAME (@var{stream}, @var{name}, @var{exp}, @var{size})
7668 A C statement (sans semicolon) to output to the stdio stream
7669 @var{stream} any text necessary for declaring the name @var{name} of a
7670 constant which is being defined. This macro is responsible for
7671 outputting the label definition (perhaps using
7672 @code{ASM_OUTPUT_LABEL}). The argument @var{exp} is the
7673 value of the constant, and @var{size} is the size of the constant
7674 in bytes. @var{name} will be an internal label.
7676 If this macro is not defined, then the @var{name} is defined in the
7677 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
7679 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition
7683 @defmac ASM_DECLARE_REGISTER_GLOBAL (@var{stream}, @var{decl}, @var{regno}, @var{name})
7684 A C statement (sans semicolon) to output to the stdio stream
7685 @var{stream} any text necessary for claiming a register @var{regno}
7686 for a global variable @var{decl} with name @var{name}.
7688 If you don't define this macro, that is equivalent to defining it to do
7692 @defmac ASM_FINISH_DECLARE_OBJECT (@var{stream}, @var{decl}, @var{toplevel}, @var{atend})
7693 A C statement (sans semicolon) to finish up declaring a variable name
7694 once the compiler has processed its initializer fully and thus has had a
7695 chance to determine the size of an array when controlled by an
7696 initializer. This is used on systems where it's necessary to declare
7697 something about the size of the object.
7699 If you don't define this macro, that is equivalent to defining it to do
7702 You may wish to use @code{ASM_OUTPUT_SIZE_DIRECTIVE} and/or
7703 @code{ASM_OUTPUT_MEASURED_SIZE} in the definition of this macro.
7706 @deftypefn {Target Hook} void TARGET_ASM_GLOBALIZE_LABEL (FILE *@var{stream}, const char *@var{name})
7707 This target hook is a function to output to the stdio stream
7708 @var{stream} some commands that will make the label @var{name} global;
7709 that is, available for reference from other files.
7711 The default implementation relies on a proper definition of
7712 @code{GLOBAL_ASM_OP}.
7715 @deftypefn {Target Hook} void TARGET_ASM_GLOBALIZE_DECL_NAME (FILE *@var{stream}, tree @var{decl})
7716 This target hook is a function to output to the stdio stream
7717 @var{stream} some commands that will make the name associated with @var{decl}
7718 global; that is, available for reference from other files.
7720 The default implementation uses the TARGET_ASM_GLOBALIZE_LABEL target hook.
7723 @defmac ASM_WEAKEN_LABEL (@var{stream}, @var{name})
7724 A C statement (sans semicolon) to output to the stdio stream
7725 @var{stream} some commands that will make the label @var{name} weak;
7726 that is, available for reference from other files but only used if
7727 no other definition is available. Use the expression
7728 @code{assemble_name (@var{stream}, @var{name})} to output the name
7729 itself; before and after that, output the additional assembler syntax
7730 for making that name weak, and a newline.
7732 If you don't define this macro or @code{ASM_WEAKEN_DECL}, GCC will not
7733 support weak symbols and you should not define the @code{SUPPORTS_WEAK}
7737 @defmac ASM_WEAKEN_DECL (@var{stream}, @var{decl}, @var{name}, @var{value})
7738 Combines (and replaces) the function of @code{ASM_WEAKEN_LABEL} and
7739 @code{ASM_OUTPUT_WEAK_ALIAS}, allowing access to the associated function
7740 or variable decl. If @var{value} is not @code{NULL}, this C statement
7741 should output to the stdio stream @var{stream} assembler code which
7742 defines (equates) the weak symbol @var{name} to have the value
7743 @var{value}. If @var{value} is @code{NULL}, it should output commands
7744 to make @var{name} weak.
7747 @defmac ASM_OUTPUT_WEAKREF (@var{stream}, @var{decl}, @var{name}, @var{value})
7748 Outputs a directive that enables @var{name} to be used to refer to
7749 symbol @var{value} with weak-symbol semantics. @code{decl} is the
7750 declaration of @code{name}.
7753 @defmac SUPPORTS_WEAK
7754 A C expression which evaluates to true if the target supports weak symbols.
7756 If you don't define this macro, @file{defaults.h} provides a default
7757 definition. If either @code{ASM_WEAKEN_LABEL} or @code{ASM_WEAKEN_DECL}
7758 is defined, the default definition is @samp{1}; otherwise, it is
7759 @samp{0}. Define this macro if you want to control weak symbol support
7760 with a compiler flag such as @option{-melf}.
7763 @defmac MAKE_DECL_ONE_ONLY (@var{decl})
7764 A C statement (sans semicolon) to mark @var{decl} to be emitted as a
7765 public symbol such that extra copies in multiple translation units will
7766 be discarded by the linker. Define this macro if your object file
7767 format provides support for this concept, such as the @samp{COMDAT}
7768 section flags in the Microsoft Windows PE/COFF format, and this support
7769 requires changes to @var{decl}, such as putting it in a separate section.
7772 @defmac SUPPORTS_ONE_ONLY
7773 A C expression which evaluates to true if the target supports one-only
7776 If you don't define this macro, @file{varasm.c} provides a default
7777 definition. If @code{MAKE_DECL_ONE_ONLY} is defined, the default
7778 definition is @samp{1}; otherwise, it is @samp{0}. Define this macro if
7779 you want to control one-only symbol support with a compiler flag, or if
7780 setting the @code{DECL_ONE_ONLY} flag is enough to mark a declaration to
7781 be emitted as one-only.
7784 @deftypefn {Target Hook} void TARGET_ASM_ASSEMBLE_VISIBILITY (tree @var{decl}, const char *@var{visibility})
7785 This target hook is a function to output to @var{asm_out_file} some
7786 commands that will make the symbol(s) associated with @var{decl} have
7787 hidden, protected or internal visibility as specified by @var{visibility}.
7790 @defmac TARGET_WEAK_NOT_IN_ARCHIVE_TOC
7791 A C expression that evaluates to true if the target's linker expects
7792 that weak symbols do not appear in a static archive's table of contents.
7793 The default is @code{0}.
7795 Leaving weak symbols out of an archive's table of contents means that,
7796 if a symbol will only have a definition in one translation unit and
7797 will have undefined references from other translation units, that
7798 symbol should not be weak. Defining this macro to be nonzero will
7799 thus have the effect that certain symbols that would normally be weak
7800 (explicit template instantiations, and vtables for polymorphic classes
7801 with noninline key methods) will instead be nonweak.
7803 The C++ ABI requires this macro to be zero. Define this macro for
7804 targets where full C++ ABI compliance is impossible and where linker
7805 restrictions require weak symbols to be left out of a static archive's
7809 @defmac ASM_OUTPUT_EXTERNAL (@var{stream}, @var{decl}, @var{name})
7810 A C statement (sans semicolon) to output to the stdio stream
7811 @var{stream} any text necessary for declaring the name of an external
7812 symbol named @var{name} which is referenced in this compilation but
7813 not defined. The value of @var{decl} is the tree node for the
7816 This macro need not be defined if it does not need to output anything.
7817 The GNU assembler and most Unix assemblers don't require anything.
7820 @deftypefn {Target Hook} void TARGET_ASM_EXTERNAL_LIBCALL (rtx @var{symref})
7821 This target hook is a function to output to @var{asm_out_file} an assembler
7822 pseudo-op to declare a library function name external. The name of the
7823 library function is given by @var{symref}, which is a @code{symbol_ref}.
7826 @deftypefn {Target Hook} void TARGET_ASM_MARK_DECL_PRESERVED (tree @var{decl})
7827 This target hook is a function to output to @var{asm_out_file} an assembler
7828 directive to annotate used symbol. Darwin target use .no_dead_code_strip
7832 @defmac ASM_OUTPUT_LABELREF (@var{stream}, @var{name})
7833 A C statement (sans semicolon) to output to the stdio stream
7834 @var{stream} a reference in assembler syntax to a label named
7835 @var{name}. This should add @samp{_} to the front of the name, if that
7836 is customary on your operating system, as it is in most Berkeley Unix
7837 systems. This macro is used in @code{assemble_name}.
7840 @defmac ASM_OUTPUT_SYMBOL_REF (@var{stream}, @var{sym})
7841 A C statement (sans semicolon) to output a reference to
7842 @code{SYMBOL_REF} @var{sym}. If not defined, @code{assemble_name}
7843 will be used to output the name of the symbol. This macro may be used
7844 to modify the way a symbol is referenced depending on information
7845 encoded by @code{TARGET_ENCODE_SECTION_INFO}.
7848 @defmac ASM_OUTPUT_LABEL_REF (@var{stream}, @var{buf})
7849 A C statement (sans semicolon) to output a reference to @var{buf}, the
7850 result of @code{ASM_GENERATE_INTERNAL_LABEL}. If not defined,
7851 @code{assemble_name} will be used to output the name of the symbol.
7852 This macro is not used by @code{output_asm_label}, or the @code{%l}
7853 specifier that calls it; the intention is that this macro should be set
7854 when it is necessary to output a label differently when its address is
7858 @deftypefn {Target Hook} void TARGET_ASM_INTERNAL_LABEL (FILE *@var{stream}, const char *@var{prefix}, unsigned long @var{labelno})
7859 A function to output to the stdio stream @var{stream} a label whose
7860 name is made from the string @var{prefix} and the number @var{labelno}.
7862 It is absolutely essential that these labels be distinct from the labels
7863 used for user-level functions and variables. Otherwise, certain programs
7864 will have name conflicts with internal labels.
7866 It is desirable to exclude internal labels from the symbol table of the
7867 object file. Most assemblers have a naming convention for labels that
7868 should be excluded; on many systems, the letter @samp{L} at the
7869 beginning of a label has this effect. You should find out what
7870 convention your system uses, and follow it.
7872 The default version of this function utilizes @code{ASM_GENERATE_INTERNAL_LABEL}.
7875 @defmac ASM_OUTPUT_DEBUG_LABEL (@var{stream}, @var{prefix}, @var{num})
7876 A C statement to output to the stdio stream @var{stream} a debug info
7877 label whose name is made from the string @var{prefix} and the number
7878 @var{num}. This is useful for VLIW targets, where debug info labels
7879 may need to be treated differently than branch target labels. On some
7880 systems, branch target labels must be at the beginning of instruction
7881 bundles, but debug info labels can occur in the middle of instruction
7884 If this macro is not defined, then @code{(*targetm.asm_out.internal_label)} will be
7888 @defmac ASM_GENERATE_INTERNAL_LABEL (@var{string}, @var{prefix}, @var{num})
7889 A C statement to store into the string @var{string} a label whose name
7890 is made from the string @var{prefix} and the number @var{num}.
7892 This string, when output subsequently by @code{assemble_name}, should
7893 produce the output that @code{(*targetm.asm_out.internal_label)} would produce
7894 with the same @var{prefix} and @var{num}.
7896 If the string begins with @samp{*}, then @code{assemble_name} will
7897 output the rest of the string unchanged. It is often convenient for
7898 @code{ASM_GENERATE_INTERNAL_LABEL} to use @samp{*} in this way. If the
7899 string doesn't start with @samp{*}, then @code{ASM_OUTPUT_LABELREF} gets
7900 to output the string, and may change it. (Of course,
7901 @code{ASM_OUTPUT_LABELREF} is also part of your machine description, so
7902 you should know what it does on your machine.)
7905 @defmac ASM_FORMAT_PRIVATE_NAME (@var{outvar}, @var{name}, @var{number})
7906 A C expression to assign to @var{outvar} (which is a variable of type
7907 @code{char *}) a newly allocated string made from the string
7908 @var{name} and the number @var{number}, with some suitable punctuation
7909 added. Use @code{alloca} to get space for the string.
7911 The string will be used as an argument to @code{ASM_OUTPUT_LABELREF} to
7912 produce an assembler label for an internal static variable whose name is
7913 @var{name}. Therefore, the string must be such as to result in valid
7914 assembler code. The argument @var{number} is different each time this
7915 macro is executed; it prevents conflicts between similarly-named
7916 internal static variables in different scopes.
7918 Ideally this string should not be a valid C identifier, to prevent any
7919 conflict with the user's own symbols. Most assemblers allow periods
7920 or percent signs in assembler symbols; putting at least one of these
7921 between the name and the number will suffice.
7923 If this macro is not defined, a default definition will be provided
7924 which is correct for most systems.
7927 @defmac ASM_OUTPUT_DEF (@var{stream}, @var{name}, @var{value})
7928 A C statement to output to the stdio stream @var{stream} assembler code
7929 which defines (equates) the symbol @var{name} to have the value @var{value}.
7932 If @code{SET_ASM_OP} is defined, a default definition is provided which is
7933 correct for most systems.
7936 @defmac ASM_OUTPUT_DEF_FROM_DECLS (@var{stream}, @var{decl_of_name}, @var{decl_of_value})
7937 A C statement to output to the stdio stream @var{stream} assembler code
7938 which defines (equates) the symbol whose tree node is @var{decl_of_name}
7939 to have the value of the tree node @var{decl_of_value}. This macro will
7940 be used in preference to @samp{ASM_OUTPUT_DEF} if it is defined and if
7941 the tree nodes are available.
7944 If @code{SET_ASM_OP} is defined, a default definition is provided which is
7945 correct for most systems.
7948 @defmac TARGET_DEFERRED_OUTPUT_DEFS (@var{decl_of_name}, @var{decl_of_value})
7949 A C statement that evaluates to true if the assembler code which defines
7950 (equates) the symbol whose tree node is @var{decl_of_name} to have the value
7951 of the tree node @var{decl_of_value} should be emitted near the end of the
7952 current compilation unit. The default is to not defer output of defines.
7953 This macro affects defines output by @samp{ASM_OUTPUT_DEF} and
7954 @samp{ASM_OUTPUT_DEF_FROM_DECLS}.
7957 @defmac ASM_OUTPUT_WEAK_ALIAS (@var{stream}, @var{name}, @var{value})
7958 A C statement to output to the stdio stream @var{stream} assembler code
7959 which defines (equates) the weak symbol @var{name} to have the value
7960 @var{value}. If @var{value} is @code{NULL}, it defines @var{name} as
7961 an undefined weak symbol.
7963 Define this macro if the target only supports weak aliases; define
7964 @code{ASM_OUTPUT_DEF} instead if possible.
7967 @defmac OBJC_GEN_METHOD_LABEL (@var{buf}, @var{is_inst}, @var{class_name}, @var{cat_name}, @var{sel_name})
7968 Define this macro to override the default assembler names used for
7969 Objective-C methods.
7971 The default name is a unique method number followed by the name of the
7972 class (e.g.@: @samp{_1_Foo}). For methods in categories, the name of
7973 the category is also included in the assembler name (e.g.@:
7976 These names are safe on most systems, but make debugging difficult since
7977 the method's selector is not present in the name. Therefore, particular
7978 systems define other ways of computing names.
7980 @var{buf} is an expression of type @code{char *} which gives you a
7981 buffer in which to store the name; its length is as long as
7982 @var{class_name}, @var{cat_name} and @var{sel_name} put together, plus
7983 50 characters extra.
7985 The argument @var{is_inst} specifies whether the method is an instance
7986 method or a class method; @var{class_name} is the name of the class;
7987 @var{cat_name} is the name of the category (or @code{NULL} if the method is not
7988 in a category); and @var{sel_name} is the name of the selector.
7990 On systems where the assembler can handle quoted names, you can use this
7991 macro to provide more human-readable names.
7994 @defmac ASM_DECLARE_CLASS_REFERENCE (@var{stream}, @var{name})
7995 A C statement (sans semicolon) to output to the stdio stream
7996 @var{stream} commands to declare that the label @var{name} is an
7997 Objective-C class reference. This is only needed for targets whose
7998 linkers have special support for NeXT-style runtimes.
8001 @defmac ASM_DECLARE_UNRESOLVED_REFERENCE (@var{stream}, @var{name})
8002 A C statement (sans semicolon) to output to the stdio stream
8003 @var{stream} commands to declare that the label @var{name} is an
8004 unresolved Objective-C class reference. This is only needed for targets
8005 whose linkers have special support for NeXT-style runtimes.
8008 @node Initialization
8009 @subsection How Initialization Functions Are Handled
8010 @cindex initialization routines
8011 @cindex termination routines
8012 @cindex constructors, output of
8013 @cindex destructors, output of
8015 The compiled code for certain languages includes @dfn{constructors}
8016 (also called @dfn{initialization routines})---functions to initialize
8017 data in the program when the program is started. These functions need
8018 to be called before the program is ``started''---that is to say, before
8019 @code{main} is called.
8021 Compiling some languages generates @dfn{destructors} (also called
8022 @dfn{termination routines}) that should be called when the program
8025 To make the initialization and termination functions work, the compiler
8026 must output something in the assembler code to cause those functions to
8027 be called at the appropriate time. When you port the compiler to a new
8028 system, you need to specify how to do this.
8030 There are two major ways that GCC currently supports the execution of
8031 initialization and termination functions. Each way has two variants.
8032 Much of the structure is common to all four variations.
8034 @findex __CTOR_LIST__
8035 @findex __DTOR_LIST__
8036 The linker must build two lists of these functions---a list of
8037 initialization functions, called @code{__CTOR_LIST__}, and a list of
8038 termination functions, called @code{__DTOR_LIST__}.
8040 Each list always begins with an ignored function pointer (which may hold
8041 0, @minus{}1, or a count of the function pointers after it, depending on
8042 the environment). This is followed by a series of zero or more function
8043 pointers to constructors (or destructors), followed by a function
8044 pointer containing zero.
8046 Depending on the operating system and its executable file format, either
8047 @file{crtstuff.c} or @file{libgcc2.c} traverses these lists at startup
8048 time and exit time. Constructors are called in reverse order of the
8049 list; destructors in forward order.
8051 The best way to handle static constructors works only for object file
8052 formats which provide arbitrarily-named sections. A section is set
8053 aside for a list of constructors, and another for a list of destructors.
8054 Traditionally these are called @samp{.ctors} and @samp{.dtors}. Each
8055 object file that defines an initialization function also puts a word in
8056 the constructor section to point to that function. The linker
8057 accumulates all these words into one contiguous @samp{.ctors} section.
8058 Termination functions are handled similarly.
8060 This method will be chosen as the default by @file{target-def.h} if
8061 @code{TARGET_ASM_NAMED_SECTION} is defined. A target that does not
8062 support arbitrary sections, but does support special designated
8063 constructor and destructor sections may define @code{CTORS_SECTION_ASM_OP}
8064 and @code{DTORS_SECTION_ASM_OP} to achieve the same effect.
8066 When arbitrary sections are available, there are two variants, depending
8067 upon how the code in @file{crtstuff.c} is called. On systems that
8068 support a @dfn{.init} section which is executed at program startup,
8069 parts of @file{crtstuff.c} are compiled into that section. The
8070 program is linked by the @command{gcc} driver like this:
8073 ld -o @var{output_file} crti.o crtbegin.o @dots{} -lgcc crtend.o crtn.o
8076 The prologue of a function (@code{__init}) appears in the @code{.init}
8077 section of @file{crti.o}; the epilogue appears in @file{crtn.o}. Likewise
8078 for the function @code{__fini} in the @dfn{.fini} section. Normally these
8079 files are provided by the operating system or by the GNU C library, but
8080 are provided by GCC for a few targets.
8082 The objects @file{crtbegin.o} and @file{crtend.o} are (for most targets)
8083 compiled from @file{crtstuff.c}. They contain, among other things, code
8084 fragments within the @code{.init} and @code{.fini} sections that branch
8085 to routines in the @code{.text} section. The linker will pull all parts
8086 of a section together, which results in a complete @code{__init} function
8087 that invokes the routines we need at startup.
8089 To use this variant, you must define the @code{INIT_SECTION_ASM_OP}
8092 If no init section is available, when GCC compiles any function called
8093 @code{main} (or more accurately, any function designated as a program
8094 entry point by the language front end calling @code{expand_main_function}),
8095 it inserts a procedure call to @code{__main} as the first executable code
8096 after the function prologue. The @code{__main} function is defined
8097 in @file{libgcc2.c} and runs the global constructors.
8099 In file formats that don't support arbitrary sections, there are again
8100 two variants. In the simplest variant, the GNU linker (GNU @code{ld})
8101 and an `a.out' format must be used. In this case,
8102 @code{TARGET_ASM_CONSTRUCTOR} is defined to produce a @code{.stabs}
8103 entry of type @samp{N_SETT}, referencing the name @code{__CTOR_LIST__},
8104 and with the address of the void function containing the initialization
8105 code as its value. The GNU linker recognizes this as a request to add
8106 the value to a @dfn{set}; the values are accumulated, and are eventually
8107 placed in the executable as a vector in the format described above, with
8108 a leading (ignored) count and a trailing zero element.
8109 @code{TARGET_ASM_DESTRUCTOR} is handled similarly. Since no init
8110 section is available, the absence of @code{INIT_SECTION_ASM_OP} causes
8111 the compilation of @code{main} to call @code{__main} as above, starting
8112 the initialization process.
8114 The last variant uses neither arbitrary sections nor the GNU linker.
8115 This is preferable when you want to do dynamic linking and when using
8116 file formats which the GNU linker does not support, such as `ECOFF'@. In
8117 this case, @code{TARGET_HAVE_CTORS_DTORS} is false, initialization and
8118 termination functions are recognized simply by their names. This requires
8119 an extra program in the linkage step, called @command{collect2}. This program
8120 pretends to be the linker, for use with GCC; it does its job by running
8121 the ordinary linker, but also arranges to include the vectors of
8122 initialization and termination functions. These functions are called
8123 via @code{__main} as described above. In order to use this method,
8124 @code{use_collect2} must be defined in the target in @file{config.gcc}.
8127 The following section describes the specific macros that control and
8128 customize the handling of initialization and termination functions.
8131 @node Macros for Initialization
8132 @subsection Macros Controlling Initialization Routines
8134 Here are the macros that control how the compiler handles initialization
8135 and termination functions:
8137 @defmac INIT_SECTION_ASM_OP
8138 If defined, a C string constant, including spacing, for the assembler
8139 operation to identify the following data as initialization code. If not
8140 defined, GCC will assume such a section does not exist. When you are
8141 using special sections for initialization and termination functions, this
8142 macro also controls how @file{crtstuff.c} and @file{libgcc2.c} arrange to
8143 run the initialization functions.
8146 @defmac HAS_INIT_SECTION
8147 If defined, @code{main} will not call @code{__main} as described above.
8148 This macro should be defined for systems that control start-up code
8149 on a symbol-by-symbol basis, such as OSF/1, and should not
8150 be defined explicitly for systems that support @code{INIT_SECTION_ASM_OP}.
8153 @defmac LD_INIT_SWITCH
8154 If defined, a C string constant for a switch that tells the linker that
8155 the following symbol is an initialization routine.
8158 @defmac LD_FINI_SWITCH
8159 If defined, a C string constant for a switch that tells the linker that
8160 the following symbol is a finalization routine.
8163 @defmac COLLECT_SHARED_INIT_FUNC (@var{stream}, @var{func})
8164 If defined, a C statement that will write a function that can be
8165 automatically called when a shared library is loaded. The function
8166 should call @var{func}, which takes no arguments. If not defined, and
8167 the object format requires an explicit initialization function, then a
8168 function called @code{_GLOBAL__DI} will be generated.
8170 This function and the following one are used by collect2 when linking a
8171 shared library that needs constructors or destructors, or has DWARF2
8172 exception tables embedded in the code.
8175 @defmac COLLECT_SHARED_FINI_FUNC (@var{stream}, @var{func})
8176 If defined, a C statement that will write a function that can be
8177 automatically called when a shared library is unloaded. The function
8178 should call @var{func}, which takes no arguments. If not defined, and
8179 the object format requires an explicit finalization function, then a
8180 function called @code{_GLOBAL__DD} will be generated.
8183 @defmac INVOKE__main
8184 If defined, @code{main} will call @code{__main} despite the presence of
8185 @code{INIT_SECTION_ASM_OP}. This macro should be defined for systems
8186 where the init section is not actually run automatically, but is still
8187 useful for collecting the lists of constructors and destructors.
8190 @defmac SUPPORTS_INIT_PRIORITY
8191 If nonzero, the C++ @code{init_priority} attribute is supported and the
8192 compiler should emit instructions to control the order of initialization
8193 of objects. If zero, the compiler will issue an error message upon
8194 encountering an @code{init_priority} attribute.
8197 @deftypefn {Target Hook} bool TARGET_HAVE_CTORS_DTORS
8198 This value is true if the target supports some ``native'' method of
8199 collecting constructors and destructors to be run at startup and exit.
8200 It is false if we must use @command{collect2}.
8203 @deftypefn {Target Hook} void TARGET_ASM_CONSTRUCTOR (rtx @var{symbol}, int @var{priority})
8204 If defined, a function that outputs assembler code to arrange to call
8205 the function referenced by @var{symbol} at initialization time.
8207 Assume that @var{symbol} is a @code{SYMBOL_REF} for a function taking
8208 no arguments and with no return value. If the target supports initialization
8209 priorities, @var{priority} is a value between 0 and @code{MAX_INIT_PRIORITY};
8210 otherwise it must be @code{DEFAULT_INIT_PRIORITY}.
8212 If this macro is not defined by the target, a suitable default will
8213 be chosen if (1) the target supports arbitrary section names, (2) the
8214 target defines @code{CTORS_SECTION_ASM_OP}, or (3) @code{USE_COLLECT2}
8218 @deftypefn {Target Hook} void TARGET_ASM_DESTRUCTOR (rtx @var{symbol}, int @var{priority})
8219 This is like @code{TARGET_ASM_CONSTRUCTOR} but used for termination
8220 functions rather than initialization functions.
8223 If @code{TARGET_HAVE_CTORS_DTORS} is true, the initialization routine
8224 generated for the generated object file will have static linkage.
8226 If your system uses @command{collect2} as the means of processing
8227 constructors, then that program normally uses @command{nm} to scan
8228 an object file for constructor functions to be called.
8230 On certain kinds of systems, you can define this macro to make
8231 @command{collect2} work faster (and, in some cases, make it work at all):
8233 @defmac OBJECT_FORMAT_COFF
8234 Define this macro if the system uses COFF (Common Object File Format)
8235 object files, so that @command{collect2} can assume this format and scan
8236 object files directly for dynamic constructor/destructor functions.
8238 This macro is effective only in a native compiler; @command{collect2} as
8239 part of a cross compiler always uses @command{nm} for the target machine.
8242 @defmac REAL_NM_FILE_NAME
8243 Define this macro as a C string constant containing the file name to use
8244 to execute @command{nm}. The default is to search the path normally for
8247 If your system supports shared libraries and has a program to list the
8248 dynamic dependencies of a given library or executable, you can define
8249 these macros to enable support for running initialization and
8250 termination functions in shared libraries:
8254 Define this macro to a C string constant containing the name of the program
8255 which lists dynamic dependencies, like @command{"ldd"} under SunOS 4.
8258 @defmac PARSE_LDD_OUTPUT (@var{ptr})
8259 Define this macro to be C code that extracts filenames from the output
8260 of the program denoted by @code{LDD_SUFFIX}. @var{ptr} is a variable
8261 of type @code{char *} that points to the beginning of a line of output
8262 from @code{LDD_SUFFIX}. If the line lists a dynamic dependency, the
8263 code must advance @var{ptr} to the beginning of the filename on that
8264 line. Otherwise, it must set @var{ptr} to @code{NULL}.
8267 @defmac SHLIB_SUFFIX
8268 Define this macro to a C string constant containing the default shared
8269 library extension of the target (e.g., @samp{".so"}). @command{collect2}
8270 strips version information after this suffix when generating global
8271 constructor and destructor names. This define is only needed on targets
8272 that use @command{collect2} to process constructors and destructors.
8275 @node Instruction Output
8276 @subsection Output of Assembler Instructions
8278 @c prevent bad page break with this line
8279 This describes assembler instruction output.
8281 @defmac REGISTER_NAMES
8282 A C initializer containing the assembler's names for the machine
8283 registers, each one as a C string constant. This is what translates
8284 register numbers in the compiler into assembler language.
8287 @defmac ADDITIONAL_REGISTER_NAMES
8288 If defined, a C initializer for an array of structures containing a name
8289 and a register number. This macro defines additional names for hard
8290 registers, thus allowing the @code{asm} option in declarations to refer
8291 to registers using alternate names.
8294 @defmac ASM_OUTPUT_OPCODE (@var{stream}, @var{ptr})
8295 Define this macro if you are using an unusual assembler that
8296 requires different names for the machine instructions.
8298 The definition is a C statement or statements which output an
8299 assembler instruction opcode to the stdio stream @var{stream}. The
8300 macro-operand @var{ptr} is a variable of type @code{char *} which
8301 points to the opcode name in its ``internal'' form---the form that is
8302 written in the machine description. The definition should output the
8303 opcode name to @var{stream}, performing any translation you desire, and
8304 increment the variable @var{ptr} to point at the end of the opcode
8305 so that it will not be output twice.
8307 In fact, your macro definition may process less than the entire opcode
8308 name, or more than the opcode name; but if you want to process text
8309 that includes @samp{%}-sequences to substitute operands, you must take
8310 care of the substitution yourself. Just be sure to increment
8311 @var{ptr} over whatever text should not be output normally.
8313 @findex recog_data.operand
8314 If you need to look at the operand values, they can be found as the
8315 elements of @code{recog_data.operand}.
8317 If the macro definition does nothing, the instruction is output
8321 @defmac FINAL_PRESCAN_INSN (@var{insn}, @var{opvec}, @var{noperands})
8322 If defined, a C statement to be executed just prior to the output of
8323 assembler code for @var{insn}, to modify the extracted operands so
8324 they will be output differently.
8326 Here the argument @var{opvec} is the vector containing the operands
8327 extracted from @var{insn}, and @var{noperands} is the number of
8328 elements of the vector which contain meaningful data for this insn.
8329 The contents of this vector are what will be used to convert the insn
8330 template into assembler code, so you can change the assembler output
8331 by changing the contents of the vector.
8333 This macro is useful when various assembler syntaxes share a single
8334 file of instruction patterns; by defining this macro differently, you
8335 can cause a large class of instructions to be output differently (such
8336 as with rearranged operands). Naturally, variations in assembler
8337 syntax affecting individual insn patterns ought to be handled by
8338 writing conditional output routines in those patterns.
8340 If this macro is not defined, it is equivalent to a null statement.
8343 @deftypefn {Target Hook} void TARGET_ASM_FINAL_POSTSCAN_INSN (FILE *@var{FILE}, rtx @var{insn}, rtx *@var{opvec}, int @var{noperands})
8344 If defined, this target hook is a function which is executed just after the
8345 output of assembler code for @var{insn}, to change the mode of the assembler
8348 Here the argument @var{opvec} is the vector containing the operands
8349 extracted from @var{insn}, and @var{noperands} is the number of
8350 elements of the vector which contain meaningful data for this insn.
8351 The contents of this vector are what was used to convert the insn
8352 template into assembler code, so you can change the assembler mode
8353 by checking the contents of the vector.
8356 @defmac PRINT_OPERAND (@var{stream}, @var{x}, @var{code})
8357 A C compound statement to output to stdio stream @var{stream} the
8358 assembler syntax for an instruction operand @var{x}. @var{x} is an
8361 @var{code} is a value that can be used to specify one of several ways
8362 of printing the operand. It is used when identical operands must be
8363 printed differently depending on the context. @var{code} comes from
8364 the @samp{%} specification that was used to request printing of the
8365 operand. If the specification was just @samp{%@var{digit}} then
8366 @var{code} is 0; if the specification was @samp{%@var{ltr}
8367 @var{digit}} then @var{code} is the ASCII code for @var{ltr}.
8370 If @var{x} is a register, this macro should print the register's name.
8371 The names can be found in an array @code{reg_names} whose type is
8372 @code{char *[]}. @code{reg_names} is initialized from
8373 @code{REGISTER_NAMES}.
8375 When the machine description has a specification @samp{%@var{punct}}
8376 (a @samp{%} followed by a punctuation character), this macro is called
8377 with a null pointer for @var{x} and the punctuation character for
8381 @defmac PRINT_OPERAND_PUNCT_VALID_P (@var{code})
8382 A C expression which evaluates to true if @var{code} is a valid
8383 punctuation character for use in the @code{PRINT_OPERAND} macro. If
8384 @code{PRINT_OPERAND_PUNCT_VALID_P} is not defined, it means that no
8385 punctuation characters (except for the standard one, @samp{%}) are used
8389 @defmac PRINT_OPERAND_ADDRESS (@var{stream}, @var{x})
8390 A C compound statement to output to stdio stream @var{stream} the
8391 assembler syntax for an instruction operand that is a memory reference
8392 whose address is @var{x}. @var{x} is an RTL expression.
8394 @cindex @code{TARGET_ENCODE_SECTION_INFO} usage
8395 On some machines, the syntax for a symbolic address depends on the
8396 section that the address refers to. On these machines, define the hook
8397 @code{TARGET_ENCODE_SECTION_INFO} to store the information into the
8398 @code{symbol_ref}, and then check for it here. @xref{Assembler
8402 @findex dbr_sequence_length
8403 @defmac DBR_OUTPUT_SEQEND (@var{file})
8404 A C statement, to be executed after all slot-filler instructions have
8405 been output. If necessary, call @code{dbr_sequence_length} to
8406 determine the number of slots filled in a sequence (zero if not
8407 currently outputting a sequence), to decide how many no-ops to output,
8410 Don't define this macro if it has nothing to do, but it is helpful in
8411 reading assembly output if the extent of the delay sequence is made
8412 explicit (e.g.@: with white space).
8415 @findex final_sequence
8416 Note that output routines for instructions with delay slots must be
8417 prepared to deal with not being output as part of a sequence
8418 (i.e.@: when the scheduling pass is not run, or when no slot fillers could be
8419 found.) The variable @code{final_sequence} is null when not
8420 processing a sequence, otherwise it contains the @code{sequence} rtx
8424 @defmac REGISTER_PREFIX
8425 @defmacx LOCAL_LABEL_PREFIX
8426 @defmacx USER_LABEL_PREFIX
8427 @defmacx IMMEDIATE_PREFIX
8428 If defined, C string expressions to be used for the @samp{%R}, @samp{%L},
8429 @samp{%U}, and @samp{%I} options of @code{asm_fprintf} (see
8430 @file{final.c}). These are useful when a single @file{md} file must
8431 support multiple assembler formats. In that case, the various @file{tm.h}
8432 files can define these macros differently.
8435 @defmac ASM_FPRINTF_EXTENSIONS (@var{file}, @var{argptr}, @var{format})
8436 If defined this macro should expand to a series of @code{case}
8437 statements which will be parsed inside the @code{switch} statement of
8438 the @code{asm_fprintf} function. This allows targets to define extra
8439 printf formats which may useful when generating their assembler
8440 statements. Note that uppercase letters are reserved for future
8441 generic extensions to asm_fprintf, and so are not available to target
8442 specific code. The output file is given by the parameter @var{file}.
8443 The varargs input pointer is @var{argptr} and the rest of the format
8444 string, starting the character after the one that is being switched
8445 upon, is pointed to by @var{format}.
8448 @defmac ASSEMBLER_DIALECT
8449 If your target supports multiple dialects of assembler language (such as
8450 different opcodes), define this macro as a C expression that gives the
8451 numeric index of the assembler language dialect to use, with zero as the
8454 If this macro is defined, you may use constructs of the form
8456 @samp{@{option0|option1|option2@dots{}@}}
8459 in the output templates of patterns (@pxref{Output Template}) or in the
8460 first argument of @code{asm_fprintf}. This construct outputs
8461 @samp{option0}, @samp{option1}, @samp{option2}, etc., if the value of
8462 @code{ASSEMBLER_DIALECT} is zero, one, two, etc. Any special characters
8463 within these strings retain their usual meaning. If there are fewer
8464 alternatives within the braces than the value of
8465 @code{ASSEMBLER_DIALECT}, the construct outputs nothing.
8467 If you do not define this macro, the characters @samp{@{}, @samp{|} and
8468 @samp{@}} do not have any special meaning when used in templates or
8469 operands to @code{asm_fprintf}.
8471 Define the macros @code{REGISTER_PREFIX}, @code{LOCAL_LABEL_PREFIX},
8472 @code{USER_LABEL_PREFIX} and @code{IMMEDIATE_PREFIX} if you can express
8473 the variations in assembler language syntax with that mechanism. Define
8474 @code{ASSEMBLER_DIALECT} and use the @samp{@{option0|option1@}} syntax
8475 if the syntax variant are larger and involve such things as different
8476 opcodes or operand order.
8479 @defmac ASM_OUTPUT_REG_PUSH (@var{stream}, @var{regno})
8480 A C expression to output to @var{stream} some assembler code
8481 which will push hard register number @var{regno} onto the stack.
8482 The code need not be optimal, since this macro is used only when
8486 @defmac ASM_OUTPUT_REG_POP (@var{stream}, @var{regno})
8487 A C expression to output to @var{stream} some assembler code
8488 which will pop hard register number @var{regno} off of the stack.
8489 The code need not be optimal, since this macro is used only when
8493 @node Dispatch Tables
8494 @subsection Output of Dispatch Tables
8496 @c prevent bad page break with this line
8497 This concerns dispatch tables.
8499 @cindex dispatch table
8500 @defmac ASM_OUTPUT_ADDR_DIFF_ELT (@var{stream}, @var{body}, @var{value}, @var{rel})
8501 A C statement to output to the stdio stream @var{stream} an assembler
8502 pseudo-instruction to generate a difference between two labels.
8503 @var{value} and @var{rel} are the numbers of two internal labels. The
8504 definitions of these labels are output using
8505 @code{(*targetm.asm_out.internal_label)}, and they must be printed in the same
8506 way here. For example,
8509 fprintf (@var{stream}, "\t.word L%d-L%d\n",
8510 @var{value}, @var{rel})
8513 You must provide this macro on machines where the addresses in a
8514 dispatch table are relative to the table's own address. If defined, GCC
8515 will also use this macro on all machines when producing PIC@.
8516 @var{body} is the body of the @code{ADDR_DIFF_VEC}; it is provided so that the
8517 mode and flags can be read.
8520 @defmac ASM_OUTPUT_ADDR_VEC_ELT (@var{stream}, @var{value})
8521 This macro should be provided on machines where the addresses
8522 in a dispatch table are absolute.
8524 The definition should be a C statement to output to the stdio stream
8525 @var{stream} an assembler pseudo-instruction to generate a reference to
8526 a label. @var{value} is the number of an internal label whose
8527 definition is output using @code{(*targetm.asm_out.internal_label)}.
8531 fprintf (@var{stream}, "\t.word L%d\n", @var{value})
8535 @defmac ASM_OUTPUT_CASE_LABEL (@var{stream}, @var{prefix}, @var{num}, @var{table})
8536 Define this if the label before a jump-table needs to be output
8537 specially. The first three arguments are the same as for
8538 @code{(*targetm.asm_out.internal_label)}; the fourth argument is the
8539 jump-table which follows (a @code{jump_insn} containing an
8540 @code{addr_vec} or @code{addr_diff_vec}).
8542 This feature is used on system V to output a @code{swbeg} statement
8545 If this macro is not defined, these labels are output with
8546 @code{(*targetm.asm_out.internal_label)}.
8549 @defmac ASM_OUTPUT_CASE_END (@var{stream}, @var{num}, @var{table})
8550 Define this if something special must be output at the end of a
8551 jump-table. The definition should be a C statement to be executed
8552 after the assembler code for the table is written. It should write
8553 the appropriate code to stdio stream @var{stream}. The argument
8554 @var{table} is the jump-table insn, and @var{num} is the label-number
8555 of the preceding label.
8557 If this macro is not defined, nothing special is output at the end of
8561 @deftypefn {Target Hook} void TARGET_ASM_EMIT_UNWIND_LABEL (@var{stream}, @var{decl}, @var{for_eh}, @var{empty})
8562 This target hook emits a label at the beginning of each FDE@. It
8563 should be defined on targets where FDEs need special labels, and it
8564 should write the appropriate label, for the FDE associated with the
8565 function declaration @var{decl}, to the stdio stream @var{stream}.
8566 The third argument, @var{for_eh}, is a boolean: true if this is for an
8567 exception table. The fourth argument, @var{empty}, is a boolean:
8568 true if this is a placeholder label for an omitted FDE@.
8570 The default is that FDEs are not given nonlocal labels.
8573 @deftypefn {Target Hook} void TARGET_ASM_EMIT_EXCEPT_TABLE_LABEL (@var{stream})
8574 This target hook emits a label at the beginning of the exception table.
8575 It should be defined on targets where it is desirable for the table
8576 to be broken up according to function.
8578 The default is that no label is emitted.
8581 @deftypefn {Target Hook} void TARGET_UNWIND_EMIT (FILE * @var{stream}, rtx @var{insn})
8582 This target hook emits and assembly directives required to unwind the
8583 given instruction. This is only used when TARGET_UNWIND_INFO is set.
8586 @node Exception Region Output
8587 @subsection Assembler Commands for Exception Regions
8589 @c prevent bad page break with this line
8591 This describes commands marking the start and the end of an exception
8594 @defmac EH_FRAME_SECTION_NAME
8595 If defined, a C string constant for the name of the section containing
8596 exception handling frame unwind information. If not defined, GCC will
8597 provide a default definition if the target supports named sections.
8598 @file{crtstuff.c} uses this macro to switch to the appropriate section.
8600 You should define this symbol if your target supports DWARF 2 frame
8601 unwind information and the default definition does not work.
8604 @defmac EH_FRAME_IN_DATA_SECTION
8605 If defined, DWARF 2 frame unwind information will be placed in the
8606 data section even though the target supports named sections. This
8607 might be necessary, for instance, if the system linker does garbage
8608 collection and sections cannot be marked as not to be collected.
8610 Do not define this macro unless @code{TARGET_ASM_NAMED_SECTION} is
8614 @defmac EH_TABLES_CAN_BE_READ_ONLY
8615 Define this macro to 1 if your target is such that no frame unwind
8616 information encoding used with non-PIC code will ever require a
8617 runtime relocation, but the linker may not support merging read-only
8618 and read-write sections into a single read-write section.
8621 @defmac MASK_RETURN_ADDR
8622 An rtx used to mask the return address found via @code{RETURN_ADDR_RTX}, so
8623 that it does not contain any extraneous set bits in it.
8626 @defmac DWARF2_UNWIND_INFO
8627 Define this macro to 0 if your target supports DWARF 2 frame unwind
8628 information, but it does not yet work with exception handling.
8629 Otherwise, if your target supports this information (if it defines
8630 @samp{INCOMING_RETURN_ADDR_RTX} and either @samp{UNALIGNED_INT_ASM_OP}
8631 or @samp{OBJECT_FORMAT_ELF}), GCC will provide a default definition of 1.
8633 If @code{TARGET_UNWIND_INFO} is defined, the target specific unwinder
8634 will be used in all cases. Defining this macro will enable the generation
8635 of DWARF 2 frame debugging information.
8637 If @code{TARGET_UNWIND_INFO} is not defined, and this macro is defined to 1,
8638 the DWARF 2 unwinder will be the default exception handling mechanism;
8639 otherwise, the @code{setjmp}/@code{longjmp}-based scheme will be used by
8643 @defmac TARGET_UNWIND_INFO
8644 Define this macro if your target has ABI specified unwind tables. Usually
8645 these will be output by @code{TARGET_UNWIND_EMIT}.
8648 @deftypevr {Target Hook} bool TARGET_UNWIND_TABLES_DEFAULT
8649 This variable should be set to @code{true} if the target ABI requires unwinding
8650 tables even when exceptions are not used.
8653 @defmac MUST_USE_SJLJ_EXCEPTIONS
8654 This macro need only be defined if @code{DWARF2_UNWIND_INFO} is
8655 runtime-variable. In that case, @file{except.h} cannot correctly
8656 determine the corresponding definition of @code{MUST_USE_SJLJ_EXCEPTIONS},
8657 so the target must provide it directly.
8660 @defmac DONT_USE_BUILTIN_SETJMP
8661 Define this macro to 1 if the @code{setjmp}/@code{longjmp}-based scheme
8662 should use the @code{setjmp}/@code{longjmp} functions from the C library
8663 instead of the @code{__builtin_setjmp}/@code{__builtin_longjmp} machinery.
8666 @defmac DWARF_CIE_DATA_ALIGNMENT
8667 This macro need only be defined if the target might save registers in the
8668 function prologue at an offset to the stack pointer that is not aligned to
8669 @code{UNITS_PER_WORD}. The definition should be the negative minimum
8670 alignment if @code{STACK_GROWS_DOWNWARD} is defined, and the positive
8671 minimum alignment otherwise. @xref{SDB and DWARF}. Only applicable if
8672 the target supports DWARF 2 frame unwind information.
8675 @deftypevr {Target Hook} bool TARGET_TERMINATE_DW2_EH_FRAME_INFO
8676 Contains the value true if the target should add a zero word onto the
8677 end of a Dwarf-2 frame info section when used for exception handling.
8678 Default value is false if @code{EH_FRAME_SECTION_NAME} is defined, and
8682 @deftypefn {Target Hook} rtx TARGET_DWARF_REGISTER_SPAN (rtx @var{reg})
8683 Given a register, this hook should return a parallel of registers to
8684 represent where to find the register pieces. Define this hook if the
8685 register and its mode are represented in Dwarf in non-contiguous
8686 locations, or if the register should be represented in more than one
8687 register in Dwarf. Otherwise, this hook should return @code{NULL_RTX}.
8688 If not defined, the default is to return @code{NULL_RTX}.
8691 @deftypefn {Target Hook} void TARGET_INIT_DWARF_REG_SIZES_EXTRA (tree @var{address})
8692 If some registers are represented in Dwarf-2 unwind information in
8693 multiple pieces, define this hook to fill in information about the
8694 sizes of those pieces in the table used by the unwinder at runtime.
8695 It will be called by @code{expand_builtin_init_dwarf_reg_sizes} after
8696 filling in a single size corresponding to each hard register;
8697 @var{address} is the address of the table.
8700 @deftypefn {Target Hook} bool TARGET_ASM_TTYPE (rtx @var{sym})
8701 This hook is used to output a reference from a frame unwinding table to
8702 the type_info object identified by @var{sym}. It should return @code{true}
8703 if the reference was output. Returning @code{false} will cause the
8704 reference to be output using the normal Dwarf2 routines.
8707 @deftypefn {Target Hook} bool TARGET_ARM_EABI_UNWINDER
8708 This hook should be set to @code{true} on targets that use an ARM EABI
8709 based unwinding library, and @code{false} on other targets. This effects
8710 the format of unwinding tables, and how the unwinder in entered after
8711 running a cleanup. The default is @code{false}.
8714 @node Alignment Output
8715 @subsection Assembler Commands for Alignment
8717 @c prevent bad page break with this line
8718 This describes commands for alignment.
8720 @defmac JUMP_ALIGN (@var{label})
8721 The alignment (log base 2) to put in front of @var{label}, which is
8722 a common destination of jumps and has no fallthru incoming edge.
8724 This macro need not be defined if you don't want any special alignment
8725 to be done at such a time. Most machine descriptions do not currently
8728 Unless it's necessary to inspect the @var{label} parameter, it is better
8729 to set the variable @var{align_jumps} in the target's
8730 @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's
8731 selection in @var{align_jumps} in a @code{JUMP_ALIGN} implementation.
8734 @defmac LABEL_ALIGN_AFTER_BARRIER (@var{label})
8735 The alignment (log base 2) to put in front of @var{label}, which follows
8738 This macro need not be defined if you don't want any special alignment
8739 to be done at such a time. Most machine descriptions do not currently
8743 @defmac LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP
8744 The maximum number of bytes to skip when applying
8745 @code{LABEL_ALIGN_AFTER_BARRIER}. This works only if
8746 @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
8749 @defmac LOOP_ALIGN (@var{label})
8750 The alignment (log base 2) to put in front of @var{label}, which follows
8751 a @code{NOTE_INSN_LOOP_BEG} note.
8753 This macro need not be defined if you don't want any special alignment
8754 to be done at such a time. Most machine descriptions do not currently
8757 Unless it's necessary to inspect the @var{label} parameter, it is better
8758 to set the variable @code{align_loops} in the target's
8759 @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's
8760 selection in @code{align_loops} in a @code{LOOP_ALIGN} implementation.
8763 @defmac LOOP_ALIGN_MAX_SKIP
8764 The maximum number of bytes to skip when applying @code{LOOP_ALIGN}.
8765 This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
8768 @defmac LABEL_ALIGN (@var{label})
8769 The alignment (log base 2) to put in front of @var{label}.
8770 If @code{LABEL_ALIGN_AFTER_BARRIER} / @code{LOOP_ALIGN} specify a different alignment,
8771 the maximum of the specified values is used.
8773 Unless it's necessary to inspect the @var{label} parameter, it is better
8774 to set the variable @code{align_labels} in the target's
8775 @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's
8776 selection in @code{align_labels} in a @code{LABEL_ALIGN} implementation.
8779 @defmac LABEL_ALIGN_MAX_SKIP
8780 The maximum number of bytes to skip when applying @code{LABEL_ALIGN}.
8781 This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
8784 @defmac ASM_OUTPUT_SKIP (@var{stream}, @var{nbytes})
8785 A C statement to output to the stdio stream @var{stream} an assembler
8786 instruction to advance the location counter by @var{nbytes} bytes.
8787 Those bytes should be zero when loaded. @var{nbytes} will be a C
8788 expression of type @code{unsigned HOST_WIDE_INT}.
8791 @defmac ASM_NO_SKIP_IN_TEXT
8792 Define this macro if @code{ASM_OUTPUT_SKIP} should not be used in the
8793 text section because it fails to put zeros in the bytes that are skipped.
8794 This is true on many Unix systems, where the pseudo--op to skip bytes
8795 produces no-op instructions rather than zeros when used in the text
8799 @defmac ASM_OUTPUT_ALIGN (@var{stream}, @var{power})
8800 A C statement to output to the stdio stream @var{stream} an assembler
8801 command to advance the location counter to a multiple of 2 to the
8802 @var{power} bytes. @var{power} will be a C expression of type @code{int}.
8805 @defmac ASM_OUTPUT_ALIGN_WITH_NOP (@var{stream}, @var{power})
8806 Like @code{ASM_OUTPUT_ALIGN}, except that the ``nop'' instruction is used
8807 for padding, if necessary.
8810 @defmac ASM_OUTPUT_MAX_SKIP_ALIGN (@var{stream}, @var{power}, @var{max_skip})
8811 A C statement to output to the stdio stream @var{stream} an assembler
8812 command to advance the location counter to a multiple of 2 to the
8813 @var{power} bytes, but only if @var{max_skip} or fewer bytes are needed to
8814 satisfy the alignment request. @var{power} and @var{max_skip} will be
8815 a C expression of type @code{int}.
8819 @node Debugging Info
8820 @section Controlling Debugging Information Format
8822 @c prevent bad page break with this line
8823 This describes how to specify debugging information.
8826 * All Debuggers:: Macros that affect all debugging formats uniformly.
8827 * DBX Options:: Macros enabling specific options in DBX format.
8828 * DBX Hooks:: Hook macros for varying DBX format.
8829 * File Names and DBX:: Macros controlling output of file names in DBX format.
8830 * SDB and DWARF:: Macros for SDB (COFF) and DWARF formats.
8831 * VMS Debug:: Macros for VMS debug format.
8835 @subsection Macros Affecting All Debugging Formats
8837 @c prevent bad page break with this line
8838 These macros affect all debugging formats.
8840 @defmac DBX_REGISTER_NUMBER (@var{regno})
8841 A C expression that returns the DBX register number for the compiler
8842 register number @var{regno}. In the default macro provided, the value
8843 of this expression will be @var{regno} itself. But sometimes there are
8844 some registers that the compiler knows about and DBX does not, or vice
8845 versa. In such cases, some register may need to have one number in the
8846 compiler and another for DBX@.
8848 If two registers have consecutive numbers inside GCC, and they can be
8849 used as a pair to hold a multiword value, then they @emph{must} have
8850 consecutive numbers after renumbering with @code{DBX_REGISTER_NUMBER}.
8851 Otherwise, debuggers will be unable to access such a pair, because they
8852 expect register pairs to be consecutive in their own numbering scheme.
8854 If you find yourself defining @code{DBX_REGISTER_NUMBER} in way that
8855 does not preserve register pairs, then what you must do instead is
8856 redefine the actual register numbering scheme.
8859 @defmac DEBUGGER_AUTO_OFFSET (@var{x})
8860 A C expression that returns the integer offset value for an automatic
8861 variable having address @var{x} (an RTL expression). The default
8862 computation assumes that @var{x} is based on the frame-pointer and
8863 gives the offset from the frame-pointer. This is required for targets
8864 that produce debugging output for DBX or COFF-style debugging output
8865 for SDB and allow the frame-pointer to be eliminated when the
8866 @option{-g} options is used.
8869 @defmac DEBUGGER_ARG_OFFSET (@var{offset}, @var{x})
8870 A C expression that returns the integer offset value for an argument
8871 having address @var{x} (an RTL expression). The nominal offset is
8875 @defmac PREFERRED_DEBUGGING_TYPE
8876 A C expression that returns the type of debugging output GCC should
8877 produce when the user specifies just @option{-g}. Define
8878 this if you have arranged for GCC to support more than one format of
8879 debugging output. Currently, the allowable values are @code{DBX_DEBUG},
8880 @code{SDB_DEBUG}, @code{DWARF_DEBUG}, @code{DWARF2_DEBUG},
8881 @code{XCOFF_DEBUG}, @code{VMS_DEBUG}, and @code{VMS_AND_DWARF2_DEBUG}.
8883 When the user specifies @option{-ggdb}, GCC normally also uses the
8884 value of this macro to select the debugging output format, but with two
8885 exceptions. If @code{DWARF2_DEBUGGING_INFO} is defined, GCC uses the
8886 value @code{DWARF2_DEBUG}. Otherwise, if @code{DBX_DEBUGGING_INFO} is
8887 defined, GCC uses @code{DBX_DEBUG}.
8889 The value of this macro only affects the default debugging output; the
8890 user can always get a specific type of output by using @option{-gstabs},
8891 @option{-gcoff}, @option{-gdwarf-2}, @option{-gxcoff}, or @option{-gvms}.
8895 @subsection Specific Options for DBX Output
8897 @c prevent bad page break with this line
8898 These are specific options for DBX output.
8900 @defmac DBX_DEBUGGING_INFO
8901 Define this macro if GCC should produce debugging output for DBX
8902 in response to the @option{-g} option.
8905 @defmac XCOFF_DEBUGGING_INFO
8906 Define this macro if GCC should produce XCOFF format debugging output
8907 in response to the @option{-g} option. This is a variant of DBX format.
8910 @defmac DEFAULT_GDB_EXTENSIONS
8911 Define this macro to control whether GCC should by default generate
8912 GDB's extended version of DBX debugging information (assuming DBX-format
8913 debugging information is enabled at all). If you don't define the
8914 macro, the default is 1: always generate the extended information
8915 if there is any occasion to.
8918 @defmac DEBUG_SYMS_TEXT
8919 Define this macro if all @code{.stabs} commands should be output while
8920 in the text section.
8923 @defmac ASM_STABS_OP
8924 A C string constant, including spacing, naming the assembler pseudo op to
8925 use instead of @code{"\t.stabs\t"} to define an ordinary debugging symbol.
8926 If you don't define this macro, @code{"\t.stabs\t"} is used. This macro
8927 applies only to DBX debugging information format.
8930 @defmac ASM_STABD_OP
8931 A C string constant, including spacing, naming the assembler pseudo op to
8932 use instead of @code{"\t.stabd\t"} to define a debugging symbol whose
8933 value is the current location. If you don't define this macro,
8934 @code{"\t.stabd\t"} is used. This macro applies only to DBX debugging
8938 @defmac ASM_STABN_OP
8939 A C string constant, including spacing, naming the assembler pseudo op to
8940 use instead of @code{"\t.stabn\t"} to define a debugging symbol with no
8941 name. If you don't define this macro, @code{"\t.stabn\t"} is used. This
8942 macro applies only to DBX debugging information format.
8945 @defmac DBX_NO_XREFS
8946 Define this macro if DBX on your system does not support the construct
8947 @samp{xs@var{tagname}}. On some systems, this construct is used to
8948 describe a forward reference to a structure named @var{tagname}.
8949 On other systems, this construct is not supported at all.
8952 @defmac DBX_CONTIN_LENGTH
8953 A symbol name in DBX-format debugging information is normally
8954 continued (split into two separate @code{.stabs} directives) when it
8955 exceeds a certain length (by default, 80 characters). On some
8956 operating systems, DBX requires this splitting; on others, splitting
8957 must not be done. You can inhibit splitting by defining this macro
8958 with the value zero. You can override the default splitting-length by
8959 defining this macro as an expression for the length you desire.
8962 @defmac DBX_CONTIN_CHAR
8963 Normally continuation is indicated by adding a @samp{\} character to
8964 the end of a @code{.stabs} string when a continuation follows. To use
8965 a different character instead, define this macro as a character
8966 constant for the character you want to use. Do not define this macro
8967 if backslash is correct for your system.
8970 @defmac DBX_STATIC_STAB_DATA_SECTION
8971 Define this macro if it is necessary to go to the data section before
8972 outputting the @samp{.stabs} pseudo-op for a non-global static
8976 @defmac DBX_TYPE_DECL_STABS_CODE
8977 The value to use in the ``code'' field of the @code{.stabs} directive
8978 for a typedef. The default is @code{N_LSYM}.
8981 @defmac DBX_STATIC_CONST_VAR_CODE
8982 The value to use in the ``code'' field of the @code{.stabs} directive
8983 for a static variable located in the text section. DBX format does not
8984 provide any ``right'' way to do this. The default is @code{N_FUN}.
8987 @defmac DBX_REGPARM_STABS_CODE
8988 The value to use in the ``code'' field of the @code{.stabs} directive
8989 for a parameter passed in registers. DBX format does not provide any
8990 ``right'' way to do this. The default is @code{N_RSYM}.
8993 @defmac DBX_REGPARM_STABS_LETTER
8994 The letter to use in DBX symbol data to identify a symbol as a parameter
8995 passed in registers. DBX format does not customarily provide any way to
8996 do this. The default is @code{'P'}.
8999 @defmac DBX_FUNCTION_FIRST
9000 Define this macro if the DBX information for a function and its
9001 arguments should precede the assembler code for the function. Normally,
9002 in DBX format, the debugging information entirely follows the assembler
9006 @defmac DBX_BLOCKS_FUNCTION_RELATIVE
9007 Define this macro, with value 1, if the value of a symbol describing
9008 the scope of a block (@code{N_LBRAC} or @code{N_RBRAC}) should be
9009 relative to the start of the enclosing function. Normally, GCC uses
9010 an absolute address.
9013 @defmac DBX_LINES_FUNCTION_RELATIVE
9014 Define this macro, with value 1, if the value of a symbol indicating
9015 the current line number (@code{N_SLINE}) should be relative to the
9016 start of the enclosing function. Normally, GCC uses an absolute address.
9019 @defmac DBX_USE_BINCL
9020 Define this macro if GCC should generate @code{N_BINCL} and
9021 @code{N_EINCL} stabs for included header files, as on Sun systems. This
9022 macro also directs GCC to output a type number as a pair of a file
9023 number and a type number within the file. Normally, GCC does not
9024 generate @code{N_BINCL} or @code{N_EINCL} stabs, and it outputs a single
9025 number for a type number.
9029 @subsection Open-Ended Hooks for DBX Format
9031 @c prevent bad page break with this line
9032 These are hooks for DBX format.
9034 @defmac DBX_OUTPUT_LBRAC (@var{stream}, @var{name})
9035 Define this macro to say how to output to @var{stream} the debugging
9036 information for the start of a scope level for variable names. The
9037 argument @var{name} is the name of an assembler symbol (for use with
9038 @code{assemble_name}) whose value is the address where the scope begins.
9041 @defmac DBX_OUTPUT_RBRAC (@var{stream}, @var{name})
9042 Like @code{DBX_OUTPUT_LBRAC}, but for the end of a scope level.
9045 @defmac DBX_OUTPUT_NFUN (@var{stream}, @var{lscope_label}, @var{decl})
9046 Define this macro if the target machine requires special handling to
9047 output an @code{N_FUN} entry for the function @var{decl}.
9050 @defmac DBX_OUTPUT_SOURCE_LINE (@var{stream}, @var{line}, @var{counter})
9051 A C statement to output DBX debugging information before code for line
9052 number @var{line} of the current source file to the stdio stream
9053 @var{stream}. @var{counter} is the number of time the macro was
9054 invoked, including the current invocation; it is intended to generate
9055 unique labels in the assembly output.
9057 This macro should not be defined if the default output is correct, or
9058 if it can be made correct by defining @code{DBX_LINES_FUNCTION_RELATIVE}.
9061 @defmac NO_DBX_FUNCTION_END
9062 Some stabs encapsulation formats (in particular ECOFF), cannot handle the
9063 @code{.stabs "",N_FUN,,0,0,Lscope-function-1} gdb dbx extension construct.
9064 On those machines, define this macro to turn this feature off without
9065 disturbing the rest of the gdb extensions.
9068 @defmac NO_DBX_BNSYM_ENSYM
9069 Some assemblers cannot handle the @code{.stabd BNSYM/ENSYM,0,0} gdb dbx
9070 extension construct. On those machines, define this macro to turn this
9071 feature off without disturbing the rest of the gdb extensions.
9074 @node File Names and DBX
9075 @subsection File Names in DBX Format
9077 @c prevent bad page break with this line
9078 This describes file names in DBX format.
9080 @defmac DBX_OUTPUT_MAIN_SOURCE_FILENAME (@var{stream}, @var{name})
9081 A C statement to output DBX debugging information to the stdio stream
9082 @var{stream}, which indicates that file @var{name} is the main source
9083 file---the file specified as the input file for compilation.
9084 This macro is called only once, at the beginning of compilation.
9086 This macro need not be defined if the standard form of output
9087 for DBX debugging information is appropriate.
9089 It may be necessary to refer to a label equal to the beginning of the
9090 text section. You can use @samp{assemble_name (stream, ltext_label_name)}
9091 to do so. If you do this, you must also set the variable
9092 @var{used_ltext_label_name} to @code{true}.
9095 @defmac NO_DBX_MAIN_SOURCE_DIRECTORY
9096 Define this macro, with value 1, if GCC should not emit an indication
9097 of the current directory for compilation and current source language at
9098 the beginning of the file.
9101 @defmac NO_DBX_GCC_MARKER
9102 Define this macro, with value 1, if GCC should not emit an indication
9103 that this object file was compiled by GCC@. The default is to emit
9104 an @code{N_OPT} stab at the beginning of every source file, with
9105 @samp{gcc2_compiled.} for the string and value 0.
9108 @defmac DBX_OUTPUT_MAIN_SOURCE_FILE_END (@var{stream}, @var{name})
9109 A C statement to output DBX debugging information at the end of
9110 compilation of the main source file @var{name}. Output should be
9111 written to the stdio stream @var{stream}.
9113 If you don't define this macro, nothing special is output at the end
9114 of compilation, which is correct for most machines.
9117 @defmac DBX_OUTPUT_NULL_N_SO_AT_MAIN_SOURCE_FILE_END
9118 Define this macro @emph{instead of} defining
9119 @code{DBX_OUTPUT_MAIN_SOURCE_FILE_END}, if what needs to be output at
9120 the end of compilation is an @code{N_SO} stab with an empty string,
9121 whose value is the highest absolute text address in the file.
9126 @subsection Macros for SDB and DWARF Output
9128 @c prevent bad page break with this line
9129 Here are macros for SDB and DWARF output.
9131 @defmac SDB_DEBUGGING_INFO
9132 Define this macro if GCC should produce COFF-style debugging output
9133 for SDB in response to the @option{-g} option.
9136 @defmac DWARF2_DEBUGGING_INFO
9137 Define this macro if GCC should produce dwarf version 2 format
9138 debugging output in response to the @option{-g} option.
9140 @deftypefn {Target Hook} int TARGET_DWARF_CALLING_CONVENTION (tree @var{function})
9141 Define this to enable the dwarf attribute @code{DW_AT_calling_convention} to
9142 be emitted for each function. Instead of an integer return the enum
9143 value for the @code{DW_CC_} tag.
9146 To support optional call frame debugging information, you must also
9147 define @code{INCOMING_RETURN_ADDR_RTX} and either set
9148 @code{RTX_FRAME_RELATED_P} on the prologue insns if you use RTL for the
9149 prologue, or call @code{dwarf2out_def_cfa} and @code{dwarf2out_reg_save}
9150 as appropriate from @code{TARGET_ASM_FUNCTION_PROLOGUE} if you don't.
9153 @defmac DWARF2_FRAME_INFO
9154 Define this macro to a nonzero value if GCC should always output
9155 Dwarf 2 frame information. If @code{DWARF2_UNWIND_INFO}
9156 (@pxref{Exception Region Output} is nonzero, GCC will output this
9157 information not matter how you define @code{DWARF2_FRAME_INFO}.
9160 @defmac DWARF2_ASM_LINE_DEBUG_INFO
9161 Define this macro to be a nonzero value if the assembler can generate Dwarf 2
9162 line debug info sections. This will result in much more compact line number
9163 tables, and hence is desirable if it works.
9166 @defmac ASM_OUTPUT_DWARF_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
9167 A C statement to issue assembly directives that create a difference
9168 @var{lab1} minus @var{lab2}, using an integer of the given @var{size}.
9171 @defmac ASM_OUTPUT_DWARF_OFFSET (@var{stream}, @var{size}, @var{label}, @var{section})
9172 A C statement to issue assembly directives that create a
9173 section-relative reference to the given @var{label}, using an integer of the
9174 given @var{size}. The label is known to be defined in the given @var{section}.
9177 @defmac ASM_OUTPUT_DWARF_PCREL (@var{stream}, @var{size}, @var{label})
9178 A C statement to issue assembly directives that create a self-relative
9179 reference to the given @var{label}, using an integer of the given @var{size}.
9182 @defmac ASM_OUTPUT_DWARF_TABLE_REF (@var{label})
9183 A C statement to issue assembly directives that create a reference to
9184 the DWARF table identifier @var{label} from the current section. This
9185 is used on some systems to avoid garbage collecting a DWARF table which
9186 is referenced by a function.
9189 @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_DWARF_DTPREL (FILE *@var{FILE}, int @var{size}, rtx @var{x})
9190 If defined, this target hook is a function which outputs a DTP-relative
9191 reference to the given TLS symbol of the specified size.
9194 @defmac PUT_SDB_@dots{}
9195 Define these macros to override the assembler syntax for the special
9196 SDB assembler directives. See @file{sdbout.c} for a list of these
9197 macros and their arguments. If the standard syntax is used, you need
9198 not define them yourself.
9202 Some assemblers do not support a semicolon as a delimiter, even between
9203 SDB assembler directives. In that case, define this macro to be the
9204 delimiter to use (usually @samp{\n}). It is not necessary to define
9205 a new set of @code{PUT_SDB_@var{op}} macros if this is the only change
9209 @defmac SDB_ALLOW_UNKNOWN_REFERENCES
9210 Define this macro to allow references to unknown structure,
9211 union, or enumeration tags to be emitted. Standard COFF does not
9212 allow handling of unknown references, MIPS ECOFF has support for
9216 @defmac SDB_ALLOW_FORWARD_REFERENCES
9217 Define this macro to allow references to structure, union, or
9218 enumeration tags that have not yet been seen to be handled. Some
9219 assemblers choke if forward tags are used, while some require it.
9222 @defmac SDB_OUTPUT_SOURCE_LINE (@var{stream}, @var{line})
9223 A C statement to output SDB debugging information before code for line
9224 number @var{line} of the current source file to the stdio stream
9225 @var{stream}. The default is to emit an @code{.ln} directive.
9230 @subsection Macros for VMS Debug Format
9232 @c prevent bad page break with this line
9233 Here are macros for VMS debug format.
9235 @defmac VMS_DEBUGGING_INFO
9236 Define this macro if GCC should produce debugging output for VMS
9237 in response to the @option{-g} option. The default behavior for VMS
9238 is to generate minimal debug info for a traceback in the absence of
9239 @option{-g} unless explicitly overridden with @option{-g0}. This
9240 behavior is controlled by @code{OPTIMIZATION_OPTIONS} and
9241 @code{OVERRIDE_OPTIONS}.
9244 @node Floating Point
9245 @section Cross Compilation and Floating Point
9246 @cindex cross compilation and floating point
9247 @cindex floating point and cross compilation
9249 While all modern machines use twos-complement representation for integers,
9250 there are a variety of representations for floating point numbers. This
9251 means that in a cross-compiler the representation of floating point numbers
9252 in the compiled program may be different from that used in the machine
9253 doing the compilation.
9255 Because different representation systems may offer different amounts of
9256 range and precision, all floating point constants must be represented in
9257 the target machine's format. Therefore, the cross compiler cannot
9258 safely use the host machine's floating point arithmetic; it must emulate
9259 the target's arithmetic. To ensure consistency, GCC always uses
9260 emulation to work with floating point values, even when the host and
9261 target floating point formats are identical.
9263 The following macros are provided by @file{real.h} for the compiler to
9264 use. All parts of the compiler which generate or optimize
9265 floating-point calculations must use these macros. They may evaluate
9266 their operands more than once, so operands must not have side effects.
9268 @defmac REAL_VALUE_TYPE
9269 The C data type to be used to hold a floating point value in the target
9270 machine's format. Typically this is a @code{struct} containing an
9271 array of @code{HOST_WIDE_INT}, but all code should treat it as an opaque
9275 @deftypefn Macro int REAL_VALUES_EQUAL (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
9276 Compares for equality the two values, @var{x} and @var{y}. If the target
9277 floating point format supports negative zeroes and/or NaNs,
9278 @samp{REAL_VALUES_EQUAL (-0.0, 0.0)} is true, and
9279 @samp{REAL_VALUES_EQUAL (NaN, NaN)} is false.
9282 @deftypefn Macro int REAL_VALUES_LESS (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
9283 Tests whether @var{x} is less than @var{y}.
9286 @deftypefn Macro HOST_WIDE_INT REAL_VALUE_FIX (REAL_VALUE_TYPE @var{x})
9287 Truncates @var{x} to a signed integer, rounding toward zero.
9290 @deftypefn Macro {unsigned HOST_WIDE_INT} REAL_VALUE_UNSIGNED_FIX (REAL_VALUE_TYPE @var{x})
9291 Truncates @var{x} to an unsigned integer, rounding toward zero. If
9292 @var{x} is negative, returns zero.
9295 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ATOF (const char *@var{string}, enum machine_mode @var{mode})
9296 Converts @var{string} into a floating point number in the target machine's
9297 representation for mode @var{mode}. This routine can handle both
9298 decimal and hexadecimal floating point constants, using the syntax
9299 defined by the C language for both.
9302 @deftypefn Macro int REAL_VALUE_NEGATIVE (REAL_VALUE_TYPE @var{x})
9303 Returns 1 if @var{x} is negative (including negative zero), 0 otherwise.
9306 @deftypefn Macro int REAL_VALUE_ISINF (REAL_VALUE_TYPE @var{x})
9307 Determines whether @var{x} represents infinity (positive or negative).
9310 @deftypefn Macro int REAL_VALUE_ISNAN (REAL_VALUE_TYPE @var{x})
9311 Determines whether @var{x} represents a ``NaN'' (not-a-number).
9314 @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})
9315 Calculates an arithmetic operation on the two floating point values
9316 @var{x} and @var{y}, storing the result in @var{output} (which must be a
9319 The operation to be performed is specified by @var{code}. Only the
9320 following codes are supported: @code{PLUS_EXPR}, @code{MINUS_EXPR},
9321 @code{MULT_EXPR}, @code{RDIV_EXPR}, @code{MAX_EXPR}, @code{MIN_EXPR}.
9323 If @code{REAL_ARITHMETIC} is asked to evaluate division by zero and the
9324 target's floating point format cannot represent infinity, it will call
9325 @code{abort}. Callers should check for this situation first, using
9326 @code{MODE_HAS_INFINITIES}. @xref{Storage Layout}.
9329 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_NEGATE (REAL_VALUE_TYPE @var{x})
9330 Returns the negative of the floating point value @var{x}.
9333 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ABS (REAL_VALUE_TYPE @var{x})
9334 Returns the absolute value of @var{x}.
9337 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_TRUNCATE (REAL_VALUE_TYPE @var{mode}, enum machine_mode @var{x})
9338 Truncates the floating point value @var{x} to fit in @var{mode}. The
9339 return value is still a full-size @code{REAL_VALUE_TYPE}, but it has an
9340 appropriate bit pattern to be output as a floating constant whose
9341 precision accords with mode @var{mode}.
9344 @deftypefn Macro void REAL_VALUE_TO_INT (HOST_WIDE_INT @var{low}, HOST_WIDE_INT @var{high}, REAL_VALUE_TYPE @var{x})
9345 Converts a floating point value @var{x} into a double-precision integer
9346 which is then stored into @var{low} and @var{high}. If the value is not
9347 integral, it is truncated.
9350 @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})
9351 Converts a double-precision integer found in @var{low} and @var{high},
9352 into a floating point value which is then stored into @var{x}. The
9353 value is truncated to fit in mode @var{mode}.
9356 @node Mode Switching
9357 @section Mode Switching Instructions
9358 @cindex mode switching
9359 The following macros control mode switching optimizations:
9361 @defmac OPTIMIZE_MODE_SWITCHING (@var{entity})
9362 Define this macro if the port needs extra instructions inserted for mode
9363 switching in an optimizing compilation.
9365 For an example, the SH4 can perform both single and double precision
9366 floating point operations, but to perform a single precision operation,
9367 the FPSCR PR bit has to be cleared, while for a double precision
9368 operation, this bit has to be set. Changing the PR bit requires a general
9369 purpose register as a scratch register, hence these FPSCR sets have to
9370 be inserted before reload, i.e.@: you can't put this into instruction emitting
9371 or @code{TARGET_MACHINE_DEPENDENT_REORG}.
9373 You can have multiple entities that are mode-switched, and select at run time
9374 which entities actually need it. @code{OPTIMIZE_MODE_SWITCHING} should
9375 return nonzero for any @var{entity} that needs mode-switching.
9376 If you define this macro, you also have to define
9377 @code{NUM_MODES_FOR_MODE_SWITCHING}, @code{MODE_NEEDED},
9378 @code{MODE_PRIORITY_TO_MODE} and @code{EMIT_MODE_SET}.
9379 @code{MODE_AFTER}, @code{MODE_ENTRY}, and @code{MODE_EXIT}
9383 @defmac NUM_MODES_FOR_MODE_SWITCHING
9384 If you define @code{OPTIMIZE_MODE_SWITCHING}, you have to define this as
9385 initializer for an array of integers. Each initializer element
9386 N refers to an entity that needs mode switching, and specifies the number
9387 of different modes that might need to be set for this entity.
9388 The position of the initializer in the initializer---starting counting at
9389 zero---determines the integer that is used to refer to the mode-switched
9391 In macros that take mode arguments / yield a mode result, modes are
9392 represented as numbers 0 @dots{} N @minus{} 1. N is used to specify that no mode
9393 switch is needed / supplied.
9396 @defmac MODE_NEEDED (@var{entity}, @var{insn})
9397 @var{entity} is an integer specifying a mode-switched entity. If
9398 @code{OPTIMIZE_MODE_SWITCHING} is defined, you must define this macro to
9399 return an integer value not larger than the corresponding element in
9400 @code{NUM_MODES_FOR_MODE_SWITCHING}, to denote the mode that @var{entity} must
9401 be switched into prior to the execution of @var{insn}.
9404 @defmac MODE_AFTER (@var{mode}, @var{insn})
9405 If this macro is defined, it is evaluated for every @var{insn} during
9406 mode switching. It determines the mode that an insn results in (if
9407 different from the incoming mode).
9410 @defmac MODE_ENTRY (@var{entity})
9411 If this macro is defined, it is evaluated for every @var{entity} that needs
9412 mode switching. It should evaluate to an integer, which is a mode that
9413 @var{entity} is assumed to be switched to at function entry. If @code{MODE_ENTRY}
9414 is defined then @code{MODE_EXIT} must be defined.
9417 @defmac MODE_EXIT (@var{entity})
9418 If this macro is defined, it is evaluated for every @var{entity} that needs
9419 mode switching. It should evaluate to an integer, which is a mode that
9420 @var{entity} is assumed to be switched to at function exit. If @code{MODE_EXIT}
9421 is defined then @code{MODE_ENTRY} must be defined.
9424 @defmac MODE_PRIORITY_TO_MODE (@var{entity}, @var{n})
9425 This macro specifies the order in which modes for @var{entity} are processed.
9426 0 is the highest priority, @code{NUM_MODES_FOR_MODE_SWITCHING[@var{entity}] - 1} the
9427 lowest. The value of the macro should be an integer designating a mode
9428 for @var{entity}. For any fixed @var{entity}, @code{mode_priority_to_mode}
9429 (@var{entity}, @var{n}) shall be a bijection in 0 @dots{}
9430 @code{num_modes_for_mode_switching[@var{entity}] - 1}.
9433 @defmac EMIT_MODE_SET (@var{entity}, @var{mode}, @var{hard_regs_live})
9434 Generate one or more insns to set @var{entity} to @var{mode}.
9435 @var{hard_reg_live} is the set of hard registers live at the point where
9436 the insn(s) are to be inserted.
9439 @node Target Attributes
9440 @section Defining target-specific uses of @code{__attribute__}
9441 @cindex target attributes
9442 @cindex machine attributes
9443 @cindex attributes, target-specific
9445 Target-specific attributes may be defined for functions, data and types.
9446 These are described using the following target hooks; they also need to
9447 be documented in @file{extend.texi}.
9449 @deftypevr {Target Hook} {const struct attribute_spec *} TARGET_ATTRIBUTE_TABLE
9450 If defined, this target hook points to an array of @samp{struct
9451 attribute_spec} (defined in @file{tree.h}) specifying the machine
9452 specific attributes for this target and some of the restrictions on the
9453 entities to which these attributes are applied and the arguments they
9457 @deftypefn {Target Hook} int TARGET_COMP_TYPE_ATTRIBUTES (tree @var{type1}, tree @var{type2})
9458 If defined, this target hook is a function which returns zero if the attributes on
9459 @var{type1} and @var{type2} are incompatible, one if they are compatible,
9460 and two if they are nearly compatible (which causes a warning to be
9461 generated). If this is not defined, machine-specific attributes are
9462 supposed always to be compatible.
9465 @deftypefn {Target Hook} void TARGET_SET_DEFAULT_TYPE_ATTRIBUTES (tree @var{type})
9466 If defined, this target hook is a function which assigns default attributes to
9467 newly defined @var{type}.
9470 @deftypefn {Target Hook} tree TARGET_MERGE_TYPE_ATTRIBUTES (tree @var{type1}, tree @var{type2})
9471 Define this target hook if the merging of type attributes needs special
9472 handling. If defined, the result is a list of the combined
9473 @code{TYPE_ATTRIBUTES} of @var{type1} and @var{type2}. It is assumed
9474 that @code{comptypes} has already been called and returned 1. This
9475 function may call @code{merge_attributes} to handle machine-independent
9479 @deftypefn {Target Hook} tree TARGET_MERGE_DECL_ATTRIBUTES (tree @var{olddecl}, tree @var{newdecl})
9480 Define this target hook if the merging of decl attributes needs special
9481 handling. If defined, the result is a list of the combined
9482 @code{DECL_ATTRIBUTES} of @var{olddecl} and @var{newdecl}.
9483 @var{newdecl} is a duplicate declaration of @var{olddecl}. Examples of
9484 when this is needed are when one attribute overrides another, or when an
9485 attribute is nullified by a subsequent definition. This function may
9486 call @code{merge_attributes} to handle machine-independent merging.
9488 @findex TARGET_DLLIMPORT_DECL_ATTRIBUTES
9489 If the only target-specific handling you require is @samp{dllimport}
9490 for Microsoft Windows targets, you should define the macro
9491 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES} to @code{1}. The compiler
9492 will then define a function called
9493 @code{merge_dllimport_decl_attributes} which can then be defined as
9494 the expansion of @code{TARGET_MERGE_DECL_ATTRIBUTES}. You can also
9495 add @code{handle_dll_attribute} in the attribute table for your port
9496 to perform initial processing of the @samp{dllimport} and
9497 @samp{dllexport} attributes. This is done in @file{i386/cygwin.h} and
9498 @file{i386/i386.c}, for example.
9501 @deftypefn {Target Hook} bool TARGET_VALID_DLLIMPORT_ATTRIBUTE_P (tree @var{decl})
9502 @var{decl} is a variable or function with @code{__attribute__((dllimport))}
9503 specified. Use this hook if the target needs to add extra validation
9504 checks to @code{handle_dll_attribute}.
9507 @defmac TARGET_DECLSPEC
9508 Define this macro to a nonzero value if you want to treat
9509 @code{__declspec(X)} as equivalent to @code{__attribute((X))}. By
9510 default, this behavior is enabled only for targets that define
9511 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES}. The current implementation
9512 of @code{__declspec} is via a built-in macro, but you should not rely
9513 on this implementation detail.
9516 @deftypefn {Target Hook} void TARGET_INSERT_ATTRIBUTES (tree @var{node}, tree *@var{attr_ptr})
9517 Define this target hook if you want to be able to add attributes to a decl
9518 when it is being created. This is normally useful for back ends which
9519 wish to implement a pragma by using the attributes which correspond to
9520 the pragma's effect. The @var{node} argument is the decl which is being
9521 created. The @var{attr_ptr} argument is a pointer to the attribute list
9522 for this decl. The list itself should not be modified, since it may be
9523 shared with other decls, but attributes may be chained on the head of
9524 the list and @code{*@var{attr_ptr}} modified to point to the new
9525 attributes, or a copy of the list may be made if further changes are
9529 @deftypefn {Target Hook} bool TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P (const_tree @var{fndecl})
9531 This target hook returns @code{true} if it is ok to inline @var{fndecl}
9532 into the current function, despite its having target-specific
9533 attributes, @code{false} otherwise. By default, if a function has a
9534 target specific attribute attached to it, it will not be inlined.
9537 @deftypefn {Target Hook} bool TARGET_VALID_OPTION_ATTRIBUTE_P (tree @var{fndecl}, tree @var{name}, tree @var{args}, int @var{flags})
9538 This hook is called to parse the @code{attribute(option("..."))}, and
9539 it allows the function to set different target machine compile time
9540 options for the current function that might be different than the
9541 options specified on the command line. The hook should return
9542 @code{true} if the options are valid.
9544 The hook should set the @var{DECL_FUNCTION_SPECIFIC_TARGET} field in
9545 the function declaration to hold a pointer to a target specific
9546 @var{struct cl_target_option} structure.
9549 @deftypefn {Target Hook} void TARGET_OPTION_SAVE (struct cl_target_option *@var{ptr})
9550 This hook is called to save any additional target specific information
9551 in the @var{struct cl_target_option} structure for function specific
9553 @xref{Option file format}.
9556 @deftypefn {Target Hook} void TARGET_OPTION_RESTORE (struct cl_target_option *@var{ptr})
9557 This hook is called to restore any additional target specific
9558 information in the @var{struct cl_target_option} structure for
9559 function specific options.
9562 @deftypefn {Target Hook} void TARGET_OPTION_PRINT (struct cl_target_option *@var{ptr})
9563 This hook is called to print any additional target specific
9564 information in the @var{struct cl_target_option} structure for
9565 function specific options.
9568 @deftypefn {Target Hook} bool TARGET_OPTION_PRAGMA_PARSE (target @var{args})
9569 This target hook parses the options for @code{#pragma GCC option} to
9570 set the machine specific options for functions that occur later in the
9571 input stream. The options should be the same as handled by the
9572 @code{TARGET_VALID_OPTION_ATTRIBUTE_P} hook.
9575 @deftypefn {Target Hook} bool TARGET_CAN_INLINE_P (tree @var{caller}, tree @var{callee})
9576 This target hook returns @code{false} if the @var{caller} function
9577 cannot inline @var{callee}, based on target specific information. By
9578 default, inlining is not allowed if the callee function has function
9579 specific target options and the caller does not use the same options.
9583 @section Emulating TLS
9584 @cindex Emulated TLS
9586 For targets whose psABI does not provide Thread Local Storage via
9587 specific relocations and instruction sequences, an emulation layer is
9588 used. A set of target hooks allows this emulation layer to be
9589 configured for the requirements of a particular target. For instance
9590 the psABI may in fact specify TLS support in terms of an emulation
9593 The emulation layer works by creating a control object for every TLS
9594 object. To access the TLS object, a lookup function is provided
9595 which, when given the address of the control object, will return the
9596 address of the current thread's instance of the TLS object.
9598 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_GET_ADDRESS
9599 Contains the name of the helper function that uses a TLS control
9600 object to locate a TLS instance. The default causes libgcc's
9601 emulated TLS helper function to be used.
9604 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_REGISTER_COMMON
9605 Contains the name of the helper function that should be used at
9606 program startup to register TLS objects that are implicitly
9607 initialized to zero. If this is @code{NULL}, all TLS objects will
9608 have explicit initializers. The default causes libgcc's emulated TLS
9609 registration function to be used.
9612 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_VAR_SECTION
9613 Contains the name of the section in which TLS control variables should
9614 be placed. The default of @code{NULL} allows these to be placed in
9618 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_TMPL_SECTION
9619 Contains the name of the section in which TLS initializers should be
9620 placed. The default of @code{NULL} allows these to be placed in any
9624 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_VAR_PREFIX
9625 Contains the prefix to be prepended to TLS control variable names.
9626 The default of @code{NULL} uses a target-specific prefix.
9629 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_TMPL_PREFIX
9630 Contains the prefix to be prepended to TLS initializer objects. The
9631 default of @code{NULL} uses a target-specific prefix.
9634 @deftypefn {Target Hook} tree TARGET_EMUTLS_VAR_FIELDS (tree @var{type}, tree *@var{name})
9635 Specifies a function that generates the FIELD_DECLs for a TLS control
9636 object type. @var{type} is the RECORD_TYPE the fields are for and
9637 @var{name} should be filled with the structure tag, if the default of
9638 @code{__emutls_object} is unsuitable. The default creates a type suitable
9639 for libgcc's emulated TLS function.
9642 @deftypefn {Target Hook} tree TARGET_EMUTLS_VAR_INIT (tree @var{var}, tree @var{decl}, tree @var{tmpl_addr})
9643 Specifies a function that generates the CONSTRUCTOR to initialize a
9644 TLS control object. @var{var} is the TLS control object, @var{decl}
9645 is the TLS object and @var{tmpl_addr} is the address of the
9646 initializer. The default initializes libgcc's emulated TLS control object.
9649 @deftypevr {Target Hook} {bool} TARGET_EMUTLS_VAR_ALIGN_FIXED
9650 Specifies whether the alignment of TLS control variable objects is
9651 fixed and should not be increased as some backends may do to optimize
9652 single objects. The default is false.
9655 @deftypevr {Target Hook} {bool} TARGET_EMUTLS_DEBUG_FORM_TLS_ADDRESS
9656 Specifies whether a DWARF @code{DW_OP_form_tls_address} location descriptor
9657 may be used to describe emulated TLS control objects.
9660 @node MIPS Coprocessors
9661 @section Defining coprocessor specifics for MIPS targets.
9662 @cindex MIPS coprocessor-definition macros
9664 The MIPS specification allows MIPS implementations to have as many as 4
9665 coprocessors, each with as many as 32 private registers. GCC supports
9666 accessing these registers and transferring values between the registers
9667 and memory using asm-ized variables. For example:
9670 register unsigned int cp0count asm ("c0r1");
9676 (``c0r1'' is the default name of register 1 in coprocessor 0; alternate
9677 names may be added as described below, or the default names may be
9678 overridden entirely in @code{SUBTARGET_CONDITIONAL_REGISTER_USAGE}.)
9680 Coprocessor registers are assumed to be epilogue-used; sets to them will
9681 be preserved even if it does not appear that the register is used again
9682 later in the function.
9684 Another note: according to the MIPS spec, coprocessor 1 (if present) is
9685 the FPU@. One accesses COP1 registers through standard mips
9686 floating-point support; they are not included in this mechanism.
9688 There is one macro used in defining the MIPS coprocessor interface which
9689 you may want to override in subtargets; it is described below.
9691 @defmac ALL_COP_ADDITIONAL_REGISTER_NAMES
9692 A comma-separated list (with leading comma) of pairs describing the
9693 alternate names of coprocessor registers. The format of each entry should be
9695 @{ @var{alternatename}, @var{register_number}@}
9701 @section Parameters for Precompiled Header Validity Checking
9702 @cindex parameters, precompiled headers
9704 @deftypefn {Target Hook} void *TARGET_GET_PCH_VALIDITY (size_t *@var{sz})
9705 This hook returns the data needed by @code{TARGET_PCH_VALID_P} and sets
9706 @samp{*@var{sz}} to the size of the data in bytes.
9709 @deftypefn {Target Hook} const char *TARGET_PCH_VALID_P (const void *@var{data}, size_t @var{sz})
9710 This hook checks whether the options used to create a PCH file are
9711 compatible with the current settings. It returns @code{NULL}
9712 if so and a suitable error message if not. Error messages will
9713 be presented to the user and must be localized using @samp{_(@var{msg})}.
9715 @var{data} is the data that was returned by @code{TARGET_GET_PCH_VALIDITY}
9716 when the PCH file was created and @var{sz} is the size of that data in bytes.
9717 It's safe to assume that the data was created by the same version of the
9718 compiler, so no format checking is needed.
9720 The default definition of @code{default_pch_valid_p} should be
9721 suitable for most targets.
9724 @deftypefn {Target Hook} const char *TARGET_CHECK_PCH_TARGET_FLAGS (int @var{pch_flags})
9725 If this hook is nonnull, the default implementation of
9726 @code{TARGET_PCH_VALID_P} will use it to check for compatible values
9727 of @code{target_flags}. @var{pch_flags} specifies the value that
9728 @code{target_flags} had when the PCH file was created. The return
9729 value is the same as for @code{TARGET_PCH_VALID_P}.
9733 @section C++ ABI parameters
9734 @cindex parameters, c++ abi
9736 @deftypefn {Target Hook} tree TARGET_CXX_GUARD_TYPE (void)
9737 Define this hook to override the integer type used for guard variables.
9738 These are used to implement one-time construction of static objects. The
9739 default is long_long_integer_type_node.
9742 @deftypefn {Target Hook} bool TARGET_CXX_GUARD_MASK_BIT (void)
9743 This hook determines how guard variables are used. It should return
9744 @code{false} (the default) if first byte should be used. A return value of
9745 @code{true} indicates the least significant bit should be used.
9748 @deftypefn {Target Hook} tree TARGET_CXX_GET_COOKIE_SIZE (tree @var{type})
9749 This hook returns the size of the cookie to use when allocating an array
9750 whose elements have the indicated @var{type}. Assumes that it is already
9751 known that a cookie is needed. The default is
9752 @code{max(sizeof (size_t), alignof(type))}, as defined in section 2.7 of the
9753 IA64/Generic C++ ABI@.
9756 @deftypefn {Target Hook} bool TARGET_CXX_COOKIE_HAS_SIZE (void)
9757 This hook should return @code{true} if the element size should be stored in
9758 array cookies. The default is to return @code{false}.
9761 @deftypefn {Target Hook} int TARGET_CXX_IMPORT_EXPORT_CLASS (tree @var{type}, int @var{import_export})
9762 If defined by a backend this hook allows the decision made to export
9763 class @var{type} to be overruled. Upon entry @var{import_export}
9764 will contain 1 if the class is going to be exported, @minus{}1 if it is going
9765 to be imported and 0 otherwise. This function should return the
9766 modified value and perform any other actions necessary to support the
9767 backend's targeted operating system.
9770 @deftypefn {Target Hook} bool TARGET_CXX_CDTOR_RETURNS_THIS (void)
9771 This hook should return @code{true} if constructors and destructors return
9772 the address of the object created/destroyed. The default is to return
9776 @deftypefn {Target Hook} bool TARGET_CXX_KEY_METHOD_MAY_BE_INLINE (void)
9777 This hook returns true if the key method for a class (i.e., the method
9778 which, if defined in the current translation unit, causes the virtual
9779 table to be emitted) may be an inline function. Under the standard
9780 Itanium C++ ABI the key method may be an inline function so long as
9781 the function is not declared inline in the class definition. Under
9782 some variants of the ABI, an inline function can never be the key
9783 method. The default is to return @code{true}.
9786 @deftypefn {Target Hook} void TARGET_CXX_DETERMINE_CLASS_DATA_VISIBILITY (tree @var{decl})
9787 @var{decl} is a virtual table, virtual table table, typeinfo object,
9788 or other similar implicit class data object that will be emitted with
9789 external linkage in this translation unit. No ELF visibility has been
9790 explicitly specified. If the target needs to specify a visibility
9791 other than that of the containing class, use this hook to set
9792 @code{DECL_VISIBILITY} and @code{DECL_VISIBILITY_SPECIFIED}.
9795 @deftypefn {Target Hook} bool TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT (void)
9796 This hook returns true (the default) if virtual tables and other
9797 similar implicit class data objects are always COMDAT if they have
9798 external linkage. If this hook returns false, then class data for
9799 classes whose virtual table will be emitted in only one translation
9800 unit will not be COMDAT.
9803 @deftypefn {Target Hook} bool TARGET_CXX_LIBRARY_RTTI_COMDAT (void)
9804 This hook returns true (the default) if the RTTI information for
9805 the basic types which is defined in the C++ runtime should always
9806 be COMDAT, false if it should not be COMDAT.
9809 @deftypefn {Target Hook} bool TARGET_CXX_USE_AEABI_ATEXIT (void)
9810 This hook returns true if @code{__aeabi_atexit} (as defined by the ARM EABI)
9811 should be used to register static destructors when @option{-fuse-cxa-atexit}
9812 is in effect. The default is to return false to use @code{__cxa_atexit}.
9815 @deftypefn {Target Hook} bool TARGET_CXX_USE_ATEXIT_FOR_CXA_ATEXIT (void)
9816 This hook returns true if the target @code{atexit} function can be used
9817 in the same manner as @code{__cxa_atexit} to register C++ static
9818 destructors. This requires that @code{atexit}-registered functions in
9819 shared libraries are run in the correct order when the libraries are
9820 unloaded. The default is to return false.
9823 @deftypefn {Target Hook} void TARGET_CXX_ADJUST_CLASS_AT_DEFINITION (tree @var{type})
9824 @var{type} is a C++ class (i.e., RECORD_TYPE or UNION_TYPE) that has just been
9825 defined. Use this hook to make adjustments to the class (eg, tweak
9826 visibility or perform any other required target modifications).
9829 @node Named Address Spaces
9830 @section Adding support for named address spaces
9831 @cindex named address spaces
9833 The draft technical report of the ISO/IEC JTC1 S22 WG14 N1275
9834 standards committee, @cite{Programming Languages - C - Extensions to
9835 support embedded processors}, specifies a syntax for embedded
9836 processors to specify alternate address spaces. You can configure a
9837 GCC port to support section 5.1 of the draft report to add support for
9838 address spaces other than the default address space. These address
9839 spaces are new keywords that are similar to the @code{volatile} and
9840 @code{const} type attributes.
9842 Pointers to named address spaces can a a different size than
9843 pointers to the generic address space.
9845 For example, the SPU port uses the @code{__ea} address space to refer
9846 to memory in the host processor, rather than memory local to the SPU
9847 processor. Access to memory in the @code{__ea} address space involves
9848 issuing DMA operations to move data between the host processor and the
9849 local processor memory address space. Pointers in the @code{__ea}
9850 address space are either 32 bits or 64 bits based on the
9851 @option{-mea32} or @option{-mea64} switches (native SPU pointers are
9854 Internally, address spaces are represented as a small integer in the
9855 range 0 to 15 with address space 0 being reserved for the generic
9858 @deftypefn {Target Hook} {bool} TARGET_ADDR_SPACE_LEGITIMATE_ADDRESS_P (enum machine_mode @var{mode}, rtx @var{exp}, bool @var{strict}, addr_space_t @var{as})
9859 Define this to return true if @var{exp} is a valid address for mode
9860 @var{mode} in the named address space @var{as}. The @var{strict}
9861 parameter says whether strict addressing is in effect after reload has
9862 finished. This target hook is the same as the
9863 @code{TARGET_LEGITIMATE_ADDRESS_P} target hook, except that it includes
9864 explicit named address space support.
9867 @deftypefn {Target Hook} {rtx} TARGET_ADDR_SPACE_LEGITIMIZE_ADDRESS (rtx @var{x}, rtx @var{oldx}, enum machine_mode @var{mode}, addr_space_t @var{as})
9868 Define this to modify an invalid address @var{x} to be a valid address
9869 with mode @var{mode} in the named address space @var{as}. This target
9870 hook is the same as the @code{TARGET_LEGITIMIZE_ADDRESS} target hook,
9871 except that it includes explicit named address space support.
9874 @deftypefn {Target Hook} {bool} TARGET_ADDR_SPACE_SUBSET_P (addr_space_t @var{superset}, addr_space_t @var{subset})
9875 Define this to return whether the @var{subset} named address space is
9876 contained within the @var{superset} named address space. Pointers to
9877 a named address space that is a subset of another named address space
9878 will be converted automatically without a cast if used together in
9879 arithmetic operations. Pointers to a superset address space can be
9880 converted to pointers to a subset address space via explict casts.
9883 @deftypefn {Target Hook} {rtx} TARGET_ADDR_SPACE_CONVERT (rtx @var{op}, tree @var{from_type}, tree @var{to_type})
9884 Define this to convert the pointer expression represented by the RTL
9885 @var{op} with type @var{from_type} that points to a named address
9886 space to a new pointer expression with type @var{to_type} that points
9887 to a different named address space. When this hook it called, it is
9888 guaranteed that one of the two address spaces is a subset of the other,
9889 as determined by the @code{TARGET_ADDR_SPACE_SUBSET_P} target hook.
9893 @section Miscellaneous Parameters
9894 @cindex parameters, miscellaneous
9896 @c prevent bad page break with this line
9897 Here are several miscellaneous parameters.
9899 @defmac HAS_LONG_COND_BRANCH
9900 Define this boolean macro to indicate whether or not your architecture
9901 has conditional branches that can span all of memory. It is used in
9902 conjunction with an optimization that partitions hot and cold basic
9903 blocks into separate sections of the executable. If this macro is
9904 set to false, gcc will convert any conditional branches that attempt
9905 to cross between sections into unconditional branches or indirect jumps.
9908 @defmac HAS_LONG_UNCOND_BRANCH
9909 Define this boolean macro to indicate whether or not your architecture
9910 has unconditional branches that can span all of memory. It is used in
9911 conjunction with an optimization that partitions hot and cold basic
9912 blocks into separate sections of the executable. If this macro is
9913 set to false, gcc will convert any unconditional branches that attempt
9914 to cross between sections into indirect jumps.
9917 @defmac CASE_VECTOR_MODE
9918 An alias for a machine mode name. This is the machine mode that
9919 elements of a jump-table should have.
9922 @defmac CASE_VECTOR_SHORTEN_MODE (@var{min_offset}, @var{max_offset}, @var{body})
9923 Optional: return the preferred mode for an @code{addr_diff_vec}
9924 when the minimum and maximum offset are known. If you define this,
9925 it enables extra code in branch shortening to deal with @code{addr_diff_vec}.
9926 To make this work, you also have to define @code{INSN_ALIGN} and
9927 make the alignment for @code{addr_diff_vec} explicit.
9928 The @var{body} argument is provided so that the offset_unsigned and scale
9929 flags can be updated.
9932 @defmac CASE_VECTOR_PC_RELATIVE
9933 Define this macro to be a C expression to indicate when jump-tables
9934 should contain relative addresses. You need not define this macro if
9935 jump-tables never contain relative addresses, or jump-tables should
9936 contain relative addresses only when @option{-fPIC} or @option{-fPIC}
9940 @deftypefn {Target Hook} unsigned int TARGET_CASE_VALUES_THRESHOLD (void)
9941 This function return the smallest number of different values for which it
9942 is best to use a jump-table instead of a tree of conditional branches.
9943 The default is four for machines with a @code{casesi} instruction and
9944 five otherwise. This is best for most machines.
9947 @defmac CASE_USE_BIT_TESTS
9948 Define this macro to be a C expression to indicate whether C switch
9949 statements may be implemented by a sequence of bit tests. This is
9950 advantageous on processors that can efficiently implement left shift
9951 of 1 by the number of bits held in a register, but inappropriate on
9952 targets that would require a loop. By default, this macro returns
9953 @code{true} if the target defines an @code{ashlsi3} pattern, and
9954 @code{false} otherwise.
9957 @defmac WORD_REGISTER_OPERATIONS
9958 Define this macro if operations between registers with integral mode
9959 smaller than a word are always performed on the entire register.
9960 Most RISC machines have this property and most CISC machines do not.
9963 @defmac LOAD_EXTEND_OP (@var{mem_mode})
9964 Define this macro to be a C expression indicating when insns that read
9965 memory in @var{mem_mode}, an integral mode narrower than a word, set the
9966 bits outside of @var{mem_mode} to be either the sign-extension or the
9967 zero-extension of the data read. Return @code{SIGN_EXTEND} for values
9968 of @var{mem_mode} for which the
9969 insn sign-extends, @code{ZERO_EXTEND} for which it zero-extends, and
9970 @code{UNKNOWN} for other modes.
9972 This macro is not called with @var{mem_mode} non-integral or with a width
9973 greater than or equal to @code{BITS_PER_WORD}, so you may return any
9974 value in this case. Do not define this macro if it would always return
9975 @code{UNKNOWN}. On machines where this macro is defined, you will normally
9976 define it as the constant @code{SIGN_EXTEND} or @code{ZERO_EXTEND}.
9978 You may return a non-@code{UNKNOWN} value even if for some hard registers
9979 the sign extension is not performed, if for the @code{REGNO_REG_CLASS}
9980 of these hard registers @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero
9981 when the @var{from} mode is @var{mem_mode} and the @var{to} mode is any
9982 integral mode larger than this but not larger than @code{word_mode}.
9984 You must return @code{UNKNOWN} if for some hard registers that allow this
9985 mode, @code{CANNOT_CHANGE_MODE_CLASS} says that they cannot change to
9986 @code{word_mode}, but that they can change to another integral mode that
9987 is larger then @var{mem_mode} but still smaller than @code{word_mode}.
9990 @defmac SHORT_IMMEDIATES_SIGN_EXTEND
9991 Define this macro if loading short immediate values into registers sign
9995 @defmac FIXUNS_TRUNC_LIKE_FIX_TRUNC
9996 Define this macro if the same instructions that convert a floating
9997 point number to a signed fixed point number also convert validly to an
10001 @deftypefn {Target Hook} int TARGET_MIN_DIVISIONS_FOR_RECIP_MUL (enum machine_mode @var{mode})
10002 When @option{-ffast-math} is in effect, GCC tries to optimize
10003 divisions by the same divisor, by turning them into multiplications by
10004 the reciprocal. This target hook specifies the minimum number of divisions
10005 that should be there for GCC to perform the optimization for a variable
10006 of mode @var{mode}. The default implementation returns 3 if the machine
10007 has an instruction for the division, and 2 if it does not.
10011 The maximum number of bytes that a single instruction can move quickly
10012 between memory and registers or between two memory locations.
10015 @defmac MAX_MOVE_MAX
10016 The maximum number of bytes that a single instruction can move quickly
10017 between memory and registers or between two memory locations. If this
10018 is undefined, the default is @code{MOVE_MAX}. Otherwise, it is the
10019 constant value that is the largest value that @code{MOVE_MAX} can have
10023 @defmac SHIFT_COUNT_TRUNCATED
10024 A C expression that is nonzero if on this machine the number of bits
10025 actually used for the count of a shift operation is equal to the number
10026 of bits needed to represent the size of the object being shifted. When
10027 this macro is nonzero, the compiler will assume that it is safe to omit
10028 a sign-extend, zero-extend, and certain bitwise `and' instructions that
10029 truncates the count of a shift operation. On machines that have
10030 instructions that act on bit-fields at variable positions, which may
10031 include `bit test' instructions, a nonzero @code{SHIFT_COUNT_TRUNCATED}
10032 also enables deletion of truncations of the values that serve as
10033 arguments to bit-field instructions.
10035 If both types of instructions truncate the count (for shifts) and
10036 position (for bit-field operations), or if no variable-position bit-field
10037 instructions exist, you should define this macro.
10039 However, on some machines, such as the 80386 and the 680x0, truncation
10040 only applies to shift operations and not the (real or pretended)
10041 bit-field operations. Define @code{SHIFT_COUNT_TRUNCATED} to be zero on
10042 such machines. Instead, add patterns to the @file{md} file that include
10043 the implied truncation of the shift instructions.
10045 You need not define this macro if it would always have the value of zero.
10048 @anchor{TARGET_SHIFT_TRUNCATION_MASK}
10049 @deftypefn {Target Hook} int TARGET_SHIFT_TRUNCATION_MASK (enum machine_mode @var{mode})
10050 This function describes how the standard shift patterns for @var{mode}
10051 deal with shifts by negative amounts or by more than the width of the mode.
10052 @xref{shift patterns}.
10054 On many machines, the shift patterns will apply a mask @var{m} to the
10055 shift count, meaning that a fixed-width shift of @var{x} by @var{y} is
10056 equivalent to an arbitrary-width shift of @var{x} by @var{y & m}. If
10057 this is true for mode @var{mode}, the function should return @var{m},
10058 otherwise it should return 0. A return value of 0 indicates that no
10059 particular behavior is guaranteed.
10061 Note that, unlike @code{SHIFT_COUNT_TRUNCATED}, this function does
10062 @emph{not} apply to general shift rtxes; it applies only to instructions
10063 that are generated by the named shift patterns.
10065 The default implementation of this function returns
10066 @code{GET_MODE_BITSIZE (@var{mode}) - 1} if @code{SHIFT_COUNT_TRUNCATED}
10067 and 0 otherwise. This definition is always safe, but if
10068 @code{SHIFT_COUNT_TRUNCATED} is false, and some shift patterns
10069 nevertheless truncate the shift count, you may get better code
10073 @defmac TRULY_NOOP_TRUNCATION (@var{outprec}, @var{inprec})
10074 A C expression which is nonzero if on this machine it is safe to
10075 ``convert'' an integer of @var{inprec} bits to one of @var{outprec}
10076 bits (where @var{outprec} is smaller than @var{inprec}) by merely
10077 operating on it as if it had only @var{outprec} bits.
10079 On many machines, this expression can be 1.
10081 @c rearranged this, removed the phrase "it is reported that". this was
10082 @c to fix an overfull hbox. --mew 10feb93
10083 When @code{TRULY_NOOP_TRUNCATION} returns 1 for a pair of sizes for
10084 modes for which @code{MODES_TIEABLE_P} is 0, suboptimal code can result.
10085 If this is the case, making @code{TRULY_NOOP_TRUNCATION} return 0 in
10086 such cases may improve things.
10089 @deftypefn {Target Hook} int TARGET_MODE_REP_EXTENDED (enum machine_mode @var{mode}, enum machine_mode @var{rep_mode})
10090 The representation of an integral mode can be such that the values
10091 are always extended to a wider integral mode. Return
10092 @code{SIGN_EXTEND} if values of @var{mode} are represented in
10093 sign-extended form to @var{rep_mode}. Return @code{UNKNOWN}
10094 otherwise. (Currently, none of the targets use zero-extended
10095 representation this way so unlike @code{LOAD_EXTEND_OP},
10096 @code{TARGET_MODE_REP_EXTENDED} is expected to return either
10097 @code{SIGN_EXTEND} or @code{UNKNOWN}. Also no target extends
10098 @var{mode} to @var{mode_rep} so that @var{mode_rep} is not the next
10099 widest integral mode and currently we take advantage of this fact.)
10101 Similarly to @code{LOAD_EXTEND_OP} you may return a non-@code{UNKNOWN}
10102 value even if the extension is not performed on certain hard registers
10103 as long as for the @code{REGNO_REG_CLASS} of these hard registers
10104 @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero.
10106 Note that @code{TARGET_MODE_REP_EXTENDED} and @code{LOAD_EXTEND_OP}
10107 describe two related properties. If you define
10108 @code{TARGET_MODE_REP_EXTENDED (mode, word_mode)} you probably also want
10109 to define @code{LOAD_EXTEND_OP (mode)} to return the same type of
10112 In order to enforce the representation of @code{mode},
10113 @code{TRULY_NOOP_TRUNCATION} should return false when truncating to
10117 @defmac STORE_FLAG_VALUE
10118 A C expression describing the value returned by a comparison operator
10119 with an integral mode and stored by a store-flag instruction
10120 (@samp{s@var{cond}}) when the condition is true. This description must
10121 apply to @emph{all} the @samp{s@var{cond}} patterns and all the
10122 comparison operators whose results have a @code{MODE_INT} mode.
10124 A value of 1 or @minus{}1 means that the instruction implementing the
10125 comparison operator returns exactly 1 or @minus{}1 when the comparison is true
10126 and 0 when the comparison is false. Otherwise, the value indicates
10127 which bits of the result are guaranteed to be 1 when the comparison is
10128 true. This value is interpreted in the mode of the comparison
10129 operation, which is given by the mode of the first operand in the
10130 @samp{s@var{cond}} pattern. Either the low bit or the sign bit of
10131 @code{STORE_FLAG_VALUE} be on. Presently, only those bits are used by
10134 If @code{STORE_FLAG_VALUE} is neither 1 or @minus{}1, the compiler will
10135 generate code that depends only on the specified bits. It can also
10136 replace comparison operators with equivalent operations if they cause
10137 the required bits to be set, even if the remaining bits are undefined.
10138 For example, on a machine whose comparison operators return an
10139 @code{SImode} value and where @code{STORE_FLAG_VALUE} is defined as
10140 @samp{0x80000000}, saying that just the sign bit is relevant, the
10144 (ne:SI (and:SI @var{x} (const_int @var{power-of-2})) (const_int 0))
10148 can be converted to
10151 (ashift:SI @var{x} (const_int @var{n}))
10155 where @var{n} is the appropriate shift count to move the bit being
10156 tested into the sign bit.
10158 There is no way to describe a machine that always sets the low-order bit
10159 for a true value, but does not guarantee the value of any other bits,
10160 but we do not know of any machine that has such an instruction. If you
10161 are trying to port GCC to such a machine, include an instruction to
10162 perform a logical-and of the result with 1 in the pattern for the
10163 comparison operators and let us know at @email{gcc@@gcc.gnu.org}.
10165 Often, a machine will have multiple instructions that obtain a value
10166 from a comparison (or the condition codes). Here are rules to guide the
10167 choice of value for @code{STORE_FLAG_VALUE}, and hence the instructions
10172 Use the shortest sequence that yields a valid definition for
10173 @code{STORE_FLAG_VALUE}. It is more efficient for the compiler to
10174 ``normalize'' the value (convert it to, e.g., 1 or 0) than for the
10175 comparison operators to do so because there may be opportunities to
10176 combine the normalization with other operations.
10179 For equal-length sequences, use a value of 1 or @minus{}1, with @minus{}1 being
10180 slightly preferred on machines with expensive jumps and 1 preferred on
10184 As a second choice, choose a value of @samp{0x80000001} if instructions
10185 exist that set both the sign and low-order bits but do not define the
10189 Otherwise, use a value of @samp{0x80000000}.
10192 Many machines can produce both the value chosen for
10193 @code{STORE_FLAG_VALUE} and its negation in the same number of
10194 instructions. On those machines, you should also define a pattern for
10195 those cases, e.g., one matching
10198 (set @var{A} (neg:@var{m} (ne:@var{m} @var{B} @var{C})))
10201 Some machines can also perform @code{and} or @code{plus} operations on
10202 condition code values with less instructions than the corresponding
10203 @samp{s@var{cond}} insn followed by @code{and} or @code{plus}. On those
10204 machines, define the appropriate patterns. Use the names @code{incscc}
10205 and @code{decscc}, respectively, for the patterns which perform
10206 @code{plus} or @code{minus} operations on condition code values. See
10207 @file{rs6000.md} for some examples. The GNU Superoptizer can be used to
10208 find such instruction sequences on other machines.
10210 If this macro is not defined, the default value, 1, is used. You need
10211 not define @code{STORE_FLAG_VALUE} if the machine has no store-flag
10212 instructions, or if the value generated by these instructions is 1.
10215 @defmac FLOAT_STORE_FLAG_VALUE (@var{mode})
10216 A C expression that gives a nonzero @code{REAL_VALUE_TYPE} value that is
10217 returned when comparison operators with floating-point results are true.
10218 Define this macro on machines that have comparison operations that return
10219 floating-point values. If there are no such operations, do not define
10223 @defmac VECTOR_STORE_FLAG_VALUE (@var{mode})
10224 A C expression that gives a rtx representing the nonzero true element
10225 for vector comparisons. The returned rtx should be valid for the inner
10226 mode of @var{mode} which is guaranteed to be a vector mode. Define
10227 this macro on machines that have vector comparison operations that
10228 return a vector result. If there are no such operations, do not define
10229 this macro. Typically, this macro is defined as @code{const1_rtx} or
10230 @code{constm1_rtx}. This macro may return @code{NULL_RTX} to prevent
10231 the compiler optimizing such vector comparison operations for the
10235 @defmac CLZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
10236 @defmacx CTZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
10237 A C expression that indicates whether the architecture defines a value
10238 for @code{clz} or @code{ctz} with a zero operand.
10239 A result of @code{0} indicates the value is undefined.
10240 If the value is defined for only the RTL expression, the macro should
10241 evaluate to @code{1}; if the value applies also to the corresponding optab
10242 entry (which is normally the case if it expands directly into
10243 the corresponding RTL), then the macro should evaluate to @code{2}.
10244 In the cases where the value is defined, @var{value} should be set to
10247 If this macro is not defined, the value of @code{clz} or
10248 @code{ctz} at zero is assumed to be undefined.
10250 This macro must be defined if the target's expansion for @code{ffs}
10251 relies on a particular value to get correct results. Otherwise it
10252 is not necessary, though it may be used to optimize some corner cases, and
10253 to provide a default expansion for the @code{ffs} optab.
10255 Note that regardless of this macro the ``definedness'' of @code{clz}
10256 and @code{ctz} at zero do @emph{not} extend to the builtin functions
10257 visible to the user. Thus one may be free to adjust the value at will
10258 to match the target expansion of these operations without fear of
10263 An alias for the machine mode for pointers. On most machines, define
10264 this to be the integer mode corresponding to the width of a hardware
10265 pointer; @code{SImode} on 32-bit machine or @code{DImode} on 64-bit machines.
10266 On some machines you must define this to be one of the partial integer
10267 modes, such as @code{PSImode}.
10269 The width of @code{Pmode} must be at least as large as the value of
10270 @code{POINTER_SIZE}. If it is not equal, you must define the macro
10271 @code{POINTERS_EXTEND_UNSIGNED} to specify how pointers are extended
10275 @defmac FUNCTION_MODE
10276 An alias for the machine mode used for memory references to functions
10277 being called, in @code{call} RTL expressions. On most CISC machines,
10278 where an instruction can begin at any byte address, this should be
10279 @code{QImode}. On most RISC machines, where all instructions have fixed
10280 size and alignment, this should be a mode with the same size and alignment
10281 as the machine instruction words - typically @code{SImode} or @code{HImode}.
10284 @defmac STDC_0_IN_SYSTEM_HEADERS
10285 In normal operation, the preprocessor expands @code{__STDC__} to the
10286 constant 1, to signify that GCC conforms to ISO Standard C@. On some
10287 hosts, like Solaris, the system compiler uses a different convention,
10288 where @code{__STDC__} is normally 0, but is 1 if the user specifies
10289 strict conformance to the C Standard.
10291 Defining @code{STDC_0_IN_SYSTEM_HEADERS} makes GNU CPP follows the host
10292 convention when processing system header files, but when processing user
10293 files @code{__STDC__} will always expand to 1.
10296 @defmac NO_IMPLICIT_EXTERN_C
10297 Define this macro if the system header files support C++ as well as C@.
10298 This macro inhibits the usual method of using system header files in
10299 C++, which is to pretend that the file's contents are enclosed in
10300 @samp{extern "C" @{@dots{}@}}.
10305 @defmac REGISTER_TARGET_PRAGMAS ()
10306 Define this macro if you want to implement any target-specific pragmas.
10307 If defined, it is a C expression which makes a series of calls to
10308 @code{c_register_pragma} or @code{c_register_pragma_with_expansion}
10309 for each pragma. The macro may also do any
10310 setup required for the pragmas.
10312 The primary reason to define this macro is to provide compatibility with
10313 other compilers for the same target. In general, we discourage
10314 definition of target-specific pragmas for GCC@.
10316 If the pragma can be implemented by attributes then you should consider
10317 defining the target hook @samp{TARGET_INSERT_ATTRIBUTES} as well.
10319 Preprocessor macros that appear on pragma lines are not expanded. All
10320 @samp{#pragma} directives that do not match any registered pragma are
10321 silently ignored, unless the user specifies @option{-Wunknown-pragmas}.
10324 @deftypefun void c_register_pragma (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
10325 @deftypefunx void c_register_pragma_with_expansion (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
10327 Each call to @code{c_register_pragma} or
10328 @code{c_register_pragma_with_expansion} establishes one pragma. The
10329 @var{callback} routine will be called when the preprocessor encounters a
10333 #pragma [@var{space}] @var{name} @dots{}
10336 @var{space} is the case-sensitive namespace of the pragma, or
10337 @code{NULL} to put the pragma in the global namespace. The callback
10338 routine receives @var{pfile} as its first argument, which can be passed
10339 on to cpplib's functions if necessary. You can lex tokens after the
10340 @var{name} by calling @code{pragma_lex}. Tokens that are not read by the
10341 callback will be silently ignored. The end of the line is indicated by
10342 a token of type @code{CPP_EOF}. Macro expansion occurs on the
10343 arguments of pragmas registered with
10344 @code{c_register_pragma_with_expansion} but not on the arguments of
10345 pragmas registered with @code{c_register_pragma}.
10347 Note that the use of @code{pragma_lex} is specific to the C and C++
10348 compilers. It will not work in the Java or Fortran compilers, or any
10349 other language compilers for that matter. Thus if @code{pragma_lex} is going
10350 to be called from target-specific code, it must only be done so when
10351 building the C and C++ compilers. This can be done by defining the
10352 variables @code{c_target_objs} and @code{cxx_target_objs} in the
10353 target entry in the @file{config.gcc} file. These variables should name
10354 the target-specific, language-specific object file which contains the
10355 code that uses @code{pragma_lex}. Note it will also be necessary to add a
10356 rule to the makefile fragment pointed to by @code{tmake_file} that shows
10357 how to build this object file.
10362 @defmac HANDLE_SYSV_PRAGMA
10363 Define this macro (to a value of 1) if you want the System V style
10364 pragmas @samp{#pragma pack(<n>)} and @samp{#pragma weak <name>
10365 [=<value>]} to be supported by gcc.
10367 The pack pragma specifies the maximum alignment (in bytes) of fields
10368 within a structure, in much the same way as the @samp{__aligned__} and
10369 @samp{__packed__} @code{__attribute__}s do. A pack value of zero resets
10370 the behavior to the default.
10372 A subtlety for Microsoft Visual C/C++ style bit-field packing
10373 (e.g.@: -mms-bitfields) for targets that support it:
10374 When a bit-field is inserted into a packed record, the whole size
10375 of the underlying type is used by one or more same-size adjacent
10376 bit-fields (that is, if its long:3, 32 bits is used in the record,
10377 and any additional adjacent long bit-fields are packed into the same
10378 chunk of 32 bits. However, if the size changes, a new field of that
10379 size is allocated).
10381 If both MS bit-fields and @samp{__attribute__((packed))} are used,
10382 the latter will take precedence. If @samp{__attribute__((packed))} is
10383 used on a single field when MS bit-fields are in use, it will take
10384 precedence for that field, but the alignment of the rest of the structure
10385 may affect its placement.
10387 The weak pragma only works if @code{SUPPORTS_WEAK} and
10388 @code{ASM_WEAKEN_LABEL} are defined. If enabled it allows the creation
10389 of specifically named weak labels, optionally with a value.
10394 @defmac HANDLE_PRAGMA_PACK_PUSH_POP
10395 Define this macro (to a value of 1) if you want to support the Win32
10396 style pragmas @samp{#pragma pack(push[,@var{n}])} and @samp{#pragma
10397 pack(pop)}. The @samp{pack(push,[@var{n}])} pragma specifies the maximum
10398 alignment (in bytes) of fields within a structure, in much the same way as
10399 the @samp{__aligned__} and @samp{__packed__} @code{__attribute__}s do. A
10400 pack value of zero resets the behavior to the default. Successive
10401 invocations of this pragma cause the previous values to be stacked, so
10402 that invocations of @samp{#pragma pack(pop)} will return to the previous
10406 @defmac HANDLE_PRAGMA_PACK_WITH_EXPANSION
10407 Define this macro, as well as
10408 @code{HANDLE_SYSV_PRAGMA}, if macros should be expanded in the
10409 arguments of @samp{#pragma pack}.
10412 @defmac TARGET_DEFAULT_PACK_STRUCT
10413 If your target requires a structure packing default other than 0 (meaning
10414 the machine default), define this macro to the necessary value (in bytes).
10415 This must be a value that would also be valid to use with
10416 @samp{#pragma pack()} (that is, a small power of two).
10421 @defmac HANDLE_PRAGMA_PUSH_POP_MACRO
10422 Define this macro if you want to support the Win32 style pragmas
10423 @samp{#pragma push_macro(macro-name-as-string)} and @samp{#pragma
10424 pop_macro(macro-name-as-string)}. The @samp{#pragma push_macro(
10425 macro-name-as-string)} pragma saves the named macro and via
10426 @samp{#pragma pop_macro(macro-name-as-string)} it will return to the
10431 @defmac DOLLARS_IN_IDENTIFIERS
10432 Define this macro to control use of the character @samp{$} in
10433 identifier names for the C family of languages. 0 means @samp{$} is
10434 not allowed by default; 1 means it is allowed. 1 is the default;
10435 there is no need to define this macro in that case.
10438 @defmac NO_DOLLAR_IN_LABEL
10439 Define this macro if the assembler does not accept the character
10440 @samp{$} in label names. By default constructors and destructors in
10441 G++ have @samp{$} in the identifiers. If this macro is defined,
10442 @samp{.} is used instead.
10445 @defmac NO_DOT_IN_LABEL
10446 Define this macro if the assembler does not accept the character
10447 @samp{.} in label names. By default constructors and destructors in G++
10448 have names that use @samp{.}. If this macro is defined, these names
10449 are rewritten to avoid @samp{.}.
10452 @defmac INSN_SETS_ARE_DELAYED (@var{insn})
10453 Define this macro as a C expression that is nonzero if it is safe for the
10454 delay slot scheduler to place instructions in the delay slot of @var{insn},
10455 even if they appear to use a resource set or clobbered in @var{insn}.
10456 @var{insn} is always a @code{jump_insn} or an @code{insn}; GCC knows that
10457 every @code{call_insn} has this behavior. On machines where some @code{insn}
10458 or @code{jump_insn} is really a function call and hence has this behavior,
10459 you should define this macro.
10461 You need not define this macro if it would always return zero.
10464 @defmac INSN_REFERENCES_ARE_DELAYED (@var{insn})
10465 Define this macro as a C expression that is nonzero if it is safe for the
10466 delay slot scheduler to place instructions in the delay slot of @var{insn},
10467 even if they appear to set or clobber a resource referenced in @var{insn}.
10468 @var{insn} is always a @code{jump_insn} or an @code{insn}. On machines where
10469 some @code{insn} or @code{jump_insn} is really a function call and its operands
10470 are registers whose use is actually in the subroutine it calls, you should
10471 define this macro. Doing so allows the delay slot scheduler to move
10472 instructions which copy arguments into the argument registers into the delay
10473 slot of @var{insn}.
10475 You need not define this macro if it would always return zero.
10478 @defmac MULTIPLE_SYMBOL_SPACES
10479 Define this macro as a C expression that is nonzero if, in some cases,
10480 global symbols from one translation unit may not be bound to undefined
10481 symbols in another translation unit without user intervention. For
10482 instance, under Microsoft Windows symbols must be explicitly imported
10483 from shared libraries (DLLs).
10485 You need not define this macro if it would always evaluate to zero.
10488 @deftypefn {Target Hook} tree TARGET_MD_ASM_CLOBBERS (tree @var{outputs}, tree @var{inputs}, tree @var{clobbers})
10489 This target hook should add to @var{clobbers} @code{STRING_CST} trees for
10490 any hard regs the port wishes to automatically clobber for an asm.
10491 It should return the result of the last @code{tree_cons} used to add a
10492 clobber. The @var{outputs}, @var{inputs} and @var{clobber} lists are the
10493 corresponding parameters to the asm and may be inspected to avoid
10494 clobbering a register that is an input or output of the asm. You can use
10495 @code{tree_overlaps_hard_reg_set}, declared in @file{tree.h}, to test
10496 for overlap with regards to asm-declared registers.
10499 @defmac MATH_LIBRARY
10500 Define this macro as a C string constant for the linker argument to link
10501 in the system math library, or @samp{""} if the target does not have a
10502 separate math library.
10504 You need only define this macro if the default of @samp{"-lm"} is wrong.
10507 @defmac LIBRARY_PATH_ENV
10508 Define this macro as a C string constant for the environment variable that
10509 specifies where the linker should look for libraries.
10511 You need only define this macro if the default of @samp{"LIBRARY_PATH"}
10515 @defmac TARGET_POSIX_IO
10516 Define this macro if the target supports the following POSIX@ file
10517 functions, access, mkdir and file locking with fcntl / F_SETLKW@.
10518 Defining @code{TARGET_POSIX_IO} will enable the test coverage code
10519 to use file locking when exiting a program, which avoids race conditions
10520 if the program has forked. It will also create directories at run-time
10521 for cross-profiling.
10524 @defmac MAX_CONDITIONAL_EXECUTE
10526 A C expression for the maximum number of instructions to execute via
10527 conditional execution instructions instead of a branch. A value of
10528 @code{BRANCH_COST}+1 is the default if the machine does not use cc0, and
10529 1 if it does use cc0.
10532 @defmac IFCVT_MODIFY_TESTS (@var{ce_info}, @var{true_expr}, @var{false_expr})
10533 Used if the target needs to perform machine-dependent modifications on the
10534 conditionals used for turning basic blocks into conditionally executed code.
10535 @var{ce_info} points to a data structure, @code{struct ce_if_block}, which
10536 contains information about the currently processed blocks. @var{true_expr}
10537 and @var{false_expr} are the tests that are used for converting the
10538 then-block and the else-block, respectively. Set either @var{true_expr} or
10539 @var{false_expr} to a null pointer if the tests cannot be converted.
10542 @defmac IFCVT_MODIFY_MULTIPLE_TESTS (@var{ce_info}, @var{bb}, @var{true_expr}, @var{false_expr})
10543 Like @code{IFCVT_MODIFY_TESTS}, but used when converting more complicated
10544 if-statements into conditions combined by @code{and} and @code{or} operations.
10545 @var{bb} contains the basic block that contains the test that is currently
10546 being processed and about to be turned into a condition.
10549 @defmac IFCVT_MODIFY_INSN (@var{ce_info}, @var{pattern}, @var{insn})
10550 A C expression to modify the @var{PATTERN} of an @var{INSN} that is to
10551 be converted to conditional execution format. @var{ce_info} points to
10552 a data structure, @code{struct ce_if_block}, which contains information
10553 about the currently processed blocks.
10556 @defmac IFCVT_MODIFY_FINAL (@var{ce_info})
10557 A C expression to perform any final machine dependent modifications in
10558 converting code to conditional execution. The involved basic blocks
10559 can be found in the @code{struct ce_if_block} structure that is pointed
10560 to by @var{ce_info}.
10563 @defmac IFCVT_MODIFY_CANCEL (@var{ce_info})
10564 A C expression to cancel any machine dependent modifications in
10565 converting code to conditional execution. The involved basic blocks
10566 can be found in the @code{struct ce_if_block} structure that is pointed
10567 to by @var{ce_info}.
10570 @defmac IFCVT_INIT_EXTRA_FIELDS (@var{ce_info})
10571 A C expression to initialize any extra fields in a @code{struct ce_if_block}
10572 structure, which are defined by the @code{IFCVT_EXTRA_FIELDS} macro.
10575 @defmac IFCVT_EXTRA_FIELDS
10576 If defined, it should expand to a set of field declarations that will be
10577 added to the @code{struct ce_if_block} structure. These should be initialized
10578 by the @code{IFCVT_INIT_EXTRA_FIELDS} macro.
10581 @deftypefn {Target Hook} void TARGET_MACHINE_DEPENDENT_REORG ()
10582 If non-null, this hook performs a target-specific pass over the
10583 instruction stream. The compiler will run it at all optimization levels,
10584 just before the point at which it normally does delayed-branch scheduling.
10586 The exact purpose of the hook varies from target to target. Some use
10587 it to do transformations that are necessary for correctness, such as
10588 laying out in-function constant pools or avoiding hardware hazards.
10589 Others use it as an opportunity to do some machine-dependent optimizations.
10591 You need not implement the hook if it has nothing to do. The default
10592 definition is null.
10595 @deftypefn {Target Hook} void TARGET_INIT_BUILTINS ()
10596 Define this hook if you have any machine-specific built-in functions
10597 that need to be defined. It should be a function that performs the
10600 Machine specific built-in functions can be useful to expand special machine
10601 instructions that would otherwise not normally be generated because
10602 they have no equivalent in the source language (for example, SIMD vector
10603 instructions or prefetch instructions).
10605 To create a built-in function, call the function
10606 @code{lang_hooks.builtin_function}
10607 which is defined by the language front end. You can use any type nodes set
10608 up by @code{build_common_tree_nodes} and @code{build_common_tree_nodes_2};
10609 only language front ends that use those two functions will call
10610 @samp{TARGET_INIT_BUILTINS}.
10613 @deftypefn {Target Hook} tree TARGET_BUILTIN_FUNCTION (unsigned @var{code}, bool @var{initialize_p})
10614 Define this hook if you have any machine-specific built-in functions
10615 that need to be defined. It should be a function that returns the
10616 builtin function declaration for the builtin function code @var{code}.
10617 If there is no such builtin and it cannot be initialized at this time
10618 if @var{initialize_p} is true the function should return @code{NULL_TREE}.
10619 If @var{code} is out of range the function should return
10620 @code{error_mark_node}.
10623 @deftypefn {Target Hook} rtx TARGET_EXPAND_BUILTIN (tree @var{exp}, rtx @var{target}, rtx @var{subtarget}, enum machine_mode @var{mode}, int @var{ignore})
10625 Expand a call to a machine specific built-in function that was set up by
10626 @samp{TARGET_INIT_BUILTINS}. @var{exp} is the expression for the
10627 function call; the result should go to @var{target} if that is
10628 convenient, and have mode @var{mode} if that is convenient.
10629 @var{subtarget} may be used as the target for computing one of
10630 @var{exp}'s operands. @var{ignore} is nonzero if the value is to be
10631 ignored. This function should return the result of the call to the
10635 @deftypefn {Target Hook} tree TARGET_RESOLVE_OVERLOADED_BUILTIN (tree @var{fndecl}, tree @var{arglist})
10637 Select a replacement for a machine specific built-in function that
10638 was set up by @samp{TARGET_INIT_BUILTINS}. This is done
10639 @emph{before} regular type checking, and so allows the target to
10640 implement a crude form of function overloading. @var{fndecl} is the
10641 declaration of the built-in function. @var{arglist} is the list of
10642 arguments passed to the built-in function. The result is a
10643 complete expression that implements the operation, usually
10644 another @code{CALL_EXPR}.
10647 @deftypefn {Target Hook} tree TARGET_FOLD_BUILTIN (tree @var{fndecl}, tree @var{arglist}, bool @var{ignore})
10649 Fold a call to a machine specific built-in function that was set up by
10650 @samp{TARGET_INIT_BUILTINS}. @var{fndecl} is the declaration of the
10651 built-in function. @var{arglist} is the list of arguments passed to
10652 the built-in function. The result is another tree containing a
10653 simplified expression for the call's result. If @var{ignore} is true
10654 the value will be ignored.
10657 @deftypefn {Target Hook} const char * TARGET_INVALID_WITHIN_DOLOOP (rtx @var{insn})
10659 Take an instruction in @var{insn} and return NULL if it is valid within a
10660 low-overhead loop, otherwise return a string why doloop could not be applied.
10662 Many targets use special registers for low-overhead looping. For any
10663 instruction that clobbers these this function should return a string indicating
10664 the reason why the doloop could not be applied.
10665 By default, the RTL loop optimizer does not use a present doloop pattern for
10666 loops containing function calls or branch on table instructions.
10669 @defmac MD_CAN_REDIRECT_BRANCH (@var{branch1}, @var{branch2})
10671 Take a branch insn in @var{branch1} and another in @var{branch2}.
10672 Return true if redirecting @var{branch1} to the destination of
10673 @var{branch2} is possible.
10675 On some targets, branches may have a limited range. Optimizing the
10676 filling of delay slots can result in branches being redirected, and this
10677 may in turn cause a branch offset to overflow.
10680 @deftypefn {Target Hook} bool TARGET_COMMUTATIVE_P (rtx @var{x}, @var{outer_code})
10681 This target hook returns @code{true} if @var{x} is considered to be commutative.
10682 Usually, this is just COMMUTATIVE_P (@var{x}), but the HP PA doesn't consider
10683 PLUS to be commutative inside a MEM@. @var{outer_code} is the rtx code
10684 of the enclosing rtl, if known, otherwise it is UNKNOWN.
10687 @deftypefn {Target Hook} rtx TARGET_ALLOCATE_INITIAL_VALUE (rtx @var{hard_reg})
10689 When the initial value of a hard register has been copied in a pseudo
10690 register, it is often not necessary to actually allocate another register
10691 to this pseudo register, because the original hard register or a stack slot
10692 it has been saved into can be used. @code{TARGET_ALLOCATE_INITIAL_VALUE}
10693 is called at the start of register allocation once for each hard register
10694 that had its initial value copied by using
10695 @code{get_func_hard_reg_initial_val} or @code{get_hard_reg_initial_val}.
10696 Possible values are @code{NULL_RTX}, if you don't want
10697 to do any special allocation, a @code{REG} rtx---that would typically be
10698 the hard register itself, if it is known not to be clobbered---or a
10700 If you are returning a @code{MEM}, this is only a hint for the allocator;
10701 it might decide to use another register anyways.
10702 You may use @code{current_function_leaf_function} in the hook, functions
10703 that use @code{REG_N_SETS}, to determine if the hard
10704 register in question will not be clobbered.
10705 The default value of this hook is @code{NULL}, which disables any special
10709 @deftypefn {Target Hook} int TARGET_UNSPEC_MAY_TRAP_P (const_rtx @var{x}, unsigned @var{flags})
10710 This target hook returns nonzero if @var{x}, an @code{unspec} or
10711 @code{unspec_volatile} operation, might cause a trap. Targets can use
10712 this hook to enhance precision of analysis for @code{unspec} and
10713 @code{unspec_volatile} operations. You may call @code{may_trap_p_1}
10714 to analyze inner elements of @var{x} in which case @var{flags} should be
10718 @deftypefn {Target Hook} void TARGET_SET_CURRENT_FUNCTION (tree @var{decl})
10719 The compiler invokes this hook whenever it changes its current function
10720 context (@code{cfun}). You can define this function if
10721 the back end needs to perform any initialization or reset actions on a
10722 per-function basis. For example, it may be used to implement function
10723 attributes that affect register usage or code generation patterns.
10724 The argument @var{decl} is the declaration for the new function context,
10725 and may be null to indicate that the compiler has left a function context
10726 and is returning to processing at the top level.
10727 The default hook function does nothing.
10729 GCC sets @code{cfun} to a dummy function context during initialization of
10730 some parts of the back end. The hook function is not invoked in this
10731 situation; you need not worry about the hook being invoked recursively,
10732 or when the back end is in a partially-initialized state.
10735 @defmac TARGET_OBJECT_SUFFIX
10736 Define this macro to be a C string representing the suffix for object
10737 files on your target machine. If you do not define this macro, GCC will
10738 use @samp{.o} as the suffix for object files.
10741 @defmac TARGET_EXECUTABLE_SUFFIX
10742 Define this macro to be a C string representing the suffix to be
10743 automatically added to executable files on your target machine. If you
10744 do not define this macro, GCC will use the null string as the suffix for
10748 @defmac COLLECT_EXPORT_LIST
10749 If defined, @code{collect2} will scan the individual object files
10750 specified on its command line and create an export list for the linker.
10751 Define this macro for systems like AIX, where the linker discards
10752 object files that are not referenced from @code{main} and uses export
10756 @defmac MODIFY_JNI_METHOD_CALL (@var{mdecl})
10757 Define this macro to a C expression representing a variant of the
10758 method call @var{mdecl}, if Java Native Interface (JNI) methods
10759 must be invoked differently from other methods on your target.
10760 For example, on 32-bit Microsoft Windows, JNI methods must be invoked using
10761 the @code{stdcall} calling convention and this macro is then
10762 defined as this expression:
10765 build_type_attribute_variant (@var{mdecl},
10767 (get_identifier ("stdcall"),
10772 @deftypefn {Target Hook} bool TARGET_CANNOT_MODIFY_JUMPS_P (void)
10773 This target hook returns @code{true} past the point in which new jump
10774 instructions could be created. On machines that require a register for
10775 every jump such as the SHmedia ISA of SH5, this point would typically be
10776 reload, so this target hook should be defined to a function such as:
10780 cannot_modify_jumps_past_reload_p ()
10782 return (reload_completed || reload_in_progress);
10787 @deftypefn {Target Hook} int TARGET_BRANCH_TARGET_REGISTER_CLASS (void)
10788 This target hook returns a register class for which branch target register
10789 optimizations should be applied. All registers in this class should be
10790 usable interchangeably. After reload, registers in this class will be
10791 re-allocated and loads will be hoisted out of loops and be subjected
10792 to inter-block scheduling.
10795 @deftypefn {Target Hook} bool TARGET_BRANCH_TARGET_REGISTER_CALLEE_SAVED (bool @var{after_prologue_epilogue_gen})
10796 Branch target register optimization will by default exclude callee-saved
10798 that are not already live during the current function; if this target hook
10799 returns true, they will be included. The target code must than make sure
10800 that all target registers in the class returned by
10801 @samp{TARGET_BRANCH_TARGET_REGISTER_CLASS} that might need saving are
10802 saved. @var{after_prologue_epilogue_gen} indicates if prologues and
10803 epilogues have already been generated. Note, even if you only return
10804 true when @var{after_prologue_epilogue_gen} is false, you still are likely
10805 to have to make special provisions in @code{INITIAL_ELIMINATION_OFFSET}
10806 to reserve space for caller-saved target registers.
10809 @defmac POWI_MAX_MULTS
10810 If defined, this macro is interpreted as a signed integer C expression
10811 that specifies the maximum number of floating point multiplications
10812 that should be emitted when expanding exponentiation by an integer
10813 constant inline. When this value is defined, exponentiation requiring
10814 more than this number of multiplications is implemented by calling the
10815 system library's @code{pow}, @code{powf} or @code{powl} routines.
10816 The default value places no upper bound on the multiplication count.
10819 @deftypefn Macro void TARGET_EXTRA_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
10820 This target hook should register any extra include files for the
10821 target. The parameter @var{stdinc} indicates if normal include files
10822 are present. The parameter @var{sysroot} is the system root directory.
10823 The parameter @var{iprefix} is the prefix for the gcc directory.
10826 @deftypefn Macro void TARGET_EXTRA_PRE_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
10827 This target hook should register any extra include files for the
10828 target before any standard headers. The parameter @var{stdinc}
10829 indicates if normal include files are present. The parameter
10830 @var{sysroot} is the system root directory. The parameter
10831 @var{iprefix} is the prefix for the gcc directory.
10834 @deftypefn Macro void TARGET_OPTF (char *@var{path})
10835 This target hook should register special include paths for the target.
10836 The parameter @var{path} is the include to register. On Darwin
10837 systems, this is used for Framework includes, which have semantics
10838 that are different from @option{-I}.
10841 @deftypefn {Target Hook} bool TARGET_USE_LOCAL_THUNK_ALIAS_P (tree @var{fndecl})
10842 This target hook returns @code{true} if it is safe to use a local alias
10843 for a virtual function @var{fndecl} when constructing thunks,
10844 @code{false} otherwise. By default, the hook returns @code{true} for all
10845 functions, if a target supports aliases (i.e.@: defines
10846 @code{ASM_OUTPUT_DEF}), @code{false} otherwise,
10849 @defmac TARGET_FORMAT_TYPES
10850 If defined, this macro is the name of a global variable containing
10851 target-specific format checking information for the @option{-Wformat}
10852 option. The default is to have no target-specific format checks.
10855 @defmac TARGET_N_FORMAT_TYPES
10856 If defined, this macro is the number of entries in
10857 @code{TARGET_FORMAT_TYPES}.
10860 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES
10861 If defined, this macro is the name of a global variable containing
10862 target-specific format overrides for the @option{-Wformat} option. The
10863 default is to have no target-specific format overrides. If defined,
10864 @code{TARGET_FORMAT_TYPES} must be defined, too.
10867 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES_COUNT
10868 If defined, this macro specifies the number of entries in
10869 @code{TARGET_OVERRIDES_FORMAT_ATTRIBUTES}.
10872 @defmac TARGET_OVERRIDES_FORMAT_INIT
10873 If defined, this macro specifies the optional initialization
10874 routine for target specific customizations of the system printf
10875 and scanf formatter settings.
10878 @deftypefn {Target Hook} bool TARGET_RELAXED_ORDERING
10879 If set to @code{true}, means that the target's memory model does not
10880 guarantee that loads which do not depend on one another will access
10881 main memory in the order of the instruction stream; if ordering is
10882 important, an explicit memory barrier must be used. This is true of
10883 many recent processors which implement a policy of ``relaxed,''
10884 ``weak,'' or ``release'' memory consistency, such as Alpha, PowerPC,
10885 and ia64. The default is @code{false}.
10888 @deftypefn {Target Hook} const char *TARGET_INVALID_ARG_FOR_UNPROTOTYPED_FN (tree @var{typelist}, tree @var{funcdecl}, tree @var{val})
10889 If defined, this macro returns the diagnostic message when it is
10890 illegal to pass argument @var{val} to function @var{funcdecl}
10891 with prototype @var{typelist}.
10894 @deftypefn {Target Hook} {const char *} TARGET_INVALID_CONVERSION (tree @var{fromtype}, tree @var{totype})
10895 If defined, this macro returns the diagnostic message when it is
10896 invalid to convert from @var{fromtype} to @var{totype}, or @code{NULL}
10897 if validity should be determined by the front end.
10900 @deftypefn {Target Hook} {const char *} TARGET_INVALID_UNARY_OP (int @var{op}, tree @var{type})
10901 If defined, this macro returns the diagnostic message when it is
10902 invalid to apply operation @var{op} (where unary plus is denoted by
10903 @code{CONVERT_EXPR}) to an operand of type @var{type}, or @code{NULL}
10904 if validity should be determined by the front end.
10907 @deftypefn {Target Hook} {const char *} TARGET_INVALID_BINARY_OP (int @var{op}, tree @var{type1}, tree @var{type2})
10908 If defined, this macro returns the diagnostic message when it is
10909 invalid to apply operation @var{op} to operands of types @var{type1}
10910 and @var{type2}, or @code{NULL} if validity should be determined by
10914 @deftypefn {Target Hook} {const char *} TARGET_INVALID_PARAMETER_TYPE (tree @var{type})
10915 If defined, this macro returns the diagnostic message when it is
10916 invalid for functions to include parameters of type @var{type},
10917 or @code{NULL} if validity should be determined by
10918 the front end. This is currently used only by the C and C++ front ends.
10921 @deftypefn {Target Hook} {const char *} TARGET_INVALID_RETURN_TYPE (tree @var{type})
10922 If defined, this macro returns the diagnostic message when it is
10923 invalid for functions to have return type @var{type},
10924 or @code{NULL} if validity should be determined by
10925 the front end. This is currently used only by the C and C++ front ends.
10928 @deftypefn {Target Hook} {tree} TARGET_PROMOTED_TYPE (tree @var{type})
10929 If defined, this target hook returns the type to which values of
10930 @var{type} should be promoted when they appear in expressions,
10931 analogous to the integer promotions, or @code{NULL_TREE} to use the
10932 front end's normal promotion rules. This hook is useful when there are
10933 target-specific types with special promotion rules.
10934 This is currently used only by the C and C++ front ends.
10937 @deftypefn {Target Hook} {tree} TARGET_CONVERT_TO_TYPE (tree @var{type}, tree @var{expr})
10938 If defined, this hook returns the result of converting @var{expr} to
10939 @var{type}. It should return the converted expression,
10940 or @code{NULL_TREE} to apply the front end's normal conversion rules.
10941 This hook is useful when there are target-specific types with special
10943 This is currently used only by the C and C++ front ends.
10946 @defmac TARGET_USE_JCR_SECTION
10947 This macro determines whether to use the JCR section to register Java
10948 classes. By default, TARGET_USE_JCR_SECTION is defined to 1 if both
10949 SUPPORTS_WEAK and TARGET_HAVE_NAMED_SECTIONS are true, else 0.
10953 This macro determines the size of the objective C jump buffer for the
10954 NeXT runtime. By default, OBJC_JBLEN is defined to an innocuous value.
10957 @defmac LIBGCC2_UNWIND_ATTRIBUTE
10958 Define this macro if any target-specific attributes need to be attached
10959 to the functions in @file{libgcc} that provide low-level support for
10960 call stack unwinding. It is used in declarations in @file{unwind-generic.h}
10961 and the associated definitions of those functions.
10964 @deftypefn {Target Hook} void TARGET_UPDATE_STACK_BOUNDARY (void)
10965 Define this macro to update the current function stack boundary if
10969 @deftypefn {Target Hook} rtx TARGET_GET_DRAP_RTX (void)
10970 Define this macro to an rtx for Dynamic Realign Argument Pointer if a
10971 different argument pointer register is needed to access the function's
10972 argument list when stack is aligned.
10975 @deftypefn {Target Hook} {bool} TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS (void)
10976 When optimization is disabled, this hook indicates whether or not
10977 arguments should be allocated to stack slots. Normally, GCC allocates
10978 stacks slots for arguments when not optimizing in order to make
10979 debugging easier. However, when a function is declared with
10980 @code{__attribute__((naked))}, there is no stack frame, and the compiler
10981 cannot safely move arguments from the registers in which they are passed
10982 to the stack. Therefore, this hook should return true in general, but
10983 false for naked functions. The default implementation always returns true.
10986 @deftypevr {Target Hook} {unsigned HOST_WIDE_INT} TARGET_CONST_ANCHOR
10987 On some architectures it can take multiple instructions to synthesize
10988 a constant. If there is another constant already in a register that
10989 is close enough in value then it is preferable that the new constant
10990 is computed from this register using immediate addition or
10991 substraction. We accomplish this through CSE. Besides the value of
10992 the constant we also add a lower and an upper constant anchor to the
10993 available expressions. These are then queried when encountering new
10994 constants. The anchors are computed by rounding the constant up and
10995 down to a multiple of the value of @code{TARGET_CONST_ANCHOR}.
10996 @code{TARGET_CONST_ANCHOR} should be the maximum positive value
10997 accepted by immediate-add plus one. We currently assume that the
10998 value of @code{TARGET_CONST_ANCHOR} is a power of 2. For example, on
10999 MIPS, where add-immediate takes a 16-bit signed value,
11000 @code{TARGET_CONST_ANCHOR} is set to @samp{0x8000}. The default value
11001 is zero, which disables this optimization. @end deftypevr