1 @c Copyright (C) 1988,1989,1992,1993,1994,1995,1996,1997,1998,1999,2000,2001,
2 @c 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009
3 @c Free Software Foundation, Inc.
4 @c This is part of the GCC manual.
5 @c For copying conditions, see the file gcc.texi.
8 @chapter Target Description Macros and Functions
9 @cindex machine description macros
10 @cindex target description macros
11 @cindex macros, target description
12 @cindex @file{tm.h} macros
14 In addition to the file @file{@var{machine}.md}, a machine description
15 includes a C header file conventionally given the name
16 @file{@var{machine}.h} and a C source file named @file{@var{machine}.c}.
17 The header file defines numerous macros that convey the information
18 about the target machine that does not fit into the scheme of the
19 @file{.md} file. The file @file{tm.h} should be a link to
20 @file{@var{machine}.h}. The header file @file{config.h} includes
21 @file{tm.h} and most compiler source files include @file{config.h}. The
22 source file defines a variable @code{targetm}, which is a structure
23 containing pointers to functions and data relating to the target
24 machine. @file{@var{machine}.c} should also contain their definitions,
25 if they are not defined elsewhere in GCC, and other functions called
26 through the macros defined in the @file{.h} file.
29 * Target Structure:: The @code{targetm} variable.
30 * Driver:: Controlling how the driver runs the compilation passes.
31 * Run-time Target:: Defining @samp{-m} options like @option{-m68000} and @option{-m68020}.
32 * Per-Function Data:: Defining data structures for per-function information.
33 * Storage Layout:: Defining sizes and alignments of data.
34 * Type Layout:: Defining sizes and properties of basic user data types.
35 * Registers:: Naming and describing the hardware registers.
36 * Register Classes:: Defining the classes of hardware registers.
37 * Old Constraints:: The old way to define machine-specific constraints.
38 * Stack and Calling:: Defining which way the stack grows and by how much.
39 * Varargs:: Defining the varargs macros.
40 * Trampolines:: Code set up at run time to enter a nested function.
41 * Library Calls:: Controlling how library routines are implicitly called.
42 * Addressing Modes:: Defining addressing modes valid for memory operands.
43 * Anchored Addresses:: Defining how @option{-fsection-anchors} should work.
44 * Condition Code:: Defining how insns update the condition code.
45 * Costs:: Defining relative costs of different operations.
46 * Scheduling:: Adjusting the behavior of the instruction scheduler.
47 * Sections:: Dividing storage into text, data, and other sections.
48 * PIC:: Macros for position independent code.
49 * Assembler Format:: Defining how to write insns and pseudo-ops to output.
50 * Debugging Info:: Defining the format of debugging output.
51 * Floating Point:: Handling floating point for cross-compilers.
52 * Mode Switching:: Insertion of mode-switching instructions.
53 * Target Attributes:: Defining target-specific uses of @code{__attribute__}.
54 * Emulated TLS:: Emulated TLS support.
55 * MIPS Coprocessors:: MIPS coprocessor support and how to customize it.
56 * PCH Target:: Validity checking for precompiled headers.
57 * C++ ABI:: Controlling C++ ABI changes.
58 * Misc:: Everything else.
61 @node Target Structure
62 @section The Global @code{targetm} Variable
64 @cindex target functions
66 @deftypevar {struct gcc_target} targetm
67 The target @file{.c} file must define the global @code{targetm} variable
68 which contains pointers to functions and data relating to the target
69 machine. The variable is declared in @file{target.h};
70 @file{target-def.h} defines the macro @code{TARGET_INITIALIZER} which is
71 used to initialize the variable, and macros for the default initializers
72 for elements of the structure. The @file{.c} file should override those
73 macros for which the default definition is inappropriate. For example:
76 #include "target-def.h"
78 /* @r{Initialize the GCC target structure.} */
80 #undef TARGET_COMP_TYPE_ATTRIBUTES
81 #define TARGET_COMP_TYPE_ATTRIBUTES @var{machine}_comp_type_attributes
83 struct gcc_target targetm = TARGET_INITIALIZER;
87 Where a macro should be defined in the @file{.c} file in this manner to
88 form part of the @code{targetm} structure, it is documented below as a
89 ``Target Hook'' with a prototype. Many macros will change in future
90 from being defined in the @file{.h} file to being part of the
91 @code{targetm} structure.
94 @section Controlling the Compilation Driver, @file{gcc}
96 @cindex controlling the compilation driver
98 @c prevent bad page break with this line
99 You can control the compilation driver.
101 @defmac SWITCH_TAKES_ARG (@var{char})
102 A C expression which determines whether the option @option{-@var{char}}
103 takes arguments. The value should be the number of arguments that
104 option takes--zero, for many options.
106 By default, this macro is defined as
107 @code{DEFAULT_SWITCH_TAKES_ARG}, which handles the standard options
108 properly. You need not define @code{SWITCH_TAKES_ARG} unless you
109 wish to add additional options which take arguments. Any redefinition
110 should call @code{DEFAULT_SWITCH_TAKES_ARG} and then check for
114 @defmac WORD_SWITCH_TAKES_ARG (@var{name})
115 A C expression which determines whether the option @option{-@var{name}}
116 takes arguments. The value should be the number of arguments that
117 option takes--zero, for many options. This macro rather than
118 @code{SWITCH_TAKES_ARG} is used for multi-character option names.
120 By default, this macro is defined as
121 @code{DEFAULT_WORD_SWITCH_TAKES_ARG}, which handles the standard options
122 properly. You need not define @code{WORD_SWITCH_TAKES_ARG} unless you
123 wish to add additional options which take arguments. Any redefinition
124 should call @code{DEFAULT_WORD_SWITCH_TAKES_ARG} and then check for
128 @defmac SWITCH_CURTAILS_COMPILATION (@var{char})
129 A C expression which determines whether the option @option{-@var{char}}
130 stops compilation before the generation of an executable. The value is
131 boolean, nonzero if the option does stop an executable from being
132 generated, zero otherwise.
134 By default, this macro is defined as
135 @code{DEFAULT_SWITCH_CURTAILS_COMPILATION}, which handles the standard
136 options properly. You need not define
137 @code{SWITCH_CURTAILS_COMPILATION} unless you wish to add additional
138 options which affect the generation of an executable. Any redefinition
139 should call @code{DEFAULT_SWITCH_CURTAILS_COMPILATION} and then check
140 for additional options.
143 @defmac SWITCHES_NEED_SPACES
144 A string-valued C expression which enumerates the options for which
145 the linker needs a space between the option and its argument.
147 If this macro is not defined, the default value is @code{""}.
150 @defmac TARGET_OPTION_TRANSLATE_TABLE
151 If defined, a list of pairs of strings, the first of which is a
152 potential command line target to the @file{gcc} driver program, and the
153 second of which is a space-separated (tabs and other whitespace are not
154 supported) list of options with which to replace the first option. The
155 target defining this list is responsible for assuring that the results
156 are valid. Replacement options may not be the @code{--opt} style, they
157 must be the @code{-opt} style. It is the intention of this macro to
158 provide a mechanism for substitution that affects the multilibs chosen,
159 such as one option that enables many options, some of which select
160 multilibs. Example nonsensical definition, where @option{-malt-abi},
161 @option{-EB}, and @option{-mspoo} cause different multilibs to be chosen:
164 #define TARGET_OPTION_TRANSLATE_TABLE \
165 @{ "-fast", "-march=fast-foo -malt-abi -I/usr/fast-foo" @}, \
166 @{ "-compat", "-EB -malign=4 -mspoo" @}
170 @defmac DRIVER_SELF_SPECS
171 A list of specs for the driver itself. It should be a suitable
172 initializer for an array of strings, with no surrounding braces.
174 The driver applies these specs to its own command line between loading
175 default @file{specs} files (but not command-line specified ones) and
176 choosing the multilib directory or running any subcommands. It
177 applies them in the order given, so each spec can depend on the
178 options added by earlier ones. It is also possible to remove options
179 using @samp{%<@var{option}} in the usual way.
181 This macro can be useful when a port has several interdependent target
182 options. It provides a way of standardizing the command line so
183 that the other specs are easier to write.
185 Do not define this macro if it does not need to do anything.
188 @defmac OPTION_DEFAULT_SPECS
189 A list of specs used to support configure-time default options (i.e.@:
190 @option{--with} options) in the driver. It should be a suitable initializer
191 for an array of structures, each containing two strings, without the
192 outermost pair of surrounding braces.
194 The first item in the pair is the name of the default. This must match
195 the code in @file{config.gcc} for the target. The second item is a spec
196 to apply if a default with this name was specified. The string
197 @samp{%(VALUE)} in the spec will be replaced by the value of the default
198 everywhere it occurs.
200 The driver will apply these specs to its own command line between loading
201 default @file{specs} files and processing @code{DRIVER_SELF_SPECS}, using
202 the same mechanism as @code{DRIVER_SELF_SPECS}.
204 Do not define this macro if it does not need to do anything.
208 A C string constant that tells the GCC driver program options to
209 pass to CPP@. It can also specify how to translate options you
210 give to GCC into options for GCC to pass to the CPP@.
212 Do not define this macro if it does not need to do anything.
215 @defmac CPLUSPLUS_CPP_SPEC
216 This macro is just like @code{CPP_SPEC}, but is used for C++, rather
217 than C@. If you do not define this macro, then the value of
218 @code{CPP_SPEC} (if any) will be used instead.
222 A C string constant that tells the GCC driver program options to
223 pass to @code{cc1}, @code{cc1plus}, @code{f771}, and the other language
225 It can also specify how to translate options you give to GCC into options
226 for GCC to pass to front ends.
228 Do not define this macro if it does not need to do anything.
232 A C string constant that tells the GCC driver program options to
233 pass to @code{cc1plus}. It can also specify how to translate options you
234 give to GCC into options for GCC to pass to the @code{cc1plus}.
236 Do not define this macro if it does not need to do anything.
237 Note that everything defined in CC1_SPEC is already passed to
238 @code{cc1plus} so there is no need to duplicate the contents of
239 CC1_SPEC in CC1PLUS_SPEC@.
243 A C string constant that tells the GCC driver program options to
244 pass to the assembler. It can also specify how to translate options
245 you give to GCC into options for GCC to pass to the assembler.
246 See the file @file{sun3.h} for an example of this.
248 Do not define this macro if it does not need to do anything.
251 @defmac ASM_FINAL_SPEC
252 A C string constant that tells the GCC driver program how to
253 run any programs which cleanup after the normal assembler.
254 Normally, this is not needed. See the file @file{mips.h} for
257 Do not define this macro if it does not need to do anything.
260 @defmac AS_NEEDS_DASH_FOR_PIPED_INPUT
261 Define this macro, with no value, if the driver should give the assembler
262 an argument consisting of a single dash, @option{-}, to instruct it to
263 read from its standard input (which will be a pipe connected to the
264 output of the compiler proper). This argument is given after any
265 @option{-o} option specifying the name of the output file.
267 If you do not define this macro, the assembler is assumed to read its
268 standard input if given no non-option arguments. If your assembler
269 cannot read standard input at all, use a @samp{%@{pipe:%e@}} construct;
270 see @file{mips.h} for instance.
274 A C string constant that tells the GCC driver program options to
275 pass to the linker. It can also specify how to translate options you
276 give to GCC into options for GCC to pass to the linker.
278 Do not define this macro if it does not need to do anything.
282 Another C string constant used much like @code{LINK_SPEC}. The difference
283 between the two is that @code{LIB_SPEC} is used at the end of the
284 command given to the linker.
286 If this macro is not defined, a default is provided that
287 loads the standard C library from the usual place. See @file{gcc.c}.
291 Another C string constant that tells the GCC driver program
292 how and when to place a reference to @file{libgcc.a} into the
293 linker command line. This constant is placed both before and after
294 the value of @code{LIB_SPEC}.
296 If this macro is not defined, the GCC driver provides a default that
297 passes the string @option{-lgcc} to the linker.
300 @defmac REAL_LIBGCC_SPEC
301 By default, if @code{ENABLE_SHARED_LIBGCC} is defined, the
302 @code{LIBGCC_SPEC} is not directly used by the driver program but is
303 instead modified to refer to different versions of @file{libgcc.a}
304 depending on the values of the command line flags @option{-static},
305 @option{-shared}, @option{-static-libgcc}, and @option{-shared-libgcc}. On
306 targets where these modifications are inappropriate, define
307 @code{REAL_LIBGCC_SPEC} instead. @code{REAL_LIBGCC_SPEC} tells the
308 driver how to place a reference to @file{libgcc} on the link command
309 line, but, unlike @code{LIBGCC_SPEC}, it is used unmodified.
312 @defmac USE_LD_AS_NEEDED
313 A macro that controls the modifications to @code{LIBGCC_SPEC}
314 mentioned in @code{REAL_LIBGCC_SPEC}. If nonzero, a spec will be
315 generated that uses --as-needed and the shared libgcc in place of the
316 static exception handler library, when linking without any of
317 @code{-static}, @code{-static-libgcc}, or @code{-shared-libgcc}.
321 If defined, this C string constant is added to @code{LINK_SPEC}.
322 When @code{USE_LD_AS_NEEDED} is zero or undefined, it also affects
323 the modifications to @code{LIBGCC_SPEC} mentioned in
324 @code{REAL_LIBGCC_SPEC}.
327 @defmac STARTFILE_SPEC
328 Another C string constant used much like @code{LINK_SPEC}. The
329 difference between the two is that @code{STARTFILE_SPEC} is used at
330 the very beginning of the command given to the linker.
332 If this macro is not defined, a default is provided that loads the
333 standard C startup file from the usual place. See @file{gcc.c}.
337 Another C string constant used much like @code{LINK_SPEC}. The
338 difference between the two is that @code{ENDFILE_SPEC} is used at
339 the very end of the command given to the linker.
341 Do not define this macro if it does not need to do anything.
344 @defmac THREAD_MODEL_SPEC
345 GCC @code{-v} will print the thread model GCC was configured to use.
346 However, this doesn't work on platforms that are multilibbed on thread
347 models, such as AIX 4.3. On such platforms, define
348 @code{THREAD_MODEL_SPEC} such that it evaluates to a string without
349 blanks that names one of the recognized thread models. @code{%*}, the
350 default value of this macro, will expand to the value of
351 @code{thread_file} set in @file{config.gcc}.
354 @defmac SYSROOT_SUFFIX_SPEC
355 Define this macro to add a suffix to the target sysroot when GCC is
356 configured with a sysroot. This will cause GCC to search for usr/lib,
357 et al, within sysroot+suffix.
360 @defmac SYSROOT_HEADERS_SUFFIX_SPEC
361 Define this macro to add a headers_suffix to the target sysroot when
362 GCC is configured with a sysroot. This will cause GCC to pass the
363 updated sysroot+headers_suffix to CPP, causing it to search for
364 usr/include, et al, within sysroot+headers_suffix.
368 Define this macro to provide additional specifications to put in the
369 @file{specs} file that can be used in various specifications like
372 The definition should be an initializer for an array of structures,
373 containing a string constant, that defines the specification name, and a
374 string constant that provides the specification.
376 Do not define this macro if it does not need to do anything.
378 @code{EXTRA_SPECS} is useful when an architecture contains several
379 related targets, which have various @code{@dots{}_SPECS} which are similar
380 to each other, and the maintainer would like one central place to keep
383 For example, the PowerPC System V.4 targets use @code{EXTRA_SPECS} to
384 define either @code{_CALL_SYSV} when the System V calling sequence is
385 used or @code{_CALL_AIX} when the older AIX-based calling sequence is
388 The @file{config/rs6000/rs6000.h} target file defines:
391 #define EXTRA_SPECS \
392 @{ "cpp_sysv_default", CPP_SYSV_DEFAULT @},
394 #define CPP_SYS_DEFAULT ""
397 The @file{config/rs6000/sysv.h} target file defines:
401 "%@{posix: -D_POSIX_SOURCE @} \
402 %@{mcall-sysv: -D_CALL_SYSV @} \
403 %@{!mcall-sysv: %(cpp_sysv_default) @} \
404 %@{msoft-float: -D_SOFT_FLOAT@} %@{mcpu=403: -D_SOFT_FLOAT@}"
406 #undef CPP_SYSV_DEFAULT
407 #define CPP_SYSV_DEFAULT "-D_CALL_SYSV"
410 while the @file{config/rs6000/eabiaix.h} target file defines
411 @code{CPP_SYSV_DEFAULT} as:
414 #undef CPP_SYSV_DEFAULT
415 #define CPP_SYSV_DEFAULT "-D_CALL_AIX"
419 @defmac LINK_LIBGCC_SPECIAL_1
420 Define this macro if the driver program should find the library
421 @file{libgcc.a}. If you do not define this macro, the driver program will pass
422 the argument @option{-lgcc} to tell the linker to do the search.
425 @defmac LINK_GCC_C_SEQUENCE_SPEC
426 The sequence in which libgcc and libc are specified to the linker.
427 By default this is @code{%G %L %G}.
430 @defmac LINK_COMMAND_SPEC
431 A C string constant giving the complete command line need to execute the
432 linker. When you do this, you will need to update your port each time a
433 change is made to the link command line within @file{gcc.c}. Therefore,
434 define this macro only if you need to completely redefine the command
435 line for invoking the linker and there is no other way to accomplish
436 the effect you need. Overriding this macro may be avoidable by overriding
437 @code{LINK_GCC_C_SEQUENCE_SPEC} instead.
440 @defmac LINK_ELIMINATE_DUPLICATE_LDIRECTORIES
441 A nonzero value causes @command{collect2} to remove duplicate @option{-L@var{directory}} search
442 directories from linking commands. Do not give it a nonzero value if
443 removing duplicate search directories changes the linker's semantics.
446 @defmac MULTILIB_DEFAULTS
447 Define this macro as a C expression for the initializer of an array of
448 string to tell the driver program which options are defaults for this
449 target and thus do not need to be handled specially when using
450 @code{MULTILIB_OPTIONS}.
452 Do not define this macro if @code{MULTILIB_OPTIONS} is not defined in
453 the target makefile fragment or if none of the options listed in
454 @code{MULTILIB_OPTIONS} are set by default.
455 @xref{Target Fragment}.
458 @defmac RELATIVE_PREFIX_NOT_LINKDIR
459 Define this macro to tell @command{gcc} that it should only translate
460 a @option{-B} prefix into a @option{-L} linker option if the prefix
461 indicates an absolute file name.
464 @defmac MD_EXEC_PREFIX
465 If defined, this macro is an additional prefix to try after
466 @code{STANDARD_EXEC_PREFIX}. @code{MD_EXEC_PREFIX} is not searched
467 when the @option{-b} option is used, or the compiler is built as a cross
468 compiler. If you define @code{MD_EXEC_PREFIX}, then be sure to add it
469 to the list of directories used to find the assembler in @file{configure.in}.
472 @defmac STANDARD_STARTFILE_PREFIX
473 Define this macro as a C string constant if you wish to override the
474 standard choice of @code{libdir} as the default prefix to
475 try when searching for startup files such as @file{crt0.o}.
476 @code{STANDARD_STARTFILE_PREFIX} is not searched when the compiler
477 is built as a cross compiler.
480 @defmac STANDARD_STARTFILE_PREFIX_1
481 Define this macro as a C string constant if you wish to override the
482 standard choice of @code{/lib} as a prefix to try after the default prefix
483 when searching for startup files such as @file{crt0.o}.
484 @code{STANDARD_STARTFILE_PREFIX_1} is not searched when the compiler
485 is built as a cross compiler.
488 @defmac STANDARD_STARTFILE_PREFIX_2
489 Define this macro as a C string constant if you wish to override the
490 standard choice of @code{/lib} as yet another prefix to try after the
491 default prefix when searching for startup files such as @file{crt0.o}.
492 @code{STANDARD_STARTFILE_PREFIX_2} is not searched when the compiler
493 is built as a cross compiler.
496 @defmac MD_STARTFILE_PREFIX
497 If defined, this macro supplies an additional prefix to try after the
498 standard prefixes. @code{MD_EXEC_PREFIX} is not searched when the
499 @option{-b} option is used, or when the compiler is built as a cross
503 @defmac MD_STARTFILE_PREFIX_1
504 If defined, this macro supplies yet another prefix to try after the
505 standard prefixes. It is not searched when the @option{-b} option is
506 used, or when the compiler is built as a cross compiler.
509 @defmac INIT_ENVIRONMENT
510 Define this macro as a C string constant if you wish to set environment
511 variables for programs called by the driver, such as the assembler and
512 loader. The driver passes the value of this macro to @code{putenv} to
513 initialize the necessary environment variables.
516 @defmac LOCAL_INCLUDE_DIR
517 Define this macro as a C string constant if you wish to override the
518 standard choice of @file{/usr/local/include} as the default prefix to
519 try when searching for local header files. @code{LOCAL_INCLUDE_DIR}
520 comes before @code{SYSTEM_INCLUDE_DIR} in the search order.
522 Cross compilers do not search either @file{/usr/local/include} or its
526 @defmac MODIFY_TARGET_NAME
527 Define this macro if you wish to define command-line switches that
528 modify the default target name.
530 For each switch, you can include a string to be appended to the first
531 part of the configuration name or a string to be deleted from the
532 configuration name, if present. The definition should be an initializer
533 for an array of structures. Each array element should have three
534 elements: the switch name (a string constant, including the initial
535 dash), one of the enumeration codes @code{ADD} or @code{DELETE} to
536 indicate whether the string should be inserted or deleted, and the string
537 to be inserted or deleted (a string constant).
539 For example, on a machine where @samp{64} at the end of the
540 configuration name denotes a 64-bit target and you want the @option{-32}
541 and @option{-64} switches to select between 32- and 64-bit targets, you would
545 #define MODIFY_TARGET_NAME \
546 @{ @{ "-32", DELETE, "64"@}, \
547 @{"-64", ADD, "64"@}@}
551 @defmac SYSTEM_INCLUDE_DIR
552 Define this macro as a C string constant if you wish to specify a
553 system-specific directory to search for header files before the standard
554 directory. @code{SYSTEM_INCLUDE_DIR} comes before
555 @code{STANDARD_INCLUDE_DIR} in the search order.
557 Cross compilers do not use this macro and do not search the directory
561 @defmac STANDARD_INCLUDE_DIR
562 Define this macro as a C string constant if you wish to override the
563 standard choice of @file{/usr/include} as the default prefix to
564 try when searching for header files.
566 Cross compilers ignore this macro and do not search either
567 @file{/usr/include} or its replacement.
570 @defmac STANDARD_INCLUDE_COMPONENT
571 The ``component'' corresponding to @code{STANDARD_INCLUDE_DIR}.
572 See @code{INCLUDE_DEFAULTS}, below, for the description of components.
573 If you do not define this macro, no component is used.
576 @defmac INCLUDE_DEFAULTS
577 Define this macro if you wish to override the entire default search path
578 for include files. For a native compiler, the default search path
579 usually consists of @code{GCC_INCLUDE_DIR}, @code{LOCAL_INCLUDE_DIR},
580 @code{SYSTEM_INCLUDE_DIR}, @code{GPLUSPLUS_INCLUDE_DIR}, and
581 @code{STANDARD_INCLUDE_DIR}. In addition, @code{GPLUSPLUS_INCLUDE_DIR}
582 and @code{GCC_INCLUDE_DIR} are defined automatically by @file{Makefile},
583 and specify private search areas for GCC@. The directory
584 @code{GPLUSPLUS_INCLUDE_DIR} is used only for C++ programs.
586 The definition should be an initializer for an array of structures.
587 Each array element should have four elements: the directory name (a
588 string constant), the component name (also a string constant), a flag
589 for C++-only directories,
590 and a flag showing that the includes in the directory don't need to be
591 wrapped in @code{extern @samp{C}} when compiling C++. Mark the end of
592 the array with a null element.
594 The component name denotes what GNU package the include file is part of,
595 if any, in all uppercase letters. For example, it might be @samp{GCC}
596 or @samp{BINUTILS}. If the package is part of a vendor-supplied
597 operating system, code the component name as @samp{0}.
599 For example, here is the definition used for VAX/VMS:
602 #define INCLUDE_DEFAULTS \
604 @{ "GNU_GXX_INCLUDE:", "G++", 1, 1@}, \
605 @{ "GNU_CC_INCLUDE:", "GCC", 0, 0@}, \
606 @{ "SYS$SYSROOT:[SYSLIB.]", 0, 0, 0@}, \
613 Here is the order of prefixes tried for exec files:
617 Any prefixes specified by the user with @option{-B}.
620 The environment variable @code{GCC_EXEC_PREFIX} or, if @code{GCC_EXEC_PREFIX}
621 is not set and the compiler has not been installed in the configure-time
622 @var{prefix}, the location in which the compiler has actually been installed.
625 The directories specified by the environment variable @code{COMPILER_PATH}.
628 The macro @code{STANDARD_EXEC_PREFIX}, if the compiler has been installed
629 in the configured-time @var{prefix}.
632 The location @file{/usr/libexec/gcc/}, but only if this is a native compiler.
635 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
638 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
642 Here is the order of prefixes tried for startfiles:
646 Any prefixes specified by the user with @option{-B}.
649 The environment variable @code{GCC_EXEC_PREFIX} or its automatically determined
650 value based on the installed toolchain location.
653 The directories specified by the environment variable @code{LIBRARY_PATH}
654 (or port-specific name; native only, cross compilers do not use this).
657 The macro @code{STANDARD_EXEC_PREFIX}, but only if the toolchain is installed
658 in the configured @var{prefix} or this is a native compiler.
661 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
664 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
668 The macro @code{MD_STARTFILE_PREFIX}, if defined, but only if this is a
669 native compiler, or we have a target system root.
672 The macro @code{MD_STARTFILE_PREFIX_1}, if defined, but only if this is a
673 native compiler, or we have a target system root.
676 The macro @code{STANDARD_STARTFILE_PREFIX}, with any sysroot modifications.
677 If this path is relative it will be prefixed by @code{GCC_EXEC_PREFIX} and
678 the machine suffix or @code{STANDARD_EXEC_PREFIX} and the machine suffix.
681 The macro @code{STANDARD_STARTFILE_PREFIX_1}, but only if this is a native
682 compiler, or we have a target system root. The default for this macro is
686 The macro @code{STANDARD_STARTFILE_PREFIX_2}, but only if this is a native
687 compiler, or we have a target system root. The default for this macro is
691 @node Run-time Target
692 @section Run-time Target Specification
693 @cindex run-time target specification
694 @cindex predefined macros
695 @cindex target specifications
697 @c prevent bad page break with this line
698 Here are run-time target specifications.
700 @defmac TARGET_CPU_CPP_BUILTINS ()
701 This function-like macro expands to a block of code that defines
702 built-in preprocessor macros and assertions for the target CPU, using
703 the functions @code{builtin_define}, @code{builtin_define_std} and
704 @code{builtin_assert}. When the front end
705 calls this macro it provides a trailing semicolon, and since it has
706 finished command line option processing your code can use those
709 @code{builtin_assert} takes a string in the form you pass to the
710 command-line option @option{-A}, such as @code{cpu=mips}, and creates
711 the assertion. @code{builtin_define} takes a string in the form
712 accepted by option @option{-D} and unconditionally defines the macro.
714 @code{builtin_define_std} takes a string representing the name of an
715 object-like macro. If it doesn't lie in the user's namespace,
716 @code{builtin_define_std} defines it unconditionally. Otherwise, it
717 defines a version with two leading underscores, and another version
718 with two leading and trailing underscores, and defines the original
719 only if an ISO standard was not requested on the command line. For
720 example, passing @code{unix} defines @code{__unix}, @code{__unix__}
721 and possibly @code{unix}; passing @code{_mips} defines @code{__mips},
722 @code{__mips__} and possibly @code{_mips}, and passing @code{_ABI64}
723 defines only @code{_ABI64}.
725 You can also test for the C dialect being compiled. The variable
726 @code{c_language} is set to one of @code{clk_c}, @code{clk_cplusplus}
727 or @code{clk_objective_c}. Note that if we are preprocessing
728 assembler, this variable will be @code{clk_c} but the function-like
729 macro @code{preprocessing_asm_p()} will return true, so you might want
730 to check for that first. If you need to check for strict ANSI, the
731 variable @code{flag_iso} can be used. The function-like macro
732 @code{preprocessing_trad_p()} can be used to check for traditional
736 @defmac TARGET_OS_CPP_BUILTINS ()
737 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
738 and is used for the target operating system instead.
741 @defmac TARGET_OBJFMT_CPP_BUILTINS ()
742 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
743 and is used for the target object format. @file{elfos.h} uses this
744 macro to define @code{__ELF__}, so you probably do not need to define
748 @deftypevar {extern int} target_flags
749 This variable is declared in @file{options.h}, which is included before
750 any target-specific headers.
753 @deftypevr {Target Hook} int TARGET_DEFAULT_TARGET_FLAGS
754 This variable specifies the initial value of @code{target_flags}.
755 Its default setting is 0.
758 @cindex optional hardware or system features
759 @cindex features, optional, in system conventions
761 @deftypefn {Target Hook} bool TARGET_HANDLE_OPTION (size_t @var{code}, const char *@var{arg}, int @var{value})
762 This hook is called whenever the user specifies one of the
763 target-specific options described by the @file{.opt} definition files
764 (@pxref{Options}). It has the opportunity to do some option-specific
765 processing and should return true if the option is valid. The default
766 definition does nothing but return true.
768 @var{code} specifies the @code{OPT_@var{name}} enumeration value
769 associated with the selected option; @var{name} is just a rendering of
770 the option name in which non-alphanumeric characters are replaced by
771 underscores. @var{arg} specifies the string argument and is null if
772 no argument was given. If the option is flagged as a @code{UInteger}
773 (@pxref{Option properties}), @var{value} is the numeric value of the
774 argument. Otherwise @var{value} is 1 if the positive form of the
775 option was used and 0 if the ``no-'' form was.
778 @deftypefn {Target Hook} bool TARGET_HANDLE_C_OPTION (size_t @var{code}, const char *@var{arg}, int @var{value})
779 This target hook is called whenever the user specifies one of the
780 target-specific C language family options described by the @file{.opt}
781 definition files(@pxref{Options}). It has the opportunity to do some
782 option-specific processing and should return true if the option is
783 valid. The default definition does nothing but return false.
785 In general, you should use @code{TARGET_HANDLE_OPTION} to handle
786 options. However, if processing an option requires routines that are
787 only available in the C (and related language) front ends, then you
788 should use @code{TARGET_HANDLE_C_OPTION} instead.
791 @defmac TARGET_VERSION
792 This macro is a C statement to print on @code{stderr} a string
793 describing the particular machine description choice. Every machine
794 description should define @code{TARGET_VERSION}. For example:
798 #define TARGET_VERSION \
799 fprintf (stderr, " (68k, Motorola syntax)");
801 #define TARGET_VERSION \
802 fprintf (stderr, " (68k, MIT syntax)");
807 @defmac OVERRIDE_OPTIONS
808 Sometimes certain combinations of command options do not make sense on
809 a particular target machine. You can define a macro
810 @code{OVERRIDE_OPTIONS} to take account of this. This macro, if
811 defined, is executed once just after all the command options have been
814 Don't use this macro to turn on various extra optimizations for
815 @option{-O}. That is what @code{OPTIMIZATION_OPTIONS} is for.
818 @defmac C_COMMON_OVERRIDE_OPTIONS
819 This is similar to @code{OVERRIDE_OPTIONS} but is only used in the C
820 language frontends (C, Objective-C, C++, Objective-C++) and so can be
821 used to alter option flag variables which only exist in those
825 @defmac OPTIMIZATION_OPTIONS (@var{level}, @var{size})
826 Some machines may desire to change what optimizations are performed for
827 various optimization levels. This macro, if defined, is executed once
828 just after the optimization level is determined and before the remainder
829 of the command options have been parsed. Values set in this macro are
830 used as the default values for the other command line options.
832 @var{level} is the optimization level specified; 2 if @option{-O2} is
833 specified, 1 if @option{-O} is specified, and 0 if neither is specified.
835 @var{size} is nonzero if @option{-Os} is specified and zero otherwise.
837 This macro is run once at program startup and when the optimization
838 options are changed via @code{#pragma GCC optimize} or by using the
839 @code{optimize} attribute.
841 @strong{Do not examine @code{write_symbols} in
842 this macro!} The debugging options are not supposed to alter the
846 @deftypefn {Target Hook} bool TARGET_HELP (void)
847 This hook is called in response to the user invoking
848 @option{--target-help} on the command line. It gives the target a
849 chance to display extra information on the target specific command
850 line options found in its @file{.opt} file.
853 @defmac CAN_DEBUG_WITHOUT_FP
854 Define this macro if debugging can be performed even without a frame
855 pointer. If this macro is defined, GCC will turn on the
856 @option{-fomit-frame-pointer} option whenever @option{-O} is specified.
859 @node Per-Function Data
860 @section Defining data structures for per-function information.
861 @cindex per-function data
862 @cindex data structures
864 If the target needs to store information on a per-function basis, GCC
865 provides a macro and a couple of variables to allow this. Note, just
866 using statics to store the information is a bad idea, since GCC supports
867 nested functions, so you can be halfway through encoding one function
868 when another one comes along.
870 GCC defines a data structure called @code{struct function} which
871 contains all of the data specific to an individual function. This
872 structure contains a field called @code{machine} whose type is
873 @code{struct machine_function *}, which can be used by targets to point
874 to their own specific data.
876 If a target needs per-function specific data it should define the type
877 @code{struct machine_function} and also the macro @code{INIT_EXPANDERS}.
878 This macro should be used to initialize the function pointer
879 @code{init_machine_status}. This pointer is explained below.
881 One typical use of per-function, target specific data is to create an
882 RTX to hold the register containing the function's return address. This
883 RTX can then be used to implement the @code{__builtin_return_address}
884 function, for level 0.
886 Note---earlier implementations of GCC used a single data area to hold
887 all of the per-function information. Thus when processing of a nested
888 function began the old per-function data had to be pushed onto a
889 stack, and when the processing was finished, it had to be popped off the
890 stack. GCC used to provide function pointers called
891 @code{save_machine_status} and @code{restore_machine_status} to handle
892 the saving and restoring of the target specific information. Since the
893 single data area approach is no longer used, these pointers are no
896 @defmac INIT_EXPANDERS
897 Macro called to initialize any target specific information. This macro
898 is called once per function, before generation of any RTL has begun.
899 The intention of this macro is to allow the initialization of the
900 function pointer @code{init_machine_status}.
903 @deftypevar {void (*)(struct function *)} init_machine_status
904 If this function pointer is non-@code{NULL} it will be called once per
905 function, before function compilation starts, in order to allow the
906 target to perform any target specific initialization of the
907 @code{struct function} structure. It is intended that this would be
908 used to initialize the @code{machine} of that structure.
910 @code{struct machine_function} structures are expected to be freed by GC@.
911 Generally, any memory that they reference must be allocated by using
912 @code{ggc_alloc}, including the structure itself.
916 @section Storage Layout
917 @cindex storage layout
919 Note that the definitions of the macros in this table which are sizes or
920 alignments measured in bits do not need to be constant. They can be C
921 expressions that refer to static variables, such as the @code{target_flags}.
922 @xref{Run-time Target}.
924 @defmac BITS_BIG_ENDIAN
925 Define this macro to have the value 1 if the most significant bit in a
926 byte has the lowest number; otherwise define it to have the value zero.
927 This means that bit-field instructions count from the most significant
928 bit. If the machine has no bit-field instructions, then this must still
929 be defined, but it doesn't matter which value it is defined to. This
930 macro need not be a constant.
932 This macro does not affect the way structure fields are packed into
933 bytes or words; that is controlled by @code{BYTES_BIG_ENDIAN}.
936 @defmac BYTES_BIG_ENDIAN
937 Define this macro to have the value 1 if the most significant byte in a
938 word has the lowest number. This macro need not be a constant.
941 @defmac WORDS_BIG_ENDIAN
942 Define this macro to have the value 1 if, in a multiword object, the
943 most significant word has the lowest number. This applies to both
944 memory locations and registers; GCC fundamentally assumes that the
945 order of words in memory is the same as the order in registers. This
946 macro need not be a constant.
949 @defmac LIBGCC2_WORDS_BIG_ENDIAN
950 Define this macro if @code{WORDS_BIG_ENDIAN} is not constant. This must be a
951 constant value with the same meaning as @code{WORDS_BIG_ENDIAN}, which will be
952 used only when compiling @file{libgcc2.c}. Typically the value will be set
953 based on preprocessor defines.
956 @defmac FLOAT_WORDS_BIG_ENDIAN
957 Define this macro to have the value 1 if @code{DFmode}, @code{XFmode} or
958 @code{TFmode} floating point numbers are stored in memory with the word
959 containing the sign bit at the lowest address; otherwise define it to
960 have the value 0. This macro need not be a constant.
962 You need not define this macro if the ordering is the same as for
966 @defmac BITS_PER_UNIT
967 Define this macro to be the number of bits in an addressable storage
968 unit (byte). If you do not define this macro the default is 8.
971 @defmac BITS_PER_WORD
972 Number of bits in a word. If you do not define this macro, the default
973 is @code{BITS_PER_UNIT * UNITS_PER_WORD}.
976 @defmac MAX_BITS_PER_WORD
977 Maximum number of bits in a word. If this is undefined, the default is
978 @code{BITS_PER_WORD}. Otherwise, it is the constant value that is the
979 largest value that @code{BITS_PER_WORD} can have at run-time.
982 @defmac UNITS_PER_WORD
983 Number of storage units in a word; normally the size of a general-purpose
984 register, a power of two from 1 or 8.
987 @defmac MIN_UNITS_PER_WORD
988 Minimum number of units in a word. If this is undefined, the default is
989 @code{UNITS_PER_WORD}. Otherwise, it is the constant value that is the
990 smallest value that @code{UNITS_PER_WORD} can have at run-time.
993 @defmac UNITS_PER_SIMD_WORD (@var{mode})
994 Number of units in the vectors that the vectorizer can produce for
995 scalar mode @var{mode}. The default is equal to @code{UNITS_PER_WORD},
996 because the vectorizer can do some transformations even in absence of
997 specialized @acronym{SIMD} hardware.
1000 @defmac POINTER_SIZE
1001 Width of a pointer, in bits. You must specify a value no wider than the
1002 width of @code{Pmode}. If it is not equal to the width of @code{Pmode},
1003 you must define @code{POINTERS_EXTEND_UNSIGNED}. If you do not specify
1004 a value the default is @code{BITS_PER_WORD}.
1007 @defmac POINTERS_EXTEND_UNSIGNED
1008 A C expression that determines how pointers should be extended from
1009 @code{ptr_mode} to either @code{Pmode} or @code{word_mode}. It is
1010 greater than zero if pointers should be zero-extended, zero if they
1011 should be sign-extended, and negative if some other sort of conversion
1012 is needed. In the last case, the extension is done by the target's
1013 @code{ptr_extend} instruction.
1015 You need not define this macro if the @code{ptr_mode}, @code{Pmode}
1016 and @code{word_mode} are all the same width.
1019 @defmac PROMOTE_MODE (@var{m}, @var{unsignedp}, @var{type})
1020 A macro to update @var{m} and @var{unsignedp} when an object whose type
1021 is @var{type} and which has the specified mode and signedness is to be
1022 stored in a register. This macro is only called when @var{type} is a
1025 On most RISC machines, which only have operations that operate on a full
1026 register, define this macro to set @var{m} to @code{word_mode} if
1027 @var{m} is an integer mode narrower than @code{BITS_PER_WORD}. In most
1028 cases, only integer modes should be widened because wider-precision
1029 floating-point operations are usually more expensive than their narrower
1032 For most machines, the macro definition does not change @var{unsignedp}.
1033 However, some machines, have instructions that preferentially handle
1034 either signed or unsigned quantities of certain modes. For example, on
1035 the DEC Alpha, 32-bit loads from memory and 32-bit add instructions
1036 sign-extend the result to 64 bits. On such machines, set
1037 @var{unsignedp} according to which kind of extension is more efficient.
1039 Do not define this macro if it would never modify @var{m}.
1042 @deftypefn {Target Hook} enum machine_mode TARGET_PROMOTE_FUNCTION_MODE (tree @var{type}, enum machine_mode @var{mode}, int *@var{punsignedp}, tree @var{funtype}, int @var{for_return})
1043 Like @code{PROMOTE_MODE}, but it is applied to outgoing function arguments or
1044 function return values. The target hook should return the new mode
1045 and possibly change @code{*@var{punsignedp}} if the promotion should
1046 change signedness. This function is called only for scalar @emph{or
1049 The default is to not promote arguments and return values. You can
1050 also define the hook to @code{default_promote_function_mode_always_promote}
1051 if you would like to apply the same rules given by @code{PROMOTE_MODE}.
1053 @var{for_return} allows to distinguish the promotion of arguments and
1054 return values. If this target hook promotes return values,
1055 @code{TARGET_FUNCTION_VALUE} must perform the same promotions done here.
1058 @defmac PARM_BOUNDARY
1059 Normal alignment required for function parameters on the stack, in
1060 bits. All stack parameters receive at least this much alignment
1061 regardless of data type. On most machines, this is the same as the
1065 @defmac STACK_BOUNDARY
1066 Define this macro to the minimum alignment enforced by hardware for the
1067 stack pointer on this machine. The definition is a C expression for the
1068 desired alignment (measured in bits). This value is used as a default
1069 if @code{PREFERRED_STACK_BOUNDARY} is not defined. On most machines,
1070 this should be the same as @code{PARM_BOUNDARY}.
1073 @defmac PREFERRED_STACK_BOUNDARY
1074 Define this macro if you wish to preserve a certain alignment for the
1075 stack pointer, greater than what the hardware enforces. The definition
1076 is a C expression for the desired alignment (measured in bits). This
1077 macro must evaluate to a value equal to or larger than
1078 @code{STACK_BOUNDARY}.
1081 @defmac INCOMING_STACK_BOUNDARY
1082 Define this macro if the incoming stack boundary may be different
1083 from @code{PREFERRED_STACK_BOUNDARY}. This macro must evaluate
1084 to a value equal to or larger than @code{STACK_BOUNDARY}.
1087 @defmac FUNCTION_BOUNDARY
1088 Alignment required for a function entry point, in bits.
1091 @defmac BIGGEST_ALIGNMENT
1092 Biggest alignment that any data type can require on this machine, in
1093 bits. Note that this is not the biggest alignment that is supported,
1094 just the biggest alignment that, when violated, may cause a fault.
1097 @defmac MALLOC_ABI_ALIGNMENT
1098 Alignment, in bits, a C conformant malloc implementation has to
1099 provide. If not defined, the default value is @code{BITS_PER_WORD}.
1102 @defmac ATTRIBUTE_ALIGNED_VALUE
1103 Alignment used by the @code{__attribute__ ((aligned))} construct. If
1104 not defined, the default value is @code{BIGGEST_ALIGNMENT}.
1107 @defmac MINIMUM_ATOMIC_ALIGNMENT
1108 If defined, the smallest alignment, in bits, that can be given to an
1109 object that can be referenced in one operation, without disturbing any
1110 nearby object. Normally, this is @code{BITS_PER_UNIT}, but may be larger
1111 on machines that don't have byte or half-word store operations.
1114 @defmac BIGGEST_FIELD_ALIGNMENT
1115 Biggest alignment that any structure or union field can require on this
1116 machine, in bits. If defined, this overrides @code{BIGGEST_ALIGNMENT} for
1117 structure and union fields only, unless the field alignment has been set
1118 by the @code{__attribute__ ((aligned (@var{n})))} construct.
1121 @defmac ADJUST_FIELD_ALIGN (@var{field}, @var{computed})
1122 An expression for the alignment of a structure field @var{field} if the
1123 alignment computed in the usual way (including applying of
1124 @code{BIGGEST_ALIGNMENT} and @code{BIGGEST_FIELD_ALIGNMENT} to the
1125 alignment) is @var{computed}. It overrides alignment only if the
1126 field alignment has not been set by the
1127 @code{__attribute__ ((aligned (@var{n})))} construct.
1130 @defmac MAX_STACK_ALIGNMENT
1131 Biggest stack alignment guaranteed by the backend. Use this macro
1132 to specify the maximum alignment of a variable on stack.
1134 If not defined, the default value is @code{STACK_BOUNDARY}.
1136 @c FIXME: The default should be @code{PREFERRED_STACK_BOUNDARY}.
1137 @c But the fix for PR 32893 indicates that we can only guarantee
1138 @c maximum stack alignment on stack up to @code{STACK_BOUNDARY}, not
1139 @c @code{PREFERRED_STACK_BOUNDARY}, if stack alignment isn't supported.
1142 @defmac MAX_OFILE_ALIGNMENT
1143 Biggest alignment supported by the object file format of this machine.
1144 Use this macro to limit the alignment which can be specified using the
1145 @code{__attribute__ ((aligned (@var{n})))} construct. If not defined,
1146 the default value is @code{BIGGEST_ALIGNMENT}.
1148 On systems that use ELF, the default (in @file{config/elfos.h}) is
1149 the largest supported 32-bit ELF section alignment representable on
1150 a 32-bit host e.g. @samp{(((unsigned HOST_WIDEST_INT) 1 << 28) * 8)}.
1151 On 32-bit ELF the largest supported section alignment in bits is
1152 @samp{(0x80000000 * 8)}, but this is not representable on 32-bit hosts.
1155 @defmac DATA_ALIGNMENT (@var{type}, @var{basic-align})
1156 If defined, a C expression to compute the alignment for a variable in
1157 the static store. @var{type} is the data type, and @var{basic-align} is
1158 the alignment that the object would ordinarily have. The value of this
1159 macro is used instead of that alignment to align the object.
1161 If this macro is not defined, then @var{basic-align} is used.
1164 One use of this macro is to increase alignment of medium-size data to
1165 make it all fit in fewer cache lines. Another is to cause character
1166 arrays to be word-aligned so that @code{strcpy} calls that copy
1167 constants to character arrays can be done inline.
1170 @defmac CONSTANT_ALIGNMENT (@var{constant}, @var{basic-align})
1171 If defined, a C expression to compute the alignment given to a constant
1172 that is being placed in memory. @var{constant} is the constant and
1173 @var{basic-align} is the alignment that the object would ordinarily
1174 have. The value of this macro is used instead of that alignment to
1177 If this macro is not defined, then @var{basic-align} is used.
1179 The typical use of this macro is to increase alignment for string
1180 constants to be word aligned so that @code{strcpy} calls that copy
1181 constants can be done inline.
1184 @defmac LOCAL_ALIGNMENT (@var{type}, @var{basic-align})
1185 If defined, a C expression to compute the alignment for a variable in
1186 the local store. @var{type} is the data type, and @var{basic-align} is
1187 the alignment that the object would ordinarily have. The value of this
1188 macro is used instead of that alignment to align the object.
1190 If this macro is not defined, then @var{basic-align} is used.
1192 One use of this macro is to increase alignment of medium-size data to
1193 make it all fit in fewer cache lines.
1196 @defmac STACK_SLOT_ALIGNMENT (@var{type}, @var{mode}, @var{basic-align})
1197 If defined, a C expression to compute the alignment for stack slot.
1198 @var{type} is the data type, @var{mode} is the widest mode available,
1199 and @var{basic-align} is the alignment that the slot would ordinarily
1200 have. The value of this macro is used instead of that alignment to
1203 If this macro is not defined, then @var{basic-align} is used when
1204 @var{type} is @code{NULL}. Otherwise, @code{LOCAL_ALIGNMENT} will
1207 This macro is to set alignment of stack slot to the maximum alignment
1208 of all possible modes which the slot may have.
1211 @defmac LOCAL_DECL_ALIGNMENT (@var{decl})
1212 If defined, a C expression to compute the alignment for a local
1213 variable @var{decl}.
1215 If this macro is not defined, then
1216 @code{LOCAL_ALIGNMENT (TREE_TYPE (@var{decl}), DECL_ALIGN (@var{decl}))}
1219 One use of this macro is to increase alignment of medium-size data to
1220 make it all fit in fewer cache lines.
1223 @defmac MINIMUM_ALIGNMENT (@var{exp}, @var{mode}, @var{align})
1224 If defined, a C expression to compute the minimum required alignment
1225 for dynamic stack realignment purposes for @var{exp} (a type or decl),
1226 @var{mode}, assuming normal alignment @var{align}.
1228 If this macro is not defined, then @var{align} will be used.
1231 @defmac EMPTY_FIELD_BOUNDARY
1232 Alignment in bits to be given to a structure bit-field that follows an
1233 empty field such as @code{int : 0;}.
1235 If @code{PCC_BITFIELD_TYPE_MATTERS} is true, it overrides this macro.
1238 @defmac STRUCTURE_SIZE_BOUNDARY
1239 Number of bits which any structure or union's size must be a multiple of.
1240 Each structure or union's size is rounded up to a multiple of this.
1242 If you do not define this macro, the default is the same as
1243 @code{BITS_PER_UNIT}.
1246 @defmac STRICT_ALIGNMENT
1247 Define this macro to be the value 1 if instructions will fail to work
1248 if given data not on the nominal alignment. If instructions will merely
1249 go slower in that case, define this macro as 0.
1252 @defmac PCC_BITFIELD_TYPE_MATTERS
1253 Define this if you wish to imitate the way many other C compilers handle
1254 alignment of bit-fields and the structures that contain them.
1256 The behavior is that the type written for a named bit-field (@code{int},
1257 @code{short}, or other integer type) imposes an alignment for the entire
1258 structure, as if the structure really did contain an ordinary field of
1259 that type. In addition, the bit-field is placed within the structure so
1260 that it would fit within such a field, not crossing a boundary for it.
1262 Thus, on most machines, a named bit-field whose type is written as
1263 @code{int} would not cross a four-byte boundary, and would force
1264 four-byte alignment for the whole structure. (The alignment used may
1265 not be four bytes; it is controlled by the other alignment parameters.)
1267 An unnamed bit-field will not affect the alignment of the containing
1270 If the macro is defined, its definition should be a C expression;
1271 a nonzero value for the expression enables this behavior.
1273 Note that if this macro is not defined, or its value is zero, some
1274 bit-fields may cross more than one alignment boundary. The compiler can
1275 support such references if there are @samp{insv}, @samp{extv}, and
1276 @samp{extzv} insns that can directly reference memory.
1278 The other known way of making bit-fields work is to define
1279 @code{STRUCTURE_SIZE_BOUNDARY} as large as @code{BIGGEST_ALIGNMENT}.
1280 Then every structure can be accessed with fullwords.
1282 Unless the machine has bit-field instructions or you define
1283 @code{STRUCTURE_SIZE_BOUNDARY} that way, you must define
1284 @code{PCC_BITFIELD_TYPE_MATTERS} to have a nonzero value.
1286 If your aim is to make GCC use the same conventions for laying out
1287 bit-fields as are used by another compiler, here is how to investigate
1288 what the other compiler does. Compile and run this program:
1307 printf ("Size of foo1 is %d\n",
1308 sizeof (struct foo1));
1309 printf ("Size of foo2 is %d\n",
1310 sizeof (struct foo2));
1315 If this prints 2 and 5, then the compiler's behavior is what you would
1316 get from @code{PCC_BITFIELD_TYPE_MATTERS}.
1319 @defmac BITFIELD_NBYTES_LIMITED
1320 Like @code{PCC_BITFIELD_TYPE_MATTERS} except that its effect is limited
1321 to aligning a bit-field within the structure.
1324 @deftypefn {Target Hook} bool TARGET_ALIGN_ANON_BITFIELD (void)
1325 When @code{PCC_BITFIELD_TYPE_MATTERS} is true this hook will determine
1326 whether unnamed bitfields affect the alignment of the containing
1327 structure. The hook should return true if the structure should inherit
1328 the alignment requirements of an unnamed bitfield's type.
1331 @deftypefn {Target Hook} bool TARGET_NARROW_VOLATILE_BITFIELD (void)
1332 This target hook should return @code{true} if accesses to volatile bitfields
1333 should use the narrowest mode possible. It should return @code{false} if
1334 these accesses should use the bitfield container type.
1336 The default is @code{!TARGET_STRICT_ALIGN}.
1339 @defmac MEMBER_TYPE_FORCES_BLK (@var{field}, @var{mode})
1340 Return 1 if a structure or array containing @var{field} should be accessed using
1343 If @var{field} is the only field in the structure, @var{mode} is its
1344 mode, otherwise @var{mode} is VOIDmode. @var{mode} is provided in the
1345 case where structures of one field would require the structure's mode to
1346 retain the field's mode.
1348 Normally, this is not needed.
1351 @defmac ROUND_TYPE_ALIGN (@var{type}, @var{computed}, @var{specified})
1352 Define this macro as an expression for the alignment of a type (given
1353 by @var{type} as a tree node) if the alignment computed in the usual
1354 way is @var{computed} and the alignment explicitly specified was
1357 The default is to use @var{specified} if it is larger; otherwise, use
1358 the smaller of @var{computed} and @code{BIGGEST_ALIGNMENT}
1361 @defmac MAX_FIXED_MODE_SIZE
1362 An integer expression for the size in bits of the largest integer
1363 machine mode that should actually be used. All integer machine modes of
1364 this size or smaller can be used for structures and unions with the
1365 appropriate sizes. If this macro is undefined, @code{GET_MODE_BITSIZE
1366 (DImode)} is assumed.
1369 @defmac STACK_SAVEAREA_MODE (@var{save_level})
1370 If defined, an expression of type @code{enum machine_mode} that
1371 specifies the mode of the save area operand of a
1372 @code{save_stack_@var{level}} named pattern (@pxref{Standard Names}).
1373 @var{save_level} is one of @code{SAVE_BLOCK}, @code{SAVE_FUNCTION}, or
1374 @code{SAVE_NONLOCAL} and selects which of the three named patterns is
1375 having its mode specified.
1377 You need not define this macro if it always returns @code{Pmode}. You
1378 would most commonly define this macro if the
1379 @code{save_stack_@var{level}} patterns need to support both a 32- and a
1383 @defmac STACK_SIZE_MODE
1384 If defined, an expression of type @code{enum machine_mode} that
1385 specifies the mode of the size increment operand of an
1386 @code{allocate_stack} named pattern (@pxref{Standard Names}).
1388 You need not define this macro if it always returns @code{word_mode}.
1389 You would most commonly define this macro if the @code{allocate_stack}
1390 pattern needs to support both a 32- and a 64-bit mode.
1393 @deftypefn {Target Hook} {enum machine_mode} TARGET_LIBGCC_CMP_RETURN_MODE ()
1394 This target hook should return the mode to be used for the return value
1395 of compare instructions expanded to libgcc calls. If not defined
1396 @code{word_mode} is returned which is the right choice for a majority of
1400 @deftypefn {Target Hook} {enum machine_mode} TARGET_LIBGCC_SHIFT_COUNT_MODE ()
1401 This target hook should return the mode to be used for the shift count operand
1402 of shift instructions expanded to libgcc calls. If not defined
1403 @code{word_mode} is returned which is the right choice for a majority of
1407 @defmac ROUND_TOWARDS_ZERO
1408 If defined, this macro should be true if the prevailing rounding
1409 mode is towards zero.
1411 Defining this macro only affects the way @file{libgcc.a} emulates
1412 floating-point arithmetic.
1414 Not defining this macro is equivalent to returning zero.
1417 @defmac LARGEST_EXPONENT_IS_NORMAL (@var{size})
1418 This macro should return true if floats with @var{size}
1419 bits do not have a NaN or infinity representation, but use the largest
1420 exponent for normal numbers instead.
1422 Defining this macro only affects the way @file{libgcc.a} emulates
1423 floating-point arithmetic.
1425 The default definition of this macro returns false for all sizes.
1428 @deftypefn {Target Hook} bool TARGET_MS_BITFIELD_LAYOUT_P (tree @var{record_type})
1429 This target hook returns @code{true} if bit-fields in the given
1430 @var{record_type} are to be laid out following the rules of Microsoft
1431 Visual C/C++, namely: (i) a bit-field won't share the same storage
1432 unit with the previous bit-field if their underlying types have
1433 different sizes, and the bit-field will be aligned to the highest
1434 alignment of the underlying types of itself and of the previous
1435 bit-field; (ii) a zero-sized bit-field will affect the alignment of
1436 the whole enclosing structure, even if it is unnamed; except that
1437 (iii) a zero-sized bit-field will be disregarded unless it follows
1438 another bit-field of nonzero size. If this hook returns @code{true},
1439 other macros that control bit-field layout are ignored.
1441 When a bit-field is inserted into a packed record, the whole size
1442 of the underlying type is used by one or more same-size adjacent
1443 bit-fields (that is, if its long:3, 32 bits is used in the record,
1444 and any additional adjacent long bit-fields are packed into the same
1445 chunk of 32 bits. However, if the size changes, a new field of that
1446 size is allocated). In an unpacked record, this is the same as using
1447 alignment, but not equivalent when packing.
1449 If both MS bit-fields and @samp{__attribute__((packed))} are used,
1450 the latter will take precedence. If @samp{__attribute__((packed))} is
1451 used on a single field when MS bit-fields are in use, it will take
1452 precedence for that field, but the alignment of the rest of the structure
1453 may affect its placement.
1456 @deftypefn {Target Hook} {bool} TARGET_DECIMAL_FLOAT_SUPPORTED_P (void)
1457 Returns true if the target supports decimal floating point.
1460 @deftypefn {Target Hook} {bool} TARGET_FIXED_POINT_SUPPORTED_P (void)
1461 Returns true if the target supports fixed-point arithmetic.
1464 @deftypefn {Target Hook} void TARGET_EXPAND_TO_RTL_HOOK (void)
1465 This hook is called just before expansion into rtl, allowing the target
1466 to perform additional initializations or analysis before the expansion.
1467 For example, the rs6000 port uses it to allocate a scratch stack slot
1468 for use in copying SDmode values between memory and floating point
1469 registers whenever the function being expanded has any SDmode
1473 @deftypefn {Target Hook} void TARGET_INSTANTIATE_DECLS (void)
1474 This hook allows the backend to perform additional instantiations on rtl
1475 that are not actually in any insns yet, but will be later.
1478 @deftypefn {Target Hook} {const char *} TARGET_MANGLE_TYPE (tree @var{type})
1479 If your target defines any fundamental types, or any types your target
1480 uses should be mangled differently from the default, define this hook
1481 to return the appropriate encoding for these types as part of a C++
1482 mangled name. The @var{type} argument is the tree structure representing
1483 the type to be mangled. The hook may be applied to trees which are
1484 not target-specific fundamental types; it should return @code{NULL}
1485 for all such types, as well as arguments it does not recognize. If the
1486 return value is not @code{NULL}, it must point to a statically-allocated
1489 Target-specific fundamental types might be new fundamental types or
1490 qualified versions of ordinary fundamental types. Encode new
1491 fundamental types as @samp{@w{u @var{n} @var{name}}}, where @var{name}
1492 is the name used for the type in source code, and @var{n} is the
1493 length of @var{name} in decimal. Encode qualified versions of
1494 ordinary types as @samp{@w{U @var{n} @var{name} @var{code}}}, where
1495 @var{name} is the name used for the type qualifier in source code,
1496 @var{n} is the length of @var{name} as above, and @var{code} is the
1497 code used to represent the unqualified version of this type. (See
1498 @code{write_builtin_type} in @file{cp/mangle.c} for the list of
1499 codes.) In both cases the spaces are for clarity; do not include any
1500 spaces in your string.
1502 This hook is applied to types prior to typedef resolution. If the mangled
1503 name for a particular type depends only on that type's main variant, you
1504 can perform typedef resolution yourself using @code{TYPE_MAIN_VARIANT}
1507 The default version of this hook always returns @code{NULL}, which is
1508 appropriate for a target that does not define any new fundamental
1513 @section Layout of Source Language Data Types
1515 These macros define the sizes and other characteristics of the standard
1516 basic data types used in programs being compiled. Unlike the macros in
1517 the previous section, these apply to specific features of C and related
1518 languages, rather than to fundamental aspects of storage layout.
1520 @defmac INT_TYPE_SIZE
1521 A C expression for the size in bits of the type @code{int} on the
1522 target machine. If you don't define this, the default is one word.
1525 @defmac SHORT_TYPE_SIZE
1526 A C expression for the size in bits of the type @code{short} on the
1527 target machine. If you don't define this, the default is half a word.
1528 (If this would be less than one storage unit, it is rounded up to one
1532 @defmac LONG_TYPE_SIZE
1533 A C expression for the size in bits of the type @code{long} on the
1534 target machine. If you don't define this, the default is one word.
1537 @defmac ADA_LONG_TYPE_SIZE
1538 On some machines, the size used for the Ada equivalent of the type
1539 @code{long} by a native Ada compiler differs from that used by C@. In
1540 that situation, define this macro to be a C expression to be used for
1541 the size of that type. If you don't define this, the default is the
1542 value of @code{LONG_TYPE_SIZE}.
1545 @defmac LONG_LONG_TYPE_SIZE
1546 A C expression for the size in bits of the type @code{long long} on the
1547 target machine. If you don't define this, the default is two
1548 words. If you want to support GNU Ada on your machine, the value of this
1549 macro must be at least 64.
1552 @defmac CHAR_TYPE_SIZE
1553 A C expression for the size in bits of the type @code{char} on the
1554 target machine. If you don't define this, the default is
1555 @code{BITS_PER_UNIT}.
1558 @defmac BOOL_TYPE_SIZE
1559 A C expression for the size in bits of the C++ type @code{bool} and
1560 C99 type @code{_Bool} on the target machine. If you don't define
1561 this, and you probably shouldn't, the default is @code{CHAR_TYPE_SIZE}.
1564 @defmac FLOAT_TYPE_SIZE
1565 A C expression for the size in bits of the type @code{float} on the
1566 target machine. If you don't define this, the default is one word.
1569 @defmac DOUBLE_TYPE_SIZE
1570 A C expression for the size in bits of the type @code{double} on the
1571 target machine. If you don't define this, the default is two
1575 @defmac LONG_DOUBLE_TYPE_SIZE
1576 A C expression for the size in bits of the type @code{long double} on
1577 the target machine. If you don't define this, the default is two
1581 @defmac SHORT_FRACT_TYPE_SIZE
1582 A C expression for the size in bits of the type @code{short _Fract} on
1583 the target machine. If you don't define this, the default is
1584 @code{BITS_PER_UNIT}.
1587 @defmac FRACT_TYPE_SIZE
1588 A C expression for the size in bits of the type @code{_Fract} on
1589 the target machine. If you don't define this, the default is
1590 @code{BITS_PER_UNIT * 2}.
1593 @defmac LONG_FRACT_TYPE_SIZE
1594 A C expression for the size in bits of the type @code{long _Fract} on
1595 the target machine. If you don't define this, the default is
1596 @code{BITS_PER_UNIT * 4}.
1599 @defmac LONG_LONG_FRACT_TYPE_SIZE
1600 A C expression for the size in bits of the type @code{long long _Fract} on
1601 the target machine. If you don't define this, the default is
1602 @code{BITS_PER_UNIT * 8}.
1605 @defmac SHORT_ACCUM_TYPE_SIZE
1606 A C expression for the size in bits of the type @code{short _Accum} on
1607 the target machine. If you don't define this, the default is
1608 @code{BITS_PER_UNIT * 2}.
1611 @defmac ACCUM_TYPE_SIZE
1612 A C expression for the size in bits of the type @code{_Accum} on
1613 the target machine. If you don't define this, the default is
1614 @code{BITS_PER_UNIT * 4}.
1617 @defmac LONG_ACCUM_TYPE_SIZE
1618 A C expression for the size in bits of the type @code{long _Accum} on
1619 the target machine. If you don't define this, the default is
1620 @code{BITS_PER_UNIT * 8}.
1623 @defmac LONG_LONG_ACCUM_TYPE_SIZE
1624 A C expression for the size in bits of the type @code{long long _Accum} on
1625 the target machine. If you don't define this, the default is
1626 @code{BITS_PER_UNIT * 16}.
1629 @defmac LIBGCC2_LONG_DOUBLE_TYPE_SIZE
1630 Define this macro if @code{LONG_DOUBLE_TYPE_SIZE} is not constant or
1631 if you want routines in @file{libgcc2.a} for a size other than
1632 @code{LONG_DOUBLE_TYPE_SIZE}. If you don't define this, the
1633 default is @code{LONG_DOUBLE_TYPE_SIZE}.
1636 @defmac LIBGCC2_HAS_DF_MODE
1637 Define this macro if neither @code{LIBGCC2_DOUBLE_TYPE_SIZE} nor
1638 @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is
1639 @code{DFmode} but you want @code{DFmode} routines in @file{libgcc2.a}
1640 anyway. If you don't define this and either @code{LIBGCC2_DOUBLE_TYPE_SIZE}
1641 or @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is 64 then the default is 1,
1645 @defmac LIBGCC2_HAS_XF_MODE
1646 Define this macro if @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is not
1647 @code{XFmode} but you want @code{XFmode} routines in @file{libgcc2.a}
1648 anyway. If you don't define this and @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE}
1649 is 80 then the default is 1, otherwise it is 0.
1652 @defmac LIBGCC2_HAS_TF_MODE
1653 Define this macro if @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is not
1654 @code{TFmode} but you want @code{TFmode} routines in @file{libgcc2.a}
1655 anyway. If you don't define this and @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE}
1656 is 128 then the default is 1, otherwise it is 0.
1663 Define these macros to be the size in bits of the mantissa of
1664 @code{SFmode}, @code{DFmode}, @code{XFmode} and @code{TFmode} values,
1665 if the defaults in @file{libgcc2.h} are inappropriate. By default,
1666 @code{FLT_MANT_DIG} is used for @code{SF_SIZE}, @code{LDBL_MANT_DIG}
1667 for @code{XF_SIZE} and @code{TF_SIZE}, and @code{DBL_MANT_DIG} or
1668 @code{LDBL_MANT_DIG} for @code{DF_SIZE} according to whether
1669 @code{LIBGCC2_DOUBLE_TYPE_SIZE} or
1670 @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is 64.
1673 @defmac TARGET_FLT_EVAL_METHOD
1674 A C expression for the value for @code{FLT_EVAL_METHOD} in @file{float.h},
1675 assuming, if applicable, that the floating-point control word is in its
1676 default state. If you do not define this macro the value of
1677 @code{FLT_EVAL_METHOD} will be zero.
1680 @defmac WIDEST_HARDWARE_FP_SIZE
1681 A C expression for the size in bits of the widest floating-point format
1682 supported by the hardware. If you define this macro, you must specify a
1683 value less than or equal to the value of @code{LONG_DOUBLE_TYPE_SIZE}.
1684 If you do not define this macro, the value of @code{LONG_DOUBLE_TYPE_SIZE}
1688 @defmac DEFAULT_SIGNED_CHAR
1689 An expression whose value is 1 or 0, according to whether the type
1690 @code{char} should be signed or unsigned by default. The user can
1691 always override this default with the options @option{-fsigned-char}
1692 and @option{-funsigned-char}.
1695 @deftypefn {Target Hook} bool TARGET_DEFAULT_SHORT_ENUMS (void)
1696 This target hook should return true if the compiler should give an
1697 @code{enum} type only as many bytes as it takes to represent the range
1698 of possible values of that type. It should return false if all
1699 @code{enum} types should be allocated like @code{int}.
1701 The default is to return false.
1705 A C expression for a string describing the name of the data type to use
1706 for size values. The typedef name @code{size_t} is defined using the
1707 contents of the string.
1709 The string can contain more than one keyword. If so, separate them with
1710 spaces, and write first any length keyword, then @code{unsigned} if
1711 appropriate, and finally @code{int}. The string must exactly match one
1712 of the data type names defined in the function
1713 @code{init_decl_processing} in the file @file{c-decl.c}. You may not
1714 omit @code{int} or change the order---that would cause the compiler to
1717 If you don't define this macro, the default is @code{"long unsigned
1721 @defmac PTRDIFF_TYPE
1722 A C expression for a string describing the name of the data type to use
1723 for the result of subtracting two pointers. The typedef name
1724 @code{ptrdiff_t} is defined using the contents of the string. See
1725 @code{SIZE_TYPE} above for more information.
1727 If you don't define this macro, the default is @code{"long int"}.
1731 A C expression for a string describing the name of the data type to use
1732 for wide characters. The typedef name @code{wchar_t} is defined using
1733 the contents of the string. See @code{SIZE_TYPE} above for more
1736 If you don't define this macro, the default is @code{"int"}.
1739 @defmac WCHAR_TYPE_SIZE
1740 A C expression for the size in bits of the data type for wide
1741 characters. This is used in @code{cpp}, which cannot make use of
1746 A C expression for a string describing the name of the data type to
1747 use for wide characters passed to @code{printf} and returned from
1748 @code{getwc}. The typedef name @code{wint_t} is defined using the
1749 contents of the string. See @code{SIZE_TYPE} above for more
1752 If you don't define this macro, the default is @code{"unsigned int"}.
1756 A C expression for a string describing the name of the data type that
1757 can represent any value of any standard or extended signed integer type.
1758 The typedef name @code{intmax_t} is defined using the contents of the
1759 string. See @code{SIZE_TYPE} above for more information.
1761 If you don't define this macro, the default is the first of
1762 @code{"int"}, @code{"long int"}, or @code{"long long int"} that has as
1763 much precision as @code{long long int}.
1766 @defmac UINTMAX_TYPE
1767 A C expression for a string describing the name of the data type that
1768 can represent any value of any standard or extended unsigned integer
1769 type. The typedef name @code{uintmax_t} is defined using the contents
1770 of the string. See @code{SIZE_TYPE} above for more information.
1772 If you don't define this macro, the default is the first of
1773 @code{"unsigned int"}, @code{"long unsigned int"}, or @code{"long long
1774 unsigned int"} that has as much precision as @code{long long unsigned
1778 @defmac SIG_ATOMIC_TYPE
1784 @defmacx UINT16_TYPE
1785 @defmacx UINT32_TYPE
1786 @defmacx UINT64_TYPE
1787 @defmacx INT_LEAST8_TYPE
1788 @defmacx INT_LEAST16_TYPE
1789 @defmacx INT_LEAST32_TYPE
1790 @defmacx INT_LEAST64_TYPE
1791 @defmacx UINT_LEAST8_TYPE
1792 @defmacx UINT_LEAST16_TYPE
1793 @defmacx UINT_LEAST32_TYPE
1794 @defmacx UINT_LEAST64_TYPE
1795 @defmacx INT_FAST8_TYPE
1796 @defmacx INT_FAST16_TYPE
1797 @defmacx INT_FAST32_TYPE
1798 @defmacx INT_FAST64_TYPE
1799 @defmacx UINT_FAST8_TYPE
1800 @defmacx UINT_FAST16_TYPE
1801 @defmacx UINT_FAST32_TYPE
1802 @defmacx UINT_FAST64_TYPE
1803 @defmacx INTPTR_TYPE
1804 @defmacx UINTPTR_TYPE
1805 C expressions for the standard types @code{sig_atomic_t},
1806 @code{int8_t}, @code{int16_t}, @code{int32_t}, @code{int64_t},
1807 @code{uint8_t}, @code{uint16_t}, @code{uint32_t}, @code{uint64_t},
1808 @code{int_least8_t}, @code{int_least16_t}, @code{int_least32_t},
1809 @code{int_least64_t}, @code{uint_least8_t}, @code{uint_least16_t},
1810 @code{uint_least32_t}, @code{uint_least64_t}, @code{int_fast8_t},
1811 @code{int_fast16_t}, @code{int_fast32_t}, @code{int_fast64_t},
1812 @code{uint_fast8_t}, @code{uint_fast16_t}, @code{uint_fast32_t},
1813 @code{uint_fast64_t}, @code{intptr_t}, and @code{uintptr_t}. See
1814 @code{SIZE_TYPE} above for more information.
1816 If any of these macros evaluates to a null pointer, the corresponding
1817 type is not supported; if GCC is configured to provide
1818 @code{<stdint.h>} in such a case, the header provided may not conform
1819 to C99, depending on the type in question. The defaults for all of
1820 these macros are null pointers.
1823 @defmac TARGET_PTRMEMFUNC_VBIT_LOCATION
1824 The C++ compiler represents a pointer-to-member-function with a struct
1831 ptrdiff_t vtable_index;
1838 The C++ compiler must use one bit to indicate whether the function that
1839 will be called through a pointer-to-member-function is virtual.
1840 Normally, we assume that the low-order bit of a function pointer must
1841 always be zero. Then, by ensuring that the vtable_index is odd, we can
1842 distinguish which variant of the union is in use. But, on some
1843 platforms function pointers can be odd, and so this doesn't work. In
1844 that case, we use the low-order bit of the @code{delta} field, and shift
1845 the remainder of the @code{delta} field to the left.
1847 GCC will automatically make the right selection about where to store
1848 this bit using the @code{FUNCTION_BOUNDARY} setting for your platform.
1849 However, some platforms such as ARM/Thumb have @code{FUNCTION_BOUNDARY}
1850 set such that functions always start at even addresses, but the lowest
1851 bit of pointers to functions indicate whether the function at that
1852 address is in ARM or Thumb mode. If this is the case of your
1853 architecture, you should define this macro to
1854 @code{ptrmemfunc_vbit_in_delta}.
1856 In general, you should not have to define this macro. On architectures
1857 in which function addresses are always even, according to
1858 @code{FUNCTION_BOUNDARY}, GCC will automatically define this macro to
1859 @code{ptrmemfunc_vbit_in_pfn}.
1862 @defmac TARGET_VTABLE_USES_DESCRIPTORS
1863 Normally, the C++ compiler uses function pointers in vtables. This
1864 macro allows the target to change to use ``function descriptors''
1865 instead. Function descriptors are found on targets for whom a
1866 function pointer is actually a small data structure. Normally the
1867 data structure consists of the actual code address plus a data
1868 pointer to which the function's data is relative.
1870 If vtables are used, the value of this macro should be the number
1871 of words that the function descriptor occupies.
1874 @defmac TARGET_VTABLE_ENTRY_ALIGN
1875 By default, the vtable entries are void pointers, the so the alignment
1876 is the same as pointer alignment. The value of this macro specifies
1877 the alignment of the vtable entry in bits. It should be defined only
1878 when special alignment is necessary. */
1881 @defmac TARGET_VTABLE_DATA_ENTRY_DISTANCE
1882 There are a few non-descriptor entries in the vtable at offsets below
1883 zero. If these entries must be padded (say, to preserve the alignment
1884 specified by @code{TARGET_VTABLE_ENTRY_ALIGN}), set this to the number
1885 of words in each data entry.
1889 @section Register Usage
1890 @cindex register usage
1892 This section explains how to describe what registers the target machine
1893 has, and how (in general) they can be used.
1895 The description of which registers a specific instruction can use is
1896 done with register classes; see @ref{Register Classes}. For information
1897 on using registers to access a stack frame, see @ref{Frame Registers}.
1898 For passing values in registers, see @ref{Register Arguments}.
1899 For returning values in registers, see @ref{Scalar Return}.
1902 * Register Basics:: Number and kinds of registers.
1903 * Allocation Order:: Order in which registers are allocated.
1904 * Values in Registers:: What kinds of values each reg can hold.
1905 * Leaf Functions:: Renumbering registers for leaf functions.
1906 * Stack Registers:: Handling a register stack such as 80387.
1909 @node Register Basics
1910 @subsection Basic Characteristics of Registers
1912 @c prevent bad page break with this line
1913 Registers have various characteristics.
1915 @defmac FIRST_PSEUDO_REGISTER
1916 Number of hardware registers known to the compiler. They receive
1917 numbers 0 through @code{FIRST_PSEUDO_REGISTER-1}; thus, the first
1918 pseudo register's number really is assigned the number
1919 @code{FIRST_PSEUDO_REGISTER}.
1922 @defmac FIXED_REGISTERS
1923 @cindex fixed register
1924 An initializer that says which registers are used for fixed purposes
1925 all throughout the compiled code and are therefore not available for
1926 general allocation. These would include the stack pointer, the frame
1927 pointer (except on machines where that can be used as a general
1928 register when no frame pointer is needed), the program counter on
1929 machines where that is considered one of the addressable registers,
1930 and any other numbered register with a standard use.
1932 This information is expressed as a sequence of numbers, separated by
1933 commas and surrounded by braces. The @var{n}th number is 1 if
1934 register @var{n} is fixed, 0 otherwise.
1936 The table initialized from this macro, and the table initialized by
1937 the following one, may be overridden at run time either automatically,
1938 by the actions of the macro @code{CONDITIONAL_REGISTER_USAGE}, or by
1939 the user with the command options @option{-ffixed-@var{reg}},
1940 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}.
1943 @defmac CALL_USED_REGISTERS
1944 @cindex call-used register
1945 @cindex call-clobbered register
1946 @cindex call-saved register
1947 Like @code{FIXED_REGISTERS} but has 1 for each register that is
1948 clobbered (in general) by function calls as well as for fixed
1949 registers. This macro therefore identifies the registers that are not
1950 available for general allocation of values that must live across
1953 If a register has 0 in @code{CALL_USED_REGISTERS}, the compiler
1954 automatically saves it on function entry and restores it on function
1955 exit, if the register is used within the function.
1958 @defmac CALL_REALLY_USED_REGISTERS
1959 @cindex call-used register
1960 @cindex call-clobbered register
1961 @cindex call-saved register
1962 Like @code{CALL_USED_REGISTERS} except this macro doesn't require
1963 that the entire set of @code{FIXED_REGISTERS} be included.
1964 (@code{CALL_USED_REGISTERS} must be a superset of @code{FIXED_REGISTERS}).
1965 This macro is optional. If not specified, it defaults to the value
1966 of @code{CALL_USED_REGISTERS}.
1969 @defmac HARD_REGNO_CALL_PART_CLOBBERED (@var{regno}, @var{mode})
1970 @cindex call-used register
1971 @cindex call-clobbered register
1972 @cindex call-saved register
1973 A C expression that is nonzero if it is not permissible to store a
1974 value of mode @var{mode} in hard register number @var{regno} across a
1975 call without some part of it being clobbered. For most machines this
1976 macro need not be defined. It is only required for machines that do not
1977 preserve the entire contents of a register across a call.
1981 @findex call_used_regs
1984 @findex reg_class_contents
1985 @defmac CONDITIONAL_REGISTER_USAGE
1986 Zero or more C statements that may conditionally modify five variables
1987 @code{fixed_regs}, @code{call_used_regs}, @code{global_regs},
1988 @code{reg_names}, and @code{reg_class_contents}, to take into account
1989 any dependence of these register sets on target flags. The first three
1990 of these are of type @code{char []} (interpreted as Boolean vectors).
1991 @code{global_regs} is a @code{const char *[]}, and
1992 @code{reg_class_contents} is a @code{HARD_REG_SET}. Before the macro is
1993 called, @code{fixed_regs}, @code{call_used_regs},
1994 @code{reg_class_contents}, and @code{reg_names} have been initialized
1995 from @code{FIXED_REGISTERS}, @code{CALL_USED_REGISTERS},
1996 @code{REG_CLASS_CONTENTS}, and @code{REGISTER_NAMES}, respectively.
1997 @code{global_regs} has been cleared, and any @option{-ffixed-@var{reg}},
1998 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}
1999 command options have been applied.
2001 You need not define this macro if it has no work to do.
2003 @cindex disabling certain registers
2004 @cindex controlling register usage
2005 If the usage of an entire class of registers depends on the target
2006 flags, you may indicate this to GCC by using this macro to modify
2007 @code{fixed_regs} and @code{call_used_regs} to 1 for each of the
2008 registers in the classes which should not be used by GCC@. Also define
2009 the macro @code{REG_CLASS_FROM_LETTER} / @code{REG_CLASS_FROM_CONSTRAINT}
2010 to return @code{NO_REGS} if it
2011 is called with a letter for a class that shouldn't be used.
2013 (However, if this class is not included in @code{GENERAL_REGS} and all
2014 of the insn patterns whose constraints permit this class are
2015 controlled by target switches, then GCC will automatically avoid using
2016 these registers when the target switches are opposed to them.)
2019 @defmac INCOMING_REGNO (@var{out})
2020 Define this macro if the target machine has register windows. This C
2021 expression returns the register number as seen by the called function
2022 corresponding to the register number @var{out} as seen by the calling
2023 function. Return @var{out} if register number @var{out} is not an
2027 @defmac OUTGOING_REGNO (@var{in})
2028 Define this macro if the target machine has register windows. This C
2029 expression returns the register number as seen by the calling function
2030 corresponding to the register number @var{in} as seen by the called
2031 function. Return @var{in} if register number @var{in} is not an inbound
2035 @defmac LOCAL_REGNO (@var{regno})
2036 Define this macro if the target machine has register windows. This C
2037 expression returns true if the register is call-saved but is in the
2038 register window. Unlike most call-saved registers, such registers
2039 need not be explicitly restored on function exit or during non-local
2044 If the program counter has a register number, define this as that
2045 register number. Otherwise, do not define it.
2048 @node Allocation Order
2049 @subsection Order of Allocation of Registers
2050 @cindex order of register allocation
2051 @cindex register allocation order
2053 @c prevent bad page break with this line
2054 Registers are allocated in order.
2056 @defmac REG_ALLOC_ORDER
2057 If defined, an initializer for a vector of integers, containing the
2058 numbers of hard registers in the order in which GCC should prefer
2059 to use them (from most preferred to least).
2061 If this macro is not defined, registers are used lowest numbered first
2062 (all else being equal).
2064 One use of this macro is on machines where the highest numbered
2065 registers must always be saved and the save-multiple-registers
2066 instruction supports only sequences of consecutive registers. On such
2067 machines, define @code{REG_ALLOC_ORDER} to be an initializer that lists
2068 the highest numbered allocable register first.
2071 @defmac ORDER_REGS_FOR_LOCAL_ALLOC
2072 A C statement (sans semicolon) to choose the order in which to allocate
2073 hard registers for pseudo-registers local to a basic block.
2075 Store the desired register order in the array @code{reg_alloc_order}.
2076 Element 0 should be the register to allocate first; element 1, the next
2077 register; and so on.
2079 The macro body should not assume anything about the contents of
2080 @code{reg_alloc_order} before execution of the macro.
2082 On most machines, it is not necessary to define this macro.
2085 @defmac IRA_HARD_REGNO_ADD_COST_MULTIPLIER (@var{regno})
2086 In some case register allocation order is not enough for the
2087 Integrated Register Allocator (@acronym{IRA}) to generate a good code.
2088 If this macro is defined, it should return a floating point value
2089 based on @var{regno}. The cost of using @var{regno} for a pseudo will
2090 be increased by approximately the pseudo's usage frequency times the
2091 value returned by this macro. Not defining this macro is equivalent
2092 to having it always return @code{0.0}.
2094 On most machines, it is not necessary to define this macro.
2097 @node Values in Registers
2098 @subsection How Values Fit in Registers
2100 This section discusses the macros that describe which kinds of values
2101 (specifically, which machine modes) each register can hold, and how many
2102 consecutive registers are needed for a given mode.
2104 @defmac HARD_REGNO_NREGS (@var{regno}, @var{mode})
2105 A C expression for the number of consecutive hard registers, starting
2106 at register number @var{regno}, required to hold a value of mode
2107 @var{mode}. This macro must never return zero, even if a register
2108 cannot hold the requested mode - indicate that with HARD_REGNO_MODE_OK
2109 and/or CANNOT_CHANGE_MODE_CLASS instead.
2111 On a machine where all registers are exactly one word, a suitable
2112 definition of this macro is
2115 #define HARD_REGNO_NREGS(REGNO, MODE) \
2116 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) \
2121 @defmac HARD_REGNO_NREGS_HAS_PADDING (@var{regno}, @var{mode})
2122 A C expression that is nonzero if a value of mode @var{mode}, stored
2123 in memory, ends with padding that causes it to take up more space than
2124 in registers starting at register number @var{regno} (as determined by
2125 multiplying GCC's notion of the size of the register when containing
2126 this mode by the number of registers returned by
2127 @code{HARD_REGNO_NREGS}). By default this is zero.
2129 For example, if a floating-point value is stored in three 32-bit
2130 registers but takes up 128 bits in memory, then this would be
2133 This macros only needs to be defined if there are cases where
2134 @code{subreg_get_info}
2135 would otherwise wrongly determine that a @code{subreg} can be
2136 represented by an offset to the register number, when in fact such a
2137 @code{subreg} would contain some of the padding not stored in
2138 registers and so not be representable.
2141 @defmac HARD_REGNO_NREGS_WITH_PADDING (@var{regno}, @var{mode})
2142 For values of @var{regno} and @var{mode} for which
2143 @code{HARD_REGNO_NREGS_HAS_PADDING} returns nonzero, a C expression
2144 returning the greater number of registers required to hold the value
2145 including any padding. In the example above, the value would be four.
2148 @defmac REGMODE_NATURAL_SIZE (@var{mode})
2149 Define this macro if the natural size of registers that hold values
2150 of mode @var{mode} is not the word size. It is a C expression that
2151 should give the natural size in bytes for the specified mode. It is
2152 used by the register allocator to try to optimize its results. This
2153 happens for example on SPARC 64-bit where the natural size of
2154 floating-point registers is still 32-bit.
2157 @defmac HARD_REGNO_MODE_OK (@var{regno}, @var{mode})
2158 A C expression that is nonzero if it is permissible to store a value
2159 of mode @var{mode} in hard register number @var{regno} (or in several
2160 registers starting with that one). For a machine where all registers
2161 are equivalent, a suitable definition is
2164 #define HARD_REGNO_MODE_OK(REGNO, MODE) 1
2167 You need not include code to check for the numbers of fixed registers,
2168 because the allocation mechanism considers them to be always occupied.
2170 @cindex register pairs
2171 On some machines, double-precision values must be kept in even/odd
2172 register pairs. You can implement that by defining this macro to reject
2173 odd register numbers for such modes.
2175 The minimum requirement for a mode to be OK in a register is that the
2176 @samp{mov@var{mode}} instruction pattern support moves between the
2177 register and other hard register in the same class and that moving a
2178 value into the register and back out not alter it.
2180 Since the same instruction used to move @code{word_mode} will work for
2181 all narrower integer modes, it is not necessary on any machine for
2182 @code{HARD_REGNO_MODE_OK} to distinguish between these modes, provided
2183 you define patterns @samp{movhi}, etc., to take advantage of this. This
2184 is useful because of the interaction between @code{HARD_REGNO_MODE_OK}
2185 and @code{MODES_TIEABLE_P}; it is very desirable for all integer modes
2188 Many machines have special registers for floating point arithmetic.
2189 Often people assume that floating point machine modes are allowed only
2190 in floating point registers. This is not true. Any registers that
2191 can hold integers can safely @emph{hold} a floating point machine
2192 mode, whether or not floating arithmetic can be done on it in those
2193 registers. Integer move instructions can be used to move the values.
2195 On some machines, though, the converse is true: fixed-point machine
2196 modes may not go in floating registers. This is true if the floating
2197 registers normalize any value stored in them, because storing a
2198 non-floating value there would garble it. In this case,
2199 @code{HARD_REGNO_MODE_OK} should reject fixed-point machine modes in
2200 floating registers. But if the floating registers do not automatically
2201 normalize, if you can store any bit pattern in one and retrieve it
2202 unchanged without a trap, then any machine mode may go in a floating
2203 register, so you can define this macro to say so.
2205 The primary significance of special floating registers is rather that
2206 they are the registers acceptable in floating point arithmetic
2207 instructions. However, this is of no concern to
2208 @code{HARD_REGNO_MODE_OK}. You handle it by writing the proper
2209 constraints for those instructions.
2211 On some machines, the floating registers are especially slow to access,
2212 so that it is better to store a value in a stack frame than in such a
2213 register if floating point arithmetic is not being done. As long as the
2214 floating registers are not in class @code{GENERAL_REGS}, they will not
2215 be used unless some pattern's constraint asks for one.
2218 @defmac HARD_REGNO_RENAME_OK (@var{from}, @var{to})
2219 A C expression that is nonzero if it is OK to rename a hard register
2220 @var{from} to another hard register @var{to}.
2222 One common use of this macro is to prevent renaming of a register to
2223 another register that is not saved by a prologue in an interrupt
2226 The default is always nonzero.
2229 @defmac MODES_TIEABLE_P (@var{mode1}, @var{mode2})
2230 A C expression that is nonzero if a value of mode
2231 @var{mode1} is accessible in mode @var{mode2} without copying.
2233 If @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode1})} and
2234 @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode2})} are always the same for
2235 any @var{r}, then @code{MODES_TIEABLE_P (@var{mode1}, @var{mode2})}
2236 should be nonzero. If they differ for any @var{r}, you should define
2237 this macro to return zero unless some other mechanism ensures the
2238 accessibility of the value in a narrower mode.
2240 You should define this macro to return nonzero in as many cases as
2241 possible since doing so will allow GCC to perform better register
2245 @deftypefn {Target Hook} bool TARGET_HARD_REGNO_SCRATCH_OK (unsigned int @var{regno})
2246 This target hook should return @code{true} if it is OK to use a hard register
2247 @var{regno} as scratch reg in peephole2.
2249 One common use of this macro is to prevent using of a register that
2250 is not saved by a prologue in an interrupt handler.
2252 The default version of this hook always returns @code{true}.
2255 @defmac AVOID_CCMODE_COPIES
2256 Define this macro if the compiler should avoid copies to/from @code{CCmode}
2257 registers. You should only define this macro if support for copying to/from
2258 @code{CCmode} is incomplete.
2261 @node Leaf Functions
2262 @subsection Handling Leaf Functions
2264 @cindex leaf functions
2265 @cindex functions, leaf
2266 On some machines, a leaf function (i.e., one which makes no calls) can run
2267 more efficiently if it does not make its own register window. Often this
2268 means it is required to receive its arguments in the registers where they
2269 are passed by the caller, instead of the registers where they would
2272 The special treatment for leaf functions generally applies only when
2273 other conditions are met; for example, often they may use only those
2274 registers for its own variables and temporaries. We use the term ``leaf
2275 function'' to mean a function that is suitable for this special
2276 handling, so that functions with no calls are not necessarily ``leaf
2279 GCC assigns register numbers before it knows whether the function is
2280 suitable for leaf function treatment. So it needs to renumber the
2281 registers in order to output a leaf function. The following macros
2284 @defmac LEAF_REGISTERS
2285 Name of a char vector, indexed by hard register number, which
2286 contains 1 for a register that is allowable in a candidate for leaf
2289 If leaf function treatment involves renumbering the registers, then the
2290 registers marked here should be the ones before renumbering---those that
2291 GCC would ordinarily allocate. The registers which will actually be
2292 used in the assembler code, after renumbering, should not be marked with 1
2295 Define this macro only if the target machine offers a way to optimize
2296 the treatment of leaf functions.
2299 @defmac LEAF_REG_REMAP (@var{regno})
2300 A C expression whose value is the register number to which @var{regno}
2301 should be renumbered, when a function is treated as a leaf function.
2303 If @var{regno} is a register number which should not appear in a leaf
2304 function before renumbering, then the expression should yield @minus{}1, which
2305 will cause the compiler to abort.
2307 Define this macro only if the target machine offers a way to optimize the
2308 treatment of leaf functions, and registers need to be renumbered to do
2312 @findex current_function_is_leaf
2313 @findex current_function_uses_only_leaf_regs
2314 @code{TARGET_ASM_FUNCTION_PROLOGUE} and
2315 @code{TARGET_ASM_FUNCTION_EPILOGUE} must usually treat leaf functions
2316 specially. They can test the C variable @code{current_function_is_leaf}
2317 which is nonzero for leaf functions. @code{current_function_is_leaf} is
2318 set prior to local register allocation and is valid for the remaining
2319 compiler passes. They can also test the C variable
2320 @code{current_function_uses_only_leaf_regs} which is nonzero for leaf
2321 functions which only use leaf registers.
2322 @code{current_function_uses_only_leaf_regs} is valid after all passes
2323 that modify the instructions have been run and is only useful if
2324 @code{LEAF_REGISTERS} is defined.
2325 @c changed this to fix overfull. ALSO: why the "it" at the beginning
2326 @c of the next paragraph?! --mew 2feb93
2328 @node Stack Registers
2329 @subsection Registers That Form a Stack
2331 There are special features to handle computers where some of the
2332 ``registers'' form a stack. Stack registers are normally written by
2333 pushing onto the stack, and are numbered relative to the top of the
2336 Currently, GCC can only handle one group of stack-like registers, and
2337 they must be consecutively numbered. Furthermore, the existing
2338 support for stack-like registers is specific to the 80387 floating
2339 point coprocessor. If you have a new architecture that uses
2340 stack-like registers, you will need to do substantial work on
2341 @file{reg-stack.c} and write your machine description to cooperate
2342 with it, as well as defining these macros.
2345 Define this if the machine has any stack-like registers.
2348 @defmac FIRST_STACK_REG
2349 The number of the first stack-like register. This one is the top
2353 @defmac LAST_STACK_REG
2354 The number of the last stack-like register. This one is the bottom of
2358 @node Register Classes
2359 @section Register Classes
2360 @cindex register class definitions
2361 @cindex class definitions, register
2363 On many machines, the numbered registers are not all equivalent.
2364 For example, certain registers may not be allowed for indexed addressing;
2365 certain registers may not be allowed in some instructions. These machine
2366 restrictions are described to the compiler using @dfn{register classes}.
2368 You define a number of register classes, giving each one a name and saying
2369 which of the registers belong to it. Then you can specify register classes
2370 that are allowed as operands to particular instruction patterns.
2374 In general, each register will belong to several classes. In fact, one
2375 class must be named @code{ALL_REGS} and contain all the registers. Another
2376 class must be named @code{NO_REGS} and contain no registers. Often the
2377 union of two classes will be another class; however, this is not required.
2379 @findex GENERAL_REGS
2380 One of the classes must be named @code{GENERAL_REGS}. There is nothing
2381 terribly special about the name, but the operand constraint letters
2382 @samp{r} and @samp{g} specify this class. If @code{GENERAL_REGS} is
2383 the same as @code{ALL_REGS}, just define it as a macro which expands
2386 Order the classes so that if class @var{x} is contained in class @var{y}
2387 then @var{x} has a lower class number than @var{y}.
2389 The way classes other than @code{GENERAL_REGS} are specified in operand
2390 constraints is through machine-dependent operand constraint letters.
2391 You can define such letters to correspond to various classes, then use
2392 them in operand constraints.
2394 You should define a class for the union of two classes whenever some
2395 instruction allows both classes. For example, if an instruction allows
2396 either a floating point (coprocessor) register or a general register for a
2397 certain operand, you should define a class @code{FLOAT_OR_GENERAL_REGS}
2398 which includes both of them. Otherwise you will get suboptimal code.
2400 You must also specify certain redundant information about the register
2401 classes: for each class, which classes contain it and which ones are
2402 contained in it; for each pair of classes, the largest class contained
2405 When a value occupying several consecutive registers is expected in a
2406 certain class, all the registers used must belong to that class.
2407 Therefore, register classes cannot be used to enforce a requirement for
2408 a register pair to start with an even-numbered register. The way to
2409 specify this requirement is with @code{HARD_REGNO_MODE_OK}.
2411 Register classes used for input-operands of bitwise-and or shift
2412 instructions have a special requirement: each such class must have, for
2413 each fixed-point machine mode, a subclass whose registers can transfer that
2414 mode to or from memory. For example, on some machines, the operations for
2415 single-byte values (@code{QImode}) are limited to certain registers. When
2416 this is so, each register class that is used in a bitwise-and or shift
2417 instruction must have a subclass consisting of registers from which
2418 single-byte values can be loaded or stored. This is so that
2419 @code{PREFERRED_RELOAD_CLASS} can always have a possible value to return.
2421 @deftp {Data type} {enum reg_class}
2422 An enumerated type that must be defined with all the register class names
2423 as enumerated values. @code{NO_REGS} must be first. @code{ALL_REGS}
2424 must be the last register class, followed by one more enumerated value,
2425 @code{LIM_REG_CLASSES}, which is not a register class but rather
2426 tells how many classes there are.
2428 Each register class has a number, which is the value of casting
2429 the class name to type @code{int}. The number serves as an index
2430 in many of the tables described below.
2433 @defmac N_REG_CLASSES
2434 The number of distinct register classes, defined as follows:
2437 #define N_REG_CLASSES (int) LIM_REG_CLASSES
2441 @defmac REG_CLASS_NAMES
2442 An initializer containing the names of the register classes as C string
2443 constants. These names are used in writing some of the debugging dumps.
2446 @defmac REG_CLASS_CONTENTS
2447 An initializer containing the contents of the register classes, as integers
2448 which are bit masks. The @var{n}th integer specifies the contents of class
2449 @var{n}. The way the integer @var{mask} is interpreted is that
2450 register @var{r} is in the class if @code{@var{mask} & (1 << @var{r})} is 1.
2452 When the machine has more than 32 registers, an integer does not suffice.
2453 Then the integers are replaced by sub-initializers, braced groupings containing
2454 several integers. Each sub-initializer must be suitable as an initializer
2455 for the type @code{HARD_REG_SET} which is defined in @file{hard-reg-set.h}.
2456 In this situation, the first integer in each sub-initializer corresponds to
2457 registers 0 through 31, the second integer to registers 32 through 63, and
2461 @defmac REGNO_REG_CLASS (@var{regno})
2462 A C expression whose value is a register class containing hard register
2463 @var{regno}. In general there is more than one such class; choose a class
2464 which is @dfn{minimal}, meaning that no smaller class also contains the
2468 @defmac BASE_REG_CLASS
2469 A macro whose definition is the name of the class to which a valid
2470 base register must belong. A base register is one used in an address
2471 which is the register value plus a displacement.
2474 @defmac MODE_BASE_REG_CLASS (@var{mode})
2475 This is a variation of the @code{BASE_REG_CLASS} macro which allows
2476 the selection of a base register in a mode dependent manner. If
2477 @var{mode} is VOIDmode then it should return the same value as
2478 @code{BASE_REG_CLASS}.
2481 @defmac MODE_BASE_REG_REG_CLASS (@var{mode})
2482 A C expression whose value is the register class to which a valid
2483 base register must belong in order to be used in a base plus index
2484 register address. You should define this macro if base plus index
2485 addresses have different requirements than other base register uses.
2488 @defmac MODE_CODE_BASE_REG_CLASS (@var{mode}, @var{outer_code}, @var{index_code})
2489 A C expression whose value is the register class to which a valid
2490 base register must belong. @var{outer_code} and @var{index_code} define the
2491 context in which the base register occurs. @var{outer_code} is the code of
2492 the immediately enclosing expression (@code{MEM} for the top level of an
2493 address, @code{ADDRESS} for something that occurs in an
2494 @code{address_operand}). @var{index_code} is the code of the corresponding
2495 index expression if @var{outer_code} is @code{PLUS}; @code{SCRATCH} otherwise.
2498 @defmac INDEX_REG_CLASS
2499 A macro whose definition is the name of the class to which a valid
2500 index register must belong. An index register is one used in an
2501 address where its value is either multiplied by a scale factor or
2502 added to another register (as well as added to a displacement).
2505 @defmac REGNO_OK_FOR_BASE_P (@var{num})
2506 A C expression which is nonzero if register number @var{num} is
2507 suitable for use as a base register in operand addresses.
2508 Like @code{TARGET_LEGITIMATE_ADDRESS_P}, this macro should also
2509 define a strict and a non-strict variant. Both variants behave
2510 the same for hard register; for pseudos, the strict variant will
2511 pass only those that have been allocated to a valid hard registers,
2512 while the non-strict variant will pass all pseudos.
2514 @findex REG_OK_STRICT
2515 Compiler source files that want to use the strict variant of this and
2516 other macros define the macro @code{REG_OK_STRICT}. You should use an
2517 @code{#ifdef REG_OK_STRICT} conditional to define the strict variant in
2518 that case and the non-strict variant otherwise.
2521 @defmac REGNO_MODE_OK_FOR_BASE_P (@var{num}, @var{mode})
2522 A C expression that is just like @code{REGNO_OK_FOR_BASE_P}, except that
2523 that expression may examine the mode of the memory reference in
2524 @var{mode}. You should define this macro if the mode of the memory
2525 reference affects whether a register may be used as a base register. If
2526 you define this macro, the compiler will use it instead of
2527 @code{REGNO_OK_FOR_BASE_P}. The mode may be @code{VOIDmode} for
2528 addresses that appear outside a @code{MEM}, i.e., as an
2529 @code{address_operand}.
2531 This macro also has strict and non-strict variants.
2534 @defmac REGNO_MODE_OK_FOR_REG_BASE_P (@var{num}, @var{mode})
2535 A C expression which is nonzero if register number @var{num} is suitable for
2536 use as a base register in base plus index operand addresses, accessing
2537 memory in mode @var{mode}. It may be either a suitable hard register or a
2538 pseudo register that has been allocated such a hard register. You should
2539 define this macro if base plus index addresses have different requirements
2540 than other base register uses.
2542 Use of this macro is deprecated; please use the more general
2543 @code{REGNO_MODE_CODE_OK_FOR_BASE_P}.
2545 This macro also has strict and non-strict variants.
2548 @defmac REGNO_MODE_CODE_OK_FOR_BASE_P (@var{num}, @var{mode}, @var{outer_code}, @var{index_code})
2549 A C expression that is just like @code{REGNO_MODE_OK_FOR_BASE_P}, except
2550 that that expression may examine the context in which the register
2551 appears in the memory reference. @var{outer_code} is the code of the
2552 immediately enclosing expression (@code{MEM} if at the top level of the
2553 address, @code{ADDRESS} for something that occurs in an
2554 @code{address_operand}). @var{index_code} is the code of the
2555 corresponding index expression if @var{outer_code} is @code{PLUS};
2556 @code{SCRATCH} otherwise. The mode may be @code{VOIDmode} for addresses
2557 that appear outside a @code{MEM}, i.e., as an @code{address_operand}.
2559 This macro also has strict and non-strict variants.
2562 @defmac REGNO_OK_FOR_INDEX_P (@var{num})
2563 A C expression which is nonzero if register number @var{num} is
2564 suitable for use as an index register in operand addresses. It may be
2565 either a suitable hard register or a pseudo register that has been
2566 allocated such a hard register.
2568 The difference between an index register and a base register is that
2569 the index register may be scaled. If an address involves the sum of
2570 two registers, neither one of them scaled, then either one may be
2571 labeled the ``base'' and the other the ``index''; but whichever
2572 labeling is used must fit the machine's constraints of which registers
2573 may serve in each capacity. The compiler will try both labelings,
2574 looking for one that is valid, and will reload one or both registers
2575 only if neither labeling works.
2577 This macro also has strict and non-strict variants.
2580 @defmac PREFERRED_RELOAD_CLASS (@var{x}, @var{class})
2581 A C expression that places additional restrictions on the register class
2582 to use when it is necessary to copy value @var{x} into a register in class
2583 @var{class}. The value is a register class; perhaps @var{class}, or perhaps
2584 another, smaller class. On many machines, the following definition is
2588 #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS
2591 Sometimes returning a more restrictive class makes better code. For
2592 example, on the 68000, when @var{x} is an integer constant that is in range
2593 for a @samp{moveq} instruction, the value of this macro is always
2594 @code{DATA_REGS} as long as @var{class} includes the data registers.
2595 Requiring a data register guarantees that a @samp{moveq} will be used.
2597 One case where @code{PREFERRED_RELOAD_CLASS} must not return
2598 @var{class} is if @var{x} is a legitimate constant which cannot be
2599 loaded into some register class. By returning @code{NO_REGS} you can
2600 force @var{x} into a memory location. For example, rs6000 can load
2601 immediate values into general-purpose registers, but does not have an
2602 instruction for loading an immediate value into a floating-point
2603 register, so @code{PREFERRED_RELOAD_CLASS} returns @code{NO_REGS} when
2604 @var{x} is a floating-point constant. If the constant can't be loaded
2605 into any kind of register, code generation will be better if
2606 @code{LEGITIMATE_CONSTANT_P} makes the constant illegitimate instead
2607 of using @code{PREFERRED_RELOAD_CLASS}.
2609 If an insn has pseudos in it after register allocation, reload will go
2610 through the alternatives and call repeatedly @code{PREFERRED_RELOAD_CLASS}
2611 to find the best one. Returning @code{NO_REGS}, in this case, makes
2612 reload add a @code{!} in front of the constraint: the x86 back-end uses
2613 this feature to discourage usage of 387 registers when math is done in
2614 the SSE registers (and vice versa).
2617 @defmac PREFERRED_OUTPUT_RELOAD_CLASS (@var{x}, @var{class})
2618 Like @code{PREFERRED_RELOAD_CLASS}, but for output reloads instead of
2619 input reloads. If you don't define this macro, the default is to use
2620 @var{class}, unchanged.
2622 You can also use @code{PREFERRED_OUTPUT_RELOAD_CLASS} to discourage
2623 reload from using some alternatives, like @code{PREFERRED_RELOAD_CLASS}.
2626 @defmac LIMIT_RELOAD_CLASS (@var{mode}, @var{class})
2627 A C expression that places additional restrictions on the register class
2628 to use when it is necessary to be able to hold a value of mode
2629 @var{mode} in a reload register for which class @var{class} would
2632 Unlike @code{PREFERRED_RELOAD_CLASS}, this macro should be used when
2633 there are certain modes that simply can't go in certain reload classes.
2635 The value is a register class; perhaps @var{class}, or perhaps another,
2638 Don't define this macro unless the target machine has limitations which
2639 require the macro to do something nontrivial.
2642 @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})
2643 Many machines have some registers that cannot be copied directly to or
2644 from memory or even from other types of registers. An example is the
2645 @samp{MQ} register, which on most machines, can only be copied to or
2646 from general registers, but not memory. Below, we shall be using the
2647 term 'intermediate register' when a move operation cannot be performed
2648 directly, but has to be done by copying the source into the intermediate
2649 register first, and then copying the intermediate register to the
2650 destination. An intermediate register always has the same mode as
2651 source and destination. Since it holds the actual value being copied,
2652 reload might apply optimizations to re-use an intermediate register
2653 and eliding the copy from the source when it can determine that the
2654 intermediate register still holds the required value.
2656 Another kind of secondary reload is required on some machines which
2657 allow copying all registers to and from memory, but require a scratch
2658 register for stores to some memory locations (e.g., those with symbolic
2659 address on the RT, and those with certain symbolic address on the SPARC
2660 when compiling PIC)@. Scratch registers need not have the same mode
2661 as the value being copied, and usually hold a different value than
2662 that being copied. Special patterns in the md file are needed to
2663 describe how the copy is performed with the help of the scratch register;
2664 these patterns also describe the number, register class(es) and mode(s)
2665 of the scratch register(s).
2667 In some cases, both an intermediate and a scratch register are required.
2669 For input reloads, this target hook is called with nonzero @var{in_p},
2670 and @var{x} is an rtx that needs to be copied to a register of class
2671 @var{reload_class} in @var{reload_mode}. For output reloads, this target
2672 hook is called with zero @var{in_p}, and a register of class @var{reload_class}
2673 needs to be copied to rtx @var{x} in @var{reload_mode}.
2675 If copying a register of @var{reload_class} from/to @var{x} requires
2676 an intermediate register, the hook @code{secondary_reload} should
2677 return the register class required for this intermediate register.
2678 If no intermediate register is required, it should return NO_REGS.
2679 If more than one intermediate register is required, describe the one
2680 that is closest in the copy chain to the reload register.
2682 If scratch registers are needed, you also have to describe how to
2683 perform the copy from/to the reload register to/from this
2684 closest intermediate register. Or if no intermediate register is
2685 required, but still a scratch register is needed, describe the
2686 copy from/to the reload register to/from the reload operand @var{x}.
2688 You do this by setting @code{sri->icode} to the instruction code of a pattern
2689 in the md file which performs the move. Operands 0 and 1 are the output
2690 and input of this copy, respectively. Operands from operand 2 onward are
2691 for scratch operands. These scratch operands must have a mode, and a
2692 single-register-class
2693 @c [later: or memory]
2696 When an intermediate register is used, the @code{secondary_reload}
2697 hook will be called again to determine how to copy the intermediate
2698 register to/from the reload operand @var{x}, so your hook must also
2699 have code to handle the register class of the intermediate operand.
2701 @c [For later: maybe we'll allow multi-alternative reload patterns -
2702 @c the port maintainer could name a mov<mode> pattern that has clobbers -
2703 @c and match the constraints of input and output to determine the required
2704 @c alternative. A restriction would be that constraints used to match
2705 @c against reloads registers would have to be written as register class
2706 @c constraints, or we need a new target macro / hook that tells us if an
2707 @c arbitrary constraint can match an unknown register of a given class.
2708 @c Such a macro / hook would also be useful in other places.]
2711 @var{x} might be a pseudo-register or a @code{subreg} of a
2712 pseudo-register, which could either be in a hard register or in memory.
2713 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2714 in memory and the hard register number if it is in a register.
2716 Scratch operands in memory (constraint @code{"=m"} / @code{"=&m"}) are
2717 currently not supported. For the time being, you will have to continue
2718 to use @code{SECONDARY_MEMORY_NEEDED} for that purpose.
2720 @code{copy_cost} also uses this target hook to find out how values are
2721 copied. If you want it to include some extra cost for the need to allocate
2722 (a) scratch register(s), set @code{sri->extra_cost} to the additional cost.
2723 Or if two dependent moves are supposed to have a lower cost than the sum
2724 of the individual moves due to expected fortuitous scheduling and/or special
2725 forwarding logic, you can set @code{sri->extra_cost} to a negative amount.
2728 @defmac SECONDARY_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2729 @defmacx SECONDARY_INPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2730 @defmacx SECONDARY_OUTPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2731 These macros are obsolete, new ports should use the target hook
2732 @code{TARGET_SECONDARY_RELOAD} instead.
2734 These are obsolete macros, replaced by the @code{TARGET_SECONDARY_RELOAD}
2735 target hook. Older ports still define these macros to indicate to the
2736 reload phase that it may
2737 need to allocate at least one register for a reload in addition to the
2738 register to contain the data. Specifically, if copying @var{x} to a
2739 register @var{class} in @var{mode} requires an intermediate register,
2740 you were supposed to define @code{SECONDARY_INPUT_RELOAD_CLASS} to return the
2741 largest register class all of whose registers can be used as
2742 intermediate registers or scratch registers.
2744 If copying a register @var{class} in @var{mode} to @var{x} requires an
2745 intermediate or scratch register, @code{SECONDARY_OUTPUT_RELOAD_CLASS}
2746 was supposed to be defined be defined to return the largest register
2747 class required. If the
2748 requirements for input and output reloads were the same, the macro
2749 @code{SECONDARY_RELOAD_CLASS} should have been used instead of defining both
2752 The values returned by these macros are often @code{GENERAL_REGS}.
2753 Return @code{NO_REGS} if no spare register is needed; i.e., if @var{x}
2754 can be directly copied to or from a register of @var{class} in
2755 @var{mode} without requiring a scratch register. Do not define this
2756 macro if it would always return @code{NO_REGS}.
2758 If a scratch register is required (either with or without an
2759 intermediate register), you were supposed to define patterns for
2760 @samp{reload_in@var{m}} or @samp{reload_out@var{m}}, as required
2761 (@pxref{Standard Names}. These patterns, which were normally
2762 implemented with a @code{define_expand}, should be similar to the
2763 @samp{mov@var{m}} patterns, except that operand 2 is the scratch
2766 These patterns need constraints for the reload register and scratch
2768 contain a single register class. If the original reload register (whose
2769 class is @var{class}) can meet the constraint given in the pattern, the
2770 value returned by these macros is used for the class of the scratch
2771 register. Otherwise, two additional reload registers are required.
2772 Their classes are obtained from the constraints in the insn pattern.
2774 @var{x} might be a pseudo-register or a @code{subreg} of a
2775 pseudo-register, which could either be in a hard register or in memory.
2776 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2777 in memory and the hard register number if it is in a register.
2779 These macros should not be used in the case where a particular class of
2780 registers can only be copied to memory and not to another class of
2781 registers. In that case, secondary reload registers are not needed and
2782 would not be helpful. Instead, a stack location must be used to perform
2783 the copy and the @code{mov@var{m}} pattern should use memory as an
2784 intermediate storage. This case often occurs between floating-point and
2788 @defmac SECONDARY_MEMORY_NEEDED (@var{class1}, @var{class2}, @var{m})
2789 Certain machines have the property that some registers cannot be copied
2790 to some other registers without using memory. Define this macro on
2791 those machines to be a C expression that is nonzero if objects of mode
2792 @var{m} in registers of @var{class1} can only be copied to registers of
2793 class @var{class2} by storing a register of @var{class1} into memory
2794 and loading that memory location into a register of @var{class2}.
2796 Do not define this macro if its value would always be zero.
2799 @defmac SECONDARY_MEMORY_NEEDED_RTX (@var{mode})
2800 Normally when @code{SECONDARY_MEMORY_NEEDED} is defined, the compiler
2801 allocates a stack slot for a memory location needed for register copies.
2802 If this macro is defined, the compiler instead uses the memory location
2803 defined by this macro.
2805 Do not define this macro if you do not define
2806 @code{SECONDARY_MEMORY_NEEDED}.
2809 @defmac SECONDARY_MEMORY_NEEDED_MODE (@var{mode})
2810 When the compiler needs a secondary memory location to copy between two
2811 registers of mode @var{mode}, it normally allocates sufficient memory to
2812 hold a quantity of @code{BITS_PER_WORD} bits and performs the store and
2813 load operations in a mode that many bits wide and whose class is the
2814 same as that of @var{mode}.
2816 This is right thing to do on most machines because it ensures that all
2817 bits of the register are copied and prevents accesses to the registers
2818 in a narrower mode, which some machines prohibit for floating-point
2821 However, this default behavior is not correct on some machines, such as
2822 the DEC Alpha, that store short integers in floating-point registers
2823 differently than in integer registers. On those machines, the default
2824 widening will not work correctly and you must define this macro to
2825 suppress that widening in some cases. See the file @file{alpha.h} for
2828 Do not define this macro if you do not define
2829 @code{SECONDARY_MEMORY_NEEDED} or if widening @var{mode} to a mode that
2830 is @code{BITS_PER_WORD} bits wide is correct for your machine.
2833 @defmac SMALL_REGISTER_CLASSES
2834 On some machines, it is risky to let hard registers live across arbitrary
2835 insns. Typically, these machines have instructions that require values
2836 to be in specific registers (like an accumulator), and reload will fail
2837 if the required hard register is used for another purpose across such an
2840 Define @code{SMALL_REGISTER_CLASSES} to be an expression with a nonzero
2841 value on these machines. When this macro has a nonzero value, the
2842 compiler will try to minimize the lifetime of hard registers.
2844 It is always safe to define this macro with a nonzero value, but if you
2845 unnecessarily define it, you will reduce the amount of optimizations
2846 that can be performed in some cases. If you do not define this macro
2847 with a nonzero value when it is required, the compiler will run out of
2848 spill registers and print a fatal error message. For most machines, you
2849 should not define this macro at all.
2852 @defmac CLASS_LIKELY_SPILLED_P (@var{class})
2853 A C expression whose value is nonzero if pseudos that have been assigned
2854 to registers of class @var{class} would likely be spilled because
2855 registers of @var{class} are needed for spill registers.
2857 The default value of this macro returns 1 if @var{class} has exactly one
2858 register and zero otherwise. On most machines, this default should be
2859 used. Only define this macro to some other expression if pseudos
2860 allocated by @file{local-alloc.c} end up in memory because their hard
2861 registers were needed for spill registers. If this macro returns nonzero
2862 for those classes, those pseudos will only be allocated by
2863 @file{global.c}, which knows how to reallocate the pseudo to another
2864 register. If there would not be another register available for
2865 reallocation, you should not change the definition of this macro since
2866 the only effect of such a definition would be to slow down register
2870 @defmac CLASS_MAX_NREGS (@var{class}, @var{mode})
2871 A C expression for the maximum number of consecutive registers
2872 of class @var{class} needed to hold a value of mode @var{mode}.
2874 This is closely related to the macro @code{HARD_REGNO_NREGS}. In fact,
2875 the value of the macro @code{CLASS_MAX_NREGS (@var{class}, @var{mode})}
2876 should be the maximum value of @code{HARD_REGNO_NREGS (@var{regno},
2877 @var{mode})} for all @var{regno} values in the class @var{class}.
2879 This macro helps control the handling of multiple-word values
2883 @defmac CANNOT_CHANGE_MODE_CLASS (@var{from}, @var{to}, @var{class})
2884 If defined, a C expression that returns nonzero for a @var{class} for which
2885 a change from mode @var{from} to mode @var{to} is invalid.
2887 For the example, loading 32-bit integer or floating-point objects into
2888 floating-point registers on the Alpha extends them to 64 bits.
2889 Therefore loading a 64-bit object and then storing it as a 32-bit object
2890 does not store the low-order 32 bits, as would be the case for a normal
2891 register. Therefore, @file{alpha.h} defines @code{CANNOT_CHANGE_MODE_CLASS}
2895 #define CANNOT_CHANGE_MODE_CLASS(FROM, TO, CLASS) \
2896 (GET_MODE_SIZE (FROM) != GET_MODE_SIZE (TO) \
2897 ? reg_classes_intersect_p (FLOAT_REGS, (CLASS)) : 0)
2901 @deftypefn {Target Hook} {const enum reg_class *} TARGET_IRA_COVER_CLASSES ()
2902 Return an array of cover classes for the Integrated Register Allocator
2903 (@acronym{IRA}). Cover classes are a set of non-intersecting register
2904 classes covering all hard registers used for register allocation
2905 purposes. If a move between two registers in the same cover class is
2906 possible, it should be cheaper than a load or store of the registers.
2907 The array is terminated by a @code{LIM_REG_CLASSES} element.
2909 The order of cover classes in the array is important. If two classes
2910 have the same cost of usage for a pseudo, the class occurred first in
2911 the array is chosen for the pseudo.
2913 This hook is called once at compiler startup, after the command-line
2914 options have been processed. It is then re-examined by every call to
2915 @code{target_reinit}.
2917 The default implementation returns @code{IRA_COVER_CLASSES}, if defined,
2918 otherwise there is no default implementation. You must define either this
2919 macro or @code{IRA_COVER_CLASSES} in order to use the integrated register
2920 allocator with Chaitin-Briggs coloring. If the macro is not defined,
2921 the only available coloring algorithm is Chow's priority coloring.
2924 @defmac IRA_COVER_CLASSES
2925 See the documentation for @code{TARGET_IRA_COVER_CLASSES}.
2928 @node Old Constraints
2929 @section Obsolete Macros for Defining Constraints
2930 @cindex defining constraints, obsolete method
2931 @cindex constraints, defining, obsolete method
2933 Machine-specific constraints can be defined with these macros instead
2934 of the machine description constructs described in @ref{Define
2935 Constraints}. This mechanism is obsolete. New ports should not use
2936 it; old ports should convert to the new mechanism.
2938 @defmac CONSTRAINT_LEN (@var{char}, @var{str})
2939 For the constraint at the start of @var{str}, which starts with the letter
2940 @var{c}, return the length. This allows you to have register class /
2941 constant / extra constraints that are longer than a single letter;
2942 you don't need to define this macro if you can do with single-letter
2943 constraints only. The definition of this macro should use
2944 DEFAULT_CONSTRAINT_LEN for all the characters that you don't want
2945 to handle specially.
2946 There are some sanity checks in genoutput.c that check the constraint lengths
2947 for the md file, so you can also use this macro to help you while you are
2948 transitioning from a byzantine single-letter-constraint scheme: when you
2949 return a negative length for a constraint you want to re-use, genoutput
2950 will complain about every instance where it is used in the md file.
2953 @defmac REG_CLASS_FROM_LETTER (@var{char})
2954 A C expression which defines the machine-dependent operand constraint
2955 letters for register classes. If @var{char} is such a letter, the
2956 value should be the register class corresponding to it. Otherwise,
2957 the value should be @code{NO_REGS}. The register letter @samp{r},
2958 corresponding to class @code{GENERAL_REGS}, will not be passed
2959 to this macro; you do not need to handle it.
2962 @defmac REG_CLASS_FROM_CONSTRAINT (@var{char}, @var{str})
2963 Like @code{REG_CLASS_FROM_LETTER}, but you also get the constraint string
2964 passed in @var{str}, so that you can use suffixes to distinguish between
2968 @defmac CONST_OK_FOR_LETTER_P (@var{value}, @var{c})
2969 A C expression that defines the machine-dependent operand constraint
2970 letters (@samp{I}, @samp{J}, @samp{K}, @dots{} @samp{P}) that specify
2971 particular ranges of integer values. If @var{c} is one of those
2972 letters, the expression should check that @var{value}, an integer, is in
2973 the appropriate range and return 1 if so, 0 otherwise. If @var{c} is
2974 not one of those letters, the value should be 0 regardless of
2978 @defmac CONST_OK_FOR_CONSTRAINT_P (@var{value}, @var{c}, @var{str})
2979 Like @code{CONST_OK_FOR_LETTER_P}, but you also get the constraint
2980 string passed in @var{str}, so that you can use suffixes to distinguish
2981 between different variants.
2984 @defmac CONST_DOUBLE_OK_FOR_LETTER_P (@var{value}, @var{c})
2985 A C expression that defines the machine-dependent operand constraint
2986 letters that specify particular ranges of @code{const_double} values
2987 (@samp{G} or @samp{H}).
2989 If @var{c} is one of those letters, the expression should check that
2990 @var{value}, an RTX of code @code{const_double}, is in the appropriate
2991 range and return 1 if so, 0 otherwise. If @var{c} is not one of those
2992 letters, the value should be 0 regardless of @var{value}.
2994 @code{const_double} is used for all floating-point constants and for
2995 @code{DImode} fixed-point constants. A given letter can accept either
2996 or both kinds of values. It can use @code{GET_MODE} to distinguish
2997 between these kinds.
3000 @defmac CONST_DOUBLE_OK_FOR_CONSTRAINT_P (@var{value}, @var{c}, @var{str})
3001 Like @code{CONST_DOUBLE_OK_FOR_LETTER_P}, but you also get the constraint
3002 string passed in @var{str}, so that you can use suffixes to distinguish
3003 between different variants.
3006 @defmac EXTRA_CONSTRAINT (@var{value}, @var{c})
3007 A C expression that defines the optional machine-dependent constraint
3008 letters that can be used to segregate specific types of operands, usually
3009 memory references, for the target machine. Any letter that is not
3010 elsewhere defined and not matched by @code{REG_CLASS_FROM_LETTER} /
3011 @code{REG_CLASS_FROM_CONSTRAINT}
3012 may be used. Normally this macro will not be defined.
3014 If it is required for a particular target machine, it should return 1
3015 if @var{value} corresponds to the operand type represented by the
3016 constraint letter @var{c}. If @var{c} is not defined as an extra
3017 constraint, the value returned should be 0 regardless of @var{value}.
3019 For example, on the ROMP, load instructions cannot have their output
3020 in r0 if the memory reference contains a symbolic address. Constraint
3021 letter @samp{Q} is defined as representing a memory address that does
3022 @emph{not} contain a symbolic address. An alternative is specified with
3023 a @samp{Q} constraint on the input and @samp{r} on the output. The next
3024 alternative specifies @samp{m} on the input and a register class that
3025 does not include r0 on the output.
3028 @defmac EXTRA_CONSTRAINT_STR (@var{value}, @var{c}, @var{str})
3029 Like @code{EXTRA_CONSTRAINT}, but you also get the constraint string passed
3030 in @var{str}, so that you can use suffixes to distinguish between different
3034 @defmac EXTRA_MEMORY_CONSTRAINT (@var{c}, @var{str})
3035 A C expression that defines the optional machine-dependent constraint
3036 letters, amongst those accepted by @code{EXTRA_CONSTRAINT}, that should
3037 be treated like memory constraints by the reload pass.
3039 It should return 1 if the operand type represented by the constraint
3040 at the start of @var{str}, the first letter of which is the letter @var{c},
3041 comprises a subset of all memory references including
3042 all those whose address is simply a base register. This allows the reload
3043 pass to reload an operand, if it does not directly correspond to the operand
3044 type of @var{c}, by copying its address into a base register.
3046 For example, on the S/390, some instructions do not accept arbitrary
3047 memory references, but only those that do not make use of an index
3048 register. The constraint letter @samp{Q} is defined via
3049 @code{EXTRA_CONSTRAINT} as representing a memory address of this type.
3050 If the letter @samp{Q} is marked as @code{EXTRA_MEMORY_CONSTRAINT},
3051 a @samp{Q} constraint can handle any memory operand, because the
3052 reload pass knows it can be reloaded by copying the memory address
3053 into a base register if required. This is analogous to the way
3054 an @samp{o} constraint can handle any memory operand.
3057 @defmac EXTRA_ADDRESS_CONSTRAINT (@var{c}, @var{str})
3058 A C expression that defines the optional machine-dependent constraint
3059 letters, amongst those accepted by @code{EXTRA_CONSTRAINT} /
3060 @code{EXTRA_CONSTRAINT_STR}, that should
3061 be treated like address constraints by the reload pass.
3063 It should return 1 if the operand type represented by the constraint
3064 at the start of @var{str}, which starts with the letter @var{c}, comprises
3065 a subset of all memory addresses including
3066 all those that consist of just a base register. This allows the reload
3067 pass to reload an operand, if it does not directly correspond to the operand
3068 type of @var{str}, by copying it into a base register.
3070 Any constraint marked as @code{EXTRA_ADDRESS_CONSTRAINT} can only
3071 be used with the @code{address_operand} predicate. It is treated
3072 analogously to the @samp{p} constraint.
3075 @node Stack and Calling
3076 @section Stack Layout and Calling Conventions
3077 @cindex calling conventions
3079 @c prevent bad page break with this line
3080 This describes the stack layout and calling conventions.
3084 * Exception Handling::
3089 * Register Arguments::
3091 * Aggregate Return::
3096 * Stack Smashing Protection::
3100 @subsection Basic Stack Layout
3101 @cindex stack frame layout
3102 @cindex frame layout
3104 @c prevent bad page break with this line
3105 Here is the basic stack layout.
3107 @defmac STACK_GROWS_DOWNWARD
3108 Define this macro if pushing a word onto the stack moves the stack
3109 pointer to a smaller address.
3111 When we say, ``define this macro if @dots{}'', it means that the
3112 compiler checks this macro only with @code{#ifdef} so the precise
3113 definition used does not matter.
3116 @defmac STACK_PUSH_CODE
3117 This macro defines the operation used when something is pushed
3118 on the stack. In RTL, a push operation will be
3119 @code{(set (mem (STACK_PUSH_CODE (reg sp))) @dots{})}
3121 The choices are @code{PRE_DEC}, @code{POST_DEC}, @code{PRE_INC},
3122 and @code{POST_INC}. Which of these is correct depends on
3123 the stack direction and on whether the stack pointer points
3124 to the last item on the stack or whether it points to the
3125 space for the next item on the stack.
3127 The default is @code{PRE_DEC} when @code{STACK_GROWS_DOWNWARD} is
3128 defined, which is almost always right, and @code{PRE_INC} otherwise,
3129 which is often wrong.
3132 @defmac FRAME_GROWS_DOWNWARD
3133 Define this macro to nonzero value if the addresses of local variable slots
3134 are at negative offsets from the frame pointer.
3137 @defmac ARGS_GROW_DOWNWARD
3138 Define this macro if successive arguments to a function occupy decreasing
3139 addresses on the stack.
3142 @defmac STARTING_FRAME_OFFSET
3143 Offset from the frame pointer to the first local variable slot to be allocated.
3145 If @code{FRAME_GROWS_DOWNWARD}, find the next slot's offset by
3146 subtracting the first slot's length from @code{STARTING_FRAME_OFFSET}.
3147 Otherwise, it is found by adding the length of the first slot to the
3148 value @code{STARTING_FRAME_OFFSET}.
3149 @c i'm not sure if the above is still correct.. had to change it to get
3150 @c rid of an overfull. --mew 2feb93
3153 @defmac STACK_ALIGNMENT_NEEDED
3154 Define to zero to disable final alignment of the stack during reload.
3155 The nonzero default for this macro is suitable for most ports.
3157 On ports where @code{STARTING_FRAME_OFFSET} is nonzero or where there
3158 is a register save block following the local block that doesn't require
3159 alignment to @code{STACK_BOUNDARY}, it may be beneficial to disable
3160 stack alignment and do it in the backend.
3163 @defmac STACK_POINTER_OFFSET
3164 Offset from the stack pointer register to the first location at which
3165 outgoing arguments are placed. If not specified, the default value of
3166 zero is used. This is the proper value for most machines.
3168 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
3169 the first location at which outgoing arguments are placed.
3172 @defmac FIRST_PARM_OFFSET (@var{fundecl})
3173 Offset from the argument pointer register to the first argument's
3174 address. On some machines it may depend on the data type of the
3177 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
3178 the first argument's address.
3181 @defmac STACK_DYNAMIC_OFFSET (@var{fundecl})
3182 Offset from the stack pointer register to an item dynamically allocated
3183 on the stack, e.g., by @code{alloca}.
3185 The default value for this macro is @code{STACK_POINTER_OFFSET} plus the
3186 length of the outgoing arguments. The default is correct for most
3187 machines. See @file{function.c} for details.
3190 @defmac INITIAL_FRAME_ADDRESS_RTX
3191 A C expression whose value is RTL representing the address of the initial
3192 stack frame. This address is passed to @code{RETURN_ADDR_RTX} and
3193 @code{DYNAMIC_CHAIN_ADDRESS}. If you don't define this macro, a reasonable
3194 default value will be used. Define this macro in order to make frame pointer
3195 elimination work in the presence of @code{__builtin_frame_address (count)} and
3196 @code{__builtin_return_address (count)} for @code{count} not equal to zero.
3199 @defmac DYNAMIC_CHAIN_ADDRESS (@var{frameaddr})
3200 A C expression whose value is RTL representing the address in a stack
3201 frame where the pointer to the caller's frame is stored. Assume that
3202 @var{frameaddr} is an RTL expression for the address of the stack frame
3205 If you don't define this macro, the default is to return the value
3206 of @var{frameaddr}---that is, the stack frame address is also the
3207 address of the stack word that points to the previous frame.
3210 @defmac SETUP_FRAME_ADDRESSES
3211 If defined, a C expression that produces the machine-specific code to
3212 setup the stack so that arbitrary frames can be accessed. For example,
3213 on the SPARC, we must flush all of the register windows to the stack
3214 before we can access arbitrary stack frames. You will seldom need to
3218 @deftypefn {Target Hook} bool TARGET_BUILTIN_SETJMP_FRAME_VALUE ()
3219 This target hook should return an rtx that is used to store
3220 the address of the current frame into the built in @code{setjmp} buffer.
3221 The default value, @code{virtual_stack_vars_rtx}, is correct for most
3222 machines. One reason you may need to define this target hook is if
3223 @code{hard_frame_pointer_rtx} is the appropriate value on your machine.
3226 @defmac FRAME_ADDR_RTX (@var{frameaddr})
3227 A C expression whose value is RTL representing the value of the frame
3228 address for the current frame. @var{frameaddr} is the frame pointer
3229 of the current frame. This is used for __builtin_frame_address.
3230 You need only define this macro if the frame address is not the same
3231 as the frame pointer. Most machines do not need to define it.
3234 @defmac RETURN_ADDR_RTX (@var{count}, @var{frameaddr})
3235 A C expression whose value is RTL representing the value of the return
3236 address for the frame @var{count} steps up from the current frame, after
3237 the prologue. @var{frameaddr} is the frame pointer of the @var{count}
3238 frame, or the frame pointer of the @var{count} @minus{} 1 frame if
3239 @code{RETURN_ADDR_IN_PREVIOUS_FRAME} is defined.
3241 The value of the expression must always be the correct address when
3242 @var{count} is zero, but may be @code{NULL_RTX} if there is no way to
3243 determine the return address of other frames.
3246 @defmac RETURN_ADDR_IN_PREVIOUS_FRAME
3247 Define this if the return address of a particular stack frame is accessed
3248 from the frame pointer of the previous stack frame.
3251 @defmac INCOMING_RETURN_ADDR_RTX
3252 A C expression whose value is RTL representing the location of the
3253 incoming return address at the beginning of any function, before the
3254 prologue. This RTL is either a @code{REG}, indicating that the return
3255 value is saved in @samp{REG}, or a @code{MEM} representing a location in
3258 You only need to define this macro if you want to support call frame
3259 debugging information like that provided by DWARF 2.
3261 If this RTL is a @code{REG}, you should also define
3262 @code{DWARF_FRAME_RETURN_COLUMN} to @code{DWARF_FRAME_REGNUM (REGNO)}.
3265 @defmac DWARF_ALT_FRAME_RETURN_COLUMN
3266 A C expression whose value is an integer giving a DWARF 2 column
3267 number that may be used as an alternative return column. The column
3268 must not correspond to any gcc hard register (that is, it must not
3269 be in the range of @code{DWARF_FRAME_REGNUM}).
3271 This macro can be useful if @code{DWARF_FRAME_RETURN_COLUMN} is set to a
3272 general register, but an alternative column needs to be used for signal
3273 frames. Some targets have also used different frame return columns
3277 @defmac DWARF_ZERO_REG
3278 A C expression whose value is an integer giving a DWARF 2 register
3279 number that is considered to always have the value zero. This should
3280 only be defined if the target has an architected zero register, and
3281 someone decided it was a good idea to use that register number to
3282 terminate the stack backtrace. New ports should avoid this.
3285 @deftypefn {Target Hook} void TARGET_DWARF_HANDLE_FRAME_UNSPEC (const char *@var{label}, rtx @var{pattern}, int @var{index})
3286 This target hook allows the backend to emit frame-related insns that
3287 contain UNSPECs or UNSPEC_VOLATILEs. The DWARF 2 call frame debugging
3288 info engine will invoke it on insns of the form
3290 (set (reg) (unspec [@dots{}] UNSPEC_INDEX))
3294 (set (reg) (unspec_volatile [@dots{}] UNSPECV_INDEX)).
3296 to let the backend emit the call frame instructions. @var{label} is
3297 the CFI label attached to the insn, @var{pattern} is the pattern of
3298 the insn and @var{index} is @code{UNSPEC_INDEX} or @code{UNSPECV_INDEX}.
3301 @defmac INCOMING_FRAME_SP_OFFSET
3302 A C expression whose value is an integer giving the offset, in bytes,
3303 from the value of the stack pointer register to the top of the stack
3304 frame at the beginning of any function, before the prologue. The top of
3305 the frame is defined to be the value of the stack pointer in the
3306 previous frame, just before the call instruction.
3308 You only need to define this macro if you want to support call frame
3309 debugging information like that provided by DWARF 2.
3312 @defmac ARG_POINTER_CFA_OFFSET (@var{fundecl})
3313 A C expression whose value is an integer giving the offset, in bytes,
3314 from the argument pointer to the canonical frame address (cfa). The
3315 final value should coincide with that calculated by
3316 @code{INCOMING_FRAME_SP_OFFSET}. Which is unfortunately not usable
3317 during virtual register instantiation.
3319 The default value for this macro is @code{FIRST_PARM_OFFSET (fundecl)},
3320 which is correct for most machines; in general, the arguments are found
3321 immediately before the stack frame. Note that this is not the case on
3322 some targets that save registers into the caller's frame, such as SPARC
3323 and rs6000, and so such targets need to define this macro.
3325 You only need to define this macro if the default is incorrect, and you
3326 want to support call frame debugging information like that provided by
3330 @defmac FRAME_POINTER_CFA_OFFSET (@var{fundecl})
3331 If defined, a C expression whose value is an integer giving the offset
3332 in bytes from the frame pointer to the canonical frame address (cfa).
3333 The final value should coincide with that calculated by
3334 @code{INCOMING_FRAME_SP_OFFSET}.
3336 Normally the CFA is calculated as an offset from the argument pointer,
3337 via @code{ARG_POINTER_CFA_OFFSET}, but if the argument pointer is
3338 variable due to the ABI, this may not be possible. If this macro is
3339 defined, it implies that the virtual register instantiation should be
3340 based on the frame pointer instead of the argument pointer. Only one
3341 of @code{FRAME_POINTER_CFA_OFFSET} and @code{ARG_POINTER_CFA_OFFSET}
3345 @defmac CFA_FRAME_BASE_OFFSET (@var{fundecl})
3346 If defined, a C expression whose value is an integer giving the offset
3347 in bytes from the canonical frame address (cfa) to the frame base used
3348 in DWARF 2 debug information. The default is zero. A different value
3349 may reduce the size of debug information on some ports.
3352 @node Exception Handling
3353 @subsection Exception Handling Support
3354 @cindex exception handling
3356 @defmac EH_RETURN_DATA_REGNO (@var{N})
3357 A C expression whose value is the @var{N}th register number used for
3358 data by exception handlers, or @code{INVALID_REGNUM} if fewer than
3359 @var{N} registers are usable.
3361 The exception handling library routines communicate with the exception
3362 handlers via a set of agreed upon registers. Ideally these registers
3363 should be call-clobbered; it is possible to use call-saved registers,
3364 but may negatively impact code size. The target must support at least
3365 2 data registers, but should define 4 if there are enough free registers.
3367 You must define this macro if you want to support call frame exception
3368 handling like that provided by DWARF 2.
3371 @defmac EH_RETURN_STACKADJ_RTX
3372 A C expression whose value is RTL representing a location in which
3373 to store a stack adjustment to be applied before function return.
3374 This is used to unwind the stack to an exception handler's call frame.
3375 It will be assigned zero on code paths that return normally.
3377 Typically this is a call-clobbered hard register that is otherwise
3378 untouched by the epilogue, but could also be a stack slot.
3380 Do not define this macro if the stack pointer is saved and restored
3381 by the regular prolog and epilog code in the call frame itself; in
3382 this case, the exception handling library routines will update the
3383 stack location to be restored in place. Otherwise, you must define
3384 this macro if you want to support call frame exception handling like
3385 that provided by DWARF 2.
3388 @defmac EH_RETURN_HANDLER_RTX
3389 A C expression whose value is RTL representing a location in which
3390 to store the address of an exception handler to which we should
3391 return. It will not be assigned on code paths that return normally.
3393 Typically this is the location in the call frame at which the normal
3394 return address is stored. For targets that return by popping an
3395 address off the stack, this might be a memory address just below
3396 the @emph{target} call frame rather than inside the current call
3397 frame. If defined, @code{EH_RETURN_STACKADJ_RTX} will have already
3398 been assigned, so it may be used to calculate the location of the
3401 Some targets have more complex requirements than storing to an
3402 address calculable during initial code generation. In that case
3403 the @code{eh_return} instruction pattern should be used instead.
3405 If you want to support call frame exception handling, you must
3406 define either this macro or the @code{eh_return} instruction pattern.
3409 @defmac RETURN_ADDR_OFFSET
3410 If defined, an integer-valued C expression for which rtl will be generated
3411 to add it to the exception handler address before it is searched in the
3412 exception handling tables, and to subtract it again from the address before
3413 using it to return to the exception handler.
3416 @defmac ASM_PREFERRED_EH_DATA_FORMAT (@var{code}, @var{global})
3417 This macro chooses the encoding of pointers embedded in the exception
3418 handling sections. If at all possible, this should be defined such
3419 that the exception handling section will not require dynamic relocations,
3420 and so may be read-only.
3422 @var{code} is 0 for data, 1 for code labels, 2 for function pointers.
3423 @var{global} is true if the symbol may be affected by dynamic relocations.
3424 The macro should return a combination of the @code{DW_EH_PE_*} defines
3425 as found in @file{dwarf2.h}.
3427 If this macro is not defined, pointers will not be encoded but
3428 represented directly.
3431 @defmac ASM_MAYBE_OUTPUT_ENCODED_ADDR_RTX (@var{file}, @var{encoding}, @var{size}, @var{addr}, @var{done})
3432 This macro allows the target to emit whatever special magic is required
3433 to represent the encoding chosen by @code{ASM_PREFERRED_EH_DATA_FORMAT}.
3434 Generic code takes care of pc-relative and indirect encodings; this must
3435 be defined if the target uses text-relative or data-relative encodings.
3437 This is a C statement that branches to @var{done} if the format was
3438 handled. @var{encoding} is the format chosen, @var{size} is the number
3439 of bytes that the format occupies, @var{addr} is the @code{SYMBOL_REF}
3443 @defmac MD_UNWIND_SUPPORT
3444 A string specifying a file to be #include'd in unwind-dw2.c. The file
3445 so included typically defines @code{MD_FALLBACK_FRAME_STATE_FOR}.
3448 @defmac MD_FALLBACK_FRAME_STATE_FOR (@var{context}, @var{fs})
3449 This macro allows the target to add CPU and operating system specific
3450 code to the call-frame unwinder for use when there is no unwind data
3451 available. The most common reason to implement this macro is to unwind
3452 through signal frames.
3454 This macro is called from @code{uw_frame_state_for} in
3455 @file{unwind-dw2.c}, @file{unwind-dw2-xtensa.c} and
3456 @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context};
3457 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{context->ra}
3458 for the address of the code being executed and @code{context->cfa} for
3459 the stack pointer value. If the frame can be decoded, the register
3460 save addresses should be updated in @var{fs} and the macro should
3461 evaluate to @code{_URC_NO_REASON}. If the frame cannot be decoded,
3462 the macro should evaluate to @code{_URC_END_OF_STACK}.
3464 For proper signal handling in Java this macro is accompanied by
3465 @code{MAKE_THROW_FRAME}, defined in @file{libjava/include/*-signal.h} headers.
3468 @defmac MD_HANDLE_UNWABI (@var{context}, @var{fs})
3469 This macro allows the target to add operating system specific code to the
3470 call-frame unwinder to handle the IA-64 @code{.unwabi} unwinding directive,
3471 usually used for signal or interrupt frames.
3473 This macro is called from @code{uw_update_context} in @file{unwind-ia64.c}.
3474 @var{context} is an @code{_Unwind_Context};
3475 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{fs->unwabi}
3476 for the abi and context in the @code{.unwabi} directive. If the
3477 @code{.unwabi} directive can be handled, the register save addresses should
3478 be updated in @var{fs}.
3481 @defmac TARGET_USES_WEAK_UNWIND_INFO
3482 A C expression that evaluates to true if the target requires unwind
3483 info to be given comdat linkage. Define it to be @code{1} if comdat
3484 linkage is necessary. The default is @code{0}.
3487 @node Stack Checking
3488 @subsection Specifying How Stack Checking is Done
3490 GCC will check that stack references are within the boundaries of the
3491 stack, if the option @option{-fstack-check} is specified, in one of
3496 If the value of the @code{STACK_CHECK_BUILTIN} macro is nonzero, GCC
3497 will assume that you have arranged for full stack checking to be done
3498 at appropriate places in the configuration files. GCC will not do
3499 other special processing.
3502 If @code{STACK_CHECK_BUILTIN} is zero and the value of the
3503 @code{STACK_CHECK_STATIC_BUILTIN} macro is nonzero, GCC will assume
3504 that you have arranged for static stack checking (checking of the
3505 static stack frame of functions) to be done at appropriate places
3506 in the configuration files. GCC will only emit code to do dynamic
3507 stack checking (checking on dynamic stack allocations) using the third
3511 If neither of the above are true, GCC will generate code to periodically
3512 ``probe'' the stack pointer using the values of the macros defined below.
3515 If neither STACK_CHECK_BUILTIN nor STACK_CHECK_STATIC_BUILTIN is defined,
3516 GCC will change its allocation strategy for large objects if the option
3517 @option{-fstack-check} is specified: they will always be allocated
3518 dynamically if their size exceeds @code{STACK_CHECK_MAX_VAR_SIZE} bytes.
3520 @defmac STACK_CHECK_BUILTIN
3521 A nonzero value if stack checking is done by the configuration files in a
3522 machine-dependent manner. You should define this macro if stack checking
3523 is require by the ABI of your machine or if you would like to do stack
3524 checking in some more efficient way than the generic approach. The default
3525 value of this macro is zero.
3528 @defmac STACK_CHECK_STATIC_BUILTIN
3529 A nonzero value if static stack checking is done by the configuration files
3530 in a machine-dependent manner. You should define this macro if you would
3531 like to do static stack checking in some more efficient way than the generic
3532 approach. The default value of this macro is zero.
3535 @defmac STACK_CHECK_PROBE_INTERVAL
3536 An integer representing the interval at which GCC must generate stack
3537 probe instructions. You will normally define this macro to be no larger
3538 than the size of the ``guard pages'' at the end of a stack area. The
3539 default value of 4096 is suitable for most systems.
3542 @defmac STACK_CHECK_PROBE_LOAD
3543 An integer which is nonzero if GCC should perform the stack probe
3544 as a load instruction and zero if GCC should use a store instruction.
3545 The default is zero, which is the most efficient choice on most systems.
3548 @defmac STACK_CHECK_PROTECT
3549 The number of bytes of stack needed to recover from a stack overflow,
3550 for languages where such a recovery is supported. The default value of
3551 75 words should be adequate for most machines.
3554 The following macros are relevant only if neither STACK_CHECK_BUILTIN
3555 nor STACK_CHECK_STATIC_BUILTIN is defined; you can omit them altogether
3556 in the opposite case.
3558 @defmac STACK_CHECK_MAX_FRAME_SIZE
3559 The maximum size of a stack frame, in bytes. GCC will generate probe
3560 instructions in non-leaf functions to ensure at least this many bytes of
3561 stack are available. If a stack frame is larger than this size, stack
3562 checking will not be reliable and GCC will issue a warning. The
3563 default is chosen so that GCC only generates one instruction on most
3564 systems. You should normally not change the default value of this macro.
3567 @defmac STACK_CHECK_FIXED_FRAME_SIZE
3568 GCC uses this value to generate the above warning message. It
3569 represents the amount of fixed frame used by a function, not including
3570 space for any callee-saved registers, temporaries and user variables.
3571 You need only specify an upper bound for this amount and will normally
3572 use the default of four words.
3575 @defmac STACK_CHECK_MAX_VAR_SIZE
3576 The maximum size, in bytes, of an object that GCC will place in the
3577 fixed area of the stack frame when the user specifies
3578 @option{-fstack-check}.
3579 GCC computed the default from the values of the above macros and you will
3580 normally not need to override that default.
3584 @node Frame Registers
3585 @subsection Registers That Address the Stack Frame
3587 @c prevent bad page break with this line
3588 This discusses registers that address the stack frame.
3590 @defmac STACK_POINTER_REGNUM
3591 The register number of the stack pointer register, which must also be a
3592 fixed register according to @code{FIXED_REGISTERS}. On most machines,
3593 the hardware determines which register this is.
3596 @defmac FRAME_POINTER_REGNUM
3597 The register number of the frame pointer register, which is used to
3598 access automatic variables in the stack frame. On some machines, the
3599 hardware determines which register this is. On other machines, you can
3600 choose any register you wish for this purpose.
3603 @defmac HARD_FRAME_POINTER_REGNUM
3604 On some machines the offset between the frame pointer and starting
3605 offset of the automatic variables is not known until after register
3606 allocation has been done (for example, because the saved registers are
3607 between these two locations). On those machines, define
3608 @code{FRAME_POINTER_REGNUM} the number of a special, fixed register to
3609 be used internally until the offset is known, and define
3610 @code{HARD_FRAME_POINTER_REGNUM} to be the actual hard register number
3611 used for the frame pointer.
3613 You should define this macro only in the very rare circumstances when it
3614 is not possible to calculate the offset between the frame pointer and
3615 the automatic variables until after register allocation has been
3616 completed. When this macro is defined, you must also indicate in your
3617 definition of @code{ELIMINABLE_REGS} how to eliminate
3618 @code{FRAME_POINTER_REGNUM} into either @code{HARD_FRAME_POINTER_REGNUM}
3619 or @code{STACK_POINTER_REGNUM}.
3621 Do not define this macro if it would be the same as
3622 @code{FRAME_POINTER_REGNUM}.
3625 @defmac ARG_POINTER_REGNUM
3626 The register number of the arg pointer register, which is used to access
3627 the function's argument list. On some machines, this is the same as the
3628 frame pointer register. On some machines, the hardware determines which
3629 register this is. On other machines, you can choose any register you
3630 wish for this purpose. If this is not the same register as the frame
3631 pointer register, then you must mark it as a fixed register according to
3632 @code{FIXED_REGISTERS}, or arrange to be able to eliminate it
3633 (@pxref{Elimination}).
3636 @defmac RETURN_ADDRESS_POINTER_REGNUM
3637 The register number of the return address pointer register, which is used to
3638 access the current function's return address from the stack. On some
3639 machines, the return address is not at a fixed offset from the frame
3640 pointer or stack pointer or argument pointer. This register can be defined
3641 to point to the return address on the stack, and then be converted by
3642 @code{ELIMINABLE_REGS} into either the frame pointer or stack pointer.
3644 Do not define this macro unless there is no other way to get the return
3645 address from the stack.
3648 @defmac STATIC_CHAIN_REGNUM
3649 @defmacx STATIC_CHAIN_INCOMING_REGNUM
3650 Register numbers used for passing a function's static chain pointer. If
3651 register windows are used, the register number as seen by the called
3652 function is @code{STATIC_CHAIN_INCOMING_REGNUM}, while the register
3653 number as seen by the calling function is @code{STATIC_CHAIN_REGNUM}. If
3654 these registers are the same, @code{STATIC_CHAIN_INCOMING_REGNUM} need
3657 The static chain register need not be a fixed register.
3659 If the static chain is passed in memory, these macros should not be
3660 defined; instead, the next two macros should be defined.
3663 @defmac STATIC_CHAIN
3664 @defmacx STATIC_CHAIN_INCOMING
3665 If the static chain is passed in memory, these macros provide rtx giving
3666 @code{mem} expressions that denote where they are stored.
3667 @code{STATIC_CHAIN} and @code{STATIC_CHAIN_INCOMING} give the locations
3668 as seen by the calling and called functions, respectively. Often the former
3669 will be at an offset from the stack pointer and the latter at an offset from
3672 @findex stack_pointer_rtx
3673 @findex frame_pointer_rtx
3674 @findex arg_pointer_rtx
3675 The variables @code{stack_pointer_rtx}, @code{frame_pointer_rtx}, and
3676 @code{arg_pointer_rtx} will have been initialized prior to the use of these
3677 macros and should be used to refer to those items.
3679 If the static chain is passed in a register, the two previous macros should
3683 @defmac DWARF_FRAME_REGISTERS
3684 This macro specifies the maximum number of hard registers that can be
3685 saved in a call frame. This is used to size data structures used in
3686 DWARF2 exception handling.
3688 Prior to GCC 3.0, this macro was needed in order to establish a stable
3689 exception handling ABI in the face of adding new hard registers for ISA
3690 extensions. In GCC 3.0 and later, the EH ABI is insulated from changes
3691 in the number of hard registers. Nevertheless, this macro can still be
3692 used to reduce the runtime memory requirements of the exception handling
3693 routines, which can be substantial if the ISA contains a lot of
3694 registers that are not call-saved.
3696 If this macro is not defined, it defaults to
3697 @code{FIRST_PSEUDO_REGISTER}.
3700 @defmac PRE_GCC3_DWARF_FRAME_REGISTERS
3702 This macro is similar to @code{DWARF_FRAME_REGISTERS}, but is provided
3703 for backward compatibility in pre GCC 3.0 compiled code.
3705 If this macro is not defined, it defaults to
3706 @code{DWARF_FRAME_REGISTERS}.
3709 @defmac DWARF_REG_TO_UNWIND_COLUMN (@var{regno})
3711 Define this macro if the target's representation for dwarf registers
3712 is different than the internal representation for unwind column.
3713 Given a dwarf register, this macro should return the internal unwind
3714 column number to use instead.
3716 See the PowerPC's SPE target for an example.
3719 @defmac DWARF_FRAME_REGNUM (@var{regno})
3721 Define this macro if the target's representation for dwarf registers
3722 used in .eh_frame or .debug_frame is different from that used in other
3723 debug info sections. Given a GCC hard register number, this macro
3724 should return the .eh_frame register number. The default is
3725 @code{DBX_REGISTER_NUMBER (@var{regno})}.
3729 @defmac DWARF2_FRAME_REG_OUT (@var{regno}, @var{for_eh})
3731 Define this macro to map register numbers held in the call frame info
3732 that GCC has collected using @code{DWARF_FRAME_REGNUM} to those that
3733 should be output in .debug_frame (@code{@var{for_eh}} is zero) and
3734 .eh_frame (@code{@var{for_eh}} is nonzero). The default is to
3735 return @code{@var{regno}}.
3740 @subsection Eliminating Frame Pointer and Arg Pointer
3742 @c prevent bad page break with this line
3743 This is about eliminating the frame pointer and arg pointer.
3745 @deftypefn {Target Hook} bool TARGET_FRAME_POINTER_REQUIRED (void)
3746 This target hook should return @code{true} if a function must have and use
3747 a frame pointer. This target hook is called in the reload pass. If its return
3748 value is @code{true} the function will have a frame pointer.
3750 This target hook can in principle examine the current function and decide
3751 according to the facts, but on most machines the constant @code{false} or the
3752 constant @code{true} suffices. Use @code{false} when the machine allows code
3753 to be generated with no frame pointer, and doing so saves some time or space.
3754 Use @code{true} when there is no possible advantage to avoiding a frame
3757 In certain cases, the compiler does not know how to produce valid code
3758 without a frame pointer. The compiler recognizes those cases and
3759 automatically gives the function a frame pointer regardless of what
3760 @code{TARGET_FRAME_POINTER_REQUIRED} returns. You don't need to worry about
3763 In a function that does not require a frame pointer, the frame pointer
3764 register can be allocated for ordinary usage, unless you mark it as a
3765 fixed register. See @code{FIXED_REGISTERS} for more information.
3767 Default return value is @code{false}.
3770 @findex get_frame_size
3771 @defmac INITIAL_FRAME_POINTER_OFFSET (@var{depth-var})
3772 A C statement to store in the variable @var{depth-var} the difference
3773 between the frame pointer and the stack pointer values immediately after
3774 the function prologue. The value would be computed from information
3775 such as the result of @code{get_frame_size ()} and the tables of
3776 registers @code{regs_ever_live} and @code{call_used_regs}.
3778 If @code{ELIMINABLE_REGS} is defined, this macro will be not be used and
3779 need not be defined. Otherwise, it must be defined even if
3780 @code{TARGET_FRAME_POINTER_REQUIRED} is always return true; in that
3781 case, you may set @var{depth-var} to anything.
3784 @defmac ELIMINABLE_REGS
3785 If defined, this macro specifies a table of register pairs used to
3786 eliminate unneeded registers that point into the stack frame. If it is not
3787 defined, the only elimination attempted by the compiler is to replace
3788 references to the frame pointer with references to the stack pointer.
3790 The definition of this macro is a list of structure initializations, each
3791 of which specifies an original and replacement register.
3793 On some machines, the position of the argument pointer is not known until
3794 the compilation is completed. In such a case, a separate hard register
3795 must be used for the argument pointer. This register can be eliminated by
3796 replacing it with either the frame pointer or the argument pointer,
3797 depending on whether or not the frame pointer has been eliminated.
3799 In this case, you might specify:
3801 #define ELIMINABLE_REGS \
3802 @{@{ARG_POINTER_REGNUM, STACK_POINTER_REGNUM@}, \
3803 @{ARG_POINTER_REGNUM, FRAME_POINTER_REGNUM@}, \
3804 @{FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM@}@}
3807 Note that the elimination of the argument pointer with the stack pointer is
3808 specified first since that is the preferred elimination.
3811 @defmac CAN_ELIMINATE (@var{from-reg}, @var{to-reg})
3812 A C expression that returns @code{true} if the compiler is allowed to try
3813 to replace register number @var{from-reg} with register number
3814 @var{to-reg}. This macro need only be defined if @code{ELIMINABLE_REGS}
3815 is defined, and will usually be @code{true}, since most of the cases
3816 preventing register elimination are things that the compiler already
3819 Default value is @code{true}.
3822 @defmac INITIAL_ELIMINATION_OFFSET (@var{from-reg}, @var{to-reg}, @var{offset-var})
3823 This macro is similar to @code{INITIAL_FRAME_POINTER_OFFSET}. It
3824 specifies the initial difference between the specified pair of
3825 registers. This macro must be defined if @code{ELIMINABLE_REGS} is
3829 @node Stack Arguments
3830 @subsection Passing Function Arguments on the Stack
3831 @cindex arguments on stack
3832 @cindex stack arguments
3834 The macros in this section control how arguments are passed
3835 on the stack. See the following section for other macros that
3836 control passing certain arguments in registers.
3838 @deftypefn {Target Hook} bool TARGET_PROMOTE_PROTOTYPES (tree @var{fntype})
3839 This target hook returns @code{true} if an argument declared in a
3840 prototype as an integral type smaller than @code{int} should actually be
3841 passed as an @code{int}. In addition to avoiding errors in certain
3842 cases of mismatch, it also makes for better code on certain machines.
3843 The default is to not promote prototypes.
3847 A C expression. If nonzero, push insns will be used to pass
3849 If the target machine does not have a push instruction, set it to zero.
3850 That directs GCC to use an alternate strategy: to
3851 allocate the entire argument block and then store the arguments into
3852 it. When @code{PUSH_ARGS} is nonzero, @code{PUSH_ROUNDING} must be defined too.
3855 @defmac PUSH_ARGS_REVERSED
3856 A C expression. If nonzero, function arguments will be evaluated from
3857 last to first, rather than from first to last. If this macro is not
3858 defined, it defaults to @code{PUSH_ARGS} on targets where the stack
3859 and args grow in opposite directions, and 0 otherwise.
3862 @defmac PUSH_ROUNDING (@var{npushed})
3863 A C expression that is the number of bytes actually pushed onto the
3864 stack when an instruction attempts to push @var{npushed} bytes.
3866 On some machines, the definition
3869 #define PUSH_ROUNDING(BYTES) (BYTES)
3873 will suffice. But on other machines, instructions that appear
3874 to push one byte actually push two bytes in an attempt to maintain
3875 alignment. Then the definition should be
3878 #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1)
3882 @findex current_function_outgoing_args_size
3883 @defmac ACCUMULATE_OUTGOING_ARGS
3884 A C expression. If nonzero, the maximum amount of space required for outgoing arguments
3885 will be computed and placed into the variable
3886 @code{current_function_outgoing_args_size}. No space will be pushed
3887 onto the stack for each call; instead, the function prologue should
3888 increase the stack frame size by this amount.
3890 Setting both @code{PUSH_ARGS} and @code{ACCUMULATE_OUTGOING_ARGS}
3894 @defmac REG_PARM_STACK_SPACE (@var{fndecl})
3895 Define this macro if functions should assume that stack space has been
3896 allocated for arguments even when their values are passed in
3899 The value of this macro is the size, in bytes, of the area reserved for
3900 arguments passed in registers for the function represented by @var{fndecl},
3901 which can be zero if GCC is calling a library function.
3902 The argument @var{fndecl} can be the FUNCTION_DECL, or the type itself
3905 This space can be allocated by the caller, or be a part of the
3906 machine-dependent stack frame: @code{OUTGOING_REG_PARM_STACK_SPACE} says
3909 @c above is overfull. not sure what to do. --mew 5feb93 did
3910 @c something, not sure if it looks good. --mew 10feb93
3912 @defmac OUTGOING_REG_PARM_STACK_SPACE (@var{fntype})
3913 Define this to a nonzero value if it is the responsibility of the
3914 caller to allocate the area reserved for arguments passed in registers
3915 when calling a function of @var{fntype}. @var{fntype} may be NULL
3916 if the function called is a library function.
3918 If @code{ACCUMULATE_OUTGOING_ARGS} is defined, this macro controls
3919 whether the space for these arguments counts in the value of
3920 @code{current_function_outgoing_args_size}.
3923 @defmac STACK_PARMS_IN_REG_PARM_AREA
3924 Define this macro if @code{REG_PARM_STACK_SPACE} is defined, but the
3925 stack parameters don't skip the area specified by it.
3926 @c i changed this, makes more sens and it should have taken care of the
3927 @c overfull.. not as specific, tho. --mew 5feb93
3929 Normally, when a parameter is not passed in registers, it is placed on the
3930 stack beyond the @code{REG_PARM_STACK_SPACE} area. Defining this macro
3931 suppresses this behavior and causes the parameter to be passed on the
3932 stack in its natural location.
3935 @defmac RETURN_POPS_ARGS (@var{fundecl}, @var{funtype}, @var{stack-size})
3936 A C expression that should indicate the number of bytes of its own
3937 arguments that a function pops on returning, or 0 if the
3938 function pops no arguments and the caller must therefore pop them all
3939 after the function returns.
3941 @var{fundecl} is a C variable whose value is a tree node that describes
3942 the function in question. Normally it is a node of type
3943 @code{FUNCTION_DECL} that describes the declaration of the function.
3944 From this you can obtain the @code{DECL_ATTRIBUTES} of the function.
3946 @var{funtype} is a C variable whose value is a tree node that
3947 describes the function in question. Normally it is a node of type
3948 @code{FUNCTION_TYPE} that describes the data type of the function.
3949 From this it is possible to obtain the data types of the value and
3950 arguments (if known).
3952 When a call to a library function is being considered, @var{fundecl}
3953 will contain an identifier node for the library function. Thus, if
3954 you need to distinguish among various library functions, you can do so
3955 by their names. Note that ``library function'' in this context means
3956 a function used to perform arithmetic, whose name is known specially
3957 in the compiler and was not mentioned in the C code being compiled.
3959 @var{stack-size} is the number of bytes of arguments passed on the
3960 stack. If a variable number of bytes is passed, it is zero, and
3961 argument popping will always be the responsibility of the calling function.
3963 On the VAX, all functions always pop their arguments, so the definition
3964 of this macro is @var{stack-size}. On the 68000, using the standard
3965 calling convention, no functions pop their arguments, so the value of
3966 the macro is always 0 in this case. But an alternative calling
3967 convention is available in which functions that take a fixed number of
3968 arguments pop them but other functions (such as @code{printf}) pop
3969 nothing (the caller pops all). When this convention is in use,
3970 @var{funtype} is examined to determine whether a function takes a fixed
3971 number of arguments.
3974 @defmac CALL_POPS_ARGS (@var{cum})
3975 A C expression that should indicate the number of bytes a call sequence
3976 pops off the stack. It is added to the value of @code{RETURN_POPS_ARGS}
3977 when compiling a function call.
3979 @var{cum} is the variable in which all arguments to the called function
3980 have been accumulated.
3982 On certain architectures, such as the SH5, a call trampoline is used
3983 that pops certain registers off the stack, depending on the arguments
3984 that have been passed to the function. Since this is a property of the
3985 call site, not of the called function, @code{RETURN_POPS_ARGS} is not
3989 @node Register Arguments
3990 @subsection Passing Arguments in Registers
3991 @cindex arguments in registers
3992 @cindex registers arguments
3994 This section describes the macros which let you control how various
3995 types of arguments are passed in registers or how they are arranged in
3998 @defmac FUNCTION_ARG (@var{cum}, @var{mode}, @var{type}, @var{named})
3999 A C expression that controls whether a function argument is passed
4000 in a register, and which register.
4002 The arguments are @var{cum}, which summarizes all the previous
4003 arguments; @var{mode}, the machine mode of the argument; @var{type},
4004 the data type of the argument as a tree node or 0 if that is not known
4005 (which happens for C support library functions); and @var{named},
4006 which is 1 for an ordinary argument and 0 for nameless arguments that
4007 correspond to @samp{@dots{}} in the called function's prototype.
4008 @var{type} can be an incomplete type if a syntax error has previously
4011 The value of the expression is usually either a @code{reg} RTX for the
4012 hard register in which to pass the argument, or zero to pass the
4013 argument on the stack.
4015 For machines like the VAX and 68000, where normally all arguments are
4016 pushed, zero suffices as a definition.
4018 The value of the expression can also be a @code{parallel} RTX@. This is
4019 used when an argument is passed in multiple locations. The mode of the
4020 @code{parallel} should be the mode of the entire argument. The
4021 @code{parallel} holds any number of @code{expr_list} pairs; each one
4022 describes where part of the argument is passed. In each
4023 @code{expr_list} the first operand must be a @code{reg} RTX for the hard
4024 register in which to pass this part of the argument, and the mode of the
4025 register RTX indicates how large this part of the argument is. The
4026 second operand of the @code{expr_list} is a @code{const_int} which gives
4027 the offset in bytes into the entire argument of where this part starts.
4028 As a special exception the first @code{expr_list} in the @code{parallel}
4029 RTX may have a first operand of zero. This indicates that the entire
4030 argument is also stored on the stack.
4032 The last time this macro is called, it is called with @code{MODE ==
4033 VOIDmode}, and its result is passed to the @code{call} or @code{call_value}
4034 pattern as operands 2 and 3 respectively.
4036 @cindex @file{stdarg.h} and register arguments
4037 The usual way to make the ISO library @file{stdarg.h} work on a machine
4038 where some arguments are usually passed in registers, is to cause
4039 nameless arguments to be passed on the stack instead. This is done
4040 by making @code{FUNCTION_ARG} return 0 whenever @var{named} is 0.
4042 @cindex @code{TARGET_MUST_PASS_IN_STACK}, and @code{FUNCTION_ARG}
4043 @cindex @code{REG_PARM_STACK_SPACE}, and @code{FUNCTION_ARG}
4044 You may use the hook @code{targetm.calls.must_pass_in_stack}
4045 in the definition of this macro to determine if this argument is of a
4046 type that must be passed in the stack. If @code{REG_PARM_STACK_SPACE}
4047 is not defined and @code{FUNCTION_ARG} returns nonzero for such an
4048 argument, the compiler will abort. If @code{REG_PARM_STACK_SPACE} is
4049 defined, the argument will be computed in the stack and then loaded into
4053 @deftypefn {Target Hook} bool TARGET_MUST_PASS_IN_STACK (enum machine_mode @var{mode}, tree @var{type})
4054 This target hook should return @code{true} if we should not pass @var{type}
4055 solely in registers. The file @file{expr.h} defines a
4056 definition that is usually appropriate, refer to @file{expr.h} for additional
4060 @defmac FUNCTION_INCOMING_ARG (@var{cum}, @var{mode}, @var{type}, @var{named})
4061 Define this macro if the target machine has ``register windows'', so
4062 that the register in which a function sees an arguments is not
4063 necessarily the same as the one in which the caller passed the
4066 For such machines, @code{FUNCTION_ARG} computes the register in which
4067 the caller passes the value, and @code{FUNCTION_INCOMING_ARG} should
4068 be defined in a similar fashion to tell the function being called
4069 where the arguments will arrive.
4071 If @code{FUNCTION_INCOMING_ARG} is not defined, @code{FUNCTION_ARG}
4072 serves both purposes.
4075 @deftypefn {Target Hook} int TARGET_ARG_PARTIAL_BYTES (CUMULATIVE_ARGS *@var{cum}, enum machine_mode @var{mode}, tree @var{type}, bool @var{named})
4076 This target hook returns the number of bytes at the beginning of an
4077 argument that must be put in registers. The value must be zero for
4078 arguments that are passed entirely in registers or that are entirely
4079 pushed on the stack.
4081 On some machines, certain arguments must be passed partially in
4082 registers and partially in memory. On these machines, typically the
4083 first few words of arguments are passed in registers, and the rest
4084 on the stack. If a multi-word argument (a @code{double} or a
4085 structure) crosses that boundary, its first few words must be passed
4086 in registers and the rest must be pushed. This macro tells the
4087 compiler when this occurs, and how many bytes should go in registers.
4089 @code{FUNCTION_ARG} for these arguments should return the first
4090 register to be used by the caller for this argument; likewise
4091 @code{FUNCTION_INCOMING_ARG}, for the called function.
4094 @deftypefn {Target Hook} bool TARGET_PASS_BY_REFERENCE (CUMULATIVE_ARGS *@var{cum}, enum machine_mode @var{mode}, tree @var{type}, bool @var{named})
4095 This target hook should return @code{true} if an argument at the
4096 position indicated by @var{cum} should be passed by reference. This
4097 predicate is queried after target independent reasons for being
4098 passed by reference, such as @code{TREE_ADDRESSABLE (type)}.
4100 If the hook returns true, a copy of that argument is made in memory and a
4101 pointer to the argument is passed instead of the argument itself.
4102 The pointer is passed in whatever way is appropriate for passing a pointer
4106 @deftypefn {Target Hook} bool TARGET_CALLEE_COPIES (CUMULATIVE_ARGS *@var{cum}, enum machine_mode @var{mode}, tree @var{type}, bool @var{named})
4107 The function argument described by the parameters to this hook is
4108 known to be passed by reference. The hook should return true if the
4109 function argument should be copied by the callee instead of copied
4112 For any argument for which the hook returns true, if it can be
4113 determined that the argument is not modified, then a copy need
4116 The default version of this hook always returns false.
4119 @defmac CUMULATIVE_ARGS
4120 A C type for declaring a variable that is used as the first argument of
4121 @code{FUNCTION_ARG} and other related values. For some target machines,
4122 the type @code{int} suffices and can hold the number of bytes of
4125 There is no need to record in @code{CUMULATIVE_ARGS} anything about the
4126 arguments that have been passed on the stack. The compiler has other
4127 variables to keep track of that. For target machines on which all
4128 arguments are passed on the stack, there is no need to store anything in
4129 @code{CUMULATIVE_ARGS}; however, the data structure must exist and
4130 should not be empty, so use @code{int}.
4133 @defmac OVERRIDE_ABI_FORMAT (@var{fndecl})
4134 If defined, this macro is called before generating any code for a
4135 function, but after the @var{cfun} descriptor for the function has been
4136 created. The back end may use this macro to update @var{cfun} to
4137 reflect an ABI other than that which would normally be used by default.
4138 If the compiler is generating code for a compiler-generated function,
4139 @var{fndecl} may be @code{NULL}.
4142 @defmac INIT_CUMULATIVE_ARGS (@var{cum}, @var{fntype}, @var{libname}, @var{fndecl}, @var{n_named_args})
4143 A C statement (sans semicolon) for initializing the variable
4144 @var{cum} for the state at the beginning of the argument list. The
4145 variable has type @code{CUMULATIVE_ARGS}. The value of @var{fntype}
4146 is the tree node for the data type of the function which will receive
4147 the args, or 0 if the args are to a compiler support library function.
4148 For direct calls that are not libcalls, @var{fndecl} contain the
4149 declaration node of the function. @var{fndecl} is also set when
4150 @code{INIT_CUMULATIVE_ARGS} is used to find arguments for the function
4151 being compiled. @var{n_named_args} is set to the number of named
4152 arguments, including a structure return address if it is passed as a
4153 parameter, when making a call. When processing incoming arguments,
4154 @var{n_named_args} is set to @minus{}1.
4156 When processing a call to a compiler support library function,
4157 @var{libname} identifies which one. It is a @code{symbol_ref} rtx which
4158 contains the name of the function, as a string. @var{libname} is 0 when
4159 an ordinary C function call is being processed. Thus, each time this
4160 macro is called, either @var{libname} or @var{fntype} is nonzero, but
4161 never both of them at once.
4164 @defmac INIT_CUMULATIVE_LIBCALL_ARGS (@var{cum}, @var{mode}, @var{libname})
4165 Like @code{INIT_CUMULATIVE_ARGS} but only used for outgoing libcalls,
4166 it gets a @code{MODE} argument instead of @var{fntype}, that would be
4167 @code{NULL}. @var{indirect} would always be zero, too. If this macro
4168 is not defined, @code{INIT_CUMULATIVE_ARGS (cum, NULL_RTX, libname,
4169 0)} is used instead.
4172 @defmac INIT_CUMULATIVE_INCOMING_ARGS (@var{cum}, @var{fntype}, @var{libname})
4173 Like @code{INIT_CUMULATIVE_ARGS} but overrides it for the purposes of
4174 finding the arguments for the function being compiled. If this macro is
4175 undefined, @code{INIT_CUMULATIVE_ARGS} is used instead.
4177 The value passed for @var{libname} is always 0, since library routines
4178 with special calling conventions are never compiled with GCC@. The
4179 argument @var{libname} exists for symmetry with
4180 @code{INIT_CUMULATIVE_ARGS}.
4181 @c could use "this macro" in place of @code{INIT_CUMULATIVE_ARGS}, maybe.
4182 @c --mew 5feb93 i switched the order of the sentences. --mew 10feb93
4185 @defmac FUNCTION_ARG_ADVANCE (@var{cum}, @var{mode}, @var{type}, @var{named})
4186 A C statement (sans semicolon) to update the summarizer variable
4187 @var{cum} to advance past an argument in the argument list. The
4188 values @var{mode}, @var{type} and @var{named} describe that argument.
4189 Once this is done, the variable @var{cum} is suitable for analyzing
4190 the @emph{following} argument with @code{FUNCTION_ARG}, etc.
4192 This macro need not do anything if the argument in question was passed
4193 on the stack. The compiler knows how to track the amount of stack space
4194 used for arguments without any special help.
4198 @defmac FUNCTION_ARG_OFFSET (@var{mode}, @var{type})
4199 If defined, a C expression that is the number of bytes to add to the
4200 offset of the argument passed in memory. This is needed for the SPU,
4201 which passes @code{char} and @code{short} arguments in the preferred
4202 slot that is in the middle of the quad word instead of starting at the
4206 @defmac FUNCTION_ARG_PADDING (@var{mode}, @var{type})
4207 If defined, a C expression which determines whether, and in which direction,
4208 to pad out an argument with extra space. The value should be of type
4209 @code{enum direction}: either @code{upward} to pad above the argument,
4210 @code{downward} to pad below, or @code{none} to inhibit padding.
4212 The @emph{amount} of padding is always just enough to reach the next
4213 multiple of @code{FUNCTION_ARG_BOUNDARY}; this macro does not control
4216 This macro has a default definition which is right for most systems.
4217 For little-endian machines, the default is to pad upward. For
4218 big-endian machines, the default is to pad downward for an argument of
4219 constant size shorter than an @code{int}, and upward otherwise.
4222 @defmac PAD_VARARGS_DOWN
4223 If defined, a C expression which determines whether the default
4224 implementation of va_arg will attempt to pad down before reading the
4225 next argument, if that argument is smaller than its aligned space as
4226 controlled by @code{PARM_BOUNDARY}. If this macro is not defined, all such
4227 arguments are padded down if @code{BYTES_BIG_ENDIAN} is true.
4230 @defmac BLOCK_REG_PADDING (@var{mode}, @var{type}, @var{first})
4231 Specify padding for the last element of a block move between registers and
4232 memory. @var{first} is nonzero if this is the only element. Defining this
4233 macro allows better control of register function parameters on big-endian
4234 machines, without using @code{PARALLEL} rtl. In particular,
4235 @code{MUST_PASS_IN_STACK} need not test padding and mode of types in
4236 registers, as there is no longer a "wrong" part of a register; For example,
4237 a three byte aggregate may be passed in the high part of a register if so
4241 @defmac FUNCTION_ARG_BOUNDARY (@var{mode}, @var{type})
4242 If defined, a C expression that gives the alignment boundary, in bits,
4243 of an argument with the specified mode and type. If it is not defined,
4244 @code{PARM_BOUNDARY} is used for all arguments.
4247 @defmac FUNCTION_ARG_REGNO_P (@var{regno})
4248 A C expression that is nonzero if @var{regno} is the number of a hard
4249 register in which function arguments are sometimes passed. This does
4250 @emph{not} include implicit arguments such as the static chain and
4251 the structure-value address. On many machines, no registers can be
4252 used for this purpose since all function arguments are pushed on the
4256 @deftypefn {Target Hook} bool TARGET_SPLIT_COMPLEX_ARG (tree @var{type})
4257 This hook should return true if parameter of type @var{type} are passed
4258 as two scalar parameters. By default, GCC will attempt to pack complex
4259 arguments into the target's word size. Some ABIs require complex arguments
4260 to be split and treated as their individual components. For example, on
4261 AIX64, complex floats should be passed in a pair of floating point
4262 registers, even though a complex float would fit in one 64-bit floating
4265 The default value of this hook is @code{NULL}, which is treated as always
4269 @deftypefn {Target Hook} tree TARGET_BUILD_BUILTIN_VA_LIST (void)
4270 This hook returns a type node for @code{va_list} for the target.
4271 The default version of the hook returns @code{void*}.
4274 @deftypefn {Target Hook} tree TARGET_FN_ABI_VA_LIST (tree @var{fndecl})
4275 This hook returns the va_list type of the calling convention specified by
4277 The default version of this hook returns @code{va_list_type_node}.
4280 @deftypefn {Target Hook} tree TARGET_CANONICAL_VA_LIST_TYPE (tree @var{type})
4281 This hook returns the va_list type of the calling convention specified by the
4282 type of @var{type}. If @var{type} is not a valid va_list type, it returns
4286 @deftypefn {Target Hook} tree TARGET_GIMPLIFY_VA_ARG_EXPR (tree @var{valist}, tree @var{type}, tree *@var{pre_p}, tree *@var{post_p})
4287 This hook performs target-specific gimplification of
4288 @code{VA_ARG_EXPR}. The first two parameters correspond to the
4289 arguments to @code{va_arg}; the latter two are as in
4290 @code{gimplify.c:gimplify_expr}.
4293 @deftypefn {Target Hook} bool TARGET_VALID_POINTER_MODE (enum machine_mode @var{mode})
4294 Define this to return nonzero if the port can handle pointers
4295 with machine mode @var{mode}. The default version of this
4296 hook returns true for both @code{ptr_mode} and @code{Pmode}.
4299 @deftypefn {Target Hook} bool TARGET_SCALAR_MODE_SUPPORTED_P (enum machine_mode @var{mode})
4300 Define this to return nonzero if the port is prepared to handle
4301 insns involving scalar mode @var{mode}. For a scalar mode to be
4302 considered supported, all the basic arithmetic and comparisons
4305 The default version of this hook returns true for any mode
4306 required to handle the basic C types (as defined by the port).
4307 Included here are the double-word arithmetic supported by the
4308 code in @file{optabs.c}.
4311 @deftypefn {Target Hook} bool TARGET_VECTOR_MODE_SUPPORTED_P (enum machine_mode @var{mode})
4312 Define this to return nonzero if the port is prepared to handle
4313 insns involving vector mode @var{mode}. At the very least, it
4314 must have move patterns for this mode.
4318 @subsection How Scalar Function Values Are Returned
4319 @cindex return values in registers
4320 @cindex values, returned by functions
4321 @cindex scalars, returned as values
4323 This section discusses the macros that control returning scalars as
4324 values---values that can fit in registers.
4326 @deftypefn {Target Hook} rtx TARGET_FUNCTION_VALUE (tree @var{ret_type}, tree @var{fn_decl_or_type}, bool @var{outgoing})
4328 Define this to return an RTX representing the place where a function
4329 returns or receives a value of data type @var{ret_type}, a tree node
4330 representing a data type. @var{fn_decl_or_type} is a tree node
4331 representing @code{FUNCTION_DECL} or @code{FUNCTION_TYPE} of a
4332 function being called. If @var{outgoing} is false, the hook should
4333 compute the register in which the caller will see the return value.
4334 Otherwise, the hook should return an RTX representing the place where
4335 a function returns a value.
4337 On many machines, only @code{TYPE_MODE (@var{ret_type})} is relevant.
4338 (Actually, on most machines, scalar values are returned in the same
4339 place regardless of mode.) The value of the expression is usually a
4340 @code{reg} RTX for the hard register where the return value is stored.
4341 The value can also be a @code{parallel} RTX, if the return value is in
4342 multiple places. See @code{FUNCTION_ARG} for an explanation of the
4343 @code{parallel} form. Note that the callee will populate every
4344 location specified in the @code{parallel}, but if the first element of
4345 the @code{parallel} contains the whole return value, callers will use
4346 that element as the canonical location and ignore the others. The m68k
4347 port uses this type of @code{parallel} to return pointers in both
4348 @samp{%a0} (the canonical location) and @samp{%d0}.
4350 If @code{TARGET_PROMOTE_FUNCTION_RETURN} returns true, you must apply
4351 the same promotion rules specified in @code{PROMOTE_MODE} if
4352 @var{valtype} is a scalar type.
4354 If the precise function being called is known, @var{func} is a tree
4355 node (@code{FUNCTION_DECL}) for it; otherwise, @var{func} is a null
4356 pointer. This makes it possible to use a different value-returning
4357 convention for specific functions when all their calls are
4360 Some target machines have ``register windows'' so that the register in
4361 which a function returns its value is not the same as the one in which
4362 the caller sees the value. For such machines, you should return
4363 different RTX depending on @var{outgoing}.
4365 @code{TARGET_FUNCTION_VALUE} is not used for return values with
4366 aggregate data types, because these are returned in another way. See
4367 @code{TARGET_STRUCT_VALUE_RTX} and related macros, below.
4370 @defmac FUNCTION_VALUE (@var{valtype}, @var{func})
4371 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE} for
4372 a new target instead.
4375 @defmac FUNCTION_OUTGOING_VALUE (@var{valtype}, @var{func})
4376 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE} for
4377 a new target instead.
4380 @defmac LIBCALL_VALUE (@var{mode})
4381 A C expression to create an RTX representing the place where a library
4382 function returns a value of mode @var{mode}.
4384 Note that ``library function'' in this context means a compiler
4385 support routine, used to perform arithmetic, whose name is known
4386 specially by the compiler and was not mentioned in the C code being
4390 @deftypefn {Target Hook} rtx TARGET_LIBCALL_VALUE (enum machine_mode
4391 @var{mode}, rtx @var{fun})
4392 Define this hook if the back-end needs to know the name of the libcall
4393 function in order to determine where the result should be returned.
4395 The mode of the result is given by @var{mode} and the name of the called
4396 library function is given by @var{fun}. The hook should return an RTX
4397 representing the place where the library function result will be returned.
4399 If this hook is not defined, then LIBCALL_VALUE will be used.
4402 @defmac FUNCTION_VALUE_REGNO_P (@var{regno})
4403 A C expression that is nonzero if @var{regno} is the number of a hard
4404 register in which the values of called function may come back.
4406 A register whose use for returning values is limited to serving as the
4407 second of a pair (for a value of type @code{double}, say) need not be
4408 recognized by this macro. So for most machines, this definition
4412 #define FUNCTION_VALUE_REGNO_P(N) ((N) == 0)
4415 If the machine has register windows, so that the caller and the called
4416 function use different registers for the return value, this macro
4417 should recognize only the caller's register numbers.
4420 @defmac TARGET_ENUM_VA_LIST (@var{idx}, @var{pname}, @var{ptype})
4421 This target macro is used in function @code{c_common_nodes_and_builtins}
4422 to iterate through the target specific builtin types for va_list. The
4423 variable @var{idx} is used as iterator. @var{pname} has to be a pointer
4424 to a @code{const char *} and @var{ptype} a pointer to a @code{tree} typed
4426 The arguments @var{pname} and @var{ptype} are used to store the result of
4427 this macro and are set to the name of the va_list builtin type and its
4429 If the return value of this macro is zero, then there is no more element.
4430 Otherwise the @var{IDX} should be increased for the next call of this
4431 macro to iterate through all types.
4434 @defmac APPLY_RESULT_SIZE
4435 Define this macro if @samp{untyped_call} and @samp{untyped_return}
4436 need more space than is implied by @code{FUNCTION_VALUE_REGNO_P} for
4437 saving and restoring an arbitrary return value.
4440 @deftypefn {Target Hook} bool TARGET_RETURN_IN_MSB (tree @var{type})
4441 This hook should return true if values of type @var{type} are returned
4442 at the most significant end of a register (in other words, if they are
4443 padded at the least significant end). You can assume that @var{type}
4444 is returned in a register; the caller is required to check this.
4446 Note that the register provided by @code{TARGET_FUNCTION_VALUE} must
4447 be able to hold the complete return value. For example, if a 1-, 2-
4448 or 3-byte structure is returned at the most significant end of a
4449 4-byte register, @code{TARGET_FUNCTION_VALUE} should provide an
4453 @node Aggregate Return
4454 @subsection How Large Values Are Returned
4455 @cindex aggregates as return values
4456 @cindex large return values
4457 @cindex returning aggregate values
4458 @cindex structure value address
4460 When a function value's mode is @code{BLKmode} (and in some other
4461 cases), the value is not returned according to
4462 @code{TARGET_FUNCTION_VALUE} (@pxref{Scalar Return}). Instead, the
4463 caller passes the address of a block of memory in which the value
4464 should be stored. This address is called the @dfn{structure value
4467 This section describes how to control returning structure values in
4470 @deftypefn {Target Hook} bool TARGET_RETURN_IN_MEMORY (tree @var{type}, tree @var{fntype})
4471 This target hook should return a nonzero value to say to return the
4472 function value in memory, just as large structures are always returned.
4473 Here @var{type} will be the data type of the value, and @var{fntype}
4474 will be the type of the function doing the returning, or @code{NULL} for
4477 Note that values of mode @code{BLKmode} must be explicitly handled
4478 by this function. Also, the option @option{-fpcc-struct-return}
4479 takes effect regardless of this macro. On most systems, it is
4480 possible to leave the hook undefined; this causes a default
4481 definition to be used, whose value is the constant 1 for @code{BLKmode}
4482 values, and 0 otherwise.
4484 Do not use this hook to indicate that structures and unions should always
4485 be returned in memory. You should instead use @code{DEFAULT_PCC_STRUCT_RETURN}
4489 @defmac DEFAULT_PCC_STRUCT_RETURN
4490 Define this macro to be 1 if all structure and union return values must be
4491 in memory. Since this results in slower code, this should be defined
4492 only if needed for compatibility with other compilers or with an ABI@.
4493 If you define this macro to be 0, then the conventions used for structure
4494 and union return values are decided by the @code{TARGET_RETURN_IN_MEMORY}
4497 If not defined, this defaults to the value 1.
4500 @deftypefn {Target Hook} rtx TARGET_STRUCT_VALUE_RTX (tree @var{fndecl}, int @var{incoming})
4501 This target hook should return the location of the structure value
4502 address (normally a @code{mem} or @code{reg}), or 0 if the address is
4503 passed as an ``invisible'' first argument. Note that @var{fndecl} may
4504 be @code{NULL}, for libcalls. You do not need to define this target
4505 hook if the address is always passed as an ``invisible'' first
4508 On some architectures the place where the structure value address
4509 is found by the called function is not the same place that the
4510 caller put it. This can be due to register windows, or it could
4511 be because the function prologue moves it to a different place.
4512 @var{incoming} is @code{1} or @code{2} when the location is needed in
4513 the context of the called function, and @code{0} in the context of
4516 If @var{incoming} is nonzero and the address is to be found on the
4517 stack, return a @code{mem} which refers to the frame pointer. If
4518 @var{incoming} is @code{2}, the result is being used to fetch the
4519 structure value address at the beginning of a function. If you need
4520 to emit adjusting code, you should do it at this point.
4523 @defmac PCC_STATIC_STRUCT_RETURN
4524 Define this macro if the usual system convention on the target machine
4525 for returning structures and unions is for the called function to return
4526 the address of a static variable containing the value.
4528 Do not define this if the usual system convention is for the caller to
4529 pass an address to the subroutine.
4531 This macro has effect in @option{-fpcc-struct-return} mode, but it does
4532 nothing when you use @option{-freg-struct-return} mode.
4536 @subsection Caller-Saves Register Allocation
4538 If you enable it, GCC can save registers around function calls. This
4539 makes it possible to use call-clobbered registers to hold variables that
4540 must live across calls.
4542 @defmac CALLER_SAVE_PROFITABLE (@var{refs}, @var{calls})
4543 A C expression to determine whether it is worthwhile to consider placing
4544 a pseudo-register in a call-clobbered hard register and saving and
4545 restoring it around each function call. The expression should be 1 when
4546 this is worth doing, and 0 otherwise.
4548 If you don't define this macro, a default is used which is good on most
4549 machines: @code{4 * @var{calls} < @var{refs}}.
4552 @defmac HARD_REGNO_CALLER_SAVE_MODE (@var{regno}, @var{nregs})
4553 A C expression specifying which mode is required for saving @var{nregs}
4554 of a pseudo-register in call-clobbered hard register @var{regno}. If
4555 @var{regno} is unsuitable for caller save, @code{VOIDmode} should be
4556 returned. For most machines this macro need not be defined since GCC
4557 will select the smallest suitable mode.
4560 @node Function Entry
4561 @subsection Function Entry and Exit
4562 @cindex function entry and exit
4566 This section describes the macros that output function entry
4567 (@dfn{prologue}) and exit (@dfn{epilogue}) code.
4569 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_PROLOGUE (FILE *@var{file}, HOST_WIDE_INT @var{size})
4570 If defined, a function that outputs the assembler code for entry to a
4571 function. The prologue is responsible for setting up the stack frame,
4572 initializing the frame pointer register, saving registers that must be
4573 saved, and allocating @var{size} additional bytes of storage for the
4574 local variables. @var{size} is an integer. @var{file} is a stdio
4575 stream to which the assembler code should be output.
4577 The label for the beginning of the function need not be output by this
4578 macro. That has already been done when the macro is run.
4580 @findex regs_ever_live
4581 To determine which registers to save, the macro can refer to the array
4582 @code{regs_ever_live}: element @var{r} is nonzero if hard register
4583 @var{r} is used anywhere within the function. This implies the function
4584 prologue should save register @var{r}, provided it is not one of the
4585 call-used registers. (@code{TARGET_ASM_FUNCTION_EPILOGUE} must likewise use
4586 @code{regs_ever_live}.)
4588 On machines that have ``register windows'', the function entry code does
4589 not save on the stack the registers that are in the windows, even if
4590 they are supposed to be preserved by function calls; instead it takes
4591 appropriate steps to ``push'' the register stack, if any non-call-used
4592 registers are used in the function.
4594 @findex frame_pointer_needed
4595 On machines where functions may or may not have frame-pointers, the
4596 function entry code must vary accordingly; it must set up the frame
4597 pointer if one is wanted, and not otherwise. To determine whether a
4598 frame pointer is in wanted, the macro can refer to the variable
4599 @code{frame_pointer_needed}. The variable's value will be 1 at run
4600 time in a function that needs a frame pointer. @xref{Elimination}.
4602 The function entry code is responsible for allocating any stack space
4603 required for the function. This stack space consists of the regions
4604 listed below. In most cases, these regions are allocated in the
4605 order listed, with the last listed region closest to the top of the
4606 stack (the lowest address if @code{STACK_GROWS_DOWNWARD} is defined, and
4607 the highest address if it is not defined). You can use a different order
4608 for a machine if doing so is more convenient or required for
4609 compatibility reasons. Except in cases where required by standard
4610 or by a debugger, there is no reason why the stack layout used by GCC
4611 need agree with that used by other compilers for a machine.
4614 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_END_PROLOGUE (FILE *@var{file})
4615 If defined, a function that outputs assembler code at the end of a
4616 prologue. This should be used when the function prologue is being
4617 emitted as RTL, and you have some extra assembler that needs to be
4618 emitted. @xref{prologue instruction pattern}.
4621 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_BEGIN_EPILOGUE (FILE *@var{file})
4622 If defined, a function that outputs assembler code at the start of an
4623 epilogue. This should be used when the function epilogue is being
4624 emitted as RTL, and you have some extra assembler that needs to be
4625 emitted. @xref{epilogue instruction pattern}.
4628 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_EPILOGUE (FILE *@var{file}, HOST_WIDE_INT @var{size})
4629 If defined, a function that outputs the assembler code for exit from a
4630 function. The epilogue is responsible for restoring the saved
4631 registers and stack pointer to their values when the function was
4632 called, and returning control to the caller. This macro takes the
4633 same arguments as the macro @code{TARGET_ASM_FUNCTION_PROLOGUE}, and the
4634 registers to restore are determined from @code{regs_ever_live} and
4635 @code{CALL_USED_REGISTERS} in the same way.
4637 On some machines, there is a single instruction that does all the work
4638 of returning from the function. On these machines, give that
4639 instruction the name @samp{return} and do not define the macro
4640 @code{TARGET_ASM_FUNCTION_EPILOGUE} at all.
4642 Do not define a pattern named @samp{return} if you want the
4643 @code{TARGET_ASM_FUNCTION_EPILOGUE} to be used. If you want the target
4644 switches to control whether return instructions or epilogues are used,
4645 define a @samp{return} pattern with a validity condition that tests the
4646 target switches appropriately. If the @samp{return} pattern's validity
4647 condition is false, epilogues will be used.
4649 On machines where functions may or may not have frame-pointers, the
4650 function exit code must vary accordingly. Sometimes the code for these
4651 two cases is completely different. To determine whether a frame pointer
4652 is wanted, the macro can refer to the variable
4653 @code{frame_pointer_needed}. The variable's value will be 1 when compiling
4654 a function that needs a frame pointer.
4656 Normally, @code{TARGET_ASM_FUNCTION_PROLOGUE} and
4657 @code{TARGET_ASM_FUNCTION_EPILOGUE} must treat leaf functions specially.
4658 The C variable @code{current_function_is_leaf} is nonzero for such a
4659 function. @xref{Leaf Functions}.
4661 On some machines, some functions pop their arguments on exit while
4662 others leave that for the caller to do. For example, the 68020 when
4663 given @option{-mrtd} pops arguments in functions that take a fixed
4664 number of arguments.
4666 @findex current_function_pops_args
4667 Your definition of the macro @code{RETURN_POPS_ARGS} decides which
4668 functions pop their own arguments. @code{TARGET_ASM_FUNCTION_EPILOGUE}
4669 needs to know what was decided. The variable that is called
4670 @code{current_function_pops_args} is the number of bytes of its
4671 arguments that a function should pop. @xref{Scalar Return}.
4672 @c what is the "its arguments" in the above sentence referring to, pray
4673 @c tell? --mew 5feb93
4678 @findex current_function_pretend_args_size
4679 A region of @code{current_function_pretend_args_size} bytes of
4680 uninitialized space just underneath the first argument arriving on the
4681 stack. (This may not be at the very start of the allocated stack region
4682 if the calling sequence has pushed anything else since pushing the stack
4683 arguments. But usually, on such machines, nothing else has been pushed
4684 yet, because the function prologue itself does all the pushing.) This
4685 region is used on machines where an argument may be passed partly in
4686 registers and partly in memory, and, in some cases to support the
4687 features in @code{<stdarg.h>}.
4690 An area of memory used to save certain registers used by the function.
4691 The size of this area, which may also include space for such things as
4692 the return address and pointers to previous stack frames, is
4693 machine-specific and usually depends on which registers have been used
4694 in the function. Machines with register windows often do not require
4698 A region of at least @var{size} bytes, possibly rounded up to an allocation
4699 boundary, to contain the local variables of the function. On some machines,
4700 this region and the save area may occur in the opposite order, with the
4701 save area closer to the top of the stack.
4704 @cindex @code{ACCUMULATE_OUTGOING_ARGS} and stack frames
4705 Optionally, when @code{ACCUMULATE_OUTGOING_ARGS} is defined, a region of
4706 @code{current_function_outgoing_args_size} bytes to be used for outgoing
4707 argument lists of the function. @xref{Stack Arguments}.
4710 @defmac EXIT_IGNORE_STACK
4711 Define this macro as a C expression that is nonzero if the return
4712 instruction or the function epilogue ignores the value of the stack
4713 pointer; in other words, if it is safe to delete an instruction to
4714 adjust the stack pointer before a return from the function. The
4717 Note that this macro's value is relevant only for functions for which
4718 frame pointers are maintained. It is never safe to delete a final
4719 stack adjustment in a function that has no frame pointer, and the
4720 compiler knows this regardless of @code{EXIT_IGNORE_STACK}.
4723 @defmac EPILOGUE_USES (@var{regno})
4724 Define this macro as a C expression that is nonzero for registers that are
4725 used by the epilogue or the @samp{return} pattern. The stack and frame
4726 pointer registers are already assumed to be used as needed.
4729 @defmac EH_USES (@var{regno})
4730 Define this macro as a C expression that is nonzero for registers that are
4731 used by the exception handling mechanism, and so should be considered live
4732 on entry to an exception edge.
4735 @defmac DELAY_SLOTS_FOR_EPILOGUE
4736 Define this macro if the function epilogue contains delay slots to which
4737 instructions from the rest of the function can be ``moved''. The
4738 definition should be a C expression whose value is an integer
4739 representing the number of delay slots there.
4742 @defmac ELIGIBLE_FOR_EPILOGUE_DELAY (@var{insn}, @var{n})
4743 A C expression that returns 1 if @var{insn} can be placed in delay
4744 slot number @var{n} of the epilogue.
4746 The argument @var{n} is an integer which identifies the delay slot now
4747 being considered (since different slots may have different rules of
4748 eligibility). It is never negative and is always less than the number
4749 of epilogue delay slots (what @code{DELAY_SLOTS_FOR_EPILOGUE} returns).
4750 If you reject a particular insn for a given delay slot, in principle, it
4751 may be reconsidered for a subsequent delay slot. Also, other insns may
4752 (at least in principle) be considered for the so far unfilled delay
4755 @findex current_function_epilogue_delay_list
4756 @findex final_scan_insn
4757 The insns accepted to fill the epilogue delay slots are put in an RTL
4758 list made with @code{insn_list} objects, stored in the variable
4759 @code{current_function_epilogue_delay_list}. The insn for the first
4760 delay slot comes first in the list. Your definition of the macro
4761 @code{TARGET_ASM_FUNCTION_EPILOGUE} should fill the delay slots by
4762 outputting the insns in this list, usually by calling
4763 @code{final_scan_insn}.
4765 You need not define this macro if you did not define
4766 @code{DELAY_SLOTS_FOR_EPILOGUE}.
4769 @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})
4770 A function that outputs the assembler code for a thunk
4771 function, used to implement C++ virtual function calls with multiple
4772 inheritance. The thunk acts as a wrapper around a virtual function,
4773 adjusting the implicit object parameter before handing control off to
4776 First, emit code to add the integer @var{delta} to the location that
4777 contains the incoming first argument. Assume that this argument
4778 contains a pointer, and is the one used to pass the @code{this} pointer
4779 in C++. This is the incoming argument @emph{before} the function prologue,
4780 e.g.@: @samp{%o0} on a sparc. The addition must preserve the values of
4781 all other incoming arguments.
4783 Then, if @var{vcall_offset} is nonzero, an additional adjustment should be
4784 made after adding @code{delta}. In particular, if @var{p} is the
4785 adjusted pointer, the following adjustment should be made:
4788 p += (*((ptrdiff_t **)p))[vcall_offset/sizeof(ptrdiff_t)]
4791 After the additions, emit code to jump to @var{function}, which is a
4792 @code{FUNCTION_DECL}. This is a direct pure jump, not a call, and does
4793 not touch the return address. Hence returning from @var{FUNCTION} will
4794 return to whoever called the current @samp{thunk}.
4796 The effect must be as if @var{function} had been called directly with
4797 the adjusted first argument. This macro is responsible for emitting all
4798 of the code for a thunk function; @code{TARGET_ASM_FUNCTION_PROLOGUE}
4799 and @code{TARGET_ASM_FUNCTION_EPILOGUE} are not invoked.
4801 The @var{thunk_fndecl} is redundant. (@var{delta} and @var{function}
4802 have already been extracted from it.) It might possibly be useful on
4803 some targets, but probably not.
4805 If you do not define this macro, the target-independent code in the C++
4806 front end will generate a less efficient heavyweight thunk that calls
4807 @var{function} instead of jumping to it. The generic approach does
4808 not support varargs.
4811 @deftypefn {Target Hook} bool TARGET_ASM_CAN_OUTPUT_MI_THUNK (tree @var{thunk_fndecl}, HOST_WIDE_INT @var{delta}, HOST_WIDE_INT @var{vcall_offset}, tree @var{function})
4812 A function that returns true if TARGET_ASM_OUTPUT_MI_THUNK would be able
4813 to output the assembler code for the thunk function specified by the
4814 arguments it is passed, and false otherwise. In the latter case, the
4815 generic approach will be used by the C++ front end, with the limitations
4820 @subsection Generating Code for Profiling
4821 @cindex profiling, code generation
4823 These macros will help you generate code for profiling.
4825 @defmac FUNCTION_PROFILER (@var{file}, @var{labelno})
4826 A C statement or compound statement to output to @var{file} some
4827 assembler code to call the profiling subroutine @code{mcount}.
4830 The details of how @code{mcount} expects to be called are determined by
4831 your operating system environment, not by GCC@. To figure them out,
4832 compile a small program for profiling using the system's installed C
4833 compiler and look at the assembler code that results.
4835 Older implementations of @code{mcount} expect the address of a counter
4836 variable to be loaded into some register. The name of this variable is
4837 @samp{LP} followed by the number @var{labelno}, so you would generate
4838 the name using @samp{LP%d} in a @code{fprintf}.
4841 @defmac PROFILE_HOOK
4842 A C statement or compound statement to output to @var{file} some assembly
4843 code to call the profiling subroutine @code{mcount} even the target does
4844 not support profiling.
4847 @defmac NO_PROFILE_COUNTERS
4848 Define this macro to be an expression with a nonzero value if the
4849 @code{mcount} subroutine on your system does not need a counter variable
4850 allocated for each function. This is true for almost all modern
4851 implementations. If you define this macro, you must not use the
4852 @var{labelno} argument to @code{FUNCTION_PROFILER}.
4855 @defmac PROFILE_BEFORE_PROLOGUE
4856 Define this macro if the code for function profiling should come before
4857 the function prologue. Normally, the profiling code comes after.
4861 @subsection Permitting tail calls
4864 @deftypefn {Target Hook} bool TARGET_FUNCTION_OK_FOR_SIBCALL (tree @var{decl}, tree @var{exp})
4865 True if it is ok to do sibling call optimization for the specified
4866 call expression @var{exp}. @var{decl} will be the called function,
4867 or @code{NULL} if this is an indirect call.
4869 It is not uncommon for limitations of calling conventions to prevent
4870 tail calls to functions outside the current unit of translation, or
4871 during PIC compilation. The hook is used to enforce these restrictions,
4872 as the @code{sibcall} md pattern can not fail, or fall over to a
4873 ``normal'' call. The criteria for successful sibling call optimization
4874 may vary greatly between different architectures.
4877 @deftypefn {Target Hook} void TARGET_EXTRA_LIVE_ON_ENTRY (bitmap *@var{regs})
4878 Add any hard registers to @var{regs} that are live on entry to the
4879 function. This hook only needs to be defined to provide registers that
4880 cannot be found by examination of FUNCTION_ARG_REGNO_P, the callee saved
4881 registers, STATIC_CHAIN_INCOMING_REGNUM, STATIC_CHAIN_REGNUM,
4882 TARGET_STRUCT_VALUE_RTX, FRAME_POINTER_REGNUM, EH_USES,
4883 FRAME_POINTER_REGNUM, ARG_POINTER_REGNUM, and the PIC_OFFSET_TABLE_REGNUM.
4886 @node Stack Smashing Protection
4887 @subsection Stack smashing protection
4888 @cindex stack smashing protection
4890 @deftypefn {Target Hook} tree TARGET_STACK_PROTECT_GUARD (void)
4891 This hook returns a @code{DECL} node for the external variable to use
4892 for the stack protection guard. This variable is initialized by the
4893 runtime to some random value and is used to initialize the guard value
4894 that is placed at the top of the local stack frame. The type of this
4895 variable must be @code{ptr_type_node}.
4897 The default version of this hook creates a variable called
4898 @samp{__stack_chk_guard}, which is normally defined in @file{libgcc2.c}.
4901 @deftypefn {Target Hook} tree TARGET_STACK_PROTECT_FAIL (void)
4902 This hook returns a tree expression that alerts the runtime that the
4903 stack protect guard variable has been modified. This expression should
4904 involve a call to a @code{noreturn} function.
4906 The default version of this hook invokes a function called
4907 @samp{__stack_chk_fail}, taking no arguments. This function is
4908 normally defined in @file{libgcc2.c}.
4912 @section Implementing the Varargs Macros
4913 @cindex varargs implementation
4915 GCC comes with an implementation of @code{<varargs.h>} and
4916 @code{<stdarg.h>} that work without change on machines that pass arguments
4917 on the stack. Other machines require their own implementations of
4918 varargs, and the two machine independent header files must have
4919 conditionals to include it.
4921 ISO @code{<stdarg.h>} differs from traditional @code{<varargs.h>} mainly in
4922 the calling convention for @code{va_start}. The traditional
4923 implementation takes just one argument, which is the variable in which
4924 to store the argument pointer. The ISO implementation of
4925 @code{va_start} takes an additional second argument. The user is
4926 supposed to write the last named argument of the function here.
4928 However, @code{va_start} should not use this argument. The way to find
4929 the end of the named arguments is with the built-in functions described
4932 @defmac __builtin_saveregs ()
4933 Use this built-in function to save the argument registers in memory so
4934 that the varargs mechanism can access them. Both ISO and traditional
4935 versions of @code{va_start} must use @code{__builtin_saveregs}, unless
4936 you use @code{TARGET_SETUP_INCOMING_VARARGS} (see below) instead.
4938 On some machines, @code{__builtin_saveregs} is open-coded under the
4939 control of the target hook @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. On
4940 other machines, it calls a routine written in assembler language,
4941 found in @file{libgcc2.c}.
4943 Code generated for the call to @code{__builtin_saveregs} appears at the
4944 beginning of the function, as opposed to where the call to
4945 @code{__builtin_saveregs} is written, regardless of what the code is.
4946 This is because the registers must be saved before the function starts
4947 to use them for its own purposes.
4948 @c i rewrote the first sentence above to fix an overfull hbox. --mew
4952 @defmac __builtin_args_info (@var{category})
4953 Use this built-in function to find the first anonymous arguments in
4956 In general, a machine may have several categories of registers used for
4957 arguments, each for a particular category of data types. (For example,
4958 on some machines, floating-point registers are used for floating-point
4959 arguments while other arguments are passed in the general registers.)
4960 To make non-varargs functions use the proper calling convention, you
4961 have defined the @code{CUMULATIVE_ARGS} data type to record how many
4962 registers in each category have been used so far
4964 @code{__builtin_args_info} accesses the same data structure of type
4965 @code{CUMULATIVE_ARGS} after the ordinary argument layout is finished
4966 with it, with @var{category} specifying which word to access. Thus, the
4967 value indicates the first unused register in a given category.
4969 Normally, you would use @code{__builtin_args_info} in the implementation
4970 of @code{va_start}, accessing each category just once and storing the
4971 value in the @code{va_list} object. This is because @code{va_list} will
4972 have to update the values, and there is no way to alter the
4973 values accessed by @code{__builtin_args_info}.
4976 @defmac __builtin_next_arg (@var{lastarg})
4977 This is the equivalent of @code{__builtin_args_info}, for stack
4978 arguments. It returns the address of the first anonymous stack
4979 argument, as type @code{void *}. If @code{ARGS_GROW_DOWNWARD}, it
4980 returns the address of the location above the first anonymous stack
4981 argument. Use it in @code{va_start} to initialize the pointer for
4982 fetching arguments from the stack. Also use it in @code{va_start} to
4983 verify that the second parameter @var{lastarg} is the last named argument
4984 of the current function.
4987 @defmac __builtin_classify_type (@var{object})
4988 Since each machine has its own conventions for which data types are
4989 passed in which kind of register, your implementation of @code{va_arg}
4990 has to embody these conventions. The easiest way to categorize the
4991 specified data type is to use @code{__builtin_classify_type} together
4992 with @code{sizeof} and @code{__alignof__}.
4994 @code{__builtin_classify_type} ignores the value of @var{object},
4995 considering only its data type. It returns an integer describing what
4996 kind of type that is---integer, floating, pointer, structure, and so on.
4998 The file @file{typeclass.h} defines an enumeration that you can use to
4999 interpret the values of @code{__builtin_classify_type}.
5002 These machine description macros help implement varargs:
5004 @deftypefn {Target Hook} rtx TARGET_EXPAND_BUILTIN_SAVEREGS (void)
5005 If defined, this hook produces the machine-specific code for a call to
5006 @code{__builtin_saveregs}. This code will be moved to the very
5007 beginning of the function, before any parameter access are made. The
5008 return value of this function should be an RTX that contains the value
5009 to use as the return of @code{__builtin_saveregs}.
5012 @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})
5013 This target hook offers an alternative to using
5014 @code{__builtin_saveregs} and defining the hook
5015 @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. Use it to store the anonymous
5016 register arguments into the stack so that all the arguments appear to
5017 have been passed consecutively on the stack. Once this is done, you can
5018 use the standard implementation of varargs that works for machines that
5019 pass all their arguments on the stack.
5021 The argument @var{args_so_far} points to the @code{CUMULATIVE_ARGS} data
5022 structure, containing the values that are obtained after processing the
5023 named arguments. The arguments @var{mode} and @var{type} describe the
5024 last named argument---its machine mode and its data type as a tree node.
5026 The target hook should do two things: first, push onto the stack all the
5027 argument registers @emph{not} used for the named arguments, and second,
5028 store the size of the data thus pushed into the @code{int}-valued
5029 variable pointed to by @var{pretend_args_size}. The value that you
5030 store here will serve as additional offset for setting up the stack
5033 Because you must generate code to push the anonymous arguments at
5034 compile time without knowing their data types,
5035 @code{TARGET_SETUP_INCOMING_VARARGS} is only useful on machines that
5036 have just a single category of argument register and use it uniformly
5039 If the argument @var{second_time} is nonzero, it means that the
5040 arguments of the function are being analyzed for the second time. This
5041 happens for an inline function, which is not actually compiled until the
5042 end of the source file. The hook @code{TARGET_SETUP_INCOMING_VARARGS} should
5043 not generate any instructions in this case.
5046 @deftypefn {Target Hook} bool TARGET_STRICT_ARGUMENT_NAMING (CUMULATIVE_ARGS *@var{ca})
5047 Define this hook to return @code{true} if the location where a function
5048 argument is passed depends on whether or not it is a named argument.
5050 This hook controls how the @var{named} argument to @code{FUNCTION_ARG}
5051 is set for varargs and stdarg functions. If this hook returns
5052 @code{true}, the @var{named} argument is always true for named
5053 arguments, and false for unnamed arguments. If it returns @code{false},
5054 but @code{TARGET_PRETEND_OUTGOING_VARARGS_NAMED} returns @code{true},
5055 then all arguments are treated as named. Otherwise, all named arguments
5056 except the last are treated as named.
5058 You need not define this hook if it always returns zero.
5061 @deftypefn {Target Hook} bool TARGET_PRETEND_OUTGOING_VARARGS_NAMED
5062 If you need to conditionally change ABIs so that one works with
5063 @code{TARGET_SETUP_INCOMING_VARARGS}, but the other works like neither
5064 @code{TARGET_SETUP_INCOMING_VARARGS} nor @code{TARGET_STRICT_ARGUMENT_NAMING} was
5065 defined, then define this hook to return @code{true} if
5066 @code{TARGET_SETUP_INCOMING_VARARGS} is used, @code{false} otherwise.
5067 Otherwise, you should not define this hook.
5071 @section Trampolines for Nested Functions
5072 @cindex trampolines for nested functions
5073 @cindex nested functions, trampolines for
5075 A @dfn{trampoline} is a small piece of code that is created at run time
5076 when the address of a nested function is taken. It normally resides on
5077 the stack, in the stack frame of the containing function. These macros
5078 tell GCC how to generate code to allocate and initialize a
5081 The instructions in the trampoline must do two things: load a constant
5082 address into the static chain register, and jump to the real address of
5083 the nested function. On CISC machines such as the m68k, this requires
5084 two instructions, a move immediate and a jump. Then the two addresses
5085 exist in the trampoline as word-long immediate operands. On RISC
5086 machines, it is often necessary to load each address into a register in
5087 two parts. Then pieces of each address form separate immediate
5090 The code generated to initialize the trampoline must store the variable
5091 parts---the static chain value and the function address---into the
5092 immediate operands of the instructions. On a CISC machine, this is
5093 simply a matter of copying each address to a memory reference at the
5094 proper offset from the start of the trampoline. On a RISC machine, it
5095 may be necessary to take out pieces of the address and store them
5098 @defmac TRAMPOLINE_TEMPLATE (@var{file})
5099 A C statement to output, on the stream @var{file}, assembler code for a
5100 block of data that contains the constant parts of a trampoline. This
5101 code should not include a label---the label is taken care of
5104 If you do not define this macro, it means no template is needed
5105 for the target. Do not define this macro on systems where the block move
5106 code to copy the trampoline into place would be larger than the code
5107 to generate it on the spot.
5110 @defmac TRAMPOLINE_SECTION
5111 Return the section into which the trampoline template is to be placed
5112 (@pxref{Sections}). The default value is @code{readonly_data_section}.
5115 @defmac TRAMPOLINE_SIZE
5116 A C expression for the size in bytes of the trampoline, as an integer.
5119 @defmac TRAMPOLINE_ALIGNMENT
5120 Alignment required for trampolines, in bits.
5122 If you don't define this macro, the value of @code{BIGGEST_ALIGNMENT}
5123 is used for aligning trampolines.
5126 @defmac INITIALIZE_TRAMPOLINE (@var{addr}, @var{fnaddr}, @var{static_chain})
5127 A C statement to initialize the variable parts of a trampoline.
5128 @var{addr} is an RTX for the address of the trampoline; @var{fnaddr} is
5129 an RTX for the address of the nested function; @var{static_chain} is an
5130 RTX for the static chain value that should be passed to the function
5134 @defmac TRAMPOLINE_ADJUST_ADDRESS (@var{addr})
5135 A C statement that should perform any machine-specific adjustment in
5136 the address of the trampoline. Its argument contains the address that
5137 was passed to @code{INITIALIZE_TRAMPOLINE}. In case the address to be
5138 used for a function call should be different from the address in which
5139 the template was stored, the different address should be assigned to
5140 @var{addr}. If this macro is not defined, @var{addr} will be used for
5143 @cindex @code{TARGET_ASM_FUNCTION_EPILOGUE} and trampolines
5144 @cindex @code{TARGET_ASM_FUNCTION_PROLOGUE} and trampolines
5145 If this macro is not defined, by default the trampoline is allocated as
5146 a stack slot. This default is right for most machines. The exceptions
5147 are machines where it is impossible to execute instructions in the stack
5148 area. On such machines, you may have to implement a separate stack,
5149 using this macro in conjunction with @code{TARGET_ASM_FUNCTION_PROLOGUE}
5150 and @code{TARGET_ASM_FUNCTION_EPILOGUE}.
5152 @var{fp} points to a data structure, a @code{struct function}, which
5153 describes the compilation status of the immediate containing function of
5154 the function which the trampoline is for. The stack slot for the
5155 trampoline is in the stack frame of this containing function. Other
5156 allocation strategies probably must do something analogous with this
5160 Implementing trampolines is difficult on many machines because they have
5161 separate instruction and data caches. Writing into a stack location
5162 fails to clear the memory in the instruction cache, so when the program
5163 jumps to that location, it executes the old contents.
5165 Here are two possible solutions. One is to clear the relevant parts of
5166 the instruction cache whenever a trampoline is set up. The other is to
5167 make all trampolines identical, by having them jump to a standard
5168 subroutine. The former technique makes trampoline execution faster; the
5169 latter makes initialization faster.
5171 To clear the instruction cache when a trampoline is initialized, define
5172 the following macro.
5174 @defmac CLEAR_INSN_CACHE (@var{beg}, @var{end})
5175 If defined, expands to a C expression clearing the @emph{instruction
5176 cache} in the specified interval. The definition of this macro would
5177 typically be a series of @code{asm} statements. Both @var{beg} and
5178 @var{end} are both pointer expressions.
5181 The operating system may also require the stack to be made executable
5182 before calling the trampoline. To implement this requirement, define
5183 the following macro.
5185 @defmac ENABLE_EXECUTE_STACK
5186 Define this macro if certain operations must be performed before executing
5187 code located on the stack. The macro should expand to a series of C
5188 file-scope constructs (e.g.@: functions) and provide a unique entry point
5189 named @code{__enable_execute_stack}. The target is responsible for
5190 emitting calls to the entry point in the code, for example from the
5191 @code{INITIALIZE_TRAMPOLINE} macro.
5194 To use a standard subroutine, define the following macro. In addition,
5195 you must make sure that the instructions in a trampoline fill an entire
5196 cache line with identical instructions, or else ensure that the
5197 beginning of the trampoline code is always aligned at the same point in
5198 its cache line. Look in @file{m68k.h} as a guide.
5200 @defmac TRANSFER_FROM_TRAMPOLINE
5201 Define this macro if trampolines need a special subroutine to do their
5202 work. The macro should expand to a series of @code{asm} statements
5203 which will be compiled with GCC@. They go in a library function named
5204 @code{__transfer_from_trampoline}.
5206 If you need to avoid executing the ordinary prologue code of a compiled
5207 C function when you jump to the subroutine, you can do so by placing a
5208 special label of your own in the assembler code. Use one @code{asm}
5209 statement to generate an assembler label, and another to make the label
5210 global. Then trampolines can use that label to jump directly to your
5211 special assembler code.
5215 @section Implicit Calls to Library Routines
5216 @cindex library subroutine names
5217 @cindex @file{libgcc.a}
5219 @c prevent bad page break with this line
5220 Here is an explanation of implicit calls to library routines.
5222 @defmac DECLARE_LIBRARY_RENAMES
5223 This macro, if defined, should expand to a piece of C code that will get
5224 expanded when compiling functions for libgcc.a. It can be used to
5225 provide alternate names for GCC's internal library functions if there
5226 are ABI-mandated names that the compiler should provide.
5229 @findex init_one_libfunc
5230 @findex set_optab_libfunc
5231 @deftypefn {Target Hook} void TARGET_INIT_LIBFUNCS (void)
5232 This hook should declare additional library routines or rename
5233 existing ones, using the functions @code{set_optab_libfunc} and
5234 @code{init_one_libfunc} defined in @file{optabs.c}.
5235 @code{init_optabs} calls this macro after initializing all the normal
5238 The default is to do nothing. Most ports don't need to define this hook.
5241 @defmac FLOAT_LIB_COMPARE_RETURNS_BOOL (@var{mode}, @var{comparison})
5242 This macro should return @code{true} if the library routine that
5243 implements the floating point comparison operator @var{comparison} in
5244 mode @var{mode} will return a boolean, and @var{false} if it will
5247 GCC's own floating point libraries return tristates from the
5248 comparison operators, so the default returns false always. Most ports
5249 don't need to define this macro.
5252 @defmac TARGET_LIB_INT_CMP_BIASED
5253 This macro should evaluate to @code{true} if the integer comparison
5254 functions (like @code{__cmpdi2}) return 0 to indicate that the first
5255 operand is smaller than the second, 1 to indicate that they are equal,
5256 and 2 to indicate that the first operand is greater than the second.
5257 If this macro evaluates to @code{false} the comparison functions return
5258 @minus{}1, 0, and 1 instead of 0, 1, and 2. If the target uses the routines
5259 in @file{libgcc.a}, you do not need to define this macro.
5262 @cindex US Software GOFAST, floating point emulation library
5263 @cindex floating point emulation library, US Software GOFAST
5264 @cindex GOFAST, floating point emulation library
5265 @findex gofast_maybe_init_libfuncs
5266 @defmac US_SOFTWARE_GOFAST
5267 Define this macro if your system C library uses the US Software GOFAST
5268 library to provide floating point emulation.
5270 In addition to defining this macro, your architecture must set
5271 @code{TARGET_INIT_LIBFUNCS} to @code{gofast_maybe_init_libfuncs}, or
5272 else call that function from its version of that hook. It is defined
5273 in @file{config/gofast.h}, which must be included by your
5274 architecture's @file{@var{cpu}.c} file. See @file{sparc/sparc.c} for
5277 If this macro is defined, the
5278 @code{TARGET_FLOAT_LIB_COMPARE_RETURNS_BOOL} target hook must return
5279 false for @code{SFmode} and @code{DFmode} comparisons.
5282 @cindex @code{EDOM}, implicit usage
5285 The value of @code{EDOM} on the target machine, as a C integer constant
5286 expression. If you don't define this macro, GCC does not attempt to
5287 deposit the value of @code{EDOM} into @code{errno} directly. Look in
5288 @file{/usr/include/errno.h} to find the value of @code{EDOM} on your
5291 If you do not define @code{TARGET_EDOM}, then compiled code reports
5292 domain errors by calling the library function and letting it report the
5293 error. If mathematical functions on your system use @code{matherr} when
5294 there is an error, then you should leave @code{TARGET_EDOM} undefined so
5295 that @code{matherr} is used normally.
5298 @cindex @code{errno}, implicit usage
5299 @defmac GEN_ERRNO_RTX
5300 Define this macro as a C expression to create an rtl expression that
5301 refers to the global ``variable'' @code{errno}. (On certain systems,
5302 @code{errno} may not actually be a variable.) If you don't define this
5303 macro, a reasonable default is used.
5306 @cindex C99 math functions, implicit usage
5307 @defmac TARGET_C99_FUNCTIONS
5308 When this macro is nonzero, GCC will implicitly optimize @code{sin} calls into
5309 @code{sinf} and similarly for other functions defined by C99 standard. The
5310 default is zero because a number of existing systems lack support for these
5311 functions in their runtime so this macro needs to be redefined to one on
5312 systems that do support the C99 runtime.
5315 @cindex sincos math function, implicit usage
5316 @defmac TARGET_HAS_SINCOS
5317 When this macro is nonzero, GCC will implicitly optimize calls to @code{sin}
5318 and @code{cos} with the same argument to a call to @code{sincos}. The
5319 default is zero. The target has to provide the following functions:
5321 void sincos(double x, double *sin, double *cos);
5322 void sincosf(float x, float *sin, float *cos);
5323 void sincosl(long double x, long double *sin, long double *cos);
5327 @defmac NEXT_OBJC_RUNTIME
5328 Define this macro to generate code for Objective-C message sending using
5329 the calling convention of the NeXT system. This calling convention
5330 involves passing the object, the selector and the method arguments all
5331 at once to the method-lookup library function.
5333 The default calling convention passes just the object and the selector
5334 to the lookup function, which returns a pointer to the method.
5337 @node Addressing Modes
5338 @section Addressing Modes
5339 @cindex addressing modes
5341 @c prevent bad page break with this line
5342 This is about addressing modes.
5344 @defmac HAVE_PRE_INCREMENT
5345 @defmacx HAVE_PRE_DECREMENT
5346 @defmacx HAVE_POST_INCREMENT
5347 @defmacx HAVE_POST_DECREMENT
5348 A C expression that is nonzero if the machine supports pre-increment,
5349 pre-decrement, post-increment, or post-decrement addressing respectively.
5352 @defmac HAVE_PRE_MODIFY_DISP
5353 @defmacx HAVE_POST_MODIFY_DISP
5354 A C expression that is nonzero if the machine supports pre- or
5355 post-address side-effect generation involving constants other than
5356 the size of the memory operand.
5359 @defmac HAVE_PRE_MODIFY_REG
5360 @defmacx HAVE_POST_MODIFY_REG
5361 A C expression that is nonzero if the machine supports pre- or
5362 post-address side-effect generation involving a register displacement.
5365 @defmac CONSTANT_ADDRESS_P (@var{x})
5366 A C expression that is 1 if the RTX @var{x} is a constant which
5367 is a valid address. On most machines, this can be defined as
5368 @code{CONSTANT_P (@var{x})}, but a few machines are more restrictive
5369 in which constant addresses are supported.
5372 @defmac CONSTANT_P (@var{x})
5373 @code{CONSTANT_P}, which is defined by target-independent code,
5374 accepts integer-values expressions whose values are not explicitly
5375 known, such as @code{symbol_ref}, @code{label_ref}, and @code{high}
5376 expressions and @code{const} arithmetic expressions, in addition to
5377 @code{const_int} and @code{const_double} expressions.
5380 @defmac MAX_REGS_PER_ADDRESS
5381 A number, the maximum number of registers that can appear in a valid
5382 memory address. Note that it is up to you to specify a value equal to
5383 the maximum number that @code{TARGET_LEGITIMATE_ADDRESS_P} would ever
5387 @deftypefn {Target Hook} TARGET_LEGITIMATE_ADDRESS_P (enum machine_mode @var{mode}, rtx @var{x}, bool @var{strict})
5388 A function that returns whether @var{x} (an RTX) is a legitimate memory
5389 address on the target machine for a memory operand of mode @var{mode}.
5391 Legitimate addresses are defined in two variants: a strict variant and a
5392 non-strict one. The @code{strict} parameter chooses which variant is
5393 desired by the caller.
5395 The strict variant is used in the reload pass. It must be defined so
5396 that any pseudo-register that has not been allocated a hard register is
5397 considered a memory reference. This is because in contexts where some
5398 kind of register is required, a pseudo-register with no hard register
5399 must be rejected. For non-hard registers, the strict variant should look
5400 up the @code{reg_renumber} array; it should then proceed using the hard
5401 register number in the array, or treat the pseudo as a memory reference
5402 if the array holds @code{-1}.
5404 The non-strict variant is used in other passes. It must be defined to
5405 accept all pseudo-registers in every context where some kind of
5406 register is required.
5408 Normally, constant addresses which are the sum of a @code{symbol_ref}
5409 and an integer are stored inside a @code{const} RTX to mark them as
5410 constant. Therefore, there is no need to recognize such sums
5411 specifically as legitimate addresses. Normally you would simply
5412 recognize any @code{const} as legitimate.
5414 Usually @code{PRINT_OPERAND_ADDRESS} is not prepared to handle constant
5415 sums that are not marked with @code{const}. It assumes that a naked
5416 @code{plus} indicates indexing. If so, then you @emph{must} reject such
5417 naked constant sums as illegitimate addresses, so that none of them will
5418 be given to @code{PRINT_OPERAND_ADDRESS}.
5420 @cindex @code{TARGET_ENCODE_SECTION_INFO} and address validation
5421 On some machines, whether a symbolic address is legitimate depends on
5422 the section that the address refers to. On these machines, define the
5423 target hook @code{TARGET_ENCODE_SECTION_INFO} to store the information
5424 into the @code{symbol_ref}, and then check for it here. When you see a
5425 @code{const}, you will have to look inside it to find the
5426 @code{symbol_ref} in order to determine the section. @xref{Assembler
5429 @cindex @code{GO_IF_LEGITIMATE_ADDRESS}
5430 Some ports are still using a deprecated legacy substitute for
5431 this hook, the @code{GO_IF_LEGITIMATE_ADDRESS} macro. This macro
5435 #define GO_IF_LEGITIMATE_ADDRESS (@var{mode}, @var{x}, @var{label})
5439 and should @code{goto @var{label}} if the address @var{x} is a valid
5440 address on the target machine for a memory operand of mode @var{mode}.
5441 Whether the strict or non-strict variants are desired is defined by
5442 the @code{REG_OK_STRICT} macro introduced earlier in this section.
5443 Using the hook is usually simpler because it limits the number of
5444 files that are recompiled when changes are made.
5447 @defmac TARGET_MEM_CONSTRAINT
5448 A single character to be used instead of the default @code{'m'}
5449 character for general memory addresses. This defines the constraint
5450 letter which matches the memory addresses accepted by
5451 @code{TARGET_LEGITIMATE_ADDRESS_P}. Define this macro if you want to
5452 support new address formats in your back end without changing the
5453 semantics of the @code{'m'} constraint. This is necessary in order to
5454 preserve functionality of inline assembly constructs using the
5455 @code{'m'} constraint.
5458 @defmac FIND_BASE_TERM (@var{x})
5459 A C expression to determine the base term of address @var{x},
5460 or to provide a simplified version of @var{x} from which @file{alias.c}
5461 can easily find the base term. This macro is used in only two places:
5462 @code{find_base_value} and @code{find_base_term} in @file{alias.c}.
5464 It is always safe for this macro to not be defined. It exists so
5465 that alias analysis can understand machine-dependent addresses.
5467 The typical use of this macro is to handle addresses containing
5468 a label_ref or symbol_ref within an UNSPEC@.
5471 @deftypefn {Target Hook} rtx TARGET_LEGITIMIZE_ADDRESS (rtx @var{x}, rtx @var{oldx}, enum machine_mode @var{mode})
5472 This hook is given an invalid memory address @var{x} for an
5473 operand of mode @var{mode} and should try to return a valid memory
5476 @findex break_out_memory_refs
5477 @var{x} will always be the result of a call to @code{break_out_memory_refs},
5478 and @var{oldx} will be the operand that was given to that function to produce
5481 The code of the hook should not alter the substructure of
5482 @var{x}. If it transforms @var{x} into a more legitimate form, it
5483 should return the new @var{x}.
5485 It is not necessary for this hook to come up with a legitimate address.
5486 The compiler has standard ways of doing so in all cases. In fact, it
5487 is safe to omit this hook or make it return @var{x} if it cannot find
5488 a valid way to legitimize the address. But often a machine-dependent
5489 strategy can generate better code.
5492 @defmac LEGITIMIZE_RELOAD_ADDRESS (@var{x}, @var{mode}, @var{opnum}, @var{type}, @var{ind_levels}, @var{win})
5493 A C compound statement that attempts to replace @var{x}, which is an address
5494 that needs reloading, with a valid memory address for an operand of mode
5495 @var{mode}. @var{win} will be a C statement label elsewhere in the code.
5496 It is not necessary to define this macro, but it might be useful for
5497 performance reasons.
5499 For example, on the i386, it is sometimes possible to use a single
5500 reload register instead of two by reloading a sum of two pseudo
5501 registers into a register. On the other hand, for number of RISC
5502 processors offsets are limited so that often an intermediate address
5503 needs to be generated in order to address a stack slot. By defining
5504 @code{LEGITIMIZE_RELOAD_ADDRESS} appropriately, the intermediate addresses
5505 generated for adjacent some stack slots can be made identical, and thus
5508 @emph{Note}: This macro should be used with caution. It is necessary
5509 to know something of how reload works in order to effectively use this,
5510 and it is quite easy to produce macros that build in too much knowledge
5511 of reload internals.
5513 @emph{Note}: This macro must be able to reload an address created by a
5514 previous invocation of this macro. If it fails to handle such addresses
5515 then the compiler may generate incorrect code or abort.
5518 The macro definition should use @code{push_reload} to indicate parts that
5519 need reloading; @var{opnum}, @var{type} and @var{ind_levels} are usually
5520 suitable to be passed unaltered to @code{push_reload}.
5522 The code generated by this macro must not alter the substructure of
5523 @var{x}. If it transforms @var{x} into a more legitimate form, it
5524 should assign @var{x} (which will always be a C variable) a new value.
5525 This also applies to parts that you change indirectly by calling
5528 @findex strict_memory_address_p
5529 The macro definition may use @code{strict_memory_address_p} to test if
5530 the address has become legitimate.
5533 If you want to change only a part of @var{x}, one standard way of doing
5534 this is to use @code{copy_rtx}. Note, however, that it unshares only a
5535 single level of rtl. Thus, if the part to be changed is not at the
5536 top level, you'll need to replace first the top level.
5537 It is not necessary for this macro to come up with a legitimate
5538 address; but often a machine-dependent strategy can generate better code.
5541 @defmac GO_IF_MODE_DEPENDENT_ADDRESS (@var{addr}, @var{label})
5542 A C statement or compound statement with a conditional @code{goto
5543 @var{label};} executed if memory address @var{x} (an RTX) can have
5544 different meanings depending on the machine mode of the memory
5545 reference it is used for or if the address is valid for some modes
5548 Autoincrement and autodecrement addresses typically have mode-dependent
5549 effects because the amount of the increment or decrement is the size
5550 of the operand being addressed. Some machines have other mode-dependent
5551 addresses. Many RISC machines have no mode-dependent addresses.
5553 You may assume that @var{addr} is a valid address for the machine.
5556 @defmac LEGITIMATE_CONSTANT_P (@var{x})
5557 A C expression that is nonzero if @var{x} is a legitimate constant for
5558 an immediate operand on the target machine. You can assume that
5559 @var{x} satisfies @code{CONSTANT_P}, so you need not check this. In fact,
5560 @samp{1} is a suitable definition for this macro on machines where
5561 anything @code{CONSTANT_P} is valid.
5564 @deftypefn {Target Hook} rtx TARGET_DELEGITIMIZE_ADDRESS (rtx @var{x})
5565 This hook is used to undo the possibly obfuscating effects of the
5566 @code{LEGITIMIZE_ADDRESS} and @code{LEGITIMIZE_RELOAD_ADDRESS} target
5567 macros. Some backend implementations of these macros wrap symbol
5568 references inside an @code{UNSPEC} rtx to represent PIC or similar
5569 addressing modes. This target hook allows GCC's optimizers to understand
5570 the semantics of these opaque @code{UNSPEC}s by converting them back
5571 into their original form.
5574 @deftypefn {Target Hook} bool TARGET_CANNOT_FORCE_CONST_MEM (rtx @var{x})
5575 This hook should return true if @var{x} is of a form that cannot (or
5576 should not) be spilled to the constant pool. The default version of
5577 this hook returns false.
5579 The primary reason to define this hook is to prevent reload from
5580 deciding that a non-legitimate constant would be better reloaded
5581 from the constant pool instead of spilling and reloading a register
5582 holding the constant. This restriction is often true of addresses
5583 of TLS symbols for various targets.
5586 @deftypefn {Target Hook} bool TARGET_USE_BLOCKS_FOR_CONSTANT_P (enum machine_mode @var{mode}, rtx @var{x})
5587 This hook should return true if pool entries for constant @var{x} can
5588 be placed in an @code{object_block} structure. @var{mode} is the mode
5591 The default version returns false for all constants.
5594 @deftypefn {Target Hook} tree TARGET_BUILTIN_RECIPROCAL (enum tree_code @var{fn}, bool @var{tm_fn}, bool @var{sqrt})
5595 This hook should return the DECL of a function that implements reciprocal of
5596 the builtin function with builtin function code @var{fn}, or
5597 @code{NULL_TREE} if such a function is not available. @var{tm_fn} is true
5598 when @var{fn} is a code of a machine-dependent builtin function. When
5599 @var{sqrt} is true, additional optimizations that apply only to the reciprocal
5600 of a square root function are performed, and only reciprocals of @code{sqrt}
5604 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_MASK_FOR_LOAD (void)
5605 This hook should return the DECL of a function @var{f} that given an
5606 address @var{addr} as an argument returns a mask @var{m} that can be
5607 used to extract from two vectors the relevant data that resides in
5608 @var{addr} in case @var{addr} is not properly aligned.
5610 The autovectorizer, when vectorizing a load operation from an address
5611 @var{addr} that may be unaligned, will generate two vector loads from
5612 the two aligned addresses around @var{addr}. It then generates a
5613 @code{REALIGN_LOAD} operation to extract the relevant data from the
5614 two loaded vectors. The first two arguments to @code{REALIGN_LOAD},
5615 @var{v1} and @var{v2}, are the two vectors, each of size @var{VS}, and
5616 the third argument, @var{OFF}, defines how the data will be extracted
5617 from these two vectors: if @var{OFF} is 0, then the returned vector is
5618 @var{v2}; otherwise, the returned vector is composed from the last
5619 @var{VS}-@var{OFF} elements of @var{v1} concatenated to the first
5620 @var{OFF} elements of @var{v2}.
5622 If this hook is defined, the autovectorizer will generate a call
5623 to @var{f} (using the DECL tree that this hook returns) and will
5624 use the return value of @var{f} as the argument @var{OFF} to
5625 @code{REALIGN_LOAD}. Therefore, the mask @var{m} returned by @var{f}
5626 should comply with the semantics expected by @code{REALIGN_LOAD}
5628 If this hook is not defined, then @var{addr} will be used as
5629 the argument @var{OFF} to @code{REALIGN_LOAD}, in which case the low
5630 log2(@var{VS})-1 bits of @var{addr} will be considered.
5633 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_EVEN (tree @var{x})
5634 This hook should return the DECL of a function @var{f} that implements
5635 widening multiplication of the even elements of two input vectors of type @var{x}.
5637 If this hook is defined, the autovectorizer will use it along with the
5638 @code{TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_ODD} target hook when vectorizing
5639 widening multiplication in cases that the order of the results does not have to be
5640 preserved (e.g.@: used only by a reduction computation). Otherwise, the
5641 @code{widen_mult_hi/lo} idioms will be used.
5644 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_ODD (tree @var{x})
5645 This hook should return the DECL of a function @var{f} that implements
5646 widening multiplication of the odd elements of two input vectors of type @var{x}.
5648 If this hook is defined, the autovectorizer will use it along with the
5649 @code{TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_EVEN} target hook when vectorizing
5650 widening multiplication in cases that the order of the results does not have to be
5651 preserved (e.g.@: used only by a reduction computation). Otherwise, the
5652 @code{widen_mult_hi/lo} idioms will be used.
5655 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_CONVERSION (enum tree_code @var{code}, tree @var{type})
5656 This hook should return the DECL of a function that implements conversion of the
5657 input vector of type @var{type}.
5658 If @var{type} is an integral type, the result of the conversion is a vector of
5659 floating-point type of the same size.
5660 If @var{type} is a floating-point type, the result of the conversion is a vector
5661 of integral type of the same size.
5662 @var{code} specifies how the conversion is to be applied
5663 (truncation, rounding, etc.).
5665 If this hook is defined, the autovectorizer will use the
5666 @code{TARGET_VECTORIZE_BUILTIN_CONVERSION} target hook when vectorizing
5667 conversion. Otherwise, it will return @code{NULL_TREE}.
5670 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION (enum built_in_function @var{code}, tree @var{vec_type_out}, tree @var{vec_type_in})
5671 This hook should return the decl of a function that implements the vectorized
5672 variant of the builtin function with builtin function code @var{code} or
5673 @code{NULL_TREE} if such a function is not available. The return type of
5674 the vectorized function shall be of vector type @var{vec_type_out} and the
5675 argument types should be @var{vec_type_in}.
5678 @node Anchored Addresses
5679 @section Anchored Addresses
5680 @cindex anchored addresses
5681 @cindex @option{-fsection-anchors}
5683 GCC usually addresses every static object as a separate entity.
5684 For example, if we have:
5688 int foo (void) @{ return a + b + c; @}
5691 the code for @code{foo} will usually calculate three separate symbolic
5692 addresses: those of @code{a}, @code{b} and @code{c}. On some targets,
5693 it would be better to calculate just one symbolic address and access
5694 the three variables relative to it. The equivalent pseudocode would
5700 register int *xr = &x;
5701 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
5705 (which isn't valid C). We refer to shared addresses like @code{x} as
5706 ``section anchors''. Their use is controlled by @option{-fsection-anchors}.
5708 The hooks below describe the target properties that GCC needs to know
5709 in order to make effective use of section anchors. It won't use
5710 section anchors at all unless either @code{TARGET_MIN_ANCHOR_OFFSET}
5711 or @code{TARGET_MAX_ANCHOR_OFFSET} is set to a nonzero value.
5713 @deftypevr {Target Hook} HOST_WIDE_INT TARGET_MIN_ANCHOR_OFFSET
5714 The minimum offset that should be applied to a section anchor.
5715 On most targets, it should be the smallest offset that can be
5716 applied to a base register while still giving a legitimate address
5717 for every mode. The default value is 0.
5720 @deftypevr {Target Hook} HOST_WIDE_INT TARGET_MAX_ANCHOR_OFFSET
5721 Like @code{TARGET_MIN_ANCHOR_OFFSET}, but the maximum (inclusive)
5722 offset that should be applied to section anchors. The default
5726 @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_ANCHOR (rtx @var{x})
5727 Write the assembly code to define section anchor @var{x}, which is a
5728 @code{SYMBOL_REF} for which @samp{SYMBOL_REF_ANCHOR_P (@var{x})} is true.
5729 The hook is called with the assembly output position set to the beginning
5730 of @code{SYMBOL_REF_BLOCK (@var{x})}.
5732 If @code{ASM_OUTPUT_DEF} is available, the hook's default definition uses
5733 it to define the symbol as @samp{. + SYMBOL_REF_BLOCK_OFFSET (@var{x})}.
5734 If @code{ASM_OUTPUT_DEF} is not available, the hook's default definition
5735 is @code{NULL}, which disables the use of section anchors altogether.
5738 @deftypefn {Target Hook} bool TARGET_USE_ANCHORS_FOR_SYMBOL_P (rtx @var{x})
5739 Return true if GCC should attempt to use anchors to access @code{SYMBOL_REF}
5740 @var{x}. You can assume @samp{SYMBOL_REF_HAS_BLOCK_INFO_P (@var{x})} and
5741 @samp{!SYMBOL_REF_ANCHOR_P (@var{x})}.
5743 The default version is correct for most targets, but you might need to
5744 intercept this hook to handle things like target-specific attributes
5745 or target-specific sections.
5748 @node Condition Code
5749 @section Condition Code Status
5750 @cindex condition code status
5752 The macros in this section can be split in two families, according to the
5753 two ways of representing condition codes in GCC.
5755 The first representation is the so called @code{(cc0)} representation
5756 (@pxref{Jump Patterns}), where all instructions can have an implicit
5757 clobber of the condition codes. The second is the condition code
5758 register representation, which provides better schedulability for
5759 architectures that do have a condition code register, but on which
5760 most instructions do not affect it. The latter category includes
5763 The implicit clobbering poses a strong restriction on the placement of
5764 the definition and use of the condition code, which need to be in adjacent
5765 insns for machines using @code{(cc0)}. This can prevent important
5766 optimizations on some machines. For example, on the IBM RS/6000, there
5767 is a delay for taken branches unless the condition code register is set
5768 three instructions earlier than the conditional branch. The instruction
5769 scheduler cannot perform this optimization if it is not permitted to
5770 separate the definition and use of the condition code register.
5772 For this reason, it is possible and suggested to use a register to
5773 represent the condition code for new ports. If there is a specific
5774 condition code register in the machine, use a hard register. If the
5775 condition code or comparison result can be placed in any general register,
5776 or if there are multiple condition registers, use a pseudo register.
5777 Registers used to store the condition code value will usually have a mode
5778 that is in class @code{MODE_CC}.
5780 Alternatively, you can use @code{BImode} if the comparison operator is
5781 specified already in the compare instruction. In this case, you are not
5782 interested in most macros in this section.
5785 * CC0 Condition Codes:: Old style representation of condition codes.
5786 * MODE_CC Condition Codes:: Modern representation of condition codes.
5787 * Cond. Exec. Macros:: Macros to control conditional execution.
5790 @node CC0 Condition Codes
5791 @subsection Representation of condition codes using @code{(cc0)}
5795 The file @file{conditions.h} defines a variable @code{cc_status} to
5796 describe how the condition code was computed (in case the interpretation of
5797 the condition code depends on the instruction that it was set by). This
5798 variable contains the RTL expressions on which the condition code is
5799 currently based, and several standard flags.
5801 Sometimes additional machine-specific flags must be defined in the machine
5802 description header file. It can also add additional machine-specific
5803 information by defining @code{CC_STATUS_MDEP}.
5805 @defmac CC_STATUS_MDEP
5806 C code for a data type which is used for declaring the @code{mdep}
5807 component of @code{cc_status}. It defaults to @code{int}.
5809 This macro is not used on machines that do not use @code{cc0}.
5812 @defmac CC_STATUS_MDEP_INIT
5813 A C expression to initialize the @code{mdep} field to ``empty''.
5814 The default definition does nothing, since most machines don't use
5815 the field anyway. If you want to use the field, you should probably
5816 define this macro to initialize it.
5818 This macro is not used on machines that do not use @code{cc0}.
5821 @defmac NOTICE_UPDATE_CC (@var{exp}, @var{insn})
5822 A C compound statement to set the components of @code{cc_status}
5823 appropriately for an insn @var{insn} whose body is @var{exp}. It is
5824 this macro's responsibility to recognize insns that set the condition
5825 code as a byproduct of other activity as well as those that explicitly
5828 This macro is not used on machines that do not use @code{cc0}.
5830 If there are insns that do not set the condition code but do alter
5831 other machine registers, this macro must check to see whether they
5832 invalidate the expressions that the condition code is recorded as
5833 reflecting. For example, on the 68000, insns that store in address
5834 registers do not set the condition code, which means that usually
5835 @code{NOTICE_UPDATE_CC} can leave @code{cc_status} unaltered for such
5836 insns. But suppose that the previous insn set the condition code
5837 based on location @samp{a4@@(102)} and the current insn stores a new
5838 value in @samp{a4}. Although the condition code is not changed by
5839 this, it will no longer be true that it reflects the contents of
5840 @samp{a4@@(102)}. Therefore, @code{NOTICE_UPDATE_CC} must alter
5841 @code{cc_status} in this case to say that nothing is known about the
5842 condition code value.
5844 The definition of @code{NOTICE_UPDATE_CC} must be prepared to deal
5845 with the results of peephole optimization: insns whose patterns are
5846 @code{parallel} RTXs containing various @code{reg}, @code{mem} or
5847 constants which are just the operands. The RTL structure of these
5848 insns is not sufficient to indicate what the insns actually do. What
5849 @code{NOTICE_UPDATE_CC} should do when it sees one is just to run
5850 @code{CC_STATUS_INIT}.
5852 A possible definition of @code{NOTICE_UPDATE_CC} is to call a function
5853 that looks at an attribute (@pxref{Insn Attributes}) named, for example,
5854 @samp{cc}. This avoids having detailed information about patterns in
5855 two places, the @file{md} file and in @code{NOTICE_UPDATE_CC}.
5858 @node MODE_CC Condition Codes
5859 @subsection Representation of condition codes using registers
5863 @defmac SELECT_CC_MODE (@var{op}, @var{x}, @var{y})
5864 On many machines, the condition code may be produced by other instructions
5865 than compares, for example the branch can use directly the condition
5866 code set by a subtract instruction. However, on some machines
5867 when the condition code is set this way some bits (such as the overflow
5868 bit) are not set in the same way as a test instruction, so that a different
5869 branch instruction must be used for some conditional branches. When
5870 this happens, use the machine mode of the condition code register to
5871 record different formats of the condition code register. Modes can
5872 also be used to record which compare instruction (e.g. a signed or an
5873 unsigned comparison) produced the condition codes.
5875 If other modes than @code{CCmode} are required, add them to
5876 @file{@var{machine}-modes.def} and define @code{SELECT_CC_MODE} to choose
5877 a mode given an operand of a compare. This is needed because the modes
5878 have to be chosen not only during RTL generation but also, for example,
5879 by instruction combination. The result of @code{SELECT_CC_MODE} should
5880 be consistent with the mode used in the patterns; for example to support
5881 the case of the add on the SPARC discussed above, we have the pattern
5885 [(set (reg:CC_NOOV 0)
5887 (plus:SI (match_operand:SI 0 "register_operand" "%r")
5888 (match_operand:SI 1 "arith_operand" "rI"))
5895 together with a @code{SELECT_CC_MODE} that returns @code{CC_NOOVmode}
5896 for comparisons whose argument is a @code{plus}:
5899 #define SELECT_CC_MODE(OP,X,Y) \
5900 (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \
5901 ? ((OP == EQ || OP == NE) ? CCFPmode : CCFPEmode) \
5902 : ((GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \
5903 || GET_CODE (X) == NEG) \
5904 ? CC_NOOVmode : CCmode))
5907 Another reason to use modes is to retain information on which operands
5908 were used by the comparison; see @code{REVERSIBLE_CC_MODE} later in
5911 You should define this macro if and only if you define extra CC modes
5912 in @file{@var{machine}-modes.def}.
5915 @defmac CANONICALIZE_COMPARISON (@var{code}, @var{op0}, @var{op1})
5916 On some machines not all possible comparisons are defined, but you can
5917 convert an invalid comparison into a valid one. For example, the Alpha
5918 does not have a @code{GT} comparison, but you can use an @code{LT}
5919 comparison instead and swap the order of the operands.
5921 On such machines, define this macro to be a C statement to do any
5922 required conversions. @var{code} is the initial comparison code
5923 and @var{op0} and @var{op1} are the left and right operands of the
5924 comparison, respectively. You should modify @var{code}, @var{op0}, and
5925 @var{op1} as required.
5927 GCC will not assume that the comparison resulting from this macro is
5928 valid but will see if the resulting insn matches a pattern in the
5931 You need not define this macro if it would never change the comparison
5935 @defmac REVERSIBLE_CC_MODE (@var{mode})
5936 A C expression whose value is one if it is always safe to reverse a
5937 comparison whose mode is @var{mode}. If @code{SELECT_CC_MODE}
5938 can ever return @var{mode} for a floating-point inequality comparison,
5939 then @code{REVERSIBLE_CC_MODE (@var{mode})} must be zero.
5941 You need not define this macro if it would always returns zero or if the
5942 floating-point format is anything other than @code{IEEE_FLOAT_FORMAT}.
5943 For example, here is the definition used on the SPARC, where floating-point
5944 inequality comparisons are always given @code{CCFPEmode}:
5947 #define REVERSIBLE_CC_MODE(MODE) ((MODE) != CCFPEmode)
5951 @defmac REVERSE_CONDITION (@var{code}, @var{mode})
5952 A C expression whose value is reversed condition code of the @var{code} for
5953 comparison done in CC_MODE @var{mode}. The macro is used only in case
5954 @code{REVERSIBLE_CC_MODE (@var{mode})} is nonzero. Define this macro in case
5955 machine has some non-standard way how to reverse certain conditionals. For
5956 instance in case all floating point conditions are non-trapping, compiler may
5957 freely convert unordered compares to ordered one. Then definition may look
5961 #define REVERSE_CONDITION(CODE, MODE) \
5962 ((MODE) != CCFPmode ? reverse_condition (CODE) \
5963 : reverse_condition_maybe_unordered (CODE))
5967 @deftypefn {Target Hook} bool TARGET_FIXED_CONDITION_CODE_REGS (unsigned int *, unsigned int *)
5968 On targets which do not use @code{(cc0)}, and which use a hard
5969 register rather than a pseudo-register to hold condition codes, the
5970 regular CSE passes are often not able to identify cases in which the
5971 hard register is set to a common value. Use this hook to enable a
5972 small pass which optimizes such cases. This hook should return true
5973 to enable this pass, and it should set the integers to which its
5974 arguments point to the hard register numbers used for condition codes.
5975 When there is only one such register, as is true on most systems, the
5976 integer pointed to by the second argument should be set to
5977 @code{INVALID_REGNUM}.
5979 The default version of this hook returns false.
5982 @deftypefn {Target Hook} enum machine_mode TARGET_CC_MODES_COMPATIBLE (enum machine_mode, enum machine_mode)
5983 On targets which use multiple condition code modes in class
5984 @code{MODE_CC}, it is sometimes the case that a comparison can be
5985 validly done in more than one mode. On such a system, define this
5986 target hook to take two mode arguments and to return a mode in which
5987 both comparisons may be validly done. If there is no such mode,
5988 return @code{VOIDmode}.
5990 The default version of this hook checks whether the modes are the
5991 same. If they are, it returns that mode. If they are different, it
5992 returns @code{VOIDmode}.
5995 @node Cond. Exec. Macros
5996 @subsection Macros to control conditional execution
5997 @findex conditional execution
6000 There is one macro that may need to be defined for targets
6001 supporting conditional execution, independent of how they
6002 represent conditional branches.
6004 @defmac REVERSE_CONDEXEC_PREDICATES_P (@var{op1}, @var{op2})
6005 A C expression that returns true if the conditional execution predicate
6006 @var{op1}, a comparison operation, is the inverse of @var{op2} and vice
6007 versa. Define this to return 0 if the target has conditional execution
6008 predicates that cannot be reversed safely. There is no need to validate
6009 that the arguments of op1 and op2 are the same, this is done separately.
6010 If no expansion is specified, this macro is defined as follows:
6013 #define REVERSE_CONDEXEC_PREDICATES_P (x, y) \
6014 (GET_CODE ((x)) == reversed_comparison_code ((y), NULL))
6019 @section Describing Relative Costs of Operations
6020 @cindex costs of instructions
6021 @cindex relative costs
6022 @cindex speed of instructions
6024 These macros let you describe the relative speed of various operations
6025 on the target machine.
6027 @defmac REGISTER_MOVE_COST (@var{mode}, @var{from}, @var{to})
6028 A C expression for the cost of moving data of mode @var{mode} from a
6029 register in class @var{from} to one in class @var{to}. The classes are
6030 expressed using the enumeration values such as @code{GENERAL_REGS}. A
6031 value of 2 is the default; other values are interpreted relative to
6034 It is not required that the cost always equal 2 when @var{from} is the
6035 same as @var{to}; on some machines it is expensive to move between
6036 registers if they are not general registers.
6038 If reload sees an insn consisting of a single @code{set} between two
6039 hard registers, and if @code{REGISTER_MOVE_COST} applied to their
6040 classes returns a value of 2, reload does not check to ensure that the
6041 constraints of the insn are met. Setting a cost of other than 2 will
6042 allow reload to verify that the constraints are met. You should do this
6043 if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
6046 @defmac MEMORY_MOVE_COST (@var{mode}, @var{class}, @var{in})
6047 A C expression for the cost of moving data of mode @var{mode} between a
6048 register of class @var{class} and memory; @var{in} is zero if the value
6049 is to be written to memory, nonzero if it is to be read in. This cost
6050 is relative to those in @code{REGISTER_MOVE_COST}. If moving between
6051 registers and memory is more expensive than between two registers, you
6052 should define this macro to express the relative cost.
6054 If you do not define this macro, GCC uses a default cost of 4 plus
6055 the cost of copying via a secondary reload register, if one is
6056 needed. If your machine requires a secondary reload register to copy
6057 between memory and a register of @var{class} but the reload mechanism is
6058 more complex than copying via an intermediate, define this macro to
6059 reflect the actual cost of the move.
6061 GCC defines the function @code{memory_move_secondary_cost} if
6062 secondary reloads are needed. It computes the costs due to copying via
6063 a secondary register. If your machine copies from memory using a
6064 secondary register in the conventional way but the default base value of
6065 4 is not correct for your machine, define this macro to add some other
6066 value to the result of that function. The arguments to that function
6067 are the same as to this macro.
6070 @defmac BRANCH_COST (@var{speed_p}, @var{predictable_p})
6071 A C expression for the cost of a branch instruction. A value of 1 is the
6072 default; other values are interpreted relative to that. Parameter @var{speed_p}
6073 is true when the branch in question should be optimized for speed. When
6074 it is false, @code{BRANCH_COST} should be returning value optimal for code size
6075 rather then performance considerations. @var{predictable_p} is true for well
6076 predictable branches. On many architectures the @code{BRANCH_COST} can be
6080 Here are additional macros which do not specify precise relative costs,
6081 but only that certain actions are more expensive than GCC would
6084 @defmac SLOW_BYTE_ACCESS
6085 Define this macro as a C expression which is nonzero if accessing less
6086 than a word of memory (i.e.@: a @code{char} or a @code{short}) is no
6087 faster than accessing a word of memory, i.e., if such access
6088 require more than one instruction or if there is no difference in cost
6089 between byte and (aligned) word loads.
6091 When this macro is not defined, the compiler will access a field by
6092 finding the smallest containing object; when it is defined, a fullword
6093 load will be used if alignment permits. Unless bytes accesses are
6094 faster than word accesses, using word accesses is preferable since it
6095 may eliminate subsequent memory access if subsequent accesses occur to
6096 other fields in the same word of the structure, but to different bytes.
6099 @defmac SLOW_UNALIGNED_ACCESS (@var{mode}, @var{alignment})
6100 Define this macro to be the value 1 if memory accesses described by the
6101 @var{mode} and @var{alignment} parameters have a cost many times greater
6102 than aligned accesses, for example if they are emulated in a trap
6105 When this macro is nonzero, the compiler will act as if
6106 @code{STRICT_ALIGNMENT} were nonzero when generating code for block
6107 moves. This can cause significantly more instructions to be produced.
6108 Therefore, do not set this macro nonzero if unaligned accesses only add a
6109 cycle or two to the time for a memory access.
6111 If the value of this macro is always zero, it need not be defined. If
6112 this macro is defined, it should produce a nonzero value when
6113 @code{STRICT_ALIGNMENT} is nonzero.
6117 The threshold of number of scalar memory-to-memory move insns, @emph{below}
6118 which a sequence of insns should be generated instead of a
6119 string move insn or a library call. Increasing the value will always
6120 make code faster, but eventually incurs high cost in increased code size.
6122 Note that on machines where the corresponding move insn is a
6123 @code{define_expand} that emits a sequence of insns, this macro counts
6124 the number of such sequences.
6126 If you don't define this, a reasonable default is used.
6129 @defmac MOVE_BY_PIECES_P (@var{size}, @var{alignment})
6130 A C expression used to determine whether @code{move_by_pieces} will be used to
6131 copy a chunk of memory, or whether some other block move mechanism
6132 will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6133 than @code{MOVE_RATIO}.
6136 @defmac MOVE_MAX_PIECES
6137 A C expression used by @code{move_by_pieces} to determine the largest unit
6138 a load or store used to copy memory is. Defaults to @code{MOVE_MAX}.
6142 The threshold of number of scalar move insns, @emph{below} which a sequence
6143 of insns should be generated to clear memory instead of a string clear insn
6144 or a library call. Increasing the value will always make code faster, but
6145 eventually incurs high cost in increased code size.
6147 If you don't define this, a reasonable default is used.
6150 @defmac CLEAR_BY_PIECES_P (@var{size}, @var{alignment})
6151 A C expression used to determine whether @code{clear_by_pieces} will be used
6152 to clear a chunk of memory, or whether some other block clear mechanism
6153 will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6154 than @code{CLEAR_RATIO}.
6158 The threshold of number of scalar move insns, @emph{below} which a sequence
6159 of insns should be generated to set memory to a constant value, instead of
6160 a block set insn or a library call.
6161 Increasing the value will always make code faster, but
6162 eventually incurs high cost in increased code size.
6164 If you don't define this, it defaults to the value of @code{MOVE_RATIO}.
6167 @defmac SET_BY_PIECES_P (@var{size}, @var{alignment})
6168 A C expression used to determine whether @code{store_by_pieces} will be
6169 used to set a chunk of memory to a constant value, or whether some
6170 other mechanism will be used. Used by @code{__builtin_memset} when
6171 storing values other than constant zero.
6172 Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6173 than @code{SET_RATIO}.
6176 @defmac STORE_BY_PIECES_P (@var{size}, @var{alignment})
6177 A C expression used to determine whether @code{store_by_pieces} will be
6178 used to set a chunk of memory to a constant string value, or whether some
6179 other mechanism will be used. Used by @code{__builtin_strcpy} when
6180 called with a constant source string.
6181 Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6182 than @code{MOVE_RATIO}.
6185 @defmac USE_LOAD_POST_INCREMENT (@var{mode})
6186 A C expression used to determine whether a load postincrement is a good
6187 thing to use for a given mode. Defaults to the value of
6188 @code{HAVE_POST_INCREMENT}.
6191 @defmac USE_LOAD_POST_DECREMENT (@var{mode})
6192 A C expression used to determine whether a load postdecrement is a good
6193 thing to use for a given mode. Defaults to the value of
6194 @code{HAVE_POST_DECREMENT}.
6197 @defmac USE_LOAD_PRE_INCREMENT (@var{mode})
6198 A C expression used to determine whether a load preincrement is a good
6199 thing to use for a given mode. Defaults to the value of
6200 @code{HAVE_PRE_INCREMENT}.
6203 @defmac USE_LOAD_PRE_DECREMENT (@var{mode})
6204 A C expression used to determine whether a load predecrement is a good
6205 thing to use for a given mode. Defaults to the value of
6206 @code{HAVE_PRE_DECREMENT}.
6209 @defmac USE_STORE_POST_INCREMENT (@var{mode})
6210 A C expression used to determine whether a store postincrement is a good
6211 thing to use for a given mode. Defaults to the value of
6212 @code{HAVE_POST_INCREMENT}.
6215 @defmac USE_STORE_POST_DECREMENT (@var{mode})
6216 A C expression used to determine whether a store postdecrement is a good
6217 thing to use for a given mode. Defaults to the value of
6218 @code{HAVE_POST_DECREMENT}.
6221 @defmac USE_STORE_PRE_INCREMENT (@var{mode})
6222 This macro is used to determine whether a store preincrement is a good
6223 thing to use for a given mode. Defaults to the value of
6224 @code{HAVE_PRE_INCREMENT}.
6227 @defmac USE_STORE_PRE_DECREMENT (@var{mode})
6228 This macro is used to determine whether a store predecrement is a good
6229 thing to use for a given mode. Defaults to the value of
6230 @code{HAVE_PRE_DECREMENT}.
6233 @defmac NO_FUNCTION_CSE
6234 Define this macro if it is as good or better to call a constant
6235 function address than to call an address kept in a register.
6238 @defmac RANGE_TEST_NON_SHORT_CIRCUIT
6239 Define this macro if a non-short-circuit operation produced by
6240 @samp{fold_range_test ()} is optimal. This macro defaults to true if
6241 @code{BRANCH_COST} is greater than or equal to the value 2.
6244 @deftypefn {Target Hook} bool TARGET_RTX_COSTS (rtx @var{x}, int @var{code}, int @var{outer_code}, int *@var{total})
6245 This target hook describes the relative costs of RTL expressions.
6247 The cost may depend on the precise form of the expression, which is
6248 available for examination in @var{x}, and the rtx code of the expression
6249 in which it is contained, found in @var{outer_code}. @var{code} is the
6250 expression code---redundant, since it can be obtained with
6251 @code{GET_CODE (@var{x})}.
6253 In implementing this hook, you can use the construct
6254 @code{COSTS_N_INSNS (@var{n})} to specify a cost equal to @var{n} fast
6257 On entry to the hook, @code{*@var{total}} contains a default estimate
6258 for the cost of the expression. The hook should modify this value as
6259 necessary. Traditionally, the default costs are @code{COSTS_N_INSNS (5)}
6260 for multiplications, @code{COSTS_N_INSNS (7)} for division and modulus
6261 operations, and @code{COSTS_N_INSNS (1)} for all other operations.
6263 When optimizing for code size, i.e.@: when @code{optimize_size} is
6264 nonzero, this target hook should be used to estimate the relative
6265 size cost of an expression, again relative to @code{COSTS_N_INSNS}.
6267 The hook returns true when all subexpressions of @var{x} have been
6268 processed, and false when @code{rtx_cost} should recurse.
6271 @deftypefn {Target Hook} int TARGET_ADDRESS_COST (rtx @var{address})
6272 This hook computes the cost of an addressing mode that contains
6273 @var{address}. If not defined, the cost is computed from
6274 the @var{address} expression and the @code{TARGET_RTX_COST} hook.
6276 For most CISC machines, the default cost is a good approximation of the
6277 true cost of the addressing mode. However, on RISC machines, all
6278 instructions normally have the same length and execution time. Hence
6279 all addresses will have equal costs.
6281 In cases where more than one form of an address is known, the form with
6282 the lowest cost will be used. If multiple forms have the same, lowest,
6283 cost, the one that is the most complex will be used.
6285 For example, suppose an address that is equal to the sum of a register
6286 and a constant is used twice in the same basic block. When this macro
6287 is not defined, the address will be computed in a register and memory
6288 references will be indirect through that register. On machines where
6289 the cost of the addressing mode containing the sum is no higher than
6290 that of a simple indirect reference, this will produce an additional
6291 instruction and possibly require an additional register. Proper
6292 specification of this macro eliminates this overhead for such machines.
6294 This hook is never called with an invalid address.
6296 On machines where an address involving more than one register is as
6297 cheap as an address computation involving only one register, defining
6298 @code{TARGET_ADDRESS_COST} to reflect this can cause two registers to
6299 be live over a region of code where only one would have been if
6300 @code{TARGET_ADDRESS_COST} were not defined in that manner. This effect
6301 should be considered in the definition of this macro. Equivalent costs
6302 should probably only be given to addresses with different numbers of
6303 registers on machines with lots of registers.
6307 @section Adjusting the Instruction Scheduler
6309 The instruction scheduler may need a fair amount of machine-specific
6310 adjustment in order to produce good code. GCC provides several target
6311 hooks for this purpose. It is usually enough to define just a few of
6312 them: try the first ones in this list first.
6314 @deftypefn {Target Hook} int TARGET_SCHED_ISSUE_RATE (void)
6315 This hook returns the maximum number of instructions that can ever
6316 issue at the same time on the target machine. The default is one.
6317 Although the insn scheduler can define itself the possibility of issue
6318 an insn on the same cycle, the value can serve as an additional
6319 constraint to issue insns on the same simulated processor cycle (see
6320 hooks @samp{TARGET_SCHED_REORDER} and @samp{TARGET_SCHED_REORDER2}).
6321 This value must be constant over the entire compilation. If you need
6322 it to vary depending on what the instructions are, you must use
6323 @samp{TARGET_SCHED_VARIABLE_ISSUE}.
6326 @deftypefn {Target Hook} int TARGET_SCHED_VARIABLE_ISSUE (FILE *@var{file}, int @var{verbose}, rtx @var{insn}, int @var{more})
6327 This hook is executed by the scheduler after it has scheduled an insn
6328 from the ready list. It should return the number of insns which can
6329 still be issued in the current cycle. The default is
6330 @samp{@w{@var{more} - 1}} for insns other than @code{CLOBBER} and
6331 @code{USE}, which normally are not counted against the issue rate.
6332 You should define this hook if some insns take more machine resources
6333 than others, so that fewer insns can follow them in the same cycle.
6334 @var{file} is either a null pointer, or a stdio stream to write any
6335 debug output to. @var{verbose} is the verbose level provided by
6336 @option{-fsched-verbose-@var{n}}. @var{insn} is the instruction that
6340 @deftypefn {Target Hook} int TARGET_SCHED_ADJUST_COST (rtx @var{insn}, rtx @var{link}, rtx @var{dep_insn}, int @var{cost})
6341 This function corrects the value of @var{cost} based on the
6342 relationship between @var{insn} and @var{dep_insn} through the
6343 dependence @var{link}. It should return the new value. The default
6344 is to make no adjustment to @var{cost}. This can be used for example
6345 to specify to the scheduler using the traditional pipeline description
6346 that an output- or anti-dependence does not incur the same cost as a
6347 data-dependence. If the scheduler using the automaton based pipeline
6348 description, the cost of anti-dependence is zero and the cost of
6349 output-dependence is maximum of one and the difference of latency
6350 times of the first and the second insns. If these values are not
6351 acceptable, you could use the hook to modify them too. See also
6352 @pxref{Processor pipeline description}.
6355 @deftypefn {Target Hook} int TARGET_SCHED_ADJUST_PRIORITY (rtx @var{insn}, int @var{priority})
6356 This hook adjusts the integer scheduling priority @var{priority} of
6357 @var{insn}. It should return the new priority. Increase the priority to
6358 execute @var{insn} earlier, reduce the priority to execute @var{insn}
6359 later. Do not define this hook if you do not need to adjust the
6360 scheduling priorities of insns.
6363 @deftypefn {Target Hook} int TARGET_SCHED_REORDER (FILE *@var{file}, int @var{verbose}, rtx *@var{ready}, int *@var{n_readyp}, int @var{clock})
6364 This hook is executed by the scheduler after it has scheduled the ready
6365 list, to allow the machine description to reorder it (for example to
6366 combine two small instructions together on @samp{VLIW} machines).
6367 @var{file} is either a null pointer, or a stdio stream to write any
6368 debug output to. @var{verbose} is the verbose level provided by
6369 @option{-fsched-verbose-@var{n}}. @var{ready} is a pointer to the ready
6370 list of instructions that are ready to be scheduled. @var{n_readyp} is
6371 a pointer to the number of elements in the ready list. The scheduler
6372 reads the ready list in reverse order, starting with
6373 @var{ready}[@var{*n_readyp}-1] and going to @var{ready}[0]. @var{clock}
6374 is the timer tick of the scheduler. You may modify the ready list and
6375 the number of ready insns. The return value is the number of insns that
6376 can issue this cycle; normally this is just @code{issue_rate}. See also
6377 @samp{TARGET_SCHED_REORDER2}.
6380 @deftypefn {Target Hook} int TARGET_SCHED_REORDER2 (FILE *@var{file}, int @var{verbose}, rtx *@var{ready}, int *@var{n_ready}, @var{clock})
6381 Like @samp{TARGET_SCHED_REORDER}, but called at a different time. That
6382 function is called whenever the scheduler starts a new cycle. This one
6383 is called once per iteration over a cycle, immediately after
6384 @samp{TARGET_SCHED_VARIABLE_ISSUE}; it can reorder the ready list and
6385 return the number of insns to be scheduled in the same cycle. Defining
6386 this hook can be useful if there are frequent situations where
6387 scheduling one insn causes other insns to become ready in the same
6388 cycle. These other insns can then be taken into account properly.
6391 @deftypefn {Target Hook} void TARGET_SCHED_DEPENDENCIES_EVALUATION_HOOK (rtx @var{head}, rtx @var{tail})
6392 This hook is called after evaluation forward dependencies of insns in
6393 chain given by two parameter values (@var{head} and @var{tail}
6394 correspondingly) but before insns scheduling of the insn chain. For
6395 example, it can be used for better insn classification if it requires
6396 analysis of dependencies. This hook can use backward and forward
6397 dependencies of the insn scheduler because they are already
6401 @deftypefn {Target Hook} void TARGET_SCHED_INIT (FILE *@var{file}, int @var{verbose}, int @var{max_ready})
6402 This hook is executed by the scheduler at the beginning of each block of
6403 instructions that are to be scheduled. @var{file} is either a null
6404 pointer, or a stdio stream to write any debug output to. @var{verbose}
6405 is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6406 @var{max_ready} is the maximum number of insns in the current scheduling
6407 region that can be live at the same time. This can be used to allocate
6408 scratch space if it is needed, e.g.@: by @samp{TARGET_SCHED_REORDER}.
6411 @deftypefn {Target Hook} void TARGET_SCHED_FINISH (FILE *@var{file}, int @var{verbose})
6412 This hook is executed by the scheduler at the end of each block of
6413 instructions that are to be scheduled. It can be used to perform
6414 cleanup of any actions done by the other scheduling hooks. @var{file}
6415 is either a null pointer, or a stdio stream to write any debug output
6416 to. @var{verbose} is the verbose level provided by
6417 @option{-fsched-verbose-@var{n}}.
6420 @deftypefn {Target Hook} void TARGET_SCHED_INIT_GLOBAL (FILE *@var{file}, int @var{verbose}, int @var{old_max_uid})
6421 This hook is executed by the scheduler after function level initializations.
6422 @var{file} is either a null pointer, or a stdio stream to write any debug output to.
6423 @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6424 @var{old_max_uid} is the maximum insn uid when scheduling begins.
6427 @deftypefn {Target Hook} void TARGET_SCHED_FINISH_GLOBAL (FILE *@var{file}, int @var{verbose})
6428 This is the cleanup hook corresponding to @code{TARGET_SCHED_INIT_GLOBAL}.
6429 @var{file} is either a null pointer, or a stdio stream to write any debug output to.
6430 @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6433 @deftypefn {Target Hook} int TARGET_SCHED_DFA_PRE_CYCLE_INSN (void)
6434 The hook returns an RTL insn. The automaton state used in the
6435 pipeline hazard recognizer is changed as if the insn were scheduled
6436 when the new simulated processor cycle starts. Usage of the hook may
6437 simplify the automaton pipeline description for some @acronym{VLIW}
6438 processors. If the hook is defined, it is used only for the automaton
6439 based pipeline description. The default is not to change the state
6440 when the new simulated processor cycle starts.
6443 @deftypefn {Target Hook} void TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN (void)
6444 The hook can be used to initialize data used by the previous hook.
6447 @deftypefn {Target Hook} int TARGET_SCHED_DFA_POST_CYCLE_INSN (void)
6448 The hook is analogous to @samp{TARGET_SCHED_DFA_PRE_CYCLE_INSN} but used
6449 to changed the state as if the insn were scheduled when the new
6450 simulated processor cycle finishes.
6453 @deftypefn {Target Hook} void TARGET_SCHED_INIT_DFA_POST_CYCLE_INSN (void)
6454 The hook is analogous to @samp{TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN} but
6455 used to initialize data used by the previous hook.
6458 @deftypefn {Target Hook} void TARGET_SCHED_DFA_PRE_CYCLE_ADVANCE (void)
6459 The hook to notify target that the current simulated cycle is about to finish.
6460 The hook is analogous to @samp{TARGET_SCHED_DFA_PRE_CYCLE_INSN} but used
6461 to change the state in more complicated situations - e.g., when advancing
6462 state on a single insn is not enough.
6465 @deftypefn {Target Hook} void TARGET_SCHED_DFA_POST_CYCLE_ADVANCE (void)
6466 The hook to notify target that new simulated cycle has just started.
6467 The hook is analogous to @samp{TARGET_SCHED_DFA_POST_CYCLE_INSN} but used
6468 to change the state in more complicated situations - e.g., when advancing
6469 state on a single insn is not enough.
6472 @deftypefn {Target Hook} int TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD (void)
6473 This hook controls better choosing an insn from the ready insn queue
6474 for the @acronym{DFA}-based insn scheduler. Usually the scheduler
6475 chooses the first insn from the queue. If the hook returns a positive
6476 value, an additional scheduler code tries all permutations of
6477 @samp{TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD ()}
6478 subsequent ready insns to choose an insn whose issue will result in
6479 maximal number of issued insns on the same cycle. For the
6480 @acronym{VLIW} processor, the code could actually solve the problem of
6481 packing simple insns into the @acronym{VLIW} insn. Of course, if the
6482 rules of @acronym{VLIW} packing are described in the automaton.
6484 This code also could be used for superscalar @acronym{RISC}
6485 processors. Let us consider a superscalar @acronym{RISC} processor
6486 with 3 pipelines. Some insns can be executed in pipelines @var{A} or
6487 @var{B}, some insns can be executed only in pipelines @var{B} or
6488 @var{C}, and one insn can be executed in pipeline @var{B}. The
6489 processor may issue the 1st insn into @var{A} and the 2nd one into
6490 @var{B}. In this case, the 3rd insn will wait for freeing @var{B}
6491 until the next cycle. If the scheduler issues the 3rd insn the first,
6492 the processor could issue all 3 insns per cycle.
6494 Actually this code demonstrates advantages of the automaton based
6495 pipeline hazard recognizer. We try quickly and easy many insn
6496 schedules to choose the best one.
6498 The default is no multipass scheduling.
6501 @deftypefn {Target Hook} int TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD (rtx)
6503 This hook controls what insns from the ready insn queue will be
6504 considered for the multipass insn scheduling. If the hook returns
6505 zero for insn passed as the parameter, the insn will be not chosen to
6508 The default is that any ready insns can be chosen to be issued.
6511 @deftypefn {Target Hook} int TARGET_SCHED_DFA_NEW_CYCLE (FILE *, int, rtx, int, int, int *)
6513 This hook is called by the insn scheduler before issuing insn passed
6514 as the third parameter on given cycle. If the hook returns nonzero,
6515 the insn is not issued on given processors cycle. Instead of that,
6516 the processor cycle is advanced. If the value passed through the last
6517 parameter is zero, the insn ready queue is not sorted on the new cycle
6518 start as usually. The first parameter passes file for debugging
6519 output. The second one passes the scheduler verbose level of the
6520 debugging output. The forth and the fifth parameter values are
6521 correspondingly processor cycle on which the previous insn has been
6522 issued and the current processor cycle.
6525 @deftypefn {Target Hook} bool TARGET_SCHED_IS_COSTLY_DEPENDENCE (struct dep_def *@var{_dep}, int @var{cost}, int @var{distance})
6526 This hook is used to define which dependences are considered costly by
6527 the target, so costly that it is not advisable to schedule the insns that
6528 are involved in the dependence too close to one another. The parameters
6529 to this hook are as follows: The first parameter @var{_dep} is the dependence
6530 being evaluated. The second parameter @var{cost} is the cost of the
6531 dependence, and the third
6532 parameter @var{distance} is the distance in cycles between the two insns.
6533 The hook returns @code{true} if considering the distance between the two
6534 insns the dependence between them is considered costly by the target,
6535 and @code{false} otherwise.
6537 Defining this hook can be useful in multiple-issue out-of-order machines,
6538 where (a) it's practically hopeless to predict the actual data/resource
6539 delays, however: (b) there's a better chance to predict the actual grouping
6540 that will be formed, and (c) correctly emulating the grouping can be very
6541 important. In such targets one may want to allow issuing dependent insns
6542 closer to one another---i.e., closer than the dependence distance; however,
6543 not in cases of "costly dependences", which this hooks allows to define.
6546 @deftypefn {Target Hook} void TARGET_SCHED_H_I_D_EXTENDED (void)
6547 This hook is called by the insn scheduler after emitting a new instruction to
6548 the instruction stream. The hook notifies a target backend to extend its
6549 per instruction data structures.
6552 @deftypefn {Target Hook} void * TARGET_SCHED_ALLOC_SCHED_CONTEXT (void)
6553 Return a pointer to a store large enough to hold target scheduling context.
6556 @deftypefn {Target Hook} void TARGET_SCHED_INIT_SCHED_CONTEXT (void *@var{tc}, bool @var{clean_p})
6557 Initialize store pointed to by @var{tc} to hold target scheduling context.
6558 It @var{clean_p} is true then initialize @var{tc} as if scheduler is at the
6559 beginning of the block. Otherwise, make a copy of the current context in
6563 @deftypefn {Target Hook} void TARGET_SCHED_SET_SCHED_CONTEXT (void *@var{tc})
6564 Copy target scheduling context pointer to by @var{tc} to the current context.
6567 @deftypefn {Target Hook} void TARGET_SCHED_CLEAR_SCHED_CONTEXT (void *@var{tc})
6568 Deallocate internal data in target scheduling context pointed to by @var{tc}.
6571 @deftypefn {Target Hook} void TARGET_SCHED_FREE_SCHED_CONTEXT (void *@var{tc})
6572 Deallocate a store for target scheduling context pointed to by @var{tc}.
6575 @deftypefn {Target Hook} void * TARGET_SCHED_ALLOC_SCHED_CONTEXT (void)
6576 Return a pointer to a store large enough to hold target scheduling context.
6579 @deftypefn {Target Hook} void TARGET_SCHED_INIT_SCHED_CONTEXT (void *@var{tc}, bool @var{clean_p})
6580 Initialize store pointed to by @var{tc} to hold target scheduling context.
6581 It @var{clean_p} is true then initialize @var{tc} as if scheduler is at the
6582 beginning of the block. Otherwise, make a copy of the current context in
6586 @deftypefn {Target Hook} void TARGET_SCHED_SET_SCHED_CONTEXT (void *@var{tc})
6587 Copy target scheduling context pointer to by @var{tc} to the current context.
6590 @deftypefn {Target Hook} void TARGET_SCHED_CLEAR_SCHED_CONTEXT (void *@var{tc})
6591 Deallocate internal data in target scheduling context pointed to by @var{tc}.
6594 @deftypefn {Target Hook} void TARGET_SCHED_FREE_SCHED_CONTEXT (void *@var{tc})
6595 Deallocate a store for target scheduling context pointed to by @var{tc}.
6598 @deftypefn {Target Hook} int TARGET_SCHED_SPECULATE_INSN (rtx @var{insn}, int @var{request}, rtx *@var{new_pat})
6599 This hook is called by the insn scheduler when @var{insn} has only
6600 speculative dependencies and therefore can be scheduled speculatively.
6601 The hook is used to check if the pattern of @var{insn} has a speculative
6602 version and, in case of successful check, to generate that speculative
6603 pattern. The hook should return 1, if the instruction has a speculative form,
6604 or -1, if it doesn't. @var{request} describes the type of requested
6605 speculation. If the return value equals 1 then @var{new_pat} is assigned
6606 the generated speculative pattern.
6609 @deftypefn {Target Hook} int TARGET_SCHED_NEEDS_BLOCK_P (rtx @var{insn})
6610 This hook is called by the insn scheduler during generation of recovery code
6611 for @var{insn}. It should return nonzero, if the corresponding check
6612 instruction should branch to recovery code, or zero otherwise.
6615 @deftypefn {Target Hook} rtx TARGET_SCHED_GEN_CHECK (rtx @var{insn}, rtx @var{label}, int @var{mutate_p})
6616 This hook is called by the insn scheduler to generate a pattern for recovery
6617 check instruction. If @var{mutate_p} is zero, then @var{insn} is a
6618 speculative instruction for which the check should be generated.
6619 @var{label} is either a label of a basic block, where recovery code should
6620 be emitted, or a null pointer, when requested check doesn't branch to
6621 recovery code (a simple check). If @var{mutate_p} is nonzero, then
6622 a pattern for a branchy check corresponding to a simple check denoted by
6623 @var{insn} should be generated. In this case @var{label} can't be null.
6626 @deftypefn {Target Hook} int TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD_SPEC (rtx @var{insn})
6627 This hook is used as a workaround for
6628 @samp{TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD} not being
6629 called on the first instruction of the ready list. The hook is used to
6630 discard speculative instruction that stand first in the ready list from
6631 being scheduled on the current cycle. For non-speculative instructions,
6632 the hook should always return nonzero. For example, in the ia64 backend
6633 the hook is used to cancel data speculative insns when the ALAT table
6637 @deftypefn {Target Hook} void TARGET_SCHED_SET_SCHED_FLAGS (unsigned int *@var{flags}, spec_info_t @var{spec_info})
6638 This hook is used by the insn scheduler to find out what features should be
6639 enabled/used. @var{flags} initially may have either the SCHED_RGN or SCHED_EBB
6640 bit set. This denotes the scheduler pass for which the data should be
6641 provided. The target backend should modify @var{flags} by modifying
6642 the bits corresponding to the following features: USE_DEPS_LIST, USE_GLAT,
6643 DETACH_LIFE_INFO, and DO_SPECULATION@. For the DO_SPECULATION feature
6644 an additional structure @var{spec_info} should be filled by the target.
6645 The structure describes speculation types that can be used in the scheduler.
6648 @deftypefn {Target Hook} int TARGET_SCHED_SMS_RES_MII (struct ddg *@var{g})
6649 This hook is called by the swing modulo scheduler to calculate a
6650 resource-based lower bound which is based on the resources available in
6651 the machine and the resources required by each instruction. The target
6652 backend can use @var{g} to calculate such bound. A very simple lower
6653 bound will be used in case this hook is not implemented: the total number
6654 of instructions divided by the issue rate.
6658 @section Dividing the Output into Sections (Texts, Data, @dots{})
6659 @c the above section title is WAY too long. maybe cut the part between
6660 @c the (...)? --mew 10feb93
6662 An object file is divided into sections containing different types of
6663 data. In the most common case, there are three sections: the @dfn{text
6664 section}, which holds instructions and read-only data; the @dfn{data
6665 section}, which holds initialized writable data; and the @dfn{bss
6666 section}, which holds uninitialized data. Some systems have other kinds
6669 @file{varasm.c} provides several well-known sections, such as
6670 @code{text_section}, @code{data_section} and @code{bss_section}.
6671 The normal way of controlling a @code{@var{foo}_section} variable
6672 is to define the associated @code{@var{FOO}_SECTION_ASM_OP} macro,
6673 as described below. The macros are only read once, when @file{varasm.c}
6674 initializes itself, so their values must be run-time constants.
6675 They may however depend on command-line flags.
6677 @emph{Note:} Some run-time files, such @file{crtstuff.c}, also make
6678 use of the @code{@var{FOO}_SECTION_ASM_OP} macros, and expect them
6679 to be string literals.
6681 Some assemblers require a different string to be written every time a
6682 section is selected. If your assembler falls into this category, you
6683 should define the @code{TARGET_ASM_INIT_SECTIONS} hook and use
6684 @code{get_unnamed_section} to set up the sections.
6686 You must always create a @code{text_section}, either by defining
6687 @code{TEXT_SECTION_ASM_OP} or by initializing @code{text_section}
6688 in @code{TARGET_ASM_INIT_SECTIONS}. The same is true of
6689 @code{data_section} and @code{DATA_SECTION_ASM_OP}. If you do not
6690 create a distinct @code{readonly_data_section}, the default is to
6691 reuse @code{text_section}.
6693 All the other @file{varasm.c} sections are optional, and are null
6694 if the target does not provide them.
6696 @defmac TEXT_SECTION_ASM_OP
6697 A C expression whose value is a string, including spacing, containing the
6698 assembler operation that should precede instructions and read-only data.
6699 Normally @code{"\t.text"} is right.
6702 @defmac HOT_TEXT_SECTION_NAME
6703 If defined, a C string constant for the name of the section containing most
6704 frequently executed functions of the program. If not defined, GCC will provide
6705 a default definition if the target supports named sections.
6708 @defmac UNLIKELY_EXECUTED_TEXT_SECTION_NAME
6709 If defined, a C string constant for the name of the section containing unlikely
6710 executed functions in the program.
6713 @defmac DATA_SECTION_ASM_OP
6714 A C expression whose value is a string, including spacing, containing the
6715 assembler operation to identify the following data as writable initialized
6716 data. Normally @code{"\t.data"} is right.
6719 @defmac SDATA_SECTION_ASM_OP
6720 If defined, a C expression whose value is a string, including spacing,
6721 containing the assembler operation to identify the following data as
6722 initialized, writable small data.
6725 @defmac READONLY_DATA_SECTION_ASM_OP
6726 A C expression whose value is a string, including spacing, containing the
6727 assembler operation to identify the following data as read-only initialized
6731 @defmac BSS_SECTION_ASM_OP
6732 If defined, a C expression whose value is a string, including spacing,
6733 containing the assembler operation to identify the following data as
6734 uninitialized global data. If not defined, and neither
6735 @code{ASM_OUTPUT_BSS} nor @code{ASM_OUTPUT_ALIGNED_BSS} are defined,
6736 uninitialized global data will be output in the data section if
6737 @option{-fno-common} is passed, otherwise @code{ASM_OUTPUT_COMMON} will be
6741 @defmac SBSS_SECTION_ASM_OP
6742 If defined, a C expression whose value is a string, including spacing,
6743 containing the assembler operation to identify the following data as
6744 uninitialized, writable small data.
6747 @defmac INIT_SECTION_ASM_OP
6748 If defined, a C expression whose value is a string, including spacing,
6749 containing the assembler operation to identify the following data as
6750 initialization code. If not defined, GCC will assume such a section does
6751 not exist. This section has no corresponding @code{init_section}
6752 variable; it is used entirely in runtime code.
6755 @defmac FINI_SECTION_ASM_OP
6756 If defined, a C expression whose value is a string, including spacing,
6757 containing the assembler operation to identify the following data as
6758 finalization code. If not defined, GCC will assume such a section does
6759 not exist. This section has no corresponding @code{fini_section}
6760 variable; it is used entirely in runtime code.
6763 @defmac INIT_ARRAY_SECTION_ASM_OP
6764 If defined, a C expression whose value is a string, including spacing,
6765 containing the assembler operation to identify the following data as
6766 part of the @code{.init_array} (or equivalent) section. If not
6767 defined, GCC will assume such a section does not exist. Do not define
6768 both this macro and @code{INIT_SECTION_ASM_OP}.
6771 @defmac FINI_ARRAY_SECTION_ASM_OP
6772 If defined, a C expression whose value is a string, including spacing,
6773 containing the assembler operation to identify the following data as
6774 part of the @code{.fini_array} (or equivalent) section. If not
6775 defined, GCC will assume such a section does not exist. Do not define
6776 both this macro and @code{FINI_SECTION_ASM_OP}.
6779 @defmac CRT_CALL_STATIC_FUNCTION (@var{section_op}, @var{function})
6780 If defined, an ASM statement that switches to a different section
6781 via @var{section_op}, calls @var{function}, and switches back to
6782 the text section. This is used in @file{crtstuff.c} if
6783 @code{INIT_SECTION_ASM_OP} or @code{FINI_SECTION_ASM_OP} to calls
6784 to initialization and finalization functions from the init and fini
6785 sections. By default, this macro uses a simple function call. Some
6786 ports need hand-crafted assembly code to avoid dependencies on
6787 registers initialized in the function prologue or to ensure that
6788 constant pools don't end up too far way in the text section.
6791 @defmac TARGET_LIBGCC_SDATA_SECTION
6792 If defined, a string which names the section into which small
6793 variables defined in crtstuff and libgcc should go. This is useful
6794 when the target has options for optimizing access to small data, and
6795 you want the crtstuff and libgcc routines to be conservative in what
6796 they expect of your application yet liberal in what your application
6797 expects. For example, for targets with a @code{.sdata} section (like
6798 MIPS), you could compile crtstuff with @code{-G 0} so that it doesn't
6799 require small data support from your application, but use this macro
6800 to put small data into @code{.sdata} so that your application can
6801 access these variables whether it uses small data or not.
6804 @defmac FORCE_CODE_SECTION_ALIGN
6805 If defined, an ASM statement that aligns a code section to some
6806 arbitrary boundary. This is used to force all fragments of the
6807 @code{.init} and @code{.fini} sections to have to same alignment
6808 and thus prevent the linker from having to add any padding.
6811 @defmac JUMP_TABLES_IN_TEXT_SECTION
6812 Define this macro to be an expression with a nonzero value if jump
6813 tables (for @code{tablejump} insns) should be output in the text
6814 section, along with the assembler instructions. Otherwise, the
6815 readonly data section is used.
6817 This macro is irrelevant if there is no separate readonly data section.
6820 @deftypefn {Target Hook} void TARGET_ASM_INIT_SECTIONS (void)
6821 Define this hook if you need to do something special to set up the
6822 @file{varasm.c} sections, or if your target has some special sections
6823 of its own that you need to create.
6825 GCC calls this hook after processing the command line, but before writing
6826 any assembly code, and before calling any of the section-returning hooks
6830 @deftypefn {Target Hook} TARGET_ASM_RELOC_RW_MASK (void)
6831 Return a mask describing how relocations should be treated when
6832 selecting sections. Bit 1 should be set if global relocations
6833 should be placed in a read-write section; bit 0 should be set if
6834 local relocations should be placed in a read-write section.
6836 The default version of this function returns 3 when @option{-fpic}
6837 is in effect, and 0 otherwise. The hook is typically redefined
6838 when the target cannot support (some kinds of) dynamic relocations
6839 in read-only sections even in executables.
6842 @deftypefn {Target Hook} {section *} TARGET_ASM_SELECT_SECTION (tree @var{exp}, int @var{reloc}, unsigned HOST_WIDE_INT @var{align})
6843 Return the section into which @var{exp} should be placed. You can
6844 assume that @var{exp} is either a @code{VAR_DECL} node or a constant of
6845 some sort. @var{reloc} indicates whether the initial value of @var{exp}
6846 requires link-time relocations. Bit 0 is set when variable contains
6847 local relocations only, while bit 1 is set for global relocations.
6848 @var{align} is the constant alignment in bits.
6850 The default version of this function takes care of putting read-only
6851 variables in @code{readonly_data_section}.
6853 See also @var{USE_SELECT_SECTION_FOR_FUNCTIONS}.
6856 @defmac USE_SELECT_SECTION_FOR_FUNCTIONS
6857 Define this macro if you wish TARGET_ASM_SELECT_SECTION to be called
6858 for @code{FUNCTION_DECL}s as well as for variables and constants.
6860 In the case of a @code{FUNCTION_DECL}, @var{reloc} will be zero if the
6861 function has been determined to be likely to be called, and nonzero if
6862 it is unlikely to be called.
6865 @deftypefn {Target Hook} void TARGET_ASM_UNIQUE_SECTION (tree @var{decl}, int @var{reloc})
6866 Build up a unique section name, expressed as a @code{STRING_CST} node,
6867 and assign it to @samp{DECL_SECTION_NAME (@var{decl})}.
6868 As with @code{TARGET_ASM_SELECT_SECTION}, @var{reloc} indicates whether
6869 the initial value of @var{exp} requires link-time relocations.
6871 The default version of this function appends the symbol name to the
6872 ELF section name that would normally be used for the symbol. For
6873 example, the function @code{foo} would be placed in @code{.text.foo}.
6874 Whatever the actual target object format, this is often good enough.
6877 @deftypefn {Target Hook} {section *} TARGET_ASM_FUNCTION_RODATA_SECTION (tree @var{decl})
6878 Return the readonly data section associated with
6879 @samp{DECL_SECTION_NAME (@var{decl})}.
6880 The default version of this function selects @code{.gnu.linkonce.r.name} if
6881 the function's section is @code{.gnu.linkonce.t.name}, @code{.rodata.name}
6882 if function is in @code{.text.name}, and the normal readonly-data section
6886 @deftypefn {Target Hook} {section *} TARGET_ASM_SELECT_RTX_SECTION (enum machine_mode @var{mode}, rtx @var{x}, unsigned HOST_WIDE_INT @var{align})
6887 Return the section into which a constant @var{x}, of mode @var{mode},
6888 should be placed. You can assume that @var{x} is some kind of
6889 constant in RTL@. The argument @var{mode} is redundant except in the
6890 case of a @code{const_int} rtx. @var{align} is the constant alignment
6893 The default version of this function takes care of putting symbolic
6894 constants in @code{flag_pic} mode in @code{data_section} and everything
6895 else in @code{readonly_data_section}.
6898 @deftypefn {Target Hook} void TARGET_MANGLE_DECL_ASSEMBLER_NAME (tree @var{decl}, tree @var{id})
6899 Define this hook if you need to postprocess the assembler name generated
6900 by target-independent code. The @var{id} provided to this hook will be
6901 the computed name (e.g., the macro @code{DECL_NAME} of the @var{decl} in C,
6902 or the mangled name of the @var{decl} in C++). The return value of the
6903 hook is an @code{IDENTIFIER_NODE} for the appropriate mangled name on
6904 your target system. The default implementation of this hook just
6905 returns the @var{id} provided.
6908 @deftypefn {Target Hook} void TARGET_ENCODE_SECTION_INFO (tree @var{decl}, rtx @var{rtl}, int @var{new_decl_p})
6909 Define this hook if references to a symbol or a constant must be
6910 treated differently depending on something about the variable or
6911 function named by the symbol (such as what section it is in).
6913 The hook is executed immediately after rtl has been created for
6914 @var{decl}, which may be a variable or function declaration or
6915 an entry in the constant pool. In either case, @var{rtl} is the
6916 rtl in question. Do @emph{not} use @code{DECL_RTL (@var{decl})}
6917 in this hook; that field may not have been initialized yet.
6919 In the case of a constant, it is safe to assume that the rtl is
6920 a @code{mem} whose address is a @code{symbol_ref}. Most decls
6921 will also have this form, but that is not guaranteed. Global
6922 register variables, for instance, will have a @code{reg} for their
6923 rtl. (Normally the right thing to do with such unusual rtl is
6926 The @var{new_decl_p} argument will be true if this is the first time
6927 that @code{TARGET_ENCODE_SECTION_INFO} has been invoked on this decl. It will
6928 be false for subsequent invocations, which will happen for duplicate
6929 declarations. Whether or not anything must be done for the duplicate
6930 declaration depends on whether the hook examines @code{DECL_ATTRIBUTES}.
6931 @var{new_decl_p} is always true when the hook is called for a constant.
6933 @cindex @code{SYMBOL_REF_FLAG}, in @code{TARGET_ENCODE_SECTION_INFO}
6934 The usual thing for this hook to do is to record flags in the
6935 @code{symbol_ref}, using @code{SYMBOL_REF_FLAG} or @code{SYMBOL_REF_FLAGS}.
6936 Historically, the name string was modified if it was necessary to
6937 encode more than one bit of information, but this practice is now
6938 discouraged; use @code{SYMBOL_REF_FLAGS}.
6940 The default definition of this hook, @code{default_encode_section_info}
6941 in @file{varasm.c}, sets a number of commonly-useful bits in
6942 @code{SYMBOL_REF_FLAGS}. Check whether the default does what you need
6943 before overriding it.
6946 @deftypefn {Target Hook} const char *TARGET_STRIP_NAME_ENCODING (const char *name)
6947 Decode @var{name} and return the real name part, sans
6948 the characters that @code{TARGET_ENCODE_SECTION_INFO}
6952 @deftypefn {Target Hook} bool TARGET_IN_SMALL_DATA_P (tree @var{exp})
6953 Returns true if @var{exp} should be placed into a ``small data'' section.
6954 The default version of this hook always returns false.
6957 @deftypevr {Target Hook} bool TARGET_HAVE_SRODATA_SECTION
6958 Contains the value true if the target places read-only
6959 ``small data'' into a separate section. The default value is false.
6962 @deftypefn {Target Hook} bool TARGET_BINDS_LOCAL_P (tree @var{exp})
6963 Returns true if @var{exp} names an object for which name resolution
6964 rules must resolve to the current ``module'' (dynamic shared library
6965 or executable image).
6967 The default version of this hook implements the name resolution rules
6968 for ELF, which has a looser model of global name binding than other
6969 currently supported object file formats.
6972 @deftypevr {Target Hook} bool TARGET_HAVE_TLS
6973 Contains the value true if the target supports thread-local storage.
6974 The default value is false.
6979 @section Position Independent Code
6980 @cindex position independent code
6983 This section describes macros that help implement generation of position
6984 independent code. Simply defining these macros is not enough to
6985 generate valid PIC; you must also add support to the hook
6986 @code{TARGET_LEGITIMATE_ADDRESS_P} and to the macro
6987 @code{PRINT_OPERAND_ADDRESS}, as well as @code{LEGITIMIZE_ADDRESS}. You
6988 must modify the definition of @samp{movsi} to do something appropriate
6989 when the source operand contains a symbolic address. You may also
6990 need to alter the handling of switch statements so that they use
6992 @c i rearranged the order of the macros above to try to force one of
6993 @c them to the next line, to eliminate an overfull hbox. --mew 10feb93
6995 @defmac PIC_OFFSET_TABLE_REGNUM
6996 The register number of the register used to address a table of static
6997 data addresses in memory. In some cases this register is defined by a
6998 processor's ``application binary interface'' (ABI)@. When this macro
6999 is defined, RTL is generated for this register once, as with the stack
7000 pointer and frame pointer registers. If this macro is not defined, it
7001 is up to the machine-dependent files to allocate such a register (if
7002 necessary). Note that this register must be fixed when in use (e.g.@:
7003 when @code{flag_pic} is true).
7006 @defmac PIC_OFFSET_TABLE_REG_CALL_CLOBBERED
7007 Define this macro if the register defined by
7008 @code{PIC_OFFSET_TABLE_REGNUM} is clobbered by calls. Do not define
7009 this macro if @code{PIC_OFFSET_TABLE_REGNUM} is not defined.
7012 @defmac LEGITIMATE_PIC_OPERAND_P (@var{x})
7013 A C expression that is nonzero if @var{x} is a legitimate immediate
7014 operand on the target machine when generating position independent code.
7015 You can assume that @var{x} satisfies @code{CONSTANT_P}, so you need not
7016 check this. You can also assume @var{flag_pic} is true, so you need not
7017 check it either. You need not define this macro if all constants
7018 (including @code{SYMBOL_REF}) can be immediate operands when generating
7019 position independent code.
7022 @node Assembler Format
7023 @section Defining the Output Assembler Language
7025 This section describes macros whose principal purpose is to describe how
7026 to write instructions in assembler language---rather than what the
7030 * File Framework:: Structural information for the assembler file.
7031 * Data Output:: Output of constants (numbers, strings, addresses).
7032 * Uninitialized Data:: Output of uninitialized variables.
7033 * Label Output:: Output and generation of labels.
7034 * Initialization:: General principles of initialization
7035 and termination routines.
7036 * Macros for Initialization::
7037 Specific macros that control the handling of
7038 initialization and termination routines.
7039 * Instruction Output:: Output of actual instructions.
7040 * Dispatch Tables:: Output of jump tables.
7041 * Exception Region Output:: Output of exception region code.
7042 * Alignment Output:: Pseudo ops for alignment and skipping data.
7045 @node File Framework
7046 @subsection The Overall Framework of an Assembler File
7047 @cindex assembler format
7048 @cindex output of assembler code
7050 @c prevent bad page break with this line
7051 This describes the overall framework of an assembly file.
7053 @deftypefn {Target Hook} void TARGET_ASM_FILE_START ()
7054 @findex default_file_start
7055 Output to @code{asm_out_file} any text which the assembler expects to
7056 find at the beginning of a file. The default behavior is controlled
7057 by two flags, documented below. Unless your target's assembler is
7058 quite unusual, if you override the default, you should call
7059 @code{default_file_start} at some point in your target hook. This
7060 lets other target files rely on these variables.
7063 @deftypevr {Target Hook} bool TARGET_ASM_FILE_START_APP_OFF
7064 If this flag is true, the text of the macro @code{ASM_APP_OFF} will be
7065 printed as the very first line in the assembly file, unless
7066 @option{-fverbose-asm} is in effect. (If that macro has been defined
7067 to the empty string, this variable has no effect.) With the normal
7068 definition of @code{ASM_APP_OFF}, the effect is to notify the GNU
7069 assembler that it need not bother stripping comments or extra
7070 whitespace from its input. This allows it to work a bit faster.
7072 The default is false. You should not set it to true unless you have
7073 verified that your port does not generate any extra whitespace or
7074 comments that will cause GAS to issue errors in NO_APP mode.
7077 @deftypevr {Target Hook} bool TARGET_ASM_FILE_START_FILE_DIRECTIVE
7078 If this flag is true, @code{output_file_directive} will be called
7079 for the primary source file, immediately after printing
7080 @code{ASM_APP_OFF} (if that is enabled). Most ELF assemblers expect
7081 this to be done. The default is false.
7084 @deftypefn {Target Hook} void TARGET_ASM_FILE_END ()
7085 Output to @code{asm_out_file} any text which the assembler expects
7086 to find at the end of a file. The default is to output nothing.
7089 @deftypefun void file_end_indicate_exec_stack ()
7090 Some systems use a common convention, the @samp{.note.GNU-stack}
7091 special section, to indicate whether or not an object file relies on
7092 the stack being executable. If your system uses this convention, you
7093 should define @code{TARGET_ASM_FILE_END} to this function. If you
7094 need to do other things in that hook, have your hook function call
7098 @defmac ASM_COMMENT_START
7099 A C string constant describing how to begin a comment in the target
7100 assembler language. The compiler assumes that the comment will end at
7101 the end of the line.
7105 A C string constant for text to be output before each @code{asm}
7106 statement or group of consecutive ones. Normally this is
7107 @code{"#APP"}, which is a comment that has no effect on most
7108 assemblers but tells the GNU assembler that it must check the lines
7109 that follow for all valid assembler constructs.
7113 A C string constant for text to be output after each @code{asm}
7114 statement or group of consecutive ones. Normally this is
7115 @code{"#NO_APP"}, which tells the GNU assembler to resume making the
7116 time-saving assumptions that are valid for ordinary compiler output.
7119 @defmac ASM_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name})
7120 A C statement to output COFF information or DWARF debugging information
7121 which indicates that filename @var{name} is the current source file to
7122 the stdio stream @var{stream}.
7124 This macro need not be defined if the standard form of output
7125 for the file format in use is appropriate.
7128 @defmac OUTPUT_QUOTED_STRING (@var{stream}, @var{string})
7129 A C statement to output the string @var{string} to the stdio stream
7130 @var{stream}. If you do not call the function @code{output_quoted_string}
7131 in your config files, GCC will only call it to output filenames to
7132 the assembler source. So you can use it to canonicalize the format
7133 of the filename using this macro.
7136 @defmac ASM_OUTPUT_IDENT (@var{stream}, @var{string})
7137 A C statement to output something to the assembler file to handle a
7138 @samp{#ident} directive containing the text @var{string}. If this
7139 macro is not defined, nothing is output for a @samp{#ident} directive.
7142 @deftypefn {Target Hook} void TARGET_ASM_NAMED_SECTION (const char *@var{name}, unsigned int @var{flags}, unsigned int @var{align})
7143 Output assembly directives to switch to section @var{name}. The section
7144 should have attributes as specified by @var{flags}, which is a bit mask
7145 of the @code{SECTION_*} flags defined in @file{output.h}. If @var{align}
7146 is nonzero, it contains an alignment in bytes to be used for the section,
7147 otherwise some target default should be used. Only targets that must
7148 specify an alignment within the section directive need pay attention to
7149 @var{align} -- we will still use @code{ASM_OUTPUT_ALIGN}.
7152 @deftypefn {Target Hook} bool TARGET_HAVE_NAMED_SECTIONS
7153 This flag is true if the target supports @code{TARGET_ASM_NAMED_SECTION}.
7156 @anchor{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}
7157 @deftypefn {Target Hook} bool TARGET_HAVE_SWITCHABLE_BSS_SECTIONS
7158 This flag is true if we can create zeroed data by switching to a BSS
7159 section and then using @code{ASM_OUTPUT_SKIP} to allocate the space.
7160 This is true on most ELF targets.
7163 @deftypefn {Target Hook} {unsigned int} TARGET_SECTION_TYPE_FLAGS (tree @var{decl}, const char *@var{name}, int @var{reloc})
7164 Choose a set of section attributes for use by @code{TARGET_ASM_NAMED_SECTION}
7165 based on a variable or function decl, a section name, and whether or not the
7166 declaration's initializer may contain runtime relocations. @var{decl} may be
7167 null, in which case read-write data should be assumed.
7169 The default version of this function handles choosing code vs data,
7170 read-only vs read-write data, and @code{flag_pic}. You should only
7171 need to override this if your target has special flags that might be
7172 set via @code{__attribute__}.
7175 @deftypefn {Target Hook} {int} TARGET_ASM_RECORD_GCC_SWITCHES (print_switch_type @var{type}, const char * @var{text})
7176 Provides the target with the ability to record the gcc command line
7177 switches that have been passed to the compiler, and options that are
7178 enabled. The @var{type} argument specifies what is being recorded.
7179 It can take the following values:
7182 @item SWITCH_TYPE_PASSED
7183 @var{text} is a command line switch that has been set by the user.
7185 @item SWITCH_TYPE_ENABLED
7186 @var{text} is an option which has been enabled. This might be as a
7187 direct result of a command line switch, or because it is enabled by
7188 default or because it has been enabled as a side effect of a different
7189 command line switch. For example, the @option{-O2} switch enables
7190 various different individual optimization passes.
7192 @item SWITCH_TYPE_DESCRIPTIVE
7193 @var{text} is either NULL or some descriptive text which should be
7194 ignored. If @var{text} is NULL then it is being used to warn the
7195 target hook that either recording is starting or ending. The first
7196 time @var{type} is SWITCH_TYPE_DESCRIPTIVE and @var{text} is NULL, the
7197 warning is for start up and the second time the warning is for
7198 wind down. This feature is to allow the target hook to make any
7199 necessary preparations before it starts to record switches and to
7200 perform any necessary tidying up after it has finished recording
7203 @item SWITCH_TYPE_LINE_START
7204 This option can be ignored by this target hook.
7206 @item SWITCH_TYPE_LINE_END
7207 This option can be ignored by this target hook.
7210 The hook's return value must be zero. Other return values may be
7211 supported in the future.
7213 By default this hook is set to NULL, but an example implementation is
7214 provided for ELF based targets. Called @var{elf_record_gcc_switches},
7215 it records the switches as ASCII text inside a new, string mergeable
7216 section in the assembler output file. The name of the new section is
7217 provided by the @code{TARGET_ASM_RECORD_GCC_SWITCHES_SECTION} target
7221 @deftypefn {Target Hook} {const char *} TARGET_ASM_RECORD_GCC_SWITCHES_SECTION
7222 This is the name of the section that will be created by the example
7223 ELF implementation of the @code{TARGET_ASM_RECORD_GCC_SWITCHES} target
7229 @subsection Output of Data
7232 @deftypevr {Target Hook} {const char *} TARGET_ASM_BYTE_OP
7233 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_HI_OP
7234 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_SI_OP
7235 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_DI_OP
7236 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_TI_OP
7237 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_HI_OP
7238 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_SI_OP
7239 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_DI_OP
7240 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_TI_OP
7241 These hooks specify assembly directives for creating certain kinds
7242 of integer object. The @code{TARGET_ASM_BYTE_OP} directive creates a
7243 byte-sized object, the @code{TARGET_ASM_ALIGNED_HI_OP} one creates an
7244 aligned two-byte object, and so on. Any of the hooks may be
7245 @code{NULL}, indicating that no suitable directive is available.
7247 The compiler will print these strings at the start of a new line,
7248 followed immediately by the object's initial value. In most cases,
7249 the string should contain a tab, a pseudo-op, and then another tab.
7252 @deftypefn {Target Hook} bool TARGET_ASM_INTEGER (rtx @var{x}, unsigned int @var{size}, int @var{aligned_p})
7253 The @code{assemble_integer} function uses this hook to output an
7254 integer object. @var{x} is the object's value, @var{size} is its size
7255 in bytes and @var{aligned_p} indicates whether it is aligned. The
7256 function should return @code{true} if it was able to output the
7257 object. If it returns false, @code{assemble_integer} will try to
7258 split the object into smaller parts.
7260 The default implementation of this hook will use the
7261 @code{TARGET_ASM_BYTE_OP} family of strings, returning @code{false}
7262 when the relevant string is @code{NULL}.
7265 @defmac OUTPUT_ADDR_CONST_EXTRA (@var{stream}, @var{x}, @var{fail})
7266 A C statement to recognize @var{rtx} patterns that
7267 @code{output_addr_const} can't deal with, and output assembly code to
7268 @var{stream} corresponding to the pattern @var{x}. This may be used to
7269 allow machine-dependent @code{UNSPEC}s to appear within constants.
7271 If @code{OUTPUT_ADDR_CONST_EXTRA} fails to recognize a pattern, it must
7272 @code{goto fail}, so that a standard error message is printed. If it
7273 prints an error message itself, by calling, for example,
7274 @code{output_operand_lossage}, it may just complete normally.
7277 @defmac ASM_OUTPUT_ASCII (@var{stream}, @var{ptr}, @var{len})
7278 A C statement to output to the stdio stream @var{stream} an assembler
7279 instruction to assemble a string constant containing the @var{len}
7280 bytes at @var{ptr}. @var{ptr} will be a C expression of type
7281 @code{char *} and @var{len} a C expression of type @code{int}.
7283 If the assembler has a @code{.ascii} pseudo-op as found in the
7284 Berkeley Unix assembler, do not define the macro
7285 @code{ASM_OUTPUT_ASCII}.
7288 @defmac ASM_OUTPUT_FDESC (@var{stream}, @var{decl}, @var{n})
7289 A C statement to output word @var{n} of a function descriptor for
7290 @var{decl}. This must be defined if @code{TARGET_VTABLE_USES_DESCRIPTORS}
7291 is defined, and is otherwise unused.
7294 @defmac CONSTANT_POOL_BEFORE_FUNCTION
7295 You may define this macro as a C expression. You should define the
7296 expression to have a nonzero value if GCC should output the constant
7297 pool for a function before the code for the function, or a zero value if
7298 GCC should output the constant pool after the function. If you do
7299 not define this macro, the usual case, GCC will output the constant
7300 pool before the function.
7303 @defmac ASM_OUTPUT_POOL_PROLOGUE (@var{file}, @var{funname}, @var{fundecl}, @var{size})
7304 A C statement to output assembler commands to define the start of the
7305 constant pool for a function. @var{funname} is a string giving
7306 the name of the function. Should the return type of the function
7307 be required, it can be obtained via @var{fundecl}. @var{size}
7308 is the size, in bytes, of the constant pool that will be written
7309 immediately after this call.
7311 If no constant-pool prefix is required, the usual case, this macro need
7315 @defmac ASM_OUTPUT_SPECIAL_POOL_ENTRY (@var{file}, @var{x}, @var{mode}, @var{align}, @var{labelno}, @var{jumpto})
7316 A C statement (with or without semicolon) to output a constant in the
7317 constant pool, if it needs special treatment. (This macro need not do
7318 anything for RTL expressions that can be output normally.)
7320 The argument @var{file} is the standard I/O stream to output the
7321 assembler code on. @var{x} is the RTL expression for the constant to
7322 output, and @var{mode} is the machine mode (in case @var{x} is a
7323 @samp{const_int}). @var{align} is the required alignment for the value
7324 @var{x}; you should output an assembler directive to force this much
7327 The argument @var{labelno} is a number to use in an internal label for
7328 the address of this pool entry. The definition of this macro is
7329 responsible for outputting the label definition at the proper place.
7330 Here is how to do this:
7333 @code{(*targetm.asm_out.internal_label)} (@var{file}, "LC", @var{labelno});
7336 When you output a pool entry specially, you should end with a
7337 @code{goto} to the label @var{jumpto}. This will prevent the same pool
7338 entry from being output a second time in the usual manner.
7340 You need not define this macro if it would do nothing.
7343 @defmac ASM_OUTPUT_POOL_EPILOGUE (@var{file} @var{funname} @var{fundecl} @var{size})
7344 A C statement to output assembler commands to at the end of the constant
7345 pool for a function. @var{funname} is a string giving the name of the
7346 function. Should the return type of the function be required, you can
7347 obtain it via @var{fundecl}. @var{size} is the size, in bytes, of the
7348 constant pool that GCC wrote immediately before this call.
7350 If no constant-pool epilogue is required, the usual case, you need not
7354 @defmac IS_ASM_LOGICAL_LINE_SEPARATOR (@var{C}, @var{STR})
7355 Define this macro as a C expression which is nonzero if @var{C} is
7356 used as a logical line separator by the assembler. @var{STR} points
7357 to the position in the string where @var{C} was found; this can be used if
7358 a line separator uses multiple characters.
7360 If you do not define this macro, the default is that only
7361 the character @samp{;} is treated as a logical line separator.
7364 @deftypevr {Target Hook} {const char *} TARGET_ASM_OPEN_PAREN
7365 @deftypevrx {Target Hook} {const char *} TARGET_ASM_CLOSE_PAREN
7366 These target hooks are C string constants, describing the syntax in the
7367 assembler for grouping arithmetic expressions. If not overridden, they
7368 default to normal parentheses, which is correct for most assemblers.
7371 These macros are provided by @file{real.h} for writing the definitions
7372 of @code{ASM_OUTPUT_DOUBLE} and the like:
7374 @defmac REAL_VALUE_TO_TARGET_SINGLE (@var{x}, @var{l})
7375 @defmacx REAL_VALUE_TO_TARGET_DOUBLE (@var{x}, @var{l})
7376 @defmacx REAL_VALUE_TO_TARGET_LONG_DOUBLE (@var{x}, @var{l})
7377 @defmacx REAL_VALUE_TO_TARGET_DECIMAL32 (@var{x}, @var{l})
7378 @defmacx REAL_VALUE_TO_TARGET_DECIMAL64 (@var{x}, @var{l})
7379 @defmacx REAL_VALUE_TO_TARGET_DECIMAL128 (@var{x}, @var{l})
7380 These translate @var{x}, of type @code{REAL_VALUE_TYPE}, to the
7381 target's floating point representation, and store its bit pattern in
7382 the variable @var{l}. For @code{REAL_VALUE_TO_TARGET_SINGLE} and
7383 @code{REAL_VALUE_TO_TARGET_DECIMAL32}, this variable should be a
7384 simple @code{long int}. For the others, it should be an array of
7385 @code{long int}. The number of elements in this array is determined
7386 by the size of the desired target floating point data type: 32 bits of
7387 it go in each @code{long int} array element. Each array element holds
7388 32 bits of the result, even if @code{long int} is wider than 32 bits
7389 on the host machine.
7391 The array element values are designed so that you can print them out
7392 using @code{fprintf} in the order they should appear in the target
7396 @node Uninitialized Data
7397 @subsection Output of Uninitialized Variables
7399 Each of the macros in this section is used to do the whole job of
7400 outputting a single uninitialized variable.
7402 @defmac ASM_OUTPUT_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
7403 A C statement (sans semicolon) to output to the stdio stream
7404 @var{stream} the assembler definition of a common-label named
7405 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7406 is the size rounded up to whatever alignment the caller wants. It is
7407 possible that @var{size} may be zero, for instance if a struct with no
7408 other member than a zero-length array is defined. In this case, the
7409 backend must output a symbol definition that allocates at least one
7410 byte, both so that the address of the resulting object does not compare
7411 equal to any other, and because some object formats cannot even express
7412 the concept of a zero-sized common symbol, as that is how they represent
7413 an ordinary undefined external.
7415 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7416 output the name itself; before and after that, output the additional
7417 assembler syntax for defining the name, and a newline.
7419 This macro controls how the assembler definitions of uninitialized
7420 common global variables are output.
7423 @defmac ASM_OUTPUT_ALIGNED_COMMON (@var{stream}, @var{name}, @var{size}, @var{alignment})
7424 Like @code{ASM_OUTPUT_COMMON} except takes the required alignment as a
7425 separate, explicit argument. If you define this macro, it is used in
7426 place of @code{ASM_OUTPUT_COMMON}, and gives you more flexibility in
7427 handling the required alignment of the variable. The alignment is specified
7428 as the number of bits.
7431 @defmac ASM_OUTPUT_ALIGNED_DECL_COMMON (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7432 Like @code{ASM_OUTPUT_ALIGNED_COMMON} except that @var{decl} of the
7433 variable to be output, if there is one, or @code{NULL_TREE} if there
7434 is no corresponding variable. If you define this macro, GCC will use it
7435 in place of both @code{ASM_OUTPUT_COMMON} and
7436 @code{ASM_OUTPUT_ALIGNED_COMMON}. Define this macro when you need to see
7437 the variable's decl in order to chose what to output.
7440 @defmac ASM_OUTPUT_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{rounded})
7441 A C statement (sans semicolon) to output to the stdio stream
7442 @var{stream} the assembler definition of uninitialized global @var{decl} named
7443 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7444 is the size rounded up to whatever alignment the caller wants.
7446 Try to use function @code{asm_output_bss} defined in @file{varasm.c} when
7447 defining this macro. If unable, use the expression
7448 @code{assemble_name (@var{stream}, @var{name})} to output the name itself;
7449 before and after that, output the additional assembler syntax for defining
7450 the name, and a newline.
7452 There are two ways of handling global BSS@. One is to define either
7453 this macro or its aligned counterpart, @code{ASM_OUTPUT_ALIGNED_BSS}.
7454 The other is to have @code{TARGET_ASM_SELECT_SECTION} return a
7455 switchable BSS section (@pxref{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}).
7456 You do not need to do both.
7458 Some languages do not have @code{common} data, and require a
7459 non-common form of global BSS in order to handle uninitialized globals
7460 efficiently. C++ is one example of this. However, if the target does
7461 not support global BSS, the front end may choose to make globals
7462 common in order to save space in the object file.
7465 @defmac ASM_OUTPUT_ALIGNED_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7466 Like @code{ASM_OUTPUT_BSS} except takes the required alignment as a
7467 separate, explicit argument. If you define this macro, it is used in
7468 place of @code{ASM_OUTPUT_BSS}, and gives you more flexibility in
7469 handling the required alignment of the variable. The alignment is specified
7470 as the number of bits.
7472 Try to use function @code{asm_output_aligned_bss} defined in file
7473 @file{varasm.c} when defining this macro.
7476 @defmac ASM_OUTPUT_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded})
7477 A C statement (sans semicolon) to output to the stdio stream
7478 @var{stream} the assembler definition of a local-common-label named
7479 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7480 is the size rounded up to whatever alignment the caller wants.
7482 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7483 output the name itself; before and after that, output the additional
7484 assembler syntax for defining the name, and a newline.
7486 This macro controls how the assembler definitions of uninitialized
7487 static variables are output.
7490 @defmac ASM_OUTPUT_ALIGNED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{alignment})
7491 Like @code{ASM_OUTPUT_LOCAL} except takes the required alignment as a
7492 separate, explicit argument. If you define this macro, it is used in
7493 place of @code{ASM_OUTPUT_LOCAL}, and gives you more flexibility in
7494 handling the required alignment of the variable. The alignment is specified
7495 as the number of bits.
7498 @defmac ASM_OUTPUT_ALIGNED_DECL_LOCAL (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7499 Like @code{ASM_OUTPUT_ALIGNED_DECL} except that @var{decl} of the
7500 variable to be output, if there is one, or @code{NULL_TREE} if there
7501 is no corresponding variable. If you define this macro, GCC will use it
7502 in place of both @code{ASM_OUTPUT_DECL} and
7503 @code{ASM_OUTPUT_ALIGNED_DECL}. Define this macro when you need to see
7504 the variable's decl in order to chose what to output.
7508 @subsection Output and Generation of Labels
7510 @c prevent bad page break with this line
7511 This is about outputting labels.
7513 @findex assemble_name
7514 @defmac ASM_OUTPUT_LABEL (@var{stream}, @var{name})
7515 A C statement (sans semicolon) to output to the stdio stream
7516 @var{stream} the assembler definition of a label named @var{name}.
7517 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7518 output the name itself; before and after that, output the additional
7519 assembler syntax for defining the name, and a newline. A default
7520 definition of this macro is provided which is correct for most systems.
7523 @findex assemble_name_raw
7524 @defmac ASM_OUTPUT_INTERNAL_LABEL (@var{stream}, @var{name})
7525 Identical to @code{ASM_OUTPUT_LABEL}, except that @var{name} is known
7526 to refer to a compiler-generated label. The default definition uses
7527 @code{assemble_name_raw}, which is like @code{assemble_name} except
7528 that it is more efficient.
7532 A C string containing the appropriate assembler directive to specify the
7533 size of a symbol, without any arguments. On systems that use ELF, the
7534 default (in @file{config/elfos.h}) is @samp{"\t.size\t"}; on other
7535 systems, the default is not to define this macro.
7537 Define this macro only if it is correct to use the default definitions
7538 of @code{ASM_OUTPUT_SIZE_DIRECTIVE} and @code{ASM_OUTPUT_MEASURED_SIZE}
7539 for your system. If you need your own custom definitions of those
7540 macros, or if you do not need explicit symbol sizes at all, do not
7544 @defmac ASM_OUTPUT_SIZE_DIRECTIVE (@var{stream}, @var{name}, @var{size})
7545 A C statement (sans semicolon) to output to the stdio stream
7546 @var{stream} a directive telling the assembler that the size of the
7547 symbol @var{name} is @var{size}. @var{size} is a @code{HOST_WIDE_INT}.
7548 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
7552 @defmac ASM_OUTPUT_MEASURED_SIZE (@var{stream}, @var{name})
7553 A C statement (sans semicolon) to output to the stdio stream
7554 @var{stream} a directive telling the assembler to calculate the size of
7555 the symbol @var{name} by subtracting its address from the current
7558 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
7559 provided. The default assumes that the assembler recognizes a special
7560 @samp{.} symbol as referring to the current address, and can calculate
7561 the difference between this and another symbol. If your assembler does
7562 not recognize @samp{.} or cannot do calculations with it, you will need
7563 to redefine @code{ASM_OUTPUT_MEASURED_SIZE} to use some other technique.
7567 A C string containing the appropriate assembler directive to specify the
7568 type of a symbol, without any arguments. On systems that use ELF, the
7569 default (in @file{config/elfos.h}) is @samp{"\t.type\t"}; on other
7570 systems, the default is not to define this macro.
7572 Define this macro only if it is correct to use the default definition of
7573 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
7574 custom definition of this macro, or if you do not need explicit symbol
7575 types at all, do not define this macro.
7578 @defmac TYPE_OPERAND_FMT
7579 A C string which specifies (using @code{printf} syntax) the format of
7580 the second operand to @code{TYPE_ASM_OP}. On systems that use ELF, the
7581 default (in @file{config/elfos.h}) is @samp{"@@%s"}; on other systems,
7582 the default is not to define this macro.
7584 Define this macro only if it is correct to use the default definition of
7585 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
7586 custom definition of this macro, or if you do not need explicit symbol
7587 types at all, do not define this macro.
7590 @defmac ASM_OUTPUT_TYPE_DIRECTIVE (@var{stream}, @var{type})
7591 A C statement (sans semicolon) to output to the stdio stream
7592 @var{stream} a directive telling the assembler that the type of the
7593 symbol @var{name} is @var{type}. @var{type} is a C string; currently,
7594 that string is always either @samp{"function"} or @samp{"object"}, but
7595 you should not count on this.
7597 If you define @code{TYPE_ASM_OP} and @code{TYPE_OPERAND_FMT}, a default
7598 definition of this macro is provided.
7601 @defmac ASM_DECLARE_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl})
7602 A C statement (sans semicolon) to output to the stdio stream
7603 @var{stream} any text necessary for declaring the name @var{name} of a
7604 function which is being defined. This macro is responsible for
7605 outputting the label definition (perhaps using
7606 @code{ASM_OUTPUT_LABEL}). The argument @var{decl} is the
7607 @code{FUNCTION_DECL} tree node representing the function.
7609 If this macro is not defined, then the function name is defined in the
7610 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
7612 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition
7616 @defmac ASM_DECLARE_FUNCTION_SIZE (@var{stream}, @var{name}, @var{decl})
7617 A C statement (sans semicolon) to output to the stdio stream
7618 @var{stream} any text necessary for declaring the size of a function
7619 which is being defined. The argument @var{name} is the name of the
7620 function. The argument @var{decl} is the @code{FUNCTION_DECL} tree node
7621 representing the function.
7623 If this macro is not defined, then the function size is not defined.
7625 You may wish to use @code{ASM_OUTPUT_MEASURED_SIZE} in the definition
7629 @defmac ASM_DECLARE_OBJECT_NAME (@var{stream}, @var{name}, @var{decl})
7630 A C statement (sans semicolon) to output to the stdio stream
7631 @var{stream} any text necessary for declaring the name @var{name} of an
7632 initialized variable which is being defined. This macro must output the
7633 label definition (perhaps using @code{ASM_OUTPUT_LABEL}). The argument
7634 @var{decl} is the @code{VAR_DECL} tree node representing the variable.
7636 If this macro is not defined, then the variable name is defined in the
7637 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
7639 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} and/or
7640 @code{ASM_OUTPUT_SIZE_DIRECTIVE} in the definition of this macro.
7643 @defmac ASM_DECLARE_CONSTANT_NAME (@var{stream}, @var{name}, @var{exp}, @var{size})
7644 A C statement (sans semicolon) to output to the stdio stream
7645 @var{stream} any text necessary for declaring the name @var{name} of a
7646 constant which is being defined. This macro is responsible for
7647 outputting the label definition (perhaps using
7648 @code{ASM_OUTPUT_LABEL}). The argument @var{exp} is the
7649 value of the constant, and @var{size} is the size of the constant
7650 in bytes. @var{name} will be an internal label.
7652 If this macro is not defined, then the @var{name} is defined in the
7653 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
7655 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition
7659 @defmac ASM_DECLARE_REGISTER_GLOBAL (@var{stream}, @var{decl}, @var{regno}, @var{name})
7660 A C statement (sans semicolon) to output to the stdio stream
7661 @var{stream} any text necessary for claiming a register @var{regno}
7662 for a global variable @var{decl} with name @var{name}.
7664 If you don't define this macro, that is equivalent to defining it to do
7668 @defmac ASM_FINISH_DECLARE_OBJECT (@var{stream}, @var{decl}, @var{toplevel}, @var{atend})
7669 A C statement (sans semicolon) to finish up declaring a variable name
7670 once the compiler has processed its initializer fully and thus has had a
7671 chance to determine the size of an array when controlled by an
7672 initializer. This is used on systems where it's necessary to declare
7673 something about the size of the object.
7675 If you don't define this macro, that is equivalent to defining it to do
7678 You may wish to use @code{ASM_OUTPUT_SIZE_DIRECTIVE} and/or
7679 @code{ASM_OUTPUT_MEASURED_SIZE} in the definition of this macro.
7682 @deftypefn {Target Hook} void TARGET_ASM_GLOBALIZE_LABEL (FILE *@var{stream}, const char *@var{name})
7683 This target hook is a function to output to the stdio stream
7684 @var{stream} some commands that will make the label @var{name} global;
7685 that is, available for reference from other files.
7687 The default implementation relies on a proper definition of
7688 @code{GLOBAL_ASM_OP}.
7691 @deftypefn {Target Hook} void TARGET_ASM_GLOBALIZE_DECL_NAME (FILE *@var{stream}, tree @var{decl})
7692 This target hook is a function to output to the stdio stream
7693 @var{stream} some commands that will make the name associated with @var{decl}
7694 global; that is, available for reference from other files.
7696 The default implementation uses the TARGET_ASM_GLOBALIZE_LABEL target hook.
7699 @defmac ASM_WEAKEN_LABEL (@var{stream}, @var{name})
7700 A C statement (sans semicolon) to output to the stdio stream
7701 @var{stream} some commands that will make the label @var{name} weak;
7702 that is, available for reference from other files but only used if
7703 no other definition is available. Use the expression
7704 @code{assemble_name (@var{stream}, @var{name})} to output the name
7705 itself; before and after that, output the additional assembler syntax
7706 for making that name weak, and a newline.
7708 If you don't define this macro or @code{ASM_WEAKEN_DECL}, GCC will not
7709 support weak symbols and you should not define the @code{SUPPORTS_WEAK}
7713 @defmac ASM_WEAKEN_DECL (@var{stream}, @var{decl}, @var{name}, @var{value})
7714 Combines (and replaces) the function of @code{ASM_WEAKEN_LABEL} and
7715 @code{ASM_OUTPUT_WEAK_ALIAS}, allowing access to the associated function
7716 or variable decl. If @var{value} is not @code{NULL}, this C statement
7717 should output to the stdio stream @var{stream} assembler code which
7718 defines (equates) the weak symbol @var{name} to have the value
7719 @var{value}. If @var{value} is @code{NULL}, it should output commands
7720 to make @var{name} weak.
7723 @defmac ASM_OUTPUT_WEAKREF (@var{stream}, @var{decl}, @var{name}, @var{value})
7724 Outputs a directive that enables @var{name} to be used to refer to
7725 symbol @var{value} with weak-symbol semantics. @code{decl} is the
7726 declaration of @code{name}.
7729 @defmac SUPPORTS_WEAK
7730 A C expression which evaluates to true if the target supports weak symbols.
7732 If you don't define this macro, @file{defaults.h} provides a default
7733 definition. If either @code{ASM_WEAKEN_LABEL} or @code{ASM_WEAKEN_DECL}
7734 is defined, the default definition is @samp{1}; otherwise, it is
7735 @samp{0}. Define this macro if you want to control weak symbol support
7736 with a compiler flag such as @option{-melf}.
7739 @defmac MAKE_DECL_ONE_ONLY (@var{decl})
7740 A C statement (sans semicolon) to mark @var{decl} to be emitted as a
7741 public symbol such that extra copies in multiple translation units will
7742 be discarded by the linker. Define this macro if your object file
7743 format provides support for this concept, such as the @samp{COMDAT}
7744 section flags in the Microsoft Windows PE/COFF format, and this support
7745 requires changes to @var{decl}, such as putting it in a separate section.
7748 @defmac SUPPORTS_ONE_ONLY
7749 A C expression which evaluates to true if the target supports one-only
7752 If you don't define this macro, @file{varasm.c} provides a default
7753 definition. If @code{MAKE_DECL_ONE_ONLY} is defined, the default
7754 definition is @samp{1}; otherwise, it is @samp{0}. Define this macro if
7755 you want to control one-only symbol support with a compiler flag, or if
7756 setting the @code{DECL_ONE_ONLY} flag is enough to mark a declaration to
7757 be emitted as one-only.
7760 @deftypefn {Target Hook} void TARGET_ASM_ASSEMBLE_VISIBILITY (tree @var{decl}, const char *@var{visibility})
7761 This target hook is a function to output to @var{asm_out_file} some
7762 commands that will make the symbol(s) associated with @var{decl} have
7763 hidden, protected or internal visibility as specified by @var{visibility}.
7766 @defmac TARGET_WEAK_NOT_IN_ARCHIVE_TOC
7767 A C expression that evaluates to true if the target's linker expects
7768 that weak symbols do not appear in a static archive's table of contents.
7769 The default is @code{0}.
7771 Leaving weak symbols out of an archive's table of contents means that,
7772 if a symbol will only have a definition in one translation unit and
7773 will have undefined references from other translation units, that
7774 symbol should not be weak. Defining this macro to be nonzero will
7775 thus have the effect that certain symbols that would normally be weak
7776 (explicit template instantiations, and vtables for polymorphic classes
7777 with noninline key methods) will instead be nonweak.
7779 The C++ ABI requires this macro to be zero. Define this macro for
7780 targets where full C++ ABI compliance is impossible and where linker
7781 restrictions require weak symbols to be left out of a static archive's
7785 @defmac ASM_OUTPUT_EXTERNAL (@var{stream}, @var{decl}, @var{name})
7786 A C statement (sans semicolon) to output to the stdio stream
7787 @var{stream} any text necessary for declaring the name of an external
7788 symbol named @var{name} which is referenced in this compilation but
7789 not defined. The value of @var{decl} is the tree node for the
7792 This macro need not be defined if it does not need to output anything.
7793 The GNU assembler and most Unix assemblers don't require anything.
7796 @deftypefn {Target Hook} void TARGET_ASM_EXTERNAL_LIBCALL (rtx @var{symref})
7797 This target hook is a function to output to @var{asm_out_file} an assembler
7798 pseudo-op to declare a library function name external. The name of the
7799 library function is given by @var{symref}, which is a @code{symbol_ref}.
7802 @deftypefn {Target Hook} void TARGET_ASM_MARK_DECL_PRESERVED (tree @var{decl})
7803 This target hook is a function to output to @var{asm_out_file} an assembler
7804 directive to annotate used symbol. Darwin target use .no_dead_code_strip
7808 @defmac ASM_OUTPUT_LABELREF (@var{stream}, @var{name})
7809 A C statement (sans semicolon) to output to the stdio stream
7810 @var{stream} a reference in assembler syntax to a label named
7811 @var{name}. This should add @samp{_} to the front of the name, if that
7812 is customary on your operating system, as it is in most Berkeley Unix
7813 systems. This macro is used in @code{assemble_name}.
7816 @defmac ASM_OUTPUT_SYMBOL_REF (@var{stream}, @var{sym})
7817 A C statement (sans semicolon) to output a reference to
7818 @code{SYMBOL_REF} @var{sym}. If not defined, @code{assemble_name}
7819 will be used to output the name of the symbol. This macro may be used
7820 to modify the way a symbol is referenced depending on information
7821 encoded by @code{TARGET_ENCODE_SECTION_INFO}.
7824 @defmac ASM_OUTPUT_LABEL_REF (@var{stream}, @var{buf})
7825 A C statement (sans semicolon) to output a reference to @var{buf}, the
7826 result of @code{ASM_GENERATE_INTERNAL_LABEL}. If not defined,
7827 @code{assemble_name} will be used to output the name of the symbol.
7828 This macro is not used by @code{output_asm_label}, or the @code{%l}
7829 specifier that calls it; the intention is that this macro should be set
7830 when it is necessary to output a label differently when its address is
7834 @deftypefn {Target Hook} void TARGET_ASM_INTERNAL_LABEL (FILE *@var{stream}, const char *@var{prefix}, unsigned long @var{labelno})
7835 A function to output to the stdio stream @var{stream} a label whose
7836 name is made from the string @var{prefix} and the number @var{labelno}.
7838 It is absolutely essential that these labels be distinct from the labels
7839 used for user-level functions and variables. Otherwise, certain programs
7840 will have name conflicts with internal labels.
7842 It is desirable to exclude internal labels from the symbol table of the
7843 object file. Most assemblers have a naming convention for labels that
7844 should be excluded; on many systems, the letter @samp{L} at the
7845 beginning of a label has this effect. You should find out what
7846 convention your system uses, and follow it.
7848 The default version of this function utilizes @code{ASM_GENERATE_INTERNAL_LABEL}.
7851 @defmac ASM_OUTPUT_DEBUG_LABEL (@var{stream}, @var{prefix}, @var{num})
7852 A C statement to output to the stdio stream @var{stream} a debug info
7853 label whose name is made from the string @var{prefix} and the number
7854 @var{num}. This is useful for VLIW targets, where debug info labels
7855 may need to be treated differently than branch target labels. On some
7856 systems, branch target labels must be at the beginning of instruction
7857 bundles, but debug info labels can occur in the middle of instruction
7860 If this macro is not defined, then @code{(*targetm.asm_out.internal_label)} will be
7864 @defmac ASM_GENERATE_INTERNAL_LABEL (@var{string}, @var{prefix}, @var{num})
7865 A C statement to store into the string @var{string} a label whose name
7866 is made from the string @var{prefix} and the number @var{num}.
7868 This string, when output subsequently by @code{assemble_name}, should
7869 produce the output that @code{(*targetm.asm_out.internal_label)} would produce
7870 with the same @var{prefix} and @var{num}.
7872 If the string begins with @samp{*}, then @code{assemble_name} will
7873 output the rest of the string unchanged. It is often convenient for
7874 @code{ASM_GENERATE_INTERNAL_LABEL} to use @samp{*} in this way. If the
7875 string doesn't start with @samp{*}, then @code{ASM_OUTPUT_LABELREF} gets
7876 to output the string, and may change it. (Of course,
7877 @code{ASM_OUTPUT_LABELREF} is also part of your machine description, so
7878 you should know what it does on your machine.)
7881 @defmac ASM_FORMAT_PRIVATE_NAME (@var{outvar}, @var{name}, @var{number})
7882 A C expression to assign to @var{outvar} (which is a variable of type
7883 @code{char *}) a newly allocated string made from the string
7884 @var{name} and the number @var{number}, with some suitable punctuation
7885 added. Use @code{alloca} to get space for the string.
7887 The string will be used as an argument to @code{ASM_OUTPUT_LABELREF} to
7888 produce an assembler label for an internal static variable whose name is
7889 @var{name}. Therefore, the string must be such as to result in valid
7890 assembler code. The argument @var{number} is different each time this
7891 macro is executed; it prevents conflicts between similarly-named
7892 internal static variables in different scopes.
7894 Ideally this string should not be a valid C identifier, to prevent any
7895 conflict with the user's own symbols. Most assemblers allow periods
7896 or percent signs in assembler symbols; putting at least one of these
7897 between the name and the number will suffice.
7899 If this macro is not defined, a default definition will be provided
7900 which is correct for most systems.
7903 @defmac ASM_OUTPUT_DEF (@var{stream}, @var{name}, @var{value})
7904 A C statement to output to the stdio stream @var{stream} assembler code
7905 which defines (equates) the symbol @var{name} to have the value @var{value}.
7908 If @code{SET_ASM_OP} is defined, a default definition is provided which is
7909 correct for most systems.
7912 @defmac ASM_OUTPUT_DEF_FROM_DECLS (@var{stream}, @var{decl_of_name}, @var{decl_of_value})
7913 A C statement to output to the stdio stream @var{stream} assembler code
7914 which defines (equates) the symbol whose tree node is @var{decl_of_name}
7915 to have the value of the tree node @var{decl_of_value}. This macro will
7916 be used in preference to @samp{ASM_OUTPUT_DEF} if it is defined and if
7917 the tree nodes are available.
7920 If @code{SET_ASM_OP} is defined, a default definition is provided which is
7921 correct for most systems.
7924 @defmac TARGET_DEFERRED_OUTPUT_DEFS (@var{decl_of_name}, @var{decl_of_value})
7925 A C statement that evaluates to true if the assembler code which defines
7926 (equates) the symbol whose tree node is @var{decl_of_name} to have the value
7927 of the tree node @var{decl_of_value} should be emitted near the end of the
7928 current compilation unit. The default is to not defer output of defines.
7929 This macro affects defines output by @samp{ASM_OUTPUT_DEF} and
7930 @samp{ASM_OUTPUT_DEF_FROM_DECLS}.
7933 @defmac ASM_OUTPUT_WEAK_ALIAS (@var{stream}, @var{name}, @var{value})
7934 A C statement to output to the stdio stream @var{stream} assembler code
7935 which defines (equates) the weak symbol @var{name} to have the value
7936 @var{value}. If @var{value} is @code{NULL}, it defines @var{name} as
7937 an undefined weak symbol.
7939 Define this macro if the target only supports weak aliases; define
7940 @code{ASM_OUTPUT_DEF} instead if possible.
7943 @defmac OBJC_GEN_METHOD_LABEL (@var{buf}, @var{is_inst}, @var{class_name}, @var{cat_name}, @var{sel_name})
7944 Define this macro to override the default assembler names used for
7945 Objective-C methods.
7947 The default name is a unique method number followed by the name of the
7948 class (e.g.@: @samp{_1_Foo}). For methods in categories, the name of
7949 the category is also included in the assembler name (e.g.@:
7952 These names are safe on most systems, but make debugging difficult since
7953 the method's selector is not present in the name. Therefore, particular
7954 systems define other ways of computing names.
7956 @var{buf} is an expression of type @code{char *} which gives you a
7957 buffer in which to store the name; its length is as long as
7958 @var{class_name}, @var{cat_name} and @var{sel_name} put together, plus
7959 50 characters extra.
7961 The argument @var{is_inst} specifies whether the method is an instance
7962 method or a class method; @var{class_name} is the name of the class;
7963 @var{cat_name} is the name of the category (or @code{NULL} if the method is not
7964 in a category); and @var{sel_name} is the name of the selector.
7966 On systems where the assembler can handle quoted names, you can use this
7967 macro to provide more human-readable names.
7970 @defmac ASM_DECLARE_CLASS_REFERENCE (@var{stream}, @var{name})
7971 A C statement (sans semicolon) to output to the stdio stream
7972 @var{stream} commands to declare that the label @var{name} is an
7973 Objective-C class reference. This is only needed for targets whose
7974 linkers have special support for NeXT-style runtimes.
7977 @defmac ASM_DECLARE_UNRESOLVED_REFERENCE (@var{stream}, @var{name})
7978 A C statement (sans semicolon) to output to the stdio stream
7979 @var{stream} commands to declare that the label @var{name} is an
7980 unresolved Objective-C class reference. This is only needed for targets
7981 whose linkers have special support for NeXT-style runtimes.
7984 @node Initialization
7985 @subsection How Initialization Functions Are Handled
7986 @cindex initialization routines
7987 @cindex termination routines
7988 @cindex constructors, output of
7989 @cindex destructors, output of
7991 The compiled code for certain languages includes @dfn{constructors}
7992 (also called @dfn{initialization routines})---functions to initialize
7993 data in the program when the program is started. These functions need
7994 to be called before the program is ``started''---that is to say, before
7995 @code{main} is called.
7997 Compiling some languages generates @dfn{destructors} (also called
7998 @dfn{termination routines}) that should be called when the program
8001 To make the initialization and termination functions work, the compiler
8002 must output something in the assembler code to cause those functions to
8003 be called at the appropriate time. When you port the compiler to a new
8004 system, you need to specify how to do this.
8006 There are two major ways that GCC currently supports the execution of
8007 initialization and termination functions. Each way has two variants.
8008 Much of the structure is common to all four variations.
8010 @findex __CTOR_LIST__
8011 @findex __DTOR_LIST__
8012 The linker must build two lists of these functions---a list of
8013 initialization functions, called @code{__CTOR_LIST__}, and a list of
8014 termination functions, called @code{__DTOR_LIST__}.
8016 Each list always begins with an ignored function pointer (which may hold
8017 0, @minus{}1, or a count of the function pointers after it, depending on
8018 the environment). This is followed by a series of zero or more function
8019 pointers to constructors (or destructors), followed by a function
8020 pointer containing zero.
8022 Depending on the operating system and its executable file format, either
8023 @file{crtstuff.c} or @file{libgcc2.c} traverses these lists at startup
8024 time and exit time. Constructors are called in reverse order of the
8025 list; destructors in forward order.
8027 The best way to handle static constructors works only for object file
8028 formats which provide arbitrarily-named sections. A section is set
8029 aside for a list of constructors, and another for a list of destructors.
8030 Traditionally these are called @samp{.ctors} and @samp{.dtors}. Each
8031 object file that defines an initialization function also puts a word in
8032 the constructor section to point to that function. The linker
8033 accumulates all these words into one contiguous @samp{.ctors} section.
8034 Termination functions are handled similarly.
8036 This method will be chosen as the default by @file{target-def.h} if
8037 @code{TARGET_ASM_NAMED_SECTION} is defined. A target that does not
8038 support arbitrary sections, but does support special designated
8039 constructor and destructor sections may define @code{CTORS_SECTION_ASM_OP}
8040 and @code{DTORS_SECTION_ASM_OP} to achieve the same effect.
8042 When arbitrary sections are available, there are two variants, depending
8043 upon how the code in @file{crtstuff.c} is called. On systems that
8044 support a @dfn{.init} section which is executed at program startup,
8045 parts of @file{crtstuff.c} are compiled into that section. The
8046 program is linked by the @command{gcc} driver like this:
8049 ld -o @var{output_file} crti.o crtbegin.o @dots{} -lgcc crtend.o crtn.o
8052 The prologue of a function (@code{__init}) appears in the @code{.init}
8053 section of @file{crti.o}; the epilogue appears in @file{crtn.o}. Likewise
8054 for the function @code{__fini} in the @dfn{.fini} section. Normally these
8055 files are provided by the operating system or by the GNU C library, but
8056 are provided by GCC for a few targets.
8058 The objects @file{crtbegin.o} and @file{crtend.o} are (for most targets)
8059 compiled from @file{crtstuff.c}. They contain, among other things, code
8060 fragments within the @code{.init} and @code{.fini} sections that branch
8061 to routines in the @code{.text} section. The linker will pull all parts
8062 of a section together, which results in a complete @code{__init} function
8063 that invokes the routines we need at startup.
8065 To use this variant, you must define the @code{INIT_SECTION_ASM_OP}
8068 If no init section is available, when GCC compiles any function called
8069 @code{main} (or more accurately, any function designated as a program
8070 entry point by the language front end calling @code{expand_main_function}),
8071 it inserts a procedure call to @code{__main} as the first executable code
8072 after the function prologue. The @code{__main} function is defined
8073 in @file{libgcc2.c} and runs the global constructors.
8075 In file formats that don't support arbitrary sections, there are again
8076 two variants. In the simplest variant, the GNU linker (GNU @code{ld})
8077 and an `a.out' format must be used. In this case,
8078 @code{TARGET_ASM_CONSTRUCTOR} is defined to produce a @code{.stabs}
8079 entry of type @samp{N_SETT}, referencing the name @code{__CTOR_LIST__},
8080 and with the address of the void function containing the initialization
8081 code as its value. The GNU linker recognizes this as a request to add
8082 the value to a @dfn{set}; the values are accumulated, and are eventually
8083 placed in the executable as a vector in the format described above, with
8084 a leading (ignored) count and a trailing zero element.
8085 @code{TARGET_ASM_DESTRUCTOR} is handled similarly. Since no init
8086 section is available, the absence of @code{INIT_SECTION_ASM_OP} causes
8087 the compilation of @code{main} to call @code{__main} as above, starting
8088 the initialization process.
8090 The last variant uses neither arbitrary sections nor the GNU linker.
8091 This is preferable when you want to do dynamic linking and when using
8092 file formats which the GNU linker does not support, such as `ECOFF'@. In
8093 this case, @code{TARGET_HAVE_CTORS_DTORS} is false, initialization and
8094 termination functions are recognized simply by their names. This requires
8095 an extra program in the linkage step, called @command{collect2}. This program
8096 pretends to be the linker, for use with GCC; it does its job by running
8097 the ordinary linker, but also arranges to include the vectors of
8098 initialization and termination functions. These functions are called
8099 via @code{__main} as described above. In order to use this method,
8100 @code{use_collect2} must be defined in the target in @file{config.gcc}.
8103 The following section describes the specific macros that control and
8104 customize the handling of initialization and termination functions.
8107 @node Macros for Initialization
8108 @subsection Macros Controlling Initialization Routines
8110 Here are the macros that control how the compiler handles initialization
8111 and termination functions:
8113 @defmac INIT_SECTION_ASM_OP
8114 If defined, a C string constant, including spacing, for the assembler
8115 operation to identify the following data as initialization code. If not
8116 defined, GCC will assume such a section does not exist. When you are
8117 using special sections for initialization and termination functions, this
8118 macro also controls how @file{crtstuff.c} and @file{libgcc2.c} arrange to
8119 run the initialization functions.
8122 @defmac HAS_INIT_SECTION
8123 If defined, @code{main} will not call @code{__main} as described above.
8124 This macro should be defined for systems that control start-up code
8125 on a symbol-by-symbol basis, such as OSF/1, and should not
8126 be defined explicitly for systems that support @code{INIT_SECTION_ASM_OP}.
8129 @defmac LD_INIT_SWITCH
8130 If defined, a C string constant for a switch that tells the linker that
8131 the following symbol is an initialization routine.
8134 @defmac LD_FINI_SWITCH
8135 If defined, a C string constant for a switch that tells the linker that
8136 the following symbol is a finalization routine.
8139 @defmac COLLECT_SHARED_INIT_FUNC (@var{stream}, @var{func})
8140 If defined, a C statement that will write a function that can be
8141 automatically called when a shared library is loaded. The function
8142 should call @var{func}, which takes no arguments. If not defined, and
8143 the object format requires an explicit initialization function, then a
8144 function called @code{_GLOBAL__DI} will be generated.
8146 This function and the following one are used by collect2 when linking a
8147 shared library that needs constructors or destructors, or has DWARF2
8148 exception tables embedded in the code.
8151 @defmac COLLECT_SHARED_FINI_FUNC (@var{stream}, @var{func})
8152 If defined, a C statement that will write a function that can be
8153 automatically called when a shared library is unloaded. The function
8154 should call @var{func}, which takes no arguments. If not defined, and
8155 the object format requires an explicit finalization function, then a
8156 function called @code{_GLOBAL__DD} will be generated.
8159 @defmac INVOKE__main
8160 If defined, @code{main} will call @code{__main} despite the presence of
8161 @code{INIT_SECTION_ASM_OP}. This macro should be defined for systems
8162 where the init section is not actually run automatically, but is still
8163 useful for collecting the lists of constructors and destructors.
8166 @defmac SUPPORTS_INIT_PRIORITY
8167 If nonzero, the C++ @code{init_priority} attribute is supported and the
8168 compiler should emit instructions to control the order of initialization
8169 of objects. If zero, the compiler will issue an error message upon
8170 encountering an @code{init_priority} attribute.
8173 @deftypefn {Target Hook} bool TARGET_HAVE_CTORS_DTORS
8174 This value is true if the target supports some ``native'' method of
8175 collecting constructors and destructors to be run at startup and exit.
8176 It is false if we must use @command{collect2}.
8179 @deftypefn {Target Hook} void TARGET_ASM_CONSTRUCTOR (rtx @var{symbol}, int @var{priority})
8180 If defined, a function that outputs assembler code to arrange to call
8181 the function referenced by @var{symbol} at initialization time.
8183 Assume that @var{symbol} is a @code{SYMBOL_REF} for a function taking
8184 no arguments and with no return value. If the target supports initialization
8185 priorities, @var{priority} is a value between 0 and @code{MAX_INIT_PRIORITY};
8186 otherwise it must be @code{DEFAULT_INIT_PRIORITY}.
8188 If this macro is not defined by the target, a suitable default will
8189 be chosen if (1) the target supports arbitrary section names, (2) the
8190 target defines @code{CTORS_SECTION_ASM_OP}, or (3) @code{USE_COLLECT2}
8194 @deftypefn {Target Hook} void TARGET_ASM_DESTRUCTOR (rtx @var{symbol}, int @var{priority})
8195 This is like @code{TARGET_ASM_CONSTRUCTOR} but used for termination
8196 functions rather than initialization functions.
8199 If @code{TARGET_HAVE_CTORS_DTORS} is true, the initialization routine
8200 generated for the generated object file will have static linkage.
8202 If your system uses @command{collect2} as the means of processing
8203 constructors, then that program normally uses @command{nm} to scan
8204 an object file for constructor functions to be called.
8206 On certain kinds of systems, you can define this macro to make
8207 @command{collect2} work faster (and, in some cases, make it work at all):
8209 @defmac OBJECT_FORMAT_COFF
8210 Define this macro if the system uses COFF (Common Object File Format)
8211 object files, so that @command{collect2} can assume this format and scan
8212 object files directly for dynamic constructor/destructor functions.
8214 This macro is effective only in a native compiler; @command{collect2} as
8215 part of a cross compiler always uses @command{nm} for the target machine.
8218 @defmac REAL_NM_FILE_NAME
8219 Define this macro as a C string constant containing the file name to use
8220 to execute @command{nm}. The default is to search the path normally for
8223 If your system supports shared libraries and has a program to list the
8224 dynamic dependencies of a given library or executable, you can define
8225 these macros to enable support for running initialization and
8226 termination functions in shared libraries:
8230 Define this macro to a C string constant containing the name of the program
8231 which lists dynamic dependencies, like @command{"ldd"} under SunOS 4.
8234 @defmac PARSE_LDD_OUTPUT (@var{ptr})
8235 Define this macro to be C code that extracts filenames from the output
8236 of the program denoted by @code{LDD_SUFFIX}. @var{ptr} is a variable
8237 of type @code{char *} that points to the beginning of a line of output
8238 from @code{LDD_SUFFIX}. If the line lists a dynamic dependency, the
8239 code must advance @var{ptr} to the beginning of the filename on that
8240 line. Otherwise, it must set @var{ptr} to @code{NULL}.
8243 @defmac SHLIB_SUFFIX
8244 Define this macro to a C string constant containing the default shared
8245 library extension of the target (e.g., @samp{".so"}). @command{collect2}
8246 strips version information after this suffix when generating global
8247 constructor and destructor names. This define is only needed on targets
8248 that use @command{collect2} to process constructors and destructors.
8251 @node Instruction Output
8252 @subsection Output of Assembler Instructions
8254 @c prevent bad page break with this line
8255 This describes assembler instruction output.
8257 @defmac REGISTER_NAMES
8258 A C initializer containing the assembler's names for the machine
8259 registers, each one as a C string constant. This is what translates
8260 register numbers in the compiler into assembler language.
8263 @defmac ADDITIONAL_REGISTER_NAMES
8264 If defined, a C initializer for an array of structures containing a name
8265 and a register number. This macro defines additional names for hard
8266 registers, thus allowing the @code{asm} option in declarations to refer
8267 to registers using alternate names.
8270 @defmac ASM_OUTPUT_OPCODE (@var{stream}, @var{ptr})
8271 Define this macro if you are using an unusual assembler that
8272 requires different names for the machine instructions.
8274 The definition is a C statement or statements which output an
8275 assembler instruction opcode to the stdio stream @var{stream}. The
8276 macro-operand @var{ptr} is a variable of type @code{char *} which
8277 points to the opcode name in its ``internal'' form---the form that is
8278 written in the machine description. The definition should output the
8279 opcode name to @var{stream}, performing any translation you desire, and
8280 increment the variable @var{ptr} to point at the end of the opcode
8281 so that it will not be output twice.
8283 In fact, your macro definition may process less than the entire opcode
8284 name, or more than the opcode name; but if you want to process text
8285 that includes @samp{%}-sequences to substitute operands, you must take
8286 care of the substitution yourself. Just be sure to increment
8287 @var{ptr} over whatever text should not be output normally.
8289 @findex recog_data.operand
8290 If you need to look at the operand values, they can be found as the
8291 elements of @code{recog_data.operand}.
8293 If the macro definition does nothing, the instruction is output
8297 @defmac FINAL_PRESCAN_INSN (@var{insn}, @var{opvec}, @var{noperands})
8298 If defined, a C statement to be executed just prior to the output of
8299 assembler code for @var{insn}, to modify the extracted operands so
8300 they will be output differently.
8302 Here the argument @var{opvec} is the vector containing the operands
8303 extracted from @var{insn}, and @var{noperands} is the number of
8304 elements of the vector which contain meaningful data for this insn.
8305 The contents of this vector are what will be used to convert the insn
8306 template into assembler code, so you can change the assembler output
8307 by changing the contents of the vector.
8309 This macro is useful when various assembler syntaxes share a single
8310 file of instruction patterns; by defining this macro differently, you
8311 can cause a large class of instructions to be output differently (such
8312 as with rearranged operands). Naturally, variations in assembler
8313 syntax affecting individual insn patterns ought to be handled by
8314 writing conditional output routines in those patterns.
8316 If this macro is not defined, it is equivalent to a null statement.
8319 @deftypefn {Target Hook} void TARGET_ASM_FINAL_POSTSCAN_INSN (FILE *@var{FILE}, rtx @var{insn}, rtx *@var{opvec}, int @var{noperands})
8320 If defined, this target hook is a function which is executed just after the
8321 output of assembler code for @var{insn}, to change the mode of the assembler
8324 Here the argument @var{opvec} is the vector containing the operands
8325 extracted from @var{insn}, and @var{noperands} is the number of
8326 elements of the vector which contain meaningful data for this insn.
8327 The contents of this vector are what was used to convert the insn
8328 template into assembler code, so you can change the assembler mode
8329 by checking the contents of the vector.
8332 @defmac PRINT_OPERAND (@var{stream}, @var{x}, @var{code})
8333 A C compound statement to output to stdio stream @var{stream} the
8334 assembler syntax for an instruction operand @var{x}. @var{x} is an
8337 @var{code} is a value that can be used to specify one of several ways
8338 of printing the operand. It is used when identical operands must be
8339 printed differently depending on the context. @var{code} comes from
8340 the @samp{%} specification that was used to request printing of the
8341 operand. If the specification was just @samp{%@var{digit}} then
8342 @var{code} is 0; if the specification was @samp{%@var{ltr}
8343 @var{digit}} then @var{code} is the ASCII code for @var{ltr}.
8346 If @var{x} is a register, this macro should print the register's name.
8347 The names can be found in an array @code{reg_names} whose type is
8348 @code{char *[]}. @code{reg_names} is initialized from
8349 @code{REGISTER_NAMES}.
8351 When the machine description has a specification @samp{%@var{punct}}
8352 (a @samp{%} followed by a punctuation character), this macro is called
8353 with a null pointer for @var{x} and the punctuation character for
8357 @defmac PRINT_OPERAND_PUNCT_VALID_P (@var{code})
8358 A C expression which evaluates to true if @var{code} is a valid
8359 punctuation character for use in the @code{PRINT_OPERAND} macro. If
8360 @code{PRINT_OPERAND_PUNCT_VALID_P} is not defined, it means that no
8361 punctuation characters (except for the standard one, @samp{%}) are used
8365 @defmac PRINT_OPERAND_ADDRESS (@var{stream}, @var{x})
8366 A C compound statement to output to stdio stream @var{stream} the
8367 assembler syntax for an instruction operand that is a memory reference
8368 whose address is @var{x}. @var{x} is an RTL expression.
8370 @cindex @code{TARGET_ENCODE_SECTION_INFO} usage
8371 On some machines, the syntax for a symbolic address depends on the
8372 section that the address refers to. On these machines, define the hook
8373 @code{TARGET_ENCODE_SECTION_INFO} to store the information into the
8374 @code{symbol_ref}, and then check for it here. @xref{Assembler
8378 @findex dbr_sequence_length
8379 @defmac DBR_OUTPUT_SEQEND (@var{file})
8380 A C statement, to be executed after all slot-filler instructions have
8381 been output. If necessary, call @code{dbr_sequence_length} to
8382 determine the number of slots filled in a sequence (zero if not
8383 currently outputting a sequence), to decide how many no-ops to output,
8386 Don't define this macro if it has nothing to do, but it is helpful in
8387 reading assembly output if the extent of the delay sequence is made
8388 explicit (e.g.@: with white space).
8391 @findex final_sequence
8392 Note that output routines for instructions with delay slots must be
8393 prepared to deal with not being output as part of a sequence
8394 (i.e.@: when the scheduling pass is not run, or when no slot fillers could be
8395 found.) The variable @code{final_sequence} is null when not
8396 processing a sequence, otherwise it contains the @code{sequence} rtx
8400 @defmac REGISTER_PREFIX
8401 @defmacx LOCAL_LABEL_PREFIX
8402 @defmacx USER_LABEL_PREFIX
8403 @defmacx IMMEDIATE_PREFIX
8404 If defined, C string expressions to be used for the @samp{%R}, @samp{%L},
8405 @samp{%U}, and @samp{%I} options of @code{asm_fprintf} (see
8406 @file{final.c}). These are useful when a single @file{md} file must
8407 support multiple assembler formats. In that case, the various @file{tm.h}
8408 files can define these macros differently.
8411 @defmac ASM_FPRINTF_EXTENSIONS (@var{file}, @var{argptr}, @var{format})
8412 If defined this macro should expand to a series of @code{case}
8413 statements which will be parsed inside the @code{switch} statement of
8414 the @code{asm_fprintf} function. This allows targets to define extra
8415 printf formats which may useful when generating their assembler
8416 statements. Note that uppercase letters are reserved for future
8417 generic extensions to asm_fprintf, and so are not available to target
8418 specific code. The output file is given by the parameter @var{file}.
8419 The varargs input pointer is @var{argptr} and the rest of the format
8420 string, starting the character after the one that is being switched
8421 upon, is pointed to by @var{format}.
8424 @defmac ASSEMBLER_DIALECT
8425 If your target supports multiple dialects of assembler language (such as
8426 different opcodes), define this macro as a C expression that gives the
8427 numeric index of the assembler language dialect to use, with zero as the
8430 If this macro is defined, you may use constructs of the form
8432 @samp{@{option0|option1|option2@dots{}@}}
8435 in the output templates of patterns (@pxref{Output Template}) or in the
8436 first argument of @code{asm_fprintf}. This construct outputs
8437 @samp{option0}, @samp{option1}, @samp{option2}, etc., if the value of
8438 @code{ASSEMBLER_DIALECT} is zero, one, two, etc. Any special characters
8439 within these strings retain their usual meaning. If there are fewer
8440 alternatives within the braces than the value of
8441 @code{ASSEMBLER_DIALECT}, the construct outputs nothing.
8443 If you do not define this macro, the characters @samp{@{}, @samp{|} and
8444 @samp{@}} do not have any special meaning when used in templates or
8445 operands to @code{asm_fprintf}.
8447 Define the macros @code{REGISTER_PREFIX}, @code{LOCAL_LABEL_PREFIX},
8448 @code{USER_LABEL_PREFIX} and @code{IMMEDIATE_PREFIX} if you can express
8449 the variations in assembler language syntax with that mechanism. Define
8450 @code{ASSEMBLER_DIALECT} and use the @samp{@{option0|option1@}} syntax
8451 if the syntax variant are larger and involve such things as different
8452 opcodes or operand order.
8455 @defmac ASM_OUTPUT_REG_PUSH (@var{stream}, @var{regno})
8456 A C expression to output to @var{stream} some assembler code
8457 which will push hard register number @var{regno} onto the stack.
8458 The code need not be optimal, since this macro is used only when
8462 @defmac ASM_OUTPUT_REG_POP (@var{stream}, @var{regno})
8463 A C expression to output to @var{stream} some assembler code
8464 which will pop hard register number @var{regno} off of the stack.
8465 The code need not be optimal, since this macro is used only when
8469 @node Dispatch Tables
8470 @subsection Output of Dispatch Tables
8472 @c prevent bad page break with this line
8473 This concerns dispatch tables.
8475 @cindex dispatch table
8476 @defmac ASM_OUTPUT_ADDR_DIFF_ELT (@var{stream}, @var{body}, @var{value}, @var{rel})
8477 A C statement to output to the stdio stream @var{stream} an assembler
8478 pseudo-instruction to generate a difference between two labels.
8479 @var{value} and @var{rel} are the numbers of two internal labels. The
8480 definitions of these labels are output using
8481 @code{(*targetm.asm_out.internal_label)}, and they must be printed in the same
8482 way here. For example,
8485 fprintf (@var{stream}, "\t.word L%d-L%d\n",
8486 @var{value}, @var{rel})
8489 You must provide this macro on machines where the addresses in a
8490 dispatch table are relative to the table's own address. If defined, GCC
8491 will also use this macro on all machines when producing PIC@.
8492 @var{body} is the body of the @code{ADDR_DIFF_VEC}; it is provided so that the
8493 mode and flags can be read.
8496 @defmac ASM_OUTPUT_ADDR_VEC_ELT (@var{stream}, @var{value})
8497 This macro should be provided on machines where the addresses
8498 in a dispatch table are absolute.
8500 The definition should be a C statement to output to the stdio stream
8501 @var{stream} an assembler pseudo-instruction to generate a reference to
8502 a label. @var{value} is the number of an internal label whose
8503 definition is output using @code{(*targetm.asm_out.internal_label)}.
8507 fprintf (@var{stream}, "\t.word L%d\n", @var{value})
8511 @defmac ASM_OUTPUT_CASE_LABEL (@var{stream}, @var{prefix}, @var{num}, @var{table})
8512 Define this if the label before a jump-table needs to be output
8513 specially. The first three arguments are the same as for
8514 @code{(*targetm.asm_out.internal_label)}; the fourth argument is the
8515 jump-table which follows (a @code{jump_insn} containing an
8516 @code{addr_vec} or @code{addr_diff_vec}).
8518 This feature is used on system V to output a @code{swbeg} statement
8521 If this macro is not defined, these labels are output with
8522 @code{(*targetm.asm_out.internal_label)}.
8525 @defmac ASM_OUTPUT_CASE_END (@var{stream}, @var{num}, @var{table})
8526 Define this if something special must be output at the end of a
8527 jump-table. The definition should be a C statement to be executed
8528 after the assembler code for the table is written. It should write
8529 the appropriate code to stdio stream @var{stream}. The argument
8530 @var{table} is the jump-table insn, and @var{num} is the label-number
8531 of the preceding label.
8533 If this macro is not defined, nothing special is output at the end of
8537 @deftypefn {Target Hook} void TARGET_ASM_EMIT_UNWIND_LABEL (@var{stream}, @var{decl}, @var{for_eh}, @var{empty})
8538 This target hook emits a label at the beginning of each FDE@. It
8539 should be defined on targets where FDEs need special labels, and it
8540 should write the appropriate label, for the FDE associated with the
8541 function declaration @var{decl}, to the stdio stream @var{stream}.
8542 The third argument, @var{for_eh}, is a boolean: true if this is for an
8543 exception table. The fourth argument, @var{empty}, is a boolean:
8544 true if this is a placeholder label for an omitted FDE@.
8546 The default is that FDEs are not given nonlocal labels.
8549 @deftypefn {Target Hook} void TARGET_ASM_EMIT_EXCEPT_TABLE_LABEL (@var{stream})
8550 This target hook emits a label at the beginning of the exception table.
8551 It should be defined on targets where it is desirable for the table
8552 to be broken up according to function.
8554 The default is that no label is emitted.
8557 @deftypefn {Target Hook} void TARGET_UNWIND_EMIT (FILE * @var{stream}, rtx @var{insn})
8558 This target hook emits and assembly directives required to unwind the
8559 given instruction. This is only used when TARGET_UNWIND_INFO is set.
8562 @node Exception Region Output
8563 @subsection Assembler Commands for Exception Regions
8565 @c prevent bad page break with this line
8567 This describes commands marking the start and the end of an exception
8570 @defmac EH_FRAME_SECTION_NAME
8571 If defined, a C string constant for the name of the section containing
8572 exception handling frame unwind information. If not defined, GCC will
8573 provide a default definition if the target supports named sections.
8574 @file{crtstuff.c} uses this macro to switch to the appropriate section.
8576 You should define this symbol if your target supports DWARF 2 frame
8577 unwind information and the default definition does not work.
8580 @defmac EH_FRAME_IN_DATA_SECTION
8581 If defined, DWARF 2 frame unwind information will be placed in the
8582 data section even though the target supports named sections. This
8583 might be necessary, for instance, if the system linker does garbage
8584 collection and sections cannot be marked as not to be collected.
8586 Do not define this macro unless @code{TARGET_ASM_NAMED_SECTION} is
8590 @defmac EH_TABLES_CAN_BE_READ_ONLY
8591 Define this macro to 1 if your target is such that no frame unwind
8592 information encoding used with non-PIC code will ever require a
8593 runtime relocation, but the linker may not support merging read-only
8594 and read-write sections into a single read-write section.
8597 @defmac MASK_RETURN_ADDR
8598 An rtx used to mask the return address found via @code{RETURN_ADDR_RTX}, so
8599 that it does not contain any extraneous set bits in it.
8602 @defmac DWARF2_UNWIND_INFO
8603 Define this macro to 0 if your target supports DWARF 2 frame unwind
8604 information, but it does not yet work with exception handling.
8605 Otherwise, if your target supports this information (if it defines
8606 @samp{INCOMING_RETURN_ADDR_RTX} and either @samp{UNALIGNED_INT_ASM_OP}
8607 or @samp{OBJECT_FORMAT_ELF}), GCC will provide a default definition of 1.
8609 If @code{TARGET_UNWIND_INFO} is defined, the target specific unwinder
8610 will be used in all cases. Defining this macro will enable the generation
8611 of DWARF 2 frame debugging information.
8613 If @code{TARGET_UNWIND_INFO} is not defined, and this macro is defined to 1,
8614 the DWARF 2 unwinder will be the default exception handling mechanism;
8615 otherwise, the @code{setjmp}/@code{longjmp}-based scheme will be used by
8619 @defmac TARGET_UNWIND_INFO
8620 Define this macro if your target has ABI specified unwind tables. Usually
8621 these will be output by @code{TARGET_UNWIND_EMIT}.
8624 @deftypevr {Target Hook} bool TARGET_UNWIND_TABLES_DEFAULT
8625 This variable should be set to @code{true} if the target ABI requires unwinding
8626 tables even when exceptions are not used.
8629 @defmac MUST_USE_SJLJ_EXCEPTIONS
8630 This macro need only be defined if @code{DWARF2_UNWIND_INFO} is
8631 runtime-variable. In that case, @file{except.h} cannot correctly
8632 determine the corresponding definition of @code{MUST_USE_SJLJ_EXCEPTIONS},
8633 so the target must provide it directly.
8636 @defmac DONT_USE_BUILTIN_SETJMP
8637 Define this macro to 1 if the @code{setjmp}/@code{longjmp}-based scheme
8638 should use the @code{setjmp}/@code{longjmp} functions from the C library
8639 instead of the @code{__builtin_setjmp}/@code{__builtin_longjmp} machinery.
8642 @defmac DWARF_CIE_DATA_ALIGNMENT
8643 This macro need only be defined if the target might save registers in the
8644 function prologue at an offset to the stack pointer that is not aligned to
8645 @code{UNITS_PER_WORD}. The definition should be the negative minimum
8646 alignment if @code{STACK_GROWS_DOWNWARD} is defined, and the positive
8647 minimum alignment otherwise. @xref{SDB and DWARF}. Only applicable if
8648 the target supports DWARF 2 frame unwind information.
8651 @deftypevr {Target Hook} bool TARGET_TERMINATE_DW2_EH_FRAME_INFO
8652 Contains the value true if the target should add a zero word onto the
8653 end of a Dwarf-2 frame info section when used for exception handling.
8654 Default value is false if @code{EH_FRAME_SECTION_NAME} is defined, and
8658 @deftypefn {Target Hook} rtx TARGET_DWARF_REGISTER_SPAN (rtx @var{reg})
8659 Given a register, this hook should return a parallel of registers to
8660 represent where to find the register pieces. Define this hook if the
8661 register and its mode are represented in Dwarf in non-contiguous
8662 locations, or if the register should be represented in more than one
8663 register in Dwarf. Otherwise, this hook should return @code{NULL_RTX}.
8664 If not defined, the default is to return @code{NULL_RTX}.
8667 @deftypefn {Target Hook} void TARGET_INIT_DWARF_REG_SIZES_EXTRA (tree @var{address})
8668 If some registers are represented in Dwarf-2 unwind information in
8669 multiple pieces, define this hook to fill in information about the
8670 sizes of those pieces in the table used by the unwinder at runtime.
8671 It will be called by @code{expand_builtin_init_dwarf_reg_sizes} after
8672 filling in a single size corresponding to each hard register;
8673 @var{address} is the address of the table.
8676 @deftypefn {Target Hook} bool TARGET_ASM_TTYPE (rtx @var{sym})
8677 This hook is used to output a reference from a frame unwinding table to
8678 the type_info object identified by @var{sym}. It should return @code{true}
8679 if the reference was output. Returning @code{false} will cause the
8680 reference to be output using the normal Dwarf2 routines.
8683 @deftypefn {Target Hook} bool TARGET_ARM_EABI_UNWINDER
8684 This hook should be set to @code{true} on targets that use an ARM EABI
8685 based unwinding library, and @code{false} on other targets. This effects
8686 the format of unwinding tables, and how the unwinder in entered after
8687 running a cleanup. The default is @code{false}.
8690 @node Alignment Output
8691 @subsection Assembler Commands for Alignment
8693 @c prevent bad page break with this line
8694 This describes commands for alignment.
8696 @defmac JUMP_ALIGN (@var{label})
8697 The alignment (log base 2) to put in front of @var{label}, which is
8698 a common destination of jumps and has no fallthru incoming edge.
8700 This macro need not be defined if you don't want any special alignment
8701 to be done at such a time. Most machine descriptions do not currently
8704 Unless it's necessary to inspect the @var{label} parameter, it is better
8705 to set the variable @var{align_jumps} in the target's
8706 @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's
8707 selection in @var{align_jumps} in a @code{JUMP_ALIGN} implementation.
8710 @defmac LABEL_ALIGN_AFTER_BARRIER (@var{label})
8711 The alignment (log base 2) to put in front of @var{label}, which follows
8714 This macro need not be defined if you don't want any special alignment
8715 to be done at such a time. Most machine descriptions do not currently
8719 @defmac LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP
8720 The maximum number of bytes to skip when applying
8721 @code{LABEL_ALIGN_AFTER_BARRIER}. This works only if
8722 @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
8725 @defmac LOOP_ALIGN (@var{label})
8726 The alignment (log base 2) to put in front of @var{label}, which follows
8727 a @code{NOTE_INSN_LOOP_BEG} note.
8729 This macro need not be defined if you don't want any special alignment
8730 to be done at such a time. Most machine descriptions do not currently
8733 Unless it's necessary to inspect the @var{label} parameter, it is better
8734 to set the variable @code{align_loops} in the target's
8735 @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's
8736 selection in @code{align_loops} in a @code{LOOP_ALIGN} implementation.
8739 @defmac LOOP_ALIGN_MAX_SKIP
8740 The maximum number of bytes to skip when applying @code{LOOP_ALIGN}.
8741 This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
8744 @defmac LABEL_ALIGN (@var{label})
8745 The alignment (log base 2) to put in front of @var{label}.
8746 If @code{LABEL_ALIGN_AFTER_BARRIER} / @code{LOOP_ALIGN} specify a different alignment,
8747 the maximum of the specified values is used.
8749 Unless it's necessary to inspect the @var{label} parameter, it is better
8750 to set the variable @code{align_labels} in the target's
8751 @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's
8752 selection in @code{align_labels} in a @code{LABEL_ALIGN} implementation.
8755 @defmac LABEL_ALIGN_MAX_SKIP
8756 The maximum number of bytes to skip when applying @code{LABEL_ALIGN}.
8757 This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
8760 @defmac ASM_OUTPUT_SKIP (@var{stream}, @var{nbytes})
8761 A C statement to output to the stdio stream @var{stream} an assembler
8762 instruction to advance the location counter by @var{nbytes} bytes.
8763 Those bytes should be zero when loaded. @var{nbytes} will be a C
8764 expression of type @code{unsigned HOST_WIDE_INT}.
8767 @defmac ASM_NO_SKIP_IN_TEXT
8768 Define this macro if @code{ASM_OUTPUT_SKIP} should not be used in the
8769 text section because it fails to put zeros in the bytes that are skipped.
8770 This is true on many Unix systems, where the pseudo--op to skip bytes
8771 produces no-op instructions rather than zeros when used in the text
8775 @defmac ASM_OUTPUT_ALIGN (@var{stream}, @var{power})
8776 A C statement to output to the stdio stream @var{stream} an assembler
8777 command to advance the location counter to a multiple of 2 to the
8778 @var{power} bytes. @var{power} will be a C expression of type @code{int}.
8781 @defmac ASM_OUTPUT_ALIGN_WITH_NOP (@var{stream}, @var{power})
8782 Like @code{ASM_OUTPUT_ALIGN}, except that the ``nop'' instruction is used
8783 for padding, if necessary.
8786 @defmac ASM_OUTPUT_MAX_SKIP_ALIGN (@var{stream}, @var{power}, @var{max_skip})
8787 A C statement to output to the stdio stream @var{stream} an assembler
8788 command to advance the location counter to a multiple of 2 to the
8789 @var{power} bytes, but only if @var{max_skip} or fewer bytes are needed to
8790 satisfy the alignment request. @var{power} and @var{max_skip} will be
8791 a C expression of type @code{int}.
8795 @node Debugging Info
8796 @section Controlling Debugging Information Format
8798 @c prevent bad page break with this line
8799 This describes how to specify debugging information.
8802 * All Debuggers:: Macros that affect all debugging formats uniformly.
8803 * DBX Options:: Macros enabling specific options in DBX format.
8804 * DBX Hooks:: Hook macros for varying DBX format.
8805 * File Names and DBX:: Macros controlling output of file names in DBX format.
8806 * SDB and DWARF:: Macros for SDB (COFF) and DWARF formats.
8807 * VMS Debug:: Macros for VMS debug format.
8811 @subsection Macros Affecting All Debugging Formats
8813 @c prevent bad page break with this line
8814 These macros affect all debugging formats.
8816 @defmac DBX_REGISTER_NUMBER (@var{regno})
8817 A C expression that returns the DBX register number for the compiler
8818 register number @var{regno}. In the default macro provided, the value
8819 of this expression will be @var{regno} itself. But sometimes there are
8820 some registers that the compiler knows about and DBX does not, or vice
8821 versa. In such cases, some register may need to have one number in the
8822 compiler and another for DBX@.
8824 If two registers have consecutive numbers inside GCC, and they can be
8825 used as a pair to hold a multiword value, then they @emph{must} have
8826 consecutive numbers after renumbering with @code{DBX_REGISTER_NUMBER}.
8827 Otherwise, debuggers will be unable to access such a pair, because they
8828 expect register pairs to be consecutive in their own numbering scheme.
8830 If you find yourself defining @code{DBX_REGISTER_NUMBER} in way that
8831 does not preserve register pairs, then what you must do instead is
8832 redefine the actual register numbering scheme.
8835 @defmac DEBUGGER_AUTO_OFFSET (@var{x})
8836 A C expression that returns the integer offset value for an automatic
8837 variable having address @var{x} (an RTL expression). The default
8838 computation assumes that @var{x} is based on the frame-pointer and
8839 gives the offset from the frame-pointer. This is required for targets
8840 that produce debugging output for DBX or COFF-style debugging output
8841 for SDB and allow the frame-pointer to be eliminated when the
8842 @option{-g} options is used.
8845 @defmac DEBUGGER_ARG_OFFSET (@var{offset}, @var{x})
8846 A C expression that returns the integer offset value for an argument
8847 having address @var{x} (an RTL expression). The nominal offset is
8851 @defmac PREFERRED_DEBUGGING_TYPE
8852 A C expression that returns the type of debugging output GCC should
8853 produce when the user specifies just @option{-g}. Define
8854 this if you have arranged for GCC to support more than one format of
8855 debugging output. Currently, the allowable values are @code{DBX_DEBUG},
8856 @code{SDB_DEBUG}, @code{DWARF_DEBUG}, @code{DWARF2_DEBUG},
8857 @code{XCOFF_DEBUG}, @code{VMS_DEBUG}, and @code{VMS_AND_DWARF2_DEBUG}.
8859 When the user specifies @option{-ggdb}, GCC normally also uses the
8860 value of this macro to select the debugging output format, but with two
8861 exceptions. If @code{DWARF2_DEBUGGING_INFO} is defined, GCC uses the
8862 value @code{DWARF2_DEBUG}. Otherwise, if @code{DBX_DEBUGGING_INFO} is
8863 defined, GCC uses @code{DBX_DEBUG}.
8865 The value of this macro only affects the default debugging output; the
8866 user can always get a specific type of output by using @option{-gstabs},
8867 @option{-gcoff}, @option{-gdwarf-2}, @option{-gxcoff}, or @option{-gvms}.
8871 @subsection Specific Options for DBX Output
8873 @c prevent bad page break with this line
8874 These are specific options for DBX output.
8876 @defmac DBX_DEBUGGING_INFO
8877 Define this macro if GCC should produce debugging output for DBX
8878 in response to the @option{-g} option.
8881 @defmac XCOFF_DEBUGGING_INFO
8882 Define this macro if GCC should produce XCOFF format debugging output
8883 in response to the @option{-g} option. This is a variant of DBX format.
8886 @defmac DEFAULT_GDB_EXTENSIONS
8887 Define this macro to control whether GCC should by default generate
8888 GDB's extended version of DBX debugging information (assuming DBX-format
8889 debugging information is enabled at all). If you don't define the
8890 macro, the default is 1: always generate the extended information
8891 if there is any occasion to.
8894 @defmac DEBUG_SYMS_TEXT
8895 Define this macro if all @code{.stabs} commands should be output while
8896 in the text section.
8899 @defmac ASM_STABS_OP
8900 A C string constant, including spacing, naming the assembler pseudo op to
8901 use instead of @code{"\t.stabs\t"} to define an ordinary debugging symbol.
8902 If you don't define this macro, @code{"\t.stabs\t"} is used. This macro
8903 applies only to DBX debugging information format.
8906 @defmac ASM_STABD_OP
8907 A C string constant, including spacing, naming the assembler pseudo op to
8908 use instead of @code{"\t.stabd\t"} to define a debugging symbol whose
8909 value is the current location. If you don't define this macro,
8910 @code{"\t.stabd\t"} is used. This macro applies only to DBX debugging
8914 @defmac ASM_STABN_OP
8915 A C string constant, including spacing, naming the assembler pseudo op to
8916 use instead of @code{"\t.stabn\t"} to define a debugging symbol with no
8917 name. If you don't define this macro, @code{"\t.stabn\t"} is used. This
8918 macro applies only to DBX debugging information format.
8921 @defmac DBX_NO_XREFS
8922 Define this macro if DBX on your system does not support the construct
8923 @samp{xs@var{tagname}}. On some systems, this construct is used to
8924 describe a forward reference to a structure named @var{tagname}.
8925 On other systems, this construct is not supported at all.
8928 @defmac DBX_CONTIN_LENGTH
8929 A symbol name in DBX-format debugging information is normally
8930 continued (split into two separate @code{.stabs} directives) when it
8931 exceeds a certain length (by default, 80 characters). On some
8932 operating systems, DBX requires this splitting; on others, splitting
8933 must not be done. You can inhibit splitting by defining this macro
8934 with the value zero. You can override the default splitting-length by
8935 defining this macro as an expression for the length you desire.
8938 @defmac DBX_CONTIN_CHAR
8939 Normally continuation is indicated by adding a @samp{\} character to
8940 the end of a @code{.stabs} string when a continuation follows. To use
8941 a different character instead, define this macro as a character
8942 constant for the character you want to use. Do not define this macro
8943 if backslash is correct for your system.
8946 @defmac DBX_STATIC_STAB_DATA_SECTION
8947 Define this macro if it is necessary to go to the data section before
8948 outputting the @samp{.stabs} pseudo-op for a non-global static
8952 @defmac DBX_TYPE_DECL_STABS_CODE
8953 The value to use in the ``code'' field of the @code{.stabs} directive
8954 for a typedef. The default is @code{N_LSYM}.
8957 @defmac DBX_STATIC_CONST_VAR_CODE
8958 The value to use in the ``code'' field of the @code{.stabs} directive
8959 for a static variable located in the text section. DBX format does not
8960 provide any ``right'' way to do this. The default is @code{N_FUN}.
8963 @defmac DBX_REGPARM_STABS_CODE
8964 The value to use in the ``code'' field of the @code{.stabs} directive
8965 for a parameter passed in registers. DBX format does not provide any
8966 ``right'' way to do this. The default is @code{N_RSYM}.
8969 @defmac DBX_REGPARM_STABS_LETTER
8970 The letter to use in DBX symbol data to identify a symbol as a parameter
8971 passed in registers. DBX format does not customarily provide any way to
8972 do this. The default is @code{'P'}.
8975 @defmac DBX_FUNCTION_FIRST
8976 Define this macro if the DBX information for a function and its
8977 arguments should precede the assembler code for the function. Normally,
8978 in DBX format, the debugging information entirely follows the assembler
8982 @defmac DBX_BLOCKS_FUNCTION_RELATIVE
8983 Define this macro, with value 1, if the value of a symbol describing
8984 the scope of a block (@code{N_LBRAC} or @code{N_RBRAC}) should be
8985 relative to the start of the enclosing function. Normally, GCC uses
8986 an absolute address.
8989 @defmac DBX_LINES_FUNCTION_RELATIVE
8990 Define this macro, with value 1, if the value of a symbol indicating
8991 the current line number (@code{N_SLINE}) should be relative to the
8992 start of the enclosing function. Normally, GCC uses an absolute address.
8995 @defmac DBX_USE_BINCL
8996 Define this macro if GCC should generate @code{N_BINCL} and
8997 @code{N_EINCL} stabs for included header files, as on Sun systems. This
8998 macro also directs GCC to output a type number as a pair of a file
8999 number and a type number within the file. Normally, GCC does not
9000 generate @code{N_BINCL} or @code{N_EINCL} stabs, and it outputs a single
9001 number for a type number.
9005 @subsection Open-Ended Hooks for DBX Format
9007 @c prevent bad page break with this line
9008 These are hooks for DBX format.
9010 @defmac DBX_OUTPUT_LBRAC (@var{stream}, @var{name})
9011 Define this macro to say how to output to @var{stream} the debugging
9012 information for the start of a scope level for variable names. The
9013 argument @var{name} is the name of an assembler symbol (for use with
9014 @code{assemble_name}) whose value is the address where the scope begins.
9017 @defmac DBX_OUTPUT_RBRAC (@var{stream}, @var{name})
9018 Like @code{DBX_OUTPUT_LBRAC}, but for the end of a scope level.
9021 @defmac DBX_OUTPUT_NFUN (@var{stream}, @var{lscope_label}, @var{decl})
9022 Define this macro if the target machine requires special handling to
9023 output an @code{N_FUN} entry for the function @var{decl}.
9026 @defmac DBX_OUTPUT_SOURCE_LINE (@var{stream}, @var{line}, @var{counter})
9027 A C statement to output DBX debugging information before code for line
9028 number @var{line} of the current source file to the stdio stream
9029 @var{stream}. @var{counter} is the number of time the macro was
9030 invoked, including the current invocation; it is intended to generate
9031 unique labels in the assembly output.
9033 This macro should not be defined if the default output is correct, or
9034 if it can be made correct by defining @code{DBX_LINES_FUNCTION_RELATIVE}.
9037 @defmac NO_DBX_FUNCTION_END
9038 Some stabs encapsulation formats (in particular ECOFF), cannot handle the
9039 @code{.stabs "",N_FUN,,0,0,Lscope-function-1} gdb dbx extension construct.
9040 On those machines, define this macro to turn this feature off without
9041 disturbing the rest of the gdb extensions.
9044 @defmac NO_DBX_BNSYM_ENSYM
9045 Some assemblers cannot handle the @code{.stabd BNSYM/ENSYM,0,0} gdb dbx
9046 extension construct. On those machines, define this macro to turn this
9047 feature off without disturbing the rest of the gdb extensions.
9050 @node File Names and DBX
9051 @subsection File Names in DBX Format
9053 @c prevent bad page break with this line
9054 This describes file names in DBX format.
9056 @defmac DBX_OUTPUT_MAIN_SOURCE_FILENAME (@var{stream}, @var{name})
9057 A C statement to output DBX debugging information to the stdio stream
9058 @var{stream}, which indicates that file @var{name} is the main source
9059 file---the file specified as the input file for compilation.
9060 This macro is called only once, at the beginning of compilation.
9062 This macro need not be defined if the standard form of output
9063 for DBX debugging information is appropriate.
9065 It may be necessary to refer to a label equal to the beginning of the
9066 text section. You can use @samp{assemble_name (stream, ltext_label_name)}
9067 to do so. If you do this, you must also set the variable
9068 @var{used_ltext_label_name} to @code{true}.
9071 @defmac NO_DBX_MAIN_SOURCE_DIRECTORY
9072 Define this macro, with value 1, if GCC should not emit an indication
9073 of the current directory for compilation and current source language at
9074 the beginning of the file.
9077 @defmac NO_DBX_GCC_MARKER
9078 Define this macro, with value 1, if GCC should not emit an indication
9079 that this object file was compiled by GCC@. The default is to emit
9080 an @code{N_OPT} stab at the beginning of every source file, with
9081 @samp{gcc2_compiled.} for the string and value 0.
9084 @defmac DBX_OUTPUT_MAIN_SOURCE_FILE_END (@var{stream}, @var{name})
9085 A C statement to output DBX debugging information at the end of
9086 compilation of the main source file @var{name}. Output should be
9087 written to the stdio stream @var{stream}.
9089 If you don't define this macro, nothing special is output at the end
9090 of compilation, which is correct for most machines.
9093 @defmac DBX_OUTPUT_NULL_N_SO_AT_MAIN_SOURCE_FILE_END
9094 Define this macro @emph{instead of} defining
9095 @code{DBX_OUTPUT_MAIN_SOURCE_FILE_END}, if what needs to be output at
9096 the end of compilation is an @code{N_SO} stab with an empty string,
9097 whose value is the highest absolute text address in the file.
9102 @subsection Macros for SDB and DWARF Output
9104 @c prevent bad page break with this line
9105 Here are macros for SDB and DWARF output.
9107 @defmac SDB_DEBUGGING_INFO
9108 Define this macro if GCC should produce COFF-style debugging output
9109 for SDB in response to the @option{-g} option.
9112 @defmac DWARF2_DEBUGGING_INFO
9113 Define this macro if GCC should produce dwarf version 2 format
9114 debugging output in response to the @option{-g} option.
9116 @deftypefn {Target Hook} int TARGET_DWARF_CALLING_CONVENTION (tree @var{function})
9117 Define this to enable the dwarf attribute @code{DW_AT_calling_convention} to
9118 be emitted for each function. Instead of an integer return the enum
9119 value for the @code{DW_CC_} tag.
9122 To support optional call frame debugging information, you must also
9123 define @code{INCOMING_RETURN_ADDR_RTX} and either set
9124 @code{RTX_FRAME_RELATED_P} on the prologue insns if you use RTL for the
9125 prologue, or call @code{dwarf2out_def_cfa} and @code{dwarf2out_reg_save}
9126 as appropriate from @code{TARGET_ASM_FUNCTION_PROLOGUE} if you don't.
9129 @defmac DWARF2_FRAME_INFO
9130 Define this macro to a nonzero value if GCC should always output
9131 Dwarf 2 frame information. If @code{DWARF2_UNWIND_INFO}
9132 (@pxref{Exception Region Output} is nonzero, GCC will output this
9133 information not matter how you define @code{DWARF2_FRAME_INFO}.
9136 @defmac DWARF2_ASM_LINE_DEBUG_INFO
9137 Define this macro to be a nonzero value if the assembler can generate Dwarf 2
9138 line debug info sections. This will result in much more compact line number
9139 tables, and hence is desirable if it works.
9142 @defmac ASM_OUTPUT_DWARF_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
9143 A C statement to issue assembly directives that create a difference
9144 @var{lab1} minus @var{lab2}, using an integer of the given @var{size}.
9147 @defmac ASM_OUTPUT_DWARF_OFFSET (@var{stream}, @var{size}, @var{label}, @var{section})
9148 A C statement to issue assembly directives that create a
9149 section-relative reference to the given @var{label}, using an integer of the
9150 given @var{size}. The label is known to be defined in the given @var{section}.
9153 @defmac ASM_OUTPUT_DWARF_PCREL (@var{stream}, @var{size}, @var{label})
9154 A C statement to issue assembly directives that create a self-relative
9155 reference to the given @var{label}, using an integer of the given @var{size}.
9158 @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_DWARF_DTPREL (FILE *@var{FILE}, int @var{size}, rtx @var{x})
9159 If defined, this target hook is a function which outputs a DTP-relative
9160 reference to the given TLS symbol of the specified size.
9163 @defmac PUT_SDB_@dots{}
9164 Define these macros to override the assembler syntax for the special
9165 SDB assembler directives. See @file{sdbout.c} for a list of these
9166 macros and their arguments. If the standard syntax is used, you need
9167 not define them yourself.
9171 Some assemblers do not support a semicolon as a delimiter, even between
9172 SDB assembler directives. In that case, define this macro to be the
9173 delimiter to use (usually @samp{\n}). It is not necessary to define
9174 a new set of @code{PUT_SDB_@var{op}} macros if this is the only change
9178 @defmac SDB_ALLOW_UNKNOWN_REFERENCES
9179 Define this macro to allow references to unknown structure,
9180 union, or enumeration tags to be emitted. Standard COFF does not
9181 allow handling of unknown references, MIPS ECOFF has support for
9185 @defmac SDB_ALLOW_FORWARD_REFERENCES
9186 Define this macro to allow references to structure, union, or
9187 enumeration tags that have not yet been seen to be handled. Some
9188 assemblers choke if forward tags are used, while some require it.
9191 @defmac SDB_OUTPUT_SOURCE_LINE (@var{stream}, @var{line})
9192 A C statement to output SDB debugging information before code for line
9193 number @var{line} of the current source file to the stdio stream
9194 @var{stream}. The default is to emit an @code{.ln} directive.
9199 @subsection Macros for VMS Debug Format
9201 @c prevent bad page break with this line
9202 Here are macros for VMS debug format.
9204 @defmac VMS_DEBUGGING_INFO
9205 Define this macro if GCC should produce debugging output for VMS
9206 in response to the @option{-g} option. The default behavior for VMS
9207 is to generate minimal debug info for a traceback in the absence of
9208 @option{-g} unless explicitly overridden with @option{-g0}. This
9209 behavior is controlled by @code{OPTIMIZATION_OPTIONS} and
9210 @code{OVERRIDE_OPTIONS}.
9213 @node Floating Point
9214 @section Cross Compilation and Floating Point
9215 @cindex cross compilation and floating point
9216 @cindex floating point and cross compilation
9218 While all modern machines use twos-complement representation for integers,
9219 there are a variety of representations for floating point numbers. This
9220 means that in a cross-compiler the representation of floating point numbers
9221 in the compiled program may be different from that used in the machine
9222 doing the compilation.
9224 Because different representation systems may offer different amounts of
9225 range and precision, all floating point constants must be represented in
9226 the target machine's format. Therefore, the cross compiler cannot
9227 safely use the host machine's floating point arithmetic; it must emulate
9228 the target's arithmetic. To ensure consistency, GCC always uses
9229 emulation to work with floating point values, even when the host and
9230 target floating point formats are identical.
9232 The following macros are provided by @file{real.h} for the compiler to
9233 use. All parts of the compiler which generate or optimize
9234 floating-point calculations must use these macros. They may evaluate
9235 their operands more than once, so operands must not have side effects.
9237 @defmac REAL_VALUE_TYPE
9238 The C data type to be used to hold a floating point value in the target
9239 machine's format. Typically this is a @code{struct} containing an
9240 array of @code{HOST_WIDE_INT}, but all code should treat it as an opaque
9244 @deftypefn Macro int REAL_VALUES_EQUAL (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
9245 Compares for equality the two values, @var{x} and @var{y}. If the target
9246 floating point format supports negative zeroes and/or NaNs,
9247 @samp{REAL_VALUES_EQUAL (-0.0, 0.0)} is true, and
9248 @samp{REAL_VALUES_EQUAL (NaN, NaN)} is false.
9251 @deftypefn Macro int REAL_VALUES_LESS (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
9252 Tests whether @var{x} is less than @var{y}.
9255 @deftypefn Macro HOST_WIDE_INT REAL_VALUE_FIX (REAL_VALUE_TYPE @var{x})
9256 Truncates @var{x} to a signed integer, rounding toward zero.
9259 @deftypefn Macro {unsigned HOST_WIDE_INT} REAL_VALUE_UNSIGNED_FIX (REAL_VALUE_TYPE @var{x})
9260 Truncates @var{x} to an unsigned integer, rounding toward zero. If
9261 @var{x} is negative, returns zero.
9264 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ATOF (const char *@var{string}, enum machine_mode @var{mode})
9265 Converts @var{string} into a floating point number in the target machine's
9266 representation for mode @var{mode}. This routine can handle both
9267 decimal and hexadecimal floating point constants, using the syntax
9268 defined by the C language for both.
9271 @deftypefn Macro int REAL_VALUE_NEGATIVE (REAL_VALUE_TYPE @var{x})
9272 Returns 1 if @var{x} is negative (including negative zero), 0 otherwise.
9275 @deftypefn Macro int REAL_VALUE_ISINF (REAL_VALUE_TYPE @var{x})
9276 Determines whether @var{x} represents infinity (positive or negative).
9279 @deftypefn Macro int REAL_VALUE_ISNAN (REAL_VALUE_TYPE @var{x})
9280 Determines whether @var{x} represents a ``NaN'' (not-a-number).
9283 @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})
9284 Calculates an arithmetic operation on the two floating point values
9285 @var{x} and @var{y}, storing the result in @var{output} (which must be a
9288 The operation to be performed is specified by @var{code}. Only the
9289 following codes are supported: @code{PLUS_EXPR}, @code{MINUS_EXPR},
9290 @code{MULT_EXPR}, @code{RDIV_EXPR}, @code{MAX_EXPR}, @code{MIN_EXPR}.
9292 If @code{REAL_ARITHMETIC} is asked to evaluate division by zero and the
9293 target's floating point format cannot represent infinity, it will call
9294 @code{abort}. Callers should check for this situation first, using
9295 @code{MODE_HAS_INFINITIES}. @xref{Storage Layout}.
9298 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_NEGATE (REAL_VALUE_TYPE @var{x})
9299 Returns the negative of the floating point value @var{x}.
9302 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ABS (REAL_VALUE_TYPE @var{x})
9303 Returns the absolute value of @var{x}.
9306 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_TRUNCATE (REAL_VALUE_TYPE @var{mode}, enum machine_mode @var{x})
9307 Truncates the floating point value @var{x} to fit in @var{mode}. The
9308 return value is still a full-size @code{REAL_VALUE_TYPE}, but it has an
9309 appropriate bit pattern to be output as a floating constant whose
9310 precision accords with mode @var{mode}.
9313 @deftypefn Macro void REAL_VALUE_TO_INT (HOST_WIDE_INT @var{low}, HOST_WIDE_INT @var{high}, REAL_VALUE_TYPE @var{x})
9314 Converts a floating point value @var{x} into a double-precision integer
9315 which is then stored into @var{low} and @var{high}. If the value is not
9316 integral, it is truncated.
9319 @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})
9320 Converts a double-precision integer found in @var{low} and @var{high},
9321 into a floating point value which is then stored into @var{x}. The
9322 value is truncated to fit in mode @var{mode}.
9325 @node Mode Switching
9326 @section Mode Switching Instructions
9327 @cindex mode switching
9328 The following macros control mode switching optimizations:
9330 @defmac OPTIMIZE_MODE_SWITCHING (@var{entity})
9331 Define this macro if the port needs extra instructions inserted for mode
9332 switching in an optimizing compilation.
9334 For an example, the SH4 can perform both single and double precision
9335 floating point operations, but to perform a single precision operation,
9336 the FPSCR PR bit has to be cleared, while for a double precision
9337 operation, this bit has to be set. Changing the PR bit requires a general
9338 purpose register as a scratch register, hence these FPSCR sets have to
9339 be inserted before reload, i.e.@: you can't put this into instruction emitting
9340 or @code{TARGET_MACHINE_DEPENDENT_REORG}.
9342 You can have multiple entities that are mode-switched, and select at run time
9343 which entities actually need it. @code{OPTIMIZE_MODE_SWITCHING} should
9344 return nonzero for any @var{entity} that needs mode-switching.
9345 If you define this macro, you also have to define
9346 @code{NUM_MODES_FOR_MODE_SWITCHING}, @code{MODE_NEEDED},
9347 @code{MODE_PRIORITY_TO_MODE} and @code{EMIT_MODE_SET}.
9348 @code{MODE_AFTER}, @code{MODE_ENTRY}, and @code{MODE_EXIT}
9352 @defmac NUM_MODES_FOR_MODE_SWITCHING
9353 If you define @code{OPTIMIZE_MODE_SWITCHING}, you have to define this as
9354 initializer for an array of integers. Each initializer element
9355 N refers to an entity that needs mode switching, and specifies the number
9356 of different modes that might need to be set for this entity.
9357 The position of the initializer in the initializer---starting counting at
9358 zero---determines the integer that is used to refer to the mode-switched
9360 In macros that take mode arguments / yield a mode result, modes are
9361 represented as numbers 0 @dots{} N @minus{} 1. N is used to specify that no mode
9362 switch is needed / supplied.
9365 @defmac MODE_NEEDED (@var{entity}, @var{insn})
9366 @var{entity} is an integer specifying a mode-switched entity. If
9367 @code{OPTIMIZE_MODE_SWITCHING} is defined, you must define this macro to
9368 return an integer value not larger than the corresponding element in
9369 @code{NUM_MODES_FOR_MODE_SWITCHING}, to denote the mode that @var{entity} must
9370 be switched into prior to the execution of @var{insn}.
9373 @defmac MODE_AFTER (@var{mode}, @var{insn})
9374 If this macro is defined, it is evaluated for every @var{insn} during
9375 mode switching. It determines the mode that an insn results in (if
9376 different from the incoming mode).
9379 @defmac MODE_ENTRY (@var{entity})
9380 If this macro is defined, it is evaluated for every @var{entity} that needs
9381 mode switching. It should evaluate to an integer, which is a mode that
9382 @var{entity} is assumed to be switched to at function entry. If @code{MODE_ENTRY}
9383 is defined then @code{MODE_EXIT} must be defined.
9386 @defmac MODE_EXIT (@var{entity})
9387 If this macro is defined, it is evaluated for every @var{entity} that needs
9388 mode switching. It should evaluate to an integer, which is a mode that
9389 @var{entity} is assumed to be switched to at function exit. If @code{MODE_EXIT}
9390 is defined then @code{MODE_ENTRY} must be defined.
9393 @defmac MODE_PRIORITY_TO_MODE (@var{entity}, @var{n})
9394 This macro specifies the order in which modes for @var{entity} are processed.
9395 0 is the highest priority, @code{NUM_MODES_FOR_MODE_SWITCHING[@var{entity}] - 1} the
9396 lowest. The value of the macro should be an integer designating a mode
9397 for @var{entity}. For any fixed @var{entity}, @code{mode_priority_to_mode}
9398 (@var{entity}, @var{n}) shall be a bijection in 0 @dots{}
9399 @code{num_modes_for_mode_switching[@var{entity}] - 1}.
9402 @defmac EMIT_MODE_SET (@var{entity}, @var{mode}, @var{hard_regs_live})
9403 Generate one or more insns to set @var{entity} to @var{mode}.
9404 @var{hard_reg_live} is the set of hard registers live at the point where
9405 the insn(s) are to be inserted.
9408 @node Target Attributes
9409 @section Defining target-specific uses of @code{__attribute__}
9410 @cindex target attributes
9411 @cindex machine attributes
9412 @cindex attributes, target-specific
9414 Target-specific attributes may be defined for functions, data and types.
9415 These are described using the following target hooks; they also need to
9416 be documented in @file{extend.texi}.
9418 @deftypevr {Target Hook} {const struct attribute_spec *} TARGET_ATTRIBUTE_TABLE
9419 If defined, this target hook points to an array of @samp{struct
9420 attribute_spec} (defined in @file{tree.h}) specifying the machine
9421 specific attributes for this target and some of the restrictions on the
9422 entities to which these attributes are applied and the arguments they
9426 @deftypefn {Target Hook} int TARGET_COMP_TYPE_ATTRIBUTES (tree @var{type1}, tree @var{type2})
9427 If defined, this target hook is a function which returns zero if the attributes on
9428 @var{type1} and @var{type2} are incompatible, one if they are compatible,
9429 and two if they are nearly compatible (which causes a warning to be
9430 generated). If this is not defined, machine-specific attributes are
9431 supposed always to be compatible.
9434 @deftypefn {Target Hook} void TARGET_SET_DEFAULT_TYPE_ATTRIBUTES (tree @var{type})
9435 If defined, this target hook is a function which assigns default attributes to
9436 newly defined @var{type}.
9439 @deftypefn {Target Hook} tree TARGET_MERGE_TYPE_ATTRIBUTES (tree @var{type1}, tree @var{type2})
9440 Define this target hook if the merging of type attributes needs special
9441 handling. If defined, the result is a list of the combined
9442 @code{TYPE_ATTRIBUTES} of @var{type1} and @var{type2}. It is assumed
9443 that @code{comptypes} has already been called and returned 1. This
9444 function may call @code{merge_attributes} to handle machine-independent
9448 @deftypefn {Target Hook} tree TARGET_MERGE_DECL_ATTRIBUTES (tree @var{olddecl}, tree @var{newdecl})
9449 Define this target hook if the merging of decl attributes needs special
9450 handling. If defined, the result is a list of the combined
9451 @code{DECL_ATTRIBUTES} of @var{olddecl} and @var{newdecl}.
9452 @var{newdecl} is a duplicate declaration of @var{olddecl}. Examples of
9453 when this is needed are when one attribute overrides another, or when an
9454 attribute is nullified by a subsequent definition. This function may
9455 call @code{merge_attributes} to handle machine-independent merging.
9457 @findex TARGET_DLLIMPORT_DECL_ATTRIBUTES
9458 If the only target-specific handling you require is @samp{dllimport}
9459 for Microsoft Windows targets, you should define the macro
9460 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES} to @code{1}. The compiler
9461 will then define a function called
9462 @code{merge_dllimport_decl_attributes} which can then be defined as
9463 the expansion of @code{TARGET_MERGE_DECL_ATTRIBUTES}. You can also
9464 add @code{handle_dll_attribute} in the attribute table for your port
9465 to perform initial processing of the @samp{dllimport} and
9466 @samp{dllexport} attributes. This is done in @file{i386/cygwin.h} and
9467 @file{i386/i386.c}, for example.
9470 @deftypefn {Target Hook} bool TARGET_VALID_DLLIMPORT_ATTRIBUTE_P (tree @var{decl})
9471 @var{decl} is a variable or function with @code{__attribute__((dllimport))}
9472 specified. Use this hook if the target needs to add extra validation
9473 checks to @code{handle_dll_attribute}.
9476 @defmac TARGET_DECLSPEC
9477 Define this macro to a nonzero value if you want to treat
9478 @code{__declspec(X)} as equivalent to @code{__attribute((X))}. By
9479 default, this behavior is enabled only for targets that define
9480 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES}. The current implementation
9481 of @code{__declspec} is via a built-in macro, but you should not rely
9482 on this implementation detail.
9485 @deftypefn {Target Hook} void TARGET_INSERT_ATTRIBUTES (tree @var{node}, tree *@var{attr_ptr})
9486 Define this target hook if you want to be able to add attributes to a decl
9487 when it is being created. This is normally useful for back ends which
9488 wish to implement a pragma by using the attributes which correspond to
9489 the pragma's effect. The @var{node} argument is the decl which is being
9490 created. The @var{attr_ptr} argument is a pointer to the attribute list
9491 for this decl. The list itself should not be modified, since it may be
9492 shared with other decls, but attributes may be chained on the head of
9493 the list and @code{*@var{attr_ptr}} modified to point to the new
9494 attributes, or a copy of the list may be made if further changes are
9498 @deftypefn {Target Hook} bool TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P (tree @var{fndecl})
9500 This target hook returns @code{true} if it is ok to inline @var{fndecl}
9501 into the current function, despite its having target-specific
9502 attributes, @code{false} otherwise. By default, if a function has a
9503 target specific attribute attached to it, it will not be inlined.
9506 @deftypefn {Target Hook} bool TARGET_VALID_OPTION_ATTRIBUTE_P (tree @var{fndecl}, tree @var{name}, tree @var{args}, int @var{flags})
9507 This hook is called to parse the @code{attribute(option("..."))}, and
9508 it allows the function to set different target machine compile time
9509 options for the current function that might be different than the
9510 options specified on the command line. The hook should return
9511 @code{true} if the options are valid.
9513 The hook should set the @var{DECL_FUNCTION_SPECIFIC_TARGET} field in
9514 the function declaration to hold a pointer to a target specific
9515 @var{struct cl_target_option} structure.
9518 @deftypefn {Target Hook} void TARGET_OPTION_SAVE (struct cl_target_option *@var{ptr})
9519 This hook is called to save any additional target specific information
9520 in the @var{struct cl_target_option} structure for function specific
9522 @xref{Option file format}.
9525 @deftypefn {Target Hook} void TARGET_OPTION_RESTORE (struct cl_target_option *@var{ptr})
9526 This hook is called to restore any additional target specific
9527 information in the @var{struct cl_target_option} structure for
9528 function specific options.
9531 @deftypefn {Target Hook} void TARGET_OPTION_PRINT (struct cl_target_option *@var{ptr})
9532 This hook is called to print any additional target specific
9533 information in the @var{struct cl_target_option} structure for
9534 function specific options.
9537 @deftypefn {Target Hook} bool TARGET_OPTION_PRAGMA_PARSE (target @var{args})
9538 This target hook parses the options for @code{#pragma GCC option} to
9539 set the machine specific options for functions that occur later in the
9540 input stream. The options should be the same as handled by the
9541 @code{TARGET_VALID_OPTION_ATTRIBUTE_P} hook.
9544 @deftypefn {Target Hook} bool TARGET_CAN_INLINE_P (tree @var{caller}, tree @var{callee})
9545 This target hook returns @code{false} if the @var{caller} function
9546 cannot inline @var{callee}, based on target specific information. By
9547 default, inlining is not allowed if the callee function has function
9548 specific target options and the caller does not use the same options.
9552 @section Emulating TLS
9553 @cindex Emulated TLS
9555 For targets whose psABI does not provide Thread Local Storage via
9556 specific relocations and instruction sequences, an emulation layer is
9557 used. A set of target hooks allows this emulation layer to be
9558 configured for the requirements of a particular target. For instance
9559 the psABI may in fact specify TLS support in terms of an emulation
9562 The emulation layer works by creating a control object for every TLS
9563 object. To access the TLS object, a lookup function is provided
9564 which, when given the address of the control object, will return the
9565 address of the current thread's instance of the TLS object.
9567 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_GET_ADDRESS
9568 Contains the name of the helper function that uses a TLS control
9569 object to locate a TLS instance. The default causes libgcc's
9570 emulated TLS helper function to be used.
9573 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_REGISTER_COMMON
9574 Contains the name of the helper function that should be used at
9575 program startup to register TLS objects that are implicitly
9576 initialized to zero. If this is @code{NULL}, all TLS objects will
9577 have explicit initializers. The default causes libgcc's emulated TLS
9578 registration function to be used.
9581 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_VAR_SECTION
9582 Contains the name of the section in which TLS control variables should
9583 be placed. The default of @code{NULL} allows these to be placed in
9587 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_TMPL_SECTION
9588 Contains the name of the section in which TLS initializers should be
9589 placed. The default of @code{NULL} allows these to be placed in any
9593 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_VAR_PREFIX
9594 Contains the prefix to be prepended to TLS control variable names.
9595 The default of @code{NULL} uses a target-specific prefix.
9598 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_TMPL_PREFIX
9599 Contains the prefix to be prepended to TLS initializer objects. The
9600 default of @code{NULL} uses a target-specific prefix.
9603 @deftypefn {Target Hook} tree TARGET_EMUTLS_VAR_FIELDS (tree @var{type}, tree *@var{name})
9604 Specifies a function that generates the FIELD_DECLs for a TLS control
9605 object type. @var{type} is the RECORD_TYPE the fields are for and
9606 @var{name} should be filled with the structure tag, if the default of
9607 @code{__emutls_object} is unsuitable. The default creates a type suitable
9608 for libgcc's emulated TLS function.
9611 @deftypefn {Target Hook} tree TARGET_EMUTLS_VAR_INIT (tree @var{var}, tree @var{decl}, tree @var{tmpl_addr})
9612 Specifies a function that generates the CONSTRUCTOR to initialize a
9613 TLS control object. @var{var} is the TLS control object, @var{decl}
9614 is the TLS object and @var{tmpl_addr} is the address of the
9615 initializer. The default initializes libgcc's emulated TLS control object.
9618 @deftypevr {Target Hook} {bool} TARGET_EMUTLS_VAR_ALIGN_FIXED
9619 Specifies whether the alignment of TLS control variable objects is
9620 fixed and should not be increased as some backends may do to optimize
9621 single objects. The default is false.
9624 @deftypevr {Target Hook} {bool} TARGET_EMUTLS_DEBUG_FORM_TLS_ADDRESS
9625 Specifies whether a DWARF @code{DW_OP_form_tls_address} location descriptor
9626 may be used to describe emulated TLS control objects.
9629 @node MIPS Coprocessors
9630 @section Defining coprocessor specifics for MIPS targets.
9631 @cindex MIPS coprocessor-definition macros
9633 The MIPS specification allows MIPS implementations to have as many as 4
9634 coprocessors, each with as many as 32 private registers. GCC supports
9635 accessing these registers and transferring values between the registers
9636 and memory using asm-ized variables. For example:
9639 register unsigned int cp0count asm ("c0r1");
9645 (``c0r1'' is the default name of register 1 in coprocessor 0; alternate
9646 names may be added as described below, or the default names may be
9647 overridden entirely in @code{SUBTARGET_CONDITIONAL_REGISTER_USAGE}.)
9649 Coprocessor registers are assumed to be epilogue-used; sets to them will
9650 be preserved even if it does not appear that the register is used again
9651 later in the function.
9653 Another note: according to the MIPS spec, coprocessor 1 (if present) is
9654 the FPU@. One accesses COP1 registers through standard mips
9655 floating-point support; they are not included in this mechanism.
9657 There is one macro used in defining the MIPS coprocessor interface which
9658 you may want to override in subtargets; it is described below.
9660 @defmac ALL_COP_ADDITIONAL_REGISTER_NAMES
9661 A comma-separated list (with leading comma) of pairs describing the
9662 alternate names of coprocessor registers. The format of each entry should be
9664 @{ @var{alternatename}, @var{register_number}@}
9670 @section Parameters for Precompiled Header Validity Checking
9671 @cindex parameters, precompiled headers
9673 @deftypefn {Target Hook} void *TARGET_GET_PCH_VALIDITY (size_t *@var{sz})
9674 This hook returns the data needed by @code{TARGET_PCH_VALID_P} and sets
9675 @samp{*@var{sz}} to the size of the data in bytes.
9678 @deftypefn {Target Hook} const char *TARGET_PCH_VALID_P (const void *@var{data}, size_t @var{sz})
9679 This hook checks whether the options used to create a PCH file are
9680 compatible with the current settings. It returns @code{NULL}
9681 if so and a suitable error message if not. Error messages will
9682 be presented to the user and must be localized using @samp{_(@var{msg})}.
9684 @var{data} is the data that was returned by @code{TARGET_GET_PCH_VALIDITY}
9685 when the PCH file was created and @var{sz} is the size of that data in bytes.
9686 It's safe to assume that the data was created by the same version of the
9687 compiler, so no format checking is needed.
9689 The default definition of @code{default_pch_valid_p} should be
9690 suitable for most targets.
9693 @deftypefn {Target Hook} const char *TARGET_CHECK_PCH_TARGET_FLAGS (int @var{pch_flags})
9694 If this hook is nonnull, the default implementation of
9695 @code{TARGET_PCH_VALID_P} will use it to check for compatible values
9696 of @code{target_flags}. @var{pch_flags} specifies the value that
9697 @code{target_flags} had when the PCH file was created. The return
9698 value is the same as for @code{TARGET_PCH_VALID_P}.
9702 @section C++ ABI parameters
9703 @cindex parameters, c++ abi
9705 @deftypefn {Target Hook} tree TARGET_CXX_GUARD_TYPE (void)
9706 Define this hook to override the integer type used for guard variables.
9707 These are used to implement one-time construction of static objects. The
9708 default is long_long_integer_type_node.
9711 @deftypefn {Target Hook} bool TARGET_CXX_GUARD_MASK_BIT (void)
9712 This hook determines how guard variables are used. It should return
9713 @code{false} (the default) if first byte should be used. A return value of
9714 @code{true} indicates the least significant bit should be used.
9717 @deftypefn {Target Hook} tree TARGET_CXX_GET_COOKIE_SIZE (tree @var{type})
9718 This hook returns the size of the cookie to use when allocating an array
9719 whose elements have the indicated @var{type}. Assumes that it is already
9720 known that a cookie is needed. The default is
9721 @code{max(sizeof (size_t), alignof(type))}, as defined in section 2.7 of the
9722 IA64/Generic C++ ABI@.
9725 @deftypefn {Target Hook} bool TARGET_CXX_COOKIE_HAS_SIZE (void)
9726 This hook should return @code{true} if the element size should be stored in
9727 array cookies. The default is to return @code{false}.
9730 @deftypefn {Target Hook} int TARGET_CXX_IMPORT_EXPORT_CLASS (tree @var{type}, int @var{import_export})
9731 If defined by a backend this hook allows the decision made to export
9732 class @var{type} to be overruled. Upon entry @var{import_export}
9733 will contain 1 if the class is going to be exported, @minus{}1 if it is going
9734 to be imported and 0 otherwise. This function should return the
9735 modified value and perform any other actions necessary to support the
9736 backend's targeted operating system.
9739 @deftypefn {Target Hook} bool TARGET_CXX_CDTOR_RETURNS_THIS (void)
9740 This hook should return @code{true} if constructors and destructors return
9741 the address of the object created/destroyed. The default is to return
9745 @deftypefn {Target Hook} bool TARGET_CXX_KEY_METHOD_MAY_BE_INLINE (void)
9746 This hook returns true if the key method for a class (i.e., the method
9747 which, if defined in the current translation unit, causes the virtual
9748 table to be emitted) may be an inline function. Under the standard
9749 Itanium C++ ABI the key method may be an inline function so long as
9750 the function is not declared inline in the class definition. Under
9751 some variants of the ABI, an inline function can never be the key
9752 method. The default is to return @code{true}.
9755 @deftypefn {Target Hook} void TARGET_CXX_DETERMINE_CLASS_DATA_VISIBILITY (tree @var{decl})
9756 @var{decl} is a virtual table, virtual table table, typeinfo object,
9757 or other similar implicit class data object that will be emitted with
9758 external linkage in this translation unit. No ELF visibility has been
9759 explicitly specified. If the target needs to specify a visibility
9760 other than that of the containing class, use this hook to set
9761 @code{DECL_VISIBILITY} and @code{DECL_VISIBILITY_SPECIFIED}.
9764 @deftypefn {Target Hook} bool TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT (void)
9765 This hook returns true (the default) if virtual tables and other
9766 similar implicit class data objects are always COMDAT if they have
9767 external linkage. If this hook returns false, then class data for
9768 classes whose virtual table will be emitted in only one translation
9769 unit will not be COMDAT.
9772 @deftypefn {Target Hook} bool TARGET_CXX_LIBRARY_RTTI_COMDAT (void)
9773 This hook returns true (the default) if the RTTI information for
9774 the basic types which is defined in the C++ runtime should always
9775 be COMDAT, false if it should not be COMDAT.
9778 @deftypefn {Target Hook} bool TARGET_CXX_USE_AEABI_ATEXIT (void)
9779 This hook returns true if @code{__aeabi_atexit} (as defined by the ARM EABI)
9780 should be used to register static destructors when @option{-fuse-cxa-atexit}
9781 is in effect. The default is to return false to use @code{__cxa_atexit}.
9784 @deftypefn {Target Hook} bool TARGET_CXX_USE_ATEXIT_FOR_CXA_ATEXIT (void)
9785 This hook returns true if the target @code{atexit} function can be used
9786 in the same manner as @code{__cxa_atexit} to register C++ static
9787 destructors. This requires that @code{atexit}-registered functions in
9788 shared libraries are run in the correct order when the libraries are
9789 unloaded. The default is to return false.
9792 @deftypefn {Target Hook} void TARGET_CXX_ADJUST_CLASS_AT_DEFINITION (tree @var{type})
9793 @var{type} is a C++ class (i.e., RECORD_TYPE or UNION_TYPE) that has just been
9794 defined. Use this hook to make adjustments to the class (eg, tweak
9795 visibility or perform any other required target modifications).
9799 @section Miscellaneous Parameters
9800 @cindex parameters, miscellaneous
9802 @c prevent bad page break with this line
9803 Here are several miscellaneous parameters.
9805 @defmac HAS_LONG_COND_BRANCH
9806 Define this boolean macro to indicate whether or not your architecture
9807 has conditional branches that can span all of memory. It is used in
9808 conjunction with an optimization that partitions hot and cold basic
9809 blocks into separate sections of the executable. If this macro is
9810 set to false, gcc will convert any conditional branches that attempt
9811 to cross between sections into unconditional branches or indirect jumps.
9814 @defmac HAS_LONG_UNCOND_BRANCH
9815 Define this boolean macro to indicate whether or not your architecture
9816 has unconditional branches that can span all of memory. It is used in
9817 conjunction with an optimization that partitions hot and cold basic
9818 blocks into separate sections of the executable. If this macro is
9819 set to false, gcc will convert any unconditional branches that attempt
9820 to cross between sections into indirect jumps.
9823 @defmac CASE_VECTOR_MODE
9824 An alias for a machine mode name. This is the machine mode that
9825 elements of a jump-table should have.
9828 @defmac CASE_VECTOR_SHORTEN_MODE (@var{min_offset}, @var{max_offset}, @var{body})
9829 Optional: return the preferred mode for an @code{addr_diff_vec}
9830 when the minimum and maximum offset are known. If you define this,
9831 it enables extra code in branch shortening to deal with @code{addr_diff_vec}.
9832 To make this work, you also have to define @code{INSN_ALIGN} and
9833 make the alignment for @code{addr_diff_vec} explicit.
9834 The @var{body} argument is provided so that the offset_unsigned and scale
9835 flags can be updated.
9838 @defmac CASE_VECTOR_PC_RELATIVE
9839 Define this macro to be a C expression to indicate when jump-tables
9840 should contain relative addresses. You need not define this macro if
9841 jump-tables never contain relative addresses, or jump-tables should
9842 contain relative addresses only when @option{-fPIC} or @option{-fPIC}
9846 @deftypefn {Target Hook} unsigned int TARGET_CASE_VALUES_THRESHOLD (void)
9847 This function return the smallest number of different values for which it
9848 is best to use a jump-table instead of a tree of conditional branches.
9849 The default is four for machines with a @code{casesi} instruction and
9850 five otherwise. This is best for most machines.
9853 @defmac CASE_USE_BIT_TESTS
9854 Define this macro to be a C expression to indicate whether C switch
9855 statements may be implemented by a sequence of bit tests. This is
9856 advantageous on processors that can efficiently implement left shift
9857 of 1 by the number of bits held in a register, but inappropriate on
9858 targets that would require a loop. By default, this macro returns
9859 @code{true} if the target defines an @code{ashlsi3} pattern, and
9860 @code{false} otherwise.
9863 @defmac WORD_REGISTER_OPERATIONS
9864 Define this macro if operations between registers with integral mode
9865 smaller than a word are always performed on the entire register.
9866 Most RISC machines have this property and most CISC machines do not.
9869 @defmac LOAD_EXTEND_OP (@var{mem_mode})
9870 Define this macro to be a C expression indicating when insns that read
9871 memory in @var{mem_mode}, an integral mode narrower than a word, set the
9872 bits outside of @var{mem_mode} to be either the sign-extension or the
9873 zero-extension of the data read. Return @code{SIGN_EXTEND} for values
9874 of @var{mem_mode} for which the
9875 insn sign-extends, @code{ZERO_EXTEND} for which it zero-extends, and
9876 @code{UNKNOWN} for other modes.
9878 This macro is not called with @var{mem_mode} non-integral or with a width
9879 greater than or equal to @code{BITS_PER_WORD}, so you may return any
9880 value in this case. Do not define this macro if it would always return
9881 @code{UNKNOWN}. On machines where this macro is defined, you will normally
9882 define it as the constant @code{SIGN_EXTEND} or @code{ZERO_EXTEND}.
9884 You may return a non-@code{UNKNOWN} value even if for some hard registers
9885 the sign extension is not performed, if for the @code{REGNO_REG_CLASS}
9886 of these hard registers @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero
9887 when the @var{from} mode is @var{mem_mode} and the @var{to} mode is any
9888 integral mode larger than this but not larger than @code{word_mode}.
9890 You must return @code{UNKNOWN} if for some hard registers that allow this
9891 mode, @code{CANNOT_CHANGE_MODE_CLASS} says that they cannot change to
9892 @code{word_mode}, but that they can change to another integral mode that
9893 is larger then @var{mem_mode} but still smaller than @code{word_mode}.
9896 @defmac SHORT_IMMEDIATES_SIGN_EXTEND
9897 Define this macro if loading short immediate values into registers sign
9901 @defmac FIXUNS_TRUNC_LIKE_FIX_TRUNC
9902 Define this macro if the same instructions that convert a floating
9903 point number to a signed fixed point number also convert validly to an
9907 @deftypefn {Target Hook} int TARGET_MIN_DIVISIONS_FOR_RECIP_MUL (enum machine_mode @var{mode})
9908 When @option{-ffast-math} is in effect, GCC tries to optimize
9909 divisions by the same divisor, by turning them into multiplications by
9910 the reciprocal. This target hook specifies the minimum number of divisions
9911 that should be there for GCC to perform the optimization for a variable
9912 of mode @var{mode}. The default implementation returns 3 if the machine
9913 has an instruction for the division, and 2 if it does not.
9917 The maximum number of bytes that a single instruction can move quickly
9918 between memory and registers or between two memory locations.
9921 @defmac MAX_MOVE_MAX
9922 The maximum number of bytes that a single instruction can move quickly
9923 between memory and registers or between two memory locations. If this
9924 is undefined, the default is @code{MOVE_MAX}. Otherwise, it is the
9925 constant value that is the largest value that @code{MOVE_MAX} can have
9929 @defmac SHIFT_COUNT_TRUNCATED
9930 A C expression that is nonzero if on this machine the number of bits
9931 actually used for the count of a shift operation is equal to the number
9932 of bits needed to represent the size of the object being shifted. When
9933 this macro is nonzero, the compiler will assume that it is safe to omit
9934 a sign-extend, zero-extend, and certain bitwise `and' instructions that
9935 truncates the count of a shift operation. On machines that have
9936 instructions that act on bit-fields at variable positions, which may
9937 include `bit test' instructions, a nonzero @code{SHIFT_COUNT_TRUNCATED}
9938 also enables deletion of truncations of the values that serve as
9939 arguments to bit-field instructions.
9941 If both types of instructions truncate the count (for shifts) and
9942 position (for bit-field operations), or if no variable-position bit-field
9943 instructions exist, you should define this macro.
9945 However, on some machines, such as the 80386 and the 680x0, truncation
9946 only applies to shift operations and not the (real or pretended)
9947 bit-field operations. Define @code{SHIFT_COUNT_TRUNCATED} to be zero on
9948 such machines. Instead, add patterns to the @file{md} file that include
9949 the implied truncation of the shift instructions.
9951 You need not define this macro if it would always have the value of zero.
9954 @anchor{TARGET_SHIFT_TRUNCATION_MASK}
9955 @deftypefn {Target Hook} int TARGET_SHIFT_TRUNCATION_MASK (enum machine_mode @var{mode})
9956 This function describes how the standard shift patterns for @var{mode}
9957 deal with shifts by negative amounts or by more than the width of the mode.
9958 @xref{shift patterns}.
9960 On many machines, the shift patterns will apply a mask @var{m} to the
9961 shift count, meaning that a fixed-width shift of @var{x} by @var{y} is
9962 equivalent to an arbitrary-width shift of @var{x} by @var{y & m}. If
9963 this is true for mode @var{mode}, the function should return @var{m},
9964 otherwise it should return 0. A return value of 0 indicates that no
9965 particular behavior is guaranteed.
9967 Note that, unlike @code{SHIFT_COUNT_TRUNCATED}, this function does
9968 @emph{not} apply to general shift rtxes; it applies only to instructions
9969 that are generated by the named shift patterns.
9971 The default implementation of this function returns
9972 @code{GET_MODE_BITSIZE (@var{mode}) - 1} if @code{SHIFT_COUNT_TRUNCATED}
9973 and 0 otherwise. This definition is always safe, but if
9974 @code{SHIFT_COUNT_TRUNCATED} is false, and some shift patterns
9975 nevertheless truncate the shift count, you may get better code
9979 @defmac TRULY_NOOP_TRUNCATION (@var{outprec}, @var{inprec})
9980 A C expression which is nonzero if on this machine it is safe to
9981 ``convert'' an integer of @var{inprec} bits to one of @var{outprec}
9982 bits (where @var{outprec} is smaller than @var{inprec}) by merely
9983 operating on it as if it had only @var{outprec} bits.
9985 On many machines, this expression can be 1.
9987 @c rearranged this, removed the phrase "it is reported that". this was
9988 @c to fix an overfull hbox. --mew 10feb93
9989 When @code{TRULY_NOOP_TRUNCATION} returns 1 for a pair of sizes for
9990 modes for which @code{MODES_TIEABLE_P} is 0, suboptimal code can result.
9991 If this is the case, making @code{TRULY_NOOP_TRUNCATION} return 0 in
9992 such cases may improve things.
9995 @deftypefn {Target Hook} int TARGET_MODE_REP_EXTENDED (enum machine_mode @var{mode}, enum machine_mode @var{rep_mode})
9996 The representation of an integral mode can be such that the values
9997 are always extended to a wider integral mode. Return
9998 @code{SIGN_EXTEND} if values of @var{mode} are represented in
9999 sign-extended form to @var{rep_mode}. Return @code{UNKNOWN}
10000 otherwise. (Currently, none of the targets use zero-extended
10001 representation this way so unlike @code{LOAD_EXTEND_OP},
10002 @code{TARGET_MODE_REP_EXTENDED} is expected to return either
10003 @code{SIGN_EXTEND} or @code{UNKNOWN}. Also no target extends
10004 @var{mode} to @var{mode_rep} so that @var{mode_rep} is not the next
10005 widest integral mode and currently we take advantage of this fact.)
10007 Similarly to @code{LOAD_EXTEND_OP} you may return a non-@code{UNKNOWN}
10008 value even if the extension is not performed on certain hard registers
10009 as long as for the @code{REGNO_REG_CLASS} of these hard registers
10010 @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero.
10012 Note that @code{TARGET_MODE_REP_EXTENDED} and @code{LOAD_EXTEND_OP}
10013 describe two related properties. If you define
10014 @code{TARGET_MODE_REP_EXTENDED (mode, word_mode)} you probably also want
10015 to define @code{LOAD_EXTEND_OP (mode)} to return the same type of
10018 In order to enforce the representation of @code{mode},
10019 @code{TRULY_NOOP_TRUNCATION} should return false when truncating to
10023 @defmac STORE_FLAG_VALUE
10024 A C expression describing the value returned by a comparison operator
10025 with an integral mode and stored by a store-flag instruction
10026 (@samp{s@var{cond}}) when the condition is true. This description must
10027 apply to @emph{all} the @samp{s@var{cond}} patterns and all the
10028 comparison operators whose results have a @code{MODE_INT} mode.
10030 A value of 1 or @minus{}1 means that the instruction implementing the
10031 comparison operator returns exactly 1 or @minus{}1 when the comparison is true
10032 and 0 when the comparison is false. Otherwise, the value indicates
10033 which bits of the result are guaranteed to be 1 when the comparison is
10034 true. This value is interpreted in the mode of the comparison
10035 operation, which is given by the mode of the first operand in the
10036 @samp{s@var{cond}} pattern. Either the low bit or the sign bit of
10037 @code{STORE_FLAG_VALUE} be on. Presently, only those bits are used by
10040 If @code{STORE_FLAG_VALUE} is neither 1 or @minus{}1, the compiler will
10041 generate code that depends only on the specified bits. It can also
10042 replace comparison operators with equivalent operations if they cause
10043 the required bits to be set, even if the remaining bits are undefined.
10044 For example, on a machine whose comparison operators return an
10045 @code{SImode} value and where @code{STORE_FLAG_VALUE} is defined as
10046 @samp{0x80000000}, saying that just the sign bit is relevant, the
10050 (ne:SI (and:SI @var{x} (const_int @var{power-of-2})) (const_int 0))
10054 can be converted to
10057 (ashift:SI @var{x} (const_int @var{n}))
10061 where @var{n} is the appropriate shift count to move the bit being
10062 tested into the sign bit.
10064 There is no way to describe a machine that always sets the low-order bit
10065 for a true value, but does not guarantee the value of any other bits,
10066 but we do not know of any machine that has such an instruction. If you
10067 are trying to port GCC to such a machine, include an instruction to
10068 perform a logical-and of the result with 1 in the pattern for the
10069 comparison operators and let us know at @email{gcc@@gcc.gnu.org}.
10071 Often, a machine will have multiple instructions that obtain a value
10072 from a comparison (or the condition codes). Here are rules to guide the
10073 choice of value for @code{STORE_FLAG_VALUE}, and hence the instructions
10078 Use the shortest sequence that yields a valid definition for
10079 @code{STORE_FLAG_VALUE}. It is more efficient for the compiler to
10080 ``normalize'' the value (convert it to, e.g., 1 or 0) than for the
10081 comparison operators to do so because there may be opportunities to
10082 combine the normalization with other operations.
10085 For equal-length sequences, use a value of 1 or @minus{}1, with @minus{}1 being
10086 slightly preferred on machines with expensive jumps and 1 preferred on
10090 As a second choice, choose a value of @samp{0x80000001} if instructions
10091 exist that set both the sign and low-order bits but do not define the
10095 Otherwise, use a value of @samp{0x80000000}.
10098 Many machines can produce both the value chosen for
10099 @code{STORE_FLAG_VALUE} and its negation in the same number of
10100 instructions. On those machines, you should also define a pattern for
10101 those cases, e.g., one matching
10104 (set @var{A} (neg:@var{m} (ne:@var{m} @var{B} @var{C})))
10107 Some machines can also perform @code{and} or @code{plus} operations on
10108 condition code values with less instructions than the corresponding
10109 @samp{s@var{cond}} insn followed by @code{and} or @code{plus}. On those
10110 machines, define the appropriate patterns. Use the names @code{incscc}
10111 and @code{decscc}, respectively, for the patterns which perform
10112 @code{plus} or @code{minus} operations on condition code values. See
10113 @file{rs6000.md} for some examples. The GNU Superoptizer can be used to
10114 find such instruction sequences on other machines.
10116 If this macro is not defined, the default value, 1, is used. You need
10117 not define @code{STORE_FLAG_VALUE} if the machine has no store-flag
10118 instructions, or if the value generated by these instructions is 1.
10121 @defmac FLOAT_STORE_FLAG_VALUE (@var{mode})
10122 A C expression that gives a nonzero @code{REAL_VALUE_TYPE} value that is
10123 returned when comparison operators with floating-point results are true.
10124 Define this macro on machines that have comparison operations that return
10125 floating-point values. If there are no such operations, do not define
10129 @defmac VECTOR_STORE_FLAG_VALUE (@var{mode})
10130 A C expression that gives a rtx representing the nonzero true element
10131 for vector comparisons. The returned rtx should be valid for the inner
10132 mode of @var{mode} which is guaranteed to be a vector mode. Define
10133 this macro on machines that have vector comparison operations that
10134 return a vector result. If there are no such operations, do not define
10135 this macro. Typically, this macro is defined as @code{const1_rtx} or
10136 @code{constm1_rtx}. This macro may return @code{NULL_RTX} to prevent
10137 the compiler optimizing such vector comparison operations for the
10141 @defmac CLZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
10142 @defmacx CTZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
10143 A C expression that indicates whether the architecture defines a value
10144 for @code{clz} or @code{ctz} with a zero operand.
10145 A result of @code{0} indicates the value is undefined.
10146 If the value is defined for only the RTL expression, the macro should
10147 evaluate to @code{1}; if the value applies also to the corresponding optab
10148 entry (which is normally the case if it expands directly into
10149 the corresponding RTL), then the macro should evaluate to @code{2}.
10150 In the cases where the value is defined, @var{value} should be set to
10153 If this macro is not defined, the value of @code{clz} or
10154 @code{ctz} at zero is assumed to be undefined.
10156 This macro must be defined if the target's expansion for @code{ffs}
10157 relies on a particular value to get correct results. Otherwise it
10158 is not necessary, though it may be used to optimize some corner cases, and
10159 to provide a default expansion for the @code{ffs} optab.
10161 Note that regardless of this macro the ``definedness'' of @code{clz}
10162 and @code{ctz} at zero do @emph{not} extend to the builtin functions
10163 visible to the user. Thus one may be free to adjust the value at will
10164 to match the target expansion of these operations without fear of
10169 An alias for the machine mode for pointers. On most machines, define
10170 this to be the integer mode corresponding to the width of a hardware
10171 pointer; @code{SImode} on 32-bit machine or @code{DImode} on 64-bit machines.
10172 On some machines you must define this to be one of the partial integer
10173 modes, such as @code{PSImode}.
10175 The width of @code{Pmode} must be at least as large as the value of
10176 @code{POINTER_SIZE}. If it is not equal, you must define the macro
10177 @code{POINTERS_EXTEND_UNSIGNED} to specify how pointers are extended
10181 @defmac FUNCTION_MODE
10182 An alias for the machine mode used for memory references to functions
10183 being called, in @code{call} RTL expressions. On most CISC machines,
10184 where an instruction can begin at any byte address, this should be
10185 @code{QImode}. On most RISC machines, where all instructions have fixed
10186 size and alignment, this should be a mode with the same size and alignment
10187 as the machine instruction words - typically @code{SImode} or @code{HImode}.
10190 @defmac STDC_0_IN_SYSTEM_HEADERS
10191 In normal operation, the preprocessor expands @code{__STDC__} to the
10192 constant 1, to signify that GCC conforms to ISO Standard C@. On some
10193 hosts, like Solaris, the system compiler uses a different convention,
10194 where @code{__STDC__} is normally 0, but is 1 if the user specifies
10195 strict conformance to the C Standard.
10197 Defining @code{STDC_0_IN_SYSTEM_HEADERS} makes GNU CPP follows the host
10198 convention when processing system header files, but when processing user
10199 files @code{__STDC__} will always expand to 1.
10202 @defmac NO_IMPLICIT_EXTERN_C
10203 Define this macro if the system header files support C++ as well as C@.
10204 This macro inhibits the usual method of using system header files in
10205 C++, which is to pretend that the file's contents are enclosed in
10206 @samp{extern "C" @{@dots{}@}}.
10211 @defmac REGISTER_TARGET_PRAGMAS ()
10212 Define this macro if you want to implement any target-specific pragmas.
10213 If defined, it is a C expression which makes a series of calls to
10214 @code{c_register_pragma} or @code{c_register_pragma_with_expansion}
10215 for each pragma. The macro may also do any
10216 setup required for the pragmas.
10218 The primary reason to define this macro is to provide compatibility with
10219 other compilers for the same target. In general, we discourage
10220 definition of target-specific pragmas for GCC@.
10222 If the pragma can be implemented by attributes then you should consider
10223 defining the target hook @samp{TARGET_INSERT_ATTRIBUTES} as well.
10225 Preprocessor macros that appear on pragma lines are not expanded. All
10226 @samp{#pragma} directives that do not match any registered pragma are
10227 silently ignored, unless the user specifies @option{-Wunknown-pragmas}.
10230 @deftypefun void c_register_pragma (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
10231 @deftypefunx void c_register_pragma_with_expansion (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
10233 Each call to @code{c_register_pragma} or
10234 @code{c_register_pragma_with_expansion} establishes one pragma. The
10235 @var{callback} routine will be called when the preprocessor encounters a
10239 #pragma [@var{space}] @var{name} @dots{}
10242 @var{space} is the case-sensitive namespace of the pragma, or
10243 @code{NULL} to put the pragma in the global namespace. The callback
10244 routine receives @var{pfile} as its first argument, which can be passed
10245 on to cpplib's functions if necessary. You can lex tokens after the
10246 @var{name} by calling @code{pragma_lex}. Tokens that are not read by the
10247 callback will be silently ignored. The end of the line is indicated by
10248 a token of type @code{CPP_EOF}. Macro expansion occurs on the
10249 arguments of pragmas registered with
10250 @code{c_register_pragma_with_expansion} but not on the arguments of
10251 pragmas registered with @code{c_register_pragma}.
10253 Note that the use of @code{pragma_lex} is specific to the C and C++
10254 compilers. It will not work in the Java or Fortran compilers, or any
10255 other language compilers for that matter. Thus if @code{pragma_lex} is going
10256 to be called from target-specific code, it must only be done so when
10257 building the C and C++ compilers. This can be done by defining the
10258 variables @code{c_target_objs} and @code{cxx_target_objs} in the
10259 target entry in the @file{config.gcc} file. These variables should name
10260 the target-specific, language-specific object file which contains the
10261 code that uses @code{pragma_lex}. Note it will also be necessary to add a
10262 rule to the makefile fragment pointed to by @code{tmake_file} that shows
10263 how to build this object file.
10268 @defmac HANDLE_SYSV_PRAGMA
10269 Define this macro (to a value of 1) if you want the System V style
10270 pragmas @samp{#pragma pack(<n>)} and @samp{#pragma weak <name>
10271 [=<value>]} to be supported by gcc.
10273 The pack pragma specifies the maximum alignment (in bytes) of fields
10274 within a structure, in much the same way as the @samp{__aligned__} and
10275 @samp{__packed__} @code{__attribute__}s do. A pack value of zero resets
10276 the behavior to the default.
10278 A subtlety for Microsoft Visual C/C++ style bit-field packing
10279 (e.g.@: -mms-bitfields) for targets that support it:
10280 When a bit-field is inserted into a packed record, the whole size
10281 of the underlying type is used by one or more same-size adjacent
10282 bit-fields (that is, if its long:3, 32 bits is used in the record,
10283 and any additional adjacent long bit-fields are packed into the same
10284 chunk of 32 bits. However, if the size changes, a new field of that
10285 size is allocated).
10287 If both MS bit-fields and @samp{__attribute__((packed))} are used,
10288 the latter will take precedence. If @samp{__attribute__((packed))} is
10289 used on a single field when MS bit-fields are in use, it will take
10290 precedence for that field, but the alignment of the rest of the structure
10291 may affect its placement.
10293 The weak pragma only works if @code{SUPPORTS_WEAK} and
10294 @code{ASM_WEAKEN_LABEL} are defined. If enabled it allows the creation
10295 of specifically named weak labels, optionally with a value.
10300 @defmac HANDLE_PRAGMA_PACK_PUSH_POP
10301 Define this macro (to a value of 1) if you want to support the Win32
10302 style pragmas @samp{#pragma pack(push[,@var{n}])} and @samp{#pragma
10303 pack(pop)}. The @samp{pack(push,[@var{n}])} pragma specifies the maximum
10304 alignment (in bytes) of fields within a structure, in much the same way as
10305 the @samp{__aligned__} and @samp{__packed__} @code{__attribute__}s do. A
10306 pack value of zero resets the behavior to the default. Successive
10307 invocations of this pragma cause the previous values to be stacked, so
10308 that invocations of @samp{#pragma pack(pop)} will return to the previous
10312 @defmac HANDLE_PRAGMA_PACK_WITH_EXPANSION
10313 Define this macro, as well as
10314 @code{HANDLE_SYSV_PRAGMA}, if macros should be expanded in the
10315 arguments of @samp{#pragma pack}.
10318 @defmac TARGET_DEFAULT_PACK_STRUCT
10319 If your target requires a structure packing default other than 0 (meaning
10320 the machine default), define this macro to the necessary value (in bytes).
10321 This must be a value that would also be valid to use with
10322 @samp{#pragma pack()} (that is, a small power of two).
10327 @defmac HANDLE_PRAGMA_PUSH_POP_MACRO
10328 Define this macro if you want to support the Win32 style pragmas
10329 @samp{#pragma push_macro(macro-name-as-string)} and @samp{#pragma
10330 pop_macro(macro-name-as-string)}. The @samp{#pragma push_macro(
10331 macro-name-as-string)} pragma saves the named macro and via
10332 @samp{#pragma pop_macro(macro-name-as-string)} it will return to the
10337 @defmac DOLLARS_IN_IDENTIFIERS
10338 Define this macro to control use of the character @samp{$} in
10339 identifier names for the C family of languages. 0 means @samp{$} is
10340 not allowed by default; 1 means it is allowed. 1 is the default;
10341 there is no need to define this macro in that case.
10344 @defmac NO_DOLLAR_IN_LABEL
10345 Define this macro if the assembler does not accept the character
10346 @samp{$} in label names. By default constructors and destructors in
10347 G++ have @samp{$} in the identifiers. If this macro is defined,
10348 @samp{.} is used instead.
10351 @defmac NO_DOT_IN_LABEL
10352 Define this macro if the assembler does not accept the character
10353 @samp{.} in label names. By default constructors and destructors in G++
10354 have names that use @samp{.}. If this macro is defined, these names
10355 are rewritten to avoid @samp{.}.
10358 @defmac INSN_SETS_ARE_DELAYED (@var{insn})
10359 Define this macro as a C expression that is nonzero if it is safe for the
10360 delay slot scheduler to place instructions in the delay slot of @var{insn},
10361 even if they appear to use a resource set or clobbered in @var{insn}.
10362 @var{insn} is always a @code{jump_insn} or an @code{insn}; GCC knows that
10363 every @code{call_insn} has this behavior. On machines where some @code{insn}
10364 or @code{jump_insn} is really a function call and hence has this behavior,
10365 you should define this macro.
10367 You need not define this macro if it would always return zero.
10370 @defmac INSN_REFERENCES_ARE_DELAYED (@var{insn})
10371 Define this macro as a C expression that is nonzero if it is safe for the
10372 delay slot scheduler to place instructions in the delay slot of @var{insn},
10373 even if they appear to set or clobber a resource referenced in @var{insn}.
10374 @var{insn} is always a @code{jump_insn} or an @code{insn}. On machines where
10375 some @code{insn} or @code{jump_insn} is really a function call and its operands
10376 are registers whose use is actually in the subroutine it calls, you should
10377 define this macro. Doing so allows the delay slot scheduler to move
10378 instructions which copy arguments into the argument registers into the delay
10379 slot of @var{insn}.
10381 You need not define this macro if it would always return zero.
10384 @defmac MULTIPLE_SYMBOL_SPACES
10385 Define this macro as a C expression that is nonzero if, in some cases,
10386 global symbols from one translation unit may not be bound to undefined
10387 symbols in another translation unit without user intervention. For
10388 instance, under Microsoft Windows symbols must be explicitly imported
10389 from shared libraries (DLLs).
10391 You need not define this macro if it would always evaluate to zero.
10394 @deftypefn {Target Hook} tree TARGET_MD_ASM_CLOBBERS (tree @var{outputs}, tree @var{inputs}, tree @var{clobbers})
10395 This target hook should add to @var{clobbers} @code{STRING_CST} trees for
10396 any hard regs the port wishes to automatically clobber for an asm.
10397 It should return the result of the last @code{tree_cons} used to add a
10398 clobber. The @var{outputs}, @var{inputs} and @var{clobber} lists are the
10399 corresponding parameters to the asm and may be inspected to avoid
10400 clobbering a register that is an input or output of the asm. You can use
10401 @code{tree_overlaps_hard_reg_set}, declared in @file{tree.h}, to test
10402 for overlap with regards to asm-declared registers.
10405 @defmac MATH_LIBRARY
10406 Define this macro as a C string constant for the linker argument to link
10407 in the system math library, or @samp{""} if the target does not have a
10408 separate math library.
10410 You need only define this macro if the default of @samp{"-lm"} is wrong.
10413 @defmac LIBRARY_PATH_ENV
10414 Define this macro as a C string constant for the environment variable that
10415 specifies where the linker should look for libraries.
10417 You need only define this macro if the default of @samp{"LIBRARY_PATH"}
10421 @defmac TARGET_POSIX_IO
10422 Define this macro if the target supports the following POSIX@ file
10423 functions, access, mkdir and file locking with fcntl / F_SETLKW@.
10424 Defining @code{TARGET_POSIX_IO} will enable the test coverage code
10425 to use file locking when exiting a program, which avoids race conditions
10426 if the program has forked. It will also create directories at run-time
10427 for cross-profiling.
10430 @defmac MAX_CONDITIONAL_EXECUTE
10432 A C expression for the maximum number of instructions to execute via
10433 conditional execution instructions instead of a branch. A value of
10434 @code{BRANCH_COST}+1 is the default if the machine does not use cc0, and
10435 1 if it does use cc0.
10438 @defmac IFCVT_MODIFY_TESTS (@var{ce_info}, @var{true_expr}, @var{false_expr})
10439 Used if the target needs to perform machine-dependent modifications on the
10440 conditionals used for turning basic blocks into conditionally executed code.
10441 @var{ce_info} points to a data structure, @code{struct ce_if_block}, which
10442 contains information about the currently processed blocks. @var{true_expr}
10443 and @var{false_expr} are the tests that are used for converting the
10444 then-block and the else-block, respectively. Set either @var{true_expr} or
10445 @var{false_expr} to a null pointer if the tests cannot be converted.
10448 @defmac IFCVT_MODIFY_MULTIPLE_TESTS (@var{ce_info}, @var{bb}, @var{true_expr}, @var{false_expr})
10449 Like @code{IFCVT_MODIFY_TESTS}, but used when converting more complicated
10450 if-statements into conditions combined by @code{and} and @code{or} operations.
10451 @var{bb} contains the basic block that contains the test that is currently
10452 being processed and about to be turned into a condition.
10455 @defmac IFCVT_MODIFY_INSN (@var{ce_info}, @var{pattern}, @var{insn})
10456 A C expression to modify the @var{PATTERN} of an @var{INSN} that is to
10457 be converted to conditional execution format. @var{ce_info} points to
10458 a data structure, @code{struct ce_if_block}, which contains information
10459 about the currently processed blocks.
10462 @defmac IFCVT_MODIFY_FINAL (@var{ce_info})
10463 A C expression to perform any final machine dependent modifications in
10464 converting code to conditional execution. The involved basic blocks
10465 can be found in the @code{struct ce_if_block} structure that is pointed
10466 to by @var{ce_info}.
10469 @defmac IFCVT_MODIFY_CANCEL (@var{ce_info})
10470 A C expression to cancel any machine dependent modifications in
10471 converting code to conditional execution. The involved basic blocks
10472 can be found in the @code{struct ce_if_block} structure that is pointed
10473 to by @var{ce_info}.
10476 @defmac IFCVT_INIT_EXTRA_FIELDS (@var{ce_info})
10477 A C expression to initialize any extra fields in a @code{struct ce_if_block}
10478 structure, which are defined by the @code{IFCVT_EXTRA_FIELDS} macro.
10481 @defmac IFCVT_EXTRA_FIELDS
10482 If defined, it should expand to a set of field declarations that will be
10483 added to the @code{struct ce_if_block} structure. These should be initialized
10484 by the @code{IFCVT_INIT_EXTRA_FIELDS} macro.
10487 @deftypefn {Target Hook} void TARGET_MACHINE_DEPENDENT_REORG ()
10488 If non-null, this hook performs a target-specific pass over the
10489 instruction stream. The compiler will run it at all optimization levels,
10490 just before the point at which it normally does delayed-branch scheduling.
10492 The exact purpose of the hook varies from target to target. Some use
10493 it to do transformations that are necessary for correctness, such as
10494 laying out in-function constant pools or avoiding hardware hazards.
10495 Others use it as an opportunity to do some machine-dependent optimizations.
10497 You need not implement the hook if it has nothing to do. The default
10498 definition is null.
10501 @deftypefn {Target Hook} void TARGET_INIT_BUILTINS ()
10502 Define this hook if you have any machine-specific built-in functions
10503 that need to be defined. It should be a function that performs the
10506 Machine specific built-in functions can be useful to expand special machine
10507 instructions that would otherwise not normally be generated because
10508 they have no equivalent in the source language (for example, SIMD vector
10509 instructions or prefetch instructions).
10511 To create a built-in function, call the function
10512 @code{lang_hooks.builtin_function}
10513 which is defined by the language front end. You can use any type nodes set
10514 up by @code{build_common_tree_nodes} and @code{build_common_tree_nodes_2};
10515 only language front ends that use those two functions will call
10516 @samp{TARGET_INIT_BUILTINS}.
10519 @deftypefn {Target Hook} rtx TARGET_EXPAND_BUILTIN (tree @var{exp}, rtx @var{target}, rtx @var{subtarget}, enum machine_mode @var{mode}, int @var{ignore})
10521 Expand a call to a machine specific built-in function that was set up by
10522 @samp{TARGET_INIT_BUILTINS}. @var{exp} is the expression for the
10523 function call; the result should go to @var{target} if that is
10524 convenient, and have mode @var{mode} if that is convenient.
10525 @var{subtarget} may be used as the target for computing one of
10526 @var{exp}'s operands. @var{ignore} is nonzero if the value is to be
10527 ignored. This function should return the result of the call to the
10531 @deftypefn {Target Hook} tree TARGET_RESOLVE_OVERLOADED_BUILTIN (tree @var{fndecl}, tree @var{arglist})
10533 Select a replacement for a machine specific built-in function that
10534 was set up by @samp{TARGET_INIT_BUILTINS}. This is done
10535 @emph{before} regular type checking, and so allows the target to
10536 implement a crude form of function overloading. @var{fndecl} is the
10537 declaration of the built-in function. @var{arglist} is the list of
10538 arguments passed to the built-in function. The result is a
10539 complete expression that implements the operation, usually
10540 another @code{CALL_EXPR}.
10543 @deftypefn {Target Hook} tree TARGET_FOLD_BUILTIN (tree @var{fndecl}, tree @var{arglist}, bool @var{ignore})
10545 Fold a call to a machine specific built-in function that was set up by
10546 @samp{TARGET_INIT_BUILTINS}. @var{fndecl} is the declaration of the
10547 built-in function. @var{arglist} is the list of arguments passed to
10548 the built-in function. The result is another tree containing a
10549 simplified expression for the call's result. If @var{ignore} is true
10550 the value will be ignored.
10553 @deftypefn {Target Hook} const char * TARGET_INVALID_WITHIN_DOLOOP (rtx @var{insn})
10555 Take an instruction in @var{insn} and return NULL if it is valid within a
10556 low-overhead loop, otherwise return a string why doloop could not be applied.
10558 Many targets use special registers for low-overhead looping. For any
10559 instruction that clobbers these this function should return a string indicating
10560 the reason why the doloop could not be applied.
10561 By default, the RTL loop optimizer does not use a present doloop pattern for
10562 loops containing function calls or branch on table instructions.
10565 @defmac MD_CAN_REDIRECT_BRANCH (@var{branch1}, @var{branch2})
10567 Take a branch insn in @var{branch1} and another in @var{branch2}.
10568 Return true if redirecting @var{branch1} to the destination of
10569 @var{branch2} is possible.
10571 On some targets, branches may have a limited range. Optimizing the
10572 filling of delay slots can result in branches being redirected, and this
10573 may in turn cause a branch offset to overflow.
10576 @deftypefn {Target Hook} bool TARGET_COMMUTATIVE_P (rtx @var{x}, @var{outer_code})
10577 This target hook returns @code{true} if @var{x} is considered to be commutative.
10578 Usually, this is just COMMUTATIVE_P (@var{x}), but the HP PA doesn't consider
10579 PLUS to be commutative inside a MEM@. @var{outer_code} is the rtx code
10580 of the enclosing rtl, if known, otherwise it is UNKNOWN.
10583 @deftypefn {Target Hook} rtx TARGET_ALLOCATE_INITIAL_VALUE (rtx @var{hard_reg})
10585 When the initial value of a hard register has been copied in a pseudo
10586 register, it is often not necessary to actually allocate another register
10587 to this pseudo register, because the original hard register or a stack slot
10588 it has been saved into can be used. @code{TARGET_ALLOCATE_INITIAL_VALUE}
10589 is called at the start of register allocation once for each hard register
10590 that had its initial value copied by using
10591 @code{get_func_hard_reg_initial_val} or @code{get_hard_reg_initial_val}.
10592 Possible values are @code{NULL_RTX}, if you don't want
10593 to do any special allocation, a @code{REG} rtx---that would typically be
10594 the hard register itself, if it is known not to be clobbered---or a
10596 If you are returning a @code{MEM}, this is only a hint for the allocator;
10597 it might decide to use another register anyways.
10598 You may use @code{current_function_leaf_function} in the hook, functions
10599 that use @code{REG_N_SETS}, to determine if the hard
10600 register in question will not be clobbered.
10601 The default value of this hook is @code{NULL}, which disables any special
10605 @deftypefn {Target Hook} int TARGET_UNSPEC_MAY_TRAP_P (const_rtx @var{x}, unsigned @var{flags})
10606 This target hook returns nonzero if @var{x}, an @code{unspec} or
10607 @code{unspec_volatile} operation, might cause a trap. Targets can use
10608 this hook to enhance precision of analysis for @code{unspec} and
10609 @code{unspec_volatile} operations. You may call @code{may_trap_p_1}
10610 to analyze inner elements of @var{x} in which case @var{flags} should be
10614 @deftypefn {Target Hook} void TARGET_SET_CURRENT_FUNCTION (tree @var{decl})
10615 The compiler invokes this hook whenever it changes its current function
10616 context (@code{cfun}). You can define this function if
10617 the back end needs to perform any initialization or reset actions on a
10618 per-function basis. For example, it may be used to implement function
10619 attributes that affect register usage or code generation patterns.
10620 The argument @var{decl} is the declaration for the new function context,
10621 and may be null to indicate that the compiler has left a function context
10622 and is returning to processing at the top level.
10623 The default hook function does nothing.
10625 GCC sets @code{cfun} to a dummy function context during initialization of
10626 some parts of the back end. The hook function is not invoked in this
10627 situation; you need not worry about the hook being invoked recursively,
10628 or when the back end is in a partially-initialized state.
10631 @defmac TARGET_OBJECT_SUFFIX
10632 Define this macro to be a C string representing the suffix for object
10633 files on your target machine. If you do not define this macro, GCC will
10634 use @samp{.o} as the suffix for object files.
10637 @defmac TARGET_EXECUTABLE_SUFFIX
10638 Define this macro to be a C string representing the suffix to be
10639 automatically added to executable files on your target machine. If you
10640 do not define this macro, GCC will use the null string as the suffix for
10644 @defmac COLLECT_EXPORT_LIST
10645 If defined, @code{collect2} will scan the individual object files
10646 specified on its command line and create an export list for the linker.
10647 Define this macro for systems like AIX, where the linker discards
10648 object files that are not referenced from @code{main} and uses export
10652 @defmac MODIFY_JNI_METHOD_CALL (@var{mdecl})
10653 Define this macro to a C expression representing a variant of the
10654 method call @var{mdecl}, if Java Native Interface (JNI) methods
10655 must be invoked differently from other methods on your target.
10656 For example, on 32-bit Microsoft Windows, JNI methods must be invoked using
10657 the @code{stdcall} calling convention and this macro is then
10658 defined as this expression:
10661 build_type_attribute_variant (@var{mdecl},
10663 (get_identifier ("stdcall"),
10668 @deftypefn {Target Hook} bool TARGET_CANNOT_MODIFY_JUMPS_P (void)
10669 This target hook returns @code{true} past the point in which new jump
10670 instructions could be created. On machines that require a register for
10671 every jump such as the SHmedia ISA of SH5, this point would typically be
10672 reload, so this target hook should be defined to a function such as:
10676 cannot_modify_jumps_past_reload_p ()
10678 return (reload_completed || reload_in_progress);
10683 @deftypefn {Target Hook} int TARGET_BRANCH_TARGET_REGISTER_CLASS (void)
10684 This target hook returns a register class for which branch target register
10685 optimizations should be applied. All registers in this class should be
10686 usable interchangeably. After reload, registers in this class will be
10687 re-allocated and loads will be hoisted out of loops and be subjected
10688 to inter-block scheduling.
10691 @deftypefn {Target Hook} bool TARGET_BRANCH_TARGET_REGISTER_CALLEE_SAVED (bool @var{after_prologue_epilogue_gen})
10692 Branch target register optimization will by default exclude callee-saved
10694 that are not already live during the current function; if this target hook
10695 returns true, they will be included. The target code must than make sure
10696 that all target registers in the class returned by
10697 @samp{TARGET_BRANCH_TARGET_REGISTER_CLASS} that might need saving are
10698 saved. @var{after_prologue_epilogue_gen} indicates if prologues and
10699 epilogues have already been generated. Note, even if you only return
10700 true when @var{after_prologue_epilogue_gen} is false, you still are likely
10701 to have to make special provisions in @code{INITIAL_ELIMINATION_OFFSET}
10702 to reserve space for caller-saved target registers.
10705 @defmac POWI_MAX_MULTS
10706 If defined, this macro is interpreted as a signed integer C expression
10707 that specifies the maximum number of floating point multiplications
10708 that should be emitted when expanding exponentiation by an integer
10709 constant inline. When this value is defined, exponentiation requiring
10710 more than this number of multiplications is implemented by calling the
10711 system library's @code{pow}, @code{powf} or @code{powl} routines.
10712 The default value places no upper bound on the multiplication count.
10715 @deftypefn Macro void TARGET_EXTRA_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
10716 This target hook should register any extra include files for the
10717 target. The parameter @var{stdinc} indicates if normal include files
10718 are present. The parameter @var{sysroot} is the system root directory.
10719 The parameter @var{iprefix} is the prefix for the gcc directory.
10722 @deftypefn Macro void TARGET_EXTRA_PRE_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
10723 This target hook should register any extra include files for the
10724 target before any standard headers. The parameter @var{stdinc}
10725 indicates if normal include files are present. The parameter
10726 @var{sysroot} is the system root directory. The parameter
10727 @var{iprefix} is the prefix for the gcc directory.
10730 @deftypefn Macro void TARGET_OPTF (char *@var{path})
10731 This target hook should register special include paths for the target.
10732 The parameter @var{path} is the include to register. On Darwin
10733 systems, this is used for Framework includes, which have semantics
10734 that are different from @option{-I}.
10737 @deftypefn {Target Hook} bool TARGET_USE_LOCAL_THUNK_ALIAS_P (tree @var{fndecl})
10738 This target hook returns @code{true} if it is safe to use a local alias
10739 for a virtual function @var{fndecl} when constructing thunks,
10740 @code{false} otherwise. By default, the hook returns @code{true} for all
10741 functions, if a target supports aliases (i.e.@: defines
10742 @code{ASM_OUTPUT_DEF}), @code{false} otherwise,
10745 @defmac TARGET_FORMAT_TYPES
10746 If defined, this macro is the name of a global variable containing
10747 target-specific format checking information for the @option{-Wformat}
10748 option. The default is to have no target-specific format checks.
10751 @defmac TARGET_N_FORMAT_TYPES
10752 If defined, this macro is the number of entries in
10753 @code{TARGET_FORMAT_TYPES}.
10756 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES
10757 If defined, this macro is the name of a global variable containing
10758 target-specific format overrides for the @option{-Wformat} option. The
10759 default is to have no target-specific format overrides. If defined,
10760 @code{TARGET_FORMAT_TYPES} must be defined, too.
10763 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES_COUNT
10764 If defined, this macro specifies the number of entries in
10765 @code{TARGET_OVERRIDES_FORMAT_ATTRIBUTES}.
10768 @defmac TARGET_OVERRIDES_FORMAT_INIT
10769 If defined, this macro specifies the optional initialization
10770 routine for target specific customizations of the system printf
10771 and scanf formatter settings.
10774 @deftypefn {Target Hook} bool TARGET_RELAXED_ORDERING
10775 If set to @code{true}, means that the target's memory model does not
10776 guarantee that loads which do not depend on one another will access
10777 main memory in the order of the instruction stream; if ordering is
10778 important, an explicit memory barrier must be used. This is true of
10779 many recent processors which implement a policy of ``relaxed,''
10780 ``weak,'' or ``release'' memory consistency, such as Alpha, PowerPC,
10781 and ia64. The default is @code{false}.
10784 @deftypefn {Target Hook} const char *TARGET_INVALID_ARG_FOR_UNPROTOTYPED_FN (tree @var{typelist}, tree @var{funcdecl}, tree @var{val})
10785 If defined, this macro returns the diagnostic message when it is
10786 illegal to pass argument @var{val} to function @var{funcdecl}
10787 with prototype @var{typelist}.
10790 @deftypefn {Target Hook} {const char *} TARGET_INVALID_CONVERSION (tree @var{fromtype}, tree @var{totype})
10791 If defined, this macro returns the diagnostic message when it is
10792 invalid to convert from @var{fromtype} to @var{totype}, or @code{NULL}
10793 if validity should be determined by the front end.
10796 @deftypefn {Target Hook} {const char *} TARGET_INVALID_UNARY_OP (int @var{op}, tree @var{type})
10797 If defined, this macro returns the diagnostic message when it is
10798 invalid to apply operation @var{op} (where unary plus is denoted by
10799 @code{CONVERT_EXPR}) to an operand of type @var{type}, or @code{NULL}
10800 if validity should be determined by the front end.
10803 @deftypefn {Target Hook} {const char *} TARGET_INVALID_BINARY_OP (int @var{op}, tree @var{type1}, tree @var{type2})
10804 If defined, this macro returns the diagnostic message when it is
10805 invalid to apply operation @var{op} to operands of types @var{type1}
10806 and @var{type2}, or @code{NULL} if validity should be determined by
10810 @deftypefn {Target Hook} {const char *} TARGET_INVALID_PARAMETER_TYPE (tree @var{type})
10811 If defined, this macro returns the diagnostic message when it is
10812 invalid for functions to include parameters of type @var{type},
10813 or @code{NULL} if validity should be determined by
10814 the front end. This is currently used only by the C and C++ front ends.
10817 @deftypefn {Target Hook} {const char *} TARGET_INVALID_RETURN_TYPE (tree @var{type})
10818 If defined, this macro returns the diagnostic message when it is
10819 invalid for functions to have return type @var{type},
10820 or @code{NULL} if validity should be determined by
10821 the front end. This is currently used only by the C and C++ front ends.
10824 @deftypefn {Target Hook} {tree} TARGET_PROMOTED_TYPE (tree @var{type})
10825 If defined, this target hook returns the type to which values of
10826 @var{type} should be promoted when they appear in expressions,
10827 analogous to the integer promotions, or @code{NULL_TREE} to use the
10828 front end's normal promotion rules. This hook is useful when there are
10829 target-specific types with special promotion rules.
10830 This is currently used only by the C and C++ front ends.
10833 @deftypefn {Target Hook} {tree} TARGET_CONVERT_TO_TYPE (tree @var{type}, tree @var{expr})
10834 If defined, this hook returns the result of converting @var{expr} to
10835 @var{type}. It should return the converted expression,
10836 or @code{NULL_TREE} to apply the front end's normal conversion rules.
10837 This hook is useful when there are target-specific types with special
10839 This is currently used only by the C and C++ front ends.
10842 @defmac TARGET_USE_JCR_SECTION
10843 This macro determines whether to use the JCR section to register Java
10844 classes. By default, TARGET_USE_JCR_SECTION is defined to 1 if both
10845 SUPPORTS_WEAK and TARGET_HAVE_NAMED_SECTIONS are true, else 0.
10849 This macro determines the size of the objective C jump buffer for the
10850 NeXT runtime. By default, OBJC_JBLEN is defined to an innocuous value.
10853 @defmac LIBGCC2_UNWIND_ATTRIBUTE
10854 Define this macro if any target-specific attributes need to be attached
10855 to the functions in @file{libgcc} that provide low-level support for
10856 call stack unwinding. It is used in declarations in @file{unwind-generic.h}
10857 and the associated definitions of those functions.
10860 @deftypefn {Target Hook} void TARGET_UPDATE_STACK_BOUNDARY (void)
10861 Define this macro to update the current function stack boundary if
10865 @deftypefn {Target Hook} rtx TARGET_GET_DRAP_RTX (void)
10866 Define this macro to an rtx for Dynamic Realign Argument Pointer if a
10867 different argument pointer register is needed to access the function's
10868 argument list when stack is aligned.
10871 @deftypefn {Target Hook} {bool} TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS (void)
10872 When optimization is disabled, this hook indicates whether or not
10873 arguments should be allocated to stack slots. Normally, GCC allocates
10874 stacks slots for arguments when not optimizing in order to make
10875 debugging easier. However, when a function is declared with
10876 @code{__attribute__((naked))}, there is no stack frame, and the compiler
10877 cannot safely move arguments from the registers in which they are passed
10878 to the stack. Therefore, this hook should return true in general, but
10879 false for naked functions. The default implementation always returns true.
10883 @deftypevr {Target Hook} {unsigned HOST_WIDE_INT} TARGET_CONST_ANCHOR
10884 On some architectures it can take multiple instructions to synthesize
10885 a constant. If there is another constant already in a register that
10886 is close enough in value then it is preferable that the new constant
10887 is computed from this register using immediate addition or
10888 substraction. We accomplish this through CSE. Besides the value of
10889 the constant we also add a lower and an upper constant anchor to the
10890 available expressions. These are then queried when encountering new
10891 constants. The anchors are computed by rounding the constant up and
10892 down to a multiple of the value of @code{TARGET_CONST_ANCHOR}.
10893 @code{TARGET_CONST_ANCHOR} should be the maximum positive value
10894 accepted by immediate-add plus one. We currently assume that the
10895 value of @code{TARGET_CONST_ANCHOR} is a power of 2. For example, on
10896 MIPS, where add-immediate takes a 16-bit signed value,
10897 @code{TARGET_CONST_ANCHOR} is set to @samp{0x8000}. The default value
10898 is zero, which disables this optimization. @end deftypevr