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, 2010
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 TARGET_OPTION_TRANSLATE_TABLE
145 If defined, a list of pairs of strings, the first of which is a
146 potential command line target to the @file{gcc} driver program, and the
147 second of which is a space-separated (tabs and other whitespace are not
148 supported) list of options with which to replace the first option. The
149 target defining this list is responsible for assuring that the results
150 are valid. Replacement options may not be the @code{--opt} style, they
151 must be the @code{-opt} style. It is the intention of this macro to
152 provide a mechanism for substitution that affects the multilibs chosen,
153 such as one option that enables many options, some of which select
154 multilibs. Example nonsensical definition, where @option{-malt-abi},
155 @option{-EB}, and @option{-mspoo} cause different multilibs to be chosen:
158 #define TARGET_OPTION_TRANSLATE_TABLE \
159 @{ "-fast", "-march=fast-foo -malt-abi -I/usr/fast-foo" @}, \
160 @{ "-compat", "-EB -malign=4 -mspoo" @}
164 @defmac DRIVER_SELF_SPECS
165 A list of specs for the driver itself. It should be a suitable
166 initializer for an array of strings, with no surrounding braces.
168 The driver applies these specs to its own command line between loading
169 default @file{specs} files (but not command-line specified ones) and
170 choosing the multilib directory or running any subcommands. It
171 applies them in the order given, so each spec can depend on the
172 options added by earlier ones. It is also possible to remove options
173 using @samp{%<@var{option}} in the usual way.
175 This macro can be useful when a port has several interdependent target
176 options. It provides a way of standardizing the command line so
177 that the other specs are easier to write.
179 Do not define this macro if it does not need to do anything.
182 @defmac OPTION_DEFAULT_SPECS
183 A list of specs used to support configure-time default options (i.e.@:
184 @option{--with} options) in the driver. It should be a suitable initializer
185 for an array of structures, each containing two strings, without the
186 outermost pair of surrounding braces.
188 The first item in the pair is the name of the default. This must match
189 the code in @file{config.gcc} for the target. The second item is a spec
190 to apply if a default with this name was specified. The string
191 @samp{%(VALUE)} in the spec will be replaced by the value of the default
192 everywhere it occurs.
194 The driver will apply these specs to its own command line between loading
195 default @file{specs} files and processing @code{DRIVER_SELF_SPECS}, using
196 the same mechanism as @code{DRIVER_SELF_SPECS}.
198 Do not define this macro if it does not need to do anything.
202 A C string constant that tells the GCC driver program options to
203 pass to CPP@. It can also specify how to translate options you
204 give to GCC into options for GCC to pass to the CPP@.
206 Do not define this macro if it does not need to do anything.
209 @defmac CPLUSPLUS_CPP_SPEC
210 This macro is just like @code{CPP_SPEC}, but is used for C++, rather
211 than C@. If you do not define this macro, then the value of
212 @code{CPP_SPEC} (if any) will be used instead.
216 A C string constant that tells the GCC driver program options to
217 pass to @code{cc1}, @code{cc1plus}, @code{f771}, and the other language
219 It can also specify how to translate options you give to GCC into options
220 for GCC to pass to front ends.
222 Do not define this macro if it does not need to do anything.
226 A C string constant that tells the GCC driver program options to
227 pass to @code{cc1plus}. It can also specify how to translate options you
228 give to GCC into options for GCC to pass to the @code{cc1plus}.
230 Do not define this macro if it does not need to do anything.
231 Note that everything defined in CC1_SPEC is already passed to
232 @code{cc1plus} so there is no need to duplicate the contents of
233 CC1_SPEC in CC1PLUS_SPEC@.
237 A C string constant that tells the GCC driver program options to
238 pass to the assembler. It can also specify how to translate options
239 you give to GCC into options for GCC to pass to the assembler.
240 See the file @file{sun3.h} for an example of this.
242 Do not define this macro if it does not need to do anything.
245 @defmac ASM_FINAL_SPEC
246 A C string constant that tells the GCC driver program how to
247 run any programs which cleanup after the normal assembler.
248 Normally, this is not needed. See the file @file{mips.h} for
251 Do not define this macro if it does not need to do anything.
254 @defmac AS_NEEDS_DASH_FOR_PIPED_INPUT
255 Define this macro, with no value, if the driver should give the assembler
256 an argument consisting of a single dash, @option{-}, to instruct it to
257 read from its standard input (which will be a pipe connected to the
258 output of the compiler proper). This argument is given after any
259 @option{-o} option specifying the name of the output file.
261 If you do not define this macro, the assembler is assumed to read its
262 standard input if given no non-option arguments. If your assembler
263 cannot read standard input at all, use a @samp{%@{pipe:%e@}} construct;
264 see @file{mips.h} for instance.
268 A C string constant that tells the GCC driver program options to
269 pass to the linker. It can also specify how to translate options you
270 give to GCC into options for GCC to pass to the linker.
272 Do not define this macro if it does not need to do anything.
276 Another C string constant used much like @code{LINK_SPEC}. The difference
277 between the two is that @code{LIB_SPEC} is used at the end of the
278 command given to the linker.
280 If this macro is not defined, a default is provided that
281 loads the standard C library from the usual place. See @file{gcc.c}.
285 Another C string constant that tells the GCC driver program
286 how and when to place a reference to @file{libgcc.a} into the
287 linker command line. This constant is placed both before and after
288 the value of @code{LIB_SPEC}.
290 If this macro is not defined, the GCC driver provides a default that
291 passes the string @option{-lgcc} to the linker.
294 @defmac REAL_LIBGCC_SPEC
295 By default, if @code{ENABLE_SHARED_LIBGCC} is defined, the
296 @code{LIBGCC_SPEC} is not directly used by the driver program but is
297 instead modified to refer to different versions of @file{libgcc.a}
298 depending on the values of the command line flags @option{-static},
299 @option{-shared}, @option{-static-libgcc}, and @option{-shared-libgcc}. On
300 targets where these modifications are inappropriate, define
301 @code{REAL_LIBGCC_SPEC} instead. @code{REAL_LIBGCC_SPEC} tells the
302 driver how to place a reference to @file{libgcc} on the link command
303 line, but, unlike @code{LIBGCC_SPEC}, it is used unmodified.
306 @defmac USE_LD_AS_NEEDED
307 A macro that controls the modifications to @code{LIBGCC_SPEC}
308 mentioned in @code{REAL_LIBGCC_SPEC}. If nonzero, a spec will be
309 generated that uses --as-needed and the shared libgcc in place of the
310 static exception handler library, when linking without any of
311 @code{-static}, @code{-static-libgcc}, or @code{-shared-libgcc}.
315 If defined, this C string constant is added to @code{LINK_SPEC}.
316 When @code{USE_LD_AS_NEEDED} is zero or undefined, it also affects
317 the modifications to @code{LIBGCC_SPEC} mentioned in
318 @code{REAL_LIBGCC_SPEC}.
321 @defmac STARTFILE_SPEC
322 Another C string constant used much like @code{LINK_SPEC}. The
323 difference between the two is that @code{STARTFILE_SPEC} is used at
324 the very beginning of the command given to the linker.
326 If this macro is not defined, a default is provided that loads the
327 standard C startup file from the usual place. See @file{gcc.c}.
331 Another C string constant used much like @code{LINK_SPEC}. The
332 difference between the two is that @code{ENDFILE_SPEC} is used at
333 the very end of the command given to the linker.
335 Do not define this macro if it does not need to do anything.
338 @defmac THREAD_MODEL_SPEC
339 GCC @code{-v} will print the thread model GCC was configured to use.
340 However, this doesn't work on platforms that are multilibbed on thread
341 models, such as AIX 4.3. On such platforms, define
342 @code{THREAD_MODEL_SPEC} such that it evaluates to a string without
343 blanks that names one of the recognized thread models. @code{%*}, the
344 default value of this macro, will expand to the value of
345 @code{thread_file} set in @file{config.gcc}.
348 @defmac SYSROOT_SUFFIX_SPEC
349 Define this macro to add a suffix to the target sysroot when GCC is
350 configured with a sysroot. This will cause GCC to search for usr/lib,
351 et al, within sysroot+suffix.
354 @defmac SYSROOT_HEADERS_SUFFIX_SPEC
355 Define this macro to add a headers_suffix to the target sysroot when
356 GCC is configured with a sysroot. This will cause GCC to pass the
357 updated sysroot+headers_suffix to CPP, causing it to search for
358 usr/include, et al, within sysroot+headers_suffix.
362 Define this macro to provide additional specifications to put in the
363 @file{specs} file that can be used in various specifications like
366 The definition should be an initializer for an array of structures,
367 containing a string constant, that defines the specification name, and a
368 string constant that provides the specification.
370 Do not define this macro if it does not need to do anything.
372 @code{EXTRA_SPECS} is useful when an architecture contains several
373 related targets, which have various @code{@dots{}_SPECS} which are similar
374 to each other, and the maintainer would like one central place to keep
377 For example, the PowerPC System V.4 targets use @code{EXTRA_SPECS} to
378 define either @code{_CALL_SYSV} when the System V calling sequence is
379 used or @code{_CALL_AIX} when the older AIX-based calling sequence is
382 The @file{config/rs6000/rs6000.h} target file defines:
385 #define EXTRA_SPECS \
386 @{ "cpp_sysv_default", CPP_SYSV_DEFAULT @},
388 #define CPP_SYS_DEFAULT ""
391 The @file{config/rs6000/sysv.h} target file defines:
395 "%@{posix: -D_POSIX_SOURCE @} \
396 %@{mcall-sysv: -D_CALL_SYSV @} \
397 %@{!mcall-sysv: %(cpp_sysv_default) @} \
398 %@{msoft-float: -D_SOFT_FLOAT@} %@{mcpu=403: -D_SOFT_FLOAT@}"
400 #undef CPP_SYSV_DEFAULT
401 #define CPP_SYSV_DEFAULT "-D_CALL_SYSV"
404 while the @file{config/rs6000/eabiaix.h} target file defines
405 @code{CPP_SYSV_DEFAULT} as:
408 #undef CPP_SYSV_DEFAULT
409 #define CPP_SYSV_DEFAULT "-D_CALL_AIX"
413 @defmac LINK_LIBGCC_SPECIAL_1
414 Define this macro if the driver program should find the library
415 @file{libgcc.a}. If you do not define this macro, the driver program will pass
416 the argument @option{-lgcc} to tell the linker to do the search.
419 @defmac LINK_GCC_C_SEQUENCE_SPEC
420 The sequence in which libgcc and libc are specified to the linker.
421 By default this is @code{%G %L %G}.
424 @defmac LINK_COMMAND_SPEC
425 A C string constant giving the complete command line need to execute the
426 linker. When you do this, you will need to update your port each time a
427 change is made to the link command line within @file{gcc.c}. Therefore,
428 define this macro only if you need to completely redefine the command
429 line for invoking the linker and there is no other way to accomplish
430 the effect you need. Overriding this macro may be avoidable by overriding
431 @code{LINK_GCC_C_SEQUENCE_SPEC} instead.
434 @defmac LINK_ELIMINATE_DUPLICATE_LDIRECTORIES
435 A nonzero value causes @command{collect2} to remove duplicate @option{-L@var{directory}} search
436 directories from linking commands. Do not give it a nonzero value if
437 removing duplicate search directories changes the linker's semantics.
440 @defmac MULTILIB_DEFAULTS
441 Define this macro as a C expression for the initializer of an array of
442 string to tell the driver program which options are defaults for this
443 target and thus do not need to be handled specially when using
444 @code{MULTILIB_OPTIONS}.
446 Do not define this macro if @code{MULTILIB_OPTIONS} is not defined in
447 the target makefile fragment or if none of the options listed in
448 @code{MULTILIB_OPTIONS} are set by default.
449 @xref{Target Fragment}.
452 @defmac RELATIVE_PREFIX_NOT_LINKDIR
453 Define this macro to tell @command{gcc} that it should only translate
454 a @option{-B} prefix into a @option{-L} linker option if the prefix
455 indicates an absolute file name.
458 @defmac MD_EXEC_PREFIX
459 If defined, this macro is an additional prefix to try after
460 @code{STANDARD_EXEC_PREFIX}. @code{MD_EXEC_PREFIX} is not searched
461 when the compiler is built as a cross
462 compiler. If you define @code{MD_EXEC_PREFIX}, then be sure to add it
463 to the list of directories used to find the assembler in @file{configure.in}.
466 @defmac STANDARD_STARTFILE_PREFIX
467 Define this macro as a C string constant if you wish to override the
468 standard choice of @code{libdir} as the default prefix to
469 try when searching for startup files such as @file{crt0.o}.
470 @code{STANDARD_STARTFILE_PREFIX} is not searched when the compiler
471 is built as a cross compiler.
474 @defmac STANDARD_STARTFILE_PREFIX_1
475 Define this macro as a C string constant if you wish to override the
476 standard choice of @code{/lib} as a prefix to try after the default prefix
477 when searching for startup files such as @file{crt0.o}.
478 @code{STANDARD_STARTFILE_PREFIX_1} is not searched when the compiler
479 is built as a cross compiler.
482 @defmac STANDARD_STARTFILE_PREFIX_2
483 Define this macro as a C string constant if you wish to override the
484 standard choice of @code{/lib} as yet another prefix to try after the
485 default prefix when searching for startup files such as @file{crt0.o}.
486 @code{STANDARD_STARTFILE_PREFIX_2} is not searched when the compiler
487 is built as a cross compiler.
490 @defmac MD_STARTFILE_PREFIX
491 If defined, this macro supplies an additional prefix to try after the
492 standard prefixes. @code{MD_EXEC_PREFIX} is not searched when the
493 compiler is built as a cross compiler.
496 @defmac MD_STARTFILE_PREFIX_1
497 If defined, this macro supplies yet another prefix to try after the
498 standard prefixes. It is not searched when the compiler is built as a
502 @defmac INIT_ENVIRONMENT
503 Define this macro as a C string constant if you wish to set environment
504 variables for programs called by the driver, such as the assembler and
505 loader. The driver passes the value of this macro to @code{putenv} to
506 initialize the necessary environment variables.
509 @defmac LOCAL_INCLUDE_DIR
510 Define this macro as a C string constant if you wish to override the
511 standard choice of @file{/usr/local/include} as the default prefix to
512 try when searching for local header files. @code{LOCAL_INCLUDE_DIR}
513 comes before @code{SYSTEM_INCLUDE_DIR} in the search order.
515 Cross compilers do not search either @file{/usr/local/include} or its
519 @defmac SYSTEM_INCLUDE_DIR
520 Define this macro as a C string constant if you wish to specify a
521 system-specific directory to search for header files before the standard
522 directory. @code{SYSTEM_INCLUDE_DIR} comes before
523 @code{STANDARD_INCLUDE_DIR} in the search order.
525 Cross compilers do not use this macro and do not search the directory
529 @defmac STANDARD_INCLUDE_DIR
530 Define this macro as a C string constant if you wish to override the
531 standard choice of @file{/usr/include} as the default prefix to
532 try when searching for header files.
534 Cross compilers ignore this macro and do not search either
535 @file{/usr/include} or its replacement.
538 @defmac STANDARD_INCLUDE_COMPONENT
539 The ``component'' corresponding to @code{STANDARD_INCLUDE_DIR}.
540 See @code{INCLUDE_DEFAULTS}, below, for the description of components.
541 If you do not define this macro, no component is used.
544 @defmac INCLUDE_DEFAULTS
545 Define this macro if you wish to override the entire default search path
546 for include files. For a native compiler, the default search path
547 usually consists of @code{GCC_INCLUDE_DIR}, @code{LOCAL_INCLUDE_DIR},
548 @code{SYSTEM_INCLUDE_DIR}, @code{GPLUSPLUS_INCLUDE_DIR}, and
549 @code{STANDARD_INCLUDE_DIR}. In addition, @code{GPLUSPLUS_INCLUDE_DIR}
550 and @code{GCC_INCLUDE_DIR} are defined automatically by @file{Makefile},
551 and specify private search areas for GCC@. The directory
552 @code{GPLUSPLUS_INCLUDE_DIR} is used only for C++ programs.
554 The definition should be an initializer for an array of structures.
555 Each array element should have four elements: the directory name (a
556 string constant), the component name (also a string constant), a flag
557 for C++-only directories,
558 and a flag showing that the includes in the directory don't need to be
559 wrapped in @code{extern @samp{C}} when compiling C++. Mark the end of
560 the array with a null element.
562 The component name denotes what GNU package the include file is part of,
563 if any, in all uppercase letters. For example, it might be @samp{GCC}
564 or @samp{BINUTILS}. If the package is part of a vendor-supplied
565 operating system, code the component name as @samp{0}.
567 For example, here is the definition used for VAX/VMS:
570 #define INCLUDE_DEFAULTS \
572 @{ "GNU_GXX_INCLUDE:", "G++", 1, 1@}, \
573 @{ "GNU_CC_INCLUDE:", "GCC", 0, 0@}, \
574 @{ "SYS$SYSROOT:[SYSLIB.]", 0, 0, 0@}, \
581 Here is the order of prefixes tried for exec files:
585 Any prefixes specified by the user with @option{-B}.
588 The environment variable @code{GCC_EXEC_PREFIX} or, if @code{GCC_EXEC_PREFIX}
589 is not set and the compiler has not been installed in the configure-time
590 @var{prefix}, the location in which the compiler has actually been installed.
593 The directories specified by the environment variable @code{COMPILER_PATH}.
596 The macro @code{STANDARD_EXEC_PREFIX}, if the compiler has been installed
597 in the configured-time @var{prefix}.
600 The location @file{/usr/libexec/gcc/}, but only if this is a native compiler.
603 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
606 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
610 Here is the order of prefixes tried for startfiles:
614 Any prefixes specified by the user with @option{-B}.
617 The environment variable @code{GCC_EXEC_PREFIX} or its automatically determined
618 value based on the installed toolchain location.
621 The directories specified by the environment variable @code{LIBRARY_PATH}
622 (or port-specific name; native only, cross compilers do not use this).
625 The macro @code{STANDARD_EXEC_PREFIX}, but only if the toolchain is installed
626 in the configured @var{prefix} or this is a native compiler.
629 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
632 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
636 The macro @code{MD_STARTFILE_PREFIX}, if defined, but only if this is a
637 native compiler, or we have a target system root.
640 The macro @code{MD_STARTFILE_PREFIX_1}, if defined, but only if this is a
641 native compiler, or we have a target system root.
644 The macro @code{STANDARD_STARTFILE_PREFIX}, with any sysroot modifications.
645 If this path is relative it will be prefixed by @code{GCC_EXEC_PREFIX} and
646 the machine suffix or @code{STANDARD_EXEC_PREFIX} and the machine suffix.
649 The macro @code{STANDARD_STARTFILE_PREFIX_1}, but only if this is a native
650 compiler, or we have a target system root. The default for this macro is
654 The macro @code{STANDARD_STARTFILE_PREFIX_2}, but only if this is a native
655 compiler, or we have a target system root. The default for this macro is
659 @node Run-time Target
660 @section Run-time Target Specification
661 @cindex run-time target specification
662 @cindex predefined macros
663 @cindex target specifications
665 @c prevent bad page break with this line
666 Here are run-time target specifications.
668 @defmac TARGET_CPU_CPP_BUILTINS ()
669 This function-like macro expands to a block of code that defines
670 built-in preprocessor macros and assertions for the target CPU, using
671 the functions @code{builtin_define}, @code{builtin_define_std} and
672 @code{builtin_assert}. When the front end
673 calls this macro it provides a trailing semicolon, and since it has
674 finished command line option processing your code can use those
677 @code{builtin_assert} takes a string in the form you pass to the
678 command-line option @option{-A}, such as @code{cpu=mips}, and creates
679 the assertion. @code{builtin_define} takes a string in the form
680 accepted by option @option{-D} and unconditionally defines the macro.
682 @code{builtin_define_std} takes a string representing the name of an
683 object-like macro. If it doesn't lie in the user's namespace,
684 @code{builtin_define_std} defines it unconditionally. Otherwise, it
685 defines a version with two leading underscores, and another version
686 with two leading and trailing underscores, and defines the original
687 only if an ISO standard was not requested on the command line. For
688 example, passing @code{unix} defines @code{__unix}, @code{__unix__}
689 and possibly @code{unix}; passing @code{_mips} defines @code{__mips},
690 @code{__mips__} and possibly @code{_mips}, and passing @code{_ABI64}
691 defines only @code{_ABI64}.
693 You can also test for the C dialect being compiled. The variable
694 @code{c_language} is set to one of @code{clk_c}, @code{clk_cplusplus}
695 or @code{clk_objective_c}. Note that if we are preprocessing
696 assembler, this variable will be @code{clk_c} but the function-like
697 macro @code{preprocessing_asm_p()} will return true, so you might want
698 to check for that first. If you need to check for strict ANSI, the
699 variable @code{flag_iso} can be used. The function-like macro
700 @code{preprocessing_trad_p()} can be used to check for traditional
704 @defmac TARGET_OS_CPP_BUILTINS ()
705 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
706 and is used for the target operating system instead.
709 @defmac TARGET_OBJFMT_CPP_BUILTINS ()
710 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
711 and is used for the target object format. @file{elfos.h} uses this
712 macro to define @code{__ELF__}, so you probably do not need to define
716 @deftypevar {extern int} target_flags
717 This variable is declared in @file{options.h}, which is included before
718 any target-specific headers.
721 @deftypevr {Target Hook} int TARGET_DEFAULT_TARGET_FLAGS
722 This variable specifies the initial value of @code{target_flags}.
723 Its default setting is 0.
726 @cindex optional hardware or system features
727 @cindex features, optional, in system conventions
729 @deftypefn {Target Hook} bool TARGET_HANDLE_OPTION (size_t @var{code}, const char *@var{arg}, int @var{value})
730 This hook is called whenever the user specifies one of the
731 target-specific options described by the @file{.opt} definition files
732 (@pxref{Options}). It has the opportunity to do some option-specific
733 processing and should return true if the option is valid. The default
734 definition does nothing but return true.
736 @var{code} specifies the @code{OPT_@var{name}} enumeration value
737 associated with the selected option; @var{name} is just a rendering of
738 the option name in which non-alphanumeric characters are replaced by
739 underscores. @var{arg} specifies the string argument and is null if
740 no argument was given. If the option is flagged as a @code{UInteger}
741 (@pxref{Option properties}), @var{value} is the numeric value of the
742 argument. Otherwise @var{value} is 1 if the positive form of the
743 option was used and 0 if the ``no-'' form was.
746 @deftypefn {Target Hook} bool TARGET_HANDLE_C_OPTION (size_t @var{code}, const char *@var{arg}, int @var{value})
747 This target hook is called whenever the user specifies one of the
748 target-specific C language family options described by the @file{.opt}
749 definition files(@pxref{Options}). It has the opportunity to do some
750 option-specific processing and should return true if the option is
751 valid. The arguments are like for @code{TARGET_HANDLE_OPTION}. The
752 default definition does nothing but return false.
754 In general, you should use @code{TARGET_HANDLE_OPTION} to handle
755 options. However, if processing an option requires routines that are
756 only available in the C (and related language) front ends, then you
757 should use @code{TARGET_HANDLE_C_OPTION} instead.
760 @defmac TARGET_VERSION
761 This macro is a C statement to print on @code{stderr} a string
762 describing the particular machine description choice. Every machine
763 description should define @code{TARGET_VERSION}. For example:
767 #define TARGET_VERSION \
768 fprintf (stderr, " (68k, Motorola syntax)");
770 #define TARGET_VERSION \
771 fprintf (stderr, " (68k, MIT syntax)");
776 @defmac OVERRIDE_OPTIONS
777 Sometimes certain combinations of command options do not make sense on
778 a particular target machine. You can define a macro
779 @code{OVERRIDE_OPTIONS} to take account of this. This macro, if
780 defined, is executed once just after all the command options have been
783 Don't use this macro to turn on various extra optimizations for
784 @option{-O}. That is what @code{OPTIMIZATION_OPTIONS} is for.
786 If you need to do something whenever the optimization level is
787 changed via the optimize attribute or pragma, see
788 @code{TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE}
791 @deftypefn {Target Hook} void TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE (void)
792 This target function is similar to the macro @code{OVERRIDE_OPTIONS}
793 but is called when the optimize level is changed via an attribute or
794 pragma or when it is reset at the end of the code affected by the
795 attribute or pragma. It is not called at the beginning of compilation
796 when @code{OVERRIDE_OPTIONS} is called so if you want to perform these
797 actions then, you should have @code{OVERRIDE_OPTIONS} call
798 @code{TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE}.
801 @defmac C_COMMON_OVERRIDE_OPTIONS
802 This is similar to @code{OVERRIDE_OPTIONS} but is only used in the C
803 language frontends (C, Objective-C, C++, Objective-C++) and so can be
804 used to alter option flag variables which only exist in those
808 @defmac OPTIMIZATION_OPTIONS (@var{level}, @var{size})
809 Some machines may desire to change what optimizations are performed for
810 various optimization levels. This macro, if defined, is executed once
811 just after the optimization level is determined and before the remainder
812 of the command options have been parsed. Values set in this macro are
813 used as the default values for the other command line options.
815 @var{level} is the optimization level specified; 2 if @option{-O2} is
816 specified, 1 if @option{-O} is specified, and 0 if neither is specified.
818 @var{size} is nonzero if @option{-Os} is specified and zero otherwise.
820 This macro is run once at program startup and when the optimization
821 options are changed via @code{#pragma GCC optimize} or by using the
822 @code{optimize} attribute.
824 @strong{Do not examine @code{write_symbols} in
825 this macro!} The debugging options are not supposed to alter the
829 @deftypefn {Target Hook} void TARGET_HELP (void)
830 This hook is called in response to the user invoking
831 @option{--target-help} on the command line. It gives the target a
832 chance to display extra information on the target specific command
833 line options found in its @file{.opt} file.
836 @defmac CAN_DEBUG_WITHOUT_FP
837 Define this macro if debugging can be performed even without a frame
838 pointer. If this macro is defined, GCC will turn on the
839 @option{-fomit-frame-pointer} option whenever @option{-O} is specified.
842 @node Per-Function Data
843 @section Defining data structures for per-function information.
844 @cindex per-function data
845 @cindex data structures
847 If the target needs to store information on a per-function basis, GCC
848 provides a macro and a couple of variables to allow this. Note, just
849 using statics to store the information is a bad idea, since GCC supports
850 nested functions, so you can be halfway through encoding one function
851 when another one comes along.
853 GCC defines a data structure called @code{struct function} which
854 contains all of the data specific to an individual function. This
855 structure contains a field called @code{machine} whose type is
856 @code{struct machine_function *}, which can be used by targets to point
857 to their own specific data.
859 If a target needs per-function specific data it should define the type
860 @code{struct machine_function} and also the macro @code{INIT_EXPANDERS}.
861 This macro should be used to initialize the function pointer
862 @code{init_machine_status}. This pointer is explained below.
864 One typical use of per-function, target specific data is to create an
865 RTX to hold the register containing the function's return address. This
866 RTX can then be used to implement the @code{__builtin_return_address}
867 function, for level 0.
869 Note---earlier implementations of GCC used a single data area to hold
870 all of the per-function information. Thus when processing of a nested
871 function began the old per-function data had to be pushed onto a
872 stack, and when the processing was finished, it had to be popped off the
873 stack. GCC used to provide function pointers called
874 @code{save_machine_status} and @code{restore_machine_status} to handle
875 the saving and restoring of the target specific information. Since the
876 single data area approach is no longer used, these pointers are no
879 @defmac INIT_EXPANDERS
880 Macro called to initialize any target specific information. This macro
881 is called once per function, before generation of any RTL has begun.
882 The intention of this macro is to allow the initialization of the
883 function pointer @code{init_machine_status}.
886 @deftypevar {void (*)(struct function *)} init_machine_status
887 If this function pointer is non-@code{NULL} it will be called once per
888 function, before function compilation starts, in order to allow the
889 target to perform any target specific initialization of the
890 @code{struct function} structure. It is intended that this would be
891 used to initialize the @code{machine} of that structure.
893 @code{struct machine_function} structures are expected to be freed by GC@.
894 Generally, any memory that they reference must be allocated by using
895 GC allocation, including the structure itself.
899 @section Storage Layout
900 @cindex storage layout
902 Note that the definitions of the macros in this table which are sizes or
903 alignments measured in bits do not need to be constant. They can be C
904 expressions that refer to static variables, such as the @code{target_flags}.
905 @xref{Run-time Target}.
907 @defmac BITS_BIG_ENDIAN
908 Define this macro to have the value 1 if the most significant bit in a
909 byte has the lowest number; otherwise define it to have the value zero.
910 This means that bit-field instructions count from the most significant
911 bit. If the machine has no bit-field instructions, then this must still
912 be defined, but it doesn't matter which value it is defined to. This
913 macro need not be a constant.
915 This macro does not affect the way structure fields are packed into
916 bytes or words; that is controlled by @code{BYTES_BIG_ENDIAN}.
919 @defmac BYTES_BIG_ENDIAN
920 Define this macro to have the value 1 if the most significant byte in a
921 word has the lowest number. This macro need not be a constant.
924 @defmac WORDS_BIG_ENDIAN
925 Define this macro to have the value 1 if, in a multiword object, the
926 most significant word has the lowest number. This applies to both
927 memory locations and registers; GCC fundamentally assumes that the
928 order of words in memory is the same as the order in registers. This
929 macro need not be a constant.
932 @defmac LIBGCC2_WORDS_BIG_ENDIAN
933 Define this macro if @code{WORDS_BIG_ENDIAN} is not constant. This must be a
934 constant value with the same meaning as @code{WORDS_BIG_ENDIAN}, which will be
935 used only when compiling @file{libgcc2.c}. Typically the value will be set
936 based on preprocessor defines.
939 @defmac FLOAT_WORDS_BIG_ENDIAN
940 Define this macro to have the value 1 if @code{DFmode}, @code{XFmode} or
941 @code{TFmode} floating point numbers are stored in memory with the word
942 containing the sign bit at the lowest address; otherwise define it to
943 have the value 0. This macro need not be a constant.
945 You need not define this macro if the ordering is the same as for
949 @defmac BITS_PER_UNIT
950 Define this macro to be the number of bits in an addressable storage
951 unit (byte). If you do not define this macro the default is 8.
954 @defmac BITS_PER_WORD
955 Number of bits in a word. If you do not define this macro, the default
956 is @code{BITS_PER_UNIT * UNITS_PER_WORD}.
959 @defmac MAX_BITS_PER_WORD
960 Maximum number of bits in a word. If this is undefined, the default is
961 @code{BITS_PER_WORD}. Otherwise, it is the constant value that is the
962 largest value that @code{BITS_PER_WORD} can have at run-time.
965 @defmac UNITS_PER_WORD
966 Number of storage units in a word; normally the size of a general-purpose
967 register, a power of two from 1 or 8.
970 @defmac MIN_UNITS_PER_WORD
971 Minimum number of units in a word. If this is undefined, the default is
972 @code{UNITS_PER_WORD}. Otherwise, it is the constant value that is the
973 smallest value that @code{UNITS_PER_WORD} can have at run-time.
976 @defmac UNITS_PER_SIMD_WORD (@var{mode})
977 Number of units in the vectors that the vectorizer can produce for
978 scalar mode @var{mode}. The default is equal to @code{UNITS_PER_WORD},
979 because the vectorizer can do some transformations even in absence of
980 specialized @acronym{SIMD} hardware.
984 Width of a pointer, in bits. You must specify a value no wider than the
985 width of @code{Pmode}. If it is not equal to the width of @code{Pmode},
986 you must define @code{POINTERS_EXTEND_UNSIGNED}. If you do not specify
987 a value the default is @code{BITS_PER_WORD}.
990 @defmac POINTERS_EXTEND_UNSIGNED
991 A C expression that determines how pointers should be extended from
992 @code{ptr_mode} to either @code{Pmode} or @code{word_mode}. It is
993 greater than zero if pointers should be zero-extended, zero if they
994 should be sign-extended, and negative if some other sort of conversion
995 is needed. In the last case, the extension is done by the target's
996 @code{ptr_extend} instruction.
998 You need not define this macro if the @code{ptr_mode}, @code{Pmode}
999 and @code{word_mode} are all the same width.
1002 @defmac PROMOTE_MODE (@var{m}, @var{unsignedp}, @var{type})
1003 A macro to update @var{m} and @var{unsignedp} when an object whose type
1004 is @var{type} and which has the specified mode and signedness is to be
1005 stored in a register. This macro is only called when @var{type} is a
1008 On most RISC machines, which only have operations that operate on a full
1009 register, define this macro to set @var{m} to @code{word_mode} if
1010 @var{m} is an integer mode narrower than @code{BITS_PER_WORD}. In most
1011 cases, only integer modes should be widened because wider-precision
1012 floating-point operations are usually more expensive than their narrower
1015 For most machines, the macro definition does not change @var{unsignedp}.
1016 However, some machines, have instructions that preferentially handle
1017 either signed or unsigned quantities of certain modes. For example, on
1018 the DEC Alpha, 32-bit loads from memory and 32-bit add instructions
1019 sign-extend the result to 64 bits. On such machines, set
1020 @var{unsignedp} according to which kind of extension is more efficient.
1022 Do not define this macro if it would never modify @var{m}.
1025 @deftypefn {Target Hook} {enum machine_mode} TARGET_PROMOTE_FUNCTION_MODE (const_tree @var{type}, enum machine_mode @var{mode}, int *@var{punsignedp}, const_tree @var{funtype}, int @var{for_return})
1026 Like @code{PROMOTE_MODE}, but it is applied to outgoing function arguments or
1027 function return values. The target hook should return the new mode
1028 and possibly change @code{*@var{punsignedp}} if the promotion should
1029 change signedness. This function is called only for scalar @emph{or
1032 @var{for_return} allows to distinguish the promotion of arguments and
1033 return values. If it is @code{1}, a return value is being promoted and
1034 @code{TARGET_FUNCTION_VALUE} must perform the same promotions done here.
1035 If it is @code{2}, the returned mode should be that of the register in
1036 which an incoming parameter is copied, or the outgoing result is computed;
1037 then the hook should return the same mode as @code{promote_mode}, though
1038 the signedness may be different.
1040 The default is to not promote arguments and return values. You can
1041 also define the hook to @code{default_promote_function_mode_always_promote}
1042 if you would like to apply the same rules given by @code{PROMOTE_MODE}.
1045 @defmac PARM_BOUNDARY
1046 Normal alignment required for function parameters on the stack, in
1047 bits. All stack parameters receive at least this much alignment
1048 regardless of data type. On most machines, this is the same as the
1052 @defmac STACK_BOUNDARY
1053 Define this macro to the minimum alignment enforced by hardware for the
1054 stack pointer on this machine. The definition is a C expression for the
1055 desired alignment (measured in bits). This value is used as a default
1056 if @code{PREFERRED_STACK_BOUNDARY} is not defined. On most machines,
1057 this should be the same as @code{PARM_BOUNDARY}.
1060 @defmac PREFERRED_STACK_BOUNDARY
1061 Define this macro if you wish to preserve a certain alignment for the
1062 stack pointer, greater than what the hardware enforces. The definition
1063 is a C expression for the desired alignment (measured in bits). This
1064 macro must evaluate to a value equal to or larger than
1065 @code{STACK_BOUNDARY}.
1068 @defmac INCOMING_STACK_BOUNDARY
1069 Define this macro if the incoming stack boundary may be different
1070 from @code{PREFERRED_STACK_BOUNDARY}. This macro must evaluate
1071 to a value equal to or larger than @code{STACK_BOUNDARY}.
1074 @defmac FUNCTION_BOUNDARY
1075 Alignment required for a function entry point, in bits.
1078 @defmac BIGGEST_ALIGNMENT
1079 Biggest alignment that any data type can require on this machine, in
1080 bits. Note that this is not the biggest alignment that is supported,
1081 just the biggest alignment that, when violated, may cause a fault.
1084 @defmac MALLOC_ABI_ALIGNMENT
1085 Alignment, in bits, a C conformant malloc implementation has to
1086 provide. If not defined, the default value is @code{BITS_PER_WORD}.
1089 @defmac ATTRIBUTE_ALIGNED_VALUE
1090 Alignment used by the @code{__attribute__ ((aligned))} construct. If
1091 not defined, the default value is @code{BIGGEST_ALIGNMENT}.
1094 @defmac MINIMUM_ATOMIC_ALIGNMENT
1095 If defined, the smallest alignment, in bits, that can be given to an
1096 object that can be referenced in one operation, without disturbing any
1097 nearby object. Normally, this is @code{BITS_PER_UNIT}, but may be larger
1098 on machines that don't have byte or half-word store operations.
1101 @defmac BIGGEST_FIELD_ALIGNMENT
1102 Biggest alignment that any structure or union field can require on this
1103 machine, in bits. If defined, this overrides @code{BIGGEST_ALIGNMENT} for
1104 structure and union fields only, unless the field alignment has been set
1105 by the @code{__attribute__ ((aligned (@var{n})))} construct.
1108 @defmac ADJUST_FIELD_ALIGN (@var{field}, @var{computed})
1109 An expression for the alignment of a structure field @var{field} if the
1110 alignment computed in the usual way (including applying of
1111 @code{BIGGEST_ALIGNMENT} and @code{BIGGEST_FIELD_ALIGNMENT} to the
1112 alignment) is @var{computed}. It overrides alignment only if the
1113 field alignment has not been set by the
1114 @code{__attribute__ ((aligned (@var{n})))} construct.
1117 @defmac MAX_STACK_ALIGNMENT
1118 Biggest stack alignment guaranteed by the backend. Use this macro
1119 to specify the maximum alignment of a variable on stack.
1121 If not defined, the default value is @code{STACK_BOUNDARY}.
1123 @c FIXME: The default should be @code{PREFERRED_STACK_BOUNDARY}.
1124 @c But the fix for PR 32893 indicates that we can only guarantee
1125 @c maximum stack alignment on stack up to @code{STACK_BOUNDARY}, not
1126 @c @code{PREFERRED_STACK_BOUNDARY}, if stack alignment isn't supported.
1129 @defmac MAX_OFILE_ALIGNMENT
1130 Biggest alignment supported by the object file format of this machine.
1131 Use this macro to limit the alignment which can be specified using the
1132 @code{__attribute__ ((aligned (@var{n})))} construct. If not defined,
1133 the default value is @code{BIGGEST_ALIGNMENT}.
1135 On systems that use ELF, the default (in @file{config/elfos.h}) is
1136 the largest supported 32-bit ELF section alignment representable on
1137 a 32-bit host e.g. @samp{(((unsigned HOST_WIDEST_INT) 1 << 28) * 8)}.
1138 On 32-bit ELF the largest supported section alignment in bits is
1139 @samp{(0x80000000 * 8)}, but this is not representable on 32-bit hosts.
1142 @defmac DATA_ALIGNMENT (@var{type}, @var{basic-align})
1143 If defined, a C expression to compute the alignment for a variable in
1144 the static store. @var{type} is the data type, and @var{basic-align} is
1145 the alignment that the object would ordinarily have. The value of this
1146 macro is used instead of that alignment to align the object.
1148 If this macro is not defined, then @var{basic-align} is used.
1151 One use of this macro is to increase alignment of medium-size data to
1152 make it all fit in fewer cache lines. Another is to cause character
1153 arrays to be word-aligned so that @code{strcpy} calls that copy
1154 constants to character arrays can be done inline.
1157 @defmac CONSTANT_ALIGNMENT (@var{constant}, @var{basic-align})
1158 If defined, a C expression to compute the alignment given to a constant
1159 that is being placed in memory. @var{constant} is the constant and
1160 @var{basic-align} is the alignment that the object would ordinarily
1161 have. The value of this macro is used instead of that alignment to
1164 If this macro is not defined, then @var{basic-align} is used.
1166 The typical use of this macro is to increase alignment for string
1167 constants to be word aligned so that @code{strcpy} calls that copy
1168 constants can be done inline.
1171 @defmac LOCAL_ALIGNMENT (@var{type}, @var{basic-align})
1172 If defined, a C expression to compute the alignment for a variable in
1173 the local store. @var{type} is the data type, and @var{basic-align} is
1174 the alignment that the object would ordinarily have. The value of this
1175 macro is used instead of that alignment to align the object.
1177 If this macro is not defined, then @var{basic-align} is used.
1179 One use of this macro is to increase alignment of medium-size data to
1180 make it all fit in fewer cache lines.
1183 @defmac STACK_SLOT_ALIGNMENT (@var{type}, @var{mode}, @var{basic-align})
1184 If defined, a C expression to compute the alignment for stack slot.
1185 @var{type} is the data type, @var{mode} is the widest mode available,
1186 and @var{basic-align} is the alignment that the slot would ordinarily
1187 have. The value of this macro is used instead of that alignment to
1190 If this macro is not defined, then @var{basic-align} is used when
1191 @var{type} is @code{NULL}. Otherwise, @code{LOCAL_ALIGNMENT} will
1194 This macro is to set alignment of stack slot to the maximum alignment
1195 of all possible modes which the slot may have.
1198 @defmac LOCAL_DECL_ALIGNMENT (@var{decl})
1199 If defined, a C expression to compute the alignment for a local
1200 variable @var{decl}.
1202 If this macro is not defined, then
1203 @code{LOCAL_ALIGNMENT (TREE_TYPE (@var{decl}), DECL_ALIGN (@var{decl}))}
1206 One use of this macro is to increase alignment of medium-size data to
1207 make it all fit in fewer cache lines.
1210 @defmac MINIMUM_ALIGNMENT (@var{exp}, @var{mode}, @var{align})
1211 If defined, a C expression to compute the minimum required alignment
1212 for dynamic stack realignment purposes for @var{exp} (a type or decl),
1213 @var{mode}, assuming normal alignment @var{align}.
1215 If this macro is not defined, then @var{align} will be used.
1218 @defmac EMPTY_FIELD_BOUNDARY
1219 Alignment in bits to be given to a structure bit-field that follows an
1220 empty field such as @code{int : 0;}.
1222 If @code{PCC_BITFIELD_TYPE_MATTERS} is true, it overrides this macro.
1225 @defmac STRUCTURE_SIZE_BOUNDARY
1226 Number of bits which any structure or union's size must be a multiple of.
1227 Each structure or union's size is rounded up to a multiple of this.
1229 If you do not define this macro, the default is the same as
1230 @code{BITS_PER_UNIT}.
1233 @defmac STRICT_ALIGNMENT
1234 Define this macro to be the value 1 if instructions will fail to work
1235 if given data not on the nominal alignment. If instructions will merely
1236 go slower in that case, define this macro as 0.
1239 @defmac PCC_BITFIELD_TYPE_MATTERS
1240 Define this if you wish to imitate the way many other C compilers handle
1241 alignment of bit-fields and the structures that contain them.
1243 The behavior is that the type written for a named bit-field (@code{int},
1244 @code{short}, or other integer type) imposes an alignment for the entire
1245 structure, as if the structure really did contain an ordinary field of
1246 that type. In addition, the bit-field is placed within the structure so
1247 that it would fit within such a field, not crossing a boundary for it.
1249 Thus, on most machines, a named bit-field whose type is written as
1250 @code{int} would not cross a four-byte boundary, and would force
1251 four-byte alignment for the whole structure. (The alignment used may
1252 not be four bytes; it is controlled by the other alignment parameters.)
1254 An unnamed bit-field will not affect the alignment of the containing
1257 If the macro is defined, its definition should be a C expression;
1258 a nonzero value for the expression enables this behavior.
1260 Note that if this macro is not defined, or its value is zero, some
1261 bit-fields may cross more than one alignment boundary. The compiler can
1262 support such references if there are @samp{insv}, @samp{extv}, and
1263 @samp{extzv} insns that can directly reference memory.
1265 The other known way of making bit-fields work is to define
1266 @code{STRUCTURE_SIZE_BOUNDARY} as large as @code{BIGGEST_ALIGNMENT}.
1267 Then every structure can be accessed with fullwords.
1269 Unless the machine has bit-field instructions or you define
1270 @code{STRUCTURE_SIZE_BOUNDARY} that way, you must define
1271 @code{PCC_BITFIELD_TYPE_MATTERS} to have a nonzero value.
1273 If your aim is to make GCC use the same conventions for laying out
1274 bit-fields as are used by another compiler, here is how to investigate
1275 what the other compiler does. Compile and run this program:
1294 printf ("Size of foo1 is %d\n",
1295 sizeof (struct foo1));
1296 printf ("Size of foo2 is %d\n",
1297 sizeof (struct foo2));
1302 If this prints 2 and 5, then the compiler's behavior is what you would
1303 get from @code{PCC_BITFIELD_TYPE_MATTERS}.
1306 @defmac BITFIELD_NBYTES_LIMITED
1307 Like @code{PCC_BITFIELD_TYPE_MATTERS} except that its effect is limited
1308 to aligning a bit-field within the structure.
1311 @deftypefn {Target Hook} bool TARGET_ALIGN_ANON_BITFIELD (void)
1312 When @code{PCC_BITFIELD_TYPE_MATTERS} is true this hook will determine
1313 whether unnamed bitfields affect the alignment of the containing
1314 structure. The hook should return true if the structure should inherit
1315 the alignment requirements of an unnamed bitfield's type.
1318 @deftypefn {Target Hook} bool TARGET_NARROW_VOLATILE_BITFIELD (void)
1319 This target hook should return @code{true} if accesses to volatile bitfields
1320 should use the narrowest mode possible. It should return @code{false} if
1321 these accesses should use the bitfield container type.
1323 The default is @code{!TARGET_STRICT_ALIGN}.
1326 @defmac MEMBER_TYPE_FORCES_BLK (@var{field}, @var{mode})
1327 Return 1 if a structure or array containing @var{field} should be accessed using
1330 If @var{field} is the only field in the structure, @var{mode} is its
1331 mode, otherwise @var{mode} is VOIDmode. @var{mode} is provided in the
1332 case where structures of one field would require the structure's mode to
1333 retain the field's mode.
1335 Normally, this is not needed.
1338 @defmac ROUND_TYPE_ALIGN (@var{type}, @var{computed}, @var{specified})
1339 Define this macro as an expression for the alignment of a type (given
1340 by @var{type} as a tree node) if the alignment computed in the usual
1341 way is @var{computed} and the alignment explicitly specified was
1344 The default is to use @var{specified} if it is larger; otherwise, use
1345 the smaller of @var{computed} and @code{BIGGEST_ALIGNMENT}
1348 @defmac MAX_FIXED_MODE_SIZE
1349 An integer expression for the size in bits of the largest integer
1350 machine mode that should actually be used. All integer machine modes of
1351 this size or smaller can be used for structures and unions with the
1352 appropriate sizes. If this macro is undefined, @code{GET_MODE_BITSIZE
1353 (DImode)} is assumed.
1356 @defmac STACK_SAVEAREA_MODE (@var{save_level})
1357 If defined, an expression of type @code{enum machine_mode} that
1358 specifies the mode of the save area operand of a
1359 @code{save_stack_@var{level}} named pattern (@pxref{Standard Names}).
1360 @var{save_level} is one of @code{SAVE_BLOCK}, @code{SAVE_FUNCTION}, or
1361 @code{SAVE_NONLOCAL} and selects which of the three named patterns is
1362 having its mode specified.
1364 You need not define this macro if it always returns @code{Pmode}. You
1365 would most commonly define this macro if the
1366 @code{save_stack_@var{level}} patterns need to support both a 32- and a
1370 @defmac STACK_SIZE_MODE
1371 If defined, an expression of type @code{enum machine_mode} that
1372 specifies the mode of the size increment operand of an
1373 @code{allocate_stack} named pattern (@pxref{Standard Names}).
1375 You need not define this macro if it always returns @code{word_mode}.
1376 You would most commonly define this macro if the @code{allocate_stack}
1377 pattern needs to support both a 32- and a 64-bit mode.
1380 @deftypefn {Target Hook} {enum machine_mode} TARGET_LIBGCC_CMP_RETURN_MODE (void)
1381 This target hook should return the mode to be used for the return value
1382 of compare instructions expanded to libgcc calls. If not defined
1383 @code{word_mode} is returned which is the right choice for a majority of
1387 @deftypefn {Target Hook} {enum machine_mode} TARGET_LIBGCC_SHIFT_COUNT_MODE (void)
1388 This target hook should return the mode to be used for the shift count operand
1389 of shift instructions expanded to libgcc calls. If not defined
1390 @code{word_mode} is returned which is the right choice for a majority of
1394 @deftypefn {Target Hook} {enum machine_mode} TARGET_UNWIND_WORD_MODE (void)
1395 Return machine mode to be used for @code{_Unwind_Word} type.
1396 The default is to use @code{word_mode}.
1399 @defmac ROUND_TOWARDS_ZERO
1400 If defined, this macro should be true if the prevailing rounding
1401 mode is towards zero.
1403 Defining this macro only affects the way @file{libgcc.a} emulates
1404 floating-point arithmetic.
1406 Not defining this macro is equivalent to returning zero.
1409 @defmac LARGEST_EXPONENT_IS_NORMAL (@var{size})
1410 This macro should return true if floats with @var{size}
1411 bits do not have a NaN or infinity representation, but use the largest
1412 exponent for normal numbers instead.
1414 Defining this macro only affects the way @file{libgcc.a} emulates
1415 floating-point arithmetic.
1417 The default definition of this macro returns false for all sizes.
1420 @deftypefn {Target Hook} bool TARGET_MS_BITFIELD_LAYOUT_P (const_tree @var{record_type})
1421 This target hook returns @code{true} if bit-fields in the given
1422 @var{record_type} are to be laid out following the rules of Microsoft
1423 Visual C/C++, namely: (i) a bit-field won't share the same storage
1424 unit with the previous bit-field if their underlying types have
1425 different sizes, and the bit-field will be aligned to the highest
1426 alignment of the underlying types of itself and of the previous
1427 bit-field; (ii) a zero-sized bit-field will affect the alignment of
1428 the whole enclosing structure, even if it is unnamed; except that
1429 (iii) a zero-sized bit-field will be disregarded unless it follows
1430 another bit-field of nonzero size. If this hook returns @code{true},
1431 other macros that control bit-field layout are ignored.
1433 When a bit-field is inserted into a packed record, the whole size
1434 of the underlying type is used by one or more same-size adjacent
1435 bit-fields (that is, if its long:3, 32 bits is used in the record,
1436 and any additional adjacent long bit-fields are packed into the same
1437 chunk of 32 bits. However, if the size changes, a new field of that
1438 size is allocated). In an unpacked record, this is the same as using
1439 alignment, but not equivalent when packing.
1441 If both MS bit-fields and @samp{__attribute__((packed))} are used,
1442 the latter will take precedence. If @samp{__attribute__((packed))} is
1443 used on a single field when MS bit-fields are in use, it will take
1444 precedence for that field, but the alignment of the rest of the structure
1445 may affect its placement.
1448 @deftypefn {Target Hook} bool TARGET_DECIMAL_FLOAT_SUPPORTED_P (void)
1449 Returns true if the target supports decimal floating point.
1452 @deftypefn {Target Hook} bool TARGET_FIXED_POINT_SUPPORTED_P (void)
1453 Returns true if the target supports fixed-point arithmetic.
1456 @deftypefn {Target Hook} void TARGET_EXPAND_TO_RTL_HOOK (void)
1457 This hook is called just before expansion into rtl, allowing the target
1458 to perform additional initializations or analysis before the expansion.
1459 For example, the rs6000 port uses it to allocate a scratch stack slot
1460 for use in copying SDmode values between memory and floating point
1461 registers whenever the function being expanded has any SDmode
1465 @deftypefn {Target Hook} void TARGET_INSTANTIATE_DECLS (void)
1466 This hook allows the backend to perform additional instantiations on rtl
1467 that are not actually in any insns yet, but will be later.
1470 @deftypefn {Target Hook} {const char *} TARGET_MANGLE_TYPE (const_tree @var{type})
1471 If your target defines any fundamental types, or any types your target
1472 uses should be mangled differently from the default, define this hook
1473 to return the appropriate encoding for these types as part of a C++
1474 mangled name. The @var{type} argument is the tree structure representing
1475 the type to be mangled. The hook may be applied to trees which are
1476 not target-specific fundamental types; it should return @code{NULL}
1477 for all such types, as well as arguments it does not recognize. If the
1478 return value is not @code{NULL}, it must point to a statically-allocated
1481 Target-specific fundamental types might be new fundamental types or
1482 qualified versions of ordinary fundamental types. Encode new
1483 fundamental types as @samp{@w{u @var{n} @var{name}}}, where @var{name}
1484 is the name used for the type in source code, and @var{n} is the
1485 length of @var{name} in decimal. Encode qualified versions of
1486 ordinary types as @samp{@w{U @var{n} @var{name} @var{code}}}, where
1487 @var{name} is the name used for the type qualifier in source code,
1488 @var{n} is the length of @var{name} as above, and @var{code} is the
1489 code used to represent the unqualified version of this type. (See
1490 @code{write_builtin_type} in @file{cp/mangle.c} for the list of
1491 codes.) In both cases the spaces are for clarity; do not include any
1492 spaces in your string.
1494 This hook is applied to types prior to typedef resolution. If the mangled
1495 name for a particular type depends only on that type's main variant, you
1496 can perform typedef resolution yourself using @code{TYPE_MAIN_VARIANT}
1499 The default version of this hook always returns @code{NULL}, which is
1500 appropriate for a target that does not define any new fundamental
1505 @section Layout of Source Language Data Types
1507 These macros define the sizes and other characteristics of the standard
1508 basic data types used in programs being compiled. Unlike the macros in
1509 the previous section, these apply to specific features of C and related
1510 languages, rather than to fundamental aspects of storage layout.
1512 @defmac INT_TYPE_SIZE
1513 A C expression for the size in bits of the type @code{int} on the
1514 target machine. If you don't define this, the default is one word.
1517 @defmac SHORT_TYPE_SIZE
1518 A C expression for the size in bits of the type @code{short} on the
1519 target machine. If you don't define this, the default is half a word.
1520 (If this would be less than one storage unit, it is rounded up to one
1524 @defmac LONG_TYPE_SIZE
1525 A C expression for the size in bits of the type @code{long} on the
1526 target machine. If you don't define this, the default is one word.
1529 @defmac ADA_LONG_TYPE_SIZE
1530 On some machines, the size used for the Ada equivalent of the type
1531 @code{long} by a native Ada compiler differs from that used by C@. In
1532 that situation, define this macro to be a C expression to be used for
1533 the size of that type. If you don't define this, the default is the
1534 value of @code{LONG_TYPE_SIZE}.
1537 @defmac LONG_LONG_TYPE_SIZE
1538 A C expression for the size in bits of the type @code{long long} on the
1539 target machine. If you don't define this, the default is two
1540 words. If you want to support GNU Ada on your machine, the value of this
1541 macro must be at least 64.
1544 @defmac CHAR_TYPE_SIZE
1545 A C expression for the size in bits of the type @code{char} on the
1546 target machine. If you don't define this, the default is
1547 @code{BITS_PER_UNIT}.
1550 @defmac BOOL_TYPE_SIZE
1551 A C expression for the size in bits of the C++ type @code{bool} and
1552 C99 type @code{_Bool} on the target machine. If you don't define
1553 this, and you probably shouldn't, the default is @code{CHAR_TYPE_SIZE}.
1556 @defmac FLOAT_TYPE_SIZE
1557 A C expression for the size in bits of the type @code{float} on the
1558 target machine. If you don't define this, the default is one word.
1561 @defmac DOUBLE_TYPE_SIZE
1562 A C expression for the size in bits of the type @code{double} on the
1563 target machine. If you don't define this, the default is two
1567 @defmac LONG_DOUBLE_TYPE_SIZE
1568 A C expression for the size in bits of the type @code{long double} on
1569 the target machine. If you don't define this, the default is two
1573 @defmac SHORT_FRACT_TYPE_SIZE
1574 A C expression for the size in bits of the type @code{short _Fract} on
1575 the target machine. If you don't define this, the default is
1576 @code{BITS_PER_UNIT}.
1579 @defmac FRACT_TYPE_SIZE
1580 A C expression for the size in bits of the type @code{_Fract} on
1581 the target machine. If you don't define this, the default is
1582 @code{BITS_PER_UNIT * 2}.
1585 @defmac LONG_FRACT_TYPE_SIZE
1586 A C expression for the size in bits of the type @code{long _Fract} on
1587 the target machine. If you don't define this, the default is
1588 @code{BITS_PER_UNIT * 4}.
1591 @defmac LONG_LONG_FRACT_TYPE_SIZE
1592 A C expression for the size in bits of the type @code{long long _Fract} on
1593 the target machine. If you don't define this, the default is
1594 @code{BITS_PER_UNIT * 8}.
1597 @defmac SHORT_ACCUM_TYPE_SIZE
1598 A C expression for the size in bits of the type @code{short _Accum} on
1599 the target machine. If you don't define this, the default is
1600 @code{BITS_PER_UNIT * 2}.
1603 @defmac ACCUM_TYPE_SIZE
1604 A C expression for the size in bits of the type @code{_Accum} on
1605 the target machine. If you don't define this, the default is
1606 @code{BITS_PER_UNIT * 4}.
1609 @defmac LONG_ACCUM_TYPE_SIZE
1610 A C expression for the size in bits of the type @code{long _Accum} on
1611 the target machine. If you don't define this, the default is
1612 @code{BITS_PER_UNIT * 8}.
1615 @defmac LONG_LONG_ACCUM_TYPE_SIZE
1616 A C expression for the size in bits of the type @code{long long _Accum} on
1617 the target machine. If you don't define this, the default is
1618 @code{BITS_PER_UNIT * 16}.
1621 @defmac LIBGCC2_LONG_DOUBLE_TYPE_SIZE
1622 Define this macro if @code{LONG_DOUBLE_TYPE_SIZE} is not constant or
1623 if you want routines in @file{libgcc2.a} for a size other than
1624 @code{LONG_DOUBLE_TYPE_SIZE}. If you don't define this, the
1625 default is @code{LONG_DOUBLE_TYPE_SIZE}.
1628 @defmac LIBGCC2_HAS_DF_MODE
1629 Define this macro if neither @code{LIBGCC2_DOUBLE_TYPE_SIZE} nor
1630 @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is
1631 @code{DFmode} but you want @code{DFmode} routines in @file{libgcc2.a}
1632 anyway. If you don't define this and either @code{LIBGCC2_DOUBLE_TYPE_SIZE}
1633 or @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is 64 then the default is 1,
1637 @defmac LIBGCC2_HAS_XF_MODE
1638 Define this macro if @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is not
1639 @code{XFmode} but you want @code{XFmode} routines in @file{libgcc2.a}
1640 anyway. If you don't define this and @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE}
1641 is 80 then the default is 1, otherwise it is 0.
1644 @defmac LIBGCC2_HAS_TF_MODE
1645 Define this macro if @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is not
1646 @code{TFmode} but you want @code{TFmode} routines in @file{libgcc2.a}
1647 anyway. If you don't define this and @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE}
1648 is 128 then the default is 1, otherwise it is 0.
1655 Define these macros to be the size in bits of the mantissa of
1656 @code{SFmode}, @code{DFmode}, @code{XFmode} and @code{TFmode} values,
1657 if the defaults in @file{libgcc2.h} are inappropriate. By default,
1658 @code{FLT_MANT_DIG} is used for @code{SF_SIZE}, @code{LDBL_MANT_DIG}
1659 for @code{XF_SIZE} and @code{TF_SIZE}, and @code{DBL_MANT_DIG} or
1660 @code{LDBL_MANT_DIG} for @code{DF_SIZE} according to whether
1661 @code{LIBGCC2_DOUBLE_TYPE_SIZE} or
1662 @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is 64.
1665 @defmac TARGET_FLT_EVAL_METHOD
1666 A C expression for the value for @code{FLT_EVAL_METHOD} in @file{float.h},
1667 assuming, if applicable, that the floating-point control word is in its
1668 default state. If you do not define this macro the value of
1669 @code{FLT_EVAL_METHOD} will be zero.
1672 @defmac WIDEST_HARDWARE_FP_SIZE
1673 A C expression for the size in bits of the widest floating-point format
1674 supported by the hardware. If you define this macro, you must specify a
1675 value less than or equal to the value of @code{LONG_DOUBLE_TYPE_SIZE}.
1676 If you do not define this macro, the value of @code{LONG_DOUBLE_TYPE_SIZE}
1680 @defmac DEFAULT_SIGNED_CHAR
1681 An expression whose value is 1 or 0, according to whether the type
1682 @code{char} should be signed or unsigned by default. The user can
1683 always override this default with the options @option{-fsigned-char}
1684 and @option{-funsigned-char}.
1687 @deftypefn {Target Hook} bool TARGET_DEFAULT_SHORT_ENUMS (void)
1688 This target hook should return true if the compiler should give an
1689 @code{enum} type only as many bytes as it takes to represent the range
1690 of possible values of that type. It should return false if all
1691 @code{enum} types should be allocated like @code{int}.
1693 The default is to return false.
1697 A C expression for a string describing the name of the data type to use
1698 for size values. The typedef name @code{size_t} is defined using the
1699 contents of the string.
1701 The string can contain more than one keyword. If so, separate them with
1702 spaces, and write first any length keyword, then @code{unsigned} if
1703 appropriate, and finally @code{int}. The string must exactly match one
1704 of the data type names defined in the function
1705 @code{init_decl_processing} in the file @file{c-decl.c}. You may not
1706 omit @code{int} or change the order---that would cause the compiler to
1709 If you don't define this macro, the default is @code{"long unsigned
1713 @defmac PTRDIFF_TYPE
1714 A C expression for a string describing the name of the data type to use
1715 for the result of subtracting two pointers. The typedef name
1716 @code{ptrdiff_t} is defined using the contents of the string. See
1717 @code{SIZE_TYPE} above for more information.
1719 If you don't define this macro, the default is @code{"long int"}.
1723 A C expression for a string describing the name of the data type to use
1724 for wide characters. The typedef name @code{wchar_t} is defined using
1725 the contents of the string. See @code{SIZE_TYPE} above for more
1728 If you don't define this macro, the default is @code{"int"}.
1731 @defmac WCHAR_TYPE_SIZE
1732 A C expression for the size in bits of the data type for wide
1733 characters. This is used in @code{cpp}, which cannot make use of
1738 A C expression for a string describing the name of the data type to
1739 use for wide characters passed to @code{printf} and returned from
1740 @code{getwc}. The typedef name @code{wint_t} is defined using the
1741 contents of the string. See @code{SIZE_TYPE} above for more
1744 If you don't define this macro, the default is @code{"unsigned int"}.
1748 A C expression for a string describing the name of the data type that
1749 can represent any value of any standard or extended signed integer type.
1750 The typedef name @code{intmax_t} is defined using the contents of the
1751 string. See @code{SIZE_TYPE} above for more information.
1753 If you don't define this macro, the default is the first of
1754 @code{"int"}, @code{"long int"}, or @code{"long long int"} that has as
1755 much precision as @code{long long int}.
1758 @defmac UINTMAX_TYPE
1759 A C expression for a string describing the name of the data type that
1760 can represent any value of any standard or extended unsigned integer
1761 type. The typedef name @code{uintmax_t} is defined using the contents
1762 of the string. See @code{SIZE_TYPE} above for more information.
1764 If you don't define this macro, the default is the first of
1765 @code{"unsigned int"}, @code{"long unsigned int"}, or @code{"long long
1766 unsigned int"} that has as much precision as @code{long long unsigned
1770 @defmac SIG_ATOMIC_TYPE
1776 @defmacx UINT16_TYPE
1777 @defmacx UINT32_TYPE
1778 @defmacx UINT64_TYPE
1779 @defmacx INT_LEAST8_TYPE
1780 @defmacx INT_LEAST16_TYPE
1781 @defmacx INT_LEAST32_TYPE
1782 @defmacx INT_LEAST64_TYPE
1783 @defmacx UINT_LEAST8_TYPE
1784 @defmacx UINT_LEAST16_TYPE
1785 @defmacx UINT_LEAST32_TYPE
1786 @defmacx UINT_LEAST64_TYPE
1787 @defmacx INT_FAST8_TYPE
1788 @defmacx INT_FAST16_TYPE
1789 @defmacx INT_FAST32_TYPE
1790 @defmacx INT_FAST64_TYPE
1791 @defmacx UINT_FAST8_TYPE
1792 @defmacx UINT_FAST16_TYPE
1793 @defmacx UINT_FAST32_TYPE
1794 @defmacx UINT_FAST64_TYPE
1795 @defmacx INTPTR_TYPE
1796 @defmacx UINTPTR_TYPE
1797 C expressions for the standard types @code{sig_atomic_t},
1798 @code{int8_t}, @code{int16_t}, @code{int32_t}, @code{int64_t},
1799 @code{uint8_t}, @code{uint16_t}, @code{uint32_t}, @code{uint64_t},
1800 @code{int_least8_t}, @code{int_least16_t}, @code{int_least32_t},
1801 @code{int_least64_t}, @code{uint_least8_t}, @code{uint_least16_t},
1802 @code{uint_least32_t}, @code{uint_least64_t}, @code{int_fast8_t},
1803 @code{int_fast16_t}, @code{int_fast32_t}, @code{int_fast64_t},
1804 @code{uint_fast8_t}, @code{uint_fast16_t}, @code{uint_fast32_t},
1805 @code{uint_fast64_t}, @code{intptr_t}, and @code{uintptr_t}. See
1806 @code{SIZE_TYPE} above for more information.
1808 If any of these macros evaluates to a null pointer, the corresponding
1809 type is not supported; if GCC is configured to provide
1810 @code{<stdint.h>} in such a case, the header provided may not conform
1811 to C99, depending on the type in question. The defaults for all of
1812 these macros are null pointers.
1815 @defmac TARGET_PTRMEMFUNC_VBIT_LOCATION
1816 The C++ compiler represents a pointer-to-member-function with a struct
1823 ptrdiff_t vtable_index;
1830 The C++ compiler must use one bit to indicate whether the function that
1831 will be called through a pointer-to-member-function is virtual.
1832 Normally, we assume that the low-order bit of a function pointer must
1833 always be zero. Then, by ensuring that the vtable_index is odd, we can
1834 distinguish which variant of the union is in use. But, on some
1835 platforms function pointers can be odd, and so this doesn't work. In
1836 that case, we use the low-order bit of the @code{delta} field, and shift
1837 the remainder of the @code{delta} field to the left.
1839 GCC will automatically make the right selection about where to store
1840 this bit using the @code{FUNCTION_BOUNDARY} setting for your platform.
1841 However, some platforms such as ARM/Thumb have @code{FUNCTION_BOUNDARY}
1842 set such that functions always start at even addresses, but the lowest
1843 bit of pointers to functions indicate whether the function at that
1844 address is in ARM or Thumb mode. If this is the case of your
1845 architecture, you should define this macro to
1846 @code{ptrmemfunc_vbit_in_delta}.
1848 In general, you should not have to define this macro. On architectures
1849 in which function addresses are always even, according to
1850 @code{FUNCTION_BOUNDARY}, GCC will automatically define this macro to
1851 @code{ptrmemfunc_vbit_in_pfn}.
1854 @defmac TARGET_VTABLE_USES_DESCRIPTORS
1855 Normally, the C++ compiler uses function pointers in vtables. This
1856 macro allows the target to change to use ``function descriptors''
1857 instead. Function descriptors are found on targets for whom a
1858 function pointer is actually a small data structure. Normally the
1859 data structure consists of the actual code address plus a data
1860 pointer to which the function's data is relative.
1862 If vtables are used, the value of this macro should be the number
1863 of words that the function descriptor occupies.
1866 @defmac TARGET_VTABLE_ENTRY_ALIGN
1867 By default, the vtable entries are void pointers, the so the alignment
1868 is the same as pointer alignment. The value of this macro specifies
1869 the alignment of the vtable entry in bits. It should be defined only
1870 when special alignment is necessary. */
1873 @defmac TARGET_VTABLE_DATA_ENTRY_DISTANCE
1874 There are a few non-descriptor entries in the vtable at offsets below
1875 zero. If these entries must be padded (say, to preserve the alignment
1876 specified by @code{TARGET_VTABLE_ENTRY_ALIGN}), set this to the number
1877 of words in each data entry.
1881 @section Register Usage
1882 @cindex register usage
1884 This section explains how to describe what registers the target machine
1885 has, and how (in general) they can be used.
1887 The description of which registers a specific instruction can use is
1888 done with register classes; see @ref{Register Classes}. For information
1889 on using registers to access a stack frame, see @ref{Frame Registers}.
1890 For passing values in registers, see @ref{Register Arguments}.
1891 For returning values in registers, see @ref{Scalar Return}.
1894 * Register Basics:: Number and kinds of registers.
1895 * Allocation Order:: Order in which registers are allocated.
1896 * Values in Registers:: What kinds of values each reg can hold.
1897 * Leaf Functions:: Renumbering registers for leaf functions.
1898 * Stack Registers:: Handling a register stack such as 80387.
1901 @node Register Basics
1902 @subsection Basic Characteristics of Registers
1904 @c prevent bad page break with this line
1905 Registers have various characteristics.
1907 @defmac FIRST_PSEUDO_REGISTER
1908 Number of hardware registers known to the compiler. They receive
1909 numbers 0 through @code{FIRST_PSEUDO_REGISTER-1}; thus, the first
1910 pseudo register's number really is assigned the number
1911 @code{FIRST_PSEUDO_REGISTER}.
1914 @defmac FIXED_REGISTERS
1915 @cindex fixed register
1916 An initializer that says which registers are used for fixed purposes
1917 all throughout the compiled code and are therefore not available for
1918 general allocation. These would include the stack pointer, the frame
1919 pointer (except on machines where that can be used as a general
1920 register when no frame pointer is needed), the program counter on
1921 machines where that is considered one of the addressable registers,
1922 and any other numbered register with a standard use.
1924 This information is expressed as a sequence of numbers, separated by
1925 commas and surrounded by braces. The @var{n}th number is 1 if
1926 register @var{n} is fixed, 0 otherwise.
1928 The table initialized from this macro, and the table initialized by
1929 the following one, may be overridden at run time either automatically,
1930 by the actions of the macro @code{CONDITIONAL_REGISTER_USAGE}, or by
1931 the user with the command options @option{-ffixed-@var{reg}},
1932 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}.
1935 @defmac CALL_USED_REGISTERS
1936 @cindex call-used register
1937 @cindex call-clobbered register
1938 @cindex call-saved register
1939 Like @code{FIXED_REGISTERS} but has 1 for each register that is
1940 clobbered (in general) by function calls as well as for fixed
1941 registers. This macro therefore identifies the registers that are not
1942 available for general allocation of values that must live across
1945 If a register has 0 in @code{CALL_USED_REGISTERS}, the compiler
1946 automatically saves it on function entry and restores it on function
1947 exit, if the register is used within the function.
1950 @defmac CALL_REALLY_USED_REGISTERS
1951 @cindex call-used register
1952 @cindex call-clobbered register
1953 @cindex call-saved register
1954 Like @code{CALL_USED_REGISTERS} except this macro doesn't require
1955 that the entire set of @code{FIXED_REGISTERS} be included.
1956 (@code{CALL_USED_REGISTERS} must be a superset of @code{FIXED_REGISTERS}).
1957 This macro is optional. If not specified, it defaults to the value
1958 of @code{CALL_USED_REGISTERS}.
1961 @defmac HARD_REGNO_CALL_PART_CLOBBERED (@var{regno}, @var{mode})
1962 @cindex call-used register
1963 @cindex call-clobbered register
1964 @cindex call-saved register
1965 A C expression that is nonzero if it is not permissible to store a
1966 value of mode @var{mode} in hard register number @var{regno} across a
1967 call without some part of it being clobbered. For most machines this
1968 macro need not be defined. It is only required for machines that do not
1969 preserve the entire contents of a register across a call.
1973 @findex call_used_regs
1976 @findex reg_class_contents
1977 @defmac CONDITIONAL_REGISTER_USAGE
1978 Zero or more C statements that may conditionally modify five variables
1979 @code{fixed_regs}, @code{call_used_regs}, @code{global_regs},
1980 @code{reg_names}, and @code{reg_class_contents}, to take into account
1981 any dependence of these register sets on target flags. The first three
1982 of these are of type @code{char []} (interpreted as Boolean vectors).
1983 @code{global_regs} is a @code{const char *[]}, and
1984 @code{reg_class_contents} is a @code{HARD_REG_SET}. Before the macro is
1985 called, @code{fixed_regs}, @code{call_used_regs},
1986 @code{reg_class_contents}, and @code{reg_names} have been initialized
1987 from @code{FIXED_REGISTERS}, @code{CALL_USED_REGISTERS},
1988 @code{REG_CLASS_CONTENTS}, and @code{REGISTER_NAMES}, respectively.
1989 @code{global_regs} has been cleared, and any @option{-ffixed-@var{reg}},
1990 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}
1991 command options have been applied.
1993 You need not define this macro if it has no work to do.
1995 @cindex disabling certain registers
1996 @cindex controlling register usage
1997 If the usage of an entire class of registers depends on the target
1998 flags, you may indicate this to GCC by using this macro to modify
1999 @code{fixed_regs} and @code{call_used_regs} to 1 for each of the
2000 registers in the classes which should not be used by GCC@. Also define
2001 the macro @code{REG_CLASS_FROM_LETTER} / @code{REG_CLASS_FROM_CONSTRAINT}
2002 to return @code{NO_REGS} if it
2003 is called with a letter for a class that shouldn't be used.
2005 (However, if this class is not included in @code{GENERAL_REGS} and all
2006 of the insn patterns whose constraints permit this class are
2007 controlled by target switches, then GCC will automatically avoid using
2008 these registers when the target switches are opposed to them.)
2011 @defmac INCOMING_REGNO (@var{out})
2012 Define this macro if the target machine has register windows. This C
2013 expression returns the register number as seen by the called function
2014 corresponding to the register number @var{out} as seen by the calling
2015 function. Return @var{out} if register number @var{out} is not an
2019 @defmac OUTGOING_REGNO (@var{in})
2020 Define this macro if the target machine has register windows. This C
2021 expression returns the register number as seen by the calling function
2022 corresponding to the register number @var{in} as seen by the called
2023 function. Return @var{in} if register number @var{in} is not an inbound
2027 @defmac LOCAL_REGNO (@var{regno})
2028 Define this macro if the target machine has register windows. This C
2029 expression returns true if the register is call-saved but is in the
2030 register window. Unlike most call-saved registers, such registers
2031 need not be explicitly restored on function exit or during non-local
2036 If the program counter has a register number, define this as that
2037 register number. Otherwise, do not define it.
2040 @node Allocation Order
2041 @subsection Order of Allocation of Registers
2042 @cindex order of register allocation
2043 @cindex register allocation order
2045 @c prevent bad page break with this line
2046 Registers are allocated in order.
2048 @defmac REG_ALLOC_ORDER
2049 If defined, an initializer for a vector of integers, containing the
2050 numbers of hard registers in the order in which GCC should prefer
2051 to use them (from most preferred to least).
2053 If this macro is not defined, registers are used lowest numbered first
2054 (all else being equal).
2056 One use of this macro is on machines where the highest numbered
2057 registers must always be saved and the save-multiple-registers
2058 instruction supports only sequences of consecutive registers. On such
2059 machines, define @code{REG_ALLOC_ORDER} to be an initializer that lists
2060 the highest numbered allocable register first.
2063 @defmac ADJUST_REG_ALLOC_ORDER
2064 A C statement (sans semicolon) to choose the order in which to allocate
2065 hard registers for pseudo-registers local to a basic block.
2067 Store the desired register order in the array @code{reg_alloc_order}.
2068 Element 0 should be the register to allocate first; element 1, the next
2069 register; and so on.
2071 The macro body should not assume anything about the contents of
2072 @code{reg_alloc_order} before execution of the macro.
2074 On most machines, it is not necessary to define this macro.
2077 @defmac HONOR_REG_ALLOC_ORDER
2078 Normally, IRA tries to estimate the costs for saving a register in the
2079 prologue and restoring it in the epilogue. This discourages it from
2080 using call-saved registers. If a machine wants to ensure that IRA
2081 allocates registers in the order given by REG_ALLOC_ORDER even if some
2082 call-saved registers appear earlier than call-used ones, this macro
2086 @defmac IRA_HARD_REGNO_ADD_COST_MULTIPLIER (@var{regno})
2087 In some case register allocation order is not enough for the
2088 Integrated Register Allocator (@acronym{IRA}) to generate a good code.
2089 If this macro is defined, it should return a floating point value
2090 based on @var{regno}. The cost of using @var{regno} for a pseudo will
2091 be increased by approximately the pseudo's usage frequency times the
2092 value returned by this macro. Not defining this macro is equivalent
2093 to having it always return @code{0.0}.
2095 On most machines, it is not necessary to define this macro.
2098 @node Values in Registers
2099 @subsection How Values Fit in Registers
2101 This section discusses the macros that describe which kinds of values
2102 (specifically, which machine modes) each register can hold, and how many
2103 consecutive registers are needed for a given mode.
2105 @defmac HARD_REGNO_NREGS (@var{regno}, @var{mode})
2106 A C expression for the number of consecutive hard registers, starting
2107 at register number @var{regno}, required to hold a value of mode
2108 @var{mode}. This macro must never return zero, even if a register
2109 cannot hold the requested mode - indicate that with HARD_REGNO_MODE_OK
2110 and/or CANNOT_CHANGE_MODE_CLASS instead.
2112 On a machine where all registers are exactly one word, a suitable
2113 definition of this macro is
2116 #define HARD_REGNO_NREGS(REGNO, MODE) \
2117 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) \
2122 @defmac HARD_REGNO_NREGS_HAS_PADDING (@var{regno}, @var{mode})
2123 A C expression that is nonzero if a value of mode @var{mode}, stored
2124 in memory, ends with padding that causes it to take up more space than
2125 in registers starting at register number @var{regno} (as determined by
2126 multiplying GCC's notion of the size of the register when containing
2127 this mode by the number of registers returned by
2128 @code{HARD_REGNO_NREGS}). By default this is zero.
2130 For example, if a floating-point value is stored in three 32-bit
2131 registers but takes up 128 bits in memory, then this would be
2134 This macros only needs to be defined if there are cases where
2135 @code{subreg_get_info}
2136 would otherwise wrongly determine that a @code{subreg} can be
2137 represented by an offset to the register number, when in fact such a
2138 @code{subreg} would contain some of the padding not stored in
2139 registers and so not be representable.
2142 @defmac HARD_REGNO_NREGS_WITH_PADDING (@var{regno}, @var{mode})
2143 For values of @var{regno} and @var{mode} for which
2144 @code{HARD_REGNO_NREGS_HAS_PADDING} returns nonzero, a C expression
2145 returning the greater number of registers required to hold the value
2146 including any padding. In the example above, the value would be four.
2149 @defmac REGMODE_NATURAL_SIZE (@var{mode})
2150 Define this macro if the natural size of registers that hold values
2151 of mode @var{mode} is not the word size. It is a C expression that
2152 should give the natural size in bytes for the specified mode. It is
2153 used by the register allocator to try to optimize its results. This
2154 happens for example on SPARC 64-bit where the natural size of
2155 floating-point registers is still 32-bit.
2158 @defmac HARD_REGNO_MODE_OK (@var{regno}, @var{mode})
2159 A C expression that is nonzero if it is permissible to store a value
2160 of mode @var{mode} in hard register number @var{regno} (or in several
2161 registers starting with that one). For a machine where all registers
2162 are equivalent, a suitable definition is
2165 #define HARD_REGNO_MODE_OK(REGNO, MODE) 1
2168 You need not include code to check for the numbers of fixed registers,
2169 because the allocation mechanism considers them to be always occupied.
2171 @cindex register pairs
2172 On some machines, double-precision values must be kept in even/odd
2173 register pairs. You can implement that by defining this macro to reject
2174 odd register numbers for such modes.
2176 The minimum requirement for a mode to be OK in a register is that the
2177 @samp{mov@var{mode}} instruction pattern support moves between the
2178 register and other hard register in the same class and that moving a
2179 value into the register and back out not alter it.
2181 Since the same instruction used to move @code{word_mode} will work for
2182 all narrower integer modes, it is not necessary on any machine for
2183 @code{HARD_REGNO_MODE_OK} to distinguish between these modes, provided
2184 you define patterns @samp{movhi}, etc., to take advantage of this. This
2185 is useful because of the interaction between @code{HARD_REGNO_MODE_OK}
2186 and @code{MODES_TIEABLE_P}; it is very desirable for all integer modes
2189 Many machines have special registers for floating point arithmetic.
2190 Often people assume that floating point machine modes are allowed only
2191 in floating point registers. This is not true. Any registers that
2192 can hold integers can safely @emph{hold} a floating point machine
2193 mode, whether or not floating arithmetic can be done on it in those
2194 registers. Integer move instructions can be used to move the values.
2196 On some machines, though, the converse is true: fixed-point machine
2197 modes may not go in floating registers. This is true if the floating
2198 registers normalize any value stored in them, because storing a
2199 non-floating value there would garble it. In this case,
2200 @code{HARD_REGNO_MODE_OK} should reject fixed-point machine modes in
2201 floating registers. But if the floating registers do not automatically
2202 normalize, if you can store any bit pattern in one and retrieve it
2203 unchanged without a trap, then any machine mode may go in a floating
2204 register, so you can define this macro to say so.
2206 The primary significance of special floating registers is rather that
2207 they are the registers acceptable in floating point arithmetic
2208 instructions. However, this is of no concern to
2209 @code{HARD_REGNO_MODE_OK}. You handle it by writing the proper
2210 constraints for those instructions.
2212 On some machines, the floating registers are especially slow to access,
2213 so that it is better to store a value in a stack frame than in such a
2214 register if floating point arithmetic is not being done. As long as the
2215 floating registers are not in class @code{GENERAL_REGS}, they will not
2216 be used unless some pattern's constraint asks for one.
2219 @defmac HARD_REGNO_RENAME_OK (@var{from}, @var{to})
2220 A C expression that is nonzero if it is OK to rename a hard register
2221 @var{from} to another hard register @var{to}.
2223 One common use of this macro is to prevent renaming of a register to
2224 another register that is not saved by a prologue in an interrupt
2227 The default is always nonzero.
2230 @defmac MODES_TIEABLE_P (@var{mode1}, @var{mode2})
2231 A C expression that is nonzero if a value of mode
2232 @var{mode1} is accessible in mode @var{mode2} without copying.
2234 If @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode1})} and
2235 @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode2})} are always the same for
2236 any @var{r}, then @code{MODES_TIEABLE_P (@var{mode1}, @var{mode2})}
2237 should be nonzero. If they differ for any @var{r}, you should define
2238 this macro to return zero unless some other mechanism ensures the
2239 accessibility of the value in a narrower mode.
2241 You should define this macro to return nonzero in as many cases as
2242 possible since doing so will allow GCC to perform better register
2246 @deftypefn {Target Hook} bool TARGET_HARD_REGNO_SCRATCH_OK (unsigned int @var{regno})
2247 This target hook should return @code{true} if it is OK to use a hard register
2248 @var{regno} as scratch reg in peephole2.
2250 One common use of this macro is to prevent using of a register that
2251 is not saved by a prologue in an interrupt handler.
2253 The default version of this hook always returns @code{true}.
2256 @defmac AVOID_CCMODE_COPIES
2257 Define this macro if the compiler should avoid copies to/from @code{CCmode}
2258 registers. You should only define this macro if support for copying to/from
2259 @code{CCmode} is incomplete.
2262 @node Leaf Functions
2263 @subsection Handling Leaf Functions
2265 @cindex leaf functions
2266 @cindex functions, leaf
2267 On some machines, a leaf function (i.e., one which makes no calls) can run
2268 more efficiently if it does not make its own register window. Often this
2269 means it is required to receive its arguments in the registers where they
2270 are passed by the caller, instead of the registers where they would
2273 The special treatment for leaf functions generally applies only when
2274 other conditions are met; for example, often they may use only those
2275 registers for its own variables and temporaries. We use the term ``leaf
2276 function'' to mean a function that is suitable for this special
2277 handling, so that functions with no calls are not necessarily ``leaf
2280 GCC assigns register numbers before it knows whether the function is
2281 suitable for leaf function treatment. So it needs to renumber the
2282 registers in order to output a leaf function. The following macros
2285 @defmac LEAF_REGISTERS
2286 Name of a char vector, indexed by hard register number, which
2287 contains 1 for a register that is allowable in a candidate for leaf
2290 If leaf function treatment involves renumbering the registers, then the
2291 registers marked here should be the ones before renumbering---those that
2292 GCC would ordinarily allocate. The registers which will actually be
2293 used in the assembler code, after renumbering, should not be marked with 1
2296 Define this macro only if the target machine offers a way to optimize
2297 the treatment of leaf functions.
2300 @defmac LEAF_REG_REMAP (@var{regno})
2301 A C expression whose value is the register number to which @var{regno}
2302 should be renumbered, when a function is treated as a leaf function.
2304 If @var{regno} is a register number which should not appear in a leaf
2305 function before renumbering, then the expression should yield @minus{}1, which
2306 will cause the compiler to abort.
2308 Define this macro only if the target machine offers a way to optimize the
2309 treatment of leaf functions, and registers need to be renumbered to do
2313 @findex current_function_is_leaf
2314 @findex current_function_uses_only_leaf_regs
2315 @code{TARGET_ASM_FUNCTION_PROLOGUE} and
2316 @code{TARGET_ASM_FUNCTION_EPILOGUE} must usually treat leaf functions
2317 specially. They can test the C variable @code{current_function_is_leaf}
2318 which is nonzero for leaf functions. @code{current_function_is_leaf} is
2319 set prior to local register allocation and is valid for the remaining
2320 compiler passes. They can also test the C variable
2321 @code{current_function_uses_only_leaf_regs} which is nonzero for leaf
2322 functions which only use leaf registers.
2323 @code{current_function_uses_only_leaf_regs} is valid after all passes
2324 that modify the instructions have been run and is only useful if
2325 @code{LEAF_REGISTERS} is defined.
2326 @c changed this to fix overfull. ALSO: why the "it" at the beginning
2327 @c of the next paragraph?! --mew 2feb93
2329 @node Stack Registers
2330 @subsection Registers That Form a Stack
2332 There are special features to handle computers where some of the
2333 ``registers'' form a stack. Stack registers are normally written by
2334 pushing onto the stack, and are numbered relative to the top of the
2337 Currently, GCC can only handle one group of stack-like registers, and
2338 they must be consecutively numbered. Furthermore, the existing
2339 support for stack-like registers is specific to the 80387 floating
2340 point coprocessor. If you have a new architecture that uses
2341 stack-like registers, you will need to do substantial work on
2342 @file{reg-stack.c} and write your machine description to cooperate
2343 with it, as well as defining these macros.
2346 Define this if the machine has any stack-like registers.
2349 @defmac STACK_REG_COVER_CLASS
2350 This is a cover class containing the stack registers. Define this if
2351 the machine has any stack-like registers.
2354 @defmac FIRST_STACK_REG
2355 The number of the first stack-like register. This one is the top
2359 @defmac LAST_STACK_REG
2360 The number of the last stack-like register. This one is the bottom of
2364 @node Register Classes
2365 @section Register Classes
2366 @cindex register class definitions
2367 @cindex class definitions, register
2369 On many machines, the numbered registers are not all equivalent.
2370 For example, certain registers may not be allowed for indexed addressing;
2371 certain registers may not be allowed in some instructions. These machine
2372 restrictions are described to the compiler using @dfn{register classes}.
2374 You define a number of register classes, giving each one a name and saying
2375 which of the registers belong to it. Then you can specify register classes
2376 that are allowed as operands to particular instruction patterns.
2380 In general, each register will belong to several classes. In fact, one
2381 class must be named @code{ALL_REGS} and contain all the registers. Another
2382 class must be named @code{NO_REGS} and contain no registers. Often the
2383 union of two classes will be another class; however, this is not required.
2385 @findex GENERAL_REGS
2386 One of the classes must be named @code{GENERAL_REGS}. There is nothing
2387 terribly special about the name, but the operand constraint letters
2388 @samp{r} and @samp{g} specify this class. If @code{GENERAL_REGS} is
2389 the same as @code{ALL_REGS}, just define it as a macro which expands
2392 Order the classes so that if class @var{x} is contained in class @var{y}
2393 then @var{x} has a lower class number than @var{y}.
2395 The way classes other than @code{GENERAL_REGS} are specified in operand
2396 constraints is through machine-dependent operand constraint letters.
2397 You can define such letters to correspond to various classes, then use
2398 them in operand constraints.
2400 You should define a class for the union of two classes whenever some
2401 instruction allows both classes. For example, if an instruction allows
2402 either a floating point (coprocessor) register or a general register for a
2403 certain operand, you should define a class @code{FLOAT_OR_GENERAL_REGS}
2404 which includes both of them. Otherwise you will get suboptimal code.
2406 You must also specify certain redundant information about the register
2407 classes: for each class, which classes contain it and which ones are
2408 contained in it; for each pair of classes, the largest class contained
2411 When a value occupying several consecutive registers is expected in a
2412 certain class, all the registers used must belong to that class.
2413 Therefore, register classes cannot be used to enforce a requirement for
2414 a register pair to start with an even-numbered register. The way to
2415 specify this requirement is with @code{HARD_REGNO_MODE_OK}.
2417 Register classes used for input-operands of bitwise-and or shift
2418 instructions have a special requirement: each such class must have, for
2419 each fixed-point machine mode, a subclass whose registers can transfer that
2420 mode to or from memory. For example, on some machines, the operations for
2421 single-byte values (@code{QImode}) are limited to certain registers. When
2422 this is so, each register class that is used in a bitwise-and or shift
2423 instruction must have a subclass consisting of registers from which
2424 single-byte values can be loaded or stored. This is so that
2425 @code{PREFERRED_RELOAD_CLASS} can always have a possible value to return.
2427 @deftp {Data type} {enum reg_class}
2428 An enumerated type that must be defined with all the register class names
2429 as enumerated values. @code{NO_REGS} must be first. @code{ALL_REGS}
2430 must be the last register class, followed by one more enumerated value,
2431 @code{LIM_REG_CLASSES}, which is not a register class but rather
2432 tells how many classes there are.
2434 Each register class has a number, which is the value of casting
2435 the class name to type @code{int}. The number serves as an index
2436 in many of the tables described below.
2439 @defmac N_REG_CLASSES
2440 The number of distinct register classes, defined as follows:
2443 #define N_REG_CLASSES (int) LIM_REG_CLASSES
2447 @defmac REG_CLASS_NAMES
2448 An initializer containing the names of the register classes as C string
2449 constants. These names are used in writing some of the debugging dumps.
2452 @defmac REG_CLASS_CONTENTS
2453 An initializer containing the contents of the register classes, as integers
2454 which are bit masks. The @var{n}th integer specifies the contents of class
2455 @var{n}. The way the integer @var{mask} is interpreted is that
2456 register @var{r} is in the class if @code{@var{mask} & (1 << @var{r})} is 1.
2458 When the machine has more than 32 registers, an integer does not suffice.
2459 Then the integers are replaced by sub-initializers, braced groupings containing
2460 several integers. Each sub-initializer must be suitable as an initializer
2461 for the type @code{HARD_REG_SET} which is defined in @file{hard-reg-set.h}.
2462 In this situation, the first integer in each sub-initializer corresponds to
2463 registers 0 through 31, the second integer to registers 32 through 63, and
2467 @defmac REGNO_REG_CLASS (@var{regno})
2468 A C expression whose value is a register class containing hard register
2469 @var{regno}. In general there is more than one such class; choose a class
2470 which is @dfn{minimal}, meaning that no smaller class also contains the
2474 @defmac BASE_REG_CLASS
2475 A macro whose definition is the name of the class to which a valid
2476 base register must belong. A base register is one used in an address
2477 which is the register value plus a displacement.
2480 @defmac MODE_BASE_REG_CLASS (@var{mode})
2481 This is a variation of the @code{BASE_REG_CLASS} macro which allows
2482 the selection of a base register in a mode dependent manner. If
2483 @var{mode} is VOIDmode then it should return the same value as
2484 @code{BASE_REG_CLASS}.
2487 @defmac MODE_BASE_REG_REG_CLASS (@var{mode})
2488 A C expression whose value is the register class to which a valid
2489 base register must belong in order to be used in a base plus index
2490 register address. You should define this macro if base plus index
2491 addresses have different requirements than other base register uses.
2494 @defmac MODE_CODE_BASE_REG_CLASS (@var{mode}, @var{outer_code}, @var{index_code})
2495 A C expression whose value is the register class to which a valid
2496 base register must belong. @var{outer_code} and @var{index_code} define the
2497 context in which the base register occurs. @var{outer_code} is the code of
2498 the immediately enclosing expression (@code{MEM} for the top level of an
2499 address, @code{ADDRESS} for something that occurs in an
2500 @code{address_operand}). @var{index_code} is the code of the corresponding
2501 index expression if @var{outer_code} is @code{PLUS}; @code{SCRATCH} otherwise.
2504 @defmac INDEX_REG_CLASS
2505 A macro whose definition is the name of the class to which a valid
2506 index register must belong. An index register is one used in an
2507 address where its value is either multiplied by a scale factor or
2508 added to another register (as well as added to a displacement).
2511 @defmac REGNO_OK_FOR_BASE_P (@var{num})
2512 A C expression which is nonzero if register number @var{num} is
2513 suitable for use as a base register in operand addresses.
2514 Like @code{TARGET_LEGITIMATE_ADDRESS_P}, this macro should also
2515 define a strict and a non-strict variant. Both variants behave
2516 the same for hard register; for pseudos, the strict variant will
2517 pass only those that have been allocated to a valid hard registers,
2518 while the non-strict variant will pass all pseudos.
2520 @findex REG_OK_STRICT
2521 Compiler source files that want to use the strict variant of this and
2522 other macros define the macro @code{REG_OK_STRICT}. You should use an
2523 @code{#ifdef REG_OK_STRICT} conditional to define the strict variant in
2524 that case and the non-strict variant otherwise.
2527 @defmac REGNO_MODE_OK_FOR_BASE_P (@var{num}, @var{mode})
2528 A C expression that is just like @code{REGNO_OK_FOR_BASE_P}, except that
2529 that expression may examine the mode of the memory reference in
2530 @var{mode}. You should define this macro if the mode of the memory
2531 reference affects whether a register may be used as a base register. If
2532 you define this macro, the compiler will use it instead of
2533 @code{REGNO_OK_FOR_BASE_P}. The mode may be @code{VOIDmode} for
2534 addresses that appear outside a @code{MEM}, i.e., as an
2535 @code{address_operand}.
2537 This macro also has strict and non-strict variants.
2540 @defmac REGNO_MODE_OK_FOR_REG_BASE_P (@var{num}, @var{mode})
2541 A C expression which is nonzero if register number @var{num} is suitable for
2542 use as a base register in base plus index operand addresses, accessing
2543 memory in mode @var{mode}. It may be either a suitable hard register or a
2544 pseudo register that has been allocated such a hard register. You should
2545 define this macro if base plus index addresses have different requirements
2546 than other base register uses.
2548 Use of this macro is deprecated; please use the more general
2549 @code{REGNO_MODE_CODE_OK_FOR_BASE_P}.
2551 This macro also has strict and non-strict variants.
2554 @defmac REGNO_MODE_CODE_OK_FOR_BASE_P (@var{num}, @var{mode}, @var{outer_code}, @var{index_code})
2555 A C expression that is just like @code{REGNO_MODE_OK_FOR_BASE_P}, except
2556 that that expression may examine the context in which the register
2557 appears in the memory reference. @var{outer_code} is the code of the
2558 immediately enclosing expression (@code{MEM} if at the top level of the
2559 address, @code{ADDRESS} for something that occurs in an
2560 @code{address_operand}). @var{index_code} is the code of the
2561 corresponding index expression if @var{outer_code} is @code{PLUS};
2562 @code{SCRATCH} otherwise. The mode may be @code{VOIDmode} for addresses
2563 that appear outside a @code{MEM}, i.e., as an @code{address_operand}.
2565 This macro also has strict and non-strict variants.
2568 @defmac REGNO_OK_FOR_INDEX_P (@var{num})
2569 A C expression which is nonzero if register number @var{num} is
2570 suitable for use as an index register in operand addresses. It may be
2571 either a suitable hard register or a pseudo register that has been
2572 allocated such a hard register.
2574 The difference between an index register and a base register is that
2575 the index register may be scaled. If an address involves the sum of
2576 two registers, neither one of them scaled, then either one may be
2577 labeled the ``base'' and the other the ``index''; but whichever
2578 labeling is used must fit the machine's constraints of which registers
2579 may serve in each capacity. The compiler will try both labelings,
2580 looking for one that is valid, and will reload one or both registers
2581 only if neither labeling works.
2583 This macro also has strict and non-strict variants.
2586 @defmac PREFERRED_RELOAD_CLASS (@var{x}, @var{class})
2587 A C expression that places additional restrictions on the register class
2588 to use when it is necessary to copy value @var{x} into a register in class
2589 @var{class}. The value is a register class; perhaps @var{class}, or perhaps
2590 another, smaller class. On many machines, the following definition is
2594 #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS
2597 Sometimes returning a more restrictive class makes better code. For
2598 example, on the 68000, when @var{x} is an integer constant that is in range
2599 for a @samp{moveq} instruction, the value of this macro is always
2600 @code{DATA_REGS} as long as @var{class} includes the data registers.
2601 Requiring a data register guarantees that a @samp{moveq} will be used.
2603 One case where @code{PREFERRED_RELOAD_CLASS} must not return
2604 @var{class} is if @var{x} is a legitimate constant which cannot be
2605 loaded into some register class. By returning @code{NO_REGS} you can
2606 force @var{x} into a memory location. For example, rs6000 can load
2607 immediate values into general-purpose registers, but does not have an
2608 instruction for loading an immediate value into a floating-point
2609 register, so @code{PREFERRED_RELOAD_CLASS} returns @code{NO_REGS} when
2610 @var{x} is a floating-point constant. If the constant can't be loaded
2611 into any kind of register, code generation will be better if
2612 @code{LEGITIMATE_CONSTANT_P} makes the constant illegitimate instead
2613 of using @code{PREFERRED_RELOAD_CLASS}.
2615 If an insn has pseudos in it after register allocation, reload will go
2616 through the alternatives and call repeatedly @code{PREFERRED_RELOAD_CLASS}
2617 to find the best one. Returning @code{NO_REGS}, in this case, makes
2618 reload add a @code{!} in front of the constraint: the x86 back-end uses
2619 this feature to discourage usage of 387 registers when math is done in
2620 the SSE registers (and vice versa).
2623 @defmac PREFERRED_OUTPUT_RELOAD_CLASS (@var{x}, @var{class})
2624 Like @code{PREFERRED_RELOAD_CLASS}, but for output reloads instead of
2625 input reloads. If you don't define this macro, the default is to use
2626 @var{class}, unchanged.
2628 You can also use @code{PREFERRED_OUTPUT_RELOAD_CLASS} to discourage
2629 reload from using some alternatives, like @code{PREFERRED_RELOAD_CLASS}.
2632 @defmac LIMIT_RELOAD_CLASS (@var{mode}, @var{class})
2633 A C expression that places additional restrictions on the register class
2634 to use when it is necessary to be able to hold a value of mode
2635 @var{mode} in a reload register for which class @var{class} would
2638 Unlike @code{PREFERRED_RELOAD_CLASS}, this macro should be used when
2639 there are certain modes that simply can't go in certain reload classes.
2641 The value is a register class; perhaps @var{class}, or perhaps another,
2644 Don't define this macro unless the target machine has limitations which
2645 require the macro to do something nontrivial.
2648 @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})
2649 Many machines have some registers that cannot be copied directly to or
2650 from memory or even from other types of registers. An example is the
2651 @samp{MQ} register, which on most machines, can only be copied to or
2652 from general registers, but not memory. Below, we shall be using the
2653 term 'intermediate register' when a move operation cannot be performed
2654 directly, but has to be done by copying the source into the intermediate
2655 register first, and then copying the intermediate register to the
2656 destination. An intermediate register always has the same mode as
2657 source and destination. Since it holds the actual value being copied,
2658 reload might apply optimizations to re-use an intermediate register
2659 and eliding the copy from the source when it can determine that the
2660 intermediate register still holds the required value.
2662 Another kind of secondary reload is required on some machines which
2663 allow copying all registers to and from memory, but require a scratch
2664 register for stores to some memory locations (e.g., those with symbolic
2665 address on the RT, and those with certain symbolic address on the SPARC
2666 when compiling PIC)@. Scratch registers need not have the same mode
2667 as the value being copied, and usually hold a different value than
2668 that being copied. Special patterns in the md file are needed to
2669 describe how the copy is performed with the help of the scratch register;
2670 these patterns also describe the number, register class(es) and mode(s)
2671 of the scratch register(s).
2673 In some cases, both an intermediate and a scratch register are required.
2675 For input reloads, this target hook is called with nonzero @var{in_p},
2676 and @var{x} is an rtx that needs to be copied to a register of class
2677 @var{reload_class} in @var{reload_mode}. For output reloads, this target
2678 hook is called with zero @var{in_p}, and a register of class @var{reload_class}
2679 needs to be copied to rtx @var{x} in @var{reload_mode}.
2681 If copying a register of @var{reload_class} from/to @var{x} requires
2682 an intermediate register, the hook @code{secondary_reload} should
2683 return the register class required for this intermediate register.
2684 If no intermediate register is required, it should return NO_REGS.
2685 If more than one intermediate register is required, describe the one
2686 that is closest in the copy chain to the reload register.
2688 If scratch registers are needed, you also have to describe how to
2689 perform the copy from/to the reload register to/from this
2690 closest intermediate register. Or if no intermediate register is
2691 required, but still a scratch register is needed, describe the
2692 copy from/to the reload register to/from the reload operand @var{x}.
2694 You do this by setting @code{sri->icode} to the instruction code of a pattern
2695 in the md file which performs the move. Operands 0 and 1 are the output
2696 and input of this copy, respectively. Operands from operand 2 onward are
2697 for scratch operands. These scratch operands must have a mode, and a
2698 single-register-class
2699 @c [later: or memory]
2702 When an intermediate register is used, the @code{secondary_reload}
2703 hook will be called again to determine how to copy the intermediate
2704 register to/from the reload operand @var{x}, so your hook must also
2705 have code to handle the register class of the intermediate operand.
2707 @c [For later: maybe we'll allow multi-alternative reload patterns -
2708 @c the port maintainer could name a mov<mode> pattern that has clobbers -
2709 @c and match the constraints of input and output to determine the required
2710 @c alternative. A restriction would be that constraints used to match
2711 @c against reloads registers would have to be written as register class
2712 @c constraints, or we need a new target macro / hook that tells us if an
2713 @c arbitrary constraint can match an unknown register of a given class.
2714 @c Such a macro / hook would also be useful in other places.]
2717 @var{x} might be a pseudo-register or a @code{subreg} of a
2718 pseudo-register, which could either be in a hard register or in memory.
2719 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2720 in memory and the hard register number if it is in a register.
2722 Scratch operands in memory (constraint @code{"=m"} / @code{"=&m"}) are
2723 currently not supported. For the time being, you will have to continue
2724 to use @code{SECONDARY_MEMORY_NEEDED} for that purpose.
2726 @code{copy_cost} also uses this target hook to find out how values are
2727 copied. If you want it to include some extra cost for the need to allocate
2728 (a) scratch register(s), set @code{sri->extra_cost} to the additional cost.
2729 Or if two dependent moves are supposed to have a lower cost than the sum
2730 of the individual moves due to expected fortuitous scheduling and/or special
2731 forwarding logic, you can set @code{sri->extra_cost} to a negative amount.
2734 @defmac SECONDARY_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2735 @defmacx SECONDARY_INPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2736 @defmacx SECONDARY_OUTPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2737 These macros are obsolete, new ports should use the target hook
2738 @code{TARGET_SECONDARY_RELOAD} instead.
2740 These are obsolete macros, replaced by the @code{TARGET_SECONDARY_RELOAD}
2741 target hook. Older ports still define these macros to indicate to the
2742 reload phase that it may
2743 need to allocate at least one register for a reload in addition to the
2744 register to contain the data. Specifically, if copying @var{x} to a
2745 register @var{class} in @var{mode} requires an intermediate register,
2746 you were supposed to define @code{SECONDARY_INPUT_RELOAD_CLASS} to return the
2747 largest register class all of whose registers can be used as
2748 intermediate registers or scratch registers.
2750 If copying a register @var{class} in @var{mode} to @var{x} requires an
2751 intermediate or scratch register, @code{SECONDARY_OUTPUT_RELOAD_CLASS}
2752 was supposed to be defined be defined to return the largest register
2753 class required. If the
2754 requirements for input and output reloads were the same, the macro
2755 @code{SECONDARY_RELOAD_CLASS} should have been used instead of defining both
2758 The values returned by these macros are often @code{GENERAL_REGS}.
2759 Return @code{NO_REGS} if no spare register is needed; i.e., if @var{x}
2760 can be directly copied to or from a register of @var{class} in
2761 @var{mode} without requiring a scratch register. Do not define this
2762 macro if it would always return @code{NO_REGS}.
2764 If a scratch register is required (either with or without an
2765 intermediate register), you were supposed to define patterns for
2766 @samp{reload_in@var{m}} or @samp{reload_out@var{m}}, as required
2767 (@pxref{Standard Names}. These patterns, which were normally
2768 implemented with a @code{define_expand}, should be similar to the
2769 @samp{mov@var{m}} patterns, except that operand 2 is the scratch
2772 These patterns need constraints for the reload register and scratch
2774 contain a single register class. If the original reload register (whose
2775 class is @var{class}) can meet the constraint given in the pattern, the
2776 value returned by these macros is used for the class of the scratch
2777 register. Otherwise, two additional reload registers are required.
2778 Their classes are obtained from the constraints in the insn pattern.
2780 @var{x} might be a pseudo-register or a @code{subreg} of a
2781 pseudo-register, which could either be in a hard register or in memory.
2782 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2783 in memory and the hard register number if it is in a register.
2785 These macros should not be used in the case where a particular class of
2786 registers can only be copied to memory and not to another class of
2787 registers. In that case, secondary reload registers are not needed and
2788 would not be helpful. Instead, a stack location must be used to perform
2789 the copy and the @code{mov@var{m}} pattern should use memory as an
2790 intermediate storage. This case often occurs between floating-point and
2794 @defmac SECONDARY_MEMORY_NEEDED (@var{class1}, @var{class2}, @var{m})
2795 Certain machines have the property that some registers cannot be copied
2796 to some other registers without using memory. Define this macro on
2797 those machines to be a C expression that is nonzero if objects of mode
2798 @var{m} in registers of @var{class1} can only be copied to registers of
2799 class @var{class2} by storing a register of @var{class1} into memory
2800 and loading that memory location into a register of @var{class2}.
2802 Do not define this macro if its value would always be zero.
2805 @defmac SECONDARY_MEMORY_NEEDED_RTX (@var{mode})
2806 Normally when @code{SECONDARY_MEMORY_NEEDED} is defined, the compiler
2807 allocates a stack slot for a memory location needed for register copies.
2808 If this macro is defined, the compiler instead uses the memory location
2809 defined by this macro.
2811 Do not define this macro if you do not define
2812 @code{SECONDARY_MEMORY_NEEDED}.
2815 @defmac SECONDARY_MEMORY_NEEDED_MODE (@var{mode})
2816 When the compiler needs a secondary memory location to copy between two
2817 registers of mode @var{mode}, it normally allocates sufficient memory to
2818 hold a quantity of @code{BITS_PER_WORD} bits and performs the store and
2819 load operations in a mode that many bits wide and whose class is the
2820 same as that of @var{mode}.
2822 This is right thing to do on most machines because it ensures that all
2823 bits of the register are copied and prevents accesses to the registers
2824 in a narrower mode, which some machines prohibit for floating-point
2827 However, this default behavior is not correct on some machines, such as
2828 the DEC Alpha, that store short integers in floating-point registers
2829 differently than in integer registers. On those machines, the default
2830 widening will not work correctly and you must define this macro to
2831 suppress that widening in some cases. See the file @file{alpha.h} for
2834 Do not define this macro if you do not define
2835 @code{SECONDARY_MEMORY_NEEDED} or if widening @var{mode} to a mode that
2836 is @code{BITS_PER_WORD} bits wide is correct for your machine.
2839 @defmac CLASS_LIKELY_SPILLED_P (@var{class})
2840 A C expression whose value is nonzero if pseudos that have been assigned
2841 to registers of class @var{class} would likely be spilled because
2842 registers of @var{class} are needed for spill registers.
2844 The default value of this macro returns 1 if @var{class} has exactly one
2845 register and zero otherwise. On most machines, this default should be
2846 used. Only define this macro to some other expression if pseudos
2847 allocated by @file{local-alloc.c} end up in memory because their hard
2848 registers were needed for spill registers. If this macro returns nonzero
2849 for those classes, those pseudos will only be allocated by
2850 @file{global.c}, which knows how to reallocate the pseudo to another
2851 register. If there would not be another register available for
2852 reallocation, you should not change the definition of this macro since
2853 the only effect of such a definition would be to slow down register
2857 @defmac CLASS_MAX_NREGS (@var{class}, @var{mode})
2858 A C expression for the maximum number of consecutive registers
2859 of class @var{class} needed to hold a value of mode @var{mode}.
2861 This is closely related to the macro @code{HARD_REGNO_NREGS}. In fact,
2862 the value of the macro @code{CLASS_MAX_NREGS (@var{class}, @var{mode})}
2863 should be the maximum value of @code{HARD_REGNO_NREGS (@var{regno},
2864 @var{mode})} for all @var{regno} values in the class @var{class}.
2866 This macro helps control the handling of multiple-word values
2870 @defmac CANNOT_CHANGE_MODE_CLASS (@var{from}, @var{to}, @var{class})
2871 If defined, a C expression that returns nonzero for a @var{class} for which
2872 a change from mode @var{from} to mode @var{to} is invalid.
2874 For the example, loading 32-bit integer or floating-point objects into
2875 floating-point registers on the Alpha extends them to 64 bits.
2876 Therefore loading a 64-bit object and then storing it as a 32-bit object
2877 does not store the low-order 32 bits, as would be the case for a normal
2878 register. Therefore, @file{alpha.h} defines @code{CANNOT_CHANGE_MODE_CLASS}
2882 #define CANNOT_CHANGE_MODE_CLASS(FROM, TO, CLASS) \
2883 (GET_MODE_SIZE (FROM) != GET_MODE_SIZE (TO) \
2884 ? reg_classes_intersect_p (FLOAT_REGS, (CLASS)) : 0)
2888 @deftypefn {Target Hook} {const enum reg_class *} TARGET_IRA_COVER_CLASSES (void)
2889 Return an array of cover classes for the Integrated Register Allocator
2890 (@acronym{IRA}). Cover classes are a set of non-intersecting register
2891 classes covering all hard registers used for register allocation
2892 purposes. If a move between two registers in the same cover class is
2893 possible, it should be cheaper than a load or store of the registers.
2894 The array is terminated by a @code{LIM_REG_CLASSES} element.
2896 The order of cover classes in the array is important. If two classes
2897 have the same cost of usage for a pseudo, the class occurred first in
2898 the array is chosen for the pseudo.
2900 This hook is called once at compiler startup, after the command-line
2901 options have been processed. It is then re-examined by every call to
2902 @code{target_reinit}.
2904 The default implementation returns @code{IRA_COVER_CLASSES}, if defined,
2905 otherwise there is no default implementation. You must define either this
2906 macro or @code{IRA_COVER_CLASSES} in order to use the integrated register
2907 allocator with Chaitin-Briggs coloring. If the macro is not defined,
2908 the only available coloring algorithm is Chow's priority coloring.
2911 @defmac IRA_COVER_CLASSES
2912 See the documentation for @code{TARGET_IRA_COVER_CLASSES}.
2915 @node Old Constraints
2916 @section Obsolete Macros for Defining Constraints
2917 @cindex defining constraints, obsolete method
2918 @cindex constraints, defining, obsolete method
2920 Machine-specific constraints can be defined with these macros instead
2921 of the machine description constructs described in @ref{Define
2922 Constraints}. This mechanism is obsolete. New ports should not use
2923 it; old ports should convert to the new mechanism.
2925 @defmac CONSTRAINT_LEN (@var{char}, @var{str})
2926 For the constraint at the start of @var{str}, which starts with the letter
2927 @var{c}, return the length. This allows you to have register class /
2928 constant / extra constraints that are longer than a single letter;
2929 you don't need to define this macro if you can do with single-letter
2930 constraints only. The definition of this macro should use
2931 DEFAULT_CONSTRAINT_LEN for all the characters that you don't want
2932 to handle specially.
2933 There are some sanity checks in genoutput.c that check the constraint lengths
2934 for the md file, so you can also use this macro to help you while you are
2935 transitioning from a byzantine single-letter-constraint scheme: when you
2936 return a negative length for a constraint you want to re-use, genoutput
2937 will complain about every instance where it is used in the md file.
2940 @defmac REG_CLASS_FROM_LETTER (@var{char})
2941 A C expression which defines the machine-dependent operand constraint
2942 letters for register classes. If @var{char} is such a letter, the
2943 value should be the register class corresponding to it. Otherwise,
2944 the value should be @code{NO_REGS}. The register letter @samp{r},
2945 corresponding to class @code{GENERAL_REGS}, will not be passed
2946 to this macro; you do not need to handle it.
2949 @defmac REG_CLASS_FROM_CONSTRAINT (@var{char}, @var{str})
2950 Like @code{REG_CLASS_FROM_LETTER}, but you also get the constraint string
2951 passed in @var{str}, so that you can use suffixes to distinguish between
2955 @defmac CONST_OK_FOR_LETTER_P (@var{value}, @var{c})
2956 A C expression that defines the machine-dependent operand constraint
2957 letters (@samp{I}, @samp{J}, @samp{K}, @dots{} @samp{P}) that specify
2958 particular ranges of integer values. If @var{c} is one of those
2959 letters, the expression should check that @var{value}, an integer, is in
2960 the appropriate range and return 1 if so, 0 otherwise. If @var{c} is
2961 not one of those letters, the value should be 0 regardless of
2965 @defmac CONST_OK_FOR_CONSTRAINT_P (@var{value}, @var{c}, @var{str})
2966 Like @code{CONST_OK_FOR_LETTER_P}, but you also get the constraint
2967 string passed in @var{str}, so that you can use suffixes to distinguish
2968 between different variants.
2971 @defmac CONST_DOUBLE_OK_FOR_LETTER_P (@var{value}, @var{c})
2972 A C expression that defines the machine-dependent operand constraint
2973 letters that specify particular ranges of @code{const_double} values
2974 (@samp{G} or @samp{H}).
2976 If @var{c} is one of those letters, the expression should check that
2977 @var{value}, an RTX of code @code{const_double}, is in the appropriate
2978 range and return 1 if so, 0 otherwise. If @var{c} is not one of those
2979 letters, the value should be 0 regardless of @var{value}.
2981 @code{const_double} is used for all floating-point constants and for
2982 @code{DImode} fixed-point constants. A given letter can accept either
2983 or both kinds of values. It can use @code{GET_MODE} to distinguish
2984 between these kinds.
2987 @defmac CONST_DOUBLE_OK_FOR_CONSTRAINT_P (@var{value}, @var{c}, @var{str})
2988 Like @code{CONST_DOUBLE_OK_FOR_LETTER_P}, but you also get the constraint
2989 string passed in @var{str}, so that you can use suffixes to distinguish
2990 between different variants.
2993 @defmac EXTRA_CONSTRAINT (@var{value}, @var{c})
2994 A C expression that defines the optional machine-dependent constraint
2995 letters that can be used to segregate specific types of operands, usually
2996 memory references, for the target machine. Any letter that is not
2997 elsewhere defined and not matched by @code{REG_CLASS_FROM_LETTER} /
2998 @code{REG_CLASS_FROM_CONSTRAINT}
2999 may be used. Normally this macro will not be defined.
3001 If it is required for a particular target machine, it should return 1
3002 if @var{value} corresponds to the operand type represented by the
3003 constraint letter @var{c}. If @var{c} is not defined as an extra
3004 constraint, the value returned should be 0 regardless of @var{value}.
3006 For example, on the ROMP, load instructions cannot have their output
3007 in r0 if the memory reference contains a symbolic address. Constraint
3008 letter @samp{Q} is defined as representing a memory address that does
3009 @emph{not} contain a symbolic address. An alternative is specified with
3010 a @samp{Q} constraint on the input and @samp{r} on the output. The next
3011 alternative specifies @samp{m} on the input and a register class that
3012 does not include r0 on the output.
3015 @defmac EXTRA_CONSTRAINT_STR (@var{value}, @var{c}, @var{str})
3016 Like @code{EXTRA_CONSTRAINT}, but you also get the constraint string passed
3017 in @var{str}, so that you can use suffixes to distinguish between different
3021 @defmac EXTRA_MEMORY_CONSTRAINT (@var{c}, @var{str})
3022 A C expression that defines the optional machine-dependent constraint
3023 letters, amongst those accepted by @code{EXTRA_CONSTRAINT}, that should
3024 be treated like memory constraints by the reload pass.
3026 It should return 1 if the operand type represented by the constraint
3027 at the start of @var{str}, the first letter of which is the letter @var{c},
3028 comprises a subset of all memory references including
3029 all those whose address is simply a base register. This allows the reload
3030 pass to reload an operand, if it does not directly correspond to the operand
3031 type of @var{c}, by copying its address into a base register.
3033 For example, on the S/390, some instructions do not accept arbitrary
3034 memory references, but only those that do not make use of an index
3035 register. The constraint letter @samp{Q} is defined via
3036 @code{EXTRA_CONSTRAINT} as representing a memory address of this type.
3037 If the letter @samp{Q} is marked as @code{EXTRA_MEMORY_CONSTRAINT},
3038 a @samp{Q} constraint can handle any memory operand, because the
3039 reload pass knows it can be reloaded by copying the memory address
3040 into a base register if required. This is analogous to the way
3041 an @samp{o} constraint can handle any memory operand.
3044 @defmac EXTRA_ADDRESS_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} /
3047 @code{EXTRA_CONSTRAINT_STR}, that should
3048 be treated like address constraints by the reload pass.
3050 It should return 1 if the operand type represented by the constraint
3051 at the start of @var{str}, which starts with the letter @var{c}, comprises
3052 a subset of all memory addresses including
3053 all those that consist of just a base register. This allows the reload
3054 pass to reload an operand, if it does not directly correspond to the operand
3055 type of @var{str}, by copying it into a base register.
3057 Any constraint marked as @code{EXTRA_ADDRESS_CONSTRAINT} can only
3058 be used with the @code{address_operand} predicate. It is treated
3059 analogously to the @samp{p} constraint.
3062 @node Stack and Calling
3063 @section Stack Layout and Calling Conventions
3064 @cindex calling conventions
3066 @c prevent bad page break with this line
3067 This describes the stack layout and calling conventions.
3071 * Exception Handling::
3076 * Register Arguments::
3078 * Aggregate Return::
3083 * Stack Smashing Protection::
3087 @subsection Basic Stack Layout
3088 @cindex stack frame layout
3089 @cindex frame layout
3091 @c prevent bad page break with this line
3092 Here is the basic stack layout.
3094 @defmac STACK_GROWS_DOWNWARD
3095 Define this macro if pushing a word onto the stack moves the stack
3096 pointer to a smaller address.
3098 When we say, ``define this macro if @dots{}'', it means that the
3099 compiler checks this macro only with @code{#ifdef} so the precise
3100 definition used does not matter.
3103 @defmac STACK_PUSH_CODE
3104 This macro defines the operation used when something is pushed
3105 on the stack. In RTL, a push operation will be
3106 @code{(set (mem (STACK_PUSH_CODE (reg sp))) @dots{})}
3108 The choices are @code{PRE_DEC}, @code{POST_DEC}, @code{PRE_INC},
3109 and @code{POST_INC}. Which of these is correct depends on
3110 the stack direction and on whether the stack pointer points
3111 to the last item on the stack or whether it points to the
3112 space for the next item on the stack.
3114 The default is @code{PRE_DEC} when @code{STACK_GROWS_DOWNWARD} is
3115 defined, which is almost always right, and @code{PRE_INC} otherwise,
3116 which is often wrong.
3119 @defmac FRAME_GROWS_DOWNWARD
3120 Define this macro to nonzero value if the addresses of local variable slots
3121 are at negative offsets from the frame pointer.
3124 @defmac ARGS_GROW_DOWNWARD
3125 Define this macro if successive arguments to a function occupy decreasing
3126 addresses on the stack.
3129 @defmac STARTING_FRAME_OFFSET
3130 Offset from the frame pointer to the first local variable slot to be allocated.
3132 If @code{FRAME_GROWS_DOWNWARD}, find the next slot's offset by
3133 subtracting the first slot's length from @code{STARTING_FRAME_OFFSET}.
3134 Otherwise, it is found by adding the length of the first slot to the
3135 value @code{STARTING_FRAME_OFFSET}.
3136 @c i'm not sure if the above is still correct.. had to change it to get
3137 @c rid of an overfull. --mew 2feb93
3140 @defmac STACK_ALIGNMENT_NEEDED
3141 Define to zero to disable final alignment of the stack during reload.
3142 The nonzero default for this macro is suitable for most ports.
3144 On ports where @code{STARTING_FRAME_OFFSET} is nonzero or where there
3145 is a register save block following the local block that doesn't require
3146 alignment to @code{STACK_BOUNDARY}, it may be beneficial to disable
3147 stack alignment and do it in the backend.
3150 @defmac STACK_POINTER_OFFSET
3151 Offset from the stack pointer register to the first location at which
3152 outgoing arguments are placed. If not specified, the default value of
3153 zero is used. This is the proper value for most machines.
3155 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
3156 the first location at which outgoing arguments are placed.
3159 @defmac FIRST_PARM_OFFSET (@var{fundecl})
3160 Offset from the argument pointer register to the first argument's
3161 address. On some machines it may depend on the data type of the
3164 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
3165 the first argument's address.
3168 @defmac STACK_DYNAMIC_OFFSET (@var{fundecl})
3169 Offset from the stack pointer register to an item dynamically allocated
3170 on the stack, e.g., by @code{alloca}.
3172 The default value for this macro is @code{STACK_POINTER_OFFSET} plus the
3173 length of the outgoing arguments. The default is correct for most
3174 machines. See @file{function.c} for details.
3177 @defmac INITIAL_FRAME_ADDRESS_RTX
3178 A C expression whose value is RTL representing the address of the initial
3179 stack frame. This address is passed to @code{RETURN_ADDR_RTX} and
3180 @code{DYNAMIC_CHAIN_ADDRESS}. If you don't define this macro, a reasonable
3181 default value will be used. Define this macro in order to make frame pointer
3182 elimination work in the presence of @code{__builtin_frame_address (count)} and
3183 @code{__builtin_return_address (count)} for @code{count} not equal to zero.
3186 @defmac DYNAMIC_CHAIN_ADDRESS (@var{frameaddr})
3187 A C expression whose value is RTL representing the address in a stack
3188 frame where the pointer to the caller's frame is stored. Assume that
3189 @var{frameaddr} is an RTL expression for the address of the stack frame
3192 If you don't define this macro, the default is to return the value
3193 of @var{frameaddr}---that is, the stack frame address is also the
3194 address of the stack word that points to the previous frame.
3197 @defmac SETUP_FRAME_ADDRESSES
3198 If defined, a C expression that produces the machine-specific code to
3199 setup the stack so that arbitrary frames can be accessed. For example,
3200 on the SPARC, we must flush all of the register windows to the stack
3201 before we can access arbitrary stack frames. You will seldom need to
3205 @deftypefn {Target Hook} rtx TARGET_BUILTIN_SETJMP_FRAME_VALUE (void)
3206 This target hook should return an rtx that is used to store
3207 the address of the current frame into the built in @code{setjmp} buffer.
3208 The default value, @code{virtual_stack_vars_rtx}, is correct for most
3209 machines. One reason you may need to define this target hook is if
3210 @code{hard_frame_pointer_rtx} is the appropriate value on your machine.
3213 @defmac FRAME_ADDR_RTX (@var{frameaddr})
3214 A C expression whose value is RTL representing the value of the frame
3215 address for the current frame. @var{frameaddr} is the frame pointer
3216 of the current frame. This is used for __builtin_frame_address.
3217 You need only define this macro if the frame address is not the same
3218 as the frame pointer. Most machines do not need to define it.
3221 @defmac RETURN_ADDR_RTX (@var{count}, @var{frameaddr})
3222 A C expression whose value is RTL representing the value of the return
3223 address for the frame @var{count} steps up from the current frame, after
3224 the prologue. @var{frameaddr} is the frame pointer of the @var{count}
3225 frame, or the frame pointer of the @var{count} @minus{} 1 frame if
3226 @code{RETURN_ADDR_IN_PREVIOUS_FRAME} is defined.
3228 The value of the expression must always be the correct address when
3229 @var{count} is zero, but may be @code{NULL_RTX} if there is no way to
3230 determine the return address of other frames.
3233 @defmac RETURN_ADDR_IN_PREVIOUS_FRAME
3234 Define this if the return address of a particular stack frame is accessed
3235 from the frame pointer of the previous stack frame.
3238 @defmac INCOMING_RETURN_ADDR_RTX
3239 A C expression whose value is RTL representing the location of the
3240 incoming return address at the beginning of any function, before the
3241 prologue. This RTL is either a @code{REG}, indicating that the return
3242 value is saved in @samp{REG}, or a @code{MEM} representing a location in
3245 You only need to define this macro if you want to support call frame
3246 debugging information like that provided by DWARF 2.
3248 If this RTL is a @code{REG}, you should also define
3249 @code{DWARF_FRAME_RETURN_COLUMN} to @code{DWARF_FRAME_REGNUM (REGNO)}.
3252 @defmac DWARF_ALT_FRAME_RETURN_COLUMN
3253 A C expression whose value is an integer giving a DWARF 2 column
3254 number that may be used as an alternative return column. The column
3255 must not correspond to any gcc hard register (that is, it must not
3256 be in the range of @code{DWARF_FRAME_REGNUM}).
3258 This macro can be useful if @code{DWARF_FRAME_RETURN_COLUMN} is set to a
3259 general register, but an alternative column needs to be used for signal
3260 frames. Some targets have also used different frame return columns
3264 @defmac DWARF_ZERO_REG
3265 A C expression whose value is an integer giving a DWARF 2 register
3266 number that is considered to always have the value zero. This should
3267 only be defined if the target has an architected zero register, and
3268 someone decided it was a good idea to use that register number to
3269 terminate the stack backtrace. New ports should avoid this.
3272 @deftypefn {Target Hook} void TARGET_DWARF_HANDLE_FRAME_UNSPEC (const char *@var{label}, rtx @var{pattern}, int @var{index})
3273 This target hook allows the backend to emit frame-related insns that
3274 contain UNSPECs or UNSPEC_VOLATILEs. The DWARF 2 call frame debugging
3275 info engine will invoke it on insns of the form
3277 (set (reg) (unspec [@dots{}] UNSPEC_INDEX))
3281 (set (reg) (unspec_volatile [@dots{}] UNSPECV_INDEX)).
3283 to let the backend emit the call frame instructions. @var{label} is
3284 the CFI label attached to the insn, @var{pattern} is the pattern of
3285 the insn and @var{index} is @code{UNSPEC_INDEX} or @code{UNSPECV_INDEX}.
3288 @defmac INCOMING_FRAME_SP_OFFSET
3289 A C expression whose value is an integer giving the offset, in bytes,
3290 from the value of the stack pointer register to the top of the stack
3291 frame at the beginning of any function, before the prologue. The top of
3292 the frame is defined to be the value of the stack pointer in the
3293 previous frame, just before the call instruction.
3295 You only need to define this macro if you want to support call frame
3296 debugging information like that provided by DWARF 2.
3299 @defmac ARG_POINTER_CFA_OFFSET (@var{fundecl})
3300 A C expression whose value is an integer giving the offset, in bytes,
3301 from the argument pointer to the canonical frame address (cfa). The
3302 final value should coincide with that calculated by
3303 @code{INCOMING_FRAME_SP_OFFSET}. Which is unfortunately not usable
3304 during virtual register instantiation.
3306 The default value for this macro is
3307 @code{FIRST_PARM_OFFSET (fundecl) + crtl->args.pretend_args_size},
3308 which is correct for most machines; in general, the arguments are found
3309 immediately before the stack frame. Note that this is not the case on
3310 some targets that save registers into the caller's frame, such as SPARC
3311 and rs6000, and so such targets need to define this macro.
3313 You only need to define this macro if the default is incorrect, and you
3314 want to support call frame debugging information like that provided by
3318 @defmac FRAME_POINTER_CFA_OFFSET (@var{fundecl})
3319 If defined, a C expression whose value is an integer giving the offset
3320 in bytes from the frame pointer to the canonical frame address (cfa).
3321 The final value should coincide with that calculated by
3322 @code{INCOMING_FRAME_SP_OFFSET}.
3324 Normally the CFA is calculated as an offset from the argument pointer,
3325 via @code{ARG_POINTER_CFA_OFFSET}, but if the argument pointer is
3326 variable due to the ABI, this may not be possible. If this macro is
3327 defined, it implies that the virtual register instantiation should be
3328 based on the frame pointer instead of the argument pointer. Only one
3329 of @code{FRAME_POINTER_CFA_OFFSET} and @code{ARG_POINTER_CFA_OFFSET}
3333 @defmac CFA_FRAME_BASE_OFFSET (@var{fundecl})
3334 If defined, a C expression whose value is an integer giving the offset
3335 in bytes from the canonical frame address (cfa) to the frame base used
3336 in DWARF 2 debug information. The default is zero. A different value
3337 may reduce the size of debug information on some ports.
3340 @node Exception Handling
3341 @subsection Exception Handling Support
3342 @cindex exception handling
3344 @defmac EH_RETURN_DATA_REGNO (@var{N})
3345 A C expression whose value is the @var{N}th register number used for
3346 data by exception handlers, or @code{INVALID_REGNUM} if fewer than
3347 @var{N} registers are usable.
3349 The exception handling library routines communicate with the exception
3350 handlers via a set of agreed upon registers. Ideally these registers
3351 should be call-clobbered; it is possible to use call-saved registers,
3352 but may negatively impact code size. The target must support at least
3353 2 data registers, but should define 4 if there are enough free registers.
3355 You must define this macro if you want to support call frame exception
3356 handling like that provided by DWARF 2.
3359 @defmac EH_RETURN_STACKADJ_RTX
3360 A C expression whose value is RTL representing a location in which
3361 to store a stack adjustment to be applied before function return.
3362 This is used to unwind the stack to an exception handler's call frame.
3363 It will be assigned zero on code paths that return normally.
3365 Typically this is a call-clobbered hard register that is otherwise
3366 untouched by the epilogue, but could also be a stack slot.
3368 Do not define this macro if the stack pointer is saved and restored
3369 by the regular prolog and epilog code in the call frame itself; in
3370 this case, the exception handling library routines will update the
3371 stack location to be restored in place. Otherwise, you must define
3372 this macro if you want to support call frame exception handling like
3373 that provided by DWARF 2.
3376 @defmac EH_RETURN_HANDLER_RTX
3377 A C expression whose value is RTL representing a location in which
3378 to store the address of an exception handler to which we should
3379 return. It will not be assigned on code paths that return normally.
3381 Typically this is the location in the call frame at which the normal
3382 return address is stored. For targets that return by popping an
3383 address off the stack, this might be a memory address just below
3384 the @emph{target} call frame rather than inside the current call
3385 frame. If defined, @code{EH_RETURN_STACKADJ_RTX} will have already
3386 been assigned, so it may be used to calculate the location of the
3389 Some targets have more complex requirements than storing to an
3390 address calculable during initial code generation. In that case
3391 the @code{eh_return} instruction pattern should be used instead.
3393 If you want to support call frame exception handling, you must
3394 define either this macro or the @code{eh_return} instruction pattern.
3397 @defmac RETURN_ADDR_OFFSET
3398 If defined, an integer-valued C expression for which rtl will be generated
3399 to add it to the exception handler address before it is searched in the
3400 exception handling tables, and to subtract it again from the address before
3401 using it to return to the exception handler.
3404 @defmac ASM_PREFERRED_EH_DATA_FORMAT (@var{code}, @var{global})
3405 This macro chooses the encoding of pointers embedded in the exception
3406 handling sections. If at all possible, this should be defined such
3407 that the exception handling section will not require dynamic relocations,
3408 and so may be read-only.
3410 @var{code} is 0 for data, 1 for code labels, 2 for function pointers.
3411 @var{global} is true if the symbol may be affected by dynamic relocations.
3412 The macro should return a combination of the @code{DW_EH_PE_*} defines
3413 as found in @file{dwarf2.h}.
3415 If this macro is not defined, pointers will not be encoded but
3416 represented directly.
3419 @defmac ASM_MAYBE_OUTPUT_ENCODED_ADDR_RTX (@var{file}, @var{encoding}, @var{size}, @var{addr}, @var{done})
3420 This macro allows the target to emit whatever special magic is required
3421 to represent the encoding chosen by @code{ASM_PREFERRED_EH_DATA_FORMAT}.
3422 Generic code takes care of pc-relative and indirect encodings; this must
3423 be defined if the target uses text-relative or data-relative encodings.
3425 This is a C statement that branches to @var{done} if the format was
3426 handled. @var{encoding} is the format chosen, @var{size} is the number
3427 of bytes that the format occupies, @var{addr} is the @code{SYMBOL_REF}
3431 @defmac MD_UNWIND_SUPPORT
3432 A string specifying a file to be #include'd in unwind-dw2.c. The file
3433 so included typically defines @code{MD_FALLBACK_FRAME_STATE_FOR}.
3436 @defmac MD_FALLBACK_FRAME_STATE_FOR (@var{context}, @var{fs})
3437 This macro allows the target to add CPU and operating system specific
3438 code to the call-frame unwinder for use when there is no unwind data
3439 available. The most common reason to implement this macro is to unwind
3440 through signal frames.
3442 This macro is called from @code{uw_frame_state_for} in
3443 @file{unwind-dw2.c}, @file{unwind-dw2-xtensa.c} and
3444 @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context};
3445 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{context->ra}
3446 for the address of the code being executed and @code{context->cfa} for
3447 the stack pointer value. If the frame can be decoded, the register
3448 save addresses should be updated in @var{fs} and the macro should
3449 evaluate to @code{_URC_NO_REASON}. If the frame cannot be decoded,
3450 the macro should evaluate to @code{_URC_END_OF_STACK}.
3452 For proper signal handling in Java this macro is accompanied by
3453 @code{MAKE_THROW_FRAME}, defined in @file{libjava/include/*-signal.h} headers.
3456 @defmac MD_HANDLE_UNWABI (@var{context}, @var{fs})
3457 This macro allows the target to add operating system specific code to the
3458 call-frame unwinder to handle the IA-64 @code{.unwabi} unwinding directive,
3459 usually used for signal or interrupt frames.
3461 This macro is called from @code{uw_update_context} in @file{unwind-ia64.c}.
3462 @var{context} is an @code{_Unwind_Context};
3463 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{fs->unwabi}
3464 for the abi and context in the @code{.unwabi} directive. If the
3465 @code{.unwabi} directive can be handled, the register save addresses should
3466 be updated in @var{fs}.
3469 @defmac TARGET_USES_WEAK_UNWIND_INFO
3470 A C expression that evaluates to true if the target requires unwind
3471 info to be given comdat linkage. Define it to be @code{1} if comdat
3472 linkage is necessary. The default is @code{0}.
3475 @node Stack Checking
3476 @subsection Specifying How Stack Checking is Done
3478 GCC will check that stack references are within the boundaries of the
3479 stack, if the option @option{-fstack-check} is specified, in one of
3484 If the value of the @code{STACK_CHECK_BUILTIN} macro is nonzero, GCC
3485 will assume that you have arranged for full stack checking to be done
3486 at appropriate places in the configuration files. GCC will not do
3487 other special processing.
3490 If @code{STACK_CHECK_BUILTIN} is zero and the value of the
3491 @code{STACK_CHECK_STATIC_BUILTIN} macro is nonzero, GCC will assume
3492 that you have arranged for static stack checking (checking of the
3493 static stack frame of functions) to be done at appropriate places
3494 in the configuration files. GCC will only emit code to do dynamic
3495 stack checking (checking on dynamic stack allocations) using the third
3499 If neither of the above are true, GCC will generate code to periodically
3500 ``probe'' the stack pointer using the values of the macros defined below.
3503 If neither STACK_CHECK_BUILTIN nor STACK_CHECK_STATIC_BUILTIN is defined,
3504 GCC will change its allocation strategy for large objects if the option
3505 @option{-fstack-check} is specified: they will always be allocated
3506 dynamically if their size exceeds @code{STACK_CHECK_MAX_VAR_SIZE} bytes.
3508 @defmac STACK_CHECK_BUILTIN
3509 A nonzero value if stack checking is done by the configuration files in a
3510 machine-dependent manner. You should define this macro if stack checking
3511 is required by the ABI of your machine or if you would like to do stack
3512 checking in some more efficient way than the generic approach. The default
3513 value of this macro is zero.
3516 @defmac STACK_CHECK_STATIC_BUILTIN
3517 A nonzero value if static stack checking is done by the configuration files
3518 in a machine-dependent manner. You should define this macro if you would
3519 like to do static stack checking in some more efficient way than the generic
3520 approach. The default value of this macro is zero.
3523 @defmac STACK_CHECK_PROBE_INTERVAL_EXP
3524 An integer specifying the interval at which GCC must generate stack probe
3525 instructions, defined as 2 raised to this integer. You will normally
3526 define this macro so that the interval be no larger than the size of
3527 the ``guard pages'' at the end of a stack area. The default value
3528 of 12 (4096-byte interval) is suitable for most systems.
3531 @defmac STACK_CHECK_MOVING_SP
3532 An integer which is nonzero if GCC should move the stack pointer page by page
3533 when doing probes. This can be necessary on systems where the stack pointer
3534 contains the bottom address of the memory area accessible to the executing
3535 thread at any point in time. In this situation an alternate signal stack
3536 is required in order to be able to recover from a stack overflow. The
3537 default value of this macro is zero.
3540 @defmac STACK_CHECK_PROTECT
3541 The number of bytes of stack needed to recover from a stack overflow, for
3542 languages where such a recovery is supported. The default value of 75 words
3543 with the @code{setjmp}/@code{longjmp}-based exception handling mechanism and
3544 8192 bytes with other exception handling mechanisms should be adequate for
3548 The following macros are relevant only if neither STACK_CHECK_BUILTIN
3549 nor STACK_CHECK_STATIC_BUILTIN is defined; you can omit them altogether
3550 in the opposite case.
3552 @defmac STACK_CHECK_MAX_FRAME_SIZE
3553 The maximum size of a stack frame, in bytes. GCC will generate probe
3554 instructions in non-leaf functions to ensure at least this many bytes of
3555 stack are available. If a stack frame is larger than this size, stack
3556 checking will not be reliable and GCC will issue a warning. The
3557 default is chosen so that GCC only generates one instruction on most
3558 systems. You should normally not change the default value of this macro.
3561 @defmac STACK_CHECK_FIXED_FRAME_SIZE
3562 GCC uses this value to generate the above warning message. It
3563 represents the amount of fixed frame used by a function, not including
3564 space for any callee-saved registers, temporaries and user variables.
3565 You need only specify an upper bound for this amount and will normally
3566 use the default of four words.
3569 @defmac STACK_CHECK_MAX_VAR_SIZE
3570 The maximum size, in bytes, of an object that GCC will place in the
3571 fixed area of the stack frame when the user specifies
3572 @option{-fstack-check}.
3573 GCC computed the default from the values of the above macros and you will
3574 normally not need to override that default.
3578 @node Frame Registers
3579 @subsection Registers That Address the Stack Frame
3581 @c prevent bad page break with this line
3582 This discusses registers that address the stack frame.
3584 @defmac STACK_POINTER_REGNUM
3585 The register number of the stack pointer register, which must also be a
3586 fixed register according to @code{FIXED_REGISTERS}. On most machines,
3587 the hardware determines which register this is.
3590 @defmac FRAME_POINTER_REGNUM
3591 The register number of the frame pointer register, which is used to
3592 access automatic variables in the stack frame. On some machines, the
3593 hardware determines which register this is. On other machines, you can
3594 choose any register you wish for this purpose.
3597 @defmac HARD_FRAME_POINTER_REGNUM
3598 On some machines the offset between the frame pointer and starting
3599 offset of the automatic variables is not known until after register
3600 allocation has been done (for example, because the saved registers are
3601 between these two locations). On those machines, define
3602 @code{FRAME_POINTER_REGNUM} the number of a special, fixed register to
3603 be used internally until the offset is known, and define
3604 @code{HARD_FRAME_POINTER_REGNUM} to be the actual hard register number
3605 used for the frame pointer.
3607 You should define this macro only in the very rare circumstances when it
3608 is not possible to calculate the offset between the frame pointer and
3609 the automatic variables until after register allocation has been
3610 completed. When this macro is defined, you must also indicate in your
3611 definition of @code{ELIMINABLE_REGS} how to eliminate
3612 @code{FRAME_POINTER_REGNUM} into either @code{HARD_FRAME_POINTER_REGNUM}
3613 or @code{STACK_POINTER_REGNUM}.
3615 Do not define this macro if it would be the same as
3616 @code{FRAME_POINTER_REGNUM}.
3619 @defmac ARG_POINTER_REGNUM
3620 The register number of the arg pointer register, which is used to access
3621 the function's argument list. On some machines, this is the same as the
3622 frame pointer register. On some machines, the hardware determines which
3623 register this is. On other machines, you can choose any register you
3624 wish for this purpose. If this is not the same register as the frame
3625 pointer register, then you must mark it as a fixed register according to
3626 @code{FIXED_REGISTERS}, or arrange to be able to eliminate it
3627 (@pxref{Elimination}).
3630 @defmac RETURN_ADDRESS_POINTER_REGNUM
3631 The register number of the return address pointer register, which is used to
3632 access the current function's return address from the stack. On some
3633 machines, the return address is not at a fixed offset from the frame
3634 pointer or stack pointer or argument pointer. This register can be defined
3635 to point to the return address on the stack, and then be converted by
3636 @code{ELIMINABLE_REGS} into either the frame pointer or stack pointer.
3638 Do not define this macro unless there is no other way to get the return
3639 address from the stack.
3642 @defmac STATIC_CHAIN_REGNUM
3643 @defmacx STATIC_CHAIN_INCOMING_REGNUM
3644 Register numbers used for passing a function's static chain pointer. If
3645 register windows are used, the register number as seen by the called
3646 function is @code{STATIC_CHAIN_INCOMING_REGNUM}, while the register
3647 number as seen by the calling function is @code{STATIC_CHAIN_REGNUM}. If
3648 these registers are the same, @code{STATIC_CHAIN_INCOMING_REGNUM} need
3651 The static chain register need not be a fixed register.
3653 If the static chain is passed in memory, these macros should not be
3654 defined; instead, the @code{TARGET_STATIC_CHAIN} hook should be used.
3657 @deftypefn {Target Hook} rtx TARGET_STATIC_CHAIN (const_tree @var{fndecl}, bool @var{incoming_p})
3658 This hook replaces the use of @code{STATIC_CHAIN_REGNUM} et al for
3659 targets that may use different static chain locations for different
3660 nested functions. This may be required if the target has function
3661 attributes that affect the calling conventions of the function and
3662 those calling conventions use different static chain locations.
3664 The default version of this hook uses @code{STATIC_CHAIN_REGNUM} et al.
3666 If the static chain is passed in memory, this hook should be used to
3667 provide rtx giving @code{mem} expressions that denote where they are stored.
3668 Often the @code{mem} expression as seen by the caller will be at an offset
3669 from the stack pointer and the @code{mem} expression as seen by the callee
3670 will be at an offset from the frame pointer.
3671 @findex stack_pointer_rtx
3672 @findex frame_pointer_rtx
3673 @findex arg_pointer_rtx
3674 The variables @code{stack_pointer_rtx}, @code{frame_pointer_rtx}, and
3675 @code{arg_pointer_rtx} will have been initialized and should be used
3676 to refer to those items.
3679 @defmac DWARF_FRAME_REGISTERS
3680 This macro specifies the maximum number of hard registers that can be
3681 saved in a call frame. This is used to size data structures used in
3682 DWARF2 exception handling.
3684 Prior to GCC 3.0, this macro was needed in order to establish a stable
3685 exception handling ABI in the face of adding new hard registers for ISA
3686 extensions. In GCC 3.0 and later, the EH ABI is insulated from changes
3687 in the number of hard registers. Nevertheless, this macro can still be
3688 used to reduce the runtime memory requirements of the exception handling
3689 routines, which can be substantial if the ISA contains a lot of
3690 registers that are not call-saved.
3692 If this macro is not defined, it defaults to
3693 @code{FIRST_PSEUDO_REGISTER}.
3696 @defmac PRE_GCC3_DWARF_FRAME_REGISTERS
3698 This macro is similar to @code{DWARF_FRAME_REGISTERS}, but is provided
3699 for backward compatibility in pre GCC 3.0 compiled code.
3701 If this macro is not defined, it defaults to
3702 @code{DWARF_FRAME_REGISTERS}.
3705 @defmac DWARF_REG_TO_UNWIND_COLUMN (@var{regno})
3707 Define this macro if the target's representation for dwarf registers
3708 is different than the internal representation for unwind column.
3709 Given a dwarf register, this macro should return the internal unwind
3710 column number to use instead.
3712 See the PowerPC's SPE target for an example.
3715 @defmac DWARF_FRAME_REGNUM (@var{regno})
3717 Define this macro if the target's representation for dwarf registers
3718 used in .eh_frame or .debug_frame is different from that used in other
3719 debug info sections. Given a GCC hard register number, this macro
3720 should return the .eh_frame register number. The default is
3721 @code{DBX_REGISTER_NUMBER (@var{regno})}.
3725 @defmac DWARF2_FRAME_REG_OUT (@var{regno}, @var{for_eh})
3727 Define this macro to map register numbers held in the call frame info
3728 that GCC has collected using @code{DWARF_FRAME_REGNUM} to those that
3729 should be output in .debug_frame (@code{@var{for_eh}} is zero) and
3730 .eh_frame (@code{@var{for_eh}} is nonzero). The default is to
3731 return @code{@var{regno}}.
3736 @subsection Eliminating Frame Pointer and Arg Pointer
3738 @c prevent bad page break with this line
3739 This is about eliminating the frame pointer and arg pointer.
3741 @deftypefn {Target Hook} bool TARGET_FRAME_POINTER_REQUIRED (void)
3742 This target hook should return @code{true} if a function must have and use
3743 a frame pointer. This target hook is called in the reload pass. If its return
3744 value is @code{true} the function will have a frame pointer.
3746 This target hook can in principle examine the current function and decide
3747 according to the facts, but on most machines the constant @code{false} or the
3748 constant @code{true} suffices. Use @code{false} when the machine allows code
3749 to be generated with no frame pointer, and doing so saves some time or space.
3750 Use @code{true} when there is no possible advantage to avoiding a frame
3753 In certain cases, the compiler does not know how to produce valid code
3754 without a frame pointer. The compiler recognizes those cases and
3755 automatically gives the function a frame pointer regardless of what
3756 @code{TARGET_FRAME_POINTER_REQUIRED} returns. You don't need to worry about
3759 In a function that does not require a frame pointer, the frame pointer
3760 register can be allocated for ordinary usage, unless you mark it as a
3761 fixed register. See @code{FIXED_REGISTERS} for more information.
3763 Default return value is @code{false}.
3766 @findex get_frame_size
3767 @defmac INITIAL_FRAME_POINTER_OFFSET (@var{depth-var})
3768 A C statement to store in the variable @var{depth-var} the difference
3769 between the frame pointer and the stack pointer values immediately after
3770 the function prologue. The value would be computed from information
3771 such as the result of @code{get_frame_size ()} and the tables of
3772 registers @code{regs_ever_live} and @code{call_used_regs}.
3774 If @code{ELIMINABLE_REGS} is defined, this macro will be not be used and
3775 need not be defined. Otherwise, it must be defined even if
3776 @code{TARGET_FRAME_POINTER_REQUIRED} always returns true; in that
3777 case, you may set @var{depth-var} to anything.
3780 @defmac ELIMINABLE_REGS
3781 If defined, this macro specifies a table of register pairs used to
3782 eliminate unneeded registers that point into the stack frame. If it is not
3783 defined, the only elimination attempted by the compiler is to replace
3784 references to the frame pointer with references to the stack pointer.
3786 The definition of this macro is a list of structure initializations, each
3787 of which specifies an original and replacement register.
3789 On some machines, the position of the argument pointer is not known until
3790 the compilation is completed. In such a case, a separate hard register
3791 must be used for the argument pointer. This register can be eliminated by
3792 replacing it with either the frame pointer or the argument pointer,
3793 depending on whether or not the frame pointer has been eliminated.
3795 In this case, you might specify:
3797 #define ELIMINABLE_REGS \
3798 @{@{ARG_POINTER_REGNUM, STACK_POINTER_REGNUM@}, \
3799 @{ARG_POINTER_REGNUM, FRAME_POINTER_REGNUM@}, \
3800 @{FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM@}@}
3803 Note that the elimination of the argument pointer with the stack pointer is
3804 specified first since that is the preferred elimination.
3807 @deftypefn {Target Hook} bool TARGET_CAN_ELIMINATE (const int @var{from_reg}, const int @var{to_reg})
3808 This target hook should returns @code{true} if the compiler is allowed to
3809 try to replace register number @var{from_reg} with register number
3810 @var{to_reg}. This target hook need only be defined if @code{ELIMINABLE_REGS}
3811 is defined, and will usually be @code{true}, since most of the cases
3812 preventing register elimination are things that the compiler already
3815 Default return value is @code{true}.
3818 @defmac INITIAL_ELIMINATION_OFFSET (@var{from-reg}, @var{to-reg}, @var{offset-var})
3819 This macro is similar to @code{INITIAL_FRAME_POINTER_OFFSET}. It
3820 specifies the initial difference between the specified pair of
3821 registers. This macro must be defined if @code{ELIMINABLE_REGS} is
3825 @node Stack Arguments
3826 @subsection Passing Function Arguments on the Stack
3827 @cindex arguments on stack
3828 @cindex stack arguments
3830 The macros in this section control how arguments are passed
3831 on the stack. See the following section for other macros that
3832 control passing certain arguments in registers.
3834 @deftypefn {Target Hook} bool TARGET_PROMOTE_PROTOTYPES (const_tree @var{fntype})
3835 This target hook returns @code{true} if an argument declared in a
3836 prototype as an integral type smaller than @code{int} should actually be
3837 passed as an @code{int}. In addition to avoiding errors in certain
3838 cases of mismatch, it also makes for better code on certain machines.
3839 The default is to not promote prototypes.
3843 A C expression. If nonzero, push insns will be used to pass
3845 If the target machine does not have a push instruction, set it to zero.
3846 That directs GCC to use an alternate strategy: to
3847 allocate the entire argument block and then store the arguments into
3848 it. When @code{PUSH_ARGS} is nonzero, @code{PUSH_ROUNDING} must be defined too.
3851 @defmac PUSH_ARGS_REVERSED
3852 A C expression. If nonzero, function arguments will be evaluated from
3853 last to first, rather than from first to last. If this macro is not
3854 defined, it defaults to @code{PUSH_ARGS} on targets where the stack
3855 and args grow in opposite directions, and 0 otherwise.
3858 @defmac PUSH_ROUNDING (@var{npushed})
3859 A C expression that is the number of bytes actually pushed onto the
3860 stack when an instruction attempts to push @var{npushed} bytes.
3862 On some machines, the definition
3865 #define PUSH_ROUNDING(BYTES) (BYTES)
3869 will suffice. But on other machines, instructions that appear
3870 to push one byte actually push two bytes in an attempt to maintain
3871 alignment. Then the definition should be
3874 #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1)
3878 @findex current_function_outgoing_args_size
3879 @defmac ACCUMULATE_OUTGOING_ARGS
3880 A C expression. If nonzero, the maximum amount of space required for outgoing arguments
3881 will be computed and placed into the variable
3882 @code{current_function_outgoing_args_size}. No space will be pushed
3883 onto the stack for each call; instead, the function prologue should
3884 increase the stack frame size by this amount.
3886 Setting both @code{PUSH_ARGS} and @code{ACCUMULATE_OUTGOING_ARGS}
3890 @defmac REG_PARM_STACK_SPACE (@var{fndecl})
3891 Define this macro if functions should assume that stack space has been
3892 allocated for arguments even when their values are passed in
3895 The value of this macro is the size, in bytes, of the area reserved for
3896 arguments passed in registers for the function represented by @var{fndecl},
3897 which can be zero if GCC is calling a library function.
3898 The argument @var{fndecl} can be the FUNCTION_DECL, or the type itself
3901 This space can be allocated by the caller, or be a part of the
3902 machine-dependent stack frame: @code{OUTGOING_REG_PARM_STACK_SPACE} says
3905 @c above is overfull. not sure what to do. --mew 5feb93 did
3906 @c something, not sure if it looks good. --mew 10feb93
3908 @defmac OUTGOING_REG_PARM_STACK_SPACE (@var{fntype})
3909 Define this to a nonzero value if it is the responsibility of the
3910 caller to allocate the area reserved for arguments passed in registers
3911 when calling a function of @var{fntype}. @var{fntype} may be NULL
3912 if the function called is a library function.
3914 If @code{ACCUMULATE_OUTGOING_ARGS} is defined, this macro controls
3915 whether the space for these arguments counts in the value of
3916 @code{current_function_outgoing_args_size}.
3919 @defmac STACK_PARMS_IN_REG_PARM_AREA
3920 Define this macro if @code{REG_PARM_STACK_SPACE} is defined, but the
3921 stack parameters don't skip the area specified by it.
3922 @c i changed this, makes more sens and it should have taken care of the
3923 @c overfull.. not as specific, tho. --mew 5feb93
3925 Normally, when a parameter is not passed in registers, it is placed on the
3926 stack beyond the @code{REG_PARM_STACK_SPACE} area. Defining this macro
3927 suppresses this behavior and causes the parameter to be passed on the
3928 stack in its natural location.
3931 @defmac RETURN_POPS_ARGS (@var{fundecl}, @var{funtype}, @var{stack-size})
3932 A C expression that should indicate the number of bytes of its own
3933 arguments that a function pops on returning, or 0 if the
3934 function pops no arguments and the caller must therefore pop them all
3935 after the function returns.
3937 @var{fundecl} is a C variable whose value is a tree node that describes
3938 the function in question. Normally it is a node of type
3939 @code{FUNCTION_DECL} that describes the declaration of the function.
3940 From this you can obtain the @code{DECL_ATTRIBUTES} of the function.
3942 @var{funtype} is a C variable whose value is a tree node that
3943 describes the function in question. Normally it is a node of type
3944 @code{FUNCTION_TYPE} that describes the data type of the function.
3945 From this it is possible to obtain the data types of the value and
3946 arguments (if known).
3948 When a call to a library function is being considered, @var{fundecl}
3949 will contain an identifier node for the library function. Thus, if
3950 you need to distinguish among various library functions, you can do so
3951 by their names. Note that ``library function'' in this context means
3952 a function used to perform arithmetic, whose name is known specially
3953 in the compiler and was not mentioned in the C code being compiled.
3955 @var{stack-size} is the number of bytes of arguments passed on the
3956 stack. If a variable number of bytes is passed, it is zero, and
3957 argument popping will always be the responsibility of the calling function.
3959 On the VAX, all functions always pop their arguments, so the definition
3960 of this macro is @var{stack-size}. On the 68000, using the standard
3961 calling convention, no functions pop their arguments, so the value of
3962 the macro is always 0 in this case. But an alternative calling
3963 convention is available in which functions that take a fixed number of
3964 arguments pop them but other functions (such as @code{printf}) pop
3965 nothing (the caller pops all). When this convention is in use,
3966 @var{funtype} is examined to determine whether a function takes a fixed
3967 number of arguments.
3970 @defmac CALL_POPS_ARGS (@var{cum})
3971 A C expression that should indicate the number of bytes a call sequence
3972 pops off the stack. It is added to the value of @code{RETURN_POPS_ARGS}
3973 when compiling a function call.
3975 @var{cum} is the variable in which all arguments to the called function
3976 have been accumulated.
3978 On certain architectures, such as the SH5, a call trampoline is used
3979 that pops certain registers off the stack, depending on the arguments
3980 that have been passed to the function. Since this is a property of the
3981 call site, not of the called function, @code{RETURN_POPS_ARGS} is not
3985 @node Register Arguments
3986 @subsection Passing Arguments in Registers
3987 @cindex arguments in registers
3988 @cindex registers arguments
3990 This section describes the macros which let you control how various
3991 types of arguments are passed in registers or how they are arranged in
3994 @defmac FUNCTION_ARG (@var{cum}, @var{mode}, @var{type}, @var{named})
3995 A C expression that controls whether a function argument is passed
3996 in a register, and which register.
3998 The arguments are @var{cum}, which summarizes all the previous
3999 arguments; @var{mode}, the machine mode of the argument; @var{type},
4000 the data type of the argument as a tree node or 0 if that is not known
4001 (which happens for C support library functions); and @var{named},
4002 which is 1 for an ordinary argument and 0 for nameless arguments that
4003 correspond to @samp{@dots{}} in the called function's prototype.
4004 @var{type} can be an incomplete type if a syntax error has previously
4007 The value of the expression is usually either a @code{reg} RTX for the
4008 hard register in which to pass the argument, or zero to pass the
4009 argument on the stack.
4011 For machines like the VAX and 68000, where normally all arguments are
4012 pushed, zero suffices as a definition.
4014 The value of the expression can also be a @code{parallel} RTX@. This is
4015 used when an argument is passed in multiple locations. The mode of the
4016 @code{parallel} should be the mode of the entire argument. The
4017 @code{parallel} holds any number of @code{expr_list} pairs; each one
4018 describes where part of the argument is passed. In each
4019 @code{expr_list} the first operand must be a @code{reg} RTX for the hard
4020 register in which to pass this part of the argument, and the mode of the
4021 register RTX indicates how large this part of the argument is. The
4022 second operand of the @code{expr_list} is a @code{const_int} which gives
4023 the offset in bytes into the entire argument of where this part starts.
4024 As a special exception the first @code{expr_list} in the @code{parallel}
4025 RTX may have a first operand of zero. This indicates that the entire
4026 argument is also stored on the stack.
4028 The last time this macro is called, it is called with @code{MODE ==
4029 VOIDmode}, and its result is passed to the @code{call} or @code{call_value}
4030 pattern as operands 2 and 3 respectively.
4032 @cindex @file{stdarg.h} and register arguments
4033 The usual way to make the ISO library @file{stdarg.h} work on a machine
4034 where some arguments are usually passed in registers, is to cause
4035 nameless arguments to be passed on the stack instead. This is done
4036 by making @code{FUNCTION_ARG} return 0 whenever @var{named} is 0.
4038 @cindex @code{TARGET_MUST_PASS_IN_STACK}, and @code{FUNCTION_ARG}
4039 @cindex @code{REG_PARM_STACK_SPACE}, and @code{FUNCTION_ARG}
4040 You may use the hook @code{targetm.calls.must_pass_in_stack}
4041 in the definition of this macro to determine if this argument is of a
4042 type that must be passed in the stack. If @code{REG_PARM_STACK_SPACE}
4043 is not defined and @code{FUNCTION_ARG} returns nonzero for such an
4044 argument, the compiler will abort. If @code{REG_PARM_STACK_SPACE} is
4045 defined, the argument will be computed in the stack and then loaded into
4049 @deftypefn {Target Hook} bool TARGET_MUST_PASS_IN_STACK (enum machine_mode @var{mode}, const_tree @var{type})
4050 This target hook should return @code{true} if we should not pass @var{type}
4051 solely in registers. The file @file{expr.h} defines a
4052 definition that is usually appropriate, refer to @file{expr.h} for additional
4056 @defmac FUNCTION_INCOMING_ARG (@var{cum}, @var{mode}, @var{type}, @var{named})
4057 Define this macro if the target machine has ``register windows'', so
4058 that the register in which a function sees an arguments is not
4059 necessarily the same as the one in which the caller passed the
4062 For such machines, @code{FUNCTION_ARG} computes the register in which
4063 the caller passes the value, and @code{FUNCTION_INCOMING_ARG} should
4064 be defined in a similar fashion to tell the function being called
4065 where the arguments will arrive.
4067 If @code{FUNCTION_INCOMING_ARG} is not defined, @code{FUNCTION_ARG}
4068 serves both purposes.
4071 @deftypefn {Target Hook} int TARGET_ARG_PARTIAL_BYTES (CUMULATIVE_ARGS *@var{cum}, enum machine_mode @var{mode}, tree @var{type}, bool @var{named})
4072 This target hook returns the number of bytes at the beginning of an
4073 argument that must be put in registers. The value must be zero for
4074 arguments that are passed entirely in registers or that are entirely
4075 pushed on the stack.
4077 On some machines, certain arguments must be passed partially in
4078 registers and partially in memory. On these machines, typically the
4079 first few words of arguments are passed in registers, and the rest
4080 on the stack. If a multi-word argument (a @code{double} or a
4081 structure) crosses that boundary, its first few words must be passed
4082 in registers and the rest must be pushed. This macro tells the
4083 compiler when this occurs, and how many bytes should go in registers.
4085 @code{FUNCTION_ARG} for these arguments should return the first
4086 register to be used by the caller for this argument; likewise
4087 @code{FUNCTION_INCOMING_ARG}, for the called function.
4090 @deftypefn {Target Hook} bool TARGET_PASS_BY_REFERENCE (CUMULATIVE_ARGS *@var{cum}, enum machine_mode @var{mode}, tree @var{type}, bool @var{named})
4091 This target hook should return @code{true} if an argument at the
4092 position indicated by @var{cum} should be passed by reference. This
4093 predicate is queried after target independent reasons for being
4094 passed by reference, such as @code{TREE_ADDRESSABLE (type)}.
4096 If the hook returns true, a copy of that argument is made in memory and a
4097 pointer to the argument is passed instead of the argument itself.
4098 The pointer is passed in whatever way is appropriate for passing a pointer
4102 @deftypefn {Target Hook} bool TARGET_CALLEE_COPIES (CUMULATIVE_ARGS *@var{cum}, enum machine_mode @var{mode}, const_tree @var{type}, bool @var{named})
4103 The function argument described by the parameters to this hook is
4104 known to be passed by reference. The hook should return true if the
4105 function argument should be copied by the callee instead of copied
4108 For any argument for which the hook returns true, if it can be
4109 determined that the argument is not modified, then a copy need
4112 The default version of this hook always returns false.
4115 @defmac CUMULATIVE_ARGS
4116 A C type for declaring a variable that is used as the first argument of
4117 @code{FUNCTION_ARG} and other related values. For some target machines,
4118 the type @code{int} suffices and can hold the number of bytes of
4121 There is no need to record in @code{CUMULATIVE_ARGS} anything about the
4122 arguments that have been passed on the stack. The compiler has other
4123 variables to keep track of that. For target machines on which all
4124 arguments are passed on the stack, there is no need to store anything in
4125 @code{CUMULATIVE_ARGS}; however, the data structure must exist and
4126 should not be empty, so use @code{int}.
4129 @defmac OVERRIDE_ABI_FORMAT (@var{fndecl})
4130 If defined, this macro is called before generating any code for a
4131 function, but after the @var{cfun} descriptor for the function has been
4132 created. The back end may use this macro to update @var{cfun} to
4133 reflect an ABI other than that which would normally be used by default.
4134 If the compiler is generating code for a compiler-generated function,
4135 @var{fndecl} may be @code{NULL}.
4138 @defmac INIT_CUMULATIVE_ARGS (@var{cum}, @var{fntype}, @var{libname}, @var{fndecl}, @var{n_named_args})
4139 A C statement (sans semicolon) for initializing the variable
4140 @var{cum} for the state at the beginning of the argument list. The
4141 variable has type @code{CUMULATIVE_ARGS}. The value of @var{fntype}
4142 is the tree node for the data type of the function which will receive
4143 the args, or 0 if the args are to a compiler support library function.
4144 For direct calls that are not libcalls, @var{fndecl} contain the
4145 declaration node of the function. @var{fndecl} is also set when
4146 @code{INIT_CUMULATIVE_ARGS} is used to find arguments for the function
4147 being compiled. @var{n_named_args} is set to the number of named
4148 arguments, including a structure return address if it is passed as a
4149 parameter, when making a call. When processing incoming arguments,
4150 @var{n_named_args} is set to @minus{}1.
4152 When processing a call to a compiler support library function,
4153 @var{libname} identifies which one. It is a @code{symbol_ref} rtx which
4154 contains the name of the function, as a string. @var{libname} is 0 when
4155 an ordinary C function call is being processed. Thus, each time this
4156 macro is called, either @var{libname} or @var{fntype} is nonzero, but
4157 never both of them at once.
4160 @defmac INIT_CUMULATIVE_LIBCALL_ARGS (@var{cum}, @var{mode}, @var{libname})
4161 Like @code{INIT_CUMULATIVE_ARGS} but only used for outgoing libcalls,
4162 it gets a @code{MODE} argument instead of @var{fntype}, that would be
4163 @code{NULL}. @var{indirect} would always be zero, too. If this macro
4164 is not defined, @code{INIT_CUMULATIVE_ARGS (cum, NULL_RTX, libname,
4165 0)} is used instead.
4168 @defmac INIT_CUMULATIVE_INCOMING_ARGS (@var{cum}, @var{fntype}, @var{libname})
4169 Like @code{INIT_CUMULATIVE_ARGS} but overrides it for the purposes of
4170 finding the arguments for the function being compiled. If this macro is
4171 undefined, @code{INIT_CUMULATIVE_ARGS} is used instead.
4173 The value passed for @var{libname} is always 0, since library routines
4174 with special calling conventions are never compiled with GCC@. The
4175 argument @var{libname} exists for symmetry with
4176 @code{INIT_CUMULATIVE_ARGS}.
4177 @c could use "this macro" in place of @code{INIT_CUMULATIVE_ARGS}, maybe.
4178 @c --mew 5feb93 i switched the order of the sentences. --mew 10feb93
4181 @defmac FUNCTION_ARG_ADVANCE (@var{cum}, @var{mode}, @var{type}, @var{named})
4182 A C statement (sans semicolon) to update the summarizer variable
4183 @var{cum} to advance past an argument in the argument list. The
4184 values @var{mode}, @var{type} and @var{named} describe that argument.
4185 Once this is done, the variable @var{cum} is suitable for analyzing
4186 the @emph{following} argument with @code{FUNCTION_ARG}, etc.
4188 This macro need not do anything if the argument in question was passed
4189 on the stack. The compiler knows how to track the amount of stack space
4190 used for arguments without any special help.
4193 @defmac FUNCTION_ARG_OFFSET (@var{mode}, @var{type})
4194 If defined, a C expression that is the number of bytes to add to the
4195 offset of the argument passed in memory. This is needed for the SPU,
4196 which passes @code{char} and @code{short} arguments in the preferred
4197 slot that is in the middle of the quad word instead of starting at the
4201 @defmac FUNCTION_ARG_PADDING (@var{mode}, @var{type})
4202 If defined, a C expression which determines whether, and in which direction,
4203 to pad out an argument with extra space. The value should be of type
4204 @code{enum direction}: either @code{upward} to pad above the argument,
4205 @code{downward} to pad below, or @code{none} to inhibit padding.
4207 The @emph{amount} of padding is always just enough to reach the next
4208 multiple of @code{FUNCTION_ARG_BOUNDARY}; this macro does not control
4211 This macro has a default definition which is right for most systems.
4212 For little-endian machines, the default is to pad upward. For
4213 big-endian machines, the default is to pad downward for an argument of
4214 constant size shorter than an @code{int}, and upward otherwise.
4217 @defmac PAD_VARARGS_DOWN
4218 If defined, a C expression which determines whether the default
4219 implementation of va_arg will attempt to pad down before reading the
4220 next argument, if that argument is smaller than its aligned space as
4221 controlled by @code{PARM_BOUNDARY}. If this macro is not defined, all such
4222 arguments are padded down if @code{BYTES_BIG_ENDIAN} is true.
4225 @defmac BLOCK_REG_PADDING (@var{mode}, @var{type}, @var{first})
4226 Specify padding for the last element of a block move between registers and
4227 memory. @var{first} is nonzero if this is the only element. Defining this
4228 macro allows better control of register function parameters on big-endian
4229 machines, without using @code{PARALLEL} rtl. In particular,
4230 @code{MUST_PASS_IN_STACK} need not test padding and mode of types in
4231 registers, as there is no longer a "wrong" part of a register; For example,
4232 a three byte aggregate may be passed in the high part of a register if so
4236 @defmac FUNCTION_ARG_BOUNDARY (@var{mode}, @var{type})
4237 If defined, a C expression that gives the alignment boundary, in bits,
4238 of an argument with the specified mode and type. If it is not defined,
4239 @code{PARM_BOUNDARY} is used for all arguments.
4242 @defmac FUNCTION_ARG_REGNO_P (@var{regno})
4243 A C expression that is nonzero if @var{regno} is the number of a hard
4244 register in which function arguments are sometimes passed. This does
4245 @emph{not} include implicit arguments such as the static chain and
4246 the structure-value address. On many machines, no registers can be
4247 used for this purpose since all function arguments are pushed on the
4251 @deftypefn {Target Hook} bool TARGET_SPLIT_COMPLEX_ARG (const_tree @var{type})
4252 This hook should return true if parameter of type @var{type} are passed
4253 as two scalar parameters. By default, GCC will attempt to pack complex
4254 arguments into the target's word size. Some ABIs require complex arguments
4255 to be split and treated as their individual components. For example, on
4256 AIX64, complex floats should be passed in a pair of floating point
4257 registers, even though a complex float would fit in one 64-bit floating
4260 The default value of this hook is @code{NULL}, which is treated as always
4264 @deftypefn {Target Hook} tree TARGET_BUILD_BUILTIN_VA_LIST (void)
4265 This hook returns a type node for @code{va_list} for the target.
4266 The default version of the hook returns @code{void*}.
4269 @deftypefn {Target Hook} int TARGET_ENUM_VA_LIST_P (int @var{idx}, const char ** @var{pname}, tree @var{ptype})
4270 This target hook is used in function @code{c_common_nodes_and_builtins}
4271 to iterate through the target specific builtin types for va_list. The
4272 variable @var{idx} is used as iterator. @var{pname} has to be a pointer
4273 to a @code{const char *} and @var{ptype} a pointer to a @code{tree} typed
4275 The arguments @var{pname} and @var{ptype} are used to store the result of
4276 this macro and are set to the name of the va_list builtin type and its
4278 If the return value of this macro is zero, then there is no more element.
4279 Otherwise the @var{IDX} should be increased for the next call of this
4280 macro to iterate through all types.
4283 @deftypefn {Target Hook} tree TARGET_FN_ABI_VA_LIST (tree @var{fndecl})
4284 This hook returns the va_list type of the calling convention specified by
4286 The default version of this hook returns @code{va_list_type_node}.
4289 @deftypefn {Target Hook} tree TARGET_CANONICAL_VA_LIST_TYPE (tree @var{type})
4290 This hook returns the va_list type of the calling convention specified by the
4291 type of @var{type}. If @var{type} is not a valid va_list type, it returns
4295 @deftypefn {Target Hook} tree TARGET_GIMPLIFY_VA_ARG_EXPR (tree @var{valist}, tree @var{type}, gimple_seq *@var{pre_p}, gimple_seq *@var{post_p})
4296 This hook performs target-specific gimplification of
4297 @code{VA_ARG_EXPR}. The first two parameters correspond to the
4298 arguments to @code{va_arg}; the latter two are as in
4299 @code{gimplify.c:gimplify_expr}.
4302 @deftypefn {Target Hook} bool TARGET_VALID_POINTER_MODE (enum machine_mode @var{mode})
4303 Define this to return nonzero if the port can handle pointers
4304 with machine mode @var{mode}. The default version of this
4305 hook returns true for both @code{ptr_mode} and @code{Pmode}.
4308 @deftypefn {Target Hook} bool TARGET_SCALAR_MODE_SUPPORTED_P (enum machine_mode @var{mode})
4309 Define this to return nonzero if the port is prepared to handle
4310 insns involving scalar mode @var{mode}. For a scalar mode to be
4311 considered supported, all the basic arithmetic and comparisons
4314 The default version of this hook returns true for any mode
4315 required to handle the basic C types (as defined by the port).
4316 Included here are the double-word arithmetic supported by the
4317 code in @file{optabs.c}.
4320 @deftypefn {Target Hook} bool TARGET_VECTOR_MODE_SUPPORTED_P (enum machine_mode @var{mode})
4321 Define this to return nonzero if the port is prepared to handle
4322 insns involving vector mode @var{mode}. At the very least, it
4323 must have move patterns for this mode.
4326 @deftypefn {Target Hook} bool TARGET_SMALL_REGISTER_CLASSES_FOR_MODE_P (enum machine_mode @var{mode})
4327 Define this to return nonzero for machine modes for which the port has
4328 small register classes. If this target hook returns nonzero for a given
4329 @var{mode}, the compiler will try to minimize the lifetime of registers
4330 in @var{mode}. The hook may be called with @code{VOIDmode} as argument.
4331 In this case, the hook is expected to return nonzero if it returns nonzero
4334 On some machines, it is risky to let hard registers live across arbitrary
4335 insns. Typically, these machines have instructions that require values
4336 to be in specific registers (like an accumulator), and reload will fail
4337 if the required hard register is used for another purpose across such an
4340 Passes before reload do not know which hard registers will be used
4341 in an instruction, but the machine modes of the registers set or used in
4342 the instruction are already known. And for some machines, register
4343 classes are small for, say, integer registers but not for floating point
4344 registers. For example, the AMD x86-64 architecture requires specific
4345 registers for the legacy x86 integer instructions, but there are many
4346 SSE registers for floating point operations. On such targets, a good
4347 strategy may be to return nonzero from this hook for @code{INTEGRAL_MODE_P}
4348 machine modes but zero for the SSE register classes.
4350 The default version of this hook retuns false for any mode. It is always
4351 safe to redefine this hook to return with a nonzero value. But if you
4352 unnecessarily define it, you will reduce the amount of optimizations
4353 that can be performed in some cases. If you do not define this hook
4354 to return a nonzero value when it is required, the compiler will run out
4355 of spill registers and print a fatal error message.
4359 @subsection How Scalar Function Values Are Returned
4360 @cindex return values in registers
4361 @cindex values, returned by functions
4362 @cindex scalars, returned as values
4364 This section discusses the macros that control returning scalars as
4365 values---values that can fit in registers.
4367 @deftypefn {Target Hook} rtx TARGET_FUNCTION_VALUE (const_tree @var{ret_type}, const_tree @var{fn_decl_or_type}, bool @var{outgoing})
4369 Define this to return an RTX representing the place where a function
4370 returns or receives a value of data type @var{ret_type}, a tree node
4371 representing a data type. @var{fn_decl_or_type} is a tree node
4372 representing @code{FUNCTION_DECL} or @code{FUNCTION_TYPE} of a
4373 function being called. If @var{outgoing} is false, the hook should
4374 compute the register in which the caller will see the return value.
4375 Otherwise, the hook should return an RTX representing the place where
4376 a function returns a value.
4378 On many machines, only @code{TYPE_MODE (@var{ret_type})} is relevant.
4379 (Actually, on most machines, scalar values are returned in the same
4380 place regardless of mode.) The value of the expression is usually a
4381 @code{reg} RTX for the hard register where the return value is stored.
4382 The value can also be a @code{parallel} RTX, if the return value is in
4383 multiple places. See @code{FUNCTION_ARG} for an explanation of the
4384 @code{parallel} form. Note that the callee will populate every
4385 location specified in the @code{parallel}, but if the first element of
4386 the @code{parallel} contains the whole return value, callers will use
4387 that element as the canonical location and ignore the others. The m68k
4388 port uses this type of @code{parallel} to return pointers in both
4389 @samp{%a0} (the canonical location) and @samp{%d0}.
4391 If @code{TARGET_PROMOTE_FUNCTION_RETURN} returns true, you must apply
4392 the same promotion rules specified in @code{PROMOTE_MODE} if
4393 @var{valtype} is a scalar type.
4395 If the precise function being called is known, @var{func} is a tree
4396 node (@code{FUNCTION_DECL}) for it; otherwise, @var{func} is a null
4397 pointer. This makes it possible to use a different value-returning
4398 convention for specific functions when all their calls are
4401 Some target machines have ``register windows'' so that the register in
4402 which a function returns its value is not the same as the one in which
4403 the caller sees the value. For such machines, you should return
4404 different RTX depending on @var{outgoing}.
4406 @code{TARGET_FUNCTION_VALUE} is not used for return values with
4407 aggregate data types, because these are returned in another way. See
4408 @code{TARGET_STRUCT_VALUE_RTX} and related macros, below.
4411 @defmac FUNCTION_VALUE (@var{valtype}, @var{func})
4412 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE} for
4413 a new target instead.
4416 @defmac LIBCALL_VALUE (@var{mode})
4417 A C expression to create an RTX representing the place where a library
4418 function returns a value of mode @var{mode}.
4420 Note that ``library function'' in this context means a compiler
4421 support routine, used to perform arithmetic, whose name is known
4422 specially by the compiler and was not mentioned in the C code being
4426 @deftypefn {Target Hook} rtx TARGET_LIBCALL_VALUE (enum machine_mode
4427 @var{mode}, const_rtx @var{fun})
4428 Define this hook if the back-end needs to know the name of the libcall
4429 function in order to determine where the result should be returned.
4431 The mode of the result is given by @var{mode} and the name of the called
4432 library function is given by @var{fun}. The hook should return an RTX
4433 representing the place where the library function result will be returned.
4435 If this hook is not defined, then LIBCALL_VALUE will be used.
4438 @defmac FUNCTION_VALUE_REGNO_P (@var{regno})
4439 A C expression that is nonzero if @var{regno} is the number of a hard
4440 register in which the values of called function may come back.
4442 A register whose use for returning values is limited to serving as the
4443 second of a pair (for a value of type @code{double}, say) need not be
4444 recognized by this macro. So for most machines, this definition
4448 #define FUNCTION_VALUE_REGNO_P(N) ((N) == 0)
4451 If the machine has register windows, so that the caller and the called
4452 function use different registers for the return value, this macro
4453 should recognize only the caller's register numbers.
4455 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE_REGNO_P}
4456 for a new target instead.
4459 @deftypefn {Target Hook} bool TARGET_FUNCTION_VALUE_REGNO_P (const unsigned int @var{regno})
4460 A target hook that return @code{true} if @var{regno} is the number of a hard
4461 register in which the values of called function may come back.
4463 A register whose use for returning values is limited to serving as the
4464 second of a pair (for a value of type @code{double}, say) need not be
4465 recognized by this target hook.
4467 If the machine has register windows, so that the caller and the called
4468 function use different registers for the return value, this target hook
4469 should recognize only the caller's register numbers.
4471 If this hook is not defined, then FUNCTION_VALUE_REGNO_P will be used.
4474 @defmac APPLY_RESULT_SIZE
4475 Define this macro if @samp{untyped_call} and @samp{untyped_return}
4476 need more space than is implied by @code{FUNCTION_VALUE_REGNO_P} for
4477 saving and restoring an arbitrary return value.
4480 @deftypefn {Target Hook} bool TARGET_RETURN_IN_MSB (const_tree @var{type})
4481 This hook should return true if values of type @var{type} are returned
4482 at the most significant end of a register (in other words, if they are
4483 padded at the least significant end). You can assume that @var{type}
4484 is returned in a register; the caller is required to check this.
4486 Note that the register provided by @code{TARGET_FUNCTION_VALUE} must
4487 be able to hold the complete return value. For example, if a 1-, 2-
4488 or 3-byte structure is returned at the most significant end of a
4489 4-byte register, @code{TARGET_FUNCTION_VALUE} should provide an
4493 @node Aggregate Return
4494 @subsection How Large Values Are Returned
4495 @cindex aggregates as return values
4496 @cindex large return values
4497 @cindex returning aggregate values
4498 @cindex structure value address
4500 When a function value's mode is @code{BLKmode} (and in some other
4501 cases), the value is not returned according to
4502 @code{TARGET_FUNCTION_VALUE} (@pxref{Scalar Return}). Instead, the
4503 caller passes the address of a block of memory in which the value
4504 should be stored. This address is called the @dfn{structure value
4507 This section describes how to control returning structure values in
4510 @deftypefn {Target Hook} bool TARGET_RETURN_IN_MEMORY (const_tree @var{type}, const_tree @var{fntype})
4511 This target hook should return a nonzero value to say to return the
4512 function value in memory, just as large structures are always returned.
4513 Here @var{type} will be the data type of the value, and @var{fntype}
4514 will be the type of the function doing the returning, or @code{NULL} for
4517 Note that values of mode @code{BLKmode} must be explicitly handled
4518 by this function. Also, the option @option{-fpcc-struct-return}
4519 takes effect regardless of this macro. On most systems, it is
4520 possible to leave the hook undefined; this causes a default
4521 definition to be used, whose value is the constant 1 for @code{BLKmode}
4522 values, and 0 otherwise.
4524 Do not use this hook to indicate that structures and unions should always
4525 be returned in memory. You should instead use @code{DEFAULT_PCC_STRUCT_RETURN}
4529 @defmac DEFAULT_PCC_STRUCT_RETURN
4530 Define this macro to be 1 if all structure and union return values must be
4531 in memory. Since this results in slower code, this should be defined
4532 only if needed for compatibility with other compilers or with an ABI@.
4533 If you define this macro to be 0, then the conventions used for structure
4534 and union return values are decided by the @code{TARGET_RETURN_IN_MEMORY}
4537 If not defined, this defaults to the value 1.
4540 @deftypefn {Target Hook} rtx TARGET_STRUCT_VALUE_RTX (tree @var{fndecl}, int @var{incoming})
4541 This target hook should return the location of the structure value
4542 address (normally a @code{mem} or @code{reg}), or 0 if the address is
4543 passed as an ``invisible'' first argument. Note that @var{fndecl} may
4544 be @code{NULL}, for libcalls. You do not need to define this target
4545 hook if the address is always passed as an ``invisible'' first
4548 On some architectures the place where the structure value address
4549 is found by the called function is not the same place that the
4550 caller put it. This can be due to register windows, or it could
4551 be because the function prologue moves it to a different place.
4552 @var{incoming} is @code{1} or @code{2} when the location is needed in
4553 the context of the called function, and @code{0} in the context of
4556 If @var{incoming} is nonzero and the address is to be found on the
4557 stack, return a @code{mem} which refers to the frame pointer. If
4558 @var{incoming} is @code{2}, the result is being used to fetch the
4559 structure value address at the beginning of a function. If you need
4560 to emit adjusting code, you should do it at this point.
4563 @defmac PCC_STATIC_STRUCT_RETURN
4564 Define this macro if the usual system convention on the target machine
4565 for returning structures and unions is for the called function to return
4566 the address of a static variable containing the value.
4568 Do not define this if the usual system convention is for the caller to
4569 pass an address to the subroutine.
4571 This macro has effect in @option{-fpcc-struct-return} mode, but it does
4572 nothing when you use @option{-freg-struct-return} mode.
4576 @subsection Caller-Saves Register Allocation
4578 If you enable it, GCC can save registers around function calls. This
4579 makes it possible to use call-clobbered registers to hold variables that
4580 must live across calls.
4582 @defmac CALLER_SAVE_PROFITABLE (@var{refs}, @var{calls})
4583 A C expression to determine whether it is worthwhile to consider placing
4584 a pseudo-register in a call-clobbered hard register and saving and
4585 restoring it around each function call. The expression should be 1 when
4586 this is worth doing, and 0 otherwise.
4588 If you don't define this macro, a default is used which is good on most
4589 machines: @code{4 * @var{calls} < @var{refs}}.
4592 @defmac HARD_REGNO_CALLER_SAVE_MODE (@var{regno}, @var{nregs})
4593 A C expression specifying which mode is required for saving @var{nregs}
4594 of a pseudo-register in call-clobbered hard register @var{regno}. If
4595 @var{regno} is unsuitable for caller save, @code{VOIDmode} should be
4596 returned. For most machines this macro need not be defined since GCC
4597 will select the smallest suitable mode.
4600 @node Function Entry
4601 @subsection Function Entry and Exit
4602 @cindex function entry and exit
4606 This section describes the macros that output function entry
4607 (@dfn{prologue}) and exit (@dfn{epilogue}) code.
4609 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_PROLOGUE (FILE *@var{file}, HOST_WIDE_INT @var{size})
4610 If defined, a function that outputs the assembler code for entry to a
4611 function. The prologue is responsible for setting up the stack frame,
4612 initializing the frame pointer register, saving registers that must be
4613 saved, and allocating @var{size} additional bytes of storage for the
4614 local variables. @var{size} is an integer. @var{file} is a stdio
4615 stream to which the assembler code should be output.
4617 The label for the beginning of the function need not be output by this
4618 macro. That has already been done when the macro is run.
4620 @findex regs_ever_live
4621 To determine which registers to save, the macro can refer to the array
4622 @code{regs_ever_live}: element @var{r} is nonzero if hard register
4623 @var{r} is used anywhere within the function. This implies the function
4624 prologue should save register @var{r}, provided it is not one of the
4625 call-used registers. (@code{TARGET_ASM_FUNCTION_EPILOGUE} must likewise use
4626 @code{regs_ever_live}.)
4628 On machines that have ``register windows'', the function entry code does
4629 not save on the stack the registers that are in the windows, even if
4630 they are supposed to be preserved by function calls; instead it takes
4631 appropriate steps to ``push'' the register stack, if any non-call-used
4632 registers are used in the function.
4634 @findex frame_pointer_needed
4635 On machines where functions may or may not have frame-pointers, the
4636 function entry code must vary accordingly; it must set up the frame
4637 pointer if one is wanted, and not otherwise. To determine whether a
4638 frame pointer is in wanted, the macro can refer to the variable
4639 @code{frame_pointer_needed}. The variable's value will be 1 at run
4640 time in a function that needs a frame pointer. @xref{Elimination}.
4642 The function entry code is responsible for allocating any stack space
4643 required for the function. This stack space consists of the regions
4644 listed below. In most cases, these regions are allocated in the
4645 order listed, with the last listed region closest to the top of the
4646 stack (the lowest address if @code{STACK_GROWS_DOWNWARD} is defined, and
4647 the highest address if it is not defined). You can use a different order
4648 for a machine if doing so is more convenient or required for
4649 compatibility reasons. Except in cases where required by standard
4650 or by a debugger, there is no reason why the stack layout used by GCC
4651 need agree with that used by other compilers for a machine.
4654 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_END_PROLOGUE (FILE *@var{file})
4655 If defined, a function that outputs assembler code at the end of a
4656 prologue. This should be used when the function prologue is being
4657 emitted as RTL, and you have some extra assembler that needs to be
4658 emitted. @xref{prologue instruction pattern}.
4661 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_BEGIN_EPILOGUE (FILE *@var{file})
4662 If defined, a function that outputs assembler code at the start of an
4663 epilogue. This should be used when the function epilogue is being
4664 emitted as RTL, and you have some extra assembler that needs to be
4665 emitted. @xref{epilogue instruction pattern}.
4668 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_EPILOGUE (FILE *@var{file}, HOST_WIDE_INT @var{size})
4669 If defined, a function that outputs the assembler code for exit from a
4670 function. The epilogue is responsible for restoring the saved
4671 registers and stack pointer to their values when the function was
4672 called, and returning control to the caller. This macro takes the
4673 same arguments as the macro @code{TARGET_ASM_FUNCTION_PROLOGUE}, and the
4674 registers to restore are determined from @code{regs_ever_live} and
4675 @code{CALL_USED_REGISTERS} in the same way.
4677 On some machines, there is a single instruction that does all the work
4678 of returning from the function. On these machines, give that
4679 instruction the name @samp{return} and do not define the macro
4680 @code{TARGET_ASM_FUNCTION_EPILOGUE} at all.
4682 Do not define a pattern named @samp{return} if you want the
4683 @code{TARGET_ASM_FUNCTION_EPILOGUE} to be used. If you want the target
4684 switches to control whether return instructions or epilogues are used,
4685 define a @samp{return} pattern with a validity condition that tests the
4686 target switches appropriately. If the @samp{return} pattern's validity
4687 condition is false, epilogues will be used.
4689 On machines where functions may or may not have frame-pointers, the
4690 function exit code must vary accordingly. Sometimes the code for these
4691 two cases is completely different. To determine whether a frame pointer
4692 is wanted, the macro can refer to the variable
4693 @code{frame_pointer_needed}. The variable's value will be 1 when compiling
4694 a function that needs a frame pointer.
4696 Normally, @code{TARGET_ASM_FUNCTION_PROLOGUE} and
4697 @code{TARGET_ASM_FUNCTION_EPILOGUE} must treat leaf functions specially.
4698 The C variable @code{current_function_is_leaf} is nonzero for such a
4699 function. @xref{Leaf Functions}.
4701 On some machines, some functions pop their arguments on exit while
4702 others leave that for the caller to do. For example, the 68020 when
4703 given @option{-mrtd} pops arguments in functions that take a fixed
4704 number of arguments.
4706 @findex current_function_pops_args
4707 Your definition of the macro @code{RETURN_POPS_ARGS} decides which
4708 functions pop their own arguments. @code{TARGET_ASM_FUNCTION_EPILOGUE}
4709 needs to know what was decided. The number of bytes of the current
4710 function's arguments that this function should pop is available in
4711 @code{crtl->args.pops_args}. @xref{Scalar Return}.
4716 @findex current_function_pretend_args_size
4717 A region of @code{current_function_pretend_args_size} bytes of
4718 uninitialized space just underneath the first argument arriving on the
4719 stack. (This may not be at the very start of the allocated stack region
4720 if the calling sequence has pushed anything else since pushing the stack
4721 arguments. But usually, on such machines, nothing else has been pushed
4722 yet, because the function prologue itself does all the pushing.) This
4723 region is used on machines where an argument may be passed partly in
4724 registers and partly in memory, and, in some cases to support the
4725 features in @code{<stdarg.h>}.
4728 An area of memory used to save certain registers used by the function.
4729 The size of this area, which may also include space for such things as
4730 the return address and pointers to previous stack frames, is
4731 machine-specific and usually depends on which registers have been used
4732 in the function. Machines with register windows often do not require
4736 A region of at least @var{size} bytes, possibly rounded up to an allocation
4737 boundary, to contain the local variables of the function. On some machines,
4738 this region and the save area may occur in the opposite order, with the
4739 save area closer to the top of the stack.
4742 @cindex @code{ACCUMULATE_OUTGOING_ARGS} and stack frames
4743 Optionally, when @code{ACCUMULATE_OUTGOING_ARGS} is defined, a region of
4744 @code{current_function_outgoing_args_size} bytes to be used for outgoing
4745 argument lists of the function. @xref{Stack Arguments}.
4748 @defmac EXIT_IGNORE_STACK
4749 Define this macro as a C expression that is nonzero if the return
4750 instruction or the function epilogue ignores the value of the stack
4751 pointer; in other words, if it is safe to delete an instruction to
4752 adjust the stack pointer before a return from the function. The
4755 Note that this macro's value is relevant only for functions for which
4756 frame pointers are maintained. It is never safe to delete a final
4757 stack adjustment in a function that has no frame pointer, and the
4758 compiler knows this regardless of @code{EXIT_IGNORE_STACK}.
4761 @defmac EPILOGUE_USES (@var{regno})
4762 Define this macro as a C expression that is nonzero for registers that are
4763 used by the epilogue or the @samp{return} pattern. The stack and frame
4764 pointer registers are already assumed to be used as needed.
4767 @defmac EH_USES (@var{regno})
4768 Define this macro as a C expression that is nonzero for registers that are
4769 used by the exception handling mechanism, and so should be considered live
4770 on entry to an exception edge.
4773 @defmac DELAY_SLOTS_FOR_EPILOGUE
4774 Define this macro if the function epilogue contains delay slots to which
4775 instructions from the rest of the function can be ``moved''. The
4776 definition should be a C expression whose value is an integer
4777 representing the number of delay slots there.
4780 @defmac ELIGIBLE_FOR_EPILOGUE_DELAY (@var{insn}, @var{n})
4781 A C expression that returns 1 if @var{insn} can be placed in delay
4782 slot number @var{n} of the epilogue.
4784 The argument @var{n} is an integer which identifies the delay slot now
4785 being considered (since different slots may have different rules of
4786 eligibility). It is never negative and is always less than the number
4787 of epilogue delay slots (what @code{DELAY_SLOTS_FOR_EPILOGUE} returns).
4788 If you reject a particular insn for a given delay slot, in principle, it
4789 may be reconsidered for a subsequent delay slot. Also, other insns may
4790 (at least in principle) be considered for the so far unfilled delay
4793 @findex current_function_epilogue_delay_list
4794 @findex final_scan_insn
4795 The insns accepted to fill the epilogue delay slots are put in an RTL
4796 list made with @code{insn_list} objects, stored in the variable
4797 @code{current_function_epilogue_delay_list}. The insn for the first
4798 delay slot comes first in the list. Your definition of the macro
4799 @code{TARGET_ASM_FUNCTION_EPILOGUE} should fill the delay slots by
4800 outputting the insns in this list, usually by calling
4801 @code{final_scan_insn}.
4803 You need not define this macro if you did not define
4804 @code{DELAY_SLOTS_FOR_EPILOGUE}.
4807 @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})
4808 A function that outputs the assembler code for a thunk
4809 function, used to implement C++ virtual function calls with multiple
4810 inheritance. The thunk acts as a wrapper around a virtual function,
4811 adjusting the implicit object parameter before handing control off to
4814 First, emit code to add the integer @var{delta} to the location that
4815 contains the incoming first argument. Assume that this argument
4816 contains a pointer, and is the one used to pass the @code{this} pointer
4817 in C++. This is the incoming argument @emph{before} the function prologue,
4818 e.g.@: @samp{%o0} on a sparc. The addition must preserve the values of
4819 all other incoming arguments.
4821 Then, if @var{vcall_offset} is nonzero, an additional adjustment should be
4822 made after adding @code{delta}. In particular, if @var{p} is the
4823 adjusted pointer, the following adjustment should be made:
4826 p += (*((ptrdiff_t **)p))[vcall_offset/sizeof(ptrdiff_t)]
4829 After the additions, emit code to jump to @var{function}, which is a
4830 @code{FUNCTION_DECL}. This is a direct pure jump, not a call, and does
4831 not touch the return address. Hence returning from @var{FUNCTION} will
4832 return to whoever called the current @samp{thunk}.
4834 The effect must be as if @var{function} had been called directly with
4835 the adjusted first argument. This macro is responsible for emitting all
4836 of the code for a thunk function; @code{TARGET_ASM_FUNCTION_PROLOGUE}
4837 and @code{TARGET_ASM_FUNCTION_EPILOGUE} are not invoked.
4839 The @var{thunk_fndecl} is redundant. (@var{delta} and @var{function}
4840 have already been extracted from it.) It might possibly be useful on
4841 some targets, but probably not.
4843 If you do not define this macro, the target-independent code in the C++
4844 front end will generate a less efficient heavyweight thunk that calls
4845 @var{function} instead of jumping to it. The generic approach does
4846 not support varargs.
4849 @deftypefn {Target Hook} bool TARGET_ASM_CAN_OUTPUT_MI_THUNK (const_tree @var{thunk_fndecl}, HOST_WIDE_INT @var{delta}, HOST_WIDE_INT @var{vcall_offset}, const_tree @var{function})
4850 A function that returns true if TARGET_ASM_OUTPUT_MI_THUNK would be able
4851 to output the assembler code for the thunk function specified by the
4852 arguments it is passed, and false otherwise. In the latter case, the
4853 generic approach will be used by the C++ front end, with the limitations
4858 @subsection Generating Code for Profiling
4859 @cindex profiling, code generation
4861 These macros will help you generate code for profiling.
4863 @defmac FUNCTION_PROFILER (@var{file}, @var{labelno})
4864 A C statement or compound statement to output to @var{file} some
4865 assembler code to call the profiling subroutine @code{mcount}.
4868 The details of how @code{mcount} expects to be called are determined by
4869 your operating system environment, not by GCC@. To figure them out,
4870 compile a small program for profiling using the system's installed C
4871 compiler and look at the assembler code that results.
4873 Older implementations of @code{mcount} expect the address of a counter
4874 variable to be loaded into some register. The name of this variable is
4875 @samp{LP} followed by the number @var{labelno}, so you would generate
4876 the name using @samp{LP%d} in a @code{fprintf}.
4879 @defmac PROFILE_HOOK
4880 A C statement or compound statement to output to @var{file} some assembly
4881 code to call the profiling subroutine @code{mcount} even the target does
4882 not support profiling.
4885 @defmac NO_PROFILE_COUNTERS
4886 Define this macro to be an expression with a nonzero value if the
4887 @code{mcount} subroutine on your system does not need a counter variable
4888 allocated for each function. This is true for almost all modern
4889 implementations. If you define this macro, you must not use the
4890 @var{labelno} argument to @code{FUNCTION_PROFILER}.
4893 @defmac PROFILE_BEFORE_PROLOGUE
4894 Define this macro if the code for function profiling should come before
4895 the function prologue. Normally, the profiling code comes after.
4899 @subsection Permitting tail calls
4902 @deftypefn {Target Hook} bool TARGET_FUNCTION_OK_FOR_SIBCALL (tree @var{decl}, tree @var{exp})
4903 True if it is ok to do sibling call optimization for the specified
4904 call expression @var{exp}. @var{decl} will be the called function,
4905 or @code{NULL} if this is an indirect call.
4907 It is not uncommon for limitations of calling conventions to prevent
4908 tail calls to functions outside the current unit of translation, or
4909 during PIC compilation. The hook is used to enforce these restrictions,
4910 as the @code{sibcall} md pattern can not fail, or fall over to a
4911 ``normal'' call. The criteria for successful sibling call optimization
4912 may vary greatly between different architectures.
4915 @deftypefn {Target Hook} void TARGET_EXTRA_LIVE_ON_ENTRY (bitmap @var{regs})
4916 Add any hard registers to @var{regs} that are live on entry to the
4917 function. This hook only needs to be defined to provide registers that
4918 cannot be found by examination of FUNCTION_ARG_REGNO_P, the callee saved
4919 registers, STATIC_CHAIN_INCOMING_REGNUM, STATIC_CHAIN_REGNUM,
4920 TARGET_STRUCT_VALUE_RTX, FRAME_POINTER_REGNUM, EH_USES,
4921 FRAME_POINTER_REGNUM, ARG_POINTER_REGNUM, and the PIC_OFFSET_TABLE_REGNUM.
4924 @node Stack Smashing Protection
4925 @subsection Stack smashing protection
4926 @cindex stack smashing protection
4928 @deftypefn {Target Hook} tree TARGET_STACK_PROTECT_GUARD (void)
4929 This hook returns a @code{DECL} node for the external variable to use
4930 for the stack protection guard. This variable is initialized by the
4931 runtime to some random value and is used to initialize the guard value
4932 that is placed at the top of the local stack frame. The type of this
4933 variable must be @code{ptr_type_node}.
4935 The default version of this hook creates a variable called
4936 @samp{__stack_chk_guard}, which is normally defined in @file{libgcc2.c}.
4939 @deftypefn {Target Hook} tree TARGET_STACK_PROTECT_FAIL (void)
4940 This hook returns a tree expression that alerts the runtime that the
4941 stack protect guard variable has been modified. This expression should
4942 involve a call to a @code{noreturn} function.
4944 The default version of this hook invokes a function called
4945 @samp{__stack_chk_fail}, taking no arguments. This function is
4946 normally defined in @file{libgcc2.c}.
4950 @section Implementing the Varargs Macros
4951 @cindex varargs implementation
4953 GCC comes with an implementation of @code{<varargs.h>} and
4954 @code{<stdarg.h>} that work without change on machines that pass arguments
4955 on the stack. Other machines require their own implementations of
4956 varargs, and the two machine independent header files must have
4957 conditionals to include it.
4959 ISO @code{<stdarg.h>} differs from traditional @code{<varargs.h>} mainly in
4960 the calling convention for @code{va_start}. The traditional
4961 implementation takes just one argument, which is the variable in which
4962 to store the argument pointer. The ISO implementation of
4963 @code{va_start} takes an additional second argument. The user is
4964 supposed to write the last named argument of the function here.
4966 However, @code{va_start} should not use this argument. The way to find
4967 the end of the named arguments is with the built-in functions described
4970 @defmac __builtin_saveregs ()
4971 Use this built-in function to save the argument registers in memory so
4972 that the varargs mechanism can access them. Both ISO and traditional
4973 versions of @code{va_start} must use @code{__builtin_saveregs}, unless
4974 you use @code{TARGET_SETUP_INCOMING_VARARGS} (see below) instead.
4976 On some machines, @code{__builtin_saveregs} is open-coded under the
4977 control of the target hook @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. On
4978 other machines, it calls a routine written in assembler language,
4979 found in @file{libgcc2.c}.
4981 Code generated for the call to @code{__builtin_saveregs} appears at the
4982 beginning of the function, as opposed to where the call to
4983 @code{__builtin_saveregs} is written, regardless of what the code is.
4984 This is because the registers must be saved before the function starts
4985 to use them for its own purposes.
4986 @c i rewrote the first sentence above to fix an overfull hbox. --mew
4990 @defmac __builtin_args_info (@var{category})
4991 Use this built-in function to find the first anonymous arguments in
4994 In general, a machine may have several categories of registers used for
4995 arguments, each for a particular category of data types. (For example,
4996 on some machines, floating-point registers are used for floating-point
4997 arguments while other arguments are passed in the general registers.)
4998 To make non-varargs functions use the proper calling convention, you
4999 have defined the @code{CUMULATIVE_ARGS} data type to record how many
5000 registers in each category have been used so far
5002 @code{__builtin_args_info} accesses the same data structure of type
5003 @code{CUMULATIVE_ARGS} after the ordinary argument layout is finished
5004 with it, with @var{category} specifying which word to access. Thus, the
5005 value indicates the first unused register in a given category.
5007 Normally, you would use @code{__builtin_args_info} in the implementation
5008 of @code{va_start}, accessing each category just once and storing the
5009 value in the @code{va_list} object. This is because @code{va_list} will
5010 have to update the values, and there is no way to alter the
5011 values accessed by @code{__builtin_args_info}.
5014 @defmac __builtin_next_arg (@var{lastarg})
5015 This is the equivalent of @code{__builtin_args_info}, for stack
5016 arguments. It returns the address of the first anonymous stack
5017 argument, as type @code{void *}. If @code{ARGS_GROW_DOWNWARD}, it
5018 returns the address of the location above the first anonymous stack
5019 argument. Use it in @code{va_start} to initialize the pointer for
5020 fetching arguments from the stack. Also use it in @code{va_start} to
5021 verify that the second parameter @var{lastarg} is the last named argument
5022 of the current function.
5025 @defmac __builtin_classify_type (@var{object})
5026 Since each machine has its own conventions for which data types are
5027 passed in which kind of register, your implementation of @code{va_arg}
5028 has to embody these conventions. The easiest way to categorize the
5029 specified data type is to use @code{__builtin_classify_type} together
5030 with @code{sizeof} and @code{__alignof__}.
5032 @code{__builtin_classify_type} ignores the value of @var{object},
5033 considering only its data type. It returns an integer describing what
5034 kind of type that is---integer, floating, pointer, structure, and so on.
5036 The file @file{typeclass.h} defines an enumeration that you can use to
5037 interpret the values of @code{__builtin_classify_type}.
5040 These machine description macros help implement varargs:
5042 @deftypefn {Target Hook} rtx TARGET_EXPAND_BUILTIN_SAVEREGS (void)
5043 If defined, this hook produces the machine-specific code for a call to
5044 @code{__builtin_saveregs}. This code will be moved to the very
5045 beginning of the function, before any parameter access are made. The
5046 return value of this function should be an RTX that contains the value
5047 to use as the return of @code{__builtin_saveregs}.
5050 @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})
5051 This target hook offers an alternative to using
5052 @code{__builtin_saveregs} and defining the hook
5053 @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. Use it to store the anonymous
5054 register arguments into the stack so that all the arguments appear to
5055 have been passed consecutively on the stack. Once this is done, you can
5056 use the standard implementation of varargs that works for machines that
5057 pass all their arguments on the stack.
5059 The argument @var{args_so_far} points to the @code{CUMULATIVE_ARGS} data
5060 structure, containing the values that are obtained after processing the
5061 named arguments. The arguments @var{mode} and @var{type} describe the
5062 last named argument---its machine mode and its data type as a tree node.
5064 The target hook should do two things: first, push onto the stack all the
5065 argument registers @emph{not} used for the named arguments, and second,
5066 store the size of the data thus pushed into the @code{int}-valued
5067 variable pointed to by @var{pretend_args_size}. The value that you
5068 store here will serve as additional offset for setting up the stack
5071 Because you must generate code to push the anonymous arguments at
5072 compile time without knowing their data types,
5073 @code{TARGET_SETUP_INCOMING_VARARGS} is only useful on machines that
5074 have just a single category of argument register and use it uniformly
5077 If the argument @var{second_time} is nonzero, it means that the
5078 arguments of the function are being analyzed for the second time. This
5079 happens for an inline function, which is not actually compiled until the
5080 end of the source file. The hook @code{TARGET_SETUP_INCOMING_VARARGS} should
5081 not generate any instructions in this case.
5084 @deftypefn {Target Hook} bool TARGET_STRICT_ARGUMENT_NAMING (CUMULATIVE_ARGS *@var{ca})
5085 Define this hook to return @code{true} if the location where a function
5086 argument is passed depends on whether or not it is a named argument.
5088 This hook controls how the @var{named} argument to @code{FUNCTION_ARG}
5089 is set for varargs and stdarg functions. If this hook returns
5090 @code{true}, the @var{named} argument is always true for named
5091 arguments, and false for unnamed arguments. If it returns @code{false},
5092 but @code{TARGET_PRETEND_OUTGOING_VARARGS_NAMED} returns @code{true},
5093 then all arguments are treated as named. Otherwise, all named arguments
5094 except the last are treated as named.
5096 You need not define this hook if it always returns @code{false}.
5099 @deftypefn {Target Hook} bool TARGET_PRETEND_OUTGOING_VARARGS_NAMED (CUMULATIVE_ARGS *@var{ca})
5100 If you need to conditionally change ABIs so that one works with
5101 @code{TARGET_SETUP_INCOMING_VARARGS}, but the other works like neither
5102 @code{TARGET_SETUP_INCOMING_VARARGS} nor @code{TARGET_STRICT_ARGUMENT_NAMING} was
5103 defined, then define this hook to return @code{true} if
5104 @code{TARGET_SETUP_INCOMING_VARARGS} is used, @code{false} otherwise.
5105 Otherwise, you should not define this hook.
5109 @section Trampolines for Nested Functions
5110 @cindex trampolines for nested functions
5111 @cindex nested functions, trampolines for
5113 A @dfn{trampoline} is a small piece of code that is created at run time
5114 when the address of a nested function is taken. It normally resides on
5115 the stack, in the stack frame of the containing function. These macros
5116 tell GCC how to generate code to allocate and initialize a
5119 The instructions in the trampoline must do two things: load a constant
5120 address into the static chain register, and jump to the real address of
5121 the nested function. On CISC machines such as the m68k, this requires
5122 two instructions, a move immediate and a jump. Then the two addresses
5123 exist in the trampoline as word-long immediate operands. On RISC
5124 machines, it is often necessary to load each address into a register in
5125 two parts. Then pieces of each address form separate immediate
5128 The code generated to initialize the trampoline must store the variable
5129 parts---the static chain value and the function address---into the
5130 immediate operands of the instructions. On a CISC machine, this is
5131 simply a matter of copying each address to a memory reference at the
5132 proper offset from the start of the trampoline. On a RISC machine, it
5133 may be necessary to take out pieces of the address and store them
5136 @deftypefn {Target Hook} void TARGET_ASM_TRAMPOLINE_TEMPLATE (FILE *@var{f})
5137 This hook is called by @code{assemble_trampoline_template} to output,
5138 on the stream @var{f}, assembler code for a block of data that contains
5139 the constant parts of a trampoline. This code should not include a
5140 label---the label is taken care of automatically.
5142 If you do not define this hook, it means no template is needed
5143 for the target. Do not define this hook on systems where the block move
5144 code to copy the trampoline into place would be larger than the code
5145 to generate it on the spot.
5148 @defmac TRAMPOLINE_SECTION
5149 Return the section into which the trampoline template is to be placed
5150 (@pxref{Sections}). The default value is @code{readonly_data_section}.
5153 @defmac TRAMPOLINE_SIZE
5154 A C expression for the size in bytes of the trampoline, as an integer.
5157 @defmac TRAMPOLINE_ALIGNMENT
5158 Alignment required for trampolines, in bits.
5160 If you don't define this macro, the value of @code{FUNCTION_ALIGNMENT}
5161 is used for aligning trampolines.
5164 @deftypefn {Target Hook} void TARGET_TRAMPOLINE_INIT (rtx @var{m_tramp}, tree @var{fndecl}, rtx @var{static_chain})
5165 This hook is called to initialize a trampoline.
5166 @var{m_tramp} is an RTX for the memory block for the trampoline; @var{fndecl}
5167 is the @code{FUNCTION_DECL} for the nested function; @var{static_chain} is an
5168 RTX for the static chain value that should be passed to the function
5171 If the target defines @code{TARGET_ASM_TRAMPOLINE_TEMPLATE}, then the
5172 first thing this hook should do is emit a block move into @var{m_tramp}
5173 from the memory block returned by @code{assemble_trampoline_template}.
5174 Note that the block move need only cover the constant parts of the
5175 trampoline. If the target isolates the variable parts of the trampoline
5176 to the end, not all @code{TRAMPOLINE_SIZE} bytes need be copied.
5178 If the target requires any other actions, such as flushing caches or
5179 enabling stack execution, these actions should be performed after
5180 initializing the trampoline proper.
5183 @deftypefn {Target Hook} rtx TARGET_TRAMPOLINE_ADJUST_ADDRESS (rtx @var{addr})
5184 This hook should perform any machine-specific adjustment in
5185 the address of the trampoline. Its argument contains the address of the
5186 memory block that was passed to @code{TARGET_TRAMPOLINE_INIT}. In case
5187 the address to be used for a function call should be different from the
5188 address at which the template was stored, the different address should
5189 be returned; otherwise @var{addr} should be returned unchanged.
5190 If this hook is not defined, @var{addr} will be used for function calls.
5193 Implementing trampolines is difficult on many machines because they have
5194 separate instruction and data caches. Writing into a stack location
5195 fails to clear the memory in the instruction cache, so when the program
5196 jumps to that location, it executes the old contents.
5198 Here are two possible solutions. One is to clear the relevant parts of
5199 the instruction cache whenever a trampoline is set up. The other is to
5200 make all trampolines identical, by having them jump to a standard
5201 subroutine. The former technique makes trampoline execution faster; the
5202 latter makes initialization faster.
5204 To clear the instruction cache when a trampoline is initialized, define
5205 the following macro.
5207 @defmac CLEAR_INSN_CACHE (@var{beg}, @var{end})
5208 If defined, expands to a C expression clearing the @emph{instruction
5209 cache} in the specified interval. The definition of this macro would
5210 typically be a series of @code{asm} statements. Both @var{beg} and
5211 @var{end} are both pointer expressions.
5214 The operating system may also require the stack to be made executable
5215 before calling the trampoline. To implement this requirement, define
5216 the following macro.
5218 @defmac ENABLE_EXECUTE_STACK
5219 Define this macro if certain operations must be performed before executing
5220 code located on the stack. The macro should expand to a series of C
5221 file-scope constructs (e.g.@: functions) and provide a unique entry point
5222 named @code{__enable_execute_stack}. The target is responsible for
5223 emitting calls to the entry point in the code, for example from the
5224 @code{TARGET_TRAMPOLINE_INIT} hook.
5227 To use a standard subroutine, define the following macro. In addition,
5228 you must make sure that the instructions in a trampoline fill an entire
5229 cache line with identical instructions, or else ensure that the
5230 beginning of the trampoline code is always aligned at the same point in
5231 its cache line. Look in @file{m68k.h} as a guide.
5233 @defmac TRANSFER_FROM_TRAMPOLINE
5234 Define this macro if trampolines need a special subroutine to do their
5235 work. The macro should expand to a series of @code{asm} statements
5236 which will be compiled with GCC@. They go in a library function named
5237 @code{__transfer_from_trampoline}.
5239 If you need to avoid executing the ordinary prologue code of a compiled
5240 C function when you jump to the subroutine, you can do so by placing a
5241 special label of your own in the assembler code. Use one @code{asm}
5242 statement to generate an assembler label, and another to make the label
5243 global. Then trampolines can use that label to jump directly to your
5244 special assembler code.
5248 @section Implicit Calls to Library Routines
5249 @cindex library subroutine names
5250 @cindex @file{libgcc.a}
5252 @c prevent bad page break with this line
5253 Here is an explanation of implicit calls to library routines.
5255 @defmac DECLARE_LIBRARY_RENAMES
5256 This macro, if defined, should expand to a piece of C code that will get
5257 expanded when compiling functions for libgcc.a. It can be used to
5258 provide alternate names for GCC's internal library functions if there
5259 are ABI-mandated names that the compiler should provide.
5262 @findex set_optab_libfunc
5263 @findex init_one_libfunc
5264 @deftypefn {Target Hook} void TARGET_INIT_LIBFUNCS (void)
5265 This hook should declare additional library routines or rename
5266 existing ones, using the functions @code{set_optab_libfunc} and
5267 @code{init_one_libfunc} defined in @file{optabs.c}.
5268 @code{init_optabs} calls this macro after initializing all the normal
5271 The default is to do nothing. Most ports don't need to define this hook.
5274 @defmac FLOAT_LIB_COMPARE_RETURNS_BOOL (@var{mode}, @var{comparison})
5275 This macro should return @code{true} if the library routine that
5276 implements the floating point comparison operator @var{comparison} in
5277 mode @var{mode} will return a boolean, and @var{false} if it will
5280 GCC's own floating point libraries return tristates from the
5281 comparison operators, so the default returns false always. Most ports
5282 don't need to define this macro.
5285 @defmac TARGET_LIB_INT_CMP_BIASED
5286 This macro should evaluate to @code{true} if the integer comparison
5287 functions (like @code{__cmpdi2}) return 0 to indicate that the first
5288 operand is smaller than the second, 1 to indicate that they are equal,
5289 and 2 to indicate that the first operand is greater than the second.
5290 If this macro evaluates to @code{false} the comparison functions return
5291 @minus{}1, 0, and 1 instead of 0, 1, and 2. If the target uses the routines
5292 in @file{libgcc.a}, you do not need to define this macro.
5295 @cindex US Software GOFAST, floating point emulation library
5296 @cindex floating point emulation library, US Software GOFAST
5297 @cindex GOFAST, floating point emulation library
5298 @findex gofast_maybe_init_libfuncs
5299 @defmac US_SOFTWARE_GOFAST
5300 Define this macro if your system C library uses the US Software GOFAST
5301 library to provide floating point emulation.
5303 In addition to defining this macro, your architecture must set
5304 @code{TARGET_INIT_LIBFUNCS} to @code{gofast_maybe_init_libfuncs}, or
5305 else call that function from its version of that hook. It is defined
5306 in @file{config/gofast.h}, which must be included by your
5307 architecture's @file{@var{cpu}.c} file. See @file{sparc/sparc.c} for
5310 If this macro is defined, the
5311 @code{TARGET_FLOAT_LIB_COMPARE_RETURNS_BOOL} target hook must return
5312 false for @code{SFmode} and @code{DFmode} comparisons.
5315 @cindex @code{EDOM}, implicit usage
5318 The value of @code{EDOM} on the target machine, as a C integer constant
5319 expression. If you don't define this macro, GCC does not attempt to
5320 deposit the value of @code{EDOM} into @code{errno} directly. Look in
5321 @file{/usr/include/errno.h} to find the value of @code{EDOM} on your
5324 If you do not define @code{TARGET_EDOM}, then compiled code reports
5325 domain errors by calling the library function and letting it report the
5326 error. If mathematical functions on your system use @code{matherr} when
5327 there is an error, then you should leave @code{TARGET_EDOM} undefined so
5328 that @code{matherr} is used normally.
5331 @cindex @code{errno}, implicit usage
5332 @defmac GEN_ERRNO_RTX
5333 Define this macro as a C expression to create an rtl expression that
5334 refers to the global ``variable'' @code{errno}. (On certain systems,
5335 @code{errno} may not actually be a variable.) If you don't define this
5336 macro, a reasonable default is used.
5339 @cindex C99 math functions, implicit usage
5340 @defmac TARGET_C99_FUNCTIONS
5341 When this macro is nonzero, GCC will implicitly optimize @code{sin} calls into
5342 @code{sinf} and similarly for other functions defined by C99 standard. The
5343 default is zero because a number of existing systems lack support for these
5344 functions in their runtime so this macro needs to be redefined to one on
5345 systems that do support the C99 runtime.
5348 @cindex sincos math function, implicit usage
5349 @defmac TARGET_HAS_SINCOS
5350 When this macro is nonzero, GCC will implicitly optimize calls to @code{sin}
5351 and @code{cos} with the same argument to a call to @code{sincos}. The
5352 default is zero. The target has to provide the following functions:
5354 void sincos(double x, double *sin, double *cos);
5355 void sincosf(float x, float *sin, float *cos);
5356 void sincosl(long double x, long double *sin, long double *cos);
5360 @defmac NEXT_OBJC_RUNTIME
5361 Define this macro to generate code for Objective-C message sending using
5362 the calling convention of the NeXT system. This calling convention
5363 involves passing the object, the selector and the method arguments all
5364 at once to the method-lookup library function.
5366 The default calling convention passes just the object and the selector
5367 to the lookup function, which returns a pointer to the method.
5370 @node Addressing Modes
5371 @section Addressing Modes
5372 @cindex addressing modes
5374 @c prevent bad page break with this line
5375 This is about addressing modes.
5377 @defmac HAVE_PRE_INCREMENT
5378 @defmacx HAVE_PRE_DECREMENT
5379 @defmacx HAVE_POST_INCREMENT
5380 @defmacx HAVE_POST_DECREMENT
5381 A C expression that is nonzero if the machine supports pre-increment,
5382 pre-decrement, post-increment, or post-decrement addressing respectively.
5385 @defmac HAVE_PRE_MODIFY_DISP
5386 @defmacx HAVE_POST_MODIFY_DISP
5387 A C expression that is nonzero if the machine supports pre- or
5388 post-address side-effect generation involving constants other than
5389 the size of the memory operand.
5392 @defmac HAVE_PRE_MODIFY_REG
5393 @defmacx HAVE_POST_MODIFY_REG
5394 A C expression that is nonzero if the machine supports pre- or
5395 post-address side-effect generation involving a register displacement.
5398 @defmac CONSTANT_ADDRESS_P (@var{x})
5399 A C expression that is 1 if the RTX @var{x} is a constant which
5400 is a valid address. On most machines the default definition of
5401 @code{(CONSTANT_P (@var{x}) && GET_CODE (@var{x}) != CONST_DOUBLE)}
5402 is acceptable, but a few machines are more restrictive as to which
5403 constant addresses are supported.
5406 @defmac CONSTANT_P (@var{x})
5407 @code{CONSTANT_P}, which is defined by target-independent code,
5408 accepts integer-values expressions whose values are not explicitly
5409 known, such as @code{symbol_ref}, @code{label_ref}, and @code{high}
5410 expressions and @code{const} arithmetic expressions, in addition to
5411 @code{const_int} and @code{const_double} expressions.
5414 @defmac MAX_REGS_PER_ADDRESS
5415 A number, the maximum number of registers that can appear in a valid
5416 memory address. Note that it is up to you to specify a value equal to
5417 the maximum number that @code{TARGET_LEGITIMATE_ADDRESS_P} would ever
5421 @deftypefn {Target Hook} bool TARGET_LEGITIMATE_ADDRESS_P (enum machine_mode @var{mode}, rtx @var{x}, bool @var{strict})
5422 A function that returns whether @var{x} (an RTX) is a legitimate memory
5423 address on the target machine for a memory operand of mode @var{mode}.
5425 Legitimate addresses are defined in two variants: a strict variant and a
5426 non-strict one. The @var{strict} parameter chooses which variant is
5427 desired by the caller.
5429 The strict variant is used in the reload pass. It must be defined so
5430 that any pseudo-register that has not been allocated a hard register is
5431 considered a memory reference. This is because in contexts where some
5432 kind of register is required, a pseudo-register with no hard register
5433 must be rejected. For non-hard registers, the strict variant should look
5434 up the @code{reg_renumber} array; it should then proceed using the hard
5435 register number in the array, or treat the pseudo as a memory reference
5436 if the array holds @code{-1}.
5438 The non-strict variant is used in other passes. It must be defined to
5439 accept all pseudo-registers in every context where some kind of
5440 register is required.
5442 Normally, constant addresses which are the sum of a @code{symbol_ref}
5443 and an integer are stored inside a @code{const} RTX to mark them as
5444 constant. Therefore, there is no need to recognize such sums
5445 specifically as legitimate addresses. Normally you would simply
5446 recognize any @code{const} as legitimate.
5448 Usually @code{PRINT_OPERAND_ADDRESS} is not prepared to handle constant
5449 sums that are not marked with @code{const}. It assumes that a naked
5450 @code{plus} indicates indexing. If so, then you @emph{must} reject such
5451 naked constant sums as illegitimate addresses, so that none of them will
5452 be given to @code{PRINT_OPERAND_ADDRESS}.
5454 @cindex @code{TARGET_ENCODE_SECTION_INFO} and address validation
5455 On some machines, whether a symbolic address is legitimate depends on
5456 the section that the address refers to. On these machines, define the
5457 target hook @code{TARGET_ENCODE_SECTION_INFO} to store the information
5458 into the @code{symbol_ref}, and then check for it here. When you see a
5459 @code{const}, you will have to look inside it to find the
5460 @code{symbol_ref} in order to determine the section. @xref{Assembler
5463 @cindex @code{GO_IF_LEGITIMATE_ADDRESS}
5464 Some ports are still using a deprecated legacy substitute for
5465 this hook, the @code{GO_IF_LEGITIMATE_ADDRESS} macro. This macro
5469 #define GO_IF_LEGITIMATE_ADDRESS (@var{mode}, @var{x}, @var{label})
5473 and should @code{goto @var{label}} if the address @var{x} is a valid
5474 address on the target machine for a memory operand of mode @var{mode}.
5475 Whether the strict or non-strict variants are desired is defined by
5476 the @code{REG_OK_STRICT} macro introduced earlier in this section.
5477 Using the hook is usually simpler because it limits the number of
5478 files that are recompiled when changes are made.
5481 @defmac TARGET_MEM_CONSTRAINT
5482 A single character to be used instead of the default @code{'m'}
5483 character for general memory addresses. This defines the constraint
5484 letter which matches the memory addresses accepted by
5485 @code{TARGET_LEGITIMATE_ADDRESS_P}. Define this macro if you want to
5486 support new address formats in your back end without changing the
5487 semantics of the @code{'m'} constraint. This is necessary in order to
5488 preserve functionality of inline assembly constructs using the
5489 @code{'m'} constraint.
5492 @defmac FIND_BASE_TERM (@var{x})
5493 A C expression to determine the base term of address @var{x},
5494 or to provide a simplified version of @var{x} from which @file{alias.c}
5495 can easily find the base term. This macro is used in only two places:
5496 @code{find_base_value} and @code{find_base_term} in @file{alias.c}.
5498 It is always safe for this macro to not be defined. It exists so
5499 that alias analysis can understand machine-dependent addresses.
5501 The typical use of this macro is to handle addresses containing
5502 a label_ref or symbol_ref within an UNSPEC@.
5505 @deftypefn {Target Hook} rtx TARGET_LEGITIMIZE_ADDRESS (rtx @var{x}, rtx @var{oldx}, enum machine_mode @var{mode})
5506 This hook is given an invalid memory address @var{x} for an
5507 operand of mode @var{mode} and should try to return a valid memory
5510 @findex break_out_memory_refs
5511 @var{x} will always be the result of a call to @code{break_out_memory_refs},
5512 and @var{oldx} will be the operand that was given to that function to produce
5515 The code of the hook should not alter the substructure of
5516 @var{x}. If it transforms @var{x} into a more legitimate form, it
5517 should return the new @var{x}.
5519 It is not necessary for this hook to come up with a legitimate address.
5520 The compiler has standard ways of doing so in all cases. In fact, it
5521 is safe to omit this hook or make it return @var{x} if it cannot find
5522 a valid way to legitimize the address. But often a machine-dependent
5523 strategy can generate better code.
5526 @defmac LEGITIMIZE_RELOAD_ADDRESS (@var{x}, @var{mode}, @var{opnum}, @var{type}, @var{ind_levels}, @var{win})
5527 A C compound statement that attempts to replace @var{x}, which is an address
5528 that needs reloading, with a valid memory address for an operand of mode
5529 @var{mode}. @var{win} will be a C statement label elsewhere in the code.
5530 It is not necessary to define this macro, but it might be useful for
5531 performance reasons.
5533 For example, on the i386, it is sometimes possible to use a single
5534 reload register instead of two by reloading a sum of two pseudo
5535 registers into a register. On the other hand, for number of RISC
5536 processors offsets are limited so that often an intermediate address
5537 needs to be generated in order to address a stack slot. By defining
5538 @code{LEGITIMIZE_RELOAD_ADDRESS} appropriately, the intermediate addresses
5539 generated for adjacent some stack slots can be made identical, and thus
5542 @emph{Note}: This macro should be used with caution. It is necessary
5543 to know something of how reload works in order to effectively use this,
5544 and it is quite easy to produce macros that build in too much knowledge
5545 of reload internals.
5547 @emph{Note}: This macro must be able to reload an address created by a
5548 previous invocation of this macro. If it fails to handle such addresses
5549 then the compiler may generate incorrect code or abort.
5552 The macro definition should use @code{push_reload} to indicate parts that
5553 need reloading; @var{opnum}, @var{type} and @var{ind_levels} are usually
5554 suitable to be passed unaltered to @code{push_reload}.
5556 The code generated by this macro must not alter the substructure of
5557 @var{x}. If it transforms @var{x} into a more legitimate form, it
5558 should assign @var{x} (which will always be a C variable) a new value.
5559 This also applies to parts that you change indirectly by calling
5562 @findex strict_memory_address_p
5563 The macro definition may use @code{strict_memory_address_p} to test if
5564 the address has become legitimate.
5567 If you want to change only a part of @var{x}, one standard way of doing
5568 this is to use @code{copy_rtx}. Note, however, that it unshares only a
5569 single level of rtl. Thus, if the part to be changed is not at the
5570 top level, you'll need to replace first the top level.
5571 It is not necessary for this macro to come up with a legitimate
5572 address; but often a machine-dependent strategy can generate better code.
5575 @deftypefn {Target Hook} bool TARGET_MODE_DEPENDENT_ADDRESS_P (const_rtx @var{addr})
5576 This hook returns @code{true} if memory address @var{addr} can have
5577 different meanings depending on the machine mode of the memory
5578 reference it is used for or if the address is valid for some modes
5581 Autoincrement and autodecrement addresses typically have mode-dependent
5582 effects because the amount of the increment or decrement is the size
5583 of the operand being addressed. Some machines have other mode-dependent
5584 addresses. Many RISC machines have no mode-dependent addresses.
5586 You may assume that @var{addr} is a valid address for the machine.
5588 The default version of this hook returns @code{false}.
5591 @defmac GO_IF_MODE_DEPENDENT_ADDRESS (@var{addr}, @var{label})
5592 A C statement or compound statement with a conditional @code{goto
5593 @var{label};} executed if memory address @var{x} (an RTX) can have
5594 different meanings depending on the machine mode of the memory
5595 reference it is used for or if the address is valid for some modes
5598 Autoincrement and autodecrement addresses typically have mode-dependent
5599 effects because the amount of the increment or decrement is the size
5600 of the operand being addressed. Some machines have other mode-dependent
5601 addresses. Many RISC machines have no mode-dependent addresses.
5603 You may assume that @var{addr} is a valid address for the machine.
5605 These are obsolete macros, replaced by the
5606 @code{TARGET_MODE_DEPENDENT_ADDRESS_P} target hook.
5609 @defmac LEGITIMATE_CONSTANT_P (@var{x})
5610 A C expression that is nonzero if @var{x} is a legitimate constant for
5611 an immediate operand on the target machine. You can assume that
5612 @var{x} satisfies @code{CONSTANT_P}, so you need not check this. In fact,
5613 @samp{1} is a suitable definition for this macro on machines where
5614 anything @code{CONSTANT_P} is valid.
5617 @deftypefn {Target Hook} rtx TARGET_DELEGITIMIZE_ADDRESS (rtx @var{x})
5618 This hook is used to undo the possibly obfuscating effects of the
5619 @code{LEGITIMIZE_ADDRESS} and @code{LEGITIMIZE_RELOAD_ADDRESS} target
5620 macros. Some backend implementations of these macros wrap symbol
5621 references inside an @code{UNSPEC} rtx to represent PIC or similar
5622 addressing modes. This target hook allows GCC's optimizers to understand
5623 the semantics of these opaque @code{UNSPEC}s by converting them back
5624 into their original form.
5627 @deftypefn {Target Hook} bool TARGET_CANNOT_FORCE_CONST_MEM (rtx @var{x})
5628 This hook should return true if @var{x} is of a form that cannot (or
5629 should not) be spilled to the constant pool. The default version of
5630 this hook returns false.
5632 The primary reason to define this hook is to prevent reload from
5633 deciding that a non-legitimate constant would be better reloaded
5634 from the constant pool instead of spilling and reloading a register
5635 holding the constant. This restriction is often true of addresses
5636 of TLS symbols for various targets.
5639 @deftypefn {Target Hook} bool TARGET_USE_BLOCKS_FOR_CONSTANT_P (enum machine_mode @var{mode}, const_rtx @var{x})
5640 This hook should return true if pool entries for constant @var{x} can
5641 be placed in an @code{object_block} structure. @var{mode} is the mode
5644 The default version returns false for all constants.
5647 @deftypefn {Target Hook} tree TARGET_BUILTIN_RECIPROCAL (unsigned @var{fn}, bool @var{md_fn}, bool @var{sqrt})
5648 This hook should return the DECL of a function that implements reciprocal of
5649 the builtin function with builtin function code @var{fn}, or
5650 @code{NULL_TREE} if such a function is not available. @var{md_fn} is true
5651 when @var{fn} is a code of a machine-dependent builtin function. When
5652 @var{sqrt} is true, additional optimizations that apply only to the reciprocal
5653 of a square root function are performed, and only reciprocals of @code{sqrt}
5657 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_MASK_FOR_LOAD (void)
5658 This hook should return the DECL of a function @var{f} that given an
5659 address @var{addr} as an argument returns a mask @var{m} that can be
5660 used to extract from two vectors the relevant data that resides in
5661 @var{addr} in case @var{addr} is not properly aligned.
5663 The autovectorizer, when vectorizing a load operation from an address
5664 @var{addr} that may be unaligned, will generate two vector loads from
5665 the two aligned addresses around @var{addr}. It then generates a
5666 @code{REALIGN_LOAD} operation to extract the relevant data from the
5667 two loaded vectors. The first two arguments to @code{REALIGN_LOAD},
5668 @var{v1} and @var{v2}, are the two vectors, each of size @var{VS}, and
5669 the third argument, @var{OFF}, defines how the data will be extracted
5670 from these two vectors: if @var{OFF} is 0, then the returned vector is
5671 @var{v2}; otherwise, the returned vector is composed from the last
5672 @var{VS}-@var{OFF} elements of @var{v1} concatenated to the first
5673 @var{OFF} elements of @var{v2}.
5675 If this hook is defined, the autovectorizer will generate a call
5676 to @var{f} (using the DECL tree that this hook returns) and will
5677 use the return value of @var{f} as the argument @var{OFF} to
5678 @code{REALIGN_LOAD}. Therefore, the mask @var{m} returned by @var{f}
5679 should comply with the semantics expected by @code{REALIGN_LOAD}
5681 If this hook is not defined, then @var{addr} will be used as
5682 the argument @var{OFF} to @code{REALIGN_LOAD}, in which case the low
5683 log2(@var{VS}) @minus{} 1 bits of @var{addr} will be considered.
5686 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_EVEN (tree @var{x})
5687 This hook should return the DECL of a function @var{f} that implements
5688 widening multiplication of the even elements of two input vectors of type @var{x}.
5690 If this hook is defined, the autovectorizer will use it along with the
5691 @code{TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_ODD} target hook when vectorizing
5692 widening multiplication in cases that the order of the results does not have to be
5693 preserved (e.g.@: used only by a reduction computation). Otherwise, the
5694 @code{widen_mult_hi/lo} idioms will be used.
5697 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_ODD (tree @var{x})
5698 This hook should return the DECL of a function @var{f} that implements
5699 widening multiplication of the odd elements of two input vectors of type @var{x}.
5701 If this hook is defined, the autovectorizer will use it along with the
5702 @code{TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_EVEN} target hook when vectorizing
5703 widening multiplication in cases that the order of the results does not have to be
5704 preserved (e.g.@: used only by a reduction computation). Otherwise, the
5705 @code{widen_mult_hi/lo} idioms will be used.
5708 @deftypefn {Target Hook} int TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST (enum vect_cost_for_stmt @var{type_of_cost})
5709 Returns cost of different scalar or vector statements for vectorization cost model.
5712 @deftypefn {Target Hook} bool TARGET_VECTORIZE_VECTOR_ALIGNMENT_REACHABLE (const_tree @var{type}, bool @var{is_packed})
5713 Return true if vector alignment is reachable (by peeling N iterations) for the given type.
5716 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_VEC_PERM (tree @var{type}, tree *@var{mask_element_type})
5717 Target builtin that implements vector permute.
5720 @deftypefn {Target Hook} bool TARGET_VECTORIZE_BUILTIN_VEC_PERM_OK (tree @var{vec_type}, tree @var{mask})
5721 Return true if a vector created for @code{builtin_vec_perm} is valid.
5724 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_CONVERSION (unsigned @var{code}, tree @var{dest_type}, tree @var{src_type})
5725 This hook should return the DECL of a function that implements conversion of the
5726 input vector of type @var{src_type} to type @var{dest_type}.
5727 The value of @var{code} is one of the enumerators in @code{enum tree_code} and
5728 specifies how the conversion is to be applied
5729 (truncation, rounding, etc.).
5731 If this hook is defined, the autovectorizer will use the
5732 @code{TARGET_VECTORIZE_BUILTIN_CONVERSION} target hook when vectorizing
5733 conversion. Otherwise, it will return @code{NULL_TREE}.
5736 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION (tree @var{fndecl}, tree @var{vec_type_out}, tree @var{vec_type_in})
5737 This hook should return the decl of a function that implements the
5738 vectorized variant of the builtin function with builtin function code
5739 @var{code} or @code{NULL_TREE} if such a function is not available.
5740 The value of @var{fndecl} is the builtin function declaration. The
5741 return type of the vectorized function shall be of vector type
5742 @var{vec_type_out} and the argument types should be @var{vec_type_in}.
5745 @deftypefn {Target Hook} bool TARGET_SUPPORT_VECTOR_MISALIGNMENT (enum machine_mode @var{mode}, const_tree @var{type}, int @var{misalignment}, bool @var{is_packed})
5746 This hook should return true if the target supports misaligned vector
5747 store/load of a specific factor denoted in the @var{misalignment}
5748 parameter. The vector store/load should be of machine mode @var{mode} and
5749 the elements in the vectors should be of type @var{type}. @var{is_packed}
5750 parameter is true if the memory access is defined in a packed struct.
5753 @node Anchored Addresses
5754 @section Anchored Addresses
5755 @cindex anchored addresses
5756 @cindex @option{-fsection-anchors}
5758 GCC usually addresses every static object as a separate entity.
5759 For example, if we have:
5763 int foo (void) @{ return a + b + c; @}
5766 the code for @code{foo} will usually calculate three separate symbolic
5767 addresses: those of @code{a}, @code{b} and @code{c}. On some targets,
5768 it would be better to calculate just one symbolic address and access
5769 the three variables relative to it. The equivalent pseudocode would
5775 register int *xr = &x;
5776 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
5780 (which isn't valid C). We refer to shared addresses like @code{x} as
5781 ``section anchors''. Their use is controlled by @option{-fsection-anchors}.
5783 The hooks below describe the target properties that GCC needs to know
5784 in order to make effective use of section anchors. It won't use
5785 section anchors at all unless either @code{TARGET_MIN_ANCHOR_OFFSET}
5786 or @code{TARGET_MAX_ANCHOR_OFFSET} is set to a nonzero value.
5788 @deftypevr {Target Hook} HOST_WIDE_INT TARGET_MIN_ANCHOR_OFFSET
5789 The minimum offset that should be applied to a section anchor.
5790 On most targets, it should be the smallest offset that can be
5791 applied to a base register while still giving a legitimate address
5792 for every mode. The default value is 0.
5795 @deftypevr {Target Hook} HOST_WIDE_INT TARGET_MAX_ANCHOR_OFFSET
5796 Like @code{TARGET_MIN_ANCHOR_OFFSET}, but the maximum (inclusive)
5797 offset that should be applied to section anchors. The default
5801 @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_ANCHOR (rtx @var{x})
5802 Write the assembly code to define section anchor @var{x}, which is a
5803 @code{SYMBOL_REF} for which @samp{SYMBOL_REF_ANCHOR_P (@var{x})} is true.
5804 The hook is called with the assembly output position set to the beginning
5805 of @code{SYMBOL_REF_BLOCK (@var{x})}.
5807 If @code{ASM_OUTPUT_DEF} is available, the hook's default definition uses
5808 it to define the symbol as @samp{. + SYMBOL_REF_BLOCK_OFFSET (@var{x})}.
5809 If @code{ASM_OUTPUT_DEF} is not available, the hook's default definition
5810 is @code{NULL}, which disables the use of section anchors altogether.
5813 @deftypefn {Target Hook} bool TARGET_USE_ANCHORS_FOR_SYMBOL_P (const_rtx @var{x})
5814 Return true if GCC should attempt to use anchors to access @code{SYMBOL_REF}
5815 @var{x}. You can assume @samp{SYMBOL_REF_HAS_BLOCK_INFO_P (@var{x})} and
5816 @samp{!SYMBOL_REF_ANCHOR_P (@var{x})}.
5818 The default version is correct for most targets, but you might need to
5819 intercept this hook to handle things like target-specific attributes
5820 or target-specific sections.
5823 @node Condition Code
5824 @section Condition Code Status
5825 @cindex condition code status
5827 The macros in this section can be split in two families, according to the
5828 two ways of representing condition codes in GCC.
5830 The first representation is the so called @code{(cc0)} representation
5831 (@pxref{Jump Patterns}), where all instructions can have an implicit
5832 clobber of the condition codes. The second is the condition code
5833 register representation, which provides better schedulability for
5834 architectures that do have a condition code register, but on which
5835 most instructions do not affect it. The latter category includes
5838 The implicit clobbering poses a strong restriction on the placement of
5839 the definition and use of the condition code, which need to be in adjacent
5840 insns for machines using @code{(cc0)}. This can prevent important
5841 optimizations on some machines. For example, on the IBM RS/6000, there
5842 is a delay for taken branches unless the condition code register is set
5843 three instructions earlier than the conditional branch. The instruction
5844 scheduler cannot perform this optimization if it is not permitted to
5845 separate the definition and use of the condition code register.
5847 For this reason, it is possible and suggested to use a register to
5848 represent the condition code for new ports. If there is a specific
5849 condition code register in the machine, use a hard register. If the
5850 condition code or comparison result can be placed in any general register,
5851 or if there are multiple condition registers, use a pseudo register.
5852 Registers used to store the condition code value will usually have a mode
5853 that is in class @code{MODE_CC}.
5855 Alternatively, you can use @code{BImode} if the comparison operator is
5856 specified already in the compare instruction. In this case, you are not
5857 interested in most macros in this section.
5860 * CC0 Condition Codes:: Old style representation of condition codes.
5861 * MODE_CC Condition Codes:: Modern representation of condition codes.
5862 * Cond. Exec. Macros:: Macros to control conditional execution.
5865 @node CC0 Condition Codes
5866 @subsection Representation of condition codes using @code{(cc0)}
5870 The file @file{conditions.h} defines a variable @code{cc_status} to
5871 describe how the condition code was computed (in case the interpretation of
5872 the condition code depends on the instruction that it was set by). This
5873 variable contains the RTL expressions on which the condition code is
5874 currently based, and several standard flags.
5876 Sometimes additional machine-specific flags must be defined in the machine
5877 description header file. It can also add additional machine-specific
5878 information by defining @code{CC_STATUS_MDEP}.
5880 @defmac CC_STATUS_MDEP
5881 C code for a data type which is used for declaring the @code{mdep}
5882 component of @code{cc_status}. It defaults to @code{int}.
5884 This macro is not used on machines that do not use @code{cc0}.
5887 @defmac CC_STATUS_MDEP_INIT
5888 A C expression to initialize the @code{mdep} field to ``empty''.
5889 The default definition does nothing, since most machines don't use
5890 the field anyway. If you want to use the field, you should probably
5891 define this macro to initialize it.
5893 This macro is not used on machines that do not use @code{cc0}.
5896 @defmac NOTICE_UPDATE_CC (@var{exp}, @var{insn})
5897 A C compound statement to set the components of @code{cc_status}
5898 appropriately for an insn @var{insn} whose body is @var{exp}. It is
5899 this macro's responsibility to recognize insns that set the condition
5900 code as a byproduct of other activity as well as those that explicitly
5903 This macro is not used on machines that do not use @code{cc0}.
5905 If there are insns that do not set the condition code but do alter
5906 other machine registers, this macro must check to see whether they
5907 invalidate the expressions that the condition code is recorded as
5908 reflecting. For example, on the 68000, insns that store in address
5909 registers do not set the condition code, which means that usually
5910 @code{NOTICE_UPDATE_CC} can leave @code{cc_status} unaltered for such
5911 insns. But suppose that the previous insn set the condition code
5912 based on location @samp{a4@@(102)} and the current insn stores a new
5913 value in @samp{a4}. Although the condition code is not changed by
5914 this, it will no longer be true that it reflects the contents of
5915 @samp{a4@@(102)}. Therefore, @code{NOTICE_UPDATE_CC} must alter
5916 @code{cc_status} in this case to say that nothing is known about the
5917 condition code value.
5919 The definition of @code{NOTICE_UPDATE_CC} must be prepared to deal
5920 with the results of peephole optimization: insns whose patterns are
5921 @code{parallel} RTXs containing various @code{reg}, @code{mem} or
5922 constants which are just the operands. The RTL structure of these
5923 insns is not sufficient to indicate what the insns actually do. What
5924 @code{NOTICE_UPDATE_CC} should do when it sees one is just to run
5925 @code{CC_STATUS_INIT}.
5927 A possible definition of @code{NOTICE_UPDATE_CC} is to call a function
5928 that looks at an attribute (@pxref{Insn Attributes}) named, for example,
5929 @samp{cc}. This avoids having detailed information about patterns in
5930 two places, the @file{md} file and in @code{NOTICE_UPDATE_CC}.
5933 @node MODE_CC Condition Codes
5934 @subsection Representation of condition codes using registers
5938 @defmac SELECT_CC_MODE (@var{op}, @var{x}, @var{y})
5939 On many machines, the condition code may be produced by other instructions
5940 than compares, for example the branch can use directly the condition
5941 code set by a subtract instruction. However, on some machines
5942 when the condition code is set this way some bits (such as the overflow
5943 bit) are not set in the same way as a test instruction, so that a different
5944 branch instruction must be used for some conditional branches. When
5945 this happens, use the machine mode of the condition code register to
5946 record different formats of the condition code register. Modes can
5947 also be used to record which compare instruction (e.g. a signed or an
5948 unsigned comparison) produced the condition codes.
5950 If other modes than @code{CCmode} are required, add them to
5951 @file{@var{machine}-modes.def} and define @code{SELECT_CC_MODE} to choose
5952 a mode given an operand of a compare. This is needed because the modes
5953 have to be chosen not only during RTL generation but also, for example,
5954 by instruction combination. The result of @code{SELECT_CC_MODE} should
5955 be consistent with the mode used in the patterns; for example to support
5956 the case of the add on the SPARC discussed above, we have the pattern
5960 [(set (reg:CC_NOOV 0)
5962 (plus:SI (match_operand:SI 0 "register_operand" "%r")
5963 (match_operand:SI 1 "arith_operand" "rI"))
5970 together with a @code{SELECT_CC_MODE} that returns @code{CC_NOOVmode}
5971 for comparisons whose argument is a @code{plus}:
5974 #define SELECT_CC_MODE(OP,X,Y) \
5975 (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \
5976 ? ((OP == EQ || OP == NE) ? CCFPmode : CCFPEmode) \
5977 : ((GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \
5978 || GET_CODE (X) == NEG) \
5979 ? CC_NOOVmode : CCmode))
5982 Another reason to use modes is to retain information on which operands
5983 were used by the comparison; see @code{REVERSIBLE_CC_MODE} later in
5986 You should define this macro if and only if you define extra CC modes
5987 in @file{@var{machine}-modes.def}.
5990 @defmac CANONICALIZE_COMPARISON (@var{code}, @var{op0}, @var{op1})
5991 On some machines not all possible comparisons are defined, but you can
5992 convert an invalid comparison into a valid one. For example, the Alpha
5993 does not have a @code{GT} comparison, but you can use an @code{LT}
5994 comparison instead and swap the order of the operands.
5996 On such machines, define this macro to be a C statement to do any
5997 required conversions. @var{code} is the initial comparison code
5998 and @var{op0} and @var{op1} are the left and right operands of the
5999 comparison, respectively. You should modify @var{code}, @var{op0}, and
6000 @var{op1} as required.
6002 GCC will not assume that the comparison resulting from this macro is
6003 valid but will see if the resulting insn matches a pattern in the
6006 You need not define this macro if it would never change the comparison
6010 @defmac REVERSIBLE_CC_MODE (@var{mode})
6011 A C expression whose value is one if it is always safe to reverse a
6012 comparison whose mode is @var{mode}. If @code{SELECT_CC_MODE}
6013 can ever return @var{mode} for a floating-point inequality comparison,
6014 then @code{REVERSIBLE_CC_MODE (@var{mode})} must be zero.
6016 You need not define this macro if it would always returns zero or if the
6017 floating-point format is anything other than @code{IEEE_FLOAT_FORMAT}.
6018 For example, here is the definition used on the SPARC, where floating-point
6019 inequality comparisons are always given @code{CCFPEmode}:
6022 #define REVERSIBLE_CC_MODE(MODE) ((MODE) != CCFPEmode)
6026 @defmac REVERSE_CONDITION (@var{code}, @var{mode})
6027 A C expression whose value is reversed condition code of the @var{code} for
6028 comparison done in CC_MODE @var{mode}. The macro is used only in case
6029 @code{REVERSIBLE_CC_MODE (@var{mode})} is nonzero. Define this macro in case
6030 machine has some non-standard way how to reverse certain conditionals. For
6031 instance in case all floating point conditions are non-trapping, compiler may
6032 freely convert unordered compares to ordered one. Then definition may look
6036 #define REVERSE_CONDITION(CODE, MODE) \
6037 ((MODE) != CCFPmode ? reverse_condition (CODE) \
6038 : reverse_condition_maybe_unordered (CODE))
6042 @deftypefn {Target Hook} bool TARGET_FIXED_CONDITION_CODE_REGS (unsigned int *@var{p1}, unsigned int *@var{p2})
6043 On targets which do not use @code{(cc0)}, and which use a hard
6044 register rather than a pseudo-register to hold condition codes, the
6045 regular CSE passes are often not able to identify cases in which the
6046 hard register is set to a common value. Use this hook to enable a
6047 small pass which optimizes such cases. This hook should return true
6048 to enable this pass, and it should set the integers to which its
6049 arguments point to the hard register numbers used for condition codes.
6050 When there is only one such register, as is true on most systems, the
6051 integer pointed to by @var{p2} should be set to
6052 @code{INVALID_REGNUM}.
6054 The default version of this hook returns false.
6057 @deftypefn {Target Hook} {enum machine_mode} TARGET_CC_MODES_COMPATIBLE (enum machine_mode @var{m1}, enum machine_mode @var{m2})
6058 On targets which use multiple condition code modes in class
6059 @code{MODE_CC}, it is sometimes the case that a comparison can be
6060 validly done in more than one mode. On such a system, define this
6061 target hook to take two mode arguments and to return a mode in which
6062 both comparisons may be validly done. If there is no such mode,
6063 return @code{VOIDmode}.
6065 The default version of this hook checks whether the modes are the
6066 same. If they are, it returns that mode. If they are different, it
6067 returns @code{VOIDmode}.
6070 @node Cond. Exec. Macros
6071 @subsection Macros to control conditional execution
6072 @findex conditional execution
6075 There is one macro that may need to be defined for targets
6076 supporting conditional execution, independent of how they
6077 represent conditional branches.
6079 @defmac REVERSE_CONDEXEC_PREDICATES_P (@var{op1}, @var{op2})
6080 A C expression that returns true if the conditional execution predicate
6081 @var{op1}, a comparison operation, is the inverse of @var{op2} and vice
6082 versa. Define this to return 0 if the target has conditional execution
6083 predicates that cannot be reversed safely. There is no need to validate
6084 that the arguments of op1 and op2 are the same, this is done separately.
6085 If no expansion is specified, this macro is defined as follows:
6088 #define REVERSE_CONDEXEC_PREDICATES_P (x, y) \
6089 (GET_CODE ((x)) == reversed_comparison_code ((y), NULL))
6094 @section Describing Relative Costs of Operations
6095 @cindex costs of instructions
6096 @cindex relative costs
6097 @cindex speed of instructions
6099 These macros let you describe the relative speed of various operations
6100 on the target machine.
6102 @defmac REGISTER_MOVE_COST (@var{mode}, @var{from}, @var{to})
6103 A C expression for the cost of moving data of mode @var{mode} from a
6104 register in class @var{from} to one in class @var{to}. The classes are
6105 expressed using the enumeration values such as @code{GENERAL_REGS}. A
6106 value of 2 is the default; other values are interpreted relative to
6109 It is not required that the cost always equal 2 when @var{from} is the
6110 same as @var{to}; on some machines it is expensive to move between
6111 registers if they are not general registers.
6113 If reload sees an insn consisting of a single @code{set} between two
6114 hard registers, and if @code{REGISTER_MOVE_COST} applied to their
6115 classes returns a value of 2, reload does not check to ensure that the
6116 constraints of the insn are met. Setting a cost of other than 2 will
6117 allow reload to verify that the constraints are met. You should do this
6118 if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
6121 @defmac MEMORY_MOVE_COST (@var{mode}, @var{class}, @var{in})
6122 A C expression for the cost of moving data of mode @var{mode} between a
6123 register of class @var{class} and memory; @var{in} is zero if the value
6124 is to be written to memory, nonzero if it is to be read in. This cost
6125 is relative to those in @code{REGISTER_MOVE_COST}. If moving between
6126 registers and memory is more expensive than between two registers, you
6127 should define this macro to express the relative cost.
6129 If you do not define this macro, GCC uses a default cost of 4 plus
6130 the cost of copying via a secondary reload register, if one is
6131 needed. If your machine requires a secondary reload register to copy
6132 between memory and a register of @var{class} but the reload mechanism is
6133 more complex than copying via an intermediate, define this macro to
6134 reflect the actual cost of the move.
6136 GCC defines the function @code{memory_move_secondary_cost} if
6137 secondary reloads are needed. It computes the costs due to copying via
6138 a secondary register. If your machine copies from memory using a
6139 secondary register in the conventional way but the default base value of
6140 4 is not correct for your machine, define this macro to add some other
6141 value to the result of that function. The arguments to that function
6142 are the same as to this macro.
6144 These macros are obsolete, new ports should use the target hook
6145 @code{TARGET_MEMORY_MOVE_COST} instead.
6148 @deftypefn {Target Hook} int TARGET_MEMORY_MOVE_COST (enum machine_mode @var{mode}, enum reg_class @var{regclass}, bool @var{in})
6149 This target hook should return the cost of moving data of mode @var{mode}
6150 between a register of class @var{class} and memory; @var{in} is @code{false}
6151 if the value is to be written to memory, @code{true} if it is to be read in.
6152 This cost is relative to those in @code{REGISTER_MOVE_COST}. If moving
6153 between registers and memory is more expensive than between two registers,
6154 you should add this target hook to express the relative cost.
6156 If you do not add this target hook, GCC uses a default cost of 4 plus
6157 the cost of copying via a secondary reload register, if one is
6158 needed. If your machine requires a secondary reload register to copy
6159 between memory and a register of @var{class} but the reload mechanism is
6160 more complex than copying via an intermediate, use this target hook to
6161 reflect the actual cost of the move.
6163 GCC defines the function @code{memory_move_secondary_cost} if
6164 secondary reloads are needed. It computes the costs due to copying via
6165 a secondary register. If your machine copies from memory using a
6166 secondary register in the conventional way but the default base value of
6167 4 is not correct for your machine, use this target hook to add some other
6168 value to the result of that function. The arguments to that function
6169 are the same as to this target hook.
6172 @defmac BRANCH_COST (@var{speed_p}, @var{predictable_p})
6173 A C expression for the cost of a branch instruction. A value of 1 is the
6174 default; other values are interpreted relative to that. Parameter @var{speed_p}
6175 is true when the branch in question should be optimized for speed. When
6176 it is false, @code{BRANCH_COST} should be returning value optimal for code size
6177 rather then performance considerations. @var{predictable_p} is true for well
6178 predictable branches. On many architectures the @code{BRANCH_COST} can be
6182 Here are additional macros which do not specify precise relative costs,
6183 but only that certain actions are more expensive than GCC would
6186 @defmac SLOW_BYTE_ACCESS
6187 Define this macro as a C expression which is nonzero if accessing less
6188 than a word of memory (i.e.@: a @code{char} or a @code{short}) is no
6189 faster than accessing a word of memory, i.e., if such access
6190 require more than one instruction or if there is no difference in cost
6191 between byte and (aligned) word loads.
6193 When this macro is not defined, the compiler will access a field by
6194 finding the smallest containing object; when it is defined, a fullword
6195 load will be used if alignment permits. Unless bytes accesses are
6196 faster than word accesses, using word accesses is preferable since it
6197 may eliminate subsequent memory access if subsequent accesses occur to
6198 other fields in the same word of the structure, but to different bytes.
6201 @defmac SLOW_UNALIGNED_ACCESS (@var{mode}, @var{alignment})
6202 Define this macro to be the value 1 if memory accesses described by the
6203 @var{mode} and @var{alignment} parameters have a cost many times greater
6204 than aligned accesses, for example if they are emulated in a trap
6207 When this macro is nonzero, the compiler will act as if
6208 @code{STRICT_ALIGNMENT} were nonzero when generating code for block
6209 moves. This can cause significantly more instructions to be produced.
6210 Therefore, do not set this macro nonzero if unaligned accesses only add a
6211 cycle or two to the time for a memory access.
6213 If the value of this macro is always zero, it need not be defined. If
6214 this macro is defined, it should produce a nonzero value when
6215 @code{STRICT_ALIGNMENT} is nonzero.
6218 @defmac MOVE_RATIO (@var{speed})
6219 The threshold of number of scalar memory-to-memory move insns, @emph{below}
6220 which a sequence of insns should be generated instead of a
6221 string move insn or a library call. Increasing the value will always
6222 make code faster, but eventually incurs high cost in increased code size.
6224 Note that on machines where the corresponding move insn is a
6225 @code{define_expand} that emits a sequence of insns, this macro counts
6226 the number of such sequences.
6228 The parameter @var{speed} is true if the code is currently being
6229 optimized for speed rather than size.
6231 If you don't define this, a reasonable default is used.
6234 @defmac MOVE_BY_PIECES_P (@var{size}, @var{alignment})
6235 A C expression used to determine whether @code{move_by_pieces} will be used to
6236 copy a chunk of memory, or whether some other block move mechanism
6237 will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6238 than @code{MOVE_RATIO}.
6241 @defmac MOVE_MAX_PIECES
6242 A C expression used by @code{move_by_pieces} to determine the largest unit
6243 a load or store used to copy memory is. Defaults to @code{MOVE_MAX}.
6246 @defmac CLEAR_RATIO (@var{speed})
6247 The threshold of number of scalar move insns, @emph{below} which a sequence
6248 of insns should be generated to clear memory instead of a string clear insn
6249 or a library call. Increasing the value will always make code faster, but
6250 eventually incurs high cost in increased code size.
6252 The parameter @var{speed} is true if the code is currently being
6253 optimized for speed rather than size.
6255 If you don't define this, a reasonable default is used.
6258 @defmac CLEAR_BY_PIECES_P (@var{size}, @var{alignment})
6259 A C expression used to determine whether @code{clear_by_pieces} will be used
6260 to clear a chunk of memory, or whether some other block clear mechanism
6261 will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6262 than @code{CLEAR_RATIO}.
6265 @defmac SET_RATIO (@var{speed})
6266 The threshold of number of scalar move insns, @emph{below} which a sequence
6267 of insns should be generated to set memory to a constant value, instead of
6268 a block set insn or a library call.
6269 Increasing the value will always make code faster, but
6270 eventually incurs high cost in increased code size.
6272 The parameter @var{speed} is true if the code is currently being
6273 optimized for speed rather than size.
6275 If you don't define this, it defaults to the value of @code{MOVE_RATIO}.
6278 @defmac SET_BY_PIECES_P (@var{size}, @var{alignment})
6279 A C expression used to determine whether @code{store_by_pieces} will be
6280 used to set a chunk of memory to a constant value, or whether some
6281 other mechanism will be used. Used by @code{__builtin_memset} when
6282 storing values other than constant zero.
6283 Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6284 than @code{SET_RATIO}.
6287 @defmac STORE_BY_PIECES_P (@var{size}, @var{alignment})
6288 A C expression used to determine whether @code{store_by_pieces} will be
6289 used to set a chunk of memory to a constant string value, or whether some
6290 other mechanism will be used. Used by @code{__builtin_strcpy} when
6291 called with a constant source string.
6292 Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6293 than @code{MOVE_RATIO}.
6296 @defmac USE_LOAD_POST_INCREMENT (@var{mode})
6297 A C expression used to determine whether a load postincrement is a good
6298 thing to use for a given mode. Defaults to the value of
6299 @code{HAVE_POST_INCREMENT}.
6302 @defmac USE_LOAD_POST_DECREMENT (@var{mode})
6303 A C expression used to determine whether a load postdecrement is a good
6304 thing to use for a given mode. Defaults to the value of
6305 @code{HAVE_POST_DECREMENT}.
6308 @defmac USE_LOAD_PRE_INCREMENT (@var{mode})
6309 A C expression used to determine whether a load preincrement is a good
6310 thing to use for a given mode. Defaults to the value of
6311 @code{HAVE_PRE_INCREMENT}.
6314 @defmac USE_LOAD_PRE_DECREMENT (@var{mode})
6315 A C expression used to determine whether a load predecrement is a good
6316 thing to use for a given mode. Defaults to the value of
6317 @code{HAVE_PRE_DECREMENT}.
6320 @defmac USE_STORE_POST_INCREMENT (@var{mode})
6321 A C expression used to determine whether a store postincrement is a good
6322 thing to use for a given mode. Defaults to the value of
6323 @code{HAVE_POST_INCREMENT}.
6326 @defmac USE_STORE_POST_DECREMENT (@var{mode})
6327 A C expression used to determine whether a store postdecrement is a good
6328 thing to use for a given mode. Defaults to the value of
6329 @code{HAVE_POST_DECREMENT}.
6332 @defmac USE_STORE_PRE_INCREMENT (@var{mode})
6333 This macro is used to determine whether a store preincrement is a good
6334 thing to use for a given mode. Defaults to the value of
6335 @code{HAVE_PRE_INCREMENT}.
6338 @defmac USE_STORE_PRE_DECREMENT (@var{mode})
6339 This macro is used to determine whether a store predecrement is a good
6340 thing to use for a given mode. Defaults to the value of
6341 @code{HAVE_PRE_DECREMENT}.
6344 @defmac NO_FUNCTION_CSE
6345 Define this macro if it is as good or better to call a constant
6346 function address than to call an address kept in a register.
6349 @defmac RANGE_TEST_NON_SHORT_CIRCUIT
6350 Define this macro if a non-short-circuit operation produced by
6351 @samp{fold_range_test ()} is optimal. This macro defaults to true if
6352 @code{BRANCH_COST} is greater than or equal to the value 2.
6355 @deftypefn {Target Hook} bool TARGET_RTX_COSTS (rtx @var{x}, int @var{code}, int @var{outer_code}, int *@var{total}, bool @var{speed})
6356 This target hook describes the relative costs of RTL expressions.
6358 The cost may depend on the precise form of the expression, which is
6359 available for examination in @var{x}, and the rtx code of the expression
6360 in which it is contained, found in @var{outer_code}. @var{code} is the
6361 expression code---redundant, since it can be obtained with
6362 @code{GET_CODE (@var{x})}.
6364 In implementing this hook, you can use the construct
6365 @code{COSTS_N_INSNS (@var{n})} to specify a cost equal to @var{n} fast
6368 On entry to the hook, @code{*@var{total}} contains a default estimate
6369 for the cost of the expression. The hook should modify this value as
6370 necessary. Traditionally, the default costs are @code{COSTS_N_INSNS (5)}
6371 for multiplications, @code{COSTS_N_INSNS (7)} for division and modulus
6372 operations, and @code{COSTS_N_INSNS (1)} for all other operations.
6374 When optimizing for code size, i.e.@: when @code{speed} is
6375 false, this target hook should be used to estimate the relative
6376 size cost of an expression, again relative to @code{COSTS_N_INSNS}.
6378 The hook returns true when all subexpressions of @var{x} have been
6379 processed, and false when @code{rtx_cost} should recurse.
6382 @deftypefn {Target Hook} int TARGET_ADDRESS_COST (rtx @var{address}, bool @var{speed})
6383 This hook computes the cost of an addressing mode that contains
6384 @var{address}. If not defined, the cost is computed from
6385 the @var{address} expression and the @code{TARGET_RTX_COST} hook.
6387 For most CISC machines, the default cost is a good approximation of the
6388 true cost of the addressing mode. However, on RISC machines, all
6389 instructions normally have the same length and execution time. Hence
6390 all addresses will have equal costs.
6392 In cases where more than one form of an address is known, the form with
6393 the lowest cost will be used. If multiple forms have the same, lowest,
6394 cost, the one that is the most complex will be used.
6396 For example, suppose an address that is equal to the sum of a register
6397 and a constant is used twice in the same basic block. When this macro
6398 is not defined, the address will be computed in a register and memory
6399 references will be indirect through that register. On machines where
6400 the cost of the addressing mode containing the sum is no higher than
6401 that of a simple indirect reference, this will produce an additional
6402 instruction and possibly require an additional register. Proper
6403 specification of this macro eliminates this overhead for such machines.
6405 This hook is never called with an invalid address.
6407 On machines where an address involving more than one register is as
6408 cheap as an address computation involving only one register, defining
6409 @code{TARGET_ADDRESS_COST} to reflect this can cause two registers to
6410 be live over a region of code where only one would have been if
6411 @code{TARGET_ADDRESS_COST} were not defined in that manner. This effect
6412 should be considered in the definition of this macro. Equivalent costs
6413 should probably only be given to addresses with different numbers of
6414 registers on machines with lots of registers.
6418 @section Adjusting the Instruction Scheduler
6420 The instruction scheduler may need a fair amount of machine-specific
6421 adjustment in order to produce good code. GCC provides several target
6422 hooks for this purpose. It is usually enough to define just a few of
6423 them: try the first ones in this list first.
6425 @deftypefn {Target Hook} int TARGET_SCHED_ISSUE_RATE (void)
6426 This hook returns the maximum number of instructions that can ever
6427 issue at the same time on the target machine. The default is one.
6428 Although the insn scheduler can define itself the possibility of issue
6429 an insn on the same cycle, the value can serve as an additional
6430 constraint to issue insns on the same simulated processor cycle (see
6431 hooks @samp{TARGET_SCHED_REORDER} and @samp{TARGET_SCHED_REORDER2}).
6432 This value must be constant over the entire compilation. If you need
6433 it to vary depending on what the instructions are, you must use
6434 @samp{TARGET_SCHED_VARIABLE_ISSUE}.
6437 @deftypefn {Target Hook} int TARGET_SCHED_VARIABLE_ISSUE (FILE *@var{file}, int @var{verbose}, rtx @var{insn}, int @var{more})
6438 This hook is executed by the scheduler after it has scheduled an insn
6439 from the ready list. It should return the number of insns which can
6440 still be issued in the current cycle. The default is
6441 @samp{@w{@var{more} - 1}} for insns other than @code{CLOBBER} and
6442 @code{USE}, which normally are not counted against the issue rate.
6443 You should define this hook if some insns take more machine resources
6444 than others, so that fewer insns can follow them in the same cycle.
6445 @var{file} is either a null pointer, or a stdio stream to write any
6446 debug output to. @var{verbose} is the verbose level provided by
6447 @option{-fsched-verbose-@var{n}}. @var{insn} is the instruction that
6451 @deftypefn {Target Hook} int TARGET_SCHED_ADJUST_COST (rtx @var{insn}, rtx @var{link}, rtx @var{dep_insn}, int @var{cost})
6452 This function corrects the value of @var{cost} based on the
6453 relationship between @var{insn} and @var{dep_insn} through the
6454 dependence @var{link}. It should return the new value. The default
6455 is to make no adjustment to @var{cost}. This can be used for example
6456 to specify to the scheduler using the traditional pipeline description
6457 that an output- or anti-dependence does not incur the same cost as a
6458 data-dependence. If the scheduler using the automaton based pipeline
6459 description, the cost of anti-dependence is zero and the cost of
6460 output-dependence is maximum of one and the difference of latency
6461 times of the first and the second insns. If these values are not
6462 acceptable, you could use the hook to modify them too. See also
6463 @pxref{Processor pipeline description}.
6466 @deftypefn {Target Hook} int TARGET_SCHED_ADJUST_PRIORITY (rtx @var{insn}, int @var{priority})
6467 This hook adjusts the integer scheduling priority @var{priority} of
6468 @var{insn}. It should return the new priority. Increase the priority to
6469 execute @var{insn} earlier, reduce the priority to execute @var{insn}
6470 later. Do not define this hook if you do not need to adjust the
6471 scheduling priorities of insns.
6474 @deftypefn {Target Hook} int TARGET_SCHED_REORDER (FILE *@var{file}, int @var{verbose}, rtx *@var{ready}, int *@var{n_readyp}, int @var{clock})
6475 This hook is executed by the scheduler after it has scheduled the ready
6476 list, to allow the machine description to reorder it (for example to
6477 combine two small instructions together on @samp{VLIW} machines).
6478 @var{file} is either a null pointer, or a stdio stream to write any
6479 debug output to. @var{verbose} is the verbose level provided by
6480 @option{-fsched-verbose-@var{n}}. @var{ready} is a pointer to the ready
6481 list of instructions that are ready to be scheduled. @var{n_readyp} is
6482 a pointer to the number of elements in the ready list. The scheduler
6483 reads the ready list in reverse order, starting with
6484 @var{ready}[@var{*n_readyp} @minus{} 1] and going to @var{ready}[0]. @var{clock}
6485 is the timer tick of the scheduler. You may modify the ready list and
6486 the number of ready insns. The return value is the number of insns that
6487 can issue this cycle; normally this is just @code{issue_rate}. See also
6488 @samp{TARGET_SCHED_REORDER2}.
6491 @deftypefn {Target Hook} int TARGET_SCHED_REORDER2 (FILE *@var{file}, int @var{verbose}, rtx *@var{ready}, int *@var{n_readyp}, int @var{clock})
6492 Like @samp{TARGET_SCHED_REORDER}, but called at a different time. That
6493 function is called whenever the scheduler starts a new cycle. This one
6494 is called once per iteration over a cycle, immediately after
6495 @samp{TARGET_SCHED_VARIABLE_ISSUE}; it can reorder the ready list and
6496 return the number of insns to be scheduled in the same cycle. Defining
6497 this hook can be useful if there are frequent situations where
6498 scheduling one insn causes other insns to become ready in the same
6499 cycle. These other insns can then be taken into account properly.
6502 @deftypefn {Target Hook} void TARGET_SCHED_DEPENDENCIES_EVALUATION_HOOK (rtx @var{head}, rtx @var{tail})
6503 This hook is called after evaluation forward dependencies of insns in
6504 chain given by two parameter values (@var{head} and @var{tail}
6505 correspondingly) but before insns scheduling of the insn chain. For
6506 example, it can be used for better insn classification if it requires
6507 analysis of dependencies. This hook can use backward and forward
6508 dependencies of the insn scheduler because they are already
6512 @deftypefn {Target Hook} void TARGET_SCHED_INIT (FILE *@var{file}, int @var{verbose}, int @var{max_ready})
6513 This hook is executed by the scheduler at the beginning of each block of
6514 instructions that are to be scheduled. @var{file} is either a null
6515 pointer, or a stdio stream to write any debug output to. @var{verbose}
6516 is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6517 @var{max_ready} is the maximum number of insns in the current scheduling
6518 region that can be live at the same time. This can be used to allocate
6519 scratch space if it is needed, e.g.@: by @samp{TARGET_SCHED_REORDER}.
6522 @deftypefn {Target Hook} void TARGET_SCHED_FINISH (FILE *@var{file}, int @var{verbose})
6523 This hook is executed by the scheduler at the end of each block of
6524 instructions that are to be scheduled. It can be used to perform
6525 cleanup of any actions done by the other scheduling hooks. @var{file}
6526 is either a null pointer, or a stdio stream to write any debug output
6527 to. @var{verbose} is the verbose level provided by
6528 @option{-fsched-verbose-@var{n}}.
6531 @deftypefn {Target Hook} void TARGET_SCHED_INIT_GLOBAL (FILE *@var{file}, int @var{verbose}, int @var{old_max_uid})
6532 This hook is executed by the scheduler after function level initializations.
6533 @var{file} is either a null pointer, or a stdio stream to write any debug output to.
6534 @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6535 @var{old_max_uid} is the maximum insn uid when scheduling begins.
6538 @deftypefn {Target Hook} void TARGET_SCHED_FINISH_GLOBAL (FILE *@var{file}, int @var{verbose})
6539 This is the cleanup hook corresponding to @code{TARGET_SCHED_INIT_GLOBAL}.
6540 @var{file} is either a null pointer, or a stdio stream to write any debug output to.
6541 @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6544 @deftypefn {Target Hook} rtx TARGET_SCHED_DFA_PRE_CYCLE_INSN (void)
6545 The hook returns an RTL insn. The automaton state used in the
6546 pipeline hazard recognizer is changed as if the insn were scheduled
6547 when the new simulated processor cycle starts. Usage of the hook may
6548 simplify the automaton pipeline description for some @acronym{VLIW}
6549 processors. If the hook is defined, it is used only for the automaton
6550 based pipeline description. The default is not to change the state
6551 when the new simulated processor cycle starts.
6554 @deftypefn {Target Hook} void TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN (void)
6555 The hook can be used to initialize data used by the previous hook.
6558 @deftypefn {Target Hook} rtx TARGET_SCHED_DFA_POST_CYCLE_INSN (void)
6559 The hook is analogous to @samp{TARGET_SCHED_DFA_PRE_CYCLE_INSN} but used
6560 to changed the state as if the insn were scheduled when the new
6561 simulated processor cycle finishes.
6564 @deftypefn {Target Hook} void TARGET_SCHED_INIT_DFA_POST_CYCLE_INSN (void)
6565 The hook is analogous to @samp{TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN} but
6566 used to initialize data used by the previous hook.
6569 @deftypefn {Target Hook} void TARGET_SCHED_DFA_PRE_ADVANCE_CYCLE (void)
6570 The hook to notify target that the current simulated cycle is about to finish.
6571 The hook is analogous to @samp{TARGET_SCHED_DFA_PRE_CYCLE_INSN} but used
6572 to change the state in more complicated situations - e.g., when advancing
6573 state on a single insn is not enough.
6576 @deftypefn {Target Hook} void TARGET_SCHED_DFA_POST_ADVANCE_CYCLE (void)
6577 The hook to notify target that new simulated cycle has just started.
6578 The hook is analogous to @samp{TARGET_SCHED_DFA_POST_CYCLE_INSN} but used
6579 to change the state in more complicated situations - e.g., when advancing
6580 state on a single insn is not enough.
6583 @deftypefn {Target Hook} int TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD (void)
6584 This hook controls better choosing an insn from the ready insn queue
6585 for the @acronym{DFA}-based insn scheduler. Usually the scheduler
6586 chooses the first insn from the queue. If the hook returns a positive
6587 value, an additional scheduler code tries all permutations of
6588 @samp{TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD ()}
6589 subsequent ready insns to choose an insn whose issue will result in
6590 maximal number of issued insns on the same cycle. For the
6591 @acronym{VLIW} processor, the code could actually solve the problem of
6592 packing simple insns into the @acronym{VLIW} insn. Of course, if the
6593 rules of @acronym{VLIW} packing are described in the automaton.
6595 This code also could be used for superscalar @acronym{RISC}
6596 processors. Let us consider a superscalar @acronym{RISC} processor
6597 with 3 pipelines. Some insns can be executed in pipelines @var{A} or
6598 @var{B}, some insns can be executed only in pipelines @var{B} or
6599 @var{C}, and one insn can be executed in pipeline @var{B}. The
6600 processor may issue the 1st insn into @var{A} and the 2nd one into
6601 @var{B}. In this case, the 3rd insn will wait for freeing @var{B}
6602 until the next cycle. If the scheduler issues the 3rd insn the first,
6603 the processor could issue all 3 insns per cycle.
6605 Actually this code demonstrates advantages of the automaton based
6606 pipeline hazard recognizer. We try quickly and easy many insn
6607 schedules to choose the best one.
6609 The default is no multipass scheduling.
6612 @deftypefn {Target Hook} int TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD (rtx @var{insn})
6614 This hook controls what insns from the ready insn queue will be
6615 considered for the multipass insn scheduling. If the hook returns
6616 zero for @var{insn}, the insn will be not chosen to
6619 The default is that any ready insns can be chosen to be issued.
6622 @deftypefn {Target Hook} int TARGET_SCHED_DFA_NEW_CYCLE (FILE *@var{dump}, int @var{verbose}, rtx @var{insn}, int @var{last_clock}, int @var{clock}, int *@var{sort_p})
6624 This hook is called by the insn scheduler before issuing @var{insn}
6625 on cycle @var{clock}. If the hook returns nonzero,
6626 @var{insn} is not issued on this processor cycle. Instead,
6627 the processor cycle is advanced. If *@var{sort_p}
6628 is zero, the insn ready queue is not sorted on the new cycle
6629 start as usually. @var{dump} and @var{verbose} specify the file and
6630 verbosity level to use for debugging output.
6631 @var{last_clock} and @var{clock} are, respectively, the
6632 processor cycle on which the previous insn has been issued,
6633 and the current processor cycle.
6636 @deftypefn {Target Hook} bool TARGET_SCHED_IS_COSTLY_DEPENDENCE (struct _dep *@var{_dep}, int @var{cost}, int @var{distance})
6637 This hook is used to define which dependences are considered costly by
6638 the target, so costly that it is not advisable to schedule the insns that
6639 are involved in the dependence too close to one another. The parameters
6640 to this hook are as follows: The first parameter @var{_dep} is the dependence
6641 being evaluated. The second parameter @var{cost} is the cost of the
6642 dependence as estimated by the scheduler, and the third
6643 parameter @var{distance} is the distance in cycles between the two insns.
6644 The hook returns @code{true} if considering the distance between the two
6645 insns the dependence between them is considered costly by the target,
6646 and @code{false} otherwise.
6648 Defining this hook can be useful in multiple-issue out-of-order machines,
6649 where (a) it's practically hopeless to predict the actual data/resource
6650 delays, however: (b) there's a better chance to predict the actual grouping
6651 that will be formed, and (c) correctly emulating the grouping can be very
6652 important. In such targets one may want to allow issuing dependent insns
6653 closer to one another---i.e., closer than the dependence distance; however,
6654 not in cases of ``costly dependences'', which this hooks allows to define.
6657 @deftypefn {Target Hook} void TARGET_SCHED_H_I_D_EXTENDED (void)
6658 This hook is called by the insn scheduler after emitting a new instruction to
6659 the instruction stream. The hook notifies a target backend to extend its
6660 per instruction data structures.
6663 @deftypefn {Target Hook} {void *} TARGET_SCHED_ALLOC_SCHED_CONTEXT (void)
6664 Return a pointer to a store large enough to hold target scheduling context.
6667 @deftypefn {Target Hook} void TARGET_SCHED_INIT_SCHED_CONTEXT (void *@var{tc}, bool @var{clean_p})
6668 Initialize store pointed to by @var{tc} to hold target scheduling context.
6669 It @var{clean_p} is true then initialize @var{tc} as if scheduler is at the
6670 beginning of the block. Otherwise, copy the current context into @var{tc}.
6673 @deftypefn {Target Hook} void TARGET_SCHED_SET_SCHED_CONTEXT (void *@var{tc})
6674 Copy target scheduling context pointed to by @var{tc} to the current context.
6677 @deftypefn {Target Hook} void TARGET_SCHED_CLEAR_SCHED_CONTEXT (void *@var{tc})
6678 Deallocate internal data in target scheduling context pointed to by @var{tc}.
6681 @deftypefn {Target Hook} void TARGET_SCHED_FREE_SCHED_CONTEXT (void *@var{tc})
6682 Deallocate a store for target scheduling context pointed to by @var{tc}.
6685 @deftypefn {Target Hook} int TARGET_SCHED_SPECULATE_INSN (rtx @var{insn}, int @var{request}, rtx *@var{new_pat})
6686 This hook is called by the insn scheduler when @var{insn} has only
6687 speculative dependencies and therefore can be scheduled speculatively.
6688 The hook is used to check if the pattern of @var{insn} has a speculative
6689 version and, in case of successful check, to generate that speculative
6690 pattern. The hook should return 1, if the instruction has a speculative form,
6691 or @minus{}1, if it doesn't. @var{request} describes the type of requested
6692 speculation. If the return value equals 1 then @var{new_pat} is assigned
6693 the generated speculative pattern.
6696 @deftypefn {Target Hook} bool TARGET_SCHED_NEEDS_BLOCK_P (int @var{dep_status})
6697 This hook is called by the insn scheduler during generation of recovery code
6698 for @var{insn}. It should return @code{true}, if the corresponding check
6699 instruction should branch to recovery code, or @code{false} otherwise.
6702 @deftypefn {Target Hook} rtx TARGET_SCHED_GEN_SPEC_CHECK (rtx @var{insn}, rtx @var{label}, int @var{mutate_p})
6703 This hook is called by the insn scheduler to generate a pattern for recovery
6704 check instruction. If @var{mutate_p} is zero, then @var{insn} is a
6705 speculative instruction for which the check should be generated.
6706 @var{label} is either a label of a basic block, where recovery code should
6707 be emitted, or a null pointer, when requested check doesn't branch to
6708 recovery code (a simple check). If @var{mutate_p} is nonzero, then
6709 a pattern for a branchy check corresponding to a simple check denoted by
6710 @var{insn} should be generated. In this case @var{label} can't be null.
6713 @deftypefn {Target Hook} bool TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD_SPEC (const_rtx @var{insn})
6714 This hook is used as a workaround for
6715 @samp{TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD} not being
6716 called on the first instruction of the ready list. The hook is used to
6717 discard speculative instructions that stand first in the ready list from
6718 being scheduled on the current cycle. If the hook returns @code{false},
6719 @var{insn} will not be chosen to be issued.
6720 For non-speculative instructions,
6721 the hook should always return @code{true}. For example, in the ia64 backend
6722 the hook is used to cancel data speculative insns when the ALAT table
6726 @deftypefn {Target Hook} void TARGET_SCHED_SET_SCHED_FLAGS (struct spec_info_def *@var{spec_info})
6727 This hook is used by the insn scheduler to find out what features should be
6729 The structure *@var{spec_info} should be filled in by the target.
6730 The structure describes speculation types that can be used in the scheduler.
6733 @deftypefn {Target Hook} int TARGET_SCHED_SMS_RES_MII (struct ddg *@var{g})
6734 This hook is called by the swing modulo scheduler to calculate a
6735 resource-based lower bound which is based on the resources available in
6736 the machine and the resources required by each instruction. The target
6737 backend can use @var{g} to calculate such bound. A very simple lower
6738 bound will be used in case this hook is not implemented: the total number
6739 of instructions divided by the issue rate.
6743 @section Dividing the Output into Sections (Texts, Data, @dots{})
6744 @c the above section title is WAY too long. maybe cut the part between
6745 @c the (...)? --mew 10feb93
6747 An object file is divided into sections containing different types of
6748 data. In the most common case, there are three sections: the @dfn{text
6749 section}, which holds instructions and read-only data; the @dfn{data
6750 section}, which holds initialized writable data; and the @dfn{bss
6751 section}, which holds uninitialized data. Some systems have other kinds
6754 @file{varasm.c} provides several well-known sections, such as
6755 @code{text_section}, @code{data_section} and @code{bss_section}.
6756 The normal way of controlling a @code{@var{foo}_section} variable
6757 is to define the associated @code{@var{FOO}_SECTION_ASM_OP} macro,
6758 as described below. The macros are only read once, when @file{varasm.c}
6759 initializes itself, so their values must be run-time constants.
6760 They may however depend on command-line flags.
6762 @emph{Note:} Some run-time files, such @file{crtstuff.c}, also make
6763 use of the @code{@var{FOO}_SECTION_ASM_OP} macros, and expect them
6764 to be string literals.
6766 Some assemblers require a different string to be written every time a
6767 section is selected. If your assembler falls into this category, you
6768 should define the @code{TARGET_ASM_INIT_SECTIONS} hook and use
6769 @code{get_unnamed_section} to set up the sections.
6771 You must always create a @code{text_section}, either by defining
6772 @code{TEXT_SECTION_ASM_OP} or by initializing @code{text_section}
6773 in @code{TARGET_ASM_INIT_SECTIONS}. The same is true of
6774 @code{data_section} and @code{DATA_SECTION_ASM_OP}. If you do not
6775 create a distinct @code{readonly_data_section}, the default is to
6776 reuse @code{text_section}.
6778 All the other @file{varasm.c} sections are optional, and are null
6779 if the target does not provide them.
6781 @defmac TEXT_SECTION_ASM_OP
6782 A C expression whose value is a string, including spacing, containing the
6783 assembler operation that should precede instructions and read-only data.
6784 Normally @code{"\t.text"} is right.
6787 @defmac HOT_TEXT_SECTION_NAME
6788 If defined, a C string constant for the name of the section containing most
6789 frequently executed functions of the program. If not defined, GCC will provide
6790 a default definition if the target supports named sections.
6793 @defmac UNLIKELY_EXECUTED_TEXT_SECTION_NAME
6794 If defined, a C string constant for the name of the section containing unlikely
6795 executed functions in the program.
6798 @defmac DATA_SECTION_ASM_OP
6799 A C expression whose value is a string, including spacing, containing the
6800 assembler operation to identify the following data as writable initialized
6801 data. Normally @code{"\t.data"} is right.
6804 @defmac SDATA_SECTION_ASM_OP
6805 If defined, a C expression whose value is a string, including spacing,
6806 containing the assembler operation to identify the following data as
6807 initialized, writable small data.
6810 @defmac READONLY_DATA_SECTION_ASM_OP
6811 A C expression whose value is a string, including spacing, containing the
6812 assembler operation to identify the following data as read-only initialized
6816 @defmac BSS_SECTION_ASM_OP
6817 If defined, a C expression whose value is a string, including spacing,
6818 containing the assembler operation to identify the following data as
6819 uninitialized global data. If not defined, and neither
6820 @code{ASM_OUTPUT_BSS} nor @code{ASM_OUTPUT_ALIGNED_BSS} are defined,
6821 uninitialized global data will be output in the data section if
6822 @option{-fno-common} is passed, otherwise @code{ASM_OUTPUT_COMMON} will be
6826 @defmac SBSS_SECTION_ASM_OP
6827 If defined, a C expression whose value is a string, including spacing,
6828 containing the assembler operation to identify the following data as
6829 uninitialized, writable small data.
6832 @defmac TLS_COMMON_ASM_OP
6833 If defined, a C expression whose value is a string containing the
6834 assembler operation to identify the following data as thread-local
6835 common data. The default is @code{".tls_common"}.
6838 @defmac TLS_SECTION_ASM_FLAG
6839 If defined, a C expression whose value is a character constant
6840 containing the flag used to mark a section as a TLS section. The
6841 default is @code{'T'}.
6844 @defmac INIT_SECTION_ASM_OP
6845 If defined, a C expression whose value is a string, including spacing,
6846 containing the assembler operation to identify the following data as
6847 initialization code. If not defined, GCC will assume such a section does
6848 not exist. This section has no corresponding @code{init_section}
6849 variable; it is used entirely in runtime code.
6852 @defmac FINI_SECTION_ASM_OP
6853 If defined, a C expression whose value is a string, including spacing,
6854 containing the assembler operation to identify the following data as
6855 finalization code. If not defined, GCC will assume such a section does
6856 not exist. This section has no corresponding @code{fini_section}
6857 variable; it is used entirely in runtime code.
6860 @defmac INIT_ARRAY_SECTION_ASM_OP
6861 If defined, a C expression whose value is a string, including spacing,
6862 containing the assembler operation to identify the following data as
6863 part of the @code{.init_array} (or equivalent) section. If not
6864 defined, GCC will assume such a section does not exist. Do not define
6865 both this macro and @code{INIT_SECTION_ASM_OP}.
6868 @defmac FINI_ARRAY_SECTION_ASM_OP
6869 If defined, a C expression whose value is a string, including spacing,
6870 containing the assembler operation to identify the following data as
6871 part of the @code{.fini_array} (or equivalent) section. If not
6872 defined, GCC will assume such a section does not exist. Do not define
6873 both this macro and @code{FINI_SECTION_ASM_OP}.
6876 @defmac CRT_CALL_STATIC_FUNCTION (@var{section_op}, @var{function})
6877 If defined, an ASM statement that switches to a different section
6878 via @var{section_op}, calls @var{function}, and switches back to
6879 the text section. This is used in @file{crtstuff.c} if
6880 @code{INIT_SECTION_ASM_OP} or @code{FINI_SECTION_ASM_OP} to calls
6881 to initialization and finalization functions from the init and fini
6882 sections. By default, this macro uses a simple function call. Some
6883 ports need hand-crafted assembly code to avoid dependencies on
6884 registers initialized in the function prologue or to ensure that
6885 constant pools don't end up too far way in the text section.
6888 @defmac TARGET_LIBGCC_SDATA_SECTION
6889 If defined, a string which names the section into which small
6890 variables defined in crtstuff and libgcc should go. This is useful
6891 when the target has options for optimizing access to small data, and
6892 you want the crtstuff and libgcc routines to be conservative in what
6893 they expect of your application yet liberal in what your application
6894 expects. For example, for targets with a @code{.sdata} section (like
6895 MIPS), you could compile crtstuff with @code{-G 0} so that it doesn't
6896 require small data support from your application, but use this macro
6897 to put small data into @code{.sdata} so that your application can
6898 access these variables whether it uses small data or not.
6901 @defmac FORCE_CODE_SECTION_ALIGN
6902 If defined, an ASM statement that aligns a code section to some
6903 arbitrary boundary. This is used to force all fragments of the
6904 @code{.init} and @code{.fini} sections to have to same alignment
6905 and thus prevent the linker from having to add any padding.
6908 @defmac JUMP_TABLES_IN_TEXT_SECTION
6909 Define this macro to be an expression with a nonzero value if jump
6910 tables (for @code{tablejump} insns) should be output in the text
6911 section, along with the assembler instructions. Otherwise, the
6912 readonly data section is used.
6914 This macro is irrelevant if there is no separate readonly data section.
6917 @deftypefn {Target Hook} void TARGET_ASM_INIT_SECTIONS (void)
6918 Define this hook if you need to do something special to set up the
6919 @file{varasm.c} sections, or if your target has some special sections
6920 of its own that you need to create.
6922 GCC calls this hook after processing the command line, but before writing
6923 any assembly code, and before calling any of the section-returning hooks
6927 @deftypefn {Target Hook} int TARGET_ASM_RELOC_RW_MASK (void)
6928 Return a mask describing how relocations should be treated when
6929 selecting sections. Bit 1 should be set if global relocations
6930 should be placed in a read-write section; bit 0 should be set if
6931 local relocations should be placed in a read-write section.
6933 The default version of this function returns 3 when @option{-fpic}
6934 is in effect, and 0 otherwise. The hook is typically redefined
6935 when the target cannot support (some kinds of) dynamic relocations
6936 in read-only sections even in executables.
6939 @deftypefn {Target Hook} {section *} TARGET_ASM_SELECT_SECTION (tree @var{exp}, int @var{reloc}, unsigned HOST_WIDE_INT @var{align})
6940 Return the section into which @var{exp} should be placed. You can
6941 assume that @var{exp} is either a @code{VAR_DECL} node or a constant of
6942 some sort. @var{reloc} indicates whether the initial value of @var{exp}
6943 requires link-time relocations. Bit 0 is set when variable contains
6944 local relocations only, while bit 1 is set for global relocations.
6945 @var{align} is the constant alignment in bits.
6947 The default version of this function takes care of putting read-only
6948 variables in @code{readonly_data_section}.
6950 See also @var{USE_SELECT_SECTION_FOR_FUNCTIONS}.
6953 @defmac USE_SELECT_SECTION_FOR_FUNCTIONS
6954 Define this macro if you wish TARGET_ASM_SELECT_SECTION to be called
6955 for @code{FUNCTION_DECL}s as well as for variables and constants.
6957 In the case of a @code{FUNCTION_DECL}, @var{reloc} will be zero if the
6958 function has been determined to be likely to be called, and nonzero if
6959 it is unlikely to be called.
6962 @deftypefn {Target Hook} void TARGET_ASM_UNIQUE_SECTION (tree @var{decl}, int @var{reloc})
6963 Build up a unique section name, expressed as a @code{STRING_CST} node,
6964 and assign it to @samp{DECL_SECTION_NAME (@var{decl})}.
6965 As with @code{TARGET_ASM_SELECT_SECTION}, @var{reloc} indicates whether
6966 the initial value of @var{exp} requires link-time relocations.
6968 The default version of this function appends the symbol name to the
6969 ELF section name that would normally be used for the symbol. For
6970 example, the function @code{foo} would be placed in @code{.text.foo}.
6971 Whatever the actual target object format, this is often good enough.
6974 @deftypefn {Target Hook} {section *} TARGET_ASM_FUNCTION_RODATA_SECTION (tree @var{decl})
6975 Return the readonly data section associated with
6976 @samp{DECL_SECTION_NAME (@var{decl})}.
6977 The default version of this function selects @code{.gnu.linkonce.r.name} if
6978 the function's section is @code{.gnu.linkonce.t.name}, @code{.rodata.name}
6979 if function is in @code{.text.name}, and the normal readonly-data section
6983 @deftypefn {Target Hook} {section *} TARGET_ASM_SELECT_RTX_SECTION (enum machine_mode @var{mode}, rtx @var{x}, unsigned HOST_WIDE_INT @var{align})
6984 Return the section into which a constant @var{x}, of mode @var{mode},
6985 should be placed. You can assume that @var{x} is some kind of
6986 constant in RTL@. The argument @var{mode} is redundant except in the
6987 case of a @code{const_int} rtx. @var{align} is the constant alignment
6990 The default version of this function takes care of putting symbolic
6991 constants in @code{flag_pic} mode in @code{data_section} and everything
6992 else in @code{readonly_data_section}.
6995 @deftypefn {Target Hook} tree TARGET_MANGLE_DECL_ASSEMBLER_NAME (tree @var{decl}, tree @var{id})
6996 Define this hook if you need to postprocess the assembler name generated
6997 by target-independent code. The @var{id} provided to this hook will be
6998 the computed name (e.g., the macro @code{DECL_NAME} of the @var{decl} in C,
6999 or the mangled name of the @var{decl} in C++). The return value of the
7000 hook is an @code{IDENTIFIER_NODE} for the appropriate mangled name on
7001 your target system. The default implementation of this hook just
7002 returns the @var{id} provided.
7005 @deftypefn {Target Hook} void TARGET_ENCODE_SECTION_INFO (tree @var{decl}, rtx @var{rtl}, int @var{new_decl_p})
7006 Define this hook if references to a symbol or a constant must be
7007 treated differently depending on something about the variable or
7008 function named by the symbol (such as what section it is in).
7010 The hook is executed immediately after rtl has been created for
7011 @var{decl}, which may be a variable or function declaration or
7012 an entry in the constant pool. In either case, @var{rtl} is the
7013 rtl in question. Do @emph{not} use @code{DECL_RTL (@var{decl})}
7014 in this hook; that field may not have been initialized yet.
7016 In the case of a constant, it is safe to assume that the rtl is
7017 a @code{mem} whose address is a @code{symbol_ref}. Most decls
7018 will also have this form, but that is not guaranteed. Global
7019 register variables, for instance, will have a @code{reg} for their
7020 rtl. (Normally the right thing to do with such unusual rtl is
7023 The @var{new_decl_p} argument will be true if this is the first time
7024 that @code{TARGET_ENCODE_SECTION_INFO} has been invoked on this decl. It will
7025 be false for subsequent invocations, which will happen for duplicate
7026 declarations. Whether or not anything must be done for the duplicate
7027 declaration depends on whether the hook examines @code{DECL_ATTRIBUTES}.
7028 @var{new_decl_p} is always true when the hook is called for a constant.
7030 @cindex @code{SYMBOL_REF_FLAG}, in @code{TARGET_ENCODE_SECTION_INFO}
7031 The usual thing for this hook to do is to record flags in the
7032 @code{symbol_ref}, using @code{SYMBOL_REF_FLAG} or @code{SYMBOL_REF_FLAGS}.
7033 Historically, the name string was modified if it was necessary to
7034 encode more than one bit of information, but this practice is now
7035 discouraged; use @code{SYMBOL_REF_FLAGS}.
7037 The default definition of this hook, @code{default_encode_section_info}
7038 in @file{varasm.c}, sets a number of commonly-useful bits in
7039 @code{SYMBOL_REF_FLAGS}. Check whether the default does what you need
7040 before overriding it.
7043 @deftypefn {Target Hook} {const char *} TARGET_STRIP_NAME_ENCODING (const char *@var{name})
7044 Decode @var{name} and return the real name part, sans
7045 the characters that @code{TARGET_ENCODE_SECTION_INFO}
7049 @deftypefn {Target Hook} bool TARGET_IN_SMALL_DATA_P (const_tree @var{exp})
7050 Returns true if @var{exp} should be placed into a ``small data'' section.
7051 The default version of this hook always returns false.
7054 @deftypevr {Target Hook} bool TARGET_HAVE_SRODATA_SECTION
7055 Contains the value true if the target places read-only
7056 ``small data'' into a separate section. The default value is false.
7059 @deftypefn {Target Hook} bool TARGET_BINDS_LOCAL_P (const_tree @var{exp})
7060 Returns true if @var{exp} names an object for which name resolution
7061 rules must resolve to the current ``module'' (dynamic shared library
7062 or executable image).
7064 The default version of this hook implements the name resolution rules
7065 for ELF, which has a looser model of global name binding than other
7066 currently supported object file formats.
7069 @deftypevr {Target Hook} bool TARGET_HAVE_TLS
7070 Contains the value true if the target supports thread-local storage.
7071 The default value is false.
7076 @section Position Independent Code
7077 @cindex position independent code
7080 This section describes macros that help implement generation of position
7081 independent code. Simply defining these macros is not enough to
7082 generate valid PIC; you must also add support to the hook
7083 @code{TARGET_LEGITIMATE_ADDRESS_P} and to the macro
7084 @code{PRINT_OPERAND_ADDRESS}, as well as @code{LEGITIMIZE_ADDRESS}. You
7085 must modify the definition of @samp{movsi} to do something appropriate
7086 when the source operand contains a symbolic address. You may also
7087 need to alter the handling of switch statements so that they use
7089 @c i rearranged the order of the macros above to try to force one of
7090 @c them to the next line, to eliminate an overfull hbox. --mew 10feb93
7092 @defmac PIC_OFFSET_TABLE_REGNUM
7093 The register number of the register used to address a table of static
7094 data addresses in memory. In some cases this register is defined by a
7095 processor's ``application binary interface'' (ABI)@. When this macro
7096 is defined, RTL is generated for this register once, as with the stack
7097 pointer and frame pointer registers. If this macro is not defined, it
7098 is up to the machine-dependent files to allocate such a register (if
7099 necessary). Note that this register must be fixed when in use (e.g.@:
7100 when @code{flag_pic} is true).
7103 @defmac PIC_OFFSET_TABLE_REG_CALL_CLOBBERED
7104 Define this macro if the register defined by
7105 @code{PIC_OFFSET_TABLE_REGNUM} is clobbered by calls. Do not define
7106 this macro if @code{PIC_OFFSET_TABLE_REGNUM} is not defined.
7109 @defmac LEGITIMATE_PIC_OPERAND_P (@var{x})
7110 A C expression that is nonzero if @var{x} is a legitimate immediate
7111 operand on the target machine when generating position independent code.
7112 You can assume that @var{x} satisfies @code{CONSTANT_P}, so you need not
7113 check this. You can also assume @var{flag_pic} is true, so you need not
7114 check it either. You need not define this macro if all constants
7115 (including @code{SYMBOL_REF}) can be immediate operands when generating
7116 position independent code.
7119 @node Assembler Format
7120 @section Defining the Output Assembler Language
7122 This section describes macros whose principal purpose is to describe how
7123 to write instructions in assembler language---rather than what the
7127 * File Framework:: Structural information for the assembler file.
7128 * Data Output:: Output of constants (numbers, strings, addresses).
7129 * Uninitialized Data:: Output of uninitialized variables.
7130 * Label Output:: Output and generation of labels.
7131 * Initialization:: General principles of initialization
7132 and termination routines.
7133 * Macros for Initialization::
7134 Specific macros that control the handling of
7135 initialization and termination routines.
7136 * Instruction Output:: Output of actual instructions.
7137 * Dispatch Tables:: Output of jump tables.
7138 * Exception Region Output:: Output of exception region code.
7139 * Alignment Output:: Pseudo ops for alignment and skipping data.
7142 @node File Framework
7143 @subsection The Overall Framework of an Assembler File
7144 @cindex assembler format
7145 @cindex output of assembler code
7147 @c prevent bad page break with this line
7148 This describes the overall framework of an assembly file.
7150 @findex default_file_start
7151 @deftypefn {Target Hook} void TARGET_ASM_FILE_START (void)
7152 Output to @code{asm_out_file} any text which the assembler expects to
7153 find at the beginning of a file. The default behavior is controlled
7154 by two flags, documented below. Unless your target's assembler is
7155 quite unusual, if you override the default, you should call
7156 @code{default_file_start} at some point in your target hook. This
7157 lets other target files rely on these variables.
7160 @deftypevr {Target Hook} bool TARGET_ASM_FILE_START_APP_OFF
7161 If this flag is true, the text of the macro @code{ASM_APP_OFF} will be
7162 printed as the very first line in the assembly file, unless
7163 @option{-fverbose-asm} is in effect. (If that macro has been defined
7164 to the empty string, this variable has no effect.) With the normal
7165 definition of @code{ASM_APP_OFF}, the effect is to notify the GNU
7166 assembler that it need not bother stripping comments or extra
7167 whitespace from its input. This allows it to work a bit faster.
7169 The default is false. You should not set it to true unless you have
7170 verified that your port does not generate any extra whitespace or
7171 comments that will cause GAS to issue errors in NO_APP mode.
7174 @deftypevr {Target Hook} bool TARGET_ASM_FILE_START_FILE_DIRECTIVE
7175 If this flag is true, @code{output_file_directive} will be called
7176 for the primary source file, immediately after printing
7177 @code{ASM_APP_OFF} (if that is enabled). Most ELF assemblers expect
7178 this to be done. The default is false.
7181 @deftypefn {Target Hook} void TARGET_ASM_FILE_END (void)
7182 Output to @code{asm_out_file} any text which the assembler expects
7183 to find at the end of a file. The default is to output nothing.
7186 @deftypefun void file_end_indicate_exec_stack ()
7187 Some systems use a common convention, the @samp{.note.GNU-stack}
7188 special section, to indicate whether or not an object file relies on
7189 the stack being executable. If your system uses this convention, you
7190 should define @code{TARGET_ASM_FILE_END} to this function. If you
7191 need to do other things in that hook, have your hook function call
7195 @deftypefn {Target Hook} void TARGET_ASM_LTO_START (void)
7196 Output to @code{asm_out_file} any text which the assembler expects
7197 to find at the start of an LTO section. The default is to output
7201 @deftypefn {Target Hook} void TARGET_ASM_LTO_END (void)
7202 Output to @code{asm_out_file} any text which the assembler expects
7203 to find at the end of an LTO section. The default is to output
7207 @deftypefn {Target Hook} void TARGET_ASM_CODE_END (void)
7208 Output to @code{asm_out_file} any text which is needed before emitting
7209 unwind info and debug info at the end of a file. Some targets emit
7210 here PIC setup thunks that cannot be emitted at the end of file,
7211 because they couldn't have unwind info then. The default is to output
7215 @defmac ASM_COMMENT_START
7216 A C string constant describing how to begin a comment in the target
7217 assembler language. The compiler assumes that the comment will end at
7218 the end of the line.
7222 A C string constant for text to be output before each @code{asm}
7223 statement or group of consecutive ones. Normally this is
7224 @code{"#APP"}, which is a comment that has no effect on most
7225 assemblers but tells the GNU assembler that it must check the lines
7226 that follow for all valid assembler constructs.
7230 A C string constant for text to be output after each @code{asm}
7231 statement or group of consecutive ones. Normally this is
7232 @code{"#NO_APP"}, which tells the GNU assembler to resume making the
7233 time-saving assumptions that are valid for ordinary compiler output.
7236 @defmac ASM_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name})
7237 A C statement to output COFF information or DWARF debugging information
7238 which indicates that filename @var{name} is the current source file to
7239 the stdio stream @var{stream}.
7241 This macro need not be defined if the standard form of output
7242 for the file format in use is appropriate.
7245 @defmac OUTPUT_QUOTED_STRING (@var{stream}, @var{string})
7246 A C statement to output the string @var{string} to the stdio stream
7247 @var{stream}. If you do not call the function @code{output_quoted_string}
7248 in your config files, GCC will only call it to output filenames to
7249 the assembler source. So you can use it to canonicalize the format
7250 of the filename using this macro.
7253 @defmac ASM_OUTPUT_IDENT (@var{stream}, @var{string})
7254 A C statement to output something to the assembler file to handle a
7255 @samp{#ident} directive containing the text @var{string}. If this
7256 macro is not defined, nothing is output for a @samp{#ident} directive.
7259 @deftypefn {Target Hook} void TARGET_ASM_NAMED_SECTION (const char *@var{name}, unsigned int @var{flags}, tree @var{decl})
7260 Output assembly directives to switch to section @var{name}. The section
7261 should have attributes as specified by @var{flags}, which is a bit mask
7262 of the @code{SECTION_*} flags defined in @file{output.h}. If @var{decl}
7263 is non-NULL, it is the @code{VAR_DECL} or @code{FUNCTION_DECL} with which
7264 this section is associated.
7267 @deftypevr {Target Hook} bool TARGET_HAVE_NAMED_SECTIONS
7268 This flag is true if the target supports @code{TARGET_ASM_NAMED_SECTION}.
7271 @anchor{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}
7272 @deftypevr {Target Hook} bool TARGET_HAVE_SWITCHABLE_BSS_SECTIONS
7273 This flag is true if we can create zeroed data by switching to a BSS
7274 section and then using @code{ASM_OUTPUT_SKIP} to allocate the space.
7275 This is true on most ELF targets.
7278 @deftypefn {Target Hook} {unsigned int} TARGET_SECTION_TYPE_FLAGS (tree @var{decl}, const char *@var{name}, int @var{reloc})
7279 Choose a set of section attributes for use by @code{TARGET_ASM_NAMED_SECTION}
7280 based on a variable or function decl, a section name, and whether or not the
7281 declaration's initializer may contain runtime relocations. @var{decl} may be
7282 null, in which case read-write data should be assumed.
7284 The default version of this function handles choosing code vs data,
7285 read-only vs read-write data, and @code{flag_pic}. You should only
7286 need to override this if your target has special flags that might be
7287 set via @code{__attribute__}.
7290 @deftypefn {Target Hook} int TARGET_ASM_RECORD_GCC_SWITCHES (print_switch_type @var{type}, const char *@var{text})
7291 Provides the target with the ability to record the gcc command line
7292 switches that have been passed to the compiler, and options that are
7293 enabled. The @var{type} argument specifies what is being recorded.
7294 It can take the following values:
7297 @item SWITCH_TYPE_PASSED
7298 @var{text} is a command line switch that has been set by the user.
7300 @item SWITCH_TYPE_ENABLED
7301 @var{text} is an option which has been enabled. This might be as a
7302 direct result of a command line switch, or because it is enabled by
7303 default or because it has been enabled as a side effect of a different
7304 command line switch. For example, the @option{-O2} switch enables
7305 various different individual optimization passes.
7307 @item SWITCH_TYPE_DESCRIPTIVE
7308 @var{text} is either NULL or some descriptive text which should be
7309 ignored. If @var{text} is NULL then it is being used to warn the
7310 target hook that either recording is starting or ending. The first
7311 time @var{type} is SWITCH_TYPE_DESCRIPTIVE and @var{text} is NULL, the
7312 warning is for start up and the second time the warning is for
7313 wind down. This feature is to allow the target hook to make any
7314 necessary preparations before it starts to record switches and to
7315 perform any necessary tidying up after it has finished recording
7318 @item SWITCH_TYPE_LINE_START
7319 This option can be ignored by this target hook.
7321 @item SWITCH_TYPE_LINE_END
7322 This option can be ignored by this target hook.
7325 The hook's return value must be zero. Other return values may be
7326 supported in the future.
7328 By default this hook is set to NULL, but an example implementation is
7329 provided for ELF based targets. Called @var{elf_record_gcc_switches},
7330 it records the switches as ASCII text inside a new, string mergeable
7331 section in the assembler output file. The name of the new section is
7332 provided by the @code{TARGET_ASM_RECORD_GCC_SWITCHES_SECTION} target
7336 @deftypevr {Target Hook} {const char *} TARGET_ASM_RECORD_GCC_SWITCHES_SECTION
7337 This is the name of the section that will be created by the example
7338 ELF implementation of the @code{TARGET_ASM_RECORD_GCC_SWITCHES} target
7344 @subsection Output of Data
7347 @deftypevr {Target Hook} {const char *} TARGET_ASM_BYTE_OP
7348 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_HI_OP
7349 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_SI_OP
7350 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_DI_OP
7351 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_TI_OP
7352 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_HI_OP
7353 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_SI_OP
7354 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_DI_OP
7355 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_TI_OP
7356 These hooks specify assembly directives for creating certain kinds
7357 of integer object. The @code{TARGET_ASM_BYTE_OP} directive creates a
7358 byte-sized object, the @code{TARGET_ASM_ALIGNED_HI_OP} one creates an
7359 aligned two-byte object, and so on. Any of the hooks may be
7360 @code{NULL}, indicating that no suitable directive is available.
7362 The compiler will print these strings at the start of a new line,
7363 followed immediately by the object's initial value. In most cases,
7364 the string should contain a tab, a pseudo-op, and then another tab.
7367 @deftypefn {Target Hook} bool TARGET_ASM_INTEGER (rtx @var{x}, unsigned int @var{size}, int @var{aligned_p})
7368 The @code{assemble_integer} function uses this hook to output an
7369 integer object. @var{x} is the object's value, @var{size} is its size
7370 in bytes and @var{aligned_p} indicates whether it is aligned. The
7371 function should return @code{true} if it was able to output the
7372 object. If it returns false, @code{assemble_integer} will try to
7373 split the object into smaller parts.
7375 The default implementation of this hook will use the
7376 @code{TARGET_ASM_BYTE_OP} family of strings, returning @code{false}
7377 when the relevant string is @code{NULL}.
7380 @defmac OUTPUT_ADDR_CONST_EXTRA (@var{stream}, @var{x}, @var{fail})
7381 A C statement to recognize @var{rtx} patterns that
7382 @code{output_addr_const} can't deal with, and output assembly code to
7383 @var{stream} corresponding to the pattern @var{x}. This may be used to
7384 allow machine-dependent @code{UNSPEC}s to appear within constants.
7386 If @code{OUTPUT_ADDR_CONST_EXTRA} fails to recognize a pattern, it must
7387 @code{goto fail}, so that a standard error message is printed. If it
7388 prints an error message itself, by calling, for example,
7389 @code{output_operand_lossage}, it may just complete normally.
7392 @defmac ASM_OUTPUT_ASCII (@var{stream}, @var{ptr}, @var{len})
7393 A C statement to output to the stdio stream @var{stream} an assembler
7394 instruction to assemble a string constant containing the @var{len}
7395 bytes at @var{ptr}. @var{ptr} will be a C expression of type
7396 @code{char *} and @var{len} a C expression of type @code{int}.
7398 If the assembler has a @code{.ascii} pseudo-op as found in the
7399 Berkeley Unix assembler, do not define the macro
7400 @code{ASM_OUTPUT_ASCII}.
7403 @defmac ASM_OUTPUT_FDESC (@var{stream}, @var{decl}, @var{n})
7404 A C statement to output word @var{n} of a function descriptor for
7405 @var{decl}. This must be defined if @code{TARGET_VTABLE_USES_DESCRIPTORS}
7406 is defined, and is otherwise unused.
7409 @defmac CONSTANT_POOL_BEFORE_FUNCTION
7410 You may define this macro as a C expression. You should define the
7411 expression to have a nonzero value if GCC should output the constant
7412 pool for a function before the code for the function, or a zero value if
7413 GCC should output the constant pool after the function. If you do
7414 not define this macro, the usual case, GCC will output the constant
7415 pool before the function.
7418 @defmac ASM_OUTPUT_POOL_PROLOGUE (@var{file}, @var{funname}, @var{fundecl}, @var{size})
7419 A C statement to output assembler commands to define the start of the
7420 constant pool for a function. @var{funname} is a string giving
7421 the name of the function. Should the return type of the function
7422 be required, it can be obtained via @var{fundecl}. @var{size}
7423 is the size, in bytes, of the constant pool that will be written
7424 immediately after this call.
7426 If no constant-pool prefix is required, the usual case, this macro need
7430 @defmac ASM_OUTPUT_SPECIAL_POOL_ENTRY (@var{file}, @var{x}, @var{mode}, @var{align}, @var{labelno}, @var{jumpto})
7431 A C statement (with or without semicolon) to output a constant in the
7432 constant pool, if it needs special treatment. (This macro need not do
7433 anything for RTL expressions that can be output normally.)
7435 The argument @var{file} is the standard I/O stream to output the
7436 assembler code on. @var{x} is the RTL expression for the constant to
7437 output, and @var{mode} is the machine mode (in case @var{x} is a
7438 @samp{const_int}). @var{align} is the required alignment for the value
7439 @var{x}; you should output an assembler directive to force this much
7442 The argument @var{labelno} is a number to use in an internal label for
7443 the address of this pool entry. The definition of this macro is
7444 responsible for outputting the label definition at the proper place.
7445 Here is how to do this:
7448 @code{(*targetm.asm_out.internal_label)} (@var{file}, "LC", @var{labelno});
7451 When you output a pool entry specially, you should end with a
7452 @code{goto} to the label @var{jumpto}. This will prevent the same pool
7453 entry from being output a second time in the usual manner.
7455 You need not define this macro if it would do nothing.
7458 @defmac ASM_OUTPUT_POOL_EPILOGUE (@var{file} @var{funname} @var{fundecl} @var{size})
7459 A C statement to output assembler commands to at the end of the constant
7460 pool for a function. @var{funname} is a string giving the name of the
7461 function. Should the return type of the function be required, you can
7462 obtain it via @var{fundecl}. @var{size} is the size, in bytes, of the
7463 constant pool that GCC wrote immediately before this call.
7465 If no constant-pool epilogue is required, the usual case, you need not
7469 @defmac IS_ASM_LOGICAL_LINE_SEPARATOR (@var{C}, @var{STR})
7470 Define this macro as a C expression which is nonzero if @var{C} is
7471 used as a logical line separator by the assembler. @var{STR} points
7472 to the position in the string where @var{C} was found; this can be used if
7473 a line separator uses multiple characters.
7475 If you do not define this macro, the default is that only
7476 the character @samp{;} is treated as a logical line separator.
7479 @deftypevr {Target Hook} {const char *} TARGET_ASM_OPEN_PAREN
7480 @deftypevrx {Target Hook} {const char *} TARGET_ASM_CLOSE_PAREN
7481 These target hooks are C string constants, describing the syntax in the
7482 assembler for grouping arithmetic expressions. If not overridden, they
7483 default to normal parentheses, which is correct for most assemblers.
7486 These macros are provided by @file{real.h} for writing the definitions
7487 of @code{ASM_OUTPUT_DOUBLE} and the like:
7489 @defmac REAL_VALUE_TO_TARGET_SINGLE (@var{x}, @var{l})
7490 @defmacx REAL_VALUE_TO_TARGET_DOUBLE (@var{x}, @var{l})
7491 @defmacx REAL_VALUE_TO_TARGET_LONG_DOUBLE (@var{x}, @var{l})
7492 @defmacx REAL_VALUE_TO_TARGET_DECIMAL32 (@var{x}, @var{l})
7493 @defmacx REAL_VALUE_TO_TARGET_DECIMAL64 (@var{x}, @var{l})
7494 @defmacx REAL_VALUE_TO_TARGET_DECIMAL128 (@var{x}, @var{l})
7495 These translate @var{x}, of type @code{REAL_VALUE_TYPE}, to the
7496 target's floating point representation, and store its bit pattern in
7497 the variable @var{l}. For @code{REAL_VALUE_TO_TARGET_SINGLE} and
7498 @code{REAL_VALUE_TO_TARGET_DECIMAL32}, this variable should be a
7499 simple @code{long int}. For the others, it should be an array of
7500 @code{long int}. The number of elements in this array is determined
7501 by the size of the desired target floating point data type: 32 bits of
7502 it go in each @code{long int} array element. Each array element holds
7503 32 bits of the result, even if @code{long int} is wider than 32 bits
7504 on the host machine.
7506 The array element values are designed so that you can print them out
7507 using @code{fprintf} in the order they should appear in the target
7511 @node Uninitialized Data
7512 @subsection Output of Uninitialized Variables
7514 Each of the macros in this section is used to do the whole job of
7515 outputting a single uninitialized variable.
7517 @defmac ASM_OUTPUT_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
7518 A C statement (sans semicolon) to output to the stdio stream
7519 @var{stream} the assembler definition of a common-label named
7520 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7521 is the size rounded up to whatever alignment the caller wants. It is
7522 possible that @var{size} may be zero, for instance if a struct with no
7523 other member than a zero-length array is defined. In this case, the
7524 backend must output a symbol definition that allocates at least one
7525 byte, both so that the address of the resulting object does not compare
7526 equal to any other, and because some object formats cannot even express
7527 the concept of a zero-sized common symbol, as that is how they represent
7528 an ordinary undefined external.
7530 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7531 output the name itself; before and after that, output the additional
7532 assembler syntax for defining the name, and a newline.
7534 This macro controls how the assembler definitions of uninitialized
7535 common global variables are output.
7538 @defmac ASM_OUTPUT_ALIGNED_COMMON (@var{stream}, @var{name}, @var{size}, @var{alignment})
7539 Like @code{ASM_OUTPUT_COMMON} except takes the required alignment as a
7540 separate, explicit argument. If you define this macro, it is used in
7541 place of @code{ASM_OUTPUT_COMMON}, and gives you more flexibility in
7542 handling the required alignment of the variable. The alignment is specified
7543 as the number of bits.
7546 @defmac ASM_OUTPUT_ALIGNED_DECL_COMMON (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7547 Like @code{ASM_OUTPUT_ALIGNED_COMMON} except that @var{decl} of the
7548 variable to be output, if there is one, or @code{NULL_TREE} if there
7549 is no corresponding variable. If you define this macro, GCC will use it
7550 in place of both @code{ASM_OUTPUT_COMMON} and
7551 @code{ASM_OUTPUT_ALIGNED_COMMON}. Define this macro when you need to see
7552 the variable's decl in order to chose what to output.
7555 @defmac ASM_OUTPUT_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{rounded})
7556 A C statement (sans semicolon) to output to the stdio stream
7557 @var{stream} the assembler definition of uninitialized global @var{decl} named
7558 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7559 is the size rounded up to whatever alignment the caller wants.
7561 Try to use function @code{asm_output_bss} defined in @file{varasm.c} when
7562 defining this macro. If unable, use the expression
7563 @code{assemble_name (@var{stream}, @var{name})} to output the name itself;
7564 before and after that, output the additional assembler syntax for defining
7565 the name, and a newline.
7567 There are two ways of handling global BSS@. One is to define either
7568 this macro or its aligned counterpart, @code{ASM_OUTPUT_ALIGNED_BSS}.
7569 The other is to have @code{TARGET_ASM_SELECT_SECTION} return a
7570 switchable BSS section (@pxref{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}).
7571 You do not need to do both.
7573 Some languages do not have @code{common} data, and require a
7574 non-common form of global BSS in order to handle uninitialized globals
7575 efficiently. C++ is one example of this. However, if the target does
7576 not support global BSS, the front end may choose to make globals
7577 common in order to save space in the object file.
7580 @defmac ASM_OUTPUT_ALIGNED_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7581 Like @code{ASM_OUTPUT_BSS} except takes the required alignment as a
7582 separate, explicit argument. If you define this macro, it is used in
7583 place of @code{ASM_OUTPUT_BSS}, and gives you more flexibility in
7584 handling the required alignment of the variable. The alignment is specified
7585 as the number of bits.
7587 Try to use function @code{asm_output_aligned_bss} defined in file
7588 @file{varasm.c} when defining this macro.
7591 @defmac ASM_OUTPUT_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded})
7592 A C statement (sans semicolon) to output to the stdio stream
7593 @var{stream} the assembler definition of a local-common-label named
7594 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7595 is the size rounded up to whatever alignment the caller wants.
7597 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7598 output the name itself; before and after that, output the additional
7599 assembler syntax for defining the name, and a newline.
7601 This macro controls how the assembler definitions of uninitialized
7602 static variables are output.
7605 @defmac ASM_OUTPUT_ALIGNED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{alignment})
7606 Like @code{ASM_OUTPUT_LOCAL} except takes the required alignment as a
7607 separate, explicit argument. If you define this macro, it is used in
7608 place of @code{ASM_OUTPUT_LOCAL}, and gives you more flexibility in
7609 handling the required alignment of the variable. The alignment is specified
7610 as the number of bits.
7613 @defmac ASM_OUTPUT_ALIGNED_DECL_LOCAL (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7614 Like @code{ASM_OUTPUT_ALIGNED_DECL} except that @var{decl} of the
7615 variable to be output, if there is one, or @code{NULL_TREE} if there
7616 is no corresponding variable. If you define this macro, GCC will use it
7617 in place of both @code{ASM_OUTPUT_DECL} and
7618 @code{ASM_OUTPUT_ALIGNED_DECL}. Define this macro when you need to see
7619 the variable's decl in order to chose what to output.
7623 @subsection Output and Generation of Labels
7625 @c prevent bad page break with this line
7626 This is about outputting labels.
7628 @findex assemble_name
7629 @defmac ASM_OUTPUT_LABEL (@var{stream}, @var{name})
7630 A C statement (sans semicolon) to output to the stdio stream
7631 @var{stream} the assembler definition of a label named @var{name}.
7632 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7633 output the name itself; before and after that, output the additional
7634 assembler syntax for defining the name, and a newline. A default
7635 definition of this macro is provided which is correct for most systems.
7638 @findex assemble_name_raw
7639 @defmac ASM_OUTPUT_INTERNAL_LABEL (@var{stream}, @var{name})
7640 Identical to @code{ASM_OUTPUT_LABEL}, except that @var{name} is known
7641 to refer to a compiler-generated label. The default definition uses
7642 @code{assemble_name_raw}, which is like @code{assemble_name} except
7643 that it is more efficient.
7647 A C string containing the appropriate assembler directive to specify the
7648 size of a symbol, without any arguments. On systems that use ELF, the
7649 default (in @file{config/elfos.h}) is @samp{"\t.size\t"}; on other
7650 systems, the default is not to define this macro.
7652 Define this macro only if it is correct to use the default definitions
7653 of @code{ASM_OUTPUT_SIZE_DIRECTIVE} and @code{ASM_OUTPUT_MEASURED_SIZE}
7654 for your system. If you need your own custom definitions of those
7655 macros, or if you do not need explicit symbol sizes at all, do not
7659 @defmac ASM_OUTPUT_SIZE_DIRECTIVE (@var{stream}, @var{name}, @var{size})
7660 A C statement (sans semicolon) to output to the stdio stream
7661 @var{stream} a directive telling the assembler that the size of the
7662 symbol @var{name} is @var{size}. @var{size} is a @code{HOST_WIDE_INT}.
7663 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
7667 @defmac ASM_OUTPUT_MEASURED_SIZE (@var{stream}, @var{name})
7668 A C statement (sans semicolon) to output to the stdio stream
7669 @var{stream} a directive telling the assembler to calculate the size of
7670 the symbol @var{name} by subtracting its address from the current
7673 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
7674 provided. The default assumes that the assembler recognizes a special
7675 @samp{.} symbol as referring to the current address, and can calculate
7676 the difference between this and another symbol. If your assembler does
7677 not recognize @samp{.} or cannot do calculations with it, you will need
7678 to redefine @code{ASM_OUTPUT_MEASURED_SIZE} to use some other technique.
7682 A C string containing the appropriate assembler directive to specify the
7683 type of a symbol, without any arguments. On systems that use ELF, the
7684 default (in @file{config/elfos.h}) is @samp{"\t.type\t"}; on other
7685 systems, the default is not to define this macro.
7687 Define this macro only if it is correct to use the default definition of
7688 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
7689 custom definition of this macro, or if you do not need explicit symbol
7690 types at all, do not define this macro.
7693 @defmac TYPE_OPERAND_FMT
7694 A C string which specifies (using @code{printf} syntax) the format of
7695 the second operand to @code{TYPE_ASM_OP}. On systems that use ELF, the
7696 default (in @file{config/elfos.h}) is @samp{"@@%s"}; on other systems,
7697 the default is not to define this macro.
7699 Define this macro only if it is correct to use the default definition of
7700 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
7701 custom definition of this macro, or if you do not need explicit symbol
7702 types at all, do not define this macro.
7705 @defmac ASM_OUTPUT_TYPE_DIRECTIVE (@var{stream}, @var{type})
7706 A C statement (sans semicolon) to output to the stdio stream
7707 @var{stream} a directive telling the assembler that the type of the
7708 symbol @var{name} is @var{type}. @var{type} is a C string; currently,
7709 that string is always either @samp{"function"} or @samp{"object"}, but
7710 you should not count on this.
7712 If you define @code{TYPE_ASM_OP} and @code{TYPE_OPERAND_FMT}, a default
7713 definition of this macro is provided.
7716 @defmac ASM_DECLARE_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl})
7717 A C statement (sans semicolon) to output to the stdio stream
7718 @var{stream} any text necessary for declaring the name @var{name} of a
7719 function which is being defined. This macro is responsible for
7720 outputting the label definition (perhaps using
7721 @code{ASM_OUTPUT_LABEL}). The argument @var{decl} is the
7722 @code{FUNCTION_DECL} tree node representing the function.
7724 If this macro is not defined, then the function name is defined in the
7725 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
7727 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition
7731 @defmac ASM_DECLARE_FUNCTION_SIZE (@var{stream}, @var{name}, @var{decl})
7732 A C statement (sans semicolon) to output to the stdio stream
7733 @var{stream} any text necessary for declaring the size of a function
7734 which is being defined. The argument @var{name} is the name of the
7735 function. The argument @var{decl} is the @code{FUNCTION_DECL} tree node
7736 representing the function.
7738 If this macro is not defined, then the function size is not defined.
7740 You may wish to use @code{ASM_OUTPUT_MEASURED_SIZE} in the definition
7744 @defmac ASM_DECLARE_OBJECT_NAME (@var{stream}, @var{name}, @var{decl})
7745 A C statement (sans semicolon) to output to the stdio stream
7746 @var{stream} any text necessary for declaring the name @var{name} of an
7747 initialized variable which is being defined. This macro must output the
7748 label definition (perhaps using @code{ASM_OUTPUT_LABEL}). The argument
7749 @var{decl} is the @code{VAR_DECL} tree node representing the variable.
7751 If this macro is not defined, then the variable name is defined in the
7752 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
7754 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} and/or
7755 @code{ASM_OUTPUT_SIZE_DIRECTIVE} in the definition of this macro.
7758 @deftypefn {Target Hook} void TARGET_ASM_DECLARE_CONSTANT_NAME (FILE * @var{file}, const char * @var{name}, const_tree @var{expr}, HOST_WIDE_INT @var{size})
7759 A target hook to output to the stdio stream @var{file} any text necessary
7760 for declaring the name @var{name} of a constant which is being defined. This
7761 target hook is responsible for outputting the label definition (perhaps using
7762 @code{assemble_label}). The argument @var{exp} is the value of the constant,
7763 and @var{size} is the size of the constant in bytes. The @var{name}
7764 will be an internal label.
7766 The default version of this target hook, define the @var{name} in the
7767 usual manner as a label (by means of @code{assemble_label}).
7769 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in this target hook.
7772 @defmac ASM_DECLARE_REGISTER_GLOBAL (@var{stream}, @var{decl}, @var{regno}, @var{name})
7773 A C statement (sans semicolon) to output to the stdio stream
7774 @var{stream} any text necessary for claiming a register @var{regno}
7775 for a global variable @var{decl} with name @var{name}.
7777 If you don't define this macro, that is equivalent to defining it to do
7781 @defmac ASM_FINISH_DECLARE_OBJECT (@var{stream}, @var{decl}, @var{toplevel}, @var{atend})
7782 A C statement (sans semicolon) to finish up declaring a variable name
7783 once the compiler has processed its initializer fully and thus has had a
7784 chance to determine the size of an array when controlled by an
7785 initializer. This is used on systems where it's necessary to declare
7786 something about the size of the object.
7788 If you don't define this macro, that is equivalent to defining it to do
7791 You may wish to use @code{ASM_OUTPUT_SIZE_DIRECTIVE} and/or
7792 @code{ASM_OUTPUT_MEASURED_SIZE} in the definition of this macro.
7795 @deftypefn {Target Hook} void TARGET_ASM_GLOBALIZE_LABEL (FILE *@var{stream}, const char *@var{name})
7796 This target hook is a function to output to the stdio stream
7797 @var{stream} some commands that will make the label @var{name} global;
7798 that is, available for reference from other files.
7800 The default implementation relies on a proper definition of
7801 @code{GLOBAL_ASM_OP}.
7804 @deftypefn {Target Hook} void TARGET_ASM_GLOBALIZE_DECL_NAME (FILE *@var{stream}, tree @var{decl})
7805 This target hook is a function to output to the stdio stream
7806 @var{stream} some commands that will make the name associated with @var{decl}
7807 global; that is, available for reference from other files.
7809 The default implementation uses the TARGET_ASM_GLOBALIZE_LABEL target hook.
7812 @defmac ASM_WEAKEN_LABEL (@var{stream}, @var{name})
7813 A C statement (sans semicolon) to output to the stdio stream
7814 @var{stream} some commands that will make the label @var{name} weak;
7815 that is, available for reference from other files but only used if
7816 no other definition is available. Use the expression
7817 @code{assemble_name (@var{stream}, @var{name})} to output the name
7818 itself; before and after that, output the additional assembler syntax
7819 for making that name weak, and a newline.
7821 If you don't define this macro or @code{ASM_WEAKEN_DECL}, GCC will not
7822 support weak symbols and you should not define the @code{SUPPORTS_WEAK}
7826 @defmac ASM_WEAKEN_DECL (@var{stream}, @var{decl}, @var{name}, @var{value})
7827 Combines (and replaces) the function of @code{ASM_WEAKEN_LABEL} and
7828 @code{ASM_OUTPUT_WEAK_ALIAS}, allowing access to the associated function
7829 or variable decl. If @var{value} is not @code{NULL}, this C statement
7830 should output to the stdio stream @var{stream} assembler code which
7831 defines (equates) the weak symbol @var{name} to have the value
7832 @var{value}. If @var{value} is @code{NULL}, it should output commands
7833 to make @var{name} weak.
7836 @defmac ASM_OUTPUT_WEAKREF (@var{stream}, @var{decl}, @var{name}, @var{value})
7837 Outputs a directive that enables @var{name} to be used to refer to
7838 symbol @var{value} with weak-symbol semantics. @code{decl} is the
7839 declaration of @code{name}.
7842 @defmac SUPPORTS_WEAK
7843 A C expression which evaluates to true if the target supports weak symbols.
7845 If you don't define this macro, @file{defaults.h} provides a default
7846 definition. If either @code{ASM_WEAKEN_LABEL} or @code{ASM_WEAKEN_DECL}
7847 is defined, the default definition is @samp{1}; otherwise, it is
7848 @samp{0}. Define this macro if you want to control weak symbol support
7849 with a compiler flag such as @option{-melf}.
7852 @defmac MAKE_DECL_ONE_ONLY (@var{decl})
7853 A C statement (sans semicolon) to mark @var{decl} to be emitted as a
7854 public symbol such that extra copies in multiple translation units will
7855 be discarded by the linker. Define this macro if your object file
7856 format provides support for this concept, such as the @samp{COMDAT}
7857 section flags in the Microsoft Windows PE/COFF format, and this support
7858 requires changes to @var{decl}, such as putting it in a separate section.
7861 @defmac SUPPORTS_ONE_ONLY
7862 A C expression which evaluates to true if the target supports one-only
7865 If you don't define this macro, @file{varasm.c} provides a default
7866 definition. If @code{MAKE_DECL_ONE_ONLY} is defined, the default
7867 definition is @samp{1}; otherwise, it is @samp{0}. Define this macro if
7868 you want to control one-only symbol support with a compiler flag, or if
7869 setting the @code{DECL_ONE_ONLY} flag is enough to mark a declaration to
7870 be emitted as one-only.
7873 @deftypefn {Target Hook} void TARGET_ASM_ASSEMBLE_VISIBILITY (tree @var{decl}, int @var{visibility})
7874 This target hook is a function to output to @var{asm_out_file} some
7875 commands that will make the symbol(s) associated with @var{decl} have
7876 hidden, protected or internal visibility as specified by @var{visibility}.
7879 @defmac TARGET_WEAK_NOT_IN_ARCHIVE_TOC
7880 A C expression that evaluates to true if the target's linker expects
7881 that weak symbols do not appear in a static archive's table of contents.
7882 The default is @code{0}.
7884 Leaving weak symbols out of an archive's table of contents means that,
7885 if a symbol will only have a definition in one translation unit and
7886 will have undefined references from other translation units, that
7887 symbol should not be weak. Defining this macro to be nonzero will
7888 thus have the effect that certain symbols that would normally be weak
7889 (explicit template instantiations, and vtables for polymorphic classes
7890 with noninline key methods) will instead be nonweak.
7892 The C++ ABI requires this macro to be zero. Define this macro for
7893 targets where full C++ ABI compliance is impossible and where linker
7894 restrictions require weak symbols to be left out of a static archive's
7898 @defmac ASM_OUTPUT_EXTERNAL (@var{stream}, @var{decl}, @var{name})
7899 A C statement (sans semicolon) to output to the stdio stream
7900 @var{stream} any text necessary for declaring the name of an external
7901 symbol named @var{name} which is referenced in this compilation but
7902 not defined. The value of @var{decl} is the tree node for the
7905 This macro need not be defined if it does not need to output anything.
7906 The GNU assembler and most Unix assemblers don't require anything.
7909 @deftypefn {Target Hook} void TARGET_ASM_EXTERNAL_LIBCALL (rtx @var{symref})
7910 This target hook is a function to output to @var{asm_out_file} an assembler
7911 pseudo-op to declare a library function name external. The name of the
7912 library function is given by @var{symref}, which is a @code{symbol_ref}.
7915 @deftypefn {Target Hook} void TARGET_ASM_MARK_DECL_PRESERVED (const char *@var{symbol})
7916 This target hook is a function to output to @var{asm_out_file} an assembler
7917 directive to annotate @var{symbol} as used. The Darwin target uses the
7918 .no_dead_code_strip directive.
7921 @defmac ASM_OUTPUT_LABELREF (@var{stream}, @var{name})
7922 A C statement (sans semicolon) to output to the stdio stream
7923 @var{stream} a reference in assembler syntax to a label named
7924 @var{name}. This should add @samp{_} to the front of the name, if that
7925 is customary on your operating system, as it is in most Berkeley Unix
7926 systems. This macro is used in @code{assemble_name}.
7929 @defmac ASM_OUTPUT_SYMBOL_REF (@var{stream}, @var{sym})
7930 A C statement (sans semicolon) to output a reference to
7931 @code{SYMBOL_REF} @var{sym}. If not defined, @code{assemble_name}
7932 will be used to output the name of the symbol. This macro may be used
7933 to modify the way a symbol is referenced depending on information
7934 encoded by @code{TARGET_ENCODE_SECTION_INFO}.
7937 @defmac ASM_OUTPUT_LABEL_REF (@var{stream}, @var{buf})
7938 A C statement (sans semicolon) to output a reference to @var{buf}, the
7939 result of @code{ASM_GENERATE_INTERNAL_LABEL}. If not defined,
7940 @code{assemble_name} will be used to output the name of the symbol.
7941 This macro is not used by @code{output_asm_label}, or the @code{%l}
7942 specifier that calls it; the intention is that this macro should be set
7943 when it is necessary to output a label differently when its address is
7947 @deftypefn {Target Hook} void TARGET_ASM_INTERNAL_LABEL (FILE *@var{stream}, const char *@var{prefix}, unsigned long @var{labelno})
7948 A function to output to the stdio stream @var{stream} a label whose
7949 name is made from the string @var{prefix} and the number @var{labelno}.
7951 It is absolutely essential that these labels be distinct from the labels
7952 used for user-level functions and variables. Otherwise, certain programs
7953 will have name conflicts with internal labels.
7955 It is desirable to exclude internal labels from the symbol table of the
7956 object file. Most assemblers have a naming convention for labels that
7957 should be excluded; on many systems, the letter @samp{L} at the
7958 beginning of a label has this effect. You should find out what
7959 convention your system uses, and follow it.
7961 The default version of this function utilizes @code{ASM_GENERATE_INTERNAL_LABEL}.
7964 @defmac ASM_OUTPUT_DEBUG_LABEL (@var{stream}, @var{prefix}, @var{num})
7965 A C statement to output to the stdio stream @var{stream} a debug info
7966 label whose name is made from the string @var{prefix} and the number
7967 @var{num}. This is useful for VLIW targets, where debug info labels
7968 may need to be treated differently than branch target labels. On some
7969 systems, branch target labels must be at the beginning of instruction
7970 bundles, but debug info labels can occur in the middle of instruction
7973 If this macro is not defined, then @code{(*targetm.asm_out.internal_label)} will be
7977 @defmac ASM_GENERATE_INTERNAL_LABEL (@var{string}, @var{prefix}, @var{num})
7978 A C statement to store into the string @var{string} a label whose name
7979 is made from the string @var{prefix} and the number @var{num}.
7981 This string, when output subsequently by @code{assemble_name}, should
7982 produce the output that @code{(*targetm.asm_out.internal_label)} would produce
7983 with the same @var{prefix} and @var{num}.
7985 If the string begins with @samp{*}, then @code{assemble_name} will
7986 output the rest of the string unchanged. It is often convenient for
7987 @code{ASM_GENERATE_INTERNAL_LABEL} to use @samp{*} in this way. If the
7988 string doesn't start with @samp{*}, then @code{ASM_OUTPUT_LABELREF} gets
7989 to output the string, and may change it. (Of course,
7990 @code{ASM_OUTPUT_LABELREF} is also part of your machine description, so
7991 you should know what it does on your machine.)
7994 @defmac ASM_FORMAT_PRIVATE_NAME (@var{outvar}, @var{name}, @var{number})
7995 A C expression to assign to @var{outvar} (which is a variable of type
7996 @code{char *}) a newly allocated string made from the string
7997 @var{name} and the number @var{number}, with some suitable punctuation
7998 added. Use @code{alloca} to get space for the string.
8000 The string will be used as an argument to @code{ASM_OUTPUT_LABELREF} to
8001 produce an assembler label for an internal static variable whose name is
8002 @var{name}. Therefore, the string must be such as to result in valid
8003 assembler code. The argument @var{number} is different each time this
8004 macro is executed; it prevents conflicts between similarly-named
8005 internal static variables in different scopes.
8007 Ideally this string should not be a valid C identifier, to prevent any
8008 conflict with the user's own symbols. Most assemblers allow periods
8009 or percent signs in assembler symbols; putting at least one of these
8010 between the name and the number will suffice.
8012 If this macro is not defined, a default definition will be provided
8013 which is correct for most systems.
8016 @defmac ASM_OUTPUT_DEF (@var{stream}, @var{name}, @var{value})
8017 A C statement to output to the stdio stream @var{stream} assembler code
8018 which defines (equates) the symbol @var{name} to have the value @var{value}.
8021 If @code{SET_ASM_OP} is defined, a default definition is provided which is
8022 correct for most systems.
8025 @defmac ASM_OUTPUT_DEF_FROM_DECLS (@var{stream}, @var{decl_of_name}, @var{decl_of_value})
8026 A C statement to output to the stdio stream @var{stream} assembler code
8027 which defines (equates) the symbol whose tree node is @var{decl_of_name}
8028 to have the value of the tree node @var{decl_of_value}. This macro will
8029 be used in preference to @samp{ASM_OUTPUT_DEF} if it is defined and if
8030 the tree nodes are available.
8033 If @code{SET_ASM_OP} is defined, a default definition is provided which is
8034 correct for most systems.
8037 @defmac TARGET_DEFERRED_OUTPUT_DEFS (@var{decl_of_name}, @var{decl_of_value})
8038 A C statement that evaluates to true if the assembler code which defines
8039 (equates) the symbol whose tree node is @var{decl_of_name} to have the value
8040 of the tree node @var{decl_of_value} should be emitted near the end of the
8041 current compilation unit. The default is to not defer output of defines.
8042 This macro affects defines output by @samp{ASM_OUTPUT_DEF} and
8043 @samp{ASM_OUTPUT_DEF_FROM_DECLS}.
8046 @defmac ASM_OUTPUT_WEAK_ALIAS (@var{stream}, @var{name}, @var{value})
8047 A C statement to output to the stdio stream @var{stream} assembler code
8048 which defines (equates) the weak symbol @var{name} to have the value
8049 @var{value}. If @var{value} is @code{NULL}, it defines @var{name} as
8050 an undefined weak symbol.
8052 Define this macro if the target only supports weak aliases; define
8053 @code{ASM_OUTPUT_DEF} instead if possible.
8056 @defmac OBJC_GEN_METHOD_LABEL (@var{buf}, @var{is_inst}, @var{class_name}, @var{cat_name}, @var{sel_name})
8057 Define this macro to override the default assembler names used for
8058 Objective-C methods.
8060 The default name is a unique method number followed by the name of the
8061 class (e.g.@: @samp{_1_Foo}). For methods in categories, the name of
8062 the category is also included in the assembler name (e.g.@:
8065 These names are safe on most systems, but make debugging difficult since
8066 the method's selector is not present in the name. Therefore, particular
8067 systems define other ways of computing names.
8069 @var{buf} is an expression of type @code{char *} which gives you a
8070 buffer in which to store the name; its length is as long as
8071 @var{class_name}, @var{cat_name} and @var{sel_name} put together, plus
8072 50 characters extra.
8074 The argument @var{is_inst} specifies whether the method is an instance
8075 method or a class method; @var{class_name} is the name of the class;
8076 @var{cat_name} is the name of the category (or @code{NULL} if the method is not
8077 in a category); and @var{sel_name} is the name of the selector.
8079 On systems where the assembler can handle quoted names, you can use this
8080 macro to provide more human-readable names.
8083 @defmac ASM_DECLARE_CLASS_REFERENCE (@var{stream}, @var{name})
8084 A C statement (sans semicolon) to output to the stdio stream
8085 @var{stream} commands to declare that the label @var{name} is an
8086 Objective-C class reference. This is only needed for targets whose
8087 linkers have special support for NeXT-style runtimes.
8090 @defmac ASM_DECLARE_UNRESOLVED_REFERENCE (@var{stream}, @var{name})
8091 A C statement (sans semicolon) to output to the stdio stream
8092 @var{stream} commands to declare that the label @var{name} is an
8093 unresolved Objective-C class reference. This is only needed for targets
8094 whose linkers have special support for NeXT-style runtimes.
8097 @node Initialization
8098 @subsection How Initialization Functions Are Handled
8099 @cindex initialization routines
8100 @cindex termination routines
8101 @cindex constructors, output of
8102 @cindex destructors, output of
8104 The compiled code for certain languages includes @dfn{constructors}
8105 (also called @dfn{initialization routines})---functions to initialize
8106 data in the program when the program is started. These functions need
8107 to be called before the program is ``started''---that is to say, before
8108 @code{main} is called.
8110 Compiling some languages generates @dfn{destructors} (also called
8111 @dfn{termination routines}) that should be called when the program
8114 To make the initialization and termination functions work, the compiler
8115 must output something in the assembler code to cause those functions to
8116 be called at the appropriate time. When you port the compiler to a new
8117 system, you need to specify how to do this.
8119 There are two major ways that GCC currently supports the execution of
8120 initialization and termination functions. Each way has two variants.
8121 Much of the structure is common to all four variations.
8123 @findex __CTOR_LIST__
8124 @findex __DTOR_LIST__
8125 The linker must build two lists of these functions---a list of
8126 initialization functions, called @code{__CTOR_LIST__}, and a list of
8127 termination functions, called @code{__DTOR_LIST__}.
8129 Each list always begins with an ignored function pointer (which may hold
8130 0, @minus{}1, or a count of the function pointers after it, depending on
8131 the environment). This is followed by a series of zero or more function
8132 pointers to constructors (or destructors), followed by a function
8133 pointer containing zero.
8135 Depending on the operating system and its executable file format, either
8136 @file{crtstuff.c} or @file{libgcc2.c} traverses these lists at startup
8137 time and exit time. Constructors are called in reverse order of the
8138 list; destructors in forward order.
8140 The best way to handle static constructors works only for object file
8141 formats which provide arbitrarily-named sections. A section is set
8142 aside for a list of constructors, and another for a list of destructors.
8143 Traditionally these are called @samp{.ctors} and @samp{.dtors}. Each
8144 object file that defines an initialization function also puts a word in
8145 the constructor section to point to that function. The linker
8146 accumulates all these words into one contiguous @samp{.ctors} section.
8147 Termination functions are handled similarly.
8149 This method will be chosen as the default by @file{target-def.h} if
8150 @code{TARGET_ASM_NAMED_SECTION} is defined. A target that does not
8151 support arbitrary sections, but does support special designated
8152 constructor and destructor sections may define @code{CTORS_SECTION_ASM_OP}
8153 and @code{DTORS_SECTION_ASM_OP} to achieve the same effect.
8155 When arbitrary sections are available, there are two variants, depending
8156 upon how the code in @file{crtstuff.c} is called. On systems that
8157 support a @dfn{.init} section which is executed at program startup,
8158 parts of @file{crtstuff.c} are compiled into that section. The
8159 program is linked by the @command{gcc} driver like this:
8162 ld -o @var{output_file} crti.o crtbegin.o @dots{} -lgcc crtend.o crtn.o
8165 The prologue of a function (@code{__init}) appears in the @code{.init}
8166 section of @file{crti.o}; the epilogue appears in @file{crtn.o}. Likewise
8167 for the function @code{__fini} in the @dfn{.fini} section. Normally these
8168 files are provided by the operating system or by the GNU C library, but
8169 are provided by GCC for a few targets.
8171 The objects @file{crtbegin.o} and @file{crtend.o} are (for most targets)
8172 compiled from @file{crtstuff.c}. They contain, among other things, code
8173 fragments within the @code{.init} and @code{.fini} sections that branch
8174 to routines in the @code{.text} section. The linker will pull all parts
8175 of a section together, which results in a complete @code{__init} function
8176 that invokes the routines we need at startup.
8178 To use this variant, you must define the @code{INIT_SECTION_ASM_OP}
8181 If no init section is available, when GCC compiles any function called
8182 @code{main} (or more accurately, any function designated as a program
8183 entry point by the language front end calling @code{expand_main_function}),
8184 it inserts a procedure call to @code{__main} as the first executable code
8185 after the function prologue. The @code{__main} function is defined
8186 in @file{libgcc2.c} and runs the global constructors.
8188 In file formats that don't support arbitrary sections, there are again
8189 two variants. In the simplest variant, the GNU linker (GNU @code{ld})
8190 and an `a.out' format must be used. In this case,
8191 @code{TARGET_ASM_CONSTRUCTOR} is defined to produce a @code{.stabs}
8192 entry of type @samp{N_SETT}, referencing the name @code{__CTOR_LIST__},
8193 and with the address of the void function containing the initialization
8194 code as its value. The GNU linker recognizes this as a request to add
8195 the value to a @dfn{set}; the values are accumulated, and are eventually
8196 placed in the executable as a vector in the format described above, with
8197 a leading (ignored) count and a trailing zero element.
8198 @code{TARGET_ASM_DESTRUCTOR} is handled similarly. Since no init
8199 section is available, the absence of @code{INIT_SECTION_ASM_OP} causes
8200 the compilation of @code{main} to call @code{__main} as above, starting
8201 the initialization process.
8203 The last variant uses neither arbitrary sections nor the GNU linker.
8204 This is preferable when you want to do dynamic linking and when using
8205 file formats which the GNU linker does not support, such as `ECOFF'@. In
8206 this case, @code{TARGET_HAVE_CTORS_DTORS} is false, initialization and
8207 termination functions are recognized simply by their names. This requires
8208 an extra program in the linkage step, called @command{collect2}. This program
8209 pretends to be the linker, for use with GCC; it does its job by running
8210 the ordinary linker, but also arranges to include the vectors of
8211 initialization and termination functions. These functions are called
8212 via @code{__main} as described above. In order to use this method,
8213 @code{use_collect2} must be defined in the target in @file{config.gcc}.
8216 The following section describes the specific macros that control and
8217 customize the handling of initialization and termination functions.
8220 @node Macros for Initialization
8221 @subsection Macros Controlling Initialization Routines
8223 Here are the macros that control how the compiler handles initialization
8224 and termination functions:
8226 @defmac INIT_SECTION_ASM_OP
8227 If defined, a C string constant, including spacing, for the assembler
8228 operation to identify the following data as initialization code. If not
8229 defined, GCC will assume such a section does not exist. When you are
8230 using special sections for initialization and termination functions, this
8231 macro also controls how @file{crtstuff.c} and @file{libgcc2.c} arrange to
8232 run the initialization functions.
8235 @defmac HAS_INIT_SECTION
8236 If defined, @code{main} will not call @code{__main} as described above.
8237 This macro should be defined for systems that control start-up code
8238 on a symbol-by-symbol basis, such as OSF/1, and should not
8239 be defined explicitly for systems that support @code{INIT_SECTION_ASM_OP}.
8242 @defmac LD_INIT_SWITCH
8243 If defined, a C string constant for a switch that tells the linker that
8244 the following symbol is an initialization routine.
8247 @defmac LD_FINI_SWITCH
8248 If defined, a C string constant for a switch that tells the linker that
8249 the following symbol is a finalization routine.
8252 @defmac COLLECT_SHARED_INIT_FUNC (@var{stream}, @var{func})
8253 If defined, a C statement that will write a function that can be
8254 automatically called when a shared library is loaded. The function
8255 should call @var{func}, which takes no arguments. If not defined, and
8256 the object format requires an explicit initialization function, then a
8257 function called @code{_GLOBAL__DI} will be generated.
8259 This function and the following one are used by collect2 when linking a
8260 shared library that needs constructors or destructors, or has DWARF2
8261 exception tables embedded in the code.
8264 @defmac COLLECT_SHARED_FINI_FUNC (@var{stream}, @var{func})
8265 If defined, a C statement that will write a function that can be
8266 automatically called when a shared library is unloaded. The function
8267 should call @var{func}, which takes no arguments. If not defined, and
8268 the object format requires an explicit finalization function, then a
8269 function called @code{_GLOBAL__DD} will be generated.
8272 @defmac INVOKE__main
8273 If defined, @code{main} will call @code{__main} despite the presence of
8274 @code{INIT_SECTION_ASM_OP}. This macro should be defined for systems
8275 where the init section is not actually run automatically, but is still
8276 useful for collecting the lists of constructors and destructors.
8279 @defmac SUPPORTS_INIT_PRIORITY
8280 If nonzero, the C++ @code{init_priority} attribute is supported and the
8281 compiler should emit instructions to control the order of initialization
8282 of objects. If zero, the compiler will issue an error message upon
8283 encountering an @code{init_priority} attribute.
8286 @deftypevr {Target Hook} bool TARGET_HAVE_CTORS_DTORS
8287 This value is true if the target supports some ``native'' method of
8288 collecting constructors and destructors to be run at startup and exit.
8289 It is false if we must use @command{collect2}.
8292 @deftypefn {Target Hook} void TARGET_ASM_CONSTRUCTOR (rtx @var{symbol}, int @var{priority})
8293 If defined, a function that outputs assembler code to arrange to call
8294 the function referenced by @var{symbol} at initialization time.
8296 Assume that @var{symbol} is a @code{SYMBOL_REF} for a function taking
8297 no arguments and with no return value. If the target supports initialization
8298 priorities, @var{priority} is a value between 0 and @code{MAX_INIT_PRIORITY};
8299 otherwise it must be @code{DEFAULT_INIT_PRIORITY}.
8301 If this macro is not defined by the target, a suitable default will
8302 be chosen if (1) the target supports arbitrary section names, (2) the
8303 target defines @code{CTORS_SECTION_ASM_OP}, or (3) @code{USE_COLLECT2}
8307 @deftypefn {Target Hook} void TARGET_ASM_DESTRUCTOR (rtx @var{symbol}, int @var{priority})
8308 This is like @code{TARGET_ASM_CONSTRUCTOR} but used for termination
8309 functions rather than initialization functions.
8312 If @code{TARGET_HAVE_CTORS_DTORS} is true, the initialization routine
8313 generated for the generated object file will have static linkage.
8315 If your system uses @command{collect2} as the means of processing
8316 constructors, then that program normally uses @command{nm} to scan
8317 an object file for constructor functions to be called.
8319 On certain kinds of systems, you can define this macro to make
8320 @command{collect2} work faster (and, in some cases, make it work at all):
8322 @defmac OBJECT_FORMAT_COFF
8323 Define this macro if the system uses COFF (Common Object File Format)
8324 object files, so that @command{collect2} can assume this format and scan
8325 object files directly for dynamic constructor/destructor functions.
8327 This macro is effective only in a native compiler; @command{collect2} as
8328 part of a cross compiler always uses @command{nm} for the target machine.
8331 @defmac REAL_NM_FILE_NAME
8332 Define this macro as a C string constant containing the file name to use
8333 to execute @command{nm}. The default is to search the path normally for
8336 If your system supports shared libraries and has a program to list the
8337 dynamic dependencies of a given library or executable, you can define
8338 these macros to enable support for running initialization and
8339 termination functions in shared libraries:
8343 Define this macro to a C string constant containing the name of the program
8344 which lists dynamic dependencies, like @command{"ldd"} under SunOS 4.
8347 @defmac PARSE_LDD_OUTPUT (@var{ptr})
8348 Define this macro to be C code that extracts filenames from the output
8349 of the program denoted by @code{LDD_SUFFIX}. @var{ptr} is a variable
8350 of type @code{char *} that points to the beginning of a line of output
8351 from @code{LDD_SUFFIX}. If the line lists a dynamic dependency, the
8352 code must advance @var{ptr} to the beginning of the filename on that
8353 line. Otherwise, it must set @var{ptr} to @code{NULL}.
8356 @defmac SHLIB_SUFFIX
8357 Define this macro to a C string constant containing the default shared
8358 library extension of the target (e.g., @samp{".so"}). @command{collect2}
8359 strips version information after this suffix when generating global
8360 constructor and destructor names. This define is only needed on targets
8361 that use @command{collect2} to process constructors and destructors.
8364 @node Instruction Output
8365 @subsection Output of Assembler Instructions
8367 @c prevent bad page break with this line
8368 This describes assembler instruction output.
8370 @defmac REGISTER_NAMES
8371 A C initializer containing the assembler's names for the machine
8372 registers, each one as a C string constant. This is what translates
8373 register numbers in the compiler into assembler language.
8376 @defmac ADDITIONAL_REGISTER_NAMES
8377 If defined, a C initializer for an array of structures containing a name
8378 and a register number. This macro defines additional names for hard
8379 registers, thus allowing the @code{asm} option in declarations to refer
8380 to registers using alternate names.
8383 @defmac ASM_OUTPUT_OPCODE (@var{stream}, @var{ptr})
8384 Define this macro if you are using an unusual assembler that
8385 requires different names for the machine instructions.
8387 The definition is a C statement or statements which output an
8388 assembler instruction opcode to the stdio stream @var{stream}. The
8389 macro-operand @var{ptr} is a variable of type @code{char *} which
8390 points to the opcode name in its ``internal'' form---the form that is
8391 written in the machine description. The definition should output the
8392 opcode name to @var{stream}, performing any translation you desire, and
8393 increment the variable @var{ptr} to point at the end of the opcode
8394 so that it will not be output twice.
8396 In fact, your macro definition may process less than the entire opcode
8397 name, or more than the opcode name; but if you want to process text
8398 that includes @samp{%}-sequences to substitute operands, you must take
8399 care of the substitution yourself. Just be sure to increment
8400 @var{ptr} over whatever text should not be output normally.
8402 @findex recog_data.operand
8403 If you need to look at the operand values, they can be found as the
8404 elements of @code{recog_data.operand}.
8406 If the macro definition does nothing, the instruction is output
8410 @defmac FINAL_PRESCAN_INSN (@var{insn}, @var{opvec}, @var{noperands})
8411 If defined, a C statement to be executed just prior to the output of
8412 assembler code for @var{insn}, to modify the extracted operands so
8413 they will be output differently.
8415 Here the argument @var{opvec} is the vector containing the operands
8416 extracted from @var{insn}, and @var{noperands} is the number of
8417 elements of the vector which contain meaningful data for this insn.
8418 The contents of this vector are what will be used to convert the insn
8419 template into assembler code, so you can change the assembler output
8420 by changing the contents of the vector.
8422 This macro is useful when various assembler syntaxes share a single
8423 file of instruction patterns; by defining this macro differently, you
8424 can cause a large class of instructions to be output differently (such
8425 as with rearranged operands). Naturally, variations in assembler
8426 syntax affecting individual insn patterns ought to be handled by
8427 writing conditional output routines in those patterns.
8429 If this macro is not defined, it is equivalent to a null statement.
8432 @deftypefn {Target Hook} void TARGET_ASM_FINAL_POSTSCAN_INSN (FILE *@var{file}, rtx @var{insn}, rtx *@var{opvec}, int @var{noperands})
8433 If defined, this target hook is a function which is executed just after the
8434 output of assembler code for @var{insn}, to change the mode of the assembler
8437 Here the argument @var{opvec} is the vector containing the operands
8438 extracted from @var{insn}, and @var{noperands} is the number of
8439 elements of the vector which contain meaningful data for this insn.
8440 The contents of this vector are what was used to convert the insn
8441 template into assembler code, so you can change the assembler mode
8442 by checking the contents of the vector.
8445 @defmac PRINT_OPERAND (@var{stream}, @var{x}, @var{code})
8446 A C compound statement to output to stdio stream @var{stream} the
8447 assembler syntax for an instruction operand @var{x}. @var{x} is an
8450 @var{code} is a value that can be used to specify one of several ways
8451 of printing the operand. It is used when identical operands must be
8452 printed differently depending on the context. @var{code} comes from
8453 the @samp{%} specification that was used to request printing of the
8454 operand. If the specification was just @samp{%@var{digit}} then
8455 @var{code} is 0; if the specification was @samp{%@var{ltr}
8456 @var{digit}} then @var{code} is the ASCII code for @var{ltr}.
8459 If @var{x} is a register, this macro should print the register's name.
8460 The names can be found in an array @code{reg_names} whose type is
8461 @code{char *[]}. @code{reg_names} is initialized from
8462 @code{REGISTER_NAMES}.
8464 When the machine description has a specification @samp{%@var{punct}}
8465 (a @samp{%} followed by a punctuation character), this macro is called
8466 with a null pointer for @var{x} and the punctuation character for
8470 @defmac PRINT_OPERAND_PUNCT_VALID_P (@var{code})
8471 A C expression which evaluates to true if @var{code} is a valid
8472 punctuation character for use in the @code{PRINT_OPERAND} macro. If
8473 @code{PRINT_OPERAND_PUNCT_VALID_P} is not defined, it means that no
8474 punctuation characters (except for the standard one, @samp{%}) are used
8478 @defmac PRINT_OPERAND_ADDRESS (@var{stream}, @var{x})
8479 A C compound statement to output to stdio stream @var{stream} the
8480 assembler syntax for an instruction operand that is a memory reference
8481 whose address is @var{x}. @var{x} is an RTL expression.
8483 @cindex @code{TARGET_ENCODE_SECTION_INFO} usage
8484 On some machines, the syntax for a symbolic address depends on the
8485 section that the address refers to. On these machines, define the hook
8486 @code{TARGET_ENCODE_SECTION_INFO} to store the information into the
8487 @code{symbol_ref}, and then check for it here. @xref{Assembler
8491 @findex dbr_sequence_length
8492 @defmac DBR_OUTPUT_SEQEND (@var{file})
8493 A C statement, to be executed after all slot-filler instructions have
8494 been output. If necessary, call @code{dbr_sequence_length} to
8495 determine the number of slots filled in a sequence (zero if not
8496 currently outputting a sequence), to decide how many no-ops to output,
8499 Don't define this macro if it has nothing to do, but it is helpful in
8500 reading assembly output if the extent of the delay sequence is made
8501 explicit (e.g.@: with white space).
8504 @findex final_sequence
8505 Note that output routines for instructions with delay slots must be
8506 prepared to deal with not being output as part of a sequence
8507 (i.e.@: when the scheduling pass is not run, or when no slot fillers could be
8508 found.) The variable @code{final_sequence} is null when not
8509 processing a sequence, otherwise it contains the @code{sequence} rtx
8513 @defmac REGISTER_PREFIX
8514 @defmacx LOCAL_LABEL_PREFIX
8515 @defmacx USER_LABEL_PREFIX
8516 @defmacx IMMEDIATE_PREFIX
8517 If defined, C string expressions to be used for the @samp{%R}, @samp{%L},
8518 @samp{%U}, and @samp{%I} options of @code{asm_fprintf} (see
8519 @file{final.c}). These are useful when a single @file{md} file must
8520 support multiple assembler formats. In that case, the various @file{tm.h}
8521 files can define these macros differently.
8524 @defmac ASM_FPRINTF_EXTENSIONS (@var{file}, @var{argptr}, @var{format})
8525 If defined this macro should expand to a series of @code{case}
8526 statements which will be parsed inside the @code{switch} statement of
8527 the @code{asm_fprintf} function. This allows targets to define extra
8528 printf formats which may useful when generating their assembler
8529 statements. Note that uppercase letters are reserved for future
8530 generic extensions to asm_fprintf, and so are not available to target
8531 specific code. The output file is given by the parameter @var{file}.
8532 The varargs input pointer is @var{argptr} and the rest of the format
8533 string, starting the character after the one that is being switched
8534 upon, is pointed to by @var{format}.
8537 @defmac ASSEMBLER_DIALECT
8538 If your target supports multiple dialects of assembler language (such as
8539 different opcodes), define this macro as a C expression that gives the
8540 numeric index of the assembler language dialect to use, with zero as the
8543 If this macro is defined, you may use constructs of the form
8545 @samp{@{option0|option1|option2@dots{}@}}
8548 in the output templates of patterns (@pxref{Output Template}) or in the
8549 first argument of @code{asm_fprintf}. This construct outputs
8550 @samp{option0}, @samp{option1}, @samp{option2}, etc., if the value of
8551 @code{ASSEMBLER_DIALECT} is zero, one, two, etc. Any special characters
8552 within these strings retain their usual meaning. If there are fewer
8553 alternatives within the braces than the value of
8554 @code{ASSEMBLER_DIALECT}, the construct outputs nothing.
8556 If you do not define this macro, the characters @samp{@{}, @samp{|} and
8557 @samp{@}} do not have any special meaning when used in templates or
8558 operands to @code{asm_fprintf}.
8560 Define the macros @code{REGISTER_PREFIX}, @code{LOCAL_LABEL_PREFIX},
8561 @code{USER_LABEL_PREFIX} and @code{IMMEDIATE_PREFIX} if you can express
8562 the variations in assembler language syntax with that mechanism. Define
8563 @code{ASSEMBLER_DIALECT} and use the @samp{@{option0|option1@}} syntax
8564 if the syntax variant are larger and involve such things as different
8565 opcodes or operand order.
8568 @defmac ASM_OUTPUT_REG_PUSH (@var{stream}, @var{regno})
8569 A C expression to output to @var{stream} some assembler code
8570 which will push hard register number @var{regno} onto the stack.
8571 The code need not be optimal, since this macro is used only when
8575 @defmac ASM_OUTPUT_REG_POP (@var{stream}, @var{regno})
8576 A C expression to output to @var{stream} some assembler code
8577 which will pop hard register number @var{regno} off of the stack.
8578 The code need not be optimal, since this macro is used only when
8582 @node Dispatch Tables
8583 @subsection Output of Dispatch Tables
8585 @c prevent bad page break with this line
8586 This concerns dispatch tables.
8588 @cindex dispatch table
8589 @defmac ASM_OUTPUT_ADDR_DIFF_ELT (@var{stream}, @var{body}, @var{value}, @var{rel})
8590 A C statement to output to the stdio stream @var{stream} an assembler
8591 pseudo-instruction to generate a difference between two labels.
8592 @var{value} and @var{rel} are the numbers of two internal labels. The
8593 definitions of these labels are output using
8594 @code{(*targetm.asm_out.internal_label)}, and they must be printed in the same
8595 way here. For example,
8598 fprintf (@var{stream}, "\t.word L%d-L%d\n",
8599 @var{value}, @var{rel})
8602 You must provide this macro on machines where the addresses in a
8603 dispatch table are relative to the table's own address. If defined, GCC
8604 will also use this macro on all machines when producing PIC@.
8605 @var{body} is the body of the @code{ADDR_DIFF_VEC}; it is provided so that the
8606 mode and flags can be read.
8609 @defmac ASM_OUTPUT_ADDR_VEC_ELT (@var{stream}, @var{value})
8610 This macro should be provided on machines where the addresses
8611 in a dispatch table are absolute.
8613 The definition should be a C statement to output to the stdio stream
8614 @var{stream} an assembler pseudo-instruction to generate a reference to
8615 a label. @var{value} is the number of an internal label whose
8616 definition is output using @code{(*targetm.asm_out.internal_label)}.
8620 fprintf (@var{stream}, "\t.word L%d\n", @var{value})
8624 @defmac ASM_OUTPUT_CASE_LABEL (@var{stream}, @var{prefix}, @var{num}, @var{table})
8625 Define this if the label before a jump-table needs to be output
8626 specially. The first three arguments are the same as for
8627 @code{(*targetm.asm_out.internal_label)}; the fourth argument is the
8628 jump-table which follows (a @code{jump_insn} containing an
8629 @code{addr_vec} or @code{addr_diff_vec}).
8631 This feature is used on system V to output a @code{swbeg} statement
8634 If this macro is not defined, these labels are output with
8635 @code{(*targetm.asm_out.internal_label)}.
8638 @defmac ASM_OUTPUT_CASE_END (@var{stream}, @var{num}, @var{table})
8639 Define this if something special must be output at the end of a
8640 jump-table. The definition should be a C statement to be executed
8641 after the assembler code for the table is written. It should write
8642 the appropriate code to stdio stream @var{stream}. The argument
8643 @var{table} is the jump-table insn, and @var{num} is the label-number
8644 of the preceding label.
8646 If this macro is not defined, nothing special is output at the end of
8650 @deftypefn {Target Hook} void TARGET_ASM_EMIT_UNWIND_LABEL (FILE *@var{stream}, tree @var{decl}, int @var{for_eh}, int @var{empty})
8651 This target hook emits a label at the beginning of each FDE@. It
8652 should be defined on targets where FDEs need special labels, and it
8653 should write the appropriate label, for the FDE associated with the
8654 function declaration @var{decl}, to the stdio stream @var{stream}.
8655 The third argument, @var{for_eh}, is a boolean: true if this is for an
8656 exception table. The fourth argument, @var{empty}, is a boolean:
8657 true if this is a placeholder label for an omitted FDE@.
8659 The default is that FDEs are not given nonlocal labels.
8662 @deftypefn {Target Hook} void TARGET_ASM_EMIT_EXCEPT_TABLE_LABEL (FILE *@var{stream})
8663 This target hook emits a label at the beginning of the exception table.
8664 It should be defined on targets where it is desirable for the table
8665 to be broken up according to function.
8667 The default is that no label is emitted.
8670 @deftypefn {Target Hook} void TARGET_UNWIND_EMIT (FILE *@var{stream}, rtx @var{insn})
8671 This target hook emits assembly directives required to unwind the
8672 given instruction. This is only used when TARGET_UNWIND_INFO is set.
8675 @node Exception Region Output
8676 @subsection Assembler Commands for Exception Regions
8678 @c prevent bad page break with this line
8680 This describes commands marking the start and the end of an exception
8683 @defmac EH_FRAME_SECTION_NAME
8684 If defined, a C string constant for the name of the section containing
8685 exception handling frame unwind information. If not defined, GCC will
8686 provide a default definition if the target supports named sections.
8687 @file{crtstuff.c} uses this macro to switch to the appropriate section.
8689 You should define this symbol if your target supports DWARF 2 frame
8690 unwind information and the default definition does not work.
8693 @defmac EH_FRAME_IN_DATA_SECTION
8694 If defined, DWARF 2 frame unwind information will be placed in the
8695 data section even though the target supports named sections. This
8696 might be necessary, for instance, if the system linker does garbage
8697 collection and sections cannot be marked as not to be collected.
8699 Do not define this macro unless @code{TARGET_ASM_NAMED_SECTION} is
8703 @defmac EH_TABLES_CAN_BE_READ_ONLY
8704 Define this macro to 1 if your target is such that no frame unwind
8705 information encoding used with non-PIC code will ever require a
8706 runtime relocation, but the linker may not support merging read-only
8707 and read-write sections into a single read-write section.
8710 @defmac MASK_RETURN_ADDR
8711 An rtx used to mask the return address found via @code{RETURN_ADDR_RTX}, so
8712 that it does not contain any extraneous set bits in it.
8715 @defmac DWARF2_UNWIND_INFO
8716 Define this macro to 0 if your target supports DWARF 2 frame unwind
8717 information, but it does not yet work with exception handling.
8718 Otherwise, if your target supports this information (if it defines
8719 @samp{INCOMING_RETURN_ADDR_RTX} and either @samp{UNALIGNED_INT_ASM_OP}
8720 or @samp{OBJECT_FORMAT_ELF}), GCC will provide a default definition of 1.
8722 If @code{TARGET_UNWIND_INFO} is defined, the target specific unwinder
8723 will be used in all cases. Defining this macro will enable the generation
8724 of DWARF 2 frame debugging information.
8726 If @code{TARGET_UNWIND_INFO} is not defined, and this macro is defined to 1,
8727 the DWARF 2 unwinder will be the default exception handling mechanism;
8728 otherwise, the @code{setjmp}/@code{longjmp}-based scheme will be used by
8732 @defmac TARGET_UNWIND_INFO
8733 Define this macro if your target has ABI specified unwind tables. Usually
8734 these will be output by @code{TARGET_UNWIND_EMIT}.
8737 @deftypevr {Target Hook} bool TARGET_UNWIND_TABLES_DEFAULT
8738 This variable should be set to @code{true} if the target ABI requires unwinding
8739 tables even when exceptions are not used.
8742 @defmac MUST_USE_SJLJ_EXCEPTIONS
8743 This macro need only be defined if @code{DWARF2_UNWIND_INFO} is
8744 runtime-variable. In that case, @file{except.h} cannot correctly
8745 determine the corresponding definition of @code{MUST_USE_SJLJ_EXCEPTIONS},
8746 so the target must provide it directly.
8749 @defmac DONT_USE_BUILTIN_SETJMP
8750 Define this macro to 1 if the @code{setjmp}/@code{longjmp}-based scheme
8751 should use the @code{setjmp}/@code{longjmp} functions from the C library
8752 instead of the @code{__builtin_setjmp}/@code{__builtin_longjmp} machinery.
8755 @defmac DWARF_CIE_DATA_ALIGNMENT
8756 This macro need only be defined if the target might save registers in the
8757 function prologue at an offset to the stack pointer that is not aligned to
8758 @code{UNITS_PER_WORD}. The definition should be the negative minimum
8759 alignment if @code{STACK_GROWS_DOWNWARD} is defined, and the positive
8760 minimum alignment otherwise. @xref{SDB and DWARF}. Only applicable if
8761 the target supports DWARF 2 frame unwind information.
8764 @deftypevr {Target Hook} bool TARGET_TERMINATE_DW2_EH_FRAME_INFO
8765 Contains the value true if the target should add a zero word onto the
8766 end of a Dwarf-2 frame info section when used for exception handling.
8767 Default value is false if @code{EH_FRAME_SECTION_NAME} is defined, and
8771 @deftypefn {Target Hook} rtx TARGET_DWARF_REGISTER_SPAN (rtx @var{reg})
8772 Given a register, this hook should return a parallel of registers to
8773 represent where to find the register pieces. Define this hook if the
8774 register and its mode are represented in Dwarf in non-contiguous
8775 locations, or if the register should be represented in more than one
8776 register in Dwarf. Otherwise, this hook should return @code{NULL_RTX}.
8777 If not defined, the default is to return @code{NULL_RTX}.
8780 @deftypefn {Target Hook} void TARGET_INIT_DWARF_REG_SIZES_EXTRA (tree @var{address})
8781 If some registers are represented in Dwarf-2 unwind information in
8782 multiple pieces, define this hook to fill in information about the
8783 sizes of those pieces in the table used by the unwinder at runtime.
8784 It will be called by @code{expand_builtin_init_dwarf_reg_sizes} after
8785 filling in a single size corresponding to each hard register;
8786 @var{address} is the address of the table.
8789 @deftypefn {Target Hook} bool TARGET_ASM_TTYPE (rtx @var{sym})
8790 This hook is used to output a reference from a frame unwinding table to
8791 the type_info object identified by @var{sym}. It should return @code{true}
8792 if the reference was output. Returning @code{false} will cause the
8793 reference to be output using the normal Dwarf2 routines.
8796 @deftypevr {Target Hook} bool TARGET_ARM_EABI_UNWINDER
8797 This flag should be set to @code{true} on targets that use an ARM EABI
8798 based unwinding library, and @code{false} on other targets. This effects
8799 the format of unwinding tables, and how the unwinder in entered after
8800 running a cleanup. The default is @code{false}.
8803 @node Alignment Output
8804 @subsection Assembler Commands for Alignment
8806 @c prevent bad page break with this line
8807 This describes commands for alignment.
8809 @defmac JUMP_ALIGN (@var{label})
8810 The alignment (log base 2) to put in front of @var{label}, which is
8811 a common destination of jumps and has no fallthru incoming edge.
8813 This macro need not be defined if you don't want any special alignment
8814 to be done at such a time. Most machine descriptions do not currently
8817 Unless it's necessary to inspect the @var{label} parameter, it is better
8818 to set the variable @var{align_jumps} in the target's
8819 @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's
8820 selection in @var{align_jumps} in a @code{JUMP_ALIGN} implementation.
8823 @defmac LABEL_ALIGN_AFTER_BARRIER (@var{label})
8824 The alignment (log base 2) to put in front of @var{label}, which follows
8827 This macro need not be defined if you don't want any special alignment
8828 to be done at such a time. Most machine descriptions do not currently
8832 @defmac LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP
8833 The maximum number of bytes to skip when applying
8834 @code{LABEL_ALIGN_AFTER_BARRIER}. This works only if
8835 @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
8838 @defmac LOOP_ALIGN (@var{label})
8839 The alignment (log base 2) to put in front of @var{label}, which follows
8840 a @code{NOTE_INSN_LOOP_BEG} note.
8842 This macro need not be defined if you don't want any special alignment
8843 to be done at such a time. Most machine descriptions do not currently
8846 Unless it's necessary to inspect the @var{label} parameter, it is better
8847 to set the variable @code{align_loops} in the target's
8848 @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's
8849 selection in @code{align_loops} in a @code{LOOP_ALIGN} implementation.
8852 @defmac LOOP_ALIGN_MAX_SKIP
8853 The maximum number of bytes to skip when applying @code{LOOP_ALIGN}.
8854 This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
8857 @defmac LABEL_ALIGN (@var{label})
8858 The alignment (log base 2) to put in front of @var{label}.
8859 If @code{LABEL_ALIGN_AFTER_BARRIER} / @code{LOOP_ALIGN} specify a different alignment,
8860 the maximum of the specified values is used.
8862 Unless it's necessary to inspect the @var{label} parameter, it is better
8863 to set the variable @code{align_labels} in the target's
8864 @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's
8865 selection in @code{align_labels} in a @code{LABEL_ALIGN} implementation.
8868 @defmac LABEL_ALIGN_MAX_SKIP
8869 The maximum number of bytes to skip when applying @code{LABEL_ALIGN}.
8870 This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
8873 @defmac ASM_OUTPUT_SKIP (@var{stream}, @var{nbytes})
8874 A C statement to output to the stdio stream @var{stream} an assembler
8875 instruction to advance the location counter by @var{nbytes} bytes.
8876 Those bytes should be zero when loaded. @var{nbytes} will be a C
8877 expression of type @code{unsigned HOST_WIDE_INT}.
8880 @defmac ASM_NO_SKIP_IN_TEXT
8881 Define this macro if @code{ASM_OUTPUT_SKIP} should not be used in the
8882 text section because it fails to put zeros in the bytes that are skipped.
8883 This is true on many Unix systems, where the pseudo--op to skip bytes
8884 produces no-op instructions rather than zeros when used in the text
8888 @defmac ASM_OUTPUT_ALIGN (@var{stream}, @var{power})
8889 A C statement to output to the stdio stream @var{stream} an assembler
8890 command to advance the location counter to a multiple of 2 to the
8891 @var{power} bytes. @var{power} will be a C expression of type @code{int}.
8894 @defmac ASM_OUTPUT_ALIGN_WITH_NOP (@var{stream}, @var{power})
8895 Like @code{ASM_OUTPUT_ALIGN}, except that the ``nop'' instruction is used
8896 for padding, if necessary.
8899 @defmac ASM_OUTPUT_MAX_SKIP_ALIGN (@var{stream}, @var{power}, @var{max_skip})
8900 A C statement to output to the stdio stream @var{stream} an assembler
8901 command to advance the location counter to a multiple of 2 to the
8902 @var{power} bytes, but only if @var{max_skip} or fewer bytes are needed to
8903 satisfy the alignment request. @var{power} and @var{max_skip} will be
8904 a C expression of type @code{int}.
8908 @node Debugging Info
8909 @section Controlling Debugging Information Format
8911 @c prevent bad page break with this line
8912 This describes how to specify debugging information.
8915 * All Debuggers:: Macros that affect all debugging formats uniformly.
8916 * DBX Options:: Macros enabling specific options in DBX format.
8917 * DBX Hooks:: Hook macros for varying DBX format.
8918 * File Names and DBX:: Macros controlling output of file names in DBX format.
8919 * SDB and DWARF:: Macros for SDB (COFF) and DWARF formats.
8920 * VMS Debug:: Macros for VMS debug format.
8924 @subsection Macros Affecting All Debugging Formats
8926 @c prevent bad page break with this line
8927 These macros affect all debugging formats.
8929 @defmac DBX_REGISTER_NUMBER (@var{regno})
8930 A C expression that returns the DBX register number for the compiler
8931 register number @var{regno}. In the default macro provided, the value
8932 of this expression will be @var{regno} itself. But sometimes there are
8933 some registers that the compiler knows about and DBX does not, or vice
8934 versa. In such cases, some register may need to have one number in the
8935 compiler and another for DBX@.
8937 If two registers have consecutive numbers inside GCC, and they can be
8938 used as a pair to hold a multiword value, then they @emph{must} have
8939 consecutive numbers after renumbering with @code{DBX_REGISTER_NUMBER}.
8940 Otherwise, debuggers will be unable to access such a pair, because they
8941 expect register pairs to be consecutive in their own numbering scheme.
8943 If you find yourself defining @code{DBX_REGISTER_NUMBER} in way that
8944 does not preserve register pairs, then what you must do instead is
8945 redefine the actual register numbering scheme.
8948 @defmac DEBUGGER_AUTO_OFFSET (@var{x})
8949 A C expression that returns the integer offset value for an automatic
8950 variable having address @var{x} (an RTL expression). The default
8951 computation assumes that @var{x} is based on the frame-pointer and
8952 gives the offset from the frame-pointer. This is required for targets
8953 that produce debugging output for DBX or COFF-style debugging output
8954 for SDB and allow the frame-pointer to be eliminated when the
8955 @option{-g} options is used.
8958 @defmac DEBUGGER_ARG_OFFSET (@var{offset}, @var{x})
8959 A C expression that returns the integer offset value for an argument
8960 having address @var{x} (an RTL expression). The nominal offset is
8964 @defmac PREFERRED_DEBUGGING_TYPE
8965 A C expression that returns the type of debugging output GCC should
8966 produce when the user specifies just @option{-g}. Define
8967 this if you have arranged for GCC to support more than one format of
8968 debugging output. Currently, the allowable values are @code{DBX_DEBUG},
8969 @code{SDB_DEBUG}, @code{DWARF_DEBUG}, @code{DWARF2_DEBUG},
8970 @code{XCOFF_DEBUG}, @code{VMS_DEBUG}, and @code{VMS_AND_DWARF2_DEBUG}.
8972 When the user specifies @option{-ggdb}, GCC normally also uses the
8973 value of this macro to select the debugging output format, but with two
8974 exceptions. If @code{DWARF2_DEBUGGING_INFO} is defined, GCC uses the
8975 value @code{DWARF2_DEBUG}. Otherwise, if @code{DBX_DEBUGGING_INFO} is
8976 defined, GCC uses @code{DBX_DEBUG}.
8978 The value of this macro only affects the default debugging output; the
8979 user can always get a specific type of output by using @option{-gstabs},
8980 @option{-gcoff}, @option{-gdwarf-2}, @option{-gxcoff}, or @option{-gvms}.
8984 @subsection Specific Options for DBX Output
8986 @c prevent bad page break with this line
8987 These are specific options for DBX output.
8989 @defmac DBX_DEBUGGING_INFO
8990 Define this macro if GCC should produce debugging output for DBX
8991 in response to the @option{-g} option.
8994 @defmac XCOFF_DEBUGGING_INFO
8995 Define this macro if GCC should produce XCOFF format debugging output
8996 in response to the @option{-g} option. This is a variant of DBX format.
8999 @defmac DEFAULT_GDB_EXTENSIONS
9000 Define this macro to control whether GCC should by default generate
9001 GDB's extended version of DBX debugging information (assuming DBX-format
9002 debugging information is enabled at all). If you don't define the
9003 macro, the default is 1: always generate the extended information
9004 if there is any occasion to.
9007 @defmac DEBUG_SYMS_TEXT
9008 Define this macro if all @code{.stabs} commands should be output while
9009 in the text section.
9012 @defmac ASM_STABS_OP
9013 A C string constant, including spacing, naming the assembler pseudo op to
9014 use instead of @code{"\t.stabs\t"} to define an ordinary debugging symbol.
9015 If you don't define this macro, @code{"\t.stabs\t"} is used. This macro
9016 applies only to DBX debugging information format.
9019 @defmac ASM_STABD_OP
9020 A C string constant, including spacing, naming the assembler pseudo op to
9021 use instead of @code{"\t.stabd\t"} to define a debugging symbol whose
9022 value is the current location. If you don't define this macro,
9023 @code{"\t.stabd\t"} is used. This macro applies only to DBX debugging
9027 @defmac ASM_STABN_OP
9028 A C string constant, including spacing, naming the assembler pseudo op to
9029 use instead of @code{"\t.stabn\t"} to define a debugging symbol with no
9030 name. If you don't define this macro, @code{"\t.stabn\t"} is used. This
9031 macro applies only to DBX debugging information format.
9034 @defmac DBX_NO_XREFS
9035 Define this macro if DBX on your system does not support the construct
9036 @samp{xs@var{tagname}}. On some systems, this construct is used to
9037 describe a forward reference to a structure named @var{tagname}.
9038 On other systems, this construct is not supported at all.
9041 @defmac DBX_CONTIN_LENGTH
9042 A symbol name in DBX-format debugging information is normally
9043 continued (split into two separate @code{.stabs} directives) when it
9044 exceeds a certain length (by default, 80 characters). On some
9045 operating systems, DBX requires this splitting; on others, splitting
9046 must not be done. You can inhibit splitting by defining this macro
9047 with the value zero. You can override the default splitting-length by
9048 defining this macro as an expression for the length you desire.
9051 @defmac DBX_CONTIN_CHAR
9052 Normally continuation is indicated by adding a @samp{\} character to
9053 the end of a @code{.stabs} string when a continuation follows. To use
9054 a different character instead, define this macro as a character
9055 constant for the character you want to use. Do not define this macro
9056 if backslash is correct for your system.
9059 @defmac DBX_STATIC_STAB_DATA_SECTION
9060 Define this macro if it is necessary to go to the data section before
9061 outputting the @samp{.stabs} pseudo-op for a non-global static
9065 @defmac DBX_TYPE_DECL_STABS_CODE
9066 The value to use in the ``code'' field of the @code{.stabs} directive
9067 for a typedef. The default is @code{N_LSYM}.
9070 @defmac DBX_STATIC_CONST_VAR_CODE
9071 The value to use in the ``code'' field of the @code{.stabs} directive
9072 for a static variable located in the text section. DBX format does not
9073 provide any ``right'' way to do this. The default is @code{N_FUN}.
9076 @defmac DBX_REGPARM_STABS_CODE
9077 The value to use in the ``code'' field of the @code{.stabs} directive
9078 for a parameter passed in registers. DBX format does not provide any
9079 ``right'' way to do this. The default is @code{N_RSYM}.
9082 @defmac DBX_REGPARM_STABS_LETTER
9083 The letter to use in DBX symbol data to identify a symbol as a parameter
9084 passed in registers. DBX format does not customarily provide any way to
9085 do this. The default is @code{'P'}.
9088 @defmac DBX_FUNCTION_FIRST
9089 Define this macro if the DBX information for a function and its
9090 arguments should precede the assembler code for the function. Normally,
9091 in DBX format, the debugging information entirely follows the assembler
9095 @defmac DBX_BLOCKS_FUNCTION_RELATIVE
9096 Define this macro, with value 1, if the value of a symbol describing
9097 the scope of a block (@code{N_LBRAC} or @code{N_RBRAC}) should be
9098 relative to the start of the enclosing function. Normally, GCC uses
9099 an absolute address.
9102 @defmac DBX_LINES_FUNCTION_RELATIVE
9103 Define this macro, with value 1, if the value of a symbol indicating
9104 the current line number (@code{N_SLINE}) should be relative to the
9105 start of the enclosing function. Normally, GCC uses an absolute address.
9108 @defmac DBX_USE_BINCL
9109 Define this macro if GCC should generate @code{N_BINCL} and
9110 @code{N_EINCL} stabs for included header files, as on Sun systems. This
9111 macro also directs GCC to output a type number as a pair of a file
9112 number and a type number within the file. Normally, GCC does not
9113 generate @code{N_BINCL} or @code{N_EINCL} stabs, and it outputs a single
9114 number for a type number.
9118 @subsection Open-Ended Hooks for DBX Format
9120 @c prevent bad page break with this line
9121 These are hooks for DBX format.
9123 @defmac DBX_OUTPUT_LBRAC (@var{stream}, @var{name})
9124 Define this macro to say how to output to @var{stream} the debugging
9125 information for the start of a scope level for variable names. The
9126 argument @var{name} is the name of an assembler symbol (for use with
9127 @code{assemble_name}) whose value is the address where the scope begins.
9130 @defmac DBX_OUTPUT_RBRAC (@var{stream}, @var{name})
9131 Like @code{DBX_OUTPUT_LBRAC}, but for the end of a scope level.
9134 @defmac DBX_OUTPUT_NFUN (@var{stream}, @var{lscope_label}, @var{decl})
9135 Define this macro if the target machine requires special handling to
9136 output an @code{N_FUN} entry for the function @var{decl}.
9139 @defmac DBX_OUTPUT_SOURCE_LINE (@var{stream}, @var{line}, @var{counter})
9140 A C statement to output DBX debugging information before code for line
9141 number @var{line} of the current source file to the stdio stream
9142 @var{stream}. @var{counter} is the number of time the macro was
9143 invoked, including the current invocation; it is intended to generate
9144 unique labels in the assembly output.
9146 This macro should not be defined if the default output is correct, or
9147 if it can be made correct by defining @code{DBX_LINES_FUNCTION_RELATIVE}.
9150 @defmac NO_DBX_FUNCTION_END
9151 Some stabs encapsulation formats (in particular ECOFF), cannot handle the
9152 @code{.stabs "",N_FUN,,0,0,Lscope-function-1} gdb dbx extension construct.
9153 On those machines, define this macro to turn this feature off without
9154 disturbing the rest of the gdb extensions.
9157 @defmac NO_DBX_BNSYM_ENSYM
9158 Some assemblers cannot handle the @code{.stabd BNSYM/ENSYM,0,0} gdb dbx
9159 extension construct. On those machines, define this macro to turn this
9160 feature off without disturbing the rest of the gdb extensions.
9163 @node File Names and DBX
9164 @subsection File Names in DBX Format
9166 @c prevent bad page break with this line
9167 This describes file names in DBX format.
9169 @defmac DBX_OUTPUT_MAIN_SOURCE_FILENAME (@var{stream}, @var{name})
9170 A C statement to output DBX debugging information to the stdio stream
9171 @var{stream}, which indicates that file @var{name} is the main source
9172 file---the file specified as the input file for compilation.
9173 This macro is called only once, at the beginning of compilation.
9175 This macro need not be defined if the standard form of output
9176 for DBX debugging information is appropriate.
9178 It may be necessary to refer to a label equal to the beginning of the
9179 text section. You can use @samp{assemble_name (stream, ltext_label_name)}
9180 to do so. If you do this, you must also set the variable
9181 @var{used_ltext_label_name} to @code{true}.
9184 @defmac NO_DBX_MAIN_SOURCE_DIRECTORY
9185 Define this macro, with value 1, if GCC should not emit an indication
9186 of the current directory for compilation and current source language at
9187 the beginning of the file.
9190 @defmac NO_DBX_GCC_MARKER
9191 Define this macro, with value 1, if GCC should not emit an indication
9192 that this object file was compiled by GCC@. The default is to emit
9193 an @code{N_OPT} stab at the beginning of every source file, with
9194 @samp{gcc2_compiled.} for the string and value 0.
9197 @defmac DBX_OUTPUT_MAIN_SOURCE_FILE_END (@var{stream}, @var{name})
9198 A C statement to output DBX debugging information at the end of
9199 compilation of the main source file @var{name}. Output should be
9200 written to the stdio stream @var{stream}.
9202 If you don't define this macro, nothing special is output at the end
9203 of compilation, which is correct for most machines.
9206 @defmac DBX_OUTPUT_NULL_N_SO_AT_MAIN_SOURCE_FILE_END
9207 Define this macro @emph{instead of} defining
9208 @code{DBX_OUTPUT_MAIN_SOURCE_FILE_END}, if what needs to be output at
9209 the end of compilation is an @code{N_SO} stab with an empty string,
9210 whose value is the highest absolute text address in the file.
9215 @subsection Macros for SDB and DWARF Output
9217 @c prevent bad page break with this line
9218 Here are macros for SDB and DWARF output.
9220 @defmac SDB_DEBUGGING_INFO
9221 Define this macro if GCC should produce COFF-style debugging output
9222 for SDB in response to the @option{-g} option.
9225 @defmac DWARF2_DEBUGGING_INFO
9226 Define this macro if GCC should produce dwarf version 2 format
9227 debugging output in response to the @option{-g} option.
9229 @deftypefn {Target Hook} int TARGET_DWARF_CALLING_CONVENTION (const_tree @var{function})
9230 Define this to enable the dwarf attribute @code{DW_AT_calling_convention} to
9231 be emitted for each function. Instead of an integer return the enum
9232 value for the @code{DW_CC_} tag.
9235 To support optional call frame debugging information, you must also
9236 define @code{INCOMING_RETURN_ADDR_RTX} and either set
9237 @code{RTX_FRAME_RELATED_P} on the prologue insns if you use RTL for the
9238 prologue, or call @code{dwarf2out_def_cfa} and @code{dwarf2out_reg_save}
9239 as appropriate from @code{TARGET_ASM_FUNCTION_PROLOGUE} if you don't.
9242 @defmac DWARF2_FRAME_INFO
9243 Define this macro to a nonzero value if GCC should always output
9244 Dwarf 2 frame information. If @code{DWARF2_UNWIND_INFO}
9245 (@pxref{Exception Region Output} is nonzero, GCC will output this
9246 information not matter how you define @code{DWARF2_FRAME_INFO}.
9249 @defmac DWARF2_ASM_LINE_DEBUG_INFO
9250 Define this macro to be a nonzero value if the assembler can generate Dwarf 2
9251 line debug info sections. This will result in much more compact line number
9252 tables, and hence is desirable if it works.
9255 @defmac ASM_OUTPUT_DWARF_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
9256 A C statement to issue assembly directives that create a difference
9257 @var{lab1} minus @var{lab2}, using an integer of the given @var{size}.
9260 @defmac ASM_OUTPUT_DWARF_VMS_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
9261 A C statement to issue assembly directives that create a difference
9262 between the two given labels in system defined units, e.g. instruction
9263 slots on IA64 VMS, using an integer of the given size.
9266 @defmac ASM_OUTPUT_DWARF_OFFSET (@var{stream}, @var{size}, @var{label}, @var{section})
9267 A C statement to issue assembly directives that create a
9268 section-relative reference to the given @var{label}, using an integer of the
9269 given @var{size}. The label is known to be defined in the given @var{section}.
9272 @defmac ASM_OUTPUT_DWARF_PCREL (@var{stream}, @var{size}, @var{label})
9273 A C statement to issue assembly directives that create a self-relative
9274 reference to the given @var{label}, using an integer of the given @var{size}.
9277 @defmac ASM_OUTPUT_DWARF_TABLE_REF (@var{label})
9278 A C statement to issue assembly directives that create a reference to
9279 the DWARF table identifier @var{label} from the current section. This
9280 is used on some systems to avoid garbage collecting a DWARF table which
9281 is referenced by a function.
9284 @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_DWARF_DTPREL (FILE *@var{file}, int @var{size}, rtx @var{x})
9285 If defined, this target hook is a function which outputs a DTP-relative
9286 reference to the given TLS symbol of the specified size.
9289 @defmac PUT_SDB_@dots{}
9290 Define these macros to override the assembler syntax for the special
9291 SDB assembler directives. See @file{sdbout.c} for a list of these
9292 macros and their arguments. If the standard syntax is used, you need
9293 not define them yourself.
9297 Some assemblers do not support a semicolon as a delimiter, even between
9298 SDB assembler directives. In that case, define this macro to be the
9299 delimiter to use (usually @samp{\n}). It is not necessary to define
9300 a new set of @code{PUT_SDB_@var{op}} macros if this is the only change
9304 @defmac SDB_ALLOW_UNKNOWN_REFERENCES
9305 Define this macro to allow references to unknown structure,
9306 union, or enumeration tags to be emitted. Standard COFF does not
9307 allow handling of unknown references, MIPS ECOFF has support for
9311 @defmac SDB_ALLOW_FORWARD_REFERENCES
9312 Define this macro to allow references to structure, union, or
9313 enumeration tags that have not yet been seen to be handled. Some
9314 assemblers choke if forward tags are used, while some require it.
9317 @defmac SDB_OUTPUT_SOURCE_LINE (@var{stream}, @var{line})
9318 A C statement to output SDB debugging information before code for line
9319 number @var{line} of the current source file to the stdio stream
9320 @var{stream}. The default is to emit an @code{.ln} directive.
9325 @subsection Macros for VMS Debug Format
9327 @c prevent bad page break with this line
9328 Here are macros for VMS debug format.
9330 @defmac VMS_DEBUGGING_INFO
9331 Define this macro if GCC should produce debugging output for VMS
9332 in response to the @option{-g} option. The default behavior for VMS
9333 is to generate minimal debug info for a traceback in the absence of
9334 @option{-g} unless explicitly overridden with @option{-g0}. This
9335 behavior is controlled by @code{OPTIMIZATION_OPTIONS} and
9336 @code{OVERRIDE_OPTIONS}.
9339 @node Floating Point
9340 @section Cross Compilation and Floating Point
9341 @cindex cross compilation and floating point
9342 @cindex floating point and cross compilation
9344 While all modern machines use twos-complement representation for integers,
9345 there are a variety of representations for floating point numbers. This
9346 means that in a cross-compiler the representation of floating point numbers
9347 in the compiled program may be different from that used in the machine
9348 doing the compilation.
9350 Because different representation systems may offer different amounts of
9351 range and precision, all floating point constants must be represented in
9352 the target machine's format. Therefore, the cross compiler cannot
9353 safely use the host machine's floating point arithmetic; it must emulate
9354 the target's arithmetic. To ensure consistency, GCC always uses
9355 emulation to work with floating point values, even when the host and
9356 target floating point formats are identical.
9358 The following macros are provided by @file{real.h} for the compiler to
9359 use. All parts of the compiler which generate or optimize
9360 floating-point calculations must use these macros. They may evaluate
9361 their operands more than once, so operands must not have side effects.
9363 @defmac REAL_VALUE_TYPE
9364 The C data type to be used to hold a floating point value in the target
9365 machine's format. Typically this is a @code{struct} containing an
9366 array of @code{HOST_WIDE_INT}, but all code should treat it as an opaque
9370 @deftypefn Macro int REAL_VALUES_EQUAL (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
9371 Compares for equality the two values, @var{x} and @var{y}. If the target
9372 floating point format supports negative zeroes and/or NaNs,
9373 @samp{REAL_VALUES_EQUAL (-0.0, 0.0)} is true, and
9374 @samp{REAL_VALUES_EQUAL (NaN, NaN)} is false.
9377 @deftypefn Macro int REAL_VALUES_LESS (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
9378 Tests whether @var{x} is less than @var{y}.
9381 @deftypefn Macro HOST_WIDE_INT REAL_VALUE_FIX (REAL_VALUE_TYPE @var{x})
9382 Truncates @var{x} to a signed integer, rounding toward zero.
9385 @deftypefn Macro {unsigned HOST_WIDE_INT} REAL_VALUE_UNSIGNED_FIX (REAL_VALUE_TYPE @var{x})
9386 Truncates @var{x} to an unsigned integer, rounding toward zero. If
9387 @var{x} is negative, returns zero.
9390 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ATOF (const char *@var{string}, enum machine_mode @var{mode})
9391 Converts @var{string} into a floating point number in the target machine's
9392 representation for mode @var{mode}. This routine can handle both
9393 decimal and hexadecimal floating point constants, using the syntax
9394 defined by the C language for both.
9397 @deftypefn Macro int REAL_VALUE_NEGATIVE (REAL_VALUE_TYPE @var{x})
9398 Returns 1 if @var{x} is negative (including negative zero), 0 otherwise.
9401 @deftypefn Macro int REAL_VALUE_ISINF (REAL_VALUE_TYPE @var{x})
9402 Determines whether @var{x} represents infinity (positive or negative).
9405 @deftypefn Macro int REAL_VALUE_ISNAN (REAL_VALUE_TYPE @var{x})
9406 Determines whether @var{x} represents a ``NaN'' (not-a-number).
9409 @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})
9410 Calculates an arithmetic operation on the two floating point values
9411 @var{x} and @var{y}, storing the result in @var{output} (which must be a
9414 The operation to be performed is specified by @var{code}. Only the
9415 following codes are supported: @code{PLUS_EXPR}, @code{MINUS_EXPR},
9416 @code{MULT_EXPR}, @code{RDIV_EXPR}, @code{MAX_EXPR}, @code{MIN_EXPR}.
9418 If @code{REAL_ARITHMETIC} is asked to evaluate division by zero and the
9419 target's floating point format cannot represent infinity, it will call
9420 @code{abort}. Callers should check for this situation first, using
9421 @code{MODE_HAS_INFINITIES}. @xref{Storage Layout}.
9424 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_NEGATE (REAL_VALUE_TYPE @var{x})
9425 Returns the negative of the floating point value @var{x}.
9428 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ABS (REAL_VALUE_TYPE @var{x})
9429 Returns the absolute value of @var{x}.
9432 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_TRUNCATE (REAL_VALUE_TYPE @var{mode}, enum machine_mode @var{x})
9433 Truncates the floating point value @var{x} to fit in @var{mode}. The
9434 return value is still a full-size @code{REAL_VALUE_TYPE}, but it has an
9435 appropriate bit pattern to be output as a floating constant whose
9436 precision accords with mode @var{mode}.
9439 @deftypefn Macro void REAL_VALUE_TO_INT (HOST_WIDE_INT @var{low}, HOST_WIDE_INT @var{high}, REAL_VALUE_TYPE @var{x})
9440 Converts a floating point value @var{x} into a double-precision integer
9441 which is then stored into @var{low} and @var{high}. If the value is not
9442 integral, it is truncated.
9445 @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})
9446 Converts a double-precision integer found in @var{low} and @var{high},
9447 into a floating point value which is then stored into @var{x}. The
9448 value is truncated to fit in mode @var{mode}.
9451 @node Mode Switching
9452 @section Mode Switching Instructions
9453 @cindex mode switching
9454 The following macros control mode switching optimizations:
9456 @defmac OPTIMIZE_MODE_SWITCHING (@var{entity})
9457 Define this macro if the port needs extra instructions inserted for mode
9458 switching in an optimizing compilation.
9460 For an example, the SH4 can perform both single and double precision
9461 floating point operations, but to perform a single precision operation,
9462 the FPSCR PR bit has to be cleared, while for a double precision
9463 operation, this bit has to be set. Changing the PR bit requires a general
9464 purpose register as a scratch register, hence these FPSCR sets have to
9465 be inserted before reload, i.e.@: you can't put this into instruction emitting
9466 or @code{TARGET_MACHINE_DEPENDENT_REORG}.
9468 You can have multiple entities that are mode-switched, and select at run time
9469 which entities actually need it. @code{OPTIMIZE_MODE_SWITCHING} should
9470 return nonzero for any @var{entity} that needs mode-switching.
9471 If you define this macro, you also have to define
9472 @code{NUM_MODES_FOR_MODE_SWITCHING}, @code{MODE_NEEDED},
9473 @code{MODE_PRIORITY_TO_MODE} and @code{EMIT_MODE_SET}.
9474 @code{MODE_AFTER}, @code{MODE_ENTRY}, and @code{MODE_EXIT}
9478 @defmac NUM_MODES_FOR_MODE_SWITCHING
9479 If you define @code{OPTIMIZE_MODE_SWITCHING}, you have to define this as
9480 initializer for an array of integers. Each initializer element
9481 N refers to an entity that needs mode switching, and specifies the number
9482 of different modes that might need to be set for this entity.
9483 The position of the initializer in the initializer---starting counting at
9484 zero---determines the integer that is used to refer to the mode-switched
9486 In macros that take mode arguments / yield a mode result, modes are
9487 represented as numbers 0 @dots{} N @minus{} 1. N is used to specify that no mode
9488 switch is needed / supplied.
9491 @defmac MODE_NEEDED (@var{entity}, @var{insn})
9492 @var{entity} is an integer specifying a mode-switched entity. If
9493 @code{OPTIMIZE_MODE_SWITCHING} is defined, you must define this macro to
9494 return an integer value not larger than the corresponding element in
9495 @code{NUM_MODES_FOR_MODE_SWITCHING}, to denote the mode that @var{entity} must
9496 be switched into prior to the execution of @var{insn}.
9499 @defmac MODE_AFTER (@var{mode}, @var{insn})
9500 If this macro is defined, it is evaluated for every @var{insn} during
9501 mode switching. It determines the mode that an insn results in (if
9502 different from the incoming mode).
9505 @defmac MODE_ENTRY (@var{entity})
9506 If this macro is defined, it is evaluated for every @var{entity} that needs
9507 mode switching. It should evaluate to an integer, which is a mode that
9508 @var{entity} is assumed to be switched to at function entry. If @code{MODE_ENTRY}
9509 is defined then @code{MODE_EXIT} must be defined.
9512 @defmac MODE_EXIT (@var{entity})
9513 If this macro is defined, it is evaluated for every @var{entity} that needs
9514 mode switching. It should evaluate to an integer, which is a mode that
9515 @var{entity} is assumed to be switched to at function exit. If @code{MODE_EXIT}
9516 is defined then @code{MODE_ENTRY} must be defined.
9519 @defmac MODE_PRIORITY_TO_MODE (@var{entity}, @var{n})
9520 This macro specifies the order in which modes for @var{entity} are processed.
9521 0 is the highest priority, @code{NUM_MODES_FOR_MODE_SWITCHING[@var{entity}] - 1} the
9522 lowest. The value of the macro should be an integer designating a mode
9523 for @var{entity}. For any fixed @var{entity}, @code{mode_priority_to_mode}
9524 (@var{entity}, @var{n}) shall be a bijection in 0 @dots{}
9525 @code{num_modes_for_mode_switching[@var{entity}] - 1}.
9528 @defmac EMIT_MODE_SET (@var{entity}, @var{mode}, @var{hard_regs_live})
9529 Generate one or more insns to set @var{entity} to @var{mode}.
9530 @var{hard_reg_live} is the set of hard registers live at the point where
9531 the insn(s) are to be inserted.
9534 @node Target Attributes
9535 @section Defining target-specific uses of @code{__attribute__}
9536 @cindex target attributes
9537 @cindex machine attributes
9538 @cindex attributes, target-specific
9540 Target-specific attributes may be defined for functions, data and types.
9541 These are described using the following target hooks; they also need to
9542 be documented in @file{extend.texi}.
9544 @deftypevr {Target Hook} {const struct attribute_spec *} TARGET_ATTRIBUTE_TABLE
9545 If defined, this target hook points to an array of @samp{struct
9546 attribute_spec} (defined in @file{tree.h}) specifying the machine
9547 specific attributes for this target and some of the restrictions on the
9548 entities to which these attributes are applied and the arguments they
9552 @deftypefn {Target Hook} bool TARGET_ATTRIBUTE_TAKES_IDENTIFIER_P (const_tree @var{name})
9553 If defined, this target hook is a function which returns true if the
9554 machine-specific attribute named @var{name} expects an identifier
9555 given as its first argument to be passed on as a plain identifier, not
9556 subjected to name lookup. If this is not defined, the default is
9557 false for all machine-specific attributes.
9560 @deftypefn {Target Hook} int TARGET_COMP_TYPE_ATTRIBUTES (const_tree @var{type1}, const_tree @var{type2})
9561 If defined, this target hook is a function which returns zero if the attributes on
9562 @var{type1} and @var{type2} are incompatible, one if they are compatible,
9563 and two if they are nearly compatible (which causes a warning to be
9564 generated). If this is not defined, machine-specific attributes are
9565 supposed always to be compatible.
9568 @deftypefn {Target Hook} void TARGET_SET_DEFAULT_TYPE_ATTRIBUTES (tree @var{type})
9569 If defined, this target hook is a function which assigns default attributes to
9570 the newly defined @var{type}.
9573 @deftypefn {Target Hook} tree TARGET_MERGE_TYPE_ATTRIBUTES (tree @var{type1}, tree @var{type2})
9574 Define this target hook if the merging of type attributes needs special
9575 handling. If defined, the result is a list of the combined
9576 @code{TYPE_ATTRIBUTES} of @var{type1} and @var{type2}. It is assumed
9577 that @code{comptypes} has already been called and returned 1. This
9578 function may call @code{merge_attributes} to handle machine-independent
9582 @deftypefn {Target Hook} tree TARGET_MERGE_DECL_ATTRIBUTES (tree @var{olddecl}, tree @var{newdecl})
9583 Define this target hook if the merging of decl attributes needs special
9584 handling. If defined, the result is a list of the combined
9585 @code{DECL_ATTRIBUTES} of @var{olddecl} and @var{newdecl}.
9586 @var{newdecl} is a duplicate declaration of @var{olddecl}. Examples of
9587 when this is needed are when one attribute overrides another, or when an
9588 attribute is nullified by a subsequent definition. This function may
9589 call @code{merge_attributes} to handle machine-independent merging.
9591 @findex TARGET_DLLIMPORT_DECL_ATTRIBUTES
9592 If the only target-specific handling you require is @samp{dllimport}
9593 for Microsoft Windows targets, you should define the macro
9594 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES} to @code{1}. The compiler
9595 will then define a function called
9596 @code{merge_dllimport_decl_attributes} which can then be defined as
9597 the expansion of @code{TARGET_MERGE_DECL_ATTRIBUTES}. You can also
9598 add @code{handle_dll_attribute} in the attribute table for your port
9599 to perform initial processing of the @samp{dllimport} and
9600 @samp{dllexport} attributes. This is done in @file{i386/cygwin.h} and
9601 @file{i386/i386.c}, for example.
9604 @deftypefn {Target Hook} bool TARGET_VALID_DLLIMPORT_ATTRIBUTE_P (const_tree @var{decl})
9605 @var{decl} is a variable or function with @code{__attribute__((dllimport))}
9606 specified. Use this hook if the target needs to add extra validation
9607 checks to @code{handle_dll_attribute}.
9610 @defmac TARGET_DECLSPEC
9611 Define this macro to a nonzero value if you want to treat
9612 @code{__declspec(X)} as equivalent to @code{__attribute((X))}. By
9613 default, this behavior is enabled only for targets that define
9614 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES}. The current implementation
9615 of @code{__declspec} is via a built-in macro, but you should not rely
9616 on this implementation detail.
9619 @deftypefn {Target Hook} void TARGET_INSERT_ATTRIBUTES (tree @var{node}, tree *@var{attr_ptr})
9620 Define this target hook if you want to be able to add attributes to a decl
9621 when it is being created. This is normally useful for back ends which
9622 wish to implement a pragma by using the attributes which correspond to
9623 the pragma's effect. The @var{node} argument is the decl which is being
9624 created. The @var{attr_ptr} argument is a pointer to the attribute list
9625 for this decl. The list itself should not be modified, since it may be
9626 shared with other decls, but attributes may be chained on the head of
9627 the list and @code{*@var{attr_ptr}} modified to point to the new
9628 attributes, or a copy of the list may be made if further changes are
9632 @deftypefn {Target Hook} bool TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P (const_tree @var{fndecl})
9634 This target hook returns @code{true} if it is ok to inline @var{fndecl}
9635 into the current function, despite its having target-specific
9636 attributes, @code{false} otherwise. By default, if a function has a
9637 target specific attribute attached to it, it will not be inlined.
9640 @deftypefn {Target Hook} bool TARGET_OPTION_VALID_ATTRIBUTE_P (tree @var{fndecl}, tree @var{name}, tree @var{args}, int @var{flags})
9641 This hook is called to parse the @code{attribute(option("..."))}, and
9642 it allows the function to set different target machine compile time
9643 options for the current function that might be different than the
9644 options specified on the command line. The hook should return
9645 @code{true} if the options are valid.
9647 The hook should set the @var{DECL_FUNCTION_SPECIFIC_TARGET} field in
9648 the function declaration to hold a pointer to a target specific
9649 @var{struct cl_target_option} structure.
9652 @deftypefn {Target Hook} void TARGET_OPTION_SAVE (struct cl_target_option *@var{ptr})
9653 This hook is called to save any additional target specific information
9654 in the @var{struct cl_target_option} structure for function specific
9656 @xref{Option file format}.
9659 @deftypefn {Target Hook} void TARGET_OPTION_RESTORE (struct cl_target_option *@var{ptr})
9660 This hook is called to restore any additional target specific
9661 information in the @var{struct cl_target_option} structure for
9662 function specific options.
9665 @deftypefn {Target Hook} void TARGET_OPTION_PRINT (FILE *@var{file}, int @var{indent}, struct cl_target_option *@var{ptr})
9666 This hook is called to print any additional target specific
9667 information in the @var{struct cl_target_option} structure for
9668 function specific options.
9671 @deftypefn {Target Hook} bool TARGET_OPTION_PRAGMA_PARSE (target @var{args})
9672 This target hook parses the options for @code{#pragma GCC option} to
9673 set the machine specific options for functions that occur later in the
9674 input stream. The options should be the same as handled by the
9675 @code{TARGET_VALID_OPTION_ATTRIBUTE_P} hook.
9678 @deftypefn {Target Hook} bool TARGET_CAN_INLINE_P (tree @var{caller}, tree @var{callee})
9679 This target hook returns @code{false} if the @var{caller} function
9680 cannot inline @var{callee}, based on target specific information. By
9681 default, inlining is not allowed if the callee function has function
9682 specific target options and the caller does not use the same options.
9686 @section Emulating TLS
9687 @cindex Emulated TLS
9689 For targets whose psABI does not provide Thread Local Storage via
9690 specific relocations and instruction sequences, an emulation layer is
9691 used. A set of target hooks allows this emulation layer to be
9692 configured for the requirements of a particular target. For instance
9693 the psABI may in fact specify TLS support in terms of an emulation
9696 The emulation layer works by creating a control object for every TLS
9697 object. To access the TLS object, a lookup function is provided
9698 which, when given the address of the control object, will return the
9699 address of the current thread's instance of the TLS object.
9701 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_GET_ADDRESS
9702 Contains the name of the helper function that uses a TLS control
9703 object to locate a TLS instance. The default causes libgcc's
9704 emulated TLS helper function to be used.
9707 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_REGISTER_COMMON
9708 Contains the name of the helper function that should be used at
9709 program startup to register TLS objects that are implicitly
9710 initialized to zero. If this is @code{NULL}, all TLS objects will
9711 have explicit initializers. The default causes libgcc's emulated TLS
9712 registration function to be used.
9715 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_VAR_SECTION
9716 Contains the name of the section in which TLS control variables should
9717 be placed. The default of @code{NULL} allows these to be placed in
9721 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_TMPL_SECTION
9722 Contains the name of the section in which TLS initializers should be
9723 placed. The default of @code{NULL} allows these to be placed in any
9727 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_VAR_PREFIX
9728 Contains the prefix to be prepended to TLS control variable names.
9729 The default of @code{NULL} uses a target-specific prefix.
9732 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_TMPL_PREFIX
9733 Contains the prefix to be prepended to TLS initializer objects. The
9734 default of @code{NULL} uses a target-specific prefix.
9737 @deftypefn {Target Hook} tree TARGET_EMUTLS_VAR_FIELDS (tree @var{type}, tree *@var{name})
9738 Specifies a function that generates the FIELD_DECLs for a TLS control
9739 object type. @var{type} is the RECORD_TYPE the fields are for and
9740 @var{name} should be filled with the structure tag, if the default of
9741 @code{__emutls_object} is unsuitable. The default creates a type suitable
9742 for libgcc's emulated TLS function.
9745 @deftypefn {Target Hook} tree TARGET_EMUTLS_VAR_INIT (tree @var{var}, tree @var{decl}, tree @var{tmpl_addr})
9746 Specifies a function that generates the CONSTRUCTOR to initialize a
9747 TLS control object. @var{var} is the TLS control object, @var{decl}
9748 is the TLS object and @var{tmpl_addr} is the address of the
9749 initializer. The default initializes libgcc's emulated TLS control object.
9752 @deftypevr {Target Hook} bool TARGET_EMUTLS_VAR_ALIGN_FIXED
9753 Specifies whether the alignment of TLS control variable objects is
9754 fixed and should not be increased as some backends may do to optimize
9755 single objects. The default is false.
9758 @deftypevr {Target Hook} bool TARGET_EMUTLS_DEBUG_FORM_TLS_ADDRESS
9759 Specifies whether a DWARF @code{DW_OP_form_tls_address} location descriptor
9760 may be used to describe emulated TLS control objects.
9763 @node MIPS Coprocessors
9764 @section Defining coprocessor specifics for MIPS targets.
9765 @cindex MIPS coprocessor-definition macros
9767 The MIPS specification allows MIPS implementations to have as many as 4
9768 coprocessors, each with as many as 32 private registers. GCC supports
9769 accessing these registers and transferring values between the registers
9770 and memory using asm-ized variables. For example:
9773 register unsigned int cp0count asm ("c0r1");
9779 (``c0r1'' is the default name of register 1 in coprocessor 0; alternate
9780 names may be added as described below, or the default names may be
9781 overridden entirely in @code{SUBTARGET_CONDITIONAL_REGISTER_USAGE}.)
9783 Coprocessor registers are assumed to be epilogue-used; sets to them will
9784 be preserved even if it does not appear that the register is used again
9785 later in the function.
9787 Another note: according to the MIPS spec, coprocessor 1 (if present) is
9788 the FPU@. One accesses COP1 registers through standard mips
9789 floating-point support; they are not included in this mechanism.
9791 There is one macro used in defining the MIPS coprocessor interface which
9792 you may want to override in subtargets; it is described below.
9794 @defmac ALL_COP_ADDITIONAL_REGISTER_NAMES
9795 A comma-separated list (with leading comma) of pairs describing the
9796 alternate names of coprocessor registers. The format of each entry should be
9798 @{ @var{alternatename}, @var{register_number}@}
9804 @section Parameters for Precompiled Header Validity Checking
9805 @cindex parameters, precompiled headers
9807 @deftypefn {Target Hook} {void *} TARGET_GET_PCH_VALIDITY (size_t *@var{sz})
9808 This hook returns a pointer to the data needed by
9809 @code{TARGET_PCH_VALID_P} and sets
9810 @samp{*@var{sz}} to the size of the data in bytes.
9813 @deftypefn {Target Hook} {const char *} TARGET_PCH_VALID_P (const void *@var{data}, size_t @var{sz})
9814 This hook checks whether the options used to create a PCH file are
9815 compatible with the current settings. It returns @code{NULL}
9816 if so and a suitable error message if not. Error messages will
9817 be presented to the user and must be localized using @samp{_(@var{msg})}.
9819 @var{data} is the data that was returned by @code{TARGET_GET_PCH_VALIDITY}
9820 when the PCH file was created and @var{sz} is the size of that data in bytes.
9821 It's safe to assume that the data was created by the same version of the
9822 compiler, so no format checking is needed.
9824 The default definition of @code{default_pch_valid_p} should be
9825 suitable for most targets.
9828 @deftypefn {Target Hook} {const char *} TARGET_CHECK_PCH_TARGET_FLAGS (int @var{pch_flags})
9829 If this hook is nonnull, the default implementation of
9830 @code{TARGET_PCH_VALID_P} will use it to check for compatible values
9831 of @code{target_flags}. @var{pch_flags} specifies the value that
9832 @code{target_flags} had when the PCH file was created. The return
9833 value is the same as for @code{TARGET_PCH_VALID_P}.
9837 @section C++ ABI parameters
9838 @cindex parameters, c++ abi
9840 @deftypefn {Target Hook} tree TARGET_CXX_GUARD_TYPE (void)
9841 Define this hook to override the integer type used for guard variables.
9842 These are used to implement one-time construction of static objects. The
9843 default is long_long_integer_type_node.
9846 @deftypefn {Target Hook} bool TARGET_CXX_GUARD_MASK_BIT (void)
9847 This hook determines how guard variables are used. It should return
9848 @code{false} (the default) if the first byte should be used. A return value of
9849 @code{true} indicates that only the least significant bit should be used.
9852 @deftypefn {Target Hook} tree TARGET_CXX_GET_COOKIE_SIZE (tree @var{type})
9853 This hook returns the size of the cookie to use when allocating an array
9854 whose elements have the indicated @var{type}. Assumes that it is already
9855 known that a cookie is needed. The default is
9856 @code{max(sizeof (size_t), alignof(type))}, as defined in section 2.7 of the
9857 IA64/Generic C++ ABI@.
9860 @deftypefn {Target Hook} bool TARGET_CXX_COOKIE_HAS_SIZE (void)
9861 This hook should return @code{true} if the element size should be stored in
9862 array cookies. The default is to return @code{false}.
9865 @deftypefn {Target Hook} int TARGET_CXX_IMPORT_EXPORT_CLASS (tree @var{type}, int @var{import_export})
9866 If defined by a backend this hook allows the decision made to export
9867 class @var{type} to be overruled. Upon entry @var{import_export}
9868 will contain 1 if the class is going to be exported, @minus{}1 if it is going
9869 to be imported and 0 otherwise. This function should return the
9870 modified value and perform any other actions necessary to support the
9871 backend's targeted operating system.
9874 @deftypefn {Target Hook} bool TARGET_CXX_CDTOR_RETURNS_THIS (void)
9875 This hook should return @code{true} if constructors and destructors return
9876 the address of the object created/destroyed. The default is to return
9880 @deftypefn {Target Hook} bool TARGET_CXX_KEY_METHOD_MAY_BE_INLINE (void)
9881 This hook returns true if the key method for a class (i.e., the method
9882 which, if defined in the current translation unit, causes the virtual
9883 table to be emitted) may be an inline function. Under the standard
9884 Itanium C++ ABI the key method may be an inline function so long as
9885 the function is not declared inline in the class definition. Under
9886 some variants of the ABI, an inline function can never be the key
9887 method. The default is to return @code{true}.
9890 @deftypefn {Target Hook} void TARGET_CXX_DETERMINE_CLASS_DATA_VISIBILITY (tree @var{decl})
9891 @var{decl} is a virtual table, virtual table table, typeinfo object,
9892 or other similar implicit class data object that will be emitted with
9893 external linkage in this translation unit. No ELF visibility has been
9894 explicitly specified. If the target needs to specify a visibility
9895 other than that of the containing class, use this hook to set
9896 @code{DECL_VISIBILITY} and @code{DECL_VISIBILITY_SPECIFIED}.
9899 @deftypefn {Target Hook} bool TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT (void)
9900 This hook returns true (the default) if virtual tables and other
9901 similar implicit class data objects are always COMDAT if they have
9902 external linkage. If this hook returns false, then class data for
9903 classes whose virtual table will be emitted in only one translation
9904 unit will not be COMDAT.
9907 @deftypefn {Target Hook} bool TARGET_CXX_LIBRARY_RTTI_COMDAT (void)
9908 This hook returns true (the default) if the RTTI information for
9909 the basic types which is defined in the C++ runtime should always
9910 be COMDAT, false if it should not be COMDAT.
9913 @deftypefn {Target Hook} bool TARGET_CXX_USE_AEABI_ATEXIT (void)
9914 This hook returns true if @code{__aeabi_atexit} (as defined by the ARM EABI)
9915 should be used to register static destructors when @option{-fuse-cxa-atexit}
9916 is in effect. The default is to return false to use @code{__cxa_atexit}.
9919 @deftypefn {Target Hook} bool TARGET_CXX_USE_ATEXIT_FOR_CXA_ATEXIT (void)
9920 This hook returns true if the target @code{atexit} function can be used
9921 in the same manner as @code{__cxa_atexit} to register C++ static
9922 destructors. This requires that @code{atexit}-registered functions in
9923 shared libraries are run in the correct order when the libraries are
9924 unloaded. The default is to return false.
9927 @deftypefn {Target Hook} void TARGET_CXX_ADJUST_CLASS_AT_DEFINITION (tree @var{type})
9928 @var{type} is a C++ class (i.e., RECORD_TYPE or UNION_TYPE) that has just been
9929 defined. Use this hook to make adjustments to the class (eg, tweak
9930 visibility or perform any other required target modifications).
9933 @node Named Address Spaces
9934 @section Adding support for named address spaces
9935 @cindex named address spaces
9937 The draft technical report of the ISO/IEC JTC1 S22 WG14 N1275
9938 standards committee, @cite{Programming Languages - C - Extensions to
9939 support embedded processors}, specifies a syntax for embedded
9940 processors to specify alternate address spaces. You can configure a
9941 GCC port to support section 5.1 of the draft report to add support for
9942 address spaces other than the default address space. These address
9943 spaces are new keywords that are similar to the @code{volatile} and
9944 @code{const} type attributes.
9946 Pointers to named address spaces can have a different size than
9947 pointers to the generic address space.
9949 For example, the SPU port uses the @code{__ea} address space to refer
9950 to memory in the host processor, rather than memory local to the SPU
9951 processor. Access to memory in the @code{__ea} address space involves
9952 issuing DMA operations to move data between the host processor and the
9953 local processor memory address space. Pointers in the @code{__ea}
9954 address space are either 32 bits or 64 bits based on the
9955 @option{-mea32} or @option{-mea64} switches (native SPU pointers are
9958 Internally, address spaces are represented as a small integer in the
9959 range 0 to 15 with address space 0 being reserved for the generic
9962 To register a named address space qualifier keyword with the C front end,
9963 the target may call the @code{c_register_addr_space} routine. For example,
9964 the SPU port uses the following to declare @code{__ea} as the keyword for
9965 named address space #1:
9967 #define ADDR_SPACE_EA 1
9968 c_register_addr_space ("__ea", ADDR_SPACE_EA);
9971 @deftypefn {Target Hook} {enum machine_mode} TARGET_ADDR_SPACE_POINTER_MODE (addr_space_t @var{address_space})
9972 Define this to return the machine mode to use for pointers to
9973 @var{address_space} if the target supports named address spaces.
9974 The default version of this hook returns @code{ptr_mode} for the
9975 generic address space only.
9978 @deftypefn {Target Hook} {enum machine_mode} TARGET_ADDR_SPACE_ADDRESS_MODE (addr_space_t @var{address_space})
9979 Define this to return the machine mode to use for addresses in
9980 @var{address_space} if the target supports named address spaces.
9981 The default version of this hook returns @code{Pmode} for the
9982 generic address space only.
9985 @deftypefn {Target Hook} bool TARGET_ADDR_SPACE_VALID_POINTER_MODE (enum machine_mode @var{mode}, addr_space_t @var{as})
9986 Define this to return nonzero if the port can handle pointers
9987 with machine mode @var{mode} to address space @var{as}. This target
9988 hook is the same as the @code{TARGET_VALID_POINTER_MODE} target hook,
9989 except that it includes explicit named address space support. The default
9990 version of this hook returns true for the modes returned by either the
9991 @code{TARGET_ADDR_SPACE_POINTER_MODE} or @code{TARGET_ADDR_SPACE_ADDRESS_MODE}
9992 target hooks for the given address space.
9995 @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})
9996 Define this to return true if @var{exp} is a valid address for mode
9997 @var{mode} in the named address space @var{as}. The @var{strict}
9998 parameter says whether strict addressing is in effect after reload has
9999 finished. This target hook is the same as the
10000 @code{TARGET_LEGITIMATE_ADDRESS_P} target hook, except that it includes
10001 explicit named address space support.
10004 @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})
10005 Define this to modify an invalid address @var{x} to be a valid address
10006 with mode @var{mode} in the named address space @var{as}. This target
10007 hook is the same as the @code{TARGET_LEGITIMIZE_ADDRESS} target hook,
10008 except that it includes explicit named address space support.
10011 @deftypefn {Target Hook} bool TARGET_ADDR_SPACE_SUBSET_P (addr_space_t @var{superset}, addr_space_t @var{subset})
10012 Define this to return whether the @var{subset} named address space is
10013 contained within the @var{superset} named address space. Pointers to
10014 a named address space that is a subset of another named address space
10015 will be converted automatically without a cast if used together in
10016 arithmetic operations. Pointers to a superset address space can be
10017 converted to pointers to a subset address space via explicit casts.
10020 @deftypefn {Target Hook} rtx TARGET_ADDR_SPACE_CONVERT (rtx @var{op}, tree @var{from_type}, tree @var{to_type})
10021 Define this to convert the pointer expression represented by the RTL
10022 @var{op} with type @var{from_type} that points to a named address
10023 space to a new pointer expression with type @var{to_type} that points
10024 to a different named address space. When this hook it called, it is
10025 guaranteed that one of the two address spaces is a subset of the other,
10026 as determined by the @code{TARGET_ADDR_SPACE_SUBSET_P} target hook.
10030 @section Miscellaneous Parameters
10031 @cindex parameters, miscellaneous
10033 @c prevent bad page break with this line
10034 Here are several miscellaneous parameters.
10036 @defmac HAS_LONG_COND_BRANCH
10037 Define this boolean macro to indicate whether or not your architecture
10038 has conditional branches that can span all of memory. It is used in
10039 conjunction with an optimization that partitions hot and cold basic
10040 blocks into separate sections of the executable. If this macro is
10041 set to false, gcc will convert any conditional branches that attempt
10042 to cross between sections into unconditional branches or indirect jumps.
10045 @defmac HAS_LONG_UNCOND_BRANCH
10046 Define this boolean macro to indicate whether or not your architecture
10047 has unconditional branches that can span all of memory. It is used in
10048 conjunction with an optimization that partitions hot and cold basic
10049 blocks into separate sections of the executable. If this macro is
10050 set to false, gcc will convert any unconditional branches that attempt
10051 to cross between sections into indirect jumps.
10054 @defmac CASE_VECTOR_MODE
10055 An alias for a machine mode name. This is the machine mode that
10056 elements of a jump-table should have.
10059 @defmac CASE_VECTOR_SHORTEN_MODE (@var{min_offset}, @var{max_offset}, @var{body})
10060 Optional: return the preferred mode for an @code{addr_diff_vec}
10061 when the minimum and maximum offset are known. If you define this,
10062 it enables extra code in branch shortening to deal with @code{addr_diff_vec}.
10063 To make this work, you also have to define @code{INSN_ALIGN} and
10064 make the alignment for @code{addr_diff_vec} explicit.
10065 The @var{body} argument is provided so that the offset_unsigned and scale
10066 flags can be updated.
10069 @defmac CASE_VECTOR_PC_RELATIVE
10070 Define this macro to be a C expression to indicate when jump-tables
10071 should contain relative addresses. You need not define this macro if
10072 jump-tables never contain relative addresses, or jump-tables should
10073 contain relative addresses only when @option{-fPIC} or @option{-fPIC}
10077 @deftypefn {Target Hook} {unsigned int} TARGET_CASE_VALUES_THRESHOLD (void)
10078 This function return the smallest number of different values for which it
10079 is best to use a jump-table instead of a tree of conditional branches.
10080 The default is four for machines with a @code{casesi} instruction and
10081 five otherwise. This is best for most machines.
10084 @defmac CASE_USE_BIT_TESTS
10085 Define this macro to be a C expression to indicate whether C switch
10086 statements may be implemented by a sequence of bit tests. This is
10087 advantageous on processors that can efficiently implement left shift
10088 of 1 by the number of bits held in a register, but inappropriate on
10089 targets that would require a loop. By default, this macro returns
10090 @code{true} if the target defines an @code{ashlsi3} pattern, and
10091 @code{false} otherwise.
10094 @defmac WORD_REGISTER_OPERATIONS
10095 Define this macro if operations between registers with integral mode
10096 smaller than a word are always performed on the entire register.
10097 Most RISC machines have this property and most CISC machines do not.
10100 @defmac LOAD_EXTEND_OP (@var{mem_mode})
10101 Define this macro to be a C expression indicating when insns that read
10102 memory in @var{mem_mode}, an integral mode narrower than a word, set the
10103 bits outside of @var{mem_mode} to be either the sign-extension or the
10104 zero-extension of the data read. Return @code{SIGN_EXTEND} for values
10105 of @var{mem_mode} for which the
10106 insn sign-extends, @code{ZERO_EXTEND} for which it zero-extends, and
10107 @code{UNKNOWN} for other modes.
10109 This macro is not called with @var{mem_mode} non-integral or with a width
10110 greater than or equal to @code{BITS_PER_WORD}, so you may return any
10111 value in this case. Do not define this macro if it would always return
10112 @code{UNKNOWN}. On machines where this macro is defined, you will normally
10113 define it as the constant @code{SIGN_EXTEND} or @code{ZERO_EXTEND}.
10115 You may return a non-@code{UNKNOWN} value even if for some hard registers
10116 the sign extension is not performed, if for the @code{REGNO_REG_CLASS}
10117 of these hard registers @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero
10118 when the @var{from} mode is @var{mem_mode} and the @var{to} mode is any
10119 integral mode larger than this but not larger than @code{word_mode}.
10121 You must return @code{UNKNOWN} if for some hard registers that allow this
10122 mode, @code{CANNOT_CHANGE_MODE_CLASS} says that they cannot change to
10123 @code{word_mode}, but that they can change to another integral mode that
10124 is larger then @var{mem_mode} but still smaller than @code{word_mode}.
10127 @defmac SHORT_IMMEDIATES_SIGN_EXTEND
10128 Define this macro if loading short immediate values into registers sign
10132 @defmac FIXUNS_TRUNC_LIKE_FIX_TRUNC
10133 Define this macro if the same instructions that convert a floating
10134 point number to a signed fixed point number also convert validly to an
10138 @deftypefn {Target Hook} {unsigned int} TARGET_MIN_DIVISIONS_FOR_RECIP_MUL (enum machine_mode @var{mode})
10139 When @option{-ffast-math} is in effect, GCC tries to optimize
10140 divisions by the same divisor, by turning them into multiplications by
10141 the reciprocal. This target hook specifies the minimum number of divisions
10142 that should be there for GCC to perform the optimization for a variable
10143 of mode @var{mode}. The default implementation returns 3 if the machine
10144 has an instruction for the division, and 2 if it does not.
10148 The maximum number of bytes that a single instruction can move quickly
10149 between memory and registers or between two memory locations.
10152 @defmac MAX_MOVE_MAX
10153 The maximum number of bytes that a single instruction can move quickly
10154 between memory and registers or between two memory locations. If this
10155 is undefined, the default is @code{MOVE_MAX}. Otherwise, it is the
10156 constant value that is the largest value that @code{MOVE_MAX} can have
10160 @defmac SHIFT_COUNT_TRUNCATED
10161 A C expression that is nonzero if on this machine the number of bits
10162 actually used for the count of a shift operation is equal to the number
10163 of bits needed to represent the size of the object being shifted. When
10164 this macro is nonzero, the compiler will assume that it is safe to omit
10165 a sign-extend, zero-extend, and certain bitwise `and' instructions that
10166 truncates the count of a shift operation. On machines that have
10167 instructions that act on bit-fields at variable positions, which may
10168 include `bit test' instructions, a nonzero @code{SHIFT_COUNT_TRUNCATED}
10169 also enables deletion of truncations of the values that serve as
10170 arguments to bit-field instructions.
10172 If both types of instructions truncate the count (for shifts) and
10173 position (for bit-field operations), or if no variable-position bit-field
10174 instructions exist, you should define this macro.
10176 However, on some machines, such as the 80386 and the 680x0, truncation
10177 only applies to shift operations and not the (real or pretended)
10178 bit-field operations. Define @code{SHIFT_COUNT_TRUNCATED} to be zero on
10179 such machines. Instead, add patterns to the @file{md} file that include
10180 the implied truncation of the shift instructions.
10182 You need not define this macro if it would always have the value of zero.
10185 @anchor{TARGET_SHIFT_TRUNCATION_MASK}
10186 @deftypefn {Target Hook} {unsigned HOST_WIDE_INT} TARGET_SHIFT_TRUNCATION_MASK (enum machine_mode @var{mode})
10187 This function describes how the standard shift patterns for @var{mode}
10188 deal with shifts by negative amounts or by more than the width of the mode.
10189 @xref{shift patterns}.
10191 On many machines, the shift patterns will apply a mask @var{m} to the
10192 shift count, meaning that a fixed-width shift of @var{x} by @var{y} is
10193 equivalent to an arbitrary-width shift of @var{x} by @var{y & m}. If
10194 this is true for mode @var{mode}, the function should return @var{m},
10195 otherwise it should return 0. A return value of 0 indicates that no
10196 particular behavior is guaranteed.
10198 Note that, unlike @code{SHIFT_COUNT_TRUNCATED}, this function does
10199 @emph{not} apply to general shift rtxes; it applies only to instructions
10200 that are generated by the named shift patterns.
10202 The default implementation of this function returns
10203 @code{GET_MODE_BITSIZE (@var{mode}) - 1} if @code{SHIFT_COUNT_TRUNCATED}
10204 and 0 otherwise. This definition is always safe, but if
10205 @code{SHIFT_COUNT_TRUNCATED} is false, and some shift patterns
10206 nevertheless truncate the shift count, you may get better code
10210 @defmac TRULY_NOOP_TRUNCATION (@var{outprec}, @var{inprec})
10211 A C expression which is nonzero if on this machine it is safe to
10212 ``convert'' an integer of @var{inprec} bits to one of @var{outprec}
10213 bits (where @var{outprec} is smaller than @var{inprec}) by merely
10214 operating on it as if it had only @var{outprec} bits.
10216 On many machines, this expression can be 1.
10218 @c rearranged this, removed the phrase "it is reported that". this was
10219 @c to fix an overfull hbox. --mew 10feb93
10220 When @code{TRULY_NOOP_TRUNCATION} returns 1 for a pair of sizes for
10221 modes for which @code{MODES_TIEABLE_P} is 0, suboptimal code can result.
10222 If this is the case, making @code{TRULY_NOOP_TRUNCATION} return 0 in
10223 such cases may improve things.
10226 @deftypefn {Target Hook} int TARGET_MODE_REP_EXTENDED (enum machine_mode @var{mode}, enum machine_mode @var{rep_mode})
10227 The representation of an integral mode can be such that the values
10228 are always extended to a wider integral mode. Return
10229 @code{SIGN_EXTEND} if values of @var{mode} are represented in
10230 sign-extended form to @var{rep_mode}. Return @code{UNKNOWN}
10231 otherwise. (Currently, none of the targets use zero-extended
10232 representation this way so unlike @code{LOAD_EXTEND_OP},
10233 @code{TARGET_MODE_REP_EXTENDED} is expected to return either
10234 @code{SIGN_EXTEND} or @code{UNKNOWN}. Also no target extends
10235 @var{mode} to @var{rep_mode} so that @var{rep_mode} is not the next
10236 widest integral mode and currently we take advantage of this fact.)
10238 Similarly to @code{LOAD_EXTEND_OP} you may return a non-@code{UNKNOWN}
10239 value even if the extension is not performed on certain hard registers
10240 as long as for the @code{REGNO_REG_CLASS} of these hard registers
10241 @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero.
10243 Note that @code{TARGET_MODE_REP_EXTENDED} and @code{LOAD_EXTEND_OP}
10244 describe two related properties. If you define
10245 @code{TARGET_MODE_REP_EXTENDED (mode, word_mode)} you probably also want
10246 to define @code{LOAD_EXTEND_OP (mode)} to return the same type of
10249 In order to enforce the representation of @code{mode},
10250 @code{TRULY_NOOP_TRUNCATION} should return false when truncating to
10254 @defmac STORE_FLAG_VALUE
10255 A C expression describing the value returned by a comparison operator
10256 with an integral mode and stored by a store-flag instruction
10257 (@samp{cstore@var{mode}4}) when the condition is true. This description must
10258 apply to @emph{all} the @samp{cstore@var{mode}4} patterns and all the
10259 comparison operators whose results have a @code{MODE_INT} mode.
10261 A value of 1 or @minus{}1 means that the instruction implementing the
10262 comparison operator returns exactly 1 or @minus{}1 when the comparison is true
10263 and 0 when the comparison is false. Otherwise, the value indicates
10264 which bits of the result are guaranteed to be 1 when the comparison is
10265 true. This value is interpreted in the mode of the comparison
10266 operation, which is given by the mode of the first operand in the
10267 @samp{cstore@var{mode}4} pattern. Either the low bit or the sign bit of
10268 @code{STORE_FLAG_VALUE} be on. Presently, only those bits are used by
10271 If @code{STORE_FLAG_VALUE} is neither 1 or @minus{}1, the compiler will
10272 generate code that depends only on the specified bits. It can also
10273 replace comparison operators with equivalent operations if they cause
10274 the required bits to be set, even if the remaining bits are undefined.
10275 For example, on a machine whose comparison operators return an
10276 @code{SImode} value and where @code{STORE_FLAG_VALUE} is defined as
10277 @samp{0x80000000}, saying that just the sign bit is relevant, the
10281 (ne:SI (and:SI @var{x} (const_int @var{power-of-2})) (const_int 0))
10285 can be converted to
10288 (ashift:SI @var{x} (const_int @var{n}))
10292 where @var{n} is the appropriate shift count to move the bit being
10293 tested into the sign bit.
10295 There is no way to describe a machine that always sets the low-order bit
10296 for a true value, but does not guarantee the value of any other bits,
10297 but we do not know of any machine that has such an instruction. If you
10298 are trying to port GCC to such a machine, include an instruction to
10299 perform a logical-and of the result with 1 in the pattern for the
10300 comparison operators and let us know at @email{gcc@@gcc.gnu.org}.
10302 Often, a machine will have multiple instructions that obtain a value
10303 from a comparison (or the condition codes). Here are rules to guide the
10304 choice of value for @code{STORE_FLAG_VALUE}, and hence the instructions
10309 Use the shortest sequence that yields a valid definition for
10310 @code{STORE_FLAG_VALUE}. It is more efficient for the compiler to
10311 ``normalize'' the value (convert it to, e.g., 1 or 0) than for the
10312 comparison operators to do so because there may be opportunities to
10313 combine the normalization with other operations.
10316 For equal-length sequences, use a value of 1 or @minus{}1, with @minus{}1 being
10317 slightly preferred on machines with expensive jumps and 1 preferred on
10321 As a second choice, choose a value of @samp{0x80000001} if instructions
10322 exist that set both the sign and low-order bits but do not define the
10326 Otherwise, use a value of @samp{0x80000000}.
10329 Many machines can produce both the value chosen for
10330 @code{STORE_FLAG_VALUE} and its negation in the same number of
10331 instructions. On those machines, you should also define a pattern for
10332 those cases, e.g., one matching
10335 (set @var{A} (neg:@var{m} (ne:@var{m} @var{B} @var{C})))
10338 Some machines can also perform @code{and} or @code{plus} operations on
10339 condition code values with less instructions than the corresponding
10340 @samp{cstore@var{mode}4} insn followed by @code{and} or @code{plus}. On those
10341 machines, define the appropriate patterns. Use the names @code{incscc}
10342 and @code{decscc}, respectively, for the patterns which perform
10343 @code{plus} or @code{minus} operations on condition code values. See
10344 @file{rs6000.md} for some examples. The GNU Superoptizer can be used to
10345 find such instruction sequences on other machines.
10347 If this macro is not defined, the default value, 1, is used. You need
10348 not define @code{STORE_FLAG_VALUE} if the machine has no store-flag
10349 instructions, or if the value generated by these instructions is 1.
10352 @defmac FLOAT_STORE_FLAG_VALUE (@var{mode})
10353 A C expression that gives a nonzero @code{REAL_VALUE_TYPE} value that is
10354 returned when comparison operators with floating-point results are true.
10355 Define this macro on machines that have comparison operations that return
10356 floating-point values. If there are no such operations, do not define
10360 @defmac VECTOR_STORE_FLAG_VALUE (@var{mode})
10361 A C expression that gives a rtx representing the nonzero true element
10362 for vector comparisons. The returned rtx should be valid for the inner
10363 mode of @var{mode} which is guaranteed to be a vector mode. Define
10364 this macro on machines that have vector comparison operations that
10365 return a vector result. If there are no such operations, do not define
10366 this macro. Typically, this macro is defined as @code{const1_rtx} or
10367 @code{constm1_rtx}. This macro may return @code{NULL_RTX} to prevent
10368 the compiler optimizing such vector comparison operations for the
10372 @defmac CLZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
10373 @defmacx CTZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
10374 A C expression that indicates whether the architecture defines a value
10375 for @code{clz} or @code{ctz} with a zero operand.
10376 A result of @code{0} indicates the value is undefined.
10377 If the value is defined for only the RTL expression, the macro should
10378 evaluate to @code{1}; if the value applies also to the corresponding optab
10379 entry (which is normally the case if it expands directly into
10380 the corresponding RTL), then the macro should evaluate to @code{2}.
10381 In the cases where the value is defined, @var{value} should be set to
10384 If this macro is not defined, the value of @code{clz} or
10385 @code{ctz} at zero is assumed to be undefined.
10387 This macro must be defined if the target's expansion for @code{ffs}
10388 relies on a particular value to get correct results. Otherwise it
10389 is not necessary, though it may be used to optimize some corner cases, and
10390 to provide a default expansion for the @code{ffs} optab.
10392 Note that regardless of this macro the ``definedness'' of @code{clz}
10393 and @code{ctz} at zero do @emph{not} extend to the builtin functions
10394 visible to the user. Thus one may be free to adjust the value at will
10395 to match the target expansion of these operations without fear of
10400 An alias for the machine mode for pointers. On most machines, define
10401 this to be the integer mode corresponding to the width of a hardware
10402 pointer; @code{SImode} on 32-bit machine or @code{DImode} on 64-bit machines.
10403 On some machines you must define this to be one of the partial integer
10404 modes, such as @code{PSImode}.
10406 The width of @code{Pmode} must be at least as large as the value of
10407 @code{POINTER_SIZE}. If it is not equal, you must define the macro
10408 @code{POINTERS_EXTEND_UNSIGNED} to specify how pointers are extended
10412 @defmac FUNCTION_MODE
10413 An alias for the machine mode used for memory references to functions
10414 being called, in @code{call} RTL expressions. On most CISC machines,
10415 where an instruction can begin at any byte address, this should be
10416 @code{QImode}. On most RISC machines, where all instructions have fixed
10417 size and alignment, this should be a mode with the same size and alignment
10418 as the machine instruction words - typically @code{SImode} or @code{HImode}.
10421 @defmac STDC_0_IN_SYSTEM_HEADERS
10422 In normal operation, the preprocessor expands @code{__STDC__} to the
10423 constant 1, to signify that GCC conforms to ISO Standard C@. On some
10424 hosts, like Solaris, the system compiler uses a different convention,
10425 where @code{__STDC__} is normally 0, but is 1 if the user specifies
10426 strict conformance to the C Standard.
10428 Defining @code{STDC_0_IN_SYSTEM_HEADERS} makes GNU CPP follows the host
10429 convention when processing system header files, but when processing user
10430 files @code{__STDC__} will always expand to 1.
10433 @defmac NO_IMPLICIT_EXTERN_C
10434 Define this macro if the system header files support C++ as well as C@.
10435 This macro inhibits the usual method of using system header files in
10436 C++, which is to pretend that the file's contents are enclosed in
10437 @samp{extern "C" @{@dots{}@}}.
10442 @defmac REGISTER_TARGET_PRAGMAS ()
10443 Define this macro if you want to implement any target-specific pragmas.
10444 If defined, it is a C expression which makes a series of calls to
10445 @code{c_register_pragma} or @code{c_register_pragma_with_expansion}
10446 for each pragma. The macro may also do any
10447 setup required for the pragmas.
10449 The primary reason to define this macro is to provide compatibility with
10450 other compilers for the same target. In general, we discourage
10451 definition of target-specific pragmas for GCC@.
10453 If the pragma can be implemented by attributes then you should consider
10454 defining the target hook @samp{TARGET_INSERT_ATTRIBUTES} as well.
10456 Preprocessor macros that appear on pragma lines are not expanded. All
10457 @samp{#pragma} directives that do not match any registered pragma are
10458 silently ignored, unless the user specifies @option{-Wunknown-pragmas}.
10461 @deftypefun void c_register_pragma (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
10462 @deftypefunx void c_register_pragma_with_expansion (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
10464 Each call to @code{c_register_pragma} or
10465 @code{c_register_pragma_with_expansion} establishes one pragma. The
10466 @var{callback} routine will be called when the preprocessor encounters a
10470 #pragma [@var{space}] @var{name} @dots{}
10473 @var{space} is the case-sensitive namespace of the pragma, or
10474 @code{NULL} to put the pragma in the global namespace. The callback
10475 routine receives @var{pfile} as its first argument, which can be passed
10476 on to cpplib's functions if necessary. You can lex tokens after the
10477 @var{name} by calling @code{pragma_lex}. Tokens that are not read by the
10478 callback will be silently ignored. The end of the line is indicated by
10479 a token of type @code{CPP_EOF}. Macro expansion occurs on the
10480 arguments of pragmas registered with
10481 @code{c_register_pragma_with_expansion} but not on the arguments of
10482 pragmas registered with @code{c_register_pragma}.
10484 Note that the use of @code{pragma_lex} is specific to the C and C++
10485 compilers. It will not work in the Java or Fortran compilers, or any
10486 other language compilers for that matter. Thus if @code{pragma_lex} is going
10487 to be called from target-specific code, it must only be done so when
10488 building the C and C++ compilers. This can be done by defining the
10489 variables @code{c_target_objs} and @code{cxx_target_objs} in the
10490 target entry in the @file{config.gcc} file. These variables should name
10491 the target-specific, language-specific object file which contains the
10492 code that uses @code{pragma_lex}. Note it will also be necessary to add a
10493 rule to the makefile fragment pointed to by @code{tmake_file} that shows
10494 how to build this object file.
10499 @defmac HANDLE_SYSV_PRAGMA
10500 Define this macro (to a value of 1) if you want the System V style
10501 pragmas @samp{#pragma pack(<n>)} and @samp{#pragma weak <name>
10502 [=<value>]} to be supported by gcc.
10504 The pack pragma specifies the maximum alignment (in bytes) of fields
10505 within a structure, in much the same way as the @samp{__aligned__} and
10506 @samp{__packed__} @code{__attribute__}s do. A pack value of zero resets
10507 the behavior to the default.
10509 A subtlety for Microsoft Visual C/C++ style bit-field packing
10510 (e.g.@: -mms-bitfields) for targets that support it:
10511 When a bit-field is inserted into a packed record, the whole size
10512 of the underlying type is used by one or more same-size adjacent
10513 bit-fields (that is, if its long:3, 32 bits is used in the record,
10514 and any additional adjacent long bit-fields are packed into the same
10515 chunk of 32 bits. However, if the size changes, a new field of that
10516 size is allocated).
10518 If both MS bit-fields and @samp{__attribute__((packed))} are used,
10519 the latter will take precedence. If @samp{__attribute__((packed))} is
10520 used on a single field when MS bit-fields are in use, it will take
10521 precedence for that field, but the alignment of the rest of the structure
10522 may affect its placement.
10524 The weak pragma only works if @code{SUPPORTS_WEAK} and
10525 @code{ASM_WEAKEN_LABEL} are defined. If enabled it allows the creation
10526 of specifically named weak labels, optionally with a value.
10531 @defmac HANDLE_PRAGMA_PACK_PUSH_POP
10532 Define this macro (to a value of 1) if you want to support the Win32
10533 style pragmas @samp{#pragma pack(push[,@var{n}])} and @samp{#pragma
10534 pack(pop)}. The @samp{pack(push,[@var{n}])} pragma specifies the maximum
10535 alignment (in bytes) of fields within a structure, in much the same way as
10536 the @samp{__aligned__} and @samp{__packed__} @code{__attribute__}s do. A
10537 pack value of zero resets the behavior to the default. Successive
10538 invocations of this pragma cause the previous values to be stacked, so
10539 that invocations of @samp{#pragma pack(pop)} will return to the previous
10543 @defmac HANDLE_PRAGMA_PACK_WITH_EXPANSION
10544 Define this macro, as well as
10545 @code{HANDLE_SYSV_PRAGMA}, if macros should be expanded in the
10546 arguments of @samp{#pragma pack}.
10549 @deftypevr {Target Hook} bool TARGET_HANDLE_PRAGMA_EXTERN_PREFIX
10550 True if @code{#pragma extern_prefix} is to be supported.
10553 @defmac TARGET_DEFAULT_PACK_STRUCT
10554 If your target requires a structure packing default other than 0 (meaning
10555 the machine default), define this macro to the necessary value (in bytes).
10556 This must be a value that would also be valid to use with
10557 @samp{#pragma pack()} (that is, a small power of two).
10560 @defmac DOLLARS_IN_IDENTIFIERS
10561 Define this macro to control use of the character @samp{$} in
10562 identifier names for the C family of languages. 0 means @samp{$} is
10563 not allowed by default; 1 means it is allowed. 1 is the default;
10564 there is no need to define this macro in that case.
10567 @defmac NO_DOLLAR_IN_LABEL
10568 Define this macro if the assembler does not accept the character
10569 @samp{$} in label names. By default constructors and destructors in
10570 G++ have @samp{$} in the identifiers. If this macro is defined,
10571 @samp{.} is used instead.
10574 @defmac NO_DOT_IN_LABEL
10575 Define this macro if the assembler does not accept the character
10576 @samp{.} in label names. By default constructors and destructors in G++
10577 have names that use @samp{.}. If this macro is defined, these names
10578 are rewritten to avoid @samp{.}.
10581 @defmac INSN_SETS_ARE_DELAYED (@var{insn})
10582 Define this macro as a C expression that is nonzero if it is safe for the
10583 delay slot scheduler to place instructions in the delay slot of @var{insn},
10584 even if they appear to use a resource set or clobbered in @var{insn}.
10585 @var{insn} is always a @code{jump_insn} or an @code{insn}; GCC knows that
10586 every @code{call_insn} has this behavior. On machines where some @code{insn}
10587 or @code{jump_insn} is really a function call and hence has this behavior,
10588 you should define this macro.
10590 You need not define this macro if it would always return zero.
10593 @defmac INSN_REFERENCES_ARE_DELAYED (@var{insn})
10594 Define this macro as a C expression that is nonzero if it is safe for the
10595 delay slot scheduler to place instructions in the delay slot of @var{insn},
10596 even if they appear to set or clobber a resource referenced in @var{insn}.
10597 @var{insn} is always a @code{jump_insn} or an @code{insn}. On machines where
10598 some @code{insn} or @code{jump_insn} is really a function call and its operands
10599 are registers whose use is actually in the subroutine it calls, you should
10600 define this macro. Doing so allows the delay slot scheduler to move
10601 instructions which copy arguments into the argument registers into the delay
10602 slot of @var{insn}.
10604 You need not define this macro if it would always return zero.
10607 @defmac MULTIPLE_SYMBOL_SPACES
10608 Define this macro as a C expression that is nonzero if, in some cases,
10609 global symbols from one translation unit may not be bound to undefined
10610 symbols in another translation unit without user intervention. For
10611 instance, under Microsoft Windows symbols must be explicitly imported
10612 from shared libraries (DLLs).
10614 You need not define this macro if it would always evaluate to zero.
10617 @deftypefn {Target Hook} tree TARGET_MD_ASM_CLOBBERS (tree @var{outputs}, tree @var{inputs}, tree @var{clobbers})
10618 This target hook should add to @var{clobbers} @code{STRING_CST} trees for
10619 any hard regs the port wishes to automatically clobber for an asm.
10620 It should return the result of the last @code{tree_cons} used to add a
10621 clobber. The @var{outputs}, @var{inputs} and @var{clobber} lists are the
10622 corresponding parameters to the asm and may be inspected to avoid
10623 clobbering a register that is an input or output of the asm. You can use
10624 @code{tree_overlaps_hard_reg_set}, declared in @file{tree.h}, to test
10625 for overlap with regards to asm-declared registers.
10628 @defmac MATH_LIBRARY
10629 Define this macro as a C string constant for the linker argument to link
10630 in the system math library, or @samp{""} if the target does not have a
10631 separate math library.
10633 You need only define this macro if the default of @samp{"-lm"} is wrong.
10636 @defmac LIBRARY_PATH_ENV
10637 Define this macro as a C string constant for the environment variable that
10638 specifies where the linker should look for libraries.
10640 You need only define this macro if the default of @samp{"LIBRARY_PATH"}
10644 @defmac TARGET_POSIX_IO
10645 Define this macro if the target supports the following POSIX@ file
10646 functions, access, mkdir and file locking with fcntl / F_SETLKW@.
10647 Defining @code{TARGET_POSIX_IO} will enable the test coverage code
10648 to use file locking when exiting a program, which avoids race conditions
10649 if the program has forked. It will also create directories at run-time
10650 for cross-profiling.
10653 @defmac MAX_CONDITIONAL_EXECUTE
10655 A C expression for the maximum number of instructions to execute via
10656 conditional execution instructions instead of a branch. A value of
10657 @code{BRANCH_COST}+1 is the default if the machine does not use cc0, and
10658 1 if it does use cc0.
10661 @defmac IFCVT_MODIFY_TESTS (@var{ce_info}, @var{true_expr}, @var{false_expr})
10662 Used if the target needs to perform machine-dependent modifications on the
10663 conditionals used for turning basic blocks into conditionally executed code.
10664 @var{ce_info} points to a data structure, @code{struct ce_if_block}, which
10665 contains information about the currently processed blocks. @var{true_expr}
10666 and @var{false_expr} are the tests that are used for converting the
10667 then-block and the else-block, respectively. Set either @var{true_expr} or
10668 @var{false_expr} to a null pointer if the tests cannot be converted.
10671 @defmac IFCVT_MODIFY_MULTIPLE_TESTS (@var{ce_info}, @var{bb}, @var{true_expr}, @var{false_expr})
10672 Like @code{IFCVT_MODIFY_TESTS}, but used when converting more complicated
10673 if-statements into conditions combined by @code{and} and @code{or} operations.
10674 @var{bb} contains the basic block that contains the test that is currently
10675 being processed and about to be turned into a condition.
10678 @defmac IFCVT_MODIFY_INSN (@var{ce_info}, @var{pattern}, @var{insn})
10679 A C expression to modify the @var{PATTERN} of an @var{INSN} that is to
10680 be converted to conditional execution format. @var{ce_info} points to
10681 a data structure, @code{struct ce_if_block}, which contains information
10682 about the currently processed blocks.
10685 @defmac IFCVT_MODIFY_FINAL (@var{ce_info})
10686 A C expression to perform any final machine dependent modifications in
10687 converting code to conditional execution. The involved basic blocks
10688 can be found in the @code{struct ce_if_block} structure that is pointed
10689 to by @var{ce_info}.
10692 @defmac IFCVT_MODIFY_CANCEL (@var{ce_info})
10693 A C expression to cancel any machine dependent modifications in
10694 converting code to conditional execution. The involved basic blocks
10695 can be found in the @code{struct ce_if_block} structure that is pointed
10696 to by @var{ce_info}.
10699 @defmac IFCVT_INIT_EXTRA_FIELDS (@var{ce_info})
10700 A C expression to initialize any extra fields in a @code{struct ce_if_block}
10701 structure, which are defined by the @code{IFCVT_EXTRA_FIELDS} macro.
10704 @defmac IFCVT_EXTRA_FIELDS
10705 If defined, it should expand to a set of field declarations that will be
10706 added to the @code{struct ce_if_block} structure. These should be initialized
10707 by the @code{IFCVT_INIT_EXTRA_FIELDS} macro.
10710 @deftypefn {Target Hook} void TARGET_MACHINE_DEPENDENT_REORG (void)
10711 If non-null, this hook performs a target-specific pass over the
10712 instruction stream. The compiler will run it at all optimization levels,
10713 just before the point at which it normally does delayed-branch scheduling.
10715 The exact purpose of the hook varies from target to target. Some use
10716 it to do transformations that are necessary for correctness, such as
10717 laying out in-function constant pools or avoiding hardware hazards.
10718 Others use it as an opportunity to do some machine-dependent optimizations.
10720 You need not implement the hook if it has nothing to do. The default
10721 definition is null.
10724 @deftypefn {Target Hook} void TARGET_INIT_BUILTINS (void)
10725 Define this hook if you have any machine-specific built-in functions
10726 that need to be defined. It should be a function that performs the
10729 Machine specific built-in functions can be useful to expand special machine
10730 instructions that would otherwise not normally be generated because
10731 they have no equivalent in the source language (for example, SIMD vector
10732 instructions or prefetch instructions).
10734 To create a built-in function, call the function
10735 @code{lang_hooks.builtin_function}
10736 which is defined by the language front end. You can use any type nodes set
10737 up by @code{build_common_tree_nodes} and @code{build_common_tree_nodes_2};
10738 only language front ends that use those two functions will call
10739 @samp{TARGET_INIT_BUILTINS}.
10742 @deftypefn {Target Hook} tree TARGET_BUILTIN_DECL (unsigned @var{code}, bool @var{initialize_p})
10743 Define this hook if you have any machine-specific built-in functions
10744 that need to be defined. It should be a function that returns the
10745 builtin function declaration for the builtin function code @var{code}.
10746 If there is no such builtin and it cannot be initialized at this time
10747 if @var{initialize_p} is true the function should return @code{NULL_TREE}.
10748 If @var{code} is out of range the function should return
10749 @code{error_mark_node}.
10752 @deftypefn {Target Hook} rtx TARGET_EXPAND_BUILTIN (tree @var{exp}, rtx @var{target}, rtx @var{subtarget}, enum machine_mode @var{mode}, int @var{ignore})
10754 Expand a call to a machine specific built-in function that was set up by
10755 @samp{TARGET_INIT_BUILTINS}. @var{exp} is the expression for the
10756 function call; the result should go to @var{target} if that is
10757 convenient, and have mode @var{mode} if that is convenient.
10758 @var{subtarget} may be used as the target for computing one of
10759 @var{exp}'s operands. @var{ignore} is nonzero if the value is to be
10760 ignored. This function should return the result of the call to the
10764 @deftypefn {Target Hook} tree TARGET_RESOLVE_OVERLOADED_BUILTIN (unsigned int @var{loc}, tree @var{fndecl}, void *@var{arglist})
10766 Select a replacement for a machine specific built-in function that
10767 was set up by @samp{TARGET_INIT_BUILTINS}. This is done
10768 @emph{before} regular type checking, and so allows the target to
10769 implement a crude form of function overloading. @var{fndecl} is the
10770 declaration of the built-in function. @var{arglist} is the list of
10771 arguments passed to the built-in function. The result is a
10772 complete expression that implements the operation, usually
10773 another @code{CALL_EXPR}.
10774 @var{arglist} really has type @samp{VEC(tree,gc)*}
10777 @deftypefn {Target Hook} tree TARGET_FOLD_BUILTIN (tree @var{fndecl}, int @var{n_args}, tree *@var{argp}, bool @var{ignore})
10779 Fold a call to a machine specific built-in function that was set up by
10780 @samp{TARGET_INIT_BUILTINS}. @var{fndecl} is the declaration of the
10781 built-in function. @var{n_args} is the number of arguments passed to
10782 the function; the arguments themselves are pointed to by @var{argp}.
10783 The result is another tree containing a simplified expression for the
10784 call's result. If @var{ignore} is true the value will be ignored.
10787 @deftypefn {Target Hook} {const char *} TARGET_INVALID_WITHIN_DOLOOP (const_rtx @var{insn})
10789 Take an instruction in @var{insn} and return NULL if it is valid within a
10790 low-overhead loop, otherwise return a string explaining why doloop
10791 could not be applied.
10793 Many targets use special registers for low-overhead looping. For any
10794 instruction that clobbers these this function should return a string indicating
10795 the reason why the doloop could not be applied.
10796 By default, the RTL loop optimizer does not use a present doloop pattern for
10797 loops containing function calls or branch on table instructions.
10800 @defmac MD_CAN_REDIRECT_BRANCH (@var{branch1}, @var{branch2})
10802 Take a branch insn in @var{branch1} and another in @var{branch2}.
10803 Return true if redirecting @var{branch1} to the destination of
10804 @var{branch2} is possible.
10806 On some targets, branches may have a limited range. Optimizing the
10807 filling of delay slots can result in branches being redirected, and this
10808 may in turn cause a branch offset to overflow.
10811 @deftypefn {Target Hook} bool TARGET_COMMUTATIVE_P (const_rtx @var{x}, int @var{outer_code})
10812 This target hook returns @code{true} if @var{x} is considered to be commutative.
10813 Usually, this is just COMMUTATIVE_P (@var{x}), but the HP PA doesn't consider
10814 PLUS to be commutative inside a MEM@. @var{outer_code} is the rtx code
10815 of the enclosing rtl, if known, otherwise it is UNKNOWN.
10818 @deftypefn {Target Hook} rtx TARGET_ALLOCATE_INITIAL_VALUE (rtx @var{hard_reg})
10820 When the initial value of a hard register has been copied in a pseudo
10821 register, it is often not necessary to actually allocate another register
10822 to this pseudo register, because the original hard register or a stack slot
10823 it has been saved into can be used. @code{TARGET_ALLOCATE_INITIAL_VALUE}
10824 is called at the start of register allocation once for each hard register
10825 that had its initial value copied by using
10826 @code{get_func_hard_reg_initial_val} or @code{get_hard_reg_initial_val}.
10827 Possible values are @code{NULL_RTX}, if you don't want
10828 to do any special allocation, a @code{REG} rtx---that would typically be
10829 the hard register itself, if it is known not to be clobbered---or a
10831 If you are returning a @code{MEM}, this is only a hint for the allocator;
10832 it might decide to use another register anyways.
10833 You may use @code{current_function_leaf_function} in the hook, functions
10834 that use @code{REG_N_SETS}, to determine if the hard
10835 register in question will not be clobbered.
10836 The default value of this hook is @code{NULL}, which disables any special
10840 @deftypefn {Target Hook} int TARGET_UNSPEC_MAY_TRAP_P (const_rtx @var{x}, unsigned @var{flags})
10841 This target hook returns nonzero if @var{x}, an @code{unspec} or
10842 @code{unspec_volatile} operation, might cause a trap. Targets can use
10843 this hook to enhance precision of analysis for @code{unspec} and
10844 @code{unspec_volatile} operations. You may call @code{may_trap_p_1}
10845 to analyze inner elements of @var{x} in which case @var{flags} should be
10849 @deftypefn {Target Hook} void TARGET_SET_CURRENT_FUNCTION (tree @var{decl})
10850 The compiler invokes this hook whenever it changes its current function
10851 context (@code{cfun}). You can define this function if
10852 the back end needs to perform any initialization or reset actions on a
10853 per-function basis. For example, it may be used to implement function
10854 attributes that affect register usage or code generation patterns.
10855 The argument @var{decl} is the declaration for the new function context,
10856 and may be null to indicate that the compiler has left a function context
10857 and is returning to processing at the top level.
10858 The default hook function does nothing.
10860 GCC sets @code{cfun} to a dummy function context during initialization of
10861 some parts of the back end. The hook function is not invoked in this
10862 situation; you need not worry about the hook being invoked recursively,
10863 or when the back end is in a partially-initialized state.
10864 @code{cfun} might be @code{NULL} to indicate processing at top level,
10865 outside of any function scope.
10868 @defmac TARGET_OBJECT_SUFFIX
10869 Define this macro to be a C string representing the suffix for object
10870 files on your target machine. If you do not define this macro, GCC will
10871 use @samp{.o} as the suffix for object files.
10874 @defmac TARGET_EXECUTABLE_SUFFIX
10875 Define this macro to be a C string representing the suffix to be
10876 automatically added to executable files on your target machine. If you
10877 do not define this macro, GCC will use the null string as the suffix for
10881 @defmac COLLECT_EXPORT_LIST
10882 If defined, @code{collect2} will scan the individual object files
10883 specified on its command line and create an export list for the linker.
10884 Define this macro for systems like AIX, where the linker discards
10885 object files that are not referenced from @code{main} and uses export
10889 @defmac MODIFY_JNI_METHOD_CALL (@var{mdecl})
10890 Define this macro to a C expression representing a variant of the
10891 method call @var{mdecl}, if Java Native Interface (JNI) methods
10892 must be invoked differently from other methods on your target.
10893 For example, on 32-bit Microsoft Windows, JNI methods must be invoked using
10894 the @code{stdcall} calling convention and this macro is then
10895 defined as this expression:
10898 build_type_attribute_variant (@var{mdecl},
10900 (get_identifier ("stdcall"),
10905 @deftypefn {Target Hook} bool TARGET_CANNOT_MODIFY_JUMPS_P (void)
10906 This target hook returns @code{true} past the point in which new jump
10907 instructions could be created. On machines that require a register for
10908 every jump such as the SHmedia ISA of SH5, this point would typically be
10909 reload, so this target hook should be defined to a function such as:
10913 cannot_modify_jumps_past_reload_p ()
10915 return (reload_completed || reload_in_progress);
10920 @deftypefn {Target Hook} {enum reg_class} TARGET_BRANCH_TARGET_REGISTER_CLASS (void)
10921 This target hook returns a register class for which branch target register
10922 optimizations should be applied. All registers in this class should be
10923 usable interchangeably. After reload, registers in this class will be
10924 re-allocated and loads will be hoisted out of loops and be subjected
10925 to inter-block scheduling.
10928 @deftypefn {Target Hook} bool TARGET_BRANCH_TARGET_REGISTER_CALLEE_SAVED (bool @var{after_prologue_epilogue_gen})
10929 Branch target register optimization will by default exclude callee-saved
10931 that are not already live during the current function; if this target hook
10932 returns true, they will be included. The target code must than make sure
10933 that all target registers in the class returned by
10934 @samp{TARGET_BRANCH_TARGET_REGISTER_CLASS} that might need saving are
10935 saved. @var{after_prologue_epilogue_gen} indicates if prologues and
10936 epilogues have already been generated. Note, even if you only return
10937 true when @var{after_prologue_epilogue_gen} is false, you still are likely
10938 to have to make special provisions in @code{INITIAL_ELIMINATION_OFFSET}
10939 to reserve space for caller-saved target registers.
10942 @deftypefn {Target Hook} bool TARGET_HAVE_CONDITIONAL_EXECUTION (void)
10943 This target hook returns true if the target supports conditional execution.
10944 This target hook is required only when the target has several different
10945 modes and they have different conditional execution capability, such as ARM.
10948 @deftypefn {Target Hook} unsigned TARGET_LOOP_UNROLL_ADJUST (unsigned @var{nunroll}, struct loop *@var{loop})
10949 This target hook returns a new value for the number of times @var{loop}
10950 should be unrolled. The parameter @var{nunroll} is the number of times
10951 the loop is to be unrolled. The parameter @var{loop} is a pointer to
10952 the loop, which is going to be checked for unrolling. This target hook
10953 is required only when the target has special constraints like maximum
10954 number of memory accesses.
10957 @defmac POWI_MAX_MULTS
10958 If defined, this macro is interpreted as a signed integer C expression
10959 that specifies the maximum number of floating point multiplications
10960 that should be emitted when expanding exponentiation by an integer
10961 constant inline. When this value is defined, exponentiation requiring
10962 more than this number of multiplications is implemented by calling the
10963 system library's @code{pow}, @code{powf} or @code{powl} routines.
10964 The default value places no upper bound on the multiplication count.
10967 @deftypefn Macro void TARGET_EXTRA_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
10968 This target hook should register any extra include files for the
10969 target. The parameter @var{stdinc} indicates if normal include files
10970 are present. The parameter @var{sysroot} is the system root directory.
10971 The parameter @var{iprefix} is the prefix for the gcc directory.
10974 @deftypefn Macro void TARGET_EXTRA_PRE_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
10975 This target hook should register any extra include files for the
10976 target before any standard headers. The parameter @var{stdinc}
10977 indicates if normal include files are present. The parameter
10978 @var{sysroot} is the system root directory. The parameter
10979 @var{iprefix} is the prefix for the gcc directory.
10982 @deftypefn Macro void TARGET_OPTF (char *@var{path})
10983 This target hook should register special include paths for the target.
10984 The parameter @var{path} is the include to register. On Darwin
10985 systems, this is used for Framework includes, which have semantics
10986 that are different from @option{-I}.
10989 @defmac bool TARGET_USE_LOCAL_THUNK_ALIAS_P (tree @var{fndecl})
10990 This target macro returns @code{true} if it is safe to use a local alias
10991 for a virtual function @var{fndecl} when constructing thunks,
10992 @code{false} otherwise. By default, the macro returns @code{true} for all
10993 functions, if a target supports aliases (i.e.@: defines
10994 @code{ASM_OUTPUT_DEF}), @code{false} otherwise,
10997 @defmac TARGET_FORMAT_TYPES
10998 If defined, this macro is the name of a global variable containing
10999 target-specific format checking information for the @option{-Wformat}
11000 option. The default is to have no target-specific format checks.
11003 @defmac TARGET_N_FORMAT_TYPES
11004 If defined, this macro is the number of entries in
11005 @code{TARGET_FORMAT_TYPES}.
11008 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES
11009 If defined, this macro is the name of a global variable containing
11010 target-specific format overrides for the @option{-Wformat} option. The
11011 default is to have no target-specific format overrides. If defined,
11012 @code{TARGET_FORMAT_TYPES} must be defined, too.
11015 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES_COUNT
11016 If defined, this macro specifies the number of entries in
11017 @code{TARGET_OVERRIDES_FORMAT_ATTRIBUTES}.
11020 @defmac TARGET_OVERRIDES_FORMAT_INIT
11021 If defined, this macro specifies the optional initialization
11022 routine for target specific customizations of the system printf
11023 and scanf formatter settings.
11026 @deftypevr {Target Hook} bool TARGET_RELAXED_ORDERING
11027 If set to @code{true}, means that the target's memory model does not
11028 guarantee that loads which do not depend on one another will access
11029 main memory in the order of the instruction stream; if ordering is
11030 important, an explicit memory barrier must be used. This is true of
11031 many recent processors which implement a policy of ``relaxed,''
11032 ``weak,'' or ``release'' memory consistency, such as Alpha, PowerPC,
11033 and ia64. The default is @code{false}.
11036 @deftypefn {Target Hook} {const char *} TARGET_INVALID_ARG_FOR_UNPROTOTYPED_FN (const_tree @var{typelist}, const_tree @var{funcdecl}, const_tree @var{val})
11037 If defined, this macro returns the diagnostic message when it is
11038 illegal to pass argument @var{val} to function @var{funcdecl}
11039 with prototype @var{typelist}.
11042 @deftypefn {Target Hook} {const char *} TARGET_INVALID_CONVERSION (const_tree @var{fromtype}, const_tree @var{totype})
11043 If defined, this macro returns the diagnostic message when it is
11044 invalid to convert from @var{fromtype} to @var{totype}, or @code{NULL}
11045 if validity should be determined by the front end.
11048 @deftypefn {Target Hook} {const char *} TARGET_INVALID_UNARY_OP (int @var{op}, const_tree @var{type})
11049 If defined, this macro returns the diagnostic message when it is
11050 invalid to apply operation @var{op} (where unary plus is denoted by
11051 @code{CONVERT_EXPR}) to an operand of type @var{type}, or @code{NULL}
11052 if validity should be determined by the front end.
11055 @deftypefn {Target Hook} {const char *} TARGET_INVALID_BINARY_OP (int @var{op}, const_tree @var{type1}, const_tree @var{type2})
11056 If defined, this macro returns the diagnostic message when it is
11057 invalid to apply operation @var{op} to operands of types @var{type1}
11058 and @var{type2}, or @code{NULL} if validity should be determined by
11062 @deftypefn {Target Hook} {const char *} TARGET_INVALID_PARAMETER_TYPE (const_tree @var{type})
11063 If defined, this macro returns the diagnostic message when it is
11064 invalid for functions to include parameters of type @var{type},
11065 or @code{NULL} if validity should be determined by
11066 the front end. This is currently used only by the C and C++ front ends.
11069 @deftypefn {Target Hook} {const char *} TARGET_INVALID_RETURN_TYPE (const_tree @var{type})
11070 If defined, this macro returns the diagnostic message when it is
11071 invalid for functions to have return type @var{type},
11072 or @code{NULL} if validity should be determined by
11073 the front end. This is currently used only by the C and C++ front ends.
11076 @deftypefn {Target Hook} tree TARGET_PROMOTED_TYPE (const_tree @var{type})
11077 If defined, this target hook returns the type to which values of
11078 @var{type} should be promoted when they appear in expressions,
11079 analogous to the integer promotions, or @code{NULL_TREE} to use the
11080 front end's normal promotion rules. This hook is useful when there are
11081 target-specific types with special promotion rules.
11082 This is currently used only by the C and C++ front ends.
11085 @deftypefn {Target Hook} tree TARGET_CONVERT_TO_TYPE (tree @var{type}, tree @var{expr})
11086 If defined, this hook returns the result of converting @var{expr} to
11087 @var{type}. It should return the converted expression,
11088 or @code{NULL_TREE} to apply the front end's normal conversion rules.
11089 This hook is useful when there are target-specific types with special
11091 This is currently used only by the C and C++ front ends.
11094 @defmac TARGET_USE_JCR_SECTION
11095 This macro determines whether to use the JCR section to register Java
11096 classes. By default, TARGET_USE_JCR_SECTION is defined to 1 if both
11097 SUPPORTS_WEAK and TARGET_HAVE_NAMED_SECTIONS are true, else 0.
11101 This macro determines the size of the objective C jump buffer for the
11102 NeXT runtime. By default, OBJC_JBLEN is defined to an innocuous value.
11105 @defmac LIBGCC2_UNWIND_ATTRIBUTE
11106 Define this macro if any target-specific attributes need to be attached
11107 to the functions in @file{libgcc} that provide low-level support for
11108 call stack unwinding. It is used in declarations in @file{unwind-generic.h}
11109 and the associated definitions of those functions.
11112 @deftypefn {Target Hook} void TARGET_UPDATE_STACK_BOUNDARY (void)
11113 Define this macro to update the current function stack boundary if
11117 @deftypefn {Target Hook} rtx TARGET_GET_DRAP_RTX (void)
11118 This hook should return an rtx for Dynamic Realign Argument Pointer (DRAP) if a
11119 different argument pointer register is needed to access the function's
11120 argument list due to stack realignment. Return @code{NULL} if no DRAP
11124 @deftypefn {Target Hook} bool TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS (void)
11125 When optimization is disabled, this hook indicates whether or not
11126 arguments should be allocated to stack slots. Normally, GCC allocates
11127 stacks slots for arguments when not optimizing in order to make
11128 debugging easier. However, when a function is declared with
11129 @code{__attribute__((naked))}, there is no stack frame, and the compiler
11130 cannot safely move arguments from the registers in which they are passed
11131 to the stack. Therefore, this hook should return true in general, but
11132 false for naked functions. The default implementation always returns true.
11135 @deftypevr {Target Hook} {unsigned HOST_WIDE_INT} TARGET_CONST_ANCHOR
11136 On some architectures it can take multiple instructions to synthesize
11137 a constant. If there is another constant already in a register that
11138 is close enough in value then it is preferable that the new constant
11139 is computed from this register using immediate addition or
11140 subtraction. We accomplish this through CSE. Besides the value of
11141 the constant we also add a lower and an upper constant anchor to the
11142 available expressions. These are then queried when encountering new
11143 constants. The anchors are computed by rounding the constant up and
11144 down to a multiple of the value of @code{TARGET_CONST_ANCHOR}.
11145 @code{TARGET_CONST_ANCHOR} should be the maximum positive value
11146 accepted by immediate-add plus one. We currently assume that the
11147 value of @code{TARGET_CONST_ANCHOR} is a power of 2. For example, on
11148 MIPS, where add-immediate takes a 16-bit signed value,
11149 @code{TARGET_CONST_ANCHOR} is set to @samp{0x8000}. The default value
11150 is zero, which disables this optimization. @end deftypevr