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 @defmac PROMOTE_FUNCTION_MODE
1043 Like @code{PROMOTE_MODE}, but is applied to outgoing function arguments or
1044 function return values, as specified by @code{TARGET_PROMOTE_FUNCTION_ARGS}
1045 and @code{TARGET_PROMOTE_FUNCTION_RETURN}, respectively.
1047 The default is @code{PROMOTE_MODE}.
1050 @deftypefn {Target Hook} bool TARGET_PROMOTE_FUNCTION_ARGS (tree @var{fntype})
1051 This target hook should return @code{true} if the promotion described by
1052 @code{PROMOTE_FUNCTION_MODE} should be done for outgoing function
1056 @deftypefn {Target Hook} bool TARGET_PROMOTE_FUNCTION_RETURN (tree @var{fntype})
1057 This target hook should return @code{true} if the promotion described by
1058 @code{PROMOTE_FUNCTION_MODE} should be done for the return value of
1061 If this target hook returns @code{true}, @code{TARGET_FUNCTION_VALUE}
1062 must perform the same promotions done by @code{PROMOTE_FUNCTION_MODE}.
1065 @defmac PARM_BOUNDARY
1066 Normal alignment required for function parameters on the stack, in
1067 bits. All stack parameters receive at least this much alignment
1068 regardless of data type. On most machines, this is the same as the
1072 @defmac STACK_BOUNDARY
1073 Define this macro to the minimum alignment enforced by hardware for the
1074 stack pointer on this machine. The definition is a C expression for the
1075 desired alignment (measured in bits). This value is used as a default
1076 if @code{PREFERRED_STACK_BOUNDARY} is not defined. On most machines,
1077 this should be the same as @code{PARM_BOUNDARY}.
1080 @defmac PREFERRED_STACK_BOUNDARY
1081 Define this macro if you wish to preserve a certain alignment for the
1082 stack pointer, greater than what the hardware enforces. The definition
1083 is a C expression for the desired alignment (measured in bits). This
1084 macro must evaluate to a value equal to or larger than
1085 @code{STACK_BOUNDARY}.
1088 @defmac INCOMING_STACK_BOUNDARY
1089 Define this macro if the incoming stack boundary may be different
1090 from @code{PREFERRED_STACK_BOUNDARY}. This macro must evaluate
1091 to a value equal to or larger than @code{STACK_BOUNDARY}.
1094 @defmac FUNCTION_BOUNDARY
1095 Alignment required for a function entry point, in bits.
1098 @defmac BIGGEST_ALIGNMENT
1099 Biggest alignment that any data type can require on this machine, in
1100 bits. Note that this is not the biggest alignment that is supported,
1101 just the biggest alignment that, when violated, may cause a fault.
1104 @defmac MALLOC_ABI_ALIGNMENT
1105 Alignment, in bits, a C conformant malloc implementation has to
1106 provide. If not defined, the default value is @code{BITS_PER_WORD}.
1109 @defmac ATTRIBUTE_ALIGNED_VALUE
1110 Alignment used by the @code{__attribute__ ((aligned))} construct. If
1111 not defined, the default value is @code{BIGGEST_ALIGNMENT}.
1114 @defmac MINIMUM_ATOMIC_ALIGNMENT
1115 If defined, the smallest alignment, in bits, that can be given to an
1116 object that can be referenced in one operation, without disturbing any
1117 nearby object. Normally, this is @code{BITS_PER_UNIT}, but may be larger
1118 on machines that don't have byte or half-word store operations.
1121 @defmac BIGGEST_FIELD_ALIGNMENT
1122 Biggest alignment that any structure or union field can require on this
1123 machine, in bits. If defined, this overrides @code{BIGGEST_ALIGNMENT} for
1124 structure and union fields only, unless the field alignment has been set
1125 by the @code{__attribute__ ((aligned (@var{n})))} construct.
1128 @defmac ADJUST_FIELD_ALIGN (@var{field}, @var{computed})
1129 An expression for the alignment of a structure field @var{field} if the
1130 alignment computed in the usual way (including applying of
1131 @code{BIGGEST_ALIGNMENT} and @code{BIGGEST_FIELD_ALIGNMENT} to the
1132 alignment) is @var{computed}. It overrides alignment only if the
1133 field alignment has not been set by the
1134 @code{__attribute__ ((aligned (@var{n})))} construct.
1137 @defmac MAX_STACK_ALIGNMENT
1138 Biggest stack alignment guaranteed by the backend. Use this macro
1139 to specify the maximum alignment of a variable on stack.
1141 If not defined, the default value is @code{STACK_BOUNDARY}.
1143 @c FIXME: The default should be @code{PREFERRED_STACK_BOUNDARY}.
1144 @c But the fix for PR 32893 indicates that we can only guarantee
1145 @c maximum stack alignment on stack up to @code{STACK_BOUNDARY}, not
1146 @c @code{PREFERRED_STACK_BOUNDARY}, if stack alignment isn't supported.
1149 @defmac MAX_OFILE_ALIGNMENT
1150 Biggest alignment supported by the object file format of this machine.
1151 Use this macro to limit the alignment which can be specified using the
1152 @code{__attribute__ ((aligned (@var{n})))} construct. If not defined,
1153 the default value is @code{BIGGEST_ALIGNMENT}.
1155 On systems that use ELF, the default (in @file{config/elfos.h}) is
1156 the largest supported 32-bit ELF section alignment representable on
1157 a 32-bit host e.g. @samp{(((unsigned HOST_WIDEST_INT) 1 << 28) * 8)}.
1158 On 32-bit ELF the largest supported section alignment in bits is
1159 @samp{(0x80000000 * 8)}, but this is not representable on 32-bit hosts.
1162 @defmac DATA_ALIGNMENT (@var{type}, @var{basic-align})
1163 If defined, a C expression to compute the alignment for a variable in
1164 the static store. @var{type} is the data type, and @var{basic-align} is
1165 the alignment that the object would ordinarily have. The value of this
1166 macro is used instead of that alignment to align the object.
1168 If this macro is not defined, then @var{basic-align} is used.
1171 One use of this macro is to increase alignment of medium-size data to
1172 make it all fit in fewer cache lines. Another is to cause character
1173 arrays to be word-aligned so that @code{strcpy} calls that copy
1174 constants to character arrays can be done inline.
1177 @defmac CONSTANT_ALIGNMENT (@var{constant}, @var{basic-align})
1178 If defined, a C expression to compute the alignment given to a constant
1179 that is being placed in memory. @var{constant} is the constant and
1180 @var{basic-align} is the alignment that the object would ordinarily
1181 have. The value of this macro is used instead of that alignment to
1184 If this macro is not defined, then @var{basic-align} is used.
1186 The typical use of this macro is to increase alignment for string
1187 constants to be word aligned so that @code{strcpy} calls that copy
1188 constants can be done inline.
1191 @defmac LOCAL_ALIGNMENT (@var{type}, @var{basic-align})
1192 If defined, a C expression to compute the alignment for a variable in
1193 the local store. @var{type} is the data type, and @var{basic-align} is
1194 the alignment that the object would ordinarily have. The value of this
1195 macro is used instead of that alignment to align the object.
1197 If this macro is not defined, then @var{basic-align} is used.
1199 One use of this macro is to increase alignment of medium-size data to
1200 make it all fit in fewer cache lines.
1203 @defmac STACK_SLOT_ALIGNMENT (@var{type}, @var{mode}, @var{basic-align})
1204 If defined, a C expression to compute the alignment for stack slot.
1205 @var{type} is the data type, @var{mode} is the widest mode available,
1206 and @var{basic-align} is the alignment that the slot would ordinarily
1207 have. The value of this macro is used instead of that alignment to
1210 If this macro is not defined, then @var{basic-align} is used when
1211 @var{type} is @code{NULL}. Otherwise, @code{LOCAL_ALIGNMENT} will
1214 This macro is to set alignment of stack slot to the maximum alignment
1215 of all possible modes which the slot may have.
1218 @defmac LOCAL_DECL_ALIGNMENT (@var{decl})
1219 If defined, a C expression to compute the alignment for a local
1220 variable @var{decl}.
1222 If this macro is not defined, then
1223 @code{LOCAL_ALIGNMENT (TREE_TYPE (@var{decl}), DECL_ALIGN (@var{decl}))}
1226 One use of this macro is to increase alignment of medium-size data to
1227 make it all fit in fewer cache lines.
1230 @defmac EMPTY_FIELD_BOUNDARY
1231 Alignment in bits to be given to a structure bit-field that follows an
1232 empty field such as @code{int : 0;}.
1234 If @code{PCC_BITFIELD_TYPE_MATTERS} is true, it overrides this macro.
1237 @defmac STRUCTURE_SIZE_BOUNDARY
1238 Number of bits which any structure or union's size must be a multiple of.
1239 Each structure or union's size is rounded up to a multiple of this.
1241 If you do not define this macro, the default is the same as
1242 @code{BITS_PER_UNIT}.
1245 @defmac STRICT_ALIGNMENT
1246 Define this macro to be the value 1 if instructions will fail to work
1247 if given data not on the nominal alignment. If instructions will merely
1248 go slower in that case, define this macro as 0.
1251 @defmac PCC_BITFIELD_TYPE_MATTERS
1252 Define this if you wish to imitate the way many other C compilers handle
1253 alignment of bit-fields and the structures that contain them.
1255 The behavior is that the type written for a named bit-field (@code{int},
1256 @code{short}, or other integer type) imposes an alignment for the entire
1257 structure, as if the structure really did contain an ordinary field of
1258 that type. In addition, the bit-field is placed within the structure so
1259 that it would fit within such a field, not crossing a boundary for it.
1261 Thus, on most machines, a named bit-field whose type is written as
1262 @code{int} would not cross a four-byte boundary, and would force
1263 four-byte alignment for the whole structure. (The alignment used may
1264 not be four bytes; it is controlled by the other alignment parameters.)
1266 An unnamed bit-field will not affect the alignment of the containing
1269 If the macro is defined, its definition should be a C expression;
1270 a nonzero value for the expression enables this behavior.
1272 Note that if this macro is not defined, or its value is zero, some
1273 bit-fields may cross more than one alignment boundary. The compiler can
1274 support such references if there are @samp{insv}, @samp{extv}, and
1275 @samp{extzv} insns that can directly reference memory.
1277 The other known way of making bit-fields work is to define
1278 @code{STRUCTURE_SIZE_BOUNDARY} as large as @code{BIGGEST_ALIGNMENT}.
1279 Then every structure can be accessed with fullwords.
1281 Unless the machine has bit-field instructions or you define
1282 @code{STRUCTURE_SIZE_BOUNDARY} that way, you must define
1283 @code{PCC_BITFIELD_TYPE_MATTERS} to have a nonzero value.
1285 If your aim is to make GCC use the same conventions for laying out
1286 bit-fields as are used by another compiler, here is how to investigate
1287 what the other compiler does. Compile and run this program:
1306 printf ("Size of foo1 is %d\n",
1307 sizeof (struct foo1));
1308 printf ("Size of foo2 is %d\n",
1309 sizeof (struct foo2));
1314 If this prints 2 and 5, then the compiler's behavior is what you would
1315 get from @code{PCC_BITFIELD_TYPE_MATTERS}.
1318 @defmac BITFIELD_NBYTES_LIMITED
1319 Like @code{PCC_BITFIELD_TYPE_MATTERS} except that its effect is limited
1320 to aligning a bit-field within the structure.
1323 @deftypefn {Target Hook} bool TARGET_ALIGN_ANON_BITFIELD (void)
1324 When @code{PCC_BITFIELD_TYPE_MATTERS} is true this hook will determine
1325 whether unnamed bitfields affect the alignment of the containing
1326 structure. The hook should return true if the structure should inherit
1327 the alignment requirements of an unnamed bitfield's type.
1330 @deftypefn {Target Hook} bool TARGET_NARROW_VOLATILE_BITFIELD (void)
1331 This target hook should return @code{true} if accesses to volatile bitfields
1332 should use the narrowest mode possible. It should return @code{false} if
1333 these accesses should use the bitfield container type.
1335 The default is @code{!TARGET_STRICT_ALIGN}.
1338 @defmac MEMBER_TYPE_FORCES_BLK (@var{field}, @var{mode})
1339 Return 1 if a structure or array containing @var{field} should be accessed using
1342 If @var{field} is the only field in the structure, @var{mode} is its
1343 mode, otherwise @var{mode} is VOIDmode. @var{mode} is provided in the
1344 case where structures of one field would require the structure's mode to
1345 retain the field's mode.
1347 Normally, this is not needed.
1350 @defmac ROUND_TYPE_ALIGN (@var{type}, @var{computed}, @var{specified})
1351 Define this macro as an expression for the alignment of a type (given
1352 by @var{type} as a tree node) if the alignment computed in the usual
1353 way is @var{computed} and the alignment explicitly specified was
1356 The default is to use @var{specified} if it is larger; otherwise, use
1357 the smaller of @var{computed} and @code{BIGGEST_ALIGNMENT}
1360 @defmac MAX_FIXED_MODE_SIZE
1361 An integer expression for the size in bits of the largest integer
1362 machine mode that should actually be used. All integer machine modes of
1363 this size or smaller can be used for structures and unions with the
1364 appropriate sizes. If this macro is undefined, @code{GET_MODE_BITSIZE
1365 (DImode)} is assumed.
1368 @defmac STACK_SAVEAREA_MODE (@var{save_level})
1369 If defined, an expression of type @code{enum machine_mode} that
1370 specifies the mode of the save area operand of a
1371 @code{save_stack_@var{level}} named pattern (@pxref{Standard Names}).
1372 @var{save_level} is one of @code{SAVE_BLOCK}, @code{SAVE_FUNCTION}, or
1373 @code{SAVE_NONLOCAL} and selects which of the three named patterns is
1374 having its mode specified.
1376 You need not define this macro if it always returns @code{Pmode}. You
1377 would most commonly define this macro if the
1378 @code{save_stack_@var{level}} patterns need to support both a 32- and a
1382 @defmac STACK_SIZE_MODE
1383 If defined, an expression of type @code{enum machine_mode} that
1384 specifies the mode of the size increment operand of an
1385 @code{allocate_stack} named pattern (@pxref{Standard Names}).
1387 You need not define this macro if it always returns @code{word_mode}.
1388 You would most commonly define this macro if the @code{allocate_stack}
1389 pattern needs to support both a 32- and a 64-bit mode.
1392 @deftypefn {Target Hook} {enum machine_mode} TARGET_LIBGCC_CMP_RETURN_MODE ()
1393 This target hook should return the mode to be used for the return value
1394 of compare instructions expanded to libgcc calls. If not defined
1395 @code{word_mode} is returned which is the right choice for a majority of
1399 @deftypefn {Target Hook} {enum machine_mode} TARGET_LIBGCC_SHIFT_COUNT_MODE ()
1400 This target hook should return the mode to be used for the shift count operand
1401 of shift instructions expanded to libgcc calls. If not defined
1402 @code{word_mode} is returned which is the right choice for a majority of
1406 @defmac ROUND_TOWARDS_ZERO
1407 If defined, this macro should be true if the prevailing rounding
1408 mode is towards zero.
1410 Defining this macro only affects the way @file{libgcc.a} emulates
1411 floating-point arithmetic.
1413 Not defining this macro is equivalent to returning zero.
1416 @defmac LARGEST_EXPONENT_IS_NORMAL (@var{size})
1417 This macro should return true if floats with @var{size}
1418 bits do not have a NaN or infinity representation, but use the largest
1419 exponent for normal numbers instead.
1421 Defining this macro only affects the way @file{libgcc.a} emulates
1422 floating-point arithmetic.
1424 The default definition of this macro returns false for all sizes.
1427 @deftypefn {Target Hook} bool TARGET_MS_BITFIELD_LAYOUT_P (tree @var{record_type})
1428 This target hook returns @code{true} if bit-fields in the given
1429 @var{record_type} are to be laid out following the rules of Microsoft
1430 Visual C/C++, namely: (i) a bit-field won't share the same storage
1431 unit with the previous bit-field if their underlying types have
1432 different sizes, and the bit-field will be aligned to the highest
1433 alignment of the underlying types of itself and of the previous
1434 bit-field; (ii) a zero-sized bit-field will affect the alignment of
1435 the whole enclosing structure, even if it is unnamed; except that
1436 (iii) a zero-sized bit-field will be disregarded unless it follows
1437 another bit-field of nonzero size. If this hook returns @code{true},
1438 other macros that control bit-field layout are ignored.
1440 When a bit-field is inserted into a packed record, the whole size
1441 of the underlying type is used by one or more same-size adjacent
1442 bit-fields (that is, if its long:3, 32 bits is used in the record,
1443 and any additional adjacent long bit-fields are packed into the same
1444 chunk of 32 bits. However, if the size changes, a new field of that
1445 size is allocated). In an unpacked record, this is the same as using
1446 alignment, but not equivalent when packing.
1448 If both MS bit-fields and @samp{__attribute__((packed))} are used,
1449 the latter will take precedence. If @samp{__attribute__((packed))} is
1450 used on a single field when MS bit-fields are in use, it will take
1451 precedence for that field, but the alignment of the rest of the structure
1452 may affect its placement.
1455 @deftypefn {Target Hook} {bool} TARGET_DECIMAL_FLOAT_SUPPORTED_P (void)
1456 Returns true if the target supports decimal floating point.
1459 @deftypefn {Target Hook} {bool} TARGET_FIXED_POINT_SUPPORTED_P (void)
1460 Returns true if the target supports fixed-point arithmetic.
1463 @deftypefn {Target Hook} void TARGET_EXPAND_TO_RTL_HOOK (void)
1464 This hook is called just before expansion into rtl, allowing the target
1465 to perform additional initializations or analysis before the expansion.
1466 For example, the rs6000 port uses it to allocate a scratch stack slot
1467 for use in copying SDmode values between memory and floating point
1468 registers whenever the function being expanded has any SDmode
1472 @deftypefn {Target Hook} void TARGET_INSTANTIATE_DECLS (void)
1473 This hook allows the backend to perform additional instantiations on rtl
1474 that are not actually in any insns yet, but will be later.
1477 @deftypefn {Target Hook} {const char *} TARGET_MANGLE_TYPE (tree @var{type})
1478 If your target defines any fundamental types, or any types your target
1479 uses should be mangled differently from the default, define this hook
1480 to return the appropriate encoding for these types as part of a C++
1481 mangled name. The @var{type} argument is the tree structure representing
1482 the type to be mangled. The hook may be applied to trees which are
1483 not target-specific fundamental types; it should return @code{NULL}
1484 for all such types, as well as arguments it does not recognize. If the
1485 return value is not @code{NULL}, it must point to a statically-allocated
1488 Target-specific fundamental types might be new fundamental types or
1489 qualified versions of ordinary fundamental types. Encode new
1490 fundamental types as @samp{@w{u @var{n} @var{name}}}, where @var{name}
1491 is the name used for the type in source code, and @var{n} is the
1492 length of @var{name} in decimal. Encode qualified versions of
1493 ordinary types as @samp{@w{U @var{n} @var{name} @var{code}}}, where
1494 @var{name} is the name used for the type qualifier in source code,
1495 @var{n} is the length of @var{name} as above, and @var{code} is the
1496 code used to represent the unqualified version of this type. (See
1497 @code{write_builtin_type} in @file{cp/mangle.c} for the list of
1498 codes.) In both cases the spaces are for clarity; do not include any
1499 spaces in your string.
1501 This hook is applied to types prior to typedef resolution. If the mangled
1502 name for a particular type depends only on that type's main variant, you
1503 can perform typedef resolution yourself using @code{TYPE_MAIN_VARIANT}
1506 The default version of this hook always returns @code{NULL}, which is
1507 appropriate for a target that does not define any new fundamental
1512 @section Layout of Source Language Data Types
1514 These macros define the sizes and other characteristics of the standard
1515 basic data types used in programs being compiled. Unlike the macros in
1516 the previous section, these apply to specific features of C and related
1517 languages, rather than to fundamental aspects of storage layout.
1519 @defmac INT_TYPE_SIZE
1520 A C expression for the size in bits of the type @code{int} on the
1521 target machine. If you don't define this, the default is one word.
1524 @defmac SHORT_TYPE_SIZE
1525 A C expression for the size in bits of the type @code{short} on the
1526 target machine. If you don't define this, the default is half a word.
1527 (If this would be less than one storage unit, it is rounded up to one
1531 @defmac LONG_TYPE_SIZE
1532 A C expression for the size in bits of the type @code{long} on the
1533 target machine. If you don't define this, the default is one word.
1536 @defmac ADA_LONG_TYPE_SIZE
1537 On some machines, the size used for the Ada equivalent of the type
1538 @code{long} by a native Ada compiler differs from that used by C@. In
1539 that situation, define this macro to be a C expression to be used for
1540 the size of that type. If you don't define this, the default is the
1541 value of @code{LONG_TYPE_SIZE}.
1544 @defmac LONG_LONG_TYPE_SIZE
1545 A C expression for the size in bits of the type @code{long long} on the
1546 target machine. If you don't define this, the default is two
1547 words. If you want to support GNU Ada on your machine, the value of this
1548 macro must be at least 64.
1551 @defmac CHAR_TYPE_SIZE
1552 A C expression for the size in bits of the type @code{char} on the
1553 target machine. If you don't define this, the default is
1554 @code{BITS_PER_UNIT}.
1557 @defmac BOOL_TYPE_SIZE
1558 A C expression for the size in bits of the C++ type @code{bool} and
1559 C99 type @code{_Bool} on the target machine. If you don't define
1560 this, and you probably shouldn't, the default is @code{CHAR_TYPE_SIZE}.
1563 @defmac FLOAT_TYPE_SIZE
1564 A C expression for the size in bits of the type @code{float} on the
1565 target machine. If you don't define this, the default is one word.
1568 @defmac DOUBLE_TYPE_SIZE
1569 A C expression for the size in bits of the type @code{double} on the
1570 target machine. If you don't define this, the default is two
1574 @defmac LONG_DOUBLE_TYPE_SIZE
1575 A C expression for the size in bits of the type @code{long double} on
1576 the target machine. If you don't define this, the default is two
1580 @defmac SHORT_FRACT_TYPE_SIZE
1581 A C expression for the size in bits of the type @code{short _Fract} on
1582 the target machine. If you don't define this, the default is
1583 @code{BITS_PER_UNIT}.
1586 @defmac FRACT_TYPE_SIZE
1587 A C expression for the size in bits of the type @code{_Fract} on
1588 the target machine. If you don't define this, the default is
1589 @code{BITS_PER_UNIT * 2}.
1592 @defmac LONG_FRACT_TYPE_SIZE
1593 A C expression for the size in bits of the type @code{long _Fract} on
1594 the target machine. If you don't define this, the default is
1595 @code{BITS_PER_UNIT * 4}.
1598 @defmac LONG_LONG_FRACT_TYPE_SIZE
1599 A C expression for the size in bits of the type @code{long long _Fract} on
1600 the target machine. If you don't define this, the default is
1601 @code{BITS_PER_UNIT * 8}.
1604 @defmac SHORT_ACCUM_TYPE_SIZE
1605 A C expression for the size in bits of the type @code{short _Accum} on
1606 the target machine. If you don't define this, the default is
1607 @code{BITS_PER_UNIT * 2}.
1610 @defmac ACCUM_TYPE_SIZE
1611 A C expression for the size in bits of the type @code{_Accum} on
1612 the target machine. If you don't define this, the default is
1613 @code{BITS_PER_UNIT * 4}.
1616 @defmac LONG_ACCUM_TYPE_SIZE
1617 A C expression for the size in bits of the type @code{long _Accum} on
1618 the target machine. If you don't define this, the default is
1619 @code{BITS_PER_UNIT * 8}.
1622 @defmac LONG_LONG_ACCUM_TYPE_SIZE
1623 A C expression for the size in bits of the type @code{long long _Accum} on
1624 the target machine. If you don't define this, the default is
1625 @code{BITS_PER_UNIT * 16}.
1628 @defmac LIBGCC2_LONG_DOUBLE_TYPE_SIZE
1629 Define this macro if @code{LONG_DOUBLE_TYPE_SIZE} is not constant or
1630 if you want routines in @file{libgcc2.a} for a size other than
1631 @code{LONG_DOUBLE_TYPE_SIZE}. If you don't define this, the
1632 default is @code{LONG_DOUBLE_TYPE_SIZE}.
1635 @defmac LIBGCC2_HAS_DF_MODE
1636 Define this macro if neither @code{LIBGCC2_DOUBLE_TYPE_SIZE} nor
1637 @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is
1638 @code{DFmode} but you want @code{DFmode} routines in @file{libgcc2.a}
1639 anyway. If you don't define this and either @code{LIBGCC2_DOUBLE_TYPE_SIZE}
1640 or @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is 64 then the default is 1,
1644 @defmac LIBGCC2_HAS_XF_MODE
1645 Define this macro if @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is not
1646 @code{XFmode} but you want @code{XFmode} routines in @file{libgcc2.a}
1647 anyway. If you don't define this and @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE}
1648 is 80 then the default is 1, otherwise it is 0.
1651 @defmac LIBGCC2_HAS_TF_MODE
1652 Define this macro if @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is not
1653 @code{TFmode} but you want @code{TFmode} routines in @file{libgcc2.a}
1654 anyway. If you don't define this and @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE}
1655 is 128 then the default is 1, otherwise it is 0.
1662 Define these macros to be the size in bits of the mantissa of
1663 @code{SFmode}, @code{DFmode}, @code{XFmode} and @code{TFmode} values,
1664 if the defaults in @file{libgcc2.h} are inappropriate. By default,
1665 @code{FLT_MANT_DIG} is used for @code{SF_SIZE}, @code{LDBL_MANT_DIG}
1666 for @code{XF_SIZE} and @code{TF_SIZE}, and @code{DBL_MANT_DIG} or
1667 @code{LDBL_MANT_DIG} for @code{DF_SIZE} according to whether
1668 @code{LIBGCC2_DOUBLE_TYPE_SIZE} or
1669 @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is 64.
1672 @defmac TARGET_FLT_EVAL_METHOD
1673 A C expression for the value for @code{FLT_EVAL_METHOD} in @file{float.h},
1674 assuming, if applicable, that the floating-point control word is in its
1675 default state. If you do not define this macro the value of
1676 @code{FLT_EVAL_METHOD} will be zero.
1679 @defmac WIDEST_HARDWARE_FP_SIZE
1680 A C expression for the size in bits of the widest floating-point format
1681 supported by the hardware. If you define this macro, you must specify a
1682 value less than or equal to the value of @code{LONG_DOUBLE_TYPE_SIZE}.
1683 If you do not define this macro, the value of @code{LONG_DOUBLE_TYPE_SIZE}
1687 @defmac DEFAULT_SIGNED_CHAR
1688 An expression whose value is 1 or 0, according to whether the type
1689 @code{char} should be signed or unsigned by default. The user can
1690 always override this default with the options @option{-fsigned-char}
1691 and @option{-funsigned-char}.
1694 @deftypefn {Target Hook} bool TARGET_DEFAULT_SHORT_ENUMS (void)
1695 This target hook should return true if the compiler should give an
1696 @code{enum} type only as many bytes as it takes to represent the range
1697 of possible values of that type. It should return false if all
1698 @code{enum} types should be allocated like @code{int}.
1700 The default is to return false.
1704 A C expression for a string describing the name of the data type to use
1705 for size values. The typedef name @code{size_t} is defined using the
1706 contents of the string.
1708 The string can contain more than one keyword. If so, separate them with
1709 spaces, and write first any length keyword, then @code{unsigned} if
1710 appropriate, and finally @code{int}. The string must exactly match one
1711 of the data type names defined in the function
1712 @code{init_decl_processing} in the file @file{c-decl.c}. You may not
1713 omit @code{int} or change the order---that would cause the compiler to
1716 If you don't define this macro, the default is @code{"long unsigned
1720 @defmac PTRDIFF_TYPE
1721 A C expression for a string describing the name of the data type to use
1722 for the result of subtracting two pointers. The typedef name
1723 @code{ptrdiff_t} is defined using the contents of the string. See
1724 @code{SIZE_TYPE} above for more information.
1726 If you don't define this macro, the default is @code{"long int"}.
1730 A C expression for a string describing the name of the data type to use
1731 for wide characters. The typedef name @code{wchar_t} is defined using
1732 the contents of the string. See @code{SIZE_TYPE} above for more
1735 If you don't define this macro, the default is @code{"int"}.
1738 @defmac WCHAR_TYPE_SIZE
1739 A C expression for the size in bits of the data type for wide
1740 characters. This is used in @code{cpp}, which cannot make use of
1745 A C expression for a string describing the name of the data type to
1746 use for wide characters passed to @code{printf} and returned from
1747 @code{getwc}. The typedef name @code{wint_t} is defined using the
1748 contents of the string. See @code{SIZE_TYPE} above for more
1751 If you don't define this macro, the default is @code{"unsigned int"}.
1755 A C expression for a string describing the name of the data type that
1756 can represent any value of any standard or extended signed integer type.
1757 The typedef name @code{intmax_t} is defined using the contents of the
1758 string. See @code{SIZE_TYPE} above for more information.
1760 If you don't define this macro, the default is the first of
1761 @code{"int"}, @code{"long int"}, or @code{"long long int"} that has as
1762 much precision as @code{long long int}.
1765 @defmac UINTMAX_TYPE
1766 A C expression for a string describing the name of the data type that
1767 can represent any value of any standard or extended unsigned integer
1768 type. The typedef name @code{uintmax_t} is defined using the contents
1769 of the string. See @code{SIZE_TYPE} above for more information.
1771 If you don't define this macro, the default is the first of
1772 @code{"unsigned int"}, @code{"long unsigned int"}, or @code{"long long
1773 unsigned int"} that has as much precision as @code{long long unsigned
1777 @defmac SIG_ATOMIC_TYPE
1783 @defmacx UINT16_TYPE
1784 @defmacx UINT32_TYPE
1785 @defmacx UINT64_TYPE
1786 @defmacx INT_LEAST8_TYPE
1787 @defmacx INT_LEAST16_TYPE
1788 @defmacx INT_LEAST32_TYPE
1789 @defmacx INT_LEAST64_TYPE
1790 @defmacx UINT_LEAST8_TYPE
1791 @defmacx UINT_LEAST16_TYPE
1792 @defmacx UINT_LEAST32_TYPE
1793 @defmacx UINT_LEAST64_TYPE
1794 @defmacx INT_FAST8_TYPE
1795 @defmacx INT_FAST16_TYPE
1796 @defmacx INT_FAST32_TYPE
1797 @defmacx INT_FAST64_TYPE
1798 @defmacx UINT_FAST8_TYPE
1799 @defmacx UINT_FAST16_TYPE
1800 @defmacx UINT_FAST32_TYPE
1801 @defmacx UINT_FAST64_TYPE
1802 @defmacx INTPTR_TYPE
1803 @defmacx UINTPTR_TYPE
1804 C expressions for the standard types @code{sig_atomic_t},
1805 @code{int8_t}, @code{int16_t}, @code{int32_t}, @code{int64_t},
1806 @code{uint8_t}, @code{uint16_t}, @code{uint32_t}, @code{uint64_t},
1807 @code{int_least8_t}, @code{int_least16_t}, @code{int_least32_t},
1808 @code{int_least64_t}, @code{uint_least8_t}, @code{uint_least16_t},
1809 @code{uint_least32_t}, @code{uint_least64_t}, @code{int_fast8_t},
1810 @code{int_fast16_t}, @code{int_fast32_t}, @code{int_fast64_t},
1811 @code{uint_fast8_t}, @code{uint_fast16_t}, @code{uint_fast32_t},
1812 @code{uint_fast64_t}, @code{intptr_t}, and @code{uintptr_t}. See
1813 @code{SIZE_TYPE} above for more information.
1815 If any of these macros evaluates to a null pointer, the corresponding
1816 type is not supported; if GCC is configured to provide
1817 @code{<stdint.h>} in such a case, the header provided may not conform
1818 to C99, depending on the type in question. The defaults for all of
1819 these macros are null pointers.
1822 @defmac TARGET_PTRMEMFUNC_VBIT_LOCATION
1823 The C++ compiler represents a pointer-to-member-function with a struct
1830 ptrdiff_t vtable_index;
1837 The C++ compiler must use one bit to indicate whether the function that
1838 will be called through a pointer-to-member-function is virtual.
1839 Normally, we assume that the low-order bit of a function pointer must
1840 always be zero. Then, by ensuring that the vtable_index is odd, we can
1841 distinguish which variant of the union is in use. But, on some
1842 platforms function pointers can be odd, and so this doesn't work. In
1843 that case, we use the low-order bit of the @code{delta} field, and shift
1844 the remainder of the @code{delta} field to the left.
1846 GCC will automatically make the right selection about where to store
1847 this bit using the @code{FUNCTION_BOUNDARY} setting for your platform.
1848 However, some platforms such as ARM/Thumb have @code{FUNCTION_BOUNDARY}
1849 set such that functions always start at even addresses, but the lowest
1850 bit of pointers to functions indicate whether the function at that
1851 address is in ARM or Thumb mode. If this is the case of your
1852 architecture, you should define this macro to
1853 @code{ptrmemfunc_vbit_in_delta}.
1855 In general, you should not have to define this macro. On architectures
1856 in which function addresses are always even, according to
1857 @code{FUNCTION_BOUNDARY}, GCC will automatically define this macro to
1858 @code{ptrmemfunc_vbit_in_pfn}.
1861 @defmac TARGET_VTABLE_USES_DESCRIPTORS
1862 Normally, the C++ compiler uses function pointers in vtables. This
1863 macro allows the target to change to use ``function descriptors''
1864 instead. Function descriptors are found on targets for whom a
1865 function pointer is actually a small data structure. Normally the
1866 data structure consists of the actual code address plus a data
1867 pointer to which the function's data is relative.
1869 If vtables are used, the value of this macro should be the number
1870 of words that the function descriptor occupies.
1873 @defmac TARGET_VTABLE_ENTRY_ALIGN
1874 By default, the vtable entries are void pointers, the so the alignment
1875 is the same as pointer alignment. The value of this macro specifies
1876 the alignment of the vtable entry in bits. It should be defined only
1877 when special alignment is necessary. */
1880 @defmac TARGET_VTABLE_DATA_ENTRY_DISTANCE
1881 There are a few non-descriptor entries in the vtable at offsets below
1882 zero. If these entries must be padded (say, to preserve the alignment
1883 specified by @code{TARGET_VTABLE_ENTRY_ALIGN}), set this to the number
1884 of words in each data entry.
1888 @section Register Usage
1889 @cindex register usage
1891 This section explains how to describe what registers the target machine
1892 has, and how (in general) they can be used.
1894 The description of which registers a specific instruction can use is
1895 done with register classes; see @ref{Register Classes}. For information
1896 on using registers to access a stack frame, see @ref{Frame Registers}.
1897 For passing values in registers, see @ref{Register Arguments}.
1898 For returning values in registers, see @ref{Scalar Return}.
1901 * Register Basics:: Number and kinds of registers.
1902 * Allocation Order:: Order in which registers are allocated.
1903 * Values in Registers:: What kinds of values each reg can hold.
1904 * Leaf Functions:: Renumbering registers for leaf functions.
1905 * Stack Registers:: Handling a register stack such as 80387.
1908 @node Register Basics
1909 @subsection Basic Characteristics of Registers
1911 @c prevent bad page break with this line
1912 Registers have various characteristics.
1914 @defmac FIRST_PSEUDO_REGISTER
1915 Number of hardware registers known to the compiler. They receive
1916 numbers 0 through @code{FIRST_PSEUDO_REGISTER-1}; thus, the first
1917 pseudo register's number really is assigned the number
1918 @code{FIRST_PSEUDO_REGISTER}.
1921 @defmac FIXED_REGISTERS
1922 @cindex fixed register
1923 An initializer that says which registers are used for fixed purposes
1924 all throughout the compiled code and are therefore not available for
1925 general allocation. These would include the stack pointer, the frame
1926 pointer (except on machines where that can be used as a general
1927 register when no frame pointer is needed), the program counter on
1928 machines where that is considered one of the addressable registers,
1929 and any other numbered register with a standard use.
1931 This information is expressed as a sequence of numbers, separated by
1932 commas and surrounded by braces. The @var{n}th number is 1 if
1933 register @var{n} is fixed, 0 otherwise.
1935 The table initialized from this macro, and the table initialized by
1936 the following one, may be overridden at run time either automatically,
1937 by the actions of the macro @code{CONDITIONAL_REGISTER_USAGE}, or by
1938 the user with the command options @option{-ffixed-@var{reg}},
1939 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}.
1942 @defmac CALL_USED_REGISTERS
1943 @cindex call-used register
1944 @cindex call-clobbered register
1945 @cindex call-saved register
1946 Like @code{FIXED_REGISTERS} but has 1 for each register that is
1947 clobbered (in general) by function calls as well as for fixed
1948 registers. This macro therefore identifies the registers that are not
1949 available for general allocation of values that must live across
1952 If a register has 0 in @code{CALL_USED_REGISTERS}, the compiler
1953 automatically saves it on function entry and restores it on function
1954 exit, if the register is used within the function.
1957 @defmac CALL_REALLY_USED_REGISTERS
1958 @cindex call-used register
1959 @cindex call-clobbered register
1960 @cindex call-saved register
1961 Like @code{CALL_USED_REGISTERS} except this macro doesn't require
1962 that the entire set of @code{FIXED_REGISTERS} be included.
1963 (@code{CALL_USED_REGISTERS} must be a superset of @code{FIXED_REGISTERS}).
1964 This macro is optional. If not specified, it defaults to the value
1965 of @code{CALL_USED_REGISTERS}.
1968 @defmac HARD_REGNO_CALL_PART_CLOBBERED (@var{regno}, @var{mode})
1969 @cindex call-used register
1970 @cindex call-clobbered register
1971 @cindex call-saved register
1972 A C expression that is nonzero if it is not permissible to store a
1973 value of mode @var{mode} in hard register number @var{regno} across a
1974 call without some part of it being clobbered. For most machines this
1975 macro need not be defined. It is only required for machines that do not
1976 preserve the entire contents of a register across a call.
1980 @findex call_used_regs
1983 @findex reg_class_contents
1984 @defmac CONDITIONAL_REGISTER_USAGE
1985 Zero or more C statements that may conditionally modify five variables
1986 @code{fixed_regs}, @code{call_used_regs}, @code{global_regs},
1987 @code{reg_names}, and @code{reg_class_contents}, to take into account
1988 any dependence of these register sets on target flags. The first three
1989 of these are of type @code{char []} (interpreted as Boolean vectors).
1990 @code{global_regs} is a @code{const char *[]}, and
1991 @code{reg_class_contents} is a @code{HARD_REG_SET}. Before the macro is
1992 called, @code{fixed_regs}, @code{call_used_regs},
1993 @code{reg_class_contents}, and @code{reg_names} have been initialized
1994 from @code{FIXED_REGISTERS}, @code{CALL_USED_REGISTERS},
1995 @code{REG_CLASS_CONTENTS}, and @code{REGISTER_NAMES}, respectively.
1996 @code{global_regs} has been cleared, and any @option{-ffixed-@var{reg}},
1997 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}
1998 command options have been applied.
2000 You need not define this macro if it has no work to do.
2002 @cindex disabling certain registers
2003 @cindex controlling register usage
2004 If the usage of an entire class of registers depends on the target
2005 flags, you may indicate this to GCC by using this macro to modify
2006 @code{fixed_regs} and @code{call_used_regs} to 1 for each of the
2007 registers in the classes which should not be used by GCC@. Also define
2008 the macro @code{REG_CLASS_FROM_LETTER} / @code{REG_CLASS_FROM_CONSTRAINT}
2009 to return @code{NO_REGS} if it
2010 is called with a letter for a class that shouldn't be used.
2012 (However, if this class is not included in @code{GENERAL_REGS} and all
2013 of the insn patterns whose constraints permit this class are
2014 controlled by target switches, then GCC will automatically avoid using
2015 these registers when the target switches are opposed to them.)
2018 @defmac INCOMING_REGNO (@var{out})
2019 Define this macro if the target machine has register windows. This C
2020 expression returns the register number as seen by the called function
2021 corresponding to the register number @var{out} as seen by the calling
2022 function. Return @var{out} if register number @var{out} is not an
2026 @defmac OUTGOING_REGNO (@var{in})
2027 Define this macro if the target machine has register windows. This C
2028 expression returns the register number as seen by the calling function
2029 corresponding to the register number @var{in} as seen by the called
2030 function. Return @var{in} if register number @var{in} is not an inbound
2034 @defmac LOCAL_REGNO (@var{regno})
2035 Define this macro if the target machine has register windows. This C
2036 expression returns true if the register is call-saved but is in the
2037 register window. Unlike most call-saved registers, such registers
2038 need not be explicitly restored on function exit or during non-local
2043 If the program counter has a register number, define this as that
2044 register number. Otherwise, do not define it.
2047 @node Allocation Order
2048 @subsection Order of Allocation of Registers
2049 @cindex order of register allocation
2050 @cindex register allocation order
2052 @c prevent bad page break with this line
2053 Registers are allocated in order.
2055 @defmac REG_ALLOC_ORDER
2056 If defined, an initializer for a vector of integers, containing the
2057 numbers of hard registers in the order in which GCC should prefer
2058 to use them (from most preferred to least).
2060 If this macro is not defined, registers are used lowest numbered first
2061 (all else being equal).
2063 One use of this macro is on machines where the highest numbered
2064 registers must always be saved and the save-multiple-registers
2065 instruction supports only sequences of consecutive registers. On such
2066 machines, define @code{REG_ALLOC_ORDER} to be an initializer that lists
2067 the highest numbered allocable register first.
2070 @defmac ORDER_REGS_FOR_LOCAL_ALLOC
2071 A C statement (sans semicolon) to choose the order in which to allocate
2072 hard registers for pseudo-registers local to a basic block.
2074 Store the desired register order in the array @code{reg_alloc_order}.
2075 Element 0 should be the register to allocate first; element 1, the next
2076 register; and so on.
2078 The macro body should not assume anything about the contents of
2079 @code{reg_alloc_order} before execution of the macro.
2081 On most machines, it is not necessary to define this macro.
2084 @defmac IRA_HARD_REGNO_ADD_COST_MULTIPLIER (@var{regno})
2085 In some case register allocation order is not enough for the
2086 Integrated Register Allocator (@acronym{IRA}) to generate a good code.
2087 If this macro is defined, it should return a floating point value
2088 based on @var{regno}. The cost of using @var{regno} for a pseudo will
2089 be increased by approximately the pseudo's usage frequency times the
2090 value returned by this macro. Not defining this macro is equivalent
2091 to having it always return @code{0.0}.
2093 On most machines, it is not necessary to define this macro.
2096 @node Values in Registers
2097 @subsection How Values Fit in Registers
2099 This section discusses the macros that describe which kinds of values
2100 (specifically, which machine modes) each register can hold, and how many
2101 consecutive registers are needed for a given mode.
2103 @defmac HARD_REGNO_NREGS (@var{regno}, @var{mode})
2104 A C expression for the number of consecutive hard registers, starting
2105 at register number @var{regno}, required to hold a value of mode
2106 @var{mode}. This macro must never return zero, even if a register
2107 cannot hold the requested mode - indicate that with HARD_REGNO_MODE_OK
2108 and/or CANNOT_CHANGE_MODE_CLASS instead.
2110 On a machine where all registers are exactly one word, a suitable
2111 definition of this macro is
2114 #define HARD_REGNO_NREGS(REGNO, MODE) \
2115 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) \
2120 @defmac HARD_REGNO_NREGS_HAS_PADDING (@var{regno}, @var{mode})
2121 A C expression that is nonzero if a value of mode @var{mode}, stored
2122 in memory, ends with padding that causes it to take up more space than
2123 in registers starting at register number @var{regno} (as determined by
2124 multiplying GCC's notion of the size of the register when containing
2125 this mode by the number of registers returned by
2126 @code{HARD_REGNO_NREGS}). By default this is zero.
2128 For example, if a floating-point value is stored in three 32-bit
2129 registers but takes up 128 bits in memory, then this would be
2132 This macros only needs to be defined if there are cases where
2133 @code{subreg_get_info}
2134 would otherwise wrongly determine that a @code{subreg} can be
2135 represented by an offset to the register number, when in fact such a
2136 @code{subreg} would contain some of the padding not stored in
2137 registers and so not be representable.
2140 @defmac HARD_REGNO_NREGS_WITH_PADDING (@var{regno}, @var{mode})
2141 For values of @var{regno} and @var{mode} for which
2142 @code{HARD_REGNO_NREGS_HAS_PADDING} returns nonzero, a C expression
2143 returning the greater number of registers required to hold the value
2144 including any padding. In the example above, the value would be four.
2147 @defmac REGMODE_NATURAL_SIZE (@var{mode})
2148 Define this macro if the natural size of registers that hold values
2149 of mode @var{mode} is not the word size. It is a C expression that
2150 should give the natural size in bytes for the specified mode. It is
2151 used by the register allocator to try to optimize its results. This
2152 happens for example on SPARC 64-bit where the natural size of
2153 floating-point registers is still 32-bit.
2156 @defmac HARD_REGNO_MODE_OK (@var{regno}, @var{mode})
2157 A C expression that is nonzero if it is permissible to store a value
2158 of mode @var{mode} in hard register number @var{regno} (or in several
2159 registers starting with that one). For a machine where all registers
2160 are equivalent, a suitable definition is
2163 #define HARD_REGNO_MODE_OK(REGNO, MODE) 1
2166 You need not include code to check for the numbers of fixed registers,
2167 because the allocation mechanism considers them to be always occupied.
2169 @cindex register pairs
2170 On some machines, double-precision values must be kept in even/odd
2171 register pairs. You can implement that by defining this macro to reject
2172 odd register numbers for such modes.
2174 The minimum requirement for a mode to be OK in a register is that the
2175 @samp{mov@var{mode}} instruction pattern support moves between the
2176 register and other hard register in the same class and that moving a
2177 value into the register and back out not alter it.
2179 Since the same instruction used to move @code{word_mode} will work for
2180 all narrower integer modes, it is not necessary on any machine for
2181 @code{HARD_REGNO_MODE_OK} to distinguish between these modes, provided
2182 you define patterns @samp{movhi}, etc., to take advantage of this. This
2183 is useful because of the interaction between @code{HARD_REGNO_MODE_OK}
2184 and @code{MODES_TIEABLE_P}; it is very desirable for all integer modes
2187 Many machines have special registers for floating point arithmetic.
2188 Often people assume that floating point machine modes are allowed only
2189 in floating point registers. This is not true. Any registers that
2190 can hold integers can safely @emph{hold} a floating point machine
2191 mode, whether or not floating arithmetic can be done on it in those
2192 registers. Integer move instructions can be used to move the values.
2194 On some machines, though, the converse is true: fixed-point machine
2195 modes may not go in floating registers. This is true if the floating
2196 registers normalize any value stored in them, because storing a
2197 non-floating value there would garble it. In this case,
2198 @code{HARD_REGNO_MODE_OK} should reject fixed-point machine modes in
2199 floating registers. But if the floating registers do not automatically
2200 normalize, if you can store any bit pattern in one and retrieve it
2201 unchanged without a trap, then any machine mode may go in a floating
2202 register, so you can define this macro to say so.
2204 The primary significance of special floating registers is rather that
2205 they are the registers acceptable in floating point arithmetic
2206 instructions. However, this is of no concern to
2207 @code{HARD_REGNO_MODE_OK}. You handle it by writing the proper
2208 constraints for those instructions.
2210 On some machines, the floating registers are especially slow to access,
2211 so that it is better to store a value in a stack frame than in such a
2212 register if floating point arithmetic is not being done. As long as the
2213 floating registers are not in class @code{GENERAL_REGS}, they will not
2214 be used unless some pattern's constraint asks for one.
2217 @defmac HARD_REGNO_RENAME_OK (@var{from}, @var{to})
2218 A C expression that is nonzero if it is OK to rename a hard register
2219 @var{from} to another hard register @var{to}.
2221 One common use of this macro is to prevent renaming of a register to
2222 another register that is not saved by a prologue in an interrupt
2225 The default is always nonzero.
2228 @defmac MODES_TIEABLE_P (@var{mode1}, @var{mode2})
2229 A C expression that is nonzero if a value of mode
2230 @var{mode1} is accessible in mode @var{mode2} without copying.
2232 If @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode1})} and
2233 @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode2})} are always the same for
2234 any @var{r}, then @code{MODES_TIEABLE_P (@var{mode1}, @var{mode2})}
2235 should be nonzero. If they differ for any @var{r}, you should define
2236 this macro to return zero unless some other mechanism ensures the
2237 accessibility of the value in a narrower mode.
2239 You should define this macro to return nonzero in as many cases as
2240 possible since doing so will allow GCC to perform better register
2244 @deftypefn {Target Hook} bool TARGET_HARD_REGNO_SCRATCH_OK (unsigned int @var{regno})
2245 This target hook should return @code{true} if it is OK to use a hard register
2246 @var{regno} as scratch reg in peephole2.
2248 One common use of this macro is to prevent using of a register that
2249 is not saved by a prologue in an interrupt handler.
2251 The default version of this hook always returns @code{true}.
2254 @defmac AVOID_CCMODE_COPIES
2255 Define this macro if the compiler should avoid copies to/from @code{CCmode}
2256 registers. You should only define this macro if support for copying to/from
2257 @code{CCmode} is incomplete.
2260 @node Leaf Functions
2261 @subsection Handling Leaf Functions
2263 @cindex leaf functions
2264 @cindex functions, leaf
2265 On some machines, a leaf function (i.e., one which makes no calls) can run
2266 more efficiently if it does not make its own register window. Often this
2267 means it is required to receive its arguments in the registers where they
2268 are passed by the caller, instead of the registers where they would
2271 The special treatment for leaf functions generally applies only when
2272 other conditions are met; for example, often they may use only those
2273 registers for its own variables and temporaries. We use the term ``leaf
2274 function'' to mean a function that is suitable for this special
2275 handling, so that functions with no calls are not necessarily ``leaf
2278 GCC assigns register numbers before it knows whether the function is
2279 suitable for leaf function treatment. So it needs to renumber the
2280 registers in order to output a leaf function. The following macros
2283 @defmac LEAF_REGISTERS
2284 Name of a char vector, indexed by hard register number, which
2285 contains 1 for a register that is allowable in a candidate for leaf
2288 If leaf function treatment involves renumbering the registers, then the
2289 registers marked here should be the ones before renumbering---those that
2290 GCC would ordinarily allocate. The registers which will actually be
2291 used in the assembler code, after renumbering, should not be marked with 1
2294 Define this macro only if the target machine offers a way to optimize
2295 the treatment of leaf functions.
2298 @defmac LEAF_REG_REMAP (@var{regno})
2299 A C expression whose value is the register number to which @var{regno}
2300 should be renumbered, when a function is treated as a leaf function.
2302 If @var{regno} is a register number which should not appear in a leaf
2303 function before renumbering, then the expression should yield @minus{}1, which
2304 will cause the compiler to abort.
2306 Define this macro only if the target machine offers a way to optimize the
2307 treatment of leaf functions, and registers need to be renumbered to do
2311 @findex current_function_is_leaf
2312 @findex current_function_uses_only_leaf_regs
2313 @code{TARGET_ASM_FUNCTION_PROLOGUE} and
2314 @code{TARGET_ASM_FUNCTION_EPILOGUE} must usually treat leaf functions
2315 specially. They can test the C variable @code{current_function_is_leaf}
2316 which is nonzero for leaf functions. @code{current_function_is_leaf} is
2317 set prior to local register allocation and is valid for the remaining
2318 compiler passes. They can also test the C variable
2319 @code{current_function_uses_only_leaf_regs} which is nonzero for leaf
2320 functions which only use leaf registers.
2321 @code{current_function_uses_only_leaf_regs} is valid after all passes
2322 that modify the instructions have been run and is only useful if
2323 @code{LEAF_REGISTERS} is defined.
2324 @c changed this to fix overfull. ALSO: why the "it" at the beginning
2325 @c of the next paragraph?! --mew 2feb93
2327 @node Stack Registers
2328 @subsection Registers That Form a Stack
2330 There are special features to handle computers where some of the
2331 ``registers'' form a stack. Stack registers are normally written by
2332 pushing onto the stack, and are numbered relative to the top of the
2335 Currently, GCC can only handle one group of stack-like registers, and
2336 they must be consecutively numbered. Furthermore, the existing
2337 support for stack-like registers is specific to the 80387 floating
2338 point coprocessor. If you have a new architecture that uses
2339 stack-like registers, you will need to do substantial work on
2340 @file{reg-stack.c} and write your machine description to cooperate
2341 with it, as well as defining these macros.
2344 Define this if the machine has any stack-like registers.
2347 @defmac FIRST_STACK_REG
2348 The number of the first stack-like register. This one is the top
2352 @defmac LAST_STACK_REG
2353 The number of the last stack-like register. This one is the bottom of
2357 @node Register Classes
2358 @section Register Classes
2359 @cindex register class definitions
2360 @cindex class definitions, register
2362 On many machines, the numbered registers are not all equivalent.
2363 For example, certain registers may not be allowed for indexed addressing;
2364 certain registers may not be allowed in some instructions. These machine
2365 restrictions are described to the compiler using @dfn{register classes}.
2367 You define a number of register classes, giving each one a name and saying
2368 which of the registers belong to it. Then you can specify register classes
2369 that are allowed as operands to particular instruction patterns.
2373 In general, each register will belong to several classes. In fact, one
2374 class must be named @code{ALL_REGS} and contain all the registers. Another
2375 class must be named @code{NO_REGS} and contain no registers. Often the
2376 union of two classes will be another class; however, this is not required.
2378 @findex GENERAL_REGS
2379 One of the classes must be named @code{GENERAL_REGS}. There is nothing
2380 terribly special about the name, but the operand constraint letters
2381 @samp{r} and @samp{g} specify this class. If @code{GENERAL_REGS} is
2382 the same as @code{ALL_REGS}, just define it as a macro which expands
2385 Order the classes so that if class @var{x} is contained in class @var{y}
2386 then @var{x} has a lower class number than @var{y}.
2388 The way classes other than @code{GENERAL_REGS} are specified in operand
2389 constraints is through machine-dependent operand constraint letters.
2390 You can define such letters to correspond to various classes, then use
2391 them in operand constraints.
2393 You should define a class for the union of two classes whenever some
2394 instruction allows both classes. For example, if an instruction allows
2395 either a floating point (coprocessor) register or a general register for a
2396 certain operand, you should define a class @code{FLOAT_OR_GENERAL_REGS}
2397 which includes both of them. Otherwise you will get suboptimal code.
2399 You must also specify certain redundant information about the register
2400 classes: for each class, which classes contain it and which ones are
2401 contained in it; for each pair of classes, the largest class contained
2404 When a value occupying several consecutive registers is expected in a
2405 certain class, all the registers used must belong to that class.
2406 Therefore, register classes cannot be used to enforce a requirement for
2407 a register pair to start with an even-numbered register. The way to
2408 specify this requirement is with @code{HARD_REGNO_MODE_OK}.
2410 Register classes used for input-operands of bitwise-and or shift
2411 instructions have a special requirement: each such class must have, for
2412 each fixed-point machine mode, a subclass whose registers can transfer that
2413 mode to or from memory. For example, on some machines, the operations for
2414 single-byte values (@code{QImode}) are limited to certain registers. When
2415 this is so, each register class that is used in a bitwise-and or shift
2416 instruction must have a subclass consisting of registers from which
2417 single-byte values can be loaded or stored. This is so that
2418 @code{PREFERRED_RELOAD_CLASS} can always have a possible value to return.
2420 @deftp {Data type} {enum reg_class}
2421 An enumerated type that must be defined with all the register class names
2422 as enumerated values. @code{NO_REGS} must be first. @code{ALL_REGS}
2423 must be the last register class, followed by one more enumerated value,
2424 @code{LIM_REG_CLASSES}, which is not a register class but rather
2425 tells how many classes there are.
2427 Each register class has a number, which is the value of casting
2428 the class name to type @code{int}. The number serves as an index
2429 in many of the tables described below.
2432 @defmac N_REG_CLASSES
2433 The number of distinct register classes, defined as follows:
2436 #define N_REG_CLASSES (int) LIM_REG_CLASSES
2440 @defmac REG_CLASS_NAMES
2441 An initializer containing the names of the register classes as C string
2442 constants. These names are used in writing some of the debugging dumps.
2445 @defmac REG_CLASS_CONTENTS
2446 An initializer containing the contents of the register classes, as integers
2447 which are bit masks. The @var{n}th integer specifies the contents of class
2448 @var{n}. The way the integer @var{mask} is interpreted is that
2449 register @var{r} is in the class if @code{@var{mask} & (1 << @var{r})} is 1.
2451 When the machine has more than 32 registers, an integer does not suffice.
2452 Then the integers are replaced by sub-initializers, braced groupings containing
2453 several integers. Each sub-initializer must be suitable as an initializer
2454 for the type @code{HARD_REG_SET} which is defined in @file{hard-reg-set.h}.
2455 In this situation, the first integer in each sub-initializer corresponds to
2456 registers 0 through 31, the second integer to registers 32 through 63, and
2460 @defmac REGNO_REG_CLASS (@var{regno})
2461 A C expression whose value is a register class containing hard register
2462 @var{regno}. In general there is more than one such class; choose a class
2463 which is @dfn{minimal}, meaning that no smaller class also contains the
2467 @defmac BASE_REG_CLASS
2468 A macro whose definition is the name of the class to which a valid
2469 base register must belong. A base register is one used in an address
2470 which is the register value plus a displacement.
2473 @defmac MODE_BASE_REG_CLASS (@var{mode})
2474 This is a variation of the @code{BASE_REG_CLASS} macro which allows
2475 the selection of a base register in a mode dependent manner. If
2476 @var{mode} is VOIDmode then it should return the same value as
2477 @code{BASE_REG_CLASS}.
2480 @defmac MODE_BASE_REG_REG_CLASS (@var{mode})
2481 A C expression whose value is the register class to which a valid
2482 base register must belong in order to be used in a base plus index
2483 register address. You should define this macro if base plus index
2484 addresses have different requirements than other base register uses.
2487 @defmac MODE_CODE_BASE_REG_CLASS (@var{mode}, @var{outer_code}, @var{index_code})
2488 A C expression whose value is the register class to which a valid
2489 base register must belong. @var{outer_code} and @var{index_code} define the
2490 context in which the base register occurs. @var{outer_code} is the code of
2491 the immediately enclosing expression (@code{MEM} for the top level of an
2492 address, @code{ADDRESS} for something that occurs in an
2493 @code{address_operand}). @var{index_code} is the code of the corresponding
2494 index expression if @var{outer_code} is @code{PLUS}; @code{SCRATCH} otherwise.
2497 @defmac INDEX_REG_CLASS
2498 A macro whose definition is the name of the class to which a valid
2499 index register must belong. An index register is one used in an
2500 address where its value is either multiplied by a scale factor or
2501 added to another register (as well as added to a displacement).
2504 @defmac REGNO_OK_FOR_BASE_P (@var{num})
2505 A C expression which is nonzero if register number @var{num} is
2506 suitable for use as a base register in operand addresses.
2507 Like @code{TARGET_LEGITIMATE_ADDRESS_P}, this macro should also
2508 define a strict and a non-strict variant. Both variants behave
2509 the same for hard register; for pseudos, the strict variant will
2510 pass only those that have been allocated to a valid hard registers,
2511 while the non-strict variant will pass all pseudos.
2513 @findex REG_OK_STRICT
2514 Compiler source files that want to use the strict variant of this and
2515 other macros define the macro @code{REG_OK_STRICT}. You should use an
2516 @code{#ifdef REG_OK_STRICT} conditional to define the strict variant in
2517 that case and the non-strict variant otherwise.
2520 @defmac REGNO_MODE_OK_FOR_BASE_P (@var{num}, @var{mode})
2521 A C expression that is just like @code{REGNO_OK_FOR_BASE_P}, except that
2522 that expression may examine the mode of the memory reference in
2523 @var{mode}. You should define this macro if the mode of the memory
2524 reference affects whether a register may be used as a base register. If
2525 you define this macro, the compiler will use it instead of
2526 @code{REGNO_OK_FOR_BASE_P}. The mode may be @code{VOIDmode} for
2527 addresses that appear outside a @code{MEM}, i.e., as an
2528 @code{address_operand}.
2530 This macro also has strict and non-strict variants.
2533 @defmac REGNO_MODE_OK_FOR_REG_BASE_P (@var{num}, @var{mode})
2534 A C expression which is nonzero if register number @var{num} is suitable for
2535 use as a base register in base plus index operand addresses, accessing
2536 memory in mode @var{mode}. It may be either a suitable hard register or a
2537 pseudo register that has been allocated such a hard register. You should
2538 define this macro if base plus index addresses have different requirements
2539 than other base register uses.
2541 Use of this macro is deprecated; please use the more general
2542 @code{REGNO_MODE_CODE_OK_FOR_BASE_P}.
2544 This macro also has strict and non-strict variants.
2547 @defmac REGNO_MODE_CODE_OK_FOR_BASE_P (@var{num}, @var{mode}, @var{outer_code}, @var{index_code})
2548 A C expression that is just like @code{REGNO_MODE_OK_FOR_BASE_P}, except
2549 that that expression may examine the context in which the register
2550 appears in the memory reference. @var{outer_code} is the code of the
2551 immediately enclosing expression (@code{MEM} if at the top level of the
2552 address, @code{ADDRESS} for something that occurs in an
2553 @code{address_operand}). @var{index_code} is the code of the
2554 corresponding index expression if @var{outer_code} is @code{PLUS};
2555 @code{SCRATCH} otherwise. The mode may be @code{VOIDmode} for addresses
2556 that appear outside a @code{MEM}, i.e., as an @code{address_operand}.
2558 This macro also has strict and non-strict variants.
2561 @defmac REGNO_OK_FOR_INDEX_P (@var{num})
2562 A C expression which is nonzero if register number @var{num} is
2563 suitable for use as an index register in operand addresses. It may be
2564 either a suitable hard register or a pseudo register that has been
2565 allocated such a hard register.
2567 The difference between an index register and a base register is that
2568 the index register may be scaled. If an address involves the sum of
2569 two registers, neither one of them scaled, then either one may be
2570 labeled the ``base'' and the other the ``index''; but whichever
2571 labeling is used must fit the machine's constraints of which registers
2572 may serve in each capacity. The compiler will try both labelings,
2573 looking for one that is valid, and will reload one or both registers
2574 only if neither labeling works.
2576 This macro also has strict and non-strict variants.
2579 @defmac PREFERRED_RELOAD_CLASS (@var{x}, @var{class})
2580 A C expression that places additional restrictions on the register class
2581 to use when it is necessary to copy value @var{x} into a register in class
2582 @var{class}. The value is a register class; perhaps @var{class}, or perhaps
2583 another, smaller class. On many machines, the following definition is
2587 #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS
2590 Sometimes returning a more restrictive class makes better code. For
2591 example, on the 68000, when @var{x} is an integer constant that is in range
2592 for a @samp{moveq} instruction, the value of this macro is always
2593 @code{DATA_REGS} as long as @var{class} includes the data registers.
2594 Requiring a data register guarantees that a @samp{moveq} will be used.
2596 One case where @code{PREFERRED_RELOAD_CLASS} must not return
2597 @var{class} is if @var{x} is a legitimate constant which cannot be
2598 loaded into some register class. By returning @code{NO_REGS} you can
2599 force @var{x} into a memory location. For example, rs6000 can load
2600 immediate values into general-purpose registers, but does not have an
2601 instruction for loading an immediate value into a floating-point
2602 register, so @code{PREFERRED_RELOAD_CLASS} returns @code{NO_REGS} when
2603 @var{x} is a floating-point constant. If the constant can't be loaded
2604 into any kind of register, code generation will be better if
2605 @code{LEGITIMATE_CONSTANT_P} makes the constant illegitimate instead
2606 of using @code{PREFERRED_RELOAD_CLASS}.
2608 If an insn has pseudos in it after register allocation, reload will go
2609 through the alternatives and call repeatedly @code{PREFERRED_RELOAD_CLASS}
2610 to find the best one. Returning @code{NO_REGS}, in this case, makes
2611 reload add a @code{!} in front of the constraint: the x86 back-end uses
2612 this feature to discourage usage of 387 registers when math is done in
2613 the SSE registers (and vice versa).
2616 @defmac PREFERRED_OUTPUT_RELOAD_CLASS (@var{x}, @var{class})
2617 Like @code{PREFERRED_RELOAD_CLASS}, but for output reloads instead of
2618 input reloads. If you don't define this macro, the default is to use
2619 @var{class}, unchanged.
2621 You can also use @code{PREFERRED_OUTPUT_RELOAD_CLASS} to discourage
2622 reload from using some alternatives, like @code{PREFERRED_RELOAD_CLASS}.
2625 @defmac LIMIT_RELOAD_CLASS (@var{mode}, @var{class})
2626 A C expression that places additional restrictions on the register class
2627 to use when it is necessary to be able to hold a value of mode
2628 @var{mode} in a reload register for which class @var{class} would
2631 Unlike @code{PREFERRED_RELOAD_CLASS}, this macro should be used when
2632 there are certain modes that simply can't go in certain reload classes.
2634 The value is a register class; perhaps @var{class}, or perhaps another,
2637 Don't define this macro unless the target machine has limitations which
2638 require the macro to do something nontrivial.
2641 @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})
2642 Many machines have some registers that cannot be copied directly to or
2643 from memory or even from other types of registers. An example is the
2644 @samp{MQ} register, which on most machines, can only be copied to or
2645 from general registers, but not memory. Below, we shall be using the
2646 term 'intermediate register' when a move operation cannot be performed
2647 directly, but has to be done by copying the source into the intermediate
2648 register first, and then copying the intermediate register to the
2649 destination. An intermediate register always has the same mode as
2650 source and destination. Since it holds the actual value being copied,
2651 reload might apply optimizations to re-use an intermediate register
2652 and eliding the copy from the source when it can determine that the
2653 intermediate register still holds the required value.
2655 Another kind of secondary reload is required on some machines which
2656 allow copying all registers to and from memory, but require a scratch
2657 register for stores to some memory locations (e.g., those with symbolic
2658 address on the RT, and those with certain symbolic address on the SPARC
2659 when compiling PIC)@. Scratch registers need not have the same mode
2660 as the value being copied, and usually hold a different value than
2661 that being copied. Special patterns in the md file are needed to
2662 describe how the copy is performed with the help of the scratch register;
2663 these patterns also describe the number, register class(es) and mode(s)
2664 of the scratch register(s).
2666 In some cases, both an intermediate and a scratch register are required.
2668 For input reloads, this target hook is called with nonzero @var{in_p},
2669 and @var{x} is an rtx that needs to be copied to a register of class
2670 @var{reload_class} in @var{reload_mode}. For output reloads, this target
2671 hook is called with zero @var{in_p}, and a register of class @var{reload_class}
2672 needs to be copied to rtx @var{x} in @var{reload_mode}.
2674 If copying a register of @var{reload_class} from/to @var{x} requires
2675 an intermediate register, the hook @code{secondary_reload} should
2676 return the register class required for this intermediate register.
2677 If no intermediate register is required, it should return NO_REGS.
2678 If more than one intermediate register is required, describe the one
2679 that is closest in the copy chain to the reload register.
2681 If scratch registers are needed, you also have to describe how to
2682 perform the copy from/to the reload register to/from this
2683 closest intermediate register. Or if no intermediate register is
2684 required, but still a scratch register is needed, describe the
2685 copy from/to the reload register to/from the reload operand @var{x}.
2687 You do this by setting @code{sri->icode} to the instruction code of a pattern
2688 in the md file which performs the move. Operands 0 and 1 are the output
2689 and input of this copy, respectively. Operands from operand 2 onward are
2690 for scratch operands. These scratch operands must have a mode, and a
2691 single-register-class
2692 @c [later: or memory]
2695 When an intermediate register is used, the @code{secondary_reload}
2696 hook will be called again to determine how to copy the intermediate
2697 register to/from the reload operand @var{x}, so your hook must also
2698 have code to handle the register class of the intermediate operand.
2700 @c [For later: maybe we'll allow multi-alternative reload patterns -
2701 @c the port maintainer could name a mov<mode> pattern that has clobbers -
2702 @c and match the constraints of input and output to determine the required
2703 @c alternative. A restriction would be that constraints used to match
2704 @c against reloads registers would have to be written as register class
2705 @c constraints, or we need a new target macro / hook that tells us if an
2706 @c arbitrary constraint can match an unknown register of a given class.
2707 @c Such a macro / hook would also be useful in other places.]
2710 @var{x} might be a pseudo-register or a @code{subreg} of a
2711 pseudo-register, which could either be in a hard register or in memory.
2712 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2713 in memory and the hard register number if it is in a register.
2715 Scratch operands in memory (constraint @code{"=m"} / @code{"=&m"}) are
2716 currently not supported. For the time being, you will have to continue
2717 to use @code{SECONDARY_MEMORY_NEEDED} for that purpose.
2719 @code{copy_cost} also uses this target hook to find out how values are
2720 copied. If you want it to include some extra cost for the need to allocate
2721 (a) scratch register(s), set @code{sri->extra_cost} to the additional cost.
2722 Or if two dependent moves are supposed to have a lower cost than the sum
2723 of the individual moves due to expected fortuitous scheduling and/or special
2724 forwarding logic, you can set @code{sri->extra_cost} to a negative amount.
2727 @defmac SECONDARY_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2728 @defmacx SECONDARY_INPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2729 @defmacx SECONDARY_OUTPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2730 These macros are obsolete, new ports should use the target hook
2731 @code{TARGET_SECONDARY_RELOAD} instead.
2733 These are obsolete macros, replaced by the @code{TARGET_SECONDARY_RELOAD}
2734 target hook. Older ports still define these macros to indicate to the
2735 reload phase that it may
2736 need to allocate at least one register for a reload in addition to the
2737 register to contain the data. Specifically, if copying @var{x} to a
2738 register @var{class} in @var{mode} requires an intermediate register,
2739 you were supposed to define @code{SECONDARY_INPUT_RELOAD_CLASS} to return the
2740 largest register class all of whose registers can be used as
2741 intermediate registers or scratch registers.
2743 If copying a register @var{class} in @var{mode} to @var{x} requires an
2744 intermediate or scratch register, @code{SECONDARY_OUTPUT_RELOAD_CLASS}
2745 was supposed to be defined be defined to return the largest register
2746 class required. If the
2747 requirements for input and output reloads were the same, the macro
2748 @code{SECONDARY_RELOAD_CLASS} should have been used instead of defining both
2751 The values returned by these macros are often @code{GENERAL_REGS}.
2752 Return @code{NO_REGS} if no spare register is needed; i.e., if @var{x}
2753 can be directly copied to or from a register of @var{class} in
2754 @var{mode} without requiring a scratch register. Do not define this
2755 macro if it would always return @code{NO_REGS}.
2757 If a scratch register is required (either with or without an
2758 intermediate register), you were supposed to define patterns for
2759 @samp{reload_in@var{m}} or @samp{reload_out@var{m}}, as required
2760 (@pxref{Standard Names}. These patterns, which were normally
2761 implemented with a @code{define_expand}, should be similar to the
2762 @samp{mov@var{m}} patterns, except that operand 2 is the scratch
2765 These patterns need constraints for the reload register and scratch
2767 contain a single register class. If the original reload register (whose
2768 class is @var{class}) can meet the constraint given in the pattern, the
2769 value returned by these macros is used for the class of the scratch
2770 register. Otherwise, two additional reload registers are required.
2771 Their classes are obtained from the constraints in the insn pattern.
2773 @var{x} might be a pseudo-register or a @code{subreg} of a
2774 pseudo-register, which could either be in a hard register or in memory.
2775 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2776 in memory and the hard register number if it is in a register.
2778 These macros should not be used in the case where a particular class of
2779 registers can only be copied to memory and not to another class of
2780 registers. In that case, secondary reload registers are not needed and
2781 would not be helpful. Instead, a stack location must be used to perform
2782 the copy and the @code{mov@var{m}} pattern should use memory as an
2783 intermediate storage. This case often occurs between floating-point and
2787 @defmac SECONDARY_MEMORY_NEEDED (@var{class1}, @var{class2}, @var{m})
2788 Certain machines have the property that some registers cannot be copied
2789 to some other registers without using memory. Define this macro on
2790 those machines to be a C expression that is nonzero if objects of mode
2791 @var{m} in registers of @var{class1} can only be copied to registers of
2792 class @var{class2} by storing a register of @var{class1} into memory
2793 and loading that memory location into a register of @var{class2}.
2795 Do not define this macro if its value would always be zero.
2798 @defmac SECONDARY_MEMORY_NEEDED_RTX (@var{mode})
2799 Normally when @code{SECONDARY_MEMORY_NEEDED} is defined, the compiler
2800 allocates a stack slot for a memory location needed for register copies.
2801 If this macro is defined, the compiler instead uses the memory location
2802 defined by this macro.
2804 Do not define this macro if you do not define
2805 @code{SECONDARY_MEMORY_NEEDED}.
2808 @defmac SECONDARY_MEMORY_NEEDED_MODE (@var{mode})
2809 When the compiler needs a secondary memory location to copy between two
2810 registers of mode @var{mode}, it normally allocates sufficient memory to
2811 hold a quantity of @code{BITS_PER_WORD} bits and performs the store and
2812 load operations in a mode that many bits wide and whose class is the
2813 same as that of @var{mode}.
2815 This is right thing to do on most machines because it ensures that all
2816 bits of the register are copied and prevents accesses to the registers
2817 in a narrower mode, which some machines prohibit for floating-point
2820 However, this default behavior is not correct on some machines, such as
2821 the DEC Alpha, that store short integers in floating-point registers
2822 differently than in integer registers. On those machines, the default
2823 widening will not work correctly and you must define this macro to
2824 suppress that widening in some cases. See the file @file{alpha.h} for
2827 Do not define this macro if you do not define
2828 @code{SECONDARY_MEMORY_NEEDED} or if widening @var{mode} to a mode that
2829 is @code{BITS_PER_WORD} bits wide is correct for your machine.
2832 @defmac SMALL_REGISTER_CLASSES
2833 On some machines, it is risky to let hard registers live across arbitrary
2834 insns. Typically, these machines have instructions that require values
2835 to be in specific registers (like an accumulator), and reload will fail
2836 if the required hard register is used for another purpose across such an
2839 Define @code{SMALL_REGISTER_CLASSES} to be an expression with a nonzero
2840 value on these machines. When this macro has a nonzero value, the
2841 compiler will try to minimize the lifetime of hard registers.
2843 It is always safe to define this macro with a nonzero value, but if you
2844 unnecessarily define it, you will reduce the amount of optimizations
2845 that can be performed in some cases. If you do not define this macro
2846 with a nonzero value when it is required, the compiler will run out of
2847 spill registers and print a fatal error message. For most machines, you
2848 should not define this macro at all.
2851 @defmac CLASS_LIKELY_SPILLED_P (@var{class})
2852 A C expression whose value is nonzero if pseudos that have been assigned
2853 to registers of class @var{class} would likely be spilled because
2854 registers of @var{class} are needed for spill registers.
2856 The default value of this macro returns 1 if @var{class} has exactly one
2857 register and zero otherwise. On most machines, this default should be
2858 used. Only define this macro to some other expression if pseudos
2859 allocated by @file{local-alloc.c} end up in memory because their hard
2860 registers were needed for spill registers. If this macro returns nonzero
2861 for those classes, those pseudos will only be allocated by
2862 @file{global.c}, which knows how to reallocate the pseudo to another
2863 register. If there would not be another register available for
2864 reallocation, you should not change the definition of this macro since
2865 the only effect of such a definition would be to slow down register
2869 @defmac CLASS_MAX_NREGS (@var{class}, @var{mode})
2870 A C expression for the maximum number of consecutive registers
2871 of class @var{class} needed to hold a value of mode @var{mode}.
2873 This is closely related to the macro @code{HARD_REGNO_NREGS}. In fact,
2874 the value of the macro @code{CLASS_MAX_NREGS (@var{class}, @var{mode})}
2875 should be the maximum value of @code{HARD_REGNO_NREGS (@var{regno},
2876 @var{mode})} for all @var{regno} values in the class @var{class}.
2878 This macro helps control the handling of multiple-word values
2882 @defmac CANNOT_CHANGE_MODE_CLASS (@var{from}, @var{to}, @var{class})
2883 If defined, a C expression that returns nonzero for a @var{class} for which
2884 a change from mode @var{from} to mode @var{to} is invalid.
2886 For the example, loading 32-bit integer or floating-point objects into
2887 floating-point registers on the Alpha extends them to 64 bits.
2888 Therefore loading a 64-bit object and then storing it as a 32-bit object
2889 does not store the low-order 32 bits, as would be the case for a normal
2890 register. Therefore, @file{alpha.h} defines @code{CANNOT_CHANGE_MODE_CLASS}
2894 #define CANNOT_CHANGE_MODE_CLASS(FROM, TO, CLASS) \
2895 (GET_MODE_SIZE (FROM) != GET_MODE_SIZE (TO) \
2896 ? reg_classes_intersect_p (FLOAT_REGS, (CLASS)) : 0)
2900 @deftypefn {Target Hook} {const enum reg_class *} TARGET_IRA_COVER_CLASSES ()
2901 Return an array of cover classes for the Integrated Register Allocator
2902 (@acronym{IRA}). Cover classes are a set of non-intersecting register
2903 classes covering all hard registers used for register allocation
2904 purposes. If a move between two registers in the same cover class is
2905 possible, it should be cheaper than a load or store of the registers.
2906 The array is terminated by a @code{LIM_REG_CLASSES} element.
2908 This hook is called once at compiler startup, after the command-line
2909 options have been processed. It is then re-examined by every call to
2910 @code{target_reinit}.
2912 The default implementation returns @code{IRA_COVER_CLASSES}, if defined,
2913 otherwise there is no default implementation. You must define either this
2914 macro or @code{IRA_COVER_CLASSES} in order to use the integrated register
2915 allocator with Chaitin-Briggs coloring. If the macro is not defined,
2916 the only available coloring algorithm is Chow's priority coloring.
2919 @defmac IRA_COVER_CLASSES
2920 See the documentation for @code{TARGET_IRA_COVER_CLASSES}.
2923 @node Old Constraints
2924 @section Obsolete Macros for Defining Constraints
2925 @cindex defining constraints, obsolete method
2926 @cindex constraints, defining, obsolete method
2928 Machine-specific constraints can be defined with these macros instead
2929 of the machine description constructs described in @ref{Define
2930 Constraints}. This mechanism is obsolete. New ports should not use
2931 it; old ports should convert to the new mechanism.
2933 @defmac CONSTRAINT_LEN (@var{char}, @var{str})
2934 For the constraint at the start of @var{str}, which starts with the letter
2935 @var{c}, return the length. This allows you to have register class /
2936 constant / extra constraints that are longer than a single letter;
2937 you don't need to define this macro if you can do with single-letter
2938 constraints only. The definition of this macro should use
2939 DEFAULT_CONSTRAINT_LEN for all the characters that you don't want
2940 to handle specially.
2941 There are some sanity checks in genoutput.c that check the constraint lengths
2942 for the md file, so you can also use this macro to help you while you are
2943 transitioning from a byzantine single-letter-constraint scheme: when you
2944 return a negative length for a constraint you want to re-use, genoutput
2945 will complain about every instance where it is used in the md file.
2948 @defmac REG_CLASS_FROM_LETTER (@var{char})
2949 A C expression which defines the machine-dependent operand constraint
2950 letters for register classes. If @var{char} is such a letter, the
2951 value should be the register class corresponding to it. Otherwise,
2952 the value should be @code{NO_REGS}. The register letter @samp{r},
2953 corresponding to class @code{GENERAL_REGS}, will not be passed
2954 to this macro; you do not need to handle it.
2957 @defmac REG_CLASS_FROM_CONSTRAINT (@var{char}, @var{str})
2958 Like @code{REG_CLASS_FROM_LETTER}, but you also get the constraint string
2959 passed in @var{str}, so that you can use suffixes to distinguish between
2963 @defmac CONST_OK_FOR_LETTER_P (@var{value}, @var{c})
2964 A C expression that defines the machine-dependent operand constraint
2965 letters (@samp{I}, @samp{J}, @samp{K}, @dots{} @samp{P}) that specify
2966 particular ranges of integer values. If @var{c} is one of those
2967 letters, the expression should check that @var{value}, an integer, is in
2968 the appropriate range and return 1 if so, 0 otherwise. If @var{c} is
2969 not one of those letters, the value should be 0 regardless of
2973 @defmac CONST_OK_FOR_CONSTRAINT_P (@var{value}, @var{c}, @var{str})
2974 Like @code{CONST_OK_FOR_LETTER_P}, but you also get the constraint
2975 string passed in @var{str}, so that you can use suffixes to distinguish
2976 between different variants.
2979 @defmac CONST_DOUBLE_OK_FOR_LETTER_P (@var{value}, @var{c})
2980 A C expression that defines the machine-dependent operand constraint
2981 letters that specify particular ranges of @code{const_double} values
2982 (@samp{G} or @samp{H}).
2984 If @var{c} is one of those letters, the expression should check that
2985 @var{value}, an RTX of code @code{const_double}, is in the appropriate
2986 range and return 1 if so, 0 otherwise. If @var{c} is not one of those
2987 letters, the value should be 0 regardless of @var{value}.
2989 @code{const_double} is used for all floating-point constants and for
2990 @code{DImode} fixed-point constants. A given letter can accept either
2991 or both kinds of values. It can use @code{GET_MODE} to distinguish
2992 between these kinds.
2995 @defmac CONST_DOUBLE_OK_FOR_CONSTRAINT_P (@var{value}, @var{c}, @var{str})
2996 Like @code{CONST_DOUBLE_OK_FOR_LETTER_P}, but you also get the constraint
2997 string passed in @var{str}, so that you can use suffixes to distinguish
2998 between different variants.
3001 @defmac EXTRA_CONSTRAINT (@var{value}, @var{c})
3002 A C expression that defines the optional machine-dependent constraint
3003 letters that can be used to segregate specific types of operands, usually
3004 memory references, for the target machine. Any letter that is not
3005 elsewhere defined and not matched by @code{REG_CLASS_FROM_LETTER} /
3006 @code{REG_CLASS_FROM_CONSTRAINT}
3007 may be used. Normally this macro will not be defined.
3009 If it is required for a particular target machine, it should return 1
3010 if @var{value} corresponds to the operand type represented by the
3011 constraint letter @var{c}. If @var{c} is not defined as an extra
3012 constraint, the value returned should be 0 regardless of @var{value}.
3014 For example, on the ROMP, load instructions cannot have their output
3015 in r0 if the memory reference contains a symbolic address. Constraint
3016 letter @samp{Q} is defined as representing a memory address that does
3017 @emph{not} contain a symbolic address. An alternative is specified with
3018 a @samp{Q} constraint on the input and @samp{r} on the output. The next
3019 alternative specifies @samp{m} on the input and a register class that
3020 does not include r0 on the output.
3023 @defmac EXTRA_CONSTRAINT_STR (@var{value}, @var{c}, @var{str})
3024 Like @code{EXTRA_CONSTRAINT}, but you also get the constraint string passed
3025 in @var{str}, so that you can use suffixes to distinguish between different
3029 @defmac EXTRA_MEMORY_CONSTRAINT (@var{c}, @var{str})
3030 A C expression that defines the optional machine-dependent constraint
3031 letters, amongst those accepted by @code{EXTRA_CONSTRAINT}, that should
3032 be treated like memory constraints by the reload pass.
3034 It should return 1 if the operand type represented by the constraint
3035 at the start of @var{str}, the first letter of which is the letter @var{c},
3036 comprises a subset of all memory references including
3037 all those whose address is simply a base register. This allows the reload
3038 pass to reload an operand, if it does not directly correspond to the operand
3039 type of @var{c}, by copying its address into a base register.
3041 For example, on the S/390, some instructions do not accept arbitrary
3042 memory references, but only those that do not make use of an index
3043 register. The constraint letter @samp{Q} is defined via
3044 @code{EXTRA_CONSTRAINT} as representing a memory address of this type.
3045 If the letter @samp{Q} is marked as @code{EXTRA_MEMORY_CONSTRAINT},
3046 a @samp{Q} constraint can handle any memory operand, because the
3047 reload pass knows it can be reloaded by copying the memory address
3048 into a base register if required. This is analogous to the way
3049 an @samp{o} constraint can handle any memory operand.
3052 @defmac EXTRA_ADDRESS_CONSTRAINT (@var{c}, @var{str})
3053 A C expression that defines the optional machine-dependent constraint
3054 letters, amongst those accepted by @code{EXTRA_CONSTRAINT} /
3055 @code{EXTRA_CONSTRAINT_STR}, that should
3056 be treated like address constraints by the reload pass.
3058 It should return 1 if the operand type represented by the constraint
3059 at the start of @var{str}, which starts with the letter @var{c}, comprises
3060 a subset of all memory addresses including
3061 all those that consist of just a base register. This allows the reload
3062 pass to reload an operand, if it does not directly correspond to the operand
3063 type of @var{str}, by copying it into a base register.
3065 Any constraint marked as @code{EXTRA_ADDRESS_CONSTRAINT} can only
3066 be used with the @code{address_operand} predicate. It is treated
3067 analogously to the @samp{p} constraint.
3070 @node Stack and Calling
3071 @section Stack Layout and Calling Conventions
3072 @cindex calling conventions
3074 @c prevent bad page break with this line
3075 This describes the stack layout and calling conventions.
3079 * Exception Handling::
3084 * Register Arguments::
3086 * Aggregate Return::
3091 * Stack Smashing Protection::
3095 @subsection Basic Stack Layout
3096 @cindex stack frame layout
3097 @cindex frame layout
3099 @c prevent bad page break with this line
3100 Here is the basic stack layout.
3102 @defmac STACK_GROWS_DOWNWARD
3103 Define this macro if pushing a word onto the stack moves the stack
3104 pointer to a smaller address.
3106 When we say, ``define this macro if @dots{}'', it means that the
3107 compiler checks this macro only with @code{#ifdef} so the precise
3108 definition used does not matter.
3111 @defmac STACK_PUSH_CODE
3112 This macro defines the operation used when something is pushed
3113 on the stack. In RTL, a push operation will be
3114 @code{(set (mem (STACK_PUSH_CODE (reg sp))) @dots{})}
3116 The choices are @code{PRE_DEC}, @code{POST_DEC}, @code{PRE_INC},
3117 and @code{POST_INC}. Which of these is correct depends on
3118 the stack direction and on whether the stack pointer points
3119 to the last item on the stack or whether it points to the
3120 space for the next item on the stack.
3122 The default is @code{PRE_DEC} when @code{STACK_GROWS_DOWNWARD} is
3123 defined, which is almost always right, and @code{PRE_INC} otherwise,
3124 which is often wrong.
3127 @defmac FRAME_GROWS_DOWNWARD
3128 Define this macro to nonzero value if the addresses of local variable slots
3129 are at negative offsets from the frame pointer.
3132 @defmac ARGS_GROW_DOWNWARD
3133 Define this macro if successive arguments to a function occupy decreasing
3134 addresses on the stack.
3137 @defmac STARTING_FRAME_OFFSET
3138 Offset from the frame pointer to the first local variable slot to be allocated.
3140 If @code{FRAME_GROWS_DOWNWARD}, find the next slot's offset by
3141 subtracting the first slot's length from @code{STARTING_FRAME_OFFSET}.
3142 Otherwise, it is found by adding the length of the first slot to the
3143 value @code{STARTING_FRAME_OFFSET}.
3144 @c i'm not sure if the above is still correct.. had to change it to get
3145 @c rid of an overfull. --mew 2feb93
3148 @defmac STACK_ALIGNMENT_NEEDED
3149 Define to zero to disable final alignment of the stack during reload.
3150 The nonzero default for this macro is suitable for most ports.
3152 On ports where @code{STARTING_FRAME_OFFSET} is nonzero or where there
3153 is a register save block following the local block that doesn't require
3154 alignment to @code{STACK_BOUNDARY}, it may be beneficial to disable
3155 stack alignment and do it in the backend.
3158 @defmac STACK_POINTER_OFFSET
3159 Offset from the stack pointer register to the first location at which
3160 outgoing arguments are placed. If not specified, the default value of
3161 zero is used. This is the proper value for most machines.
3163 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
3164 the first location at which outgoing arguments are placed.
3167 @defmac FIRST_PARM_OFFSET (@var{fundecl})
3168 Offset from the argument pointer register to the first argument's
3169 address. On some machines it may depend on the data type of the
3172 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
3173 the first argument's address.
3176 @defmac STACK_DYNAMIC_OFFSET (@var{fundecl})
3177 Offset from the stack pointer register to an item dynamically allocated
3178 on the stack, e.g., by @code{alloca}.
3180 The default value for this macro is @code{STACK_POINTER_OFFSET} plus the
3181 length of the outgoing arguments. The default is correct for most
3182 machines. See @file{function.c} for details.
3185 @defmac INITIAL_FRAME_ADDRESS_RTX
3186 A C expression whose value is RTL representing the address of the initial
3187 stack frame. This address is passed to @code{RETURN_ADDR_RTX} and
3188 @code{DYNAMIC_CHAIN_ADDRESS}. If you don't define this macro, a reasonable
3189 default value will be used. Define this macro in order to make frame pointer
3190 elimination work in the presence of @code{__builtin_frame_address (count)} and
3191 @code{__builtin_return_address (count)} for @code{count} not equal to zero.
3194 @defmac DYNAMIC_CHAIN_ADDRESS (@var{frameaddr})
3195 A C expression whose value is RTL representing the address in a stack
3196 frame where the pointer to the caller's frame is stored. Assume that
3197 @var{frameaddr} is an RTL expression for the address of the stack frame
3200 If you don't define this macro, the default is to return the value
3201 of @var{frameaddr}---that is, the stack frame address is also the
3202 address of the stack word that points to the previous frame.
3205 @defmac SETUP_FRAME_ADDRESSES
3206 If defined, a C expression that produces the machine-specific code to
3207 setup the stack so that arbitrary frames can be accessed. For example,
3208 on the SPARC, we must flush all of the register windows to the stack
3209 before we can access arbitrary stack frames. You will seldom need to
3213 @deftypefn {Target Hook} bool TARGET_BUILTIN_SETJMP_FRAME_VALUE ()
3214 This target hook should return an rtx that is used to store
3215 the address of the current frame into the built in @code{setjmp} buffer.
3216 The default value, @code{virtual_stack_vars_rtx}, is correct for most
3217 machines. One reason you may need to define this target hook is if
3218 @code{hard_frame_pointer_rtx} is the appropriate value on your machine.
3221 @defmac FRAME_ADDR_RTX (@var{frameaddr})
3222 A C expression whose value is RTL representing the value of the frame
3223 address for the current frame. @var{frameaddr} is the frame pointer
3224 of the current frame. This is used for __builtin_frame_address.
3225 You need only define this macro if the frame address is not the same
3226 as the frame pointer. Most machines do not need to define it.
3229 @defmac RETURN_ADDR_RTX (@var{count}, @var{frameaddr})
3230 A C expression whose value is RTL representing the value of the return
3231 address for the frame @var{count} steps up from the current frame, after
3232 the prologue. @var{frameaddr} is the frame pointer of the @var{count}
3233 frame, or the frame pointer of the @var{count} @minus{} 1 frame if
3234 @code{RETURN_ADDR_IN_PREVIOUS_FRAME} is defined.
3236 The value of the expression must always be the correct address when
3237 @var{count} is zero, but may be @code{NULL_RTX} if there is no way to
3238 determine the return address of other frames.
3241 @defmac RETURN_ADDR_IN_PREVIOUS_FRAME
3242 Define this if the return address of a particular stack frame is accessed
3243 from the frame pointer of the previous stack frame.
3246 @defmac INCOMING_RETURN_ADDR_RTX
3247 A C expression whose value is RTL representing the location of the
3248 incoming return address at the beginning of any function, before the
3249 prologue. This RTL is either a @code{REG}, indicating that the return
3250 value is saved in @samp{REG}, or a @code{MEM} representing a location in
3253 You only need to define this macro if you want to support call frame
3254 debugging information like that provided by DWARF 2.
3256 If this RTL is a @code{REG}, you should also define
3257 @code{DWARF_FRAME_RETURN_COLUMN} to @code{DWARF_FRAME_REGNUM (REGNO)}.
3260 @defmac DWARF_ALT_FRAME_RETURN_COLUMN
3261 A C expression whose value is an integer giving a DWARF 2 column
3262 number that may be used as an alternative return column. The column
3263 must not correspond to any gcc hard register (that is, it must not
3264 be in the range of @code{DWARF_FRAME_REGNUM}).
3266 This macro can be useful if @code{DWARF_FRAME_RETURN_COLUMN} is set to a
3267 general register, but an alternative column needs to be used for signal
3268 frames. Some targets have also used different frame return columns
3272 @defmac DWARF_ZERO_REG
3273 A C expression whose value is an integer giving a DWARF 2 register
3274 number that is considered to always have the value zero. This should
3275 only be defined if the target has an architected zero register, and
3276 someone decided it was a good idea to use that register number to
3277 terminate the stack backtrace. New ports should avoid this.
3280 @deftypefn {Target Hook} void TARGET_DWARF_HANDLE_FRAME_UNSPEC (const char *@var{label}, rtx @var{pattern}, int @var{index})
3281 This target hook allows the backend to emit frame-related insns that
3282 contain UNSPECs or UNSPEC_VOLATILEs. The DWARF 2 call frame debugging
3283 info engine will invoke it on insns of the form
3285 (set (reg) (unspec [@dots{}] UNSPEC_INDEX))
3289 (set (reg) (unspec_volatile [@dots{}] UNSPECV_INDEX)).
3291 to let the backend emit the call frame instructions. @var{label} is
3292 the CFI label attached to the insn, @var{pattern} is the pattern of
3293 the insn and @var{index} is @code{UNSPEC_INDEX} or @code{UNSPECV_INDEX}.
3296 @defmac INCOMING_FRAME_SP_OFFSET
3297 A C expression whose value is an integer giving the offset, in bytes,
3298 from the value of the stack pointer register to the top of the stack
3299 frame at the beginning of any function, before the prologue. The top of
3300 the frame is defined to be the value of the stack pointer in the
3301 previous frame, just before the call instruction.
3303 You only need to define this macro if you want to support call frame
3304 debugging information like that provided by DWARF 2.
3307 @defmac ARG_POINTER_CFA_OFFSET (@var{fundecl})
3308 A C expression whose value is an integer giving the offset, in bytes,
3309 from the argument pointer to the canonical frame address (cfa). The
3310 final value should coincide with that calculated by
3311 @code{INCOMING_FRAME_SP_OFFSET}. Which is unfortunately not usable
3312 during virtual register instantiation.
3314 The default value for this macro is @code{FIRST_PARM_OFFSET (fundecl)},
3315 which is correct for most machines; in general, the arguments are found
3316 immediately before the stack frame. Note that this is not the case on
3317 some targets that save registers into the caller's frame, such as SPARC
3318 and rs6000, and so such targets need to define this macro.
3320 You only need to define this macro if the default is incorrect, and you
3321 want to support call frame debugging information like that provided by
3325 @defmac FRAME_POINTER_CFA_OFFSET (@var{fundecl})
3326 If defined, a C expression whose value is an integer giving the offset
3327 in bytes from the frame pointer to the canonical frame address (cfa).
3328 The final value should coincide with that calculated by
3329 @code{INCOMING_FRAME_SP_OFFSET}.
3331 Normally the CFA is calculated as an offset from the argument pointer,
3332 via @code{ARG_POINTER_CFA_OFFSET}, but if the argument pointer is
3333 variable due to the ABI, this may not be possible. If this macro is
3334 defined, it implies that the virtual register instantiation should be
3335 based on the frame pointer instead of the argument pointer. Only one
3336 of @code{FRAME_POINTER_CFA_OFFSET} and @code{ARG_POINTER_CFA_OFFSET}
3340 @defmac CFA_FRAME_BASE_OFFSET (@var{fundecl})
3341 If defined, a C expression whose value is an integer giving the offset
3342 in bytes from the canonical frame address (cfa) to the frame base used
3343 in DWARF 2 debug information. The default is zero. A different value
3344 may reduce the size of debug information on some ports.
3347 @node Exception Handling
3348 @subsection Exception Handling Support
3349 @cindex exception handling
3351 @defmac EH_RETURN_DATA_REGNO (@var{N})
3352 A C expression whose value is the @var{N}th register number used for
3353 data by exception handlers, or @code{INVALID_REGNUM} if fewer than
3354 @var{N} registers are usable.
3356 The exception handling library routines communicate with the exception
3357 handlers via a set of agreed upon registers. Ideally these registers
3358 should be call-clobbered; it is possible to use call-saved registers,
3359 but may negatively impact code size. The target must support at least
3360 2 data registers, but should define 4 if there are enough free registers.
3362 You must define this macro if you want to support call frame exception
3363 handling like that provided by DWARF 2.
3366 @defmac EH_RETURN_STACKADJ_RTX
3367 A C expression whose value is RTL representing a location in which
3368 to store a stack adjustment to be applied before function return.
3369 This is used to unwind the stack to an exception handler's call frame.
3370 It will be assigned zero on code paths that return normally.
3372 Typically this is a call-clobbered hard register that is otherwise
3373 untouched by the epilogue, but could also be a stack slot.
3375 Do not define this macro if the stack pointer is saved and restored
3376 by the regular prolog and epilog code in the call frame itself; in
3377 this case, the exception handling library routines will update the
3378 stack location to be restored in place. Otherwise, you must define
3379 this macro if you want to support call frame exception handling like
3380 that provided by DWARF 2.
3383 @defmac EH_RETURN_HANDLER_RTX
3384 A C expression whose value is RTL representing a location in which
3385 to store the address of an exception handler to which we should
3386 return. It will not be assigned on code paths that return normally.
3388 Typically this is the location in the call frame at which the normal
3389 return address is stored. For targets that return by popping an
3390 address off the stack, this might be a memory address just below
3391 the @emph{target} call frame rather than inside the current call
3392 frame. If defined, @code{EH_RETURN_STACKADJ_RTX} will have already
3393 been assigned, so it may be used to calculate the location of the
3396 Some targets have more complex requirements than storing to an
3397 address calculable during initial code generation. In that case
3398 the @code{eh_return} instruction pattern should be used instead.
3400 If you want to support call frame exception handling, you must
3401 define either this macro or the @code{eh_return} instruction pattern.
3404 @defmac RETURN_ADDR_OFFSET
3405 If defined, an integer-valued C expression for which rtl will be generated
3406 to add it to the exception handler address before it is searched in the
3407 exception handling tables, and to subtract it again from the address before
3408 using it to return to the exception handler.
3411 @defmac ASM_PREFERRED_EH_DATA_FORMAT (@var{code}, @var{global})
3412 This macro chooses the encoding of pointers embedded in the exception
3413 handling sections. If at all possible, this should be defined such
3414 that the exception handling section will not require dynamic relocations,
3415 and so may be read-only.
3417 @var{code} is 0 for data, 1 for code labels, 2 for function pointers.
3418 @var{global} is true if the symbol may be affected by dynamic relocations.
3419 The macro should return a combination of the @code{DW_EH_PE_*} defines
3420 as found in @file{dwarf2.h}.
3422 If this macro is not defined, pointers will not be encoded but
3423 represented directly.
3426 @defmac ASM_MAYBE_OUTPUT_ENCODED_ADDR_RTX (@var{file}, @var{encoding}, @var{size}, @var{addr}, @var{done})
3427 This macro allows the target to emit whatever special magic is required
3428 to represent the encoding chosen by @code{ASM_PREFERRED_EH_DATA_FORMAT}.
3429 Generic code takes care of pc-relative and indirect encodings; this must
3430 be defined if the target uses text-relative or data-relative encodings.
3432 This is a C statement that branches to @var{done} if the format was
3433 handled. @var{encoding} is the format chosen, @var{size} is the number
3434 of bytes that the format occupies, @var{addr} is the @code{SYMBOL_REF}
3438 @defmac MD_UNWIND_SUPPORT
3439 A string specifying a file to be #include'd in unwind-dw2.c. The file
3440 so included typically defines @code{MD_FALLBACK_FRAME_STATE_FOR}.
3443 @defmac MD_FALLBACK_FRAME_STATE_FOR (@var{context}, @var{fs})
3444 This macro allows the target to add CPU and operating system specific
3445 code to the call-frame unwinder for use when there is no unwind data
3446 available. The most common reason to implement this macro is to unwind
3447 through signal frames.
3449 This macro is called from @code{uw_frame_state_for} in
3450 @file{unwind-dw2.c}, @file{unwind-dw2-xtensa.c} and
3451 @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context};
3452 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{context->ra}
3453 for the address of the code being executed and @code{context->cfa} for
3454 the stack pointer value. If the frame can be decoded, the register
3455 save addresses should be updated in @var{fs} and the macro should
3456 evaluate to @code{_URC_NO_REASON}. If the frame cannot be decoded,
3457 the macro should evaluate to @code{_URC_END_OF_STACK}.
3459 For proper signal handling in Java this macro is accompanied by
3460 @code{MAKE_THROW_FRAME}, defined in @file{libjava/include/*-signal.h} headers.
3463 @defmac MD_HANDLE_UNWABI (@var{context}, @var{fs})
3464 This macro allows the target to add operating system specific code to the
3465 call-frame unwinder to handle the IA-64 @code{.unwabi} unwinding directive,
3466 usually used for signal or interrupt frames.
3468 This macro is called from @code{uw_update_context} in @file{unwind-ia64.c}.
3469 @var{context} is an @code{_Unwind_Context};
3470 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{fs->unwabi}
3471 for the abi and context in the @code{.unwabi} directive. If the
3472 @code{.unwabi} directive can be handled, the register save addresses should
3473 be updated in @var{fs}.
3476 @defmac TARGET_USES_WEAK_UNWIND_INFO
3477 A C expression that evaluates to true if the target requires unwind
3478 info to be given comdat linkage. Define it to be @code{1} if comdat
3479 linkage is necessary. The default is @code{0}.
3482 @node Stack Checking
3483 @subsection Specifying How Stack Checking is Done
3485 GCC will check that stack references are within the boundaries of the
3486 stack, if the option @option{-fstack-check} is specified, in one of
3491 If the value of the @code{STACK_CHECK_BUILTIN} macro is nonzero, GCC
3492 will assume that you have arranged for full stack checking to be done
3493 at appropriate places in the configuration files. GCC will not do
3494 other special processing.
3497 If @code{STACK_CHECK_BUILTIN} is zero and the value of the
3498 @code{STACK_CHECK_STATIC_BUILTIN} macro is nonzero, GCC will assume
3499 that you have arranged for static stack checking (checking of the
3500 static stack frame of functions) to be done at appropriate places
3501 in the configuration files. GCC will only emit code to do dynamic
3502 stack checking (checking on dynamic stack allocations) using the third
3506 If neither of the above are true, GCC will generate code to periodically
3507 ``probe'' the stack pointer using the values of the macros defined below.
3510 If neither STACK_CHECK_BUILTIN nor STACK_CHECK_STATIC_BUILTIN is defined,
3511 GCC will change its allocation strategy for large objects if the option
3512 @option{-fstack-check} is specified: they will always be allocated
3513 dynamically if their size exceeds @code{STACK_CHECK_MAX_VAR_SIZE} bytes.
3515 @defmac STACK_CHECK_BUILTIN
3516 A nonzero value if stack checking is done by the configuration files in a
3517 machine-dependent manner. You should define this macro if stack checking
3518 is require by the ABI of your machine or if you would like to do stack
3519 checking in some more efficient way than the generic approach. The default
3520 value of this macro is zero.
3523 @defmac STACK_CHECK_STATIC_BUILTIN
3524 A nonzero value if static stack checking is done by the configuration files
3525 in a machine-dependent manner. You should define this macro if you would
3526 like to do static stack checking in some more efficient way than the generic
3527 approach. The default value of this macro is zero.
3530 @defmac STACK_CHECK_PROBE_INTERVAL
3531 An integer representing the interval at which GCC must generate stack
3532 probe instructions. You will normally define this macro to be no larger
3533 than the size of the ``guard pages'' at the end of a stack area. The
3534 default value of 4096 is suitable for most systems.
3537 @defmac STACK_CHECK_PROBE_LOAD
3538 An integer which is nonzero if GCC should perform the stack probe
3539 as a load instruction and zero if GCC should use a store instruction.
3540 The default is zero, which is the most efficient choice on most systems.
3543 @defmac STACK_CHECK_PROTECT
3544 The number of bytes of stack needed to recover from a stack overflow,
3545 for languages where such a recovery is supported. The default value of
3546 75 words should be adequate for most machines.
3549 The following macros are relevant only if neither STACK_CHECK_BUILTIN
3550 nor STACK_CHECK_STATIC_BUILTIN is defined; you can omit them altogether
3551 in the opposite case.
3553 @defmac STACK_CHECK_MAX_FRAME_SIZE
3554 The maximum size of a stack frame, in bytes. GCC will generate probe
3555 instructions in non-leaf functions to ensure at least this many bytes of
3556 stack are available. If a stack frame is larger than this size, stack
3557 checking will not be reliable and GCC will issue a warning. The
3558 default is chosen so that GCC only generates one instruction on most
3559 systems. You should normally not change the default value of this macro.
3562 @defmac STACK_CHECK_FIXED_FRAME_SIZE
3563 GCC uses this value to generate the above warning message. It
3564 represents the amount of fixed frame used by a function, not including
3565 space for any callee-saved registers, temporaries and user variables.
3566 You need only specify an upper bound for this amount and will normally
3567 use the default of four words.
3570 @defmac STACK_CHECK_MAX_VAR_SIZE
3571 The maximum size, in bytes, of an object that GCC will place in the
3572 fixed area of the stack frame when the user specifies
3573 @option{-fstack-check}.
3574 GCC computed the default from the values of the above macros and you will
3575 normally not need to override that default.
3579 @node Frame Registers
3580 @subsection Registers That Address the Stack Frame
3582 @c prevent bad page break with this line
3583 This discusses registers that address the stack frame.
3585 @defmac STACK_POINTER_REGNUM
3586 The register number of the stack pointer register, which must also be a
3587 fixed register according to @code{FIXED_REGISTERS}. On most machines,
3588 the hardware determines which register this is.
3591 @defmac FRAME_POINTER_REGNUM
3592 The register number of the frame pointer register, which is used to
3593 access automatic variables in the stack frame. On some machines, the
3594 hardware determines which register this is. On other machines, you can
3595 choose any register you wish for this purpose.
3598 @defmac HARD_FRAME_POINTER_REGNUM
3599 On some machines the offset between the frame pointer and starting
3600 offset of the automatic variables is not known until after register
3601 allocation has been done (for example, because the saved registers are
3602 between these two locations). On those machines, define
3603 @code{FRAME_POINTER_REGNUM} the number of a special, fixed register to
3604 be used internally until the offset is known, and define
3605 @code{HARD_FRAME_POINTER_REGNUM} to be the actual hard register number
3606 used for the frame pointer.
3608 You should define this macro only in the very rare circumstances when it
3609 is not possible to calculate the offset between the frame pointer and
3610 the automatic variables until after register allocation has been
3611 completed. When this macro is defined, you must also indicate in your
3612 definition of @code{ELIMINABLE_REGS} how to eliminate
3613 @code{FRAME_POINTER_REGNUM} into either @code{HARD_FRAME_POINTER_REGNUM}
3614 or @code{STACK_POINTER_REGNUM}.
3616 Do not define this macro if it would be the same as
3617 @code{FRAME_POINTER_REGNUM}.
3620 @defmac ARG_POINTER_REGNUM
3621 The register number of the arg pointer register, which is used to access
3622 the function's argument list. On some machines, this is the same as the
3623 frame pointer register. On some machines, the hardware determines which
3624 register this is. On other machines, you can choose any register you
3625 wish for this purpose. If this is not the same register as the frame
3626 pointer register, then you must mark it as a fixed register according to
3627 @code{FIXED_REGISTERS}, or arrange to be able to eliminate it
3628 (@pxref{Elimination}).
3631 @defmac RETURN_ADDRESS_POINTER_REGNUM
3632 The register number of the return address pointer register, which is used to
3633 access the current function's return address from the stack. On some
3634 machines, the return address is not at a fixed offset from the frame
3635 pointer or stack pointer or argument pointer. This register can be defined
3636 to point to the return address on the stack, and then be converted by
3637 @code{ELIMINABLE_REGS} into either the frame pointer or stack pointer.
3639 Do not define this macro unless there is no other way to get the return
3640 address from the stack.
3643 @defmac STATIC_CHAIN_REGNUM
3644 @defmacx STATIC_CHAIN_INCOMING_REGNUM
3645 Register numbers used for passing a function's static chain pointer. If
3646 register windows are used, the register number as seen by the called
3647 function is @code{STATIC_CHAIN_INCOMING_REGNUM}, while the register
3648 number as seen by the calling function is @code{STATIC_CHAIN_REGNUM}. If
3649 these registers are the same, @code{STATIC_CHAIN_INCOMING_REGNUM} need
3652 The static chain register need not be a fixed register.
3654 If the static chain is passed in memory, these macros should not be
3655 defined; instead, the next two macros should be defined.
3658 @defmac STATIC_CHAIN
3659 @defmacx STATIC_CHAIN_INCOMING
3660 If the static chain is passed in memory, these macros provide rtx giving
3661 @code{mem} expressions that denote where they are stored.
3662 @code{STATIC_CHAIN} and @code{STATIC_CHAIN_INCOMING} give the locations
3663 as seen by the calling and called functions, respectively. Often the former
3664 will be at an offset from the stack pointer and the latter at an offset from
3667 @findex stack_pointer_rtx
3668 @findex frame_pointer_rtx
3669 @findex arg_pointer_rtx
3670 The variables @code{stack_pointer_rtx}, @code{frame_pointer_rtx}, and
3671 @code{arg_pointer_rtx} will have been initialized prior to the use of these
3672 macros and should be used to refer to those items.
3674 If the static chain is passed in a register, the two previous macros should
3678 @defmac DWARF_FRAME_REGISTERS
3679 This macro specifies the maximum number of hard registers that can be
3680 saved in a call frame. This is used to size data structures used in
3681 DWARF2 exception handling.
3683 Prior to GCC 3.0, this macro was needed in order to establish a stable
3684 exception handling ABI in the face of adding new hard registers for ISA
3685 extensions. In GCC 3.0 and later, the EH ABI is insulated from changes
3686 in the number of hard registers. Nevertheless, this macro can still be
3687 used to reduce the runtime memory requirements of the exception handling
3688 routines, which can be substantial if the ISA contains a lot of
3689 registers that are not call-saved.
3691 If this macro is not defined, it defaults to
3692 @code{FIRST_PSEUDO_REGISTER}.
3695 @defmac PRE_GCC3_DWARF_FRAME_REGISTERS
3697 This macro is similar to @code{DWARF_FRAME_REGISTERS}, but is provided
3698 for backward compatibility in pre GCC 3.0 compiled code.
3700 If this macro is not defined, it defaults to
3701 @code{DWARF_FRAME_REGISTERS}.
3704 @defmac DWARF_REG_TO_UNWIND_COLUMN (@var{regno})
3706 Define this macro if the target's representation for dwarf registers
3707 is different than the internal representation for unwind column.
3708 Given a dwarf register, this macro should return the internal unwind
3709 column number to use instead.
3711 See the PowerPC's SPE target for an example.
3714 @defmac DWARF_FRAME_REGNUM (@var{regno})
3716 Define this macro if the target's representation for dwarf registers
3717 used in .eh_frame or .debug_frame is different from that used in other
3718 debug info sections. Given a GCC hard register number, this macro
3719 should return the .eh_frame register number. The default is
3720 @code{DBX_REGISTER_NUMBER (@var{regno})}.
3724 @defmac DWARF2_FRAME_REG_OUT (@var{regno}, @var{for_eh})
3726 Define this macro to map register numbers held in the call frame info
3727 that GCC has collected using @code{DWARF_FRAME_REGNUM} to those that
3728 should be output in .debug_frame (@code{@var{for_eh}} is zero) and
3729 .eh_frame (@code{@var{for_eh}} is nonzero). The default is to
3730 return @code{@var{regno}}.
3735 @subsection Eliminating Frame Pointer and Arg Pointer
3737 @c prevent bad page break with this line
3738 This is about eliminating the frame pointer and arg pointer.
3740 @defmac FRAME_POINTER_REQUIRED
3741 A C expression which is @code{true} if a function must have and use a frame
3742 pointer. This expression is evaluated in the reload pass. If its value is
3743 @code{true} the function will have a frame pointer.
3745 The expression can in principle examine the current function and decide
3746 according to the facts, but on most machines the constant @code{false} or the
3747 constant @code{true} suffices. Use @code{false} when the machine allows code
3748 to be generated with no frame pointer, and doing so saves some time or space.
3749 Use @code{true} when there is no possible advantage to avoiding a frame
3752 In certain cases, the compiler does not know how to produce valid code
3753 without a frame pointer. The compiler recognizes those cases and
3754 automatically gives the function a frame pointer regardless of what
3755 @code{FRAME_POINTER_REQUIRED} says. You don't need to worry about
3758 In a function that does not require a frame pointer, the frame pointer
3759 register can be allocated for ordinary usage, unless you mark it as a
3760 fixed register. See @code{FIXED_REGISTERS} for more information.
3762 Default value is @code{false}.
3765 @findex get_frame_size
3766 @defmac INITIAL_FRAME_POINTER_OFFSET (@var{depth-var})
3767 A C statement to store in the variable @var{depth-var} the difference
3768 between the frame pointer and the stack pointer values immediately after
3769 the function prologue. The value would be computed from information
3770 such as the result of @code{get_frame_size ()} and the tables of
3771 registers @code{regs_ever_live} and @code{call_used_regs}.
3773 If @code{ELIMINABLE_REGS} is defined, this macro will be not be used and
3774 need not be defined. Otherwise, it must be defined even if
3775 @code{FRAME_POINTER_REQUIRED} is defined to always be true; in that
3776 case, you may set @var{depth-var} to anything.
3779 @defmac ELIMINABLE_REGS
3780 If defined, this macro specifies a table of register pairs used to
3781 eliminate unneeded registers that point into the stack frame. If it is not
3782 defined, the only elimination attempted by the compiler is to replace
3783 references to the frame pointer with references to the stack pointer.
3785 The definition of this macro is a list of structure initializations, each
3786 of which specifies an original and replacement register.
3788 On some machines, the position of the argument pointer is not known until
3789 the compilation is completed. In such a case, a separate hard register
3790 must be used for the argument pointer. This register can be eliminated by
3791 replacing it with either the frame pointer or the argument pointer,
3792 depending on whether or not the frame pointer has been eliminated.
3794 In this case, you might specify:
3796 #define ELIMINABLE_REGS \
3797 @{@{ARG_POINTER_REGNUM, STACK_POINTER_REGNUM@}, \
3798 @{ARG_POINTER_REGNUM, FRAME_POINTER_REGNUM@}, \
3799 @{FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM@}@}
3802 Note that the elimination of the argument pointer with the stack pointer is
3803 specified first since that is the preferred elimination.
3806 @defmac CAN_ELIMINATE (@var{from-reg}, @var{to-reg})
3807 A C expression that returns nonzero if the compiler is allowed to try
3808 to replace register number @var{from-reg} with register number
3809 @var{to-reg}. This macro need only be defined if @code{ELIMINABLE_REGS}
3810 is defined, and will usually be the constant 1, since most of the cases
3811 preventing register elimination are things that the compiler already
3815 @defmac INITIAL_ELIMINATION_OFFSET (@var{from-reg}, @var{to-reg}, @var{offset-var})
3816 This macro is similar to @code{INITIAL_FRAME_POINTER_OFFSET}. It
3817 specifies the initial difference between the specified pair of
3818 registers. This macro must be defined if @code{ELIMINABLE_REGS} is
3822 @node Stack Arguments
3823 @subsection Passing Function Arguments on the Stack
3824 @cindex arguments on stack
3825 @cindex stack arguments
3827 The macros in this section control how arguments are passed
3828 on the stack. See the following section for other macros that
3829 control passing certain arguments in registers.
3831 @deftypefn {Target Hook} bool TARGET_PROMOTE_PROTOTYPES (tree @var{fntype})
3832 This target hook returns @code{true} if an argument declared in a
3833 prototype as an integral type smaller than @code{int} should actually be
3834 passed as an @code{int}. In addition to avoiding errors in certain
3835 cases of mismatch, it also makes for better code on certain machines.
3836 The default is to not promote prototypes.
3840 A C expression. If nonzero, push insns will be used to pass
3842 If the target machine does not have a push instruction, set it to zero.
3843 That directs GCC to use an alternate strategy: to
3844 allocate the entire argument block and then store the arguments into
3845 it. When @code{PUSH_ARGS} is nonzero, @code{PUSH_ROUNDING} must be defined too.
3848 @defmac PUSH_ARGS_REVERSED
3849 A C expression. If nonzero, function arguments will be evaluated from
3850 last to first, rather than from first to last. If this macro is not
3851 defined, it defaults to @code{PUSH_ARGS} on targets where the stack
3852 and args grow in opposite directions, and 0 otherwise.
3855 @defmac PUSH_ROUNDING (@var{npushed})
3856 A C expression that is the number of bytes actually pushed onto the
3857 stack when an instruction attempts to push @var{npushed} bytes.
3859 On some machines, the definition
3862 #define PUSH_ROUNDING(BYTES) (BYTES)
3866 will suffice. But on other machines, instructions that appear
3867 to push one byte actually push two bytes in an attempt to maintain
3868 alignment. Then the definition should be
3871 #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1)
3875 @findex current_function_outgoing_args_size
3876 @defmac ACCUMULATE_OUTGOING_ARGS
3877 A C expression. If nonzero, the maximum amount of space required for outgoing arguments
3878 will be computed and placed into the variable
3879 @code{current_function_outgoing_args_size}. No space will be pushed
3880 onto the stack for each call; instead, the function prologue should
3881 increase the stack frame size by this amount.
3883 Setting both @code{PUSH_ARGS} and @code{ACCUMULATE_OUTGOING_ARGS}
3887 @defmac REG_PARM_STACK_SPACE (@var{fndecl})
3888 Define this macro if functions should assume that stack space has been
3889 allocated for arguments even when their values are passed in
3892 The value of this macro is the size, in bytes, of the area reserved for
3893 arguments passed in registers for the function represented by @var{fndecl},
3894 which can be zero if GCC is calling a library function.
3895 The argument @var{fndecl} can be the FUNCTION_DECL, or the type itself
3898 This space can be allocated by the caller, or be a part of the
3899 machine-dependent stack frame: @code{OUTGOING_REG_PARM_STACK_SPACE} says
3902 @c above is overfull. not sure what to do. --mew 5feb93 did
3903 @c something, not sure if it looks good. --mew 10feb93
3905 @defmac OUTGOING_REG_PARM_STACK_SPACE (@var{fntype})
3906 Define this to a nonzero value if it is the responsibility of the
3907 caller to allocate the area reserved for arguments passed in registers
3908 when calling a function of @var{fntype}. @var{fntype} may be NULL
3909 if the function called is a library function.
3911 If @code{ACCUMULATE_OUTGOING_ARGS} is defined, this macro controls
3912 whether the space for these arguments counts in the value of
3913 @code{current_function_outgoing_args_size}.
3916 @defmac STACK_PARMS_IN_REG_PARM_AREA
3917 Define this macro if @code{REG_PARM_STACK_SPACE} is defined, but the
3918 stack parameters don't skip the area specified by it.
3919 @c i changed this, makes more sens and it should have taken care of the
3920 @c overfull.. not as specific, tho. --mew 5feb93
3922 Normally, when a parameter is not passed in registers, it is placed on the
3923 stack beyond the @code{REG_PARM_STACK_SPACE} area. Defining this macro
3924 suppresses this behavior and causes the parameter to be passed on the
3925 stack in its natural location.
3928 @defmac RETURN_POPS_ARGS (@var{fundecl}, @var{funtype}, @var{stack-size})
3929 A C expression that should indicate the number of bytes of its own
3930 arguments that a function pops on returning, or 0 if the
3931 function pops no arguments and the caller must therefore pop them all
3932 after the function returns.
3934 @var{fundecl} is a C variable whose value is a tree node that describes
3935 the function in question. Normally it is a node of type
3936 @code{FUNCTION_DECL} that describes the declaration of the function.
3937 From this you can obtain the @code{DECL_ATTRIBUTES} of the function.
3939 @var{funtype} is a C variable whose value is a tree node that
3940 describes the function in question. Normally it is a node of type
3941 @code{FUNCTION_TYPE} that describes the data type of the function.
3942 From this it is possible to obtain the data types of the value and
3943 arguments (if known).
3945 When a call to a library function is being considered, @var{fundecl}
3946 will contain an identifier node for the library function. Thus, if
3947 you need to distinguish among various library functions, you can do so
3948 by their names. Note that ``library function'' in this context means
3949 a function used to perform arithmetic, whose name is known specially
3950 in the compiler and was not mentioned in the C code being compiled.
3952 @var{stack-size} is the number of bytes of arguments passed on the
3953 stack. If a variable number of bytes is passed, it is zero, and
3954 argument popping will always be the responsibility of the calling function.
3956 On the VAX, all functions always pop their arguments, so the definition
3957 of this macro is @var{stack-size}. On the 68000, using the standard
3958 calling convention, no functions pop their arguments, so the value of
3959 the macro is always 0 in this case. But an alternative calling
3960 convention is available in which functions that take a fixed number of
3961 arguments pop them but other functions (such as @code{printf}) pop
3962 nothing (the caller pops all). When this convention is in use,
3963 @var{funtype} is examined to determine whether a function takes a fixed
3964 number of arguments.
3967 @defmac CALL_POPS_ARGS (@var{cum})
3968 A C expression that should indicate the number of bytes a call sequence
3969 pops off the stack. It is added to the value of @code{RETURN_POPS_ARGS}
3970 when compiling a function call.
3972 @var{cum} is the variable in which all arguments to the called function
3973 have been accumulated.
3975 On certain architectures, such as the SH5, a call trampoline is used
3976 that pops certain registers off the stack, depending on the arguments
3977 that have been passed to the function. Since this is a property of the
3978 call site, not of the called function, @code{RETURN_POPS_ARGS} is not
3982 @node Register Arguments
3983 @subsection Passing Arguments in Registers
3984 @cindex arguments in registers
3985 @cindex registers arguments
3987 This section describes the macros which let you control how various
3988 types of arguments are passed in registers or how they are arranged in
3991 @defmac FUNCTION_ARG (@var{cum}, @var{mode}, @var{type}, @var{named})
3992 A C expression that controls whether a function argument is passed
3993 in a register, and which register.
3995 The arguments are @var{cum}, which summarizes all the previous
3996 arguments; @var{mode}, the machine mode of the argument; @var{type},
3997 the data type of the argument as a tree node or 0 if that is not known
3998 (which happens for C support library functions); and @var{named},
3999 which is 1 for an ordinary argument and 0 for nameless arguments that
4000 correspond to @samp{@dots{}} in the called function's prototype.
4001 @var{type} can be an incomplete type if a syntax error has previously
4004 The value of the expression is usually either a @code{reg} RTX for the
4005 hard register in which to pass the argument, or zero to pass the
4006 argument on the stack.
4008 For machines like the VAX and 68000, where normally all arguments are
4009 pushed, zero suffices as a definition.
4011 The value of the expression can also be a @code{parallel} RTX@. This is
4012 used when an argument is passed in multiple locations. The mode of the
4013 @code{parallel} should be the mode of the entire argument. The
4014 @code{parallel} holds any number of @code{expr_list} pairs; each one
4015 describes where part of the argument is passed. In each
4016 @code{expr_list} the first operand must be a @code{reg} RTX for the hard
4017 register in which to pass this part of the argument, and the mode of the
4018 register RTX indicates how large this part of the argument is. The
4019 second operand of the @code{expr_list} is a @code{const_int} which gives
4020 the offset in bytes into the entire argument of where this part starts.
4021 As a special exception the first @code{expr_list} in the @code{parallel}
4022 RTX may have a first operand of zero. This indicates that the entire
4023 argument is also stored on the stack.
4025 The last time this macro is called, it is called with @code{MODE ==
4026 VOIDmode}, and its result is passed to the @code{call} or @code{call_value}
4027 pattern as operands 2 and 3 respectively.
4029 @cindex @file{stdarg.h} and register arguments
4030 The usual way to make the ISO library @file{stdarg.h} work on a machine
4031 where some arguments are usually passed in registers, is to cause
4032 nameless arguments to be passed on the stack instead. This is done
4033 by making @code{FUNCTION_ARG} return 0 whenever @var{named} is 0.
4035 @cindex @code{TARGET_MUST_PASS_IN_STACK}, and @code{FUNCTION_ARG}
4036 @cindex @code{REG_PARM_STACK_SPACE}, and @code{FUNCTION_ARG}
4037 You may use the hook @code{targetm.calls.must_pass_in_stack}
4038 in the definition of this macro to determine if this argument is of a
4039 type that must be passed in the stack. If @code{REG_PARM_STACK_SPACE}
4040 is not defined and @code{FUNCTION_ARG} returns nonzero for such an
4041 argument, the compiler will abort. If @code{REG_PARM_STACK_SPACE} is
4042 defined, the argument will be computed in the stack and then loaded into
4046 @deftypefn {Target Hook} bool TARGET_MUST_PASS_IN_STACK (enum machine_mode @var{mode}, tree @var{type})
4047 This target hook should return @code{true} if we should not pass @var{type}
4048 solely in registers. The file @file{expr.h} defines a
4049 definition that is usually appropriate, refer to @file{expr.h} for additional
4053 @defmac FUNCTION_INCOMING_ARG (@var{cum}, @var{mode}, @var{type}, @var{named})
4054 Define this macro if the target machine has ``register windows'', so
4055 that the register in which a function sees an arguments is not
4056 necessarily the same as the one in which the caller passed the
4059 For such machines, @code{FUNCTION_ARG} computes the register in which
4060 the caller passes the value, and @code{FUNCTION_INCOMING_ARG} should
4061 be defined in a similar fashion to tell the function being called
4062 where the arguments will arrive.
4064 If @code{FUNCTION_INCOMING_ARG} is not defined, @code{FUNCTION_ARG}
4065 serves both purposes.
4068 @deftypefn {Target Hook} int TARGET_ARG_PARTIAL_BYTES (CUMULATIVE_ARGS *@var{cum}, enum machine_mode @var{mode}, tree @var{type}, bool @var{named})
4069 This target hook returns the number of bytes at the beginning of an
4070 argument that must be put in registers. The value must be zero for
4071 arguments that are passed entirely in registers or that are entirely
4072 pushed on the stack.
4074 On some machines, certain arguments must be passed partially in
4075 registers and partially in memory. On these machines, typically the
4076 first few words of arguments are passed in registers, and the rest
4077 on the stack. If a multi-word argument (a @code{double} or a
4078 structure) crosses that boundary, its first few words must be passed
4079 in registers and the rest must be pushed. This macro tells the
4080 compiler when this occurs, and how many bytes should go in registers.
4082 @code{FUNCTION_ARG} for these arguments should return the first
4083 register to be used by the caller for this argument; likewise
4084 @code{FUNCTION_INCOMING_ARG}, for the called function.
4087 @deftypefn {Target Hook} bool TARGET_PASS_BY_REFERENCE (CUMULATIVE_ARGS *@var{cum}, enum machine_mode @var{mode}, tree @var{type}, bool @var{named})
4088 This target hook should return @code{true} if an argument at the
4089 position indicated by @var{cum} should be passed by reference. This
4090 predicate is queried after target independent reasons for being
4091 passed by reference, such as @code{TREE_ADDRESSABLE (type)}.
4093 If the hook returns true, a copy of that argument is made in memory and a
4094 pointer to the argument is passed instead of the argument itself.
4095 The pointer is passed in whatever way is appropriate for passing a pointer
4099 @deftypefn {Target Hook} bool TARGET_CALLEE_COPIES (CUMULATIVE_ARGS *@var{cum}, enum machine_mode @var{mode}, tree @var{type}, bool @var{named})
4100 The function argument described by the parameters to this hook is
4101 known to be passed by reference. The hook should return true if the
4102 function argument should be copied by the callee instead of copied
4105 For any argument for which the hook returns true, if it can be
4106 determined that the argument is not modified, then a copy need
4109 The default version of this hook always returns false.
4112 @defmac CUMULATIVE_ARGS
4113 A C type for declaring a variable that is used as the first argument of
4114 @code{FUNCTION_ARG} and other related values. For some target machines,
4115 the type @code{int} suffices and can hold the number of bytes of
4118 There is no need to record in @code{CUMULATIVE_ARGS} anything about the
4119 arguments that have been passed on the stack. The compiler has other
4120 variables to keep track of that. For target machines on which all
4121 arguments are passed on the stack, there is no need to store anything in
4122 @code{CUMULATIVE_ARGS}; however, the data structure must exist and
4123 should not be empty, so use @code{int}.
4126 @defmac OVERRIDE_ABI_FORMAT (@var{fndecl})
4127 If defined, this macro is called before generating any code for a
4128 function, but after the @var{cfun} descriptor for the function has been
4129 created. The back end may use this macro to update @var{cfun} to
4130 reflect an ABI other than that which would normally be used by default.
4131 If the compiler is generating code for a compiler-generated function,
4132 @var{fndecl} may be @code{NULL}.
4135 @defmac INIT_CUMULATIVE_ARGS (@var{cum}, @var{fntype}, @var{libname}, @var{fndecl}, @var{n_named_args})
4136 A C statement (sans semicolon) for initializing the variable
4137 @var{cum} for the state at the beginning of the argument list. The
4138 variable has type @code{CUMULATIVE_ARGS}. The value of @var{fntype}
4139 is the tree node for the data type of the function which will receive
4140 the args, or 0 if the args are to a compiler support library function.
4141 For direct calls that are not libcalls, @var{fndecl} contain the
4142 declaration node of the function. @var{fndecl} is also set when
4143 @code{INIT_CUMULATIVE_ARGS} is used to find arguments for the function
4144 being compiled. @var{n_named_args} is set to the number of named
4145 arguments, including a structure return address if it is passed as a
4146 parameter, when making a call. When processing incoming arguments,
4147 @var{n_named_args} is set to @minus{}1.
4149 When processing a call to a compiler support library function,
4150 @var{libname} identifies which one. It is a @code{symbol_ref} rtx which
4151 contains the name of the function, as a string. @var{libname} is 0 when
4152 an ordinary C function call is being processed. Thus, each time this
4153 macro is called, either @var{libname} or @var{fntype} is nonzero, but
4154 never both of them at once.
4157 @defmac INIT_CUMULATIVE_LIBCALL_ARGS (@var{cum}, @var{mode}, @var{libname})
4158 Like @code{INIT_CUMULATIVE_ARGS} but only used for outgoing libcalls,
4159 it gets a @code{MODE} argument instead of @var{fntype}, that would be
4160 @code{NULL}. @var{indirect} would always be zero, too. If this macro
4161 is not defined, @code{INIT_CUMULATIVE_ARGS (cum, NULL_RTX, libname,
4162 0)} is used instead.
4165 @defmac INIT_CUMULATIVE_INCOMING_ARGS (@var{cum}, @var{fntype}, @var{libname})
4166 Like @code{INIT_CUMULATIVE_ARGS} but overrides it for the purposes of
4167 finding the arguments for the function being compiled. If this macro is
4168 undefined, @code{INIT_CUMULATIVE_ARGS} is used instead.
4170 The value passed for @var{libname} is always 0, since library routines
4171 with special calling conventions are never compiled with GCC@. The
4172 argument @var{libname} exists for symmetry with
4173 @code{INIT_CUMULATIVE_ARGS}.
4174 @c could use "this macro" in place of @code{INIT_CUMULATIVE_ARGS}, maybe.
4175 @c --mew 5feb93 i switched the order of the sentences. --mew 10feb93
4178 @defmac FUNCTION_ARG_ADVANCE (@var{cum}, @var{mode}, @var{type}, @var{named})
4179 A C statement (sans semicolon) to update the summarizer variable
4180 @var{cum} to advance past an argument in the argument list. The
4181 values @var{mode}, @var{type} and @var{named} describe that argument.
4182 Once this is done, the variable @var{cum} is suitable for analyzing
4183 the @emph{following} argument with @code{FUNCTION_ARG}, etc.
4185 This macro need not do anything if the argument in question was passed
4186 on the stack. The compiler knows how to track the amount of stack space
4187 used for arguments without any special help.
4191 @defmac FUNCTION_ARG_OFFSET (@var{mode}, @var{type})
4192 If defined, a C expression that is the number of bytes to add to the
4193 offset of the argument passed in memory. This is needed for the SPU,
4194 which passes @code{char} and @code{short} arguments in the preferred
4195 slot that is in the middle of the quad word instead of starting at the
4199 @defmac FUNCTION_ARG_PADDING (@var{mode}, @var{type})
4200 If defined, a C expression which determines whether, and in which direction,
4201 to pad out an argument with extra space. The value should be of type
4202 @code{enum direction}: either @code{upward} to pad above the argument,
4203 @code{downward} to pad below, or @code{none} to inhibit padding.
4205 The @emph{amount} of padding is always just enough to reach the next
4206 multiple of @code{FUNCTION_ARG_BOUNDARY}; this macro does not control
4209 This macro has a default definition which is right for most systems.
4210 For little-endian machines, the default is to pad upward. For
4211 big-endian machines, the default is to pad downward for an argument of
4212 constant size shorter than an @code{int}, and upward otherwise.
4215 @defmac PAD_VARARGS_DOWN
4216 If defined, a C expression which determines whether the default
4217 implementation of va_arg will attempt to pad down before reading the
4218 next argument, if that argument is smaller than its aligned space as
4219 controlled by @code{PARM_BOUNDARY}. If this macro is not defined, all such
4220 arguments are padded down if @code{BYTES_BIG_ENDIAN} is true.
4223 @defmac BLOCK_REG_PADDING (@var{mode}, @var{type}, @var{first})
4224 Specify padding for the last element of a block move between registers and
4225 memory. @var{first} is nonzero if this is the only element. Defining this
4226 macro allows better control of register function parameters on big-endian
4227 machines, without using @code{PARALLEL} rtl. In particular,
4228 @code{MUST_PASS_IN_STACK} need not test padding and mode of types in
4229 registers, as there is no longer a "wrong" part of a register; For example,
4230 a three byte aggregate may be passed in the high part of a register if so
4234 @defmac FUNCTION_ARG_BOUNDARY (@var{mode}, @var{type})
4235 If defined, a C expression that gives the alignment boundary, in bits,
4236 of an argument with the specified mode and type. If it is not defined,
4237 @code{PARM_BOUNDARY} is used for all arguments.
4240 @defmac FUNCTION_ARG_REGNO_P (@var{regno})
4241 A C expression that is nonzero if @var{regno} is the number of a hard
4242 register in which function arguments are sometimes passed. This does
4243 @emph{not} include implicit arguments such as the static chain and
4244 the structure-value address. On many machines, no registers can be
4245 used for this purpose since all function arguments are pushed on the
4249 @deftypefn {Target Hook} bool TARGET_SPLIT_COMPLEX_ARG (tree @var{type})
4250 This hook should return true if parameter of type @var{type} are passed
4251 as two scalar parameters. By default, GCC will attempt to pack complex
4252 arguments into the target's word size. Some ABIs require complex arguments
4253 to be split and treated as their individual components. For example, on
4254 AIX64, complex floats should be passed in a pair of floating point
4255 registers, even though a complex float would fit in one 64-bit floating
4258 The default value of this hook is @code{NULL}, which is treated as always
4262 @deftypefn {Target Hook} tree TARGET_BUILD_BUILTIN_VA_LIST (void)
4263 This hook returns a type node for @code{va_list} for the target.
4264 The default version of the hook returns @code{void*}.
4267 @deftypefn {Target Hook} tree TARGET_FN_ABI_VA_LIST (tree @var{fndecl})
4268 This hook returns the va_list type of the calling convention specified by
4270 The default version of this hook returns @code{va_list_type_node}.
4273 @deftypefn {Target Hook} tree TARGET_CANONICAL_VA_LIST_TYPE (tree @var{type})
4274 This hook returns the va_list type of the calling convention specified by the
4275 type of @var{type}. If @var{type} is not a valid va_list type, it returns
4279 @deftypefn {Target Hook} tree TARGET_GIMPLIFY_VA_ARG_EXPR (tree @var{valist}, tree @var{type}, tree *@var{pre_p}, tree *@var{post_p})
4280 This hook performs target-specific gimplification of
4281 @code{VA_ARG_EXPR}. The first two parameters correspond to the
4282 arguments to @code{va_arg}; the latter two are as in
4283 @code{gimplify.c:gimplify_expr}.
4286 @deftypefn {Target Hook} bool TARGET_VALID_POINTER_MODE (enum machine_mode @var{mode})
4287 Define this to return nonzero if the port can handle pointers
4288 with machine mode @var{mode}. The default version of this
4289 hook returns true for both @code{ptr_mode} and @code{Pmode}.
4292 @deftypefn {Target Hook} bool TARGET_SCALAR_MODE_SUPPORTED_P (enum machine_mode @var{mode})
4293 Define this to return nonzero if the port is prepared to handle
4294 insns involving scalar mode @var{mode}. For a scalar mode to be
4295 considered supported, all the basic arithmetic and comparisons
4298 The default version of this hook returns true for any mode
4299 required to handle the basic C types (as defined by the port).
4300 Included here are the double-word arithmetic supported by the
4301 code in @file{optabs.c}.
4304 @deftypefn {Target Hook} bool TARGET_VECTOR_MODE_SUPPORTED_P (enum machine_mode @var{mode})
4305 Define this to return nonzero if the port is prepared to handle
4306 insns involving vector mode @var{mode}. At the very least, it
4307 must have move patterns for this mode.
4311 @subsection How Scalar Function Values Are Returned
4312 @cindex return values in registers
4313 @cindex values, returned by functions
4314 @cindex scalars, returned as values
4316 This section discusses the macros that control returning scalars as
4317 values---values that can fit in registers.
4319 @deftypefn {Target Hook} rtx TARGET_FUNCTION_VALUE (tree @var{ret_type}, tree @var{fn_decl_or_type}, bool @var{outgoing})
4321 Define this to return an RTX representing the place where a function
4322 returns or receives a value of data type @var{ret_type}, a tree node
4323 representing a data type. @var{fn_decl_or_type} is a tree node
4324 representing @code{FUNCTION_DECL} or @code{FUNCTION_TYPE} of a
4325 function being called. If @var{outgoing} is false, the hook should
4326 compute the register in which the caller will see the return value.
4327 Otherwise, the hook should return an RTX representing the place where
4328 a function returns a value.
4330 On many machines, only @code{TYPE_MODE (@var{ret_type})} is relevant.
4331 (Actually, on most machines, scalar values are returned in the same
4332 place regardless of mode.) The value of the expression is usually a
4333 @code{reg} RTX for the hard register where the return value is stored.
4334 The value can also be a @code{parallel} RTX, if the return value is in
4335 multiple places. See @code{FUNCTION_ARG} for an explanation of the
4336 @code{parallel} form. Note that the callee will populate every
4337 location specified in the @code{parallel}, but if the first element of
4338 the @code{parallel} contains the whole return value, callers will use
4339 that element as the canonical location and ignore the others. The m68k
4340 port uses this type of @code{parallel} to return pointers in both
4341 @samp{%a0} (the canonical location) and @samp{%d0}.
4343 If @code{TARGET_PROMOTE_FUNCTION_RETURN} returns true, you must apply
4344 the same promotion rules specified in @code{PROMOTE_MODE} if
4345 @var{valtype} is a scalar type.
4347 If the precise function being called is known, @var{func} is a tree
4348 node (@code{FUNCTION_DECL}) for it; otherwise, @var{func} is a null
4349 pointer. This makes it possible to use a different value-returning
4350 convention for specific functions when all their calls are
4353 Some target machines have ``register windows'' so that the register in
4354 which a function returns its value is not the same as the one in which
4355 the caller sees the value. For such machines, you should return
4356 different RTX depending on @var{outgoing}.
4358 @code{TARGET_FUNCTION_VALUE} is not used for return values with
4359 aggregate data types, because these are returned in another way. See
4360 @code{TARGET_STRUCT_VALUE_RTX} and related macros, below.
4363 @defmac FUNCTION_VALUE (@var{valtype}, @var{func})
4364 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE} for
4365 a new target instead.
4368 @defmac FUNCTION_OUTGOING_VALUE (@var{valtype}, @var{func})
4369 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE} for
4370 a new target instead.
4373 @defmac LIBCALL_VALUE (@var{mode})
4374 A C expression to create an RTX representing the place where a library
4375 function returns a value of mode @var{mode}.
4377 Note that ``library function'' in this context means a compiler
4378 support routine, used to perform arithmetic, whose name is known
4379 specially by the compiler and was not mentioned in the C code being
4383 @defmac FUNCTION_VALUE_REGNO_P (@var{regno})
4384 A C expression that is nonzero if @var{regno} is the number of a hard
4385 register in which the values of called function may come back.
4387 A register whose use for returning values is limited to serving as the
4388 second of a pair (for a value of type @code{double}, say) need not be
4389 recognized by this macro. So for most machines, this definition
4393 #define FUNCTION_VALUE_REGNO_P(N) ((N) == 0)
4396 If the machine has register windows, so that the caller and the called
4397 function use different registers for the return value, this macro
4398 should recognize only the caller's register numbers.
4401 @defmac TARGET_ENUM_VA_LIST (@var{idx}, @var{pname}, @var{ptype})
4402 This target macro is used in function @code{c_common_nodes_and_builtins}
4403 to iterate through the target specific builtin types for va_list. The
4404 variable @var{idx} is used as iterator. @var{pname} has to be a pointer
4405 to a @code{const char *} and @var{ptype} a pointer to a @code{tree} typed
4407 The arguments @var{pname} and @var{ptype} are used to store the result of
4408 this macro and are set to the name of the va_list builtin type and its
4410 If the return value of this macro is zero, then there is no more element.
4411 Otherwise the @var{IDX} should be increased for the next call of this
4412 macro to iterate through all types.
4415 @defmac APPLY_RESULT_SIZE
4416 Define this macro if @samp{untyped_call} and @samp{untyped_return}
4417 need more space than is implied by @code{FUNCTION_VALUE_REGNO_P} for
4418 saving and restoring an arbitrary return value.
4421 @deftypefn {Target Hook} bool TARGET_RETURN_IN_MSB (tree @var{type})
4422 This hook should return true if values of type @var{type} are returned
4423 at the most significant end of a register (in other words, if they are
4424 padded at the least significant end). You can assume that @var{type}
4425 is returned in a register; the caller is required to check this.
4427 Note that the register provided by @code{TARGET_FUNCTION_VALUE} must
4428 be able to hold the complete return value. For example, if a 1-, 2-
4429 or 3-byte structure is returned at the most significant end of a
4430 4-byte register, @code{TARGET_FUNCTION_VALUE} should provide an
4434 @node Aggregate Return
4435 @subsection How Large Values Are Returned
4436 @cindex aggregates as return values
4437 @cindex large return values
4438 @cindex returning aggregate values
4439 @cindex structure value address
4441 When a function value's mode is @code{BLKmode} (and in some other
4442 cases), the value is not returned according to
4443 @code{TARGET_FUNCTION_VALUE} (@pxref{Scalar Return}). Instead, the
4444 caller passes the address of a block of memory in which the value
4445 should be stored. This address is called the @dfn{structure value
4448 This section describes how to control returning structure values in
4451 @deftypefn {Target Hook} bool TARGET_RETURN_IN_MEMORY (tree @var{type}, tree @var{fntype})
4452 This target hook should return a nonzero value to say to return the
4453 function value in memory, just as large structures are always returned.
4454 Here @var{type} will be the data type of the value, and @var{fntype}
4455 will be the type of the function doing the returning, or @code{NULL} for
4458 Note that values of mode @code{BLKmode} must be explicitly handled
4459 by this function. Also, the option @option{-fpcc-struct-return}
4460 takes effect regardless of this macro. On most systems, it is
4461 possible to leave the hook undefined; this causes a default
4462 definition to be used, whose value is the constant 1 for @code{BLKmode}
4463 values, and 0 otherwise.
4465 Do not use this hook to indicate that structures and unions should always
4466 be returned in memory. You should instead use @code{DEFAULT_PCC_STRUCT_RETURN}
4470 @defmac DEFAULT_PCC_STRUCT_RETURN
4471 Define this macro to be 1 if all structure and union return values must be
4472 in memory. Since this results in slower code, this should be defined
4473 only if needed for compatibility with other compilers or with an ABI@.
4474 If you define this macro to be 0, then the conventions used for structure
4475 and union return values are decided by the @code{TARGET_RETURN_IN_MEMORY}
4478 If not defined, this defaults to the value 1.
4481 @deftypefn {Target Hook} rtx TARGET_STRUCT_VALUE_RTX (tree @var{fndecl}, int @var{incoming})
4482 This target hook should return the location of the structure value
4483 address (normally a @code{mem} or @code{reg}), or 0 if the address is
4484 passed as an ``invisible'' first argument. Note that @var{fndecl} may
4485 be @code{NULL}, for libcalls. You do not need to define this target
4486 hook if the address is always passed as an ``invisible'' first
4489 On some architectures the place where the structure value address
4490 is found by the called function is not the same place that the
4491 caller put it. This can be due to register windows, or it could
4492 be because the function prologue moves it to a different place.
4493 @var{incoming} is @code{1} or @code{2} when the location is needed in
4494 the context of the called function, and @code{0} in the context of
4497 If @var{incoming} is nonzero and the address is to be found on the
4498 stack, return a @code{mem} which refers to the frame pointer. If
4499 @var{incoming} is @code{2}, the result is being used to fetch the
4500 structure value address at the beginning of a function. If you need
4501 to emit adjusting code, you should do it at this point.
4504 @defmac PCC_STATIC_STRUCT_RETURN
4505 Define this macro if the usual system convention on the target machine
4506 for returning structures and unions is for the called function to return
4507 the address of a static variable containing the value.
4509 Do not define this if the usual system convention is for the caller to
4510 pass an address to the subroutine.
4512 This macro has effect in @option{-fpcc-struct-return} mode, but it does
4513 nothing when you use @option{-freg-struct-return} mode.
4517 @subsection Caller-Saves Register Allocation
4519 If you enable it, GCC can save registers around function calls. This
4520 makes it possible to use call-clobbered registers to hold variables that
4521 must live across calls.
4523 @defmac CALLER_SAVE_PROFITABLE (@var{refs}, @var{calls})
4524 A C expression to determine whether it is worthwhile to consider placing
4525 a pseudo-register in a call-clobbered hard register and saving and
4526 restoring it around each function call. The expression should be 1 when
4527 this is worth doing, and 0 otherwise.
4529 If you don't define this macro, a default is used which is good on most
4530 machines: @code{4 * @var{calls} < @var{refs}}.
4533 @defmac HARD_REGNO_CALLER_SAVE_MODE (@var{regno}, @var{nregs})
4534 A C expression specifying which mode is required for saving @var{nregs}
4535 of a pseudo-register in call-clobbered hard register @var{regno}. If
4536 @var{regno} is unsuitable for caller save, @code{VOIDmode} should be
4537 returned. For most machines this macro need not be defined since GCC
4538 will select the smallest suitable mode.
4541 @node Function Entry
4542 @subsection Function Entry and Exit
4543 @cindex function entry and exit
4547 This section describes the macros that output function entry
4548 (@dfn{prologue}) and exit (@dfn{epilogue}) code.
4550 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_PROLOGUE (FILE *@var{file}, HOST_WIDE_INT @var{size})
4551 If defined, a function that outputs the assembler code for entry to a
4552 function. The prologue is responsible for setting up the stack frame,
4553 initializing the frame pointer register, saving registers that must be
4554 saved, and allocating @var{size} additional bytes of storage for the
4555 local variables. @var{size} is an integer. @var{file} is a stdio
4556 stream to which the assembler code should be output.
4558 The label for the beginning of the function need not be output by this
4559 macro. That has already been done when the macro is run.
4561 @findex regs_ever_live
4562 To determine which registers to save, the macro can refer to the array
4563 @code{regs_ever_live}: element @var{r} is nonzero if hard register
4564 @var{r} is used anywhere within the function. This implies the function
4565 prologue should save register @var{r}, provided it is not one of the
4566 call-used registers. (@code{TARGET_ASM_FUNCTION_EPILOGUE} must likewise use
4567 @code{regs_ever_live}.)
4569 On machines that have ``register windows'', the function entry code does
4570 not save on the stack the registers that are in the windows, even if
4571 they are supposed to be preserved by function calls; instead it takes
4572 appropriate steps to ``push'' the register stack, if any non-call-used
4573 registers are used in the function.
4575 @findex frame_pointer_needed
4576 On machines where functions may or may not have frame-pointers, the
4577 function entry code must vary accordingly; it must set up the frame
4578 pointer if one is wanted, and not otherwise. To determine whether a
4579 frame pointer is in wanted, the macro can refer to the variable
4580 @code{frame_pointer_needed}. The variable's value will be 1 at run
4581 time in a function that needs a frame pointer. @xref{Elimination}.
4583 The function entry code is responsible for allocating any stack space
4584 required for the function. This stack space consists of the regions
4585 listed below. In most cases, these regions are allocated in the
4586 order listed, with the last listed region closest to the top of the
4587 stack (the lowest address if @code{STACK_GROWS_DOWNWARD} is defined, and
4588 the highest address if it is not defined). You can use a different order
4589 for a machine if doing so is more convenient or required for
4590 compatibility reasons. Except in cases where required by standard
4591 or by a debugger, there is no reason why the stack layout used by GCC
4592 need agree with that used by other compilers for a machine.
4595 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_END_PROLOGUE (FILE *@var{file})
4596 If defined, a function that outputs assembler code at the end of a
4597 prologue. This should be used when the function prologue is being
4598 emitted as RTL, and you have some extra assembler that needs to be
4599 emitted. @xref{prologue instruction pattern}.
4602 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_BEGIN_EPILOGUE (FILE *@var{file})
4603 If defined, a function that outputs assembler code at the start of an
4604 epilogue. This should be used when the function epilogue is being
4605 emitted as RTL, and you have some extra assembler that needs to be
4606 emitted. @xref{epilogue instruction pattern}.
4609 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_EPILOGUE (FILE *@var{file}, HOST_WIDE_INT @var{size})
4610 If defined, a function that outputs the assembler code for exit from a
4611 function. The epilogue is responsible for restoring the saved
4612 registers and stack pointer to their values when the function was
4613 called, and returning control to the caller. This macro takes the
4614 same arguments as the macro @code{TARGET_ASM_FUNCTION_PROLOGUE}, and the
4615 registers to restore are determined from @code{regs_ever_live} and
4616 @code{CALL_USED_REGISTERS} in the same way.
4618 On some machines, there is a single instruction that does all the work
4619 of returning from the function. On these machines, give that
4620 instruction the name @samp{return} and do not define the macro
4621 @code{TARGET_ASM_FUNCTION_EPILOGUE} at all.
4623 Do not define a pattern named @samp{return} if you want the
4624 @code{TARGET_ASM_FUNCTION_EPILOGUE} to be used. If you want the target
4625 switches to control whether return instructions or epilogues are used,
4626 define a @samp{return} pattern with a validity condition that tests the
4627 target switches appropriately. If the @samp{return} pattern's validity
4628 condition is false, epilogues will be used.
4630 On machines where functions may or may not have frame-pointers, the
4631 function exit code must vary accordingly. Sometimes the code for these
4632 two cases is completely different. To determine whether a frame pointer
4633 is wanted, the macro can refer to the variable
4634 @code{frame_pointer_needed}. The variable's value will be 1 when compiling
4635 a function that needs a frame pointer.
4637 Normally, @code{TARGET_ASM_FUNCTION_PROLOGUE} and
4638 @code{TARGET_ASM_FUNCTION_EPILOGUE} must treat leaf functions specially.
4639 The C variable @code{current_function_is_leaf} is nonzero for such a
4640 function. @xref{Leaf Functions}.
4642 On some machines, some functions pop their arguments on exit while
4643 others leave that for the caller to do. For example, the 68020 when
4644 given @option{-mrtd} pops arguments in functions that take a fixed
4645 number of arguments.
4647 @findex current_function_pops_args
4648 Your definition of the macro @code{RETURN_POPS_ARGS} decides which
4649 functions pop their own arguments. @code{TARGET_ASM_FUNCTION_EPILOGUE}
4650 needs to know what was decided. The variable that is called
4651 @code{current_function_pops_args} is the number of bytes of its
4652 arguments that a function should pop. @xref{Scalar Return}.
4653 @c what is the "its arguments" in the above sentence referring to, pray
4654 @c tell? --mew 5feb93
4659 @findex current_function_pretend_args_size
4660 A region of @code{current_function_pretend_args_size} bytes of
4661 uninitialized space just underneath the first argument arriving on the
4662 stack. (This may not be at the very start of the allocated stack region
4663 if the calling sequence has pushed anything else since pushing the stack
4664 arguments. But usually, on such machines, nothing else has been pushed
4665 yet, because the function prologue itself does all the pushing.) This
4666 region is used on machines where an argument may be passed partly in
4667 registers and partly in memory, and, in some cases to support the
4668 features in @code{<stdarg.h>}.
4671 An area of memory used to save certain registers used by the function.
4672 The size of this area, which may also include space for such things as
4673 the return address and pointers to previous stack frames, is
4674 machine-specific and usually depends on which registers have been used
4675 in the function. Machines with register windows often do not require
4679 A region of at least @var{size} bytes, possibly rounded up to an allocation
4680 boundary, to contain the local variables of the function. On some machines,
4681 this region and the save area may occur in the opposite order, with the
4682 save area closer to the top of the stack.
4685 @cindex @code{ACCUMULATE_OUTGOING_ARGS} and stack frames
4686 Optionally, when @code{ACCUMULATE_OUTGOING_ARGS} is defined, a region of
4687 @code{current_function_outgoing_args_size} bytes to be used for outgoing
4688 argument lists of the function. @xref{Stack Arguments}.
4691 @defmac EXIT_IGNORE_STACK
4692 Define this macro as a C expression that is nonzero if the return
4693 instruction or the function epilogue ignores the value of the stack
4694 pointer; in other words, if it is safe to delete an instruction to
4695 adjust the stack pointer before a return from the function. The
4698 Note that this macro's value is relevant only for functions for which
4699 frame pointers are maintained. It is never safe to delete a final
4700 stack adjustment in a function that has no frame pointer, and the
4701 compiler knows this regardless of @code{EXIT_IGNORE_STACK}.
4704 @defmac EPILOGUE_USES (@var{regno})
4705 Define this macro as a C expression that is nonzero for registers that are
4706 used by the epilogue or the @samp{return} pattern. The stack and frame
4707 pointer registers are already assumed to be used as needed.
4710 @defmac EH_USES (@var{regno})
4711 Define this macro as a C expression that is nonzero for registers that are
4712 used by the exception handling mechanism, and so should be considered live
4713 on entry to an exception edge.
4716 @defmac DELAY_SLOTS_FOR_EPILOGUE
4717 Define this macro if the function epilogue contains delay slots to which
4718 instructions from the rest of the function can be ``moved''. The
4719 definition should be a C expression whose value is an integer
4720 representing the number of delay slots there.
4723 @defmac ELIGIBLE_FOR_EPILOGUE_DELAY (@var{insn}, @var{n})
4724 A C expression that returns 1 if @var{insn} can be placed in delay
4725 slot number @var{n} of the epilogue.
4727 The argument @var{n} is an integer which identifies the delay slot now
4728 being considered (since different slots may have different rules of
4729 eligibility). It is never negative and is always less than the number
4730 of epilogue delay slots (what @code{DELAY_SLOTS_FOR_EPILOGUE} returns).
4731 If you reject a particular insn for a given delay slot, in principle, it
4732 may be reconsidered for a subsequent delay slot. Also, other insns may
4733 (at least in principle) be considered for the so far unfilled delay
4736 @findex current_function_epilogue_delay_list
4737 @findex final_scan_insn
4738 The insns accepted to fill the epilogue delay slots are put in an RTL
4739 list made with @code{insn_list} objects, stored in the variable
4740 @code{current_function_epilogue_delay_list}. The insn for the first
4741 delay slot comes first in the list. Your definition of the macro
4742 @code{TARGET_ASM_FUNCTION_EPILOGUE} should fill the delay slots by
4743 outputting the insns in this list, usually by calling
4744 @code{final_scan_insn}.
4746 You need not define this macro if you did not define
4747 @code{DELAY_SLOTS_FOR_EPILOGUE}.
4750 @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})
4751 A function that outputs the assembler code for a thunk
4752 function, used to implement C++ virtual function calls with multiple
4753 inheritance. The thunk acts as a wrapper around a virtual function,
4754 adjusting the implicit object parameter before handing control off to
4757 First, emit code to add the integer @var{delta} to the location that
4758 contains the incoming first argument. Assume that this argument
4759 contains a pointer, and is the one used to pass the @code{this} pointer
4760 in C++. This is the incoming argument @emph{before} the function prologue,
4761 e.g.@: @samp{%o0} on a sparc. The addition must preserve the values of
4762 all other incoming arguments.
4764 Then, if @var{vcall_offset} is nonzero, an additional adjustment should be
4765 made after adding @code{delta}. In particular, if @var{p} is the
4766 adjusted pointer, the following adjustment should be made:
4769 p += (*((ptrdiff_t **)p))[vcall_offset/sizeof(ptrdiff_t)]
4772 After the additions, emit code to jump to @var{function}, which is a
4773 @code{FUNCTION_DECL}. This is a direct pure jump, not a call, and does
4774 not touch the return address. Hence returning from @var{FUNCTION} will
4775 return to whoever called the current @samp{thunk}.
4777 The effect must be as if @var{function} had been called directly with
4778 the adjusted first argument. This macro is responsible for emitting all
4779 of the code for a thunk function; @code{TARGET_ASM_FUNCTION_PROLOGUE}
4780 and @code{TARGET_ASM_FUNCTION_EPILOGUE} are not invoked.
4782 The @var{thunk_fndecl} is redundant. (@var{delta} and @var{function}
4783 have already been extracted from it.) It might possibly be useful on
4784 some targets, but probably not.
4786 If you do not define this macro, the target-independent code in the C++
4787 front end will generate a less efficient heavyweight thunk that calls
4788 @var{function} instead of jumping to it. The generic approach does
4789 not support varargs.
4792 @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})
4793 A function that returns true if TARGET_ASM_OUTPUT_MI_THUNK would be able
4794 to output the assembler code for the thunk function specified by the
4795 arguments it is passed, and false otherwise. In the latter case, the
4796 generic approach will be used by the C++ front end, with the limitations
4801 @subsection Generating Code for Profiling
4802 @cindex profiling, code generation
4804 These macros will help you generate code for profiling.
4806 @defmac FUNCTION_PROFILER (@var{file}, @var{labelno})
4807 A C statement or compound statement to output to @var{file} some
4808 assembler code to call the profiling subroutine @code{mcount}.
4811 The details of how @code{mcount} expects to be called are determined by
4812 your operating system environment, not by GCC@. To figure them out,
4813 compile a small program for profiling using the system's installed C
4814 compiler and look at the assembler code that results.
4816 Older implementations of @code{mcount} expect the address of a counter
4817 variable to be loaded into some register. The name of this variable is
4818 @samp{LP} followed by the number @var{labelno}, so you would generate
4819 the name using @samp{LP%d} in a @code{fprintf}.
4822 @defmac PROFILE_HOOK
4823 A C statement or compound statement to output to @var{file} some assembly
4824 code to call the profiling subroutine @code{mcount} even the target does
4825 not support profiling.
4828 @defmac NO_PROFILE_COUNTERS
4829 Define this macro to be an expression with a nonzero value if the
4830 @code{mcount} subroutine on your system does not need a counter variable
4831 allocated for each function. This is true for almost all modern
4832 implementations. If you define this macro, you must not use the
4833 @var{labelno} argument to @code{FUNCTION_PROFILER}.
4836 @defmac PROFILE_BEFORE_PROLOGUE
4837 Define this macro if the code for function profiling should come before
4838 the function prologue. Normally, the profiling code comes after.
4842 @subsection Permitting tail calls
4845 @deftypefn {Target Hook} bool TARGET_FUNCTION_OK_FOR_SIBCALL (tree @var{decl}, tree @var{exp})
4846 True if it is ok to do sibling call optimization for the specified
4847 call expression @var{exp}. @var{decl} will be the called function,
4848 or @code{NULL} if this is an indirect call.
4850 It is not uncommon for limitations of calling conventions to prevent
4851 tail calls to functions outside the current unit of translation, or
4852 during PIC compilation. The hook is used to enforce these restrictions,
4853 as the @code{sibcall} md pattern can not fail, or fall over to a
4854 ``normal'' call. The criteria for successful sibling call optimization
4855 may vary greatly between different architectures.
4858 @deftypefn {Target Hook} void TARGET_EXTRA_LIVE_ON_ENTRY (bitmap *@var{regs})
4859 Add any hard registers to @var{regs} that are live on entry to the
4860 function. This hook only needs to be defined to provide registers that
4861 cannot be found by examination of FUNCTION_ARG_REGNO_P, the callee saved
4862 registers, STATIC_CHAIN_INCOMING_REGNUM, STATIC_CHAIN_REGNUM,
4863 TARGET_STRUCT_VALUE_RTX, FRAME_POINTER_REGNUM, EH_USES,
4864 FRAME_POINTER_REGNUM, ARG_POINTER_REGNUM, and the PIC_OFFSET_TABLE_REGNUM.
4867 @node Stack Smashing Protection
4868 @subsection Stack smashing protection
4869 @cindex stack smashing protection
4871 @deftypefn {Target Hook} tree TARGET_STACK_PROTECT_GUARD (void)
4872 This hook returns a @code{DECL} node for the external variable to use
4873 for the stack protection guard. This variable is initialized by the
4874 runtime to some random value and is used to initialize the guard value
4875 that is placed at the top of the local stack frame. The type of this
4876 variable must be @code{ptr_type_node}.
4878 The default version of this hook creates a variable called
4879 @samp{__stack_chk_guard}, which is normally defined in @file{libgcc2.c}.
4882 @deftypefn {Target Hook} tree TARGET_STACK_PROTECT_FAIL (void)
4883 This hook returns a tree expression that alerts the runtime that the
4884 stack protect guard variable has been modified. This expression should
4885 involve a call to a @code{noreturn} function.
4887 The default version of this hook invokes a function called
4888 @samp{__stack_chk_fail}, taking no arguments. This function is
4889 normally defined in @file{libgcc2.c}.
4893 @section Implementing the Varargs Macros
4894 @cindex varargs implementation
4896 GCC comes with an implementation of @code{<varargs.h>} and
4897 @code{<stdarg.h>} that work without change on machines that pass arguments
4898 on the stack. Other machines require their own implementations of
4899 varargs, and the two machine independent header files must have
4900 conditionals to include it.
4902 ISO @code{<stdarg.h>} differs from traditional @code{<varargs.h>} mainly in
4903 the calling convention for @code{va_start}. The traditional
4904 implementation takes just one argument, which is the variable in which
4905 to store the argument pointer. The ISO implementation of
4906 @code{va_start} takes an additional second argument. The user is
4907 supposed to write the last named argument of the function here.
4909 However, @code{va_start} should not use this argument. The way to find
4910 the end of the named arguments is with the built-in functions described
4913 @defmac __builtin_saveregs ()
4914 Use this built-in function to save the argument registers in memory so
4915 that the varargs mechanism can access them. Both ISO and traditional
4916 versions of @code{va_start} must use @code{__builtin_saveregs}, unless
4917 you use @code{TARGET_SETUP_INCOMING_VARARGS} (see below) instead.
4919 On some machines, @code{__builtin_saveregs} is open-coded under the
4920 control of the target hook @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. On
4921 other machines, it calls a routine written in assembler language,
4922 found in @file{libgcc2.c}.
4924 Code generated for the call to @code{__builtin_saveregs} appears at the
4925 beginning of the function, as opposed to where the call to
4926 @code{__builtin_saveregs} is written, regardless of what the code is.
4927 This is because the registers must be saved before the function starts
4928 to use them for its own purposes.
4929 @c i rewrote the first sentence above to fix an overfull hbox. --mew
4933 @defmac __builtin_args_info (@var{category})
4934 Use this built-in function to find the first anonymous arguments in
4937 In general, a machine may have several categories of registers used for
4938 arguments, each for a particular category of data types. (For example,
4939 on some machines, floating-point registers are used for floating-point
4940 arguments while other arguments are passed in the general registers.)
4941 To make non-varargs functions use the proper calling convention, you
4942 have defined the @code{CUMULATIVE_ARGS} data type to record how many
4943 registers in each category have been used so far
4945 @code{__builtin_args_info} accesses the same data structure of type
4946 @code{CUMULATIVE_ARGS} after the ordinary argument layout is finished
4947 with it, with @var{category} specifying which word to access. Thus, the
4948 value indicates the first unused register in a given category.
4950 Normally, you would use @code{__builtin_args_info} in the implementation
4951 of @code{va_start}, accessing each category just once and storing the
4952 value in the @code{va_list} object. This is because @code{va_list} will
4953 have to update the values, and there is no way to alter the
4954 values accessed by @code{__builtin_args_info}.
4957 @defmac __builtin_next_arg (@var{lastarg})
4958 This is the equivalent of @code{__builtin_args_info}, for stack
4959 arguments. It returns the address of the first anonymous stack
4960 argument, as type @code{void *}. If @code{ARGS_GROW_DOWNWARD}, it
4961 returns the address of the location above the first anonymous stack
4962 argument. Use it in @code{va_start} to initialize the pointer for
4963 fetching arguments from the stack. Also use it in @code{va_start} to
4964 verify that the second parameter @var{lastarg} is the last named argument
4965 of the current function.
4968 @defmac __builtin_classify_type (@var{object})
4969 Since each machine has its own conventions for which data types are
4970 passed in which kind of register, your implementation of @code{va_arg}
4971 has to embody these conventions. The easiest way to categorize the
4972 specified data type is to use @code{__builtin_classify_type} together
4973 with @code{sizeof} and @code{__alignof__}.
4975 @code{__builtin_classify_type} ignores the value of @var{object},
4976 considering only its data type. It returns an integer describing what
4977 kind of type that is---integer, floating, pointer, structure, and so on.
4979 The file @file{typeclass.h} defines an enumeration that you can use to
4980 interpret the values of @code{__builtin_classify_type}.
4983 These machine description macros help implement varargs:
4985 @deftypefn {Target Hook} rtx TARGET_EXPAND_BUILTIN_SAVEREGS (void)
4986 If defined, this hook produces the machine-specific code for a call to
4987 @code{__builtin_saveregs}. This code will be moved to the very
4988 beginning of the function, before any parameter access are made. The
4989 return value of this function should be an RTX that contains the value
4990 to use as the return of @code{__builtin_saveregs}.
4993 @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})
4994 This target hook offers an alternative to using
4995 @code{__builtin_saveregs} and defining the hook
4996 @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. Use it to store the anonymous
4997 register arguments into the stack so that all the arguments appear to
4998 have been passed consecutively on the stack. Once this is done, you can
4999 use the standard implementation of varargs that works for machines that
5000 pass all their arguments on the stack.
5002 The argument @var{args_so_far} points to the @code{CUMULATIVE_ARGS} data
5003 structure, containing the values that are obtained after processing the
5004 named arguments. The arguments @var{mode} and @var{type} describe the
5005 last named argument---its machine mode and its data type as a tree node.
5007 The target hook should do two things: first, push onto the stack all the
5008 argument registers @emph{not} used for the named arguments, and second,
5009 store the size of the data thus pushed into the @code{int}-valued
5010 variable pointed to by @var{pretend_args_size}. The value that you
5011 store here will serve as additional offset for setting up the stack
5014 Because you must generate code to push the anonymous arguments at
5015 compile time without knowing their data types,
5016 @code{TARGET_SETUP_INCOMING_VARARGS} is only useful on machines that
5017 have just a single category of argument register and use it uniformly
5020 If the argument @var{second_time} is nonzero, it means that the
5021 arguments of the function are being analyzed for the second time. This
5022 happens for an inline function, which is not actually compiled until the
5023 end of the source file. The hook @code{TARGET_SETUP_INCOMING_VARARGS} should
5024 not generate any instructions in this case.
5027 @deftypefn {Target Hook} bool TARGET_STRICT_ARGUMENT_NAMING (CUMULATIVE_ARGS *@var{ca})
5028 Define this hook to return @code{true} if the location where a function
5029 argument is passed depends on whether or not it is a named argument.
5031 This hook controls how the @var{named} argument to @code{FUNCTION_ARG}
5032 is set for varargs and stdarg functions. If this hook returns
5033 @code{true}, the @var{named} argument is always true for named
5034 arguments, and false for unnamed arguments. If it returns @code{false},
5035 but @code{TARGET_PRETEND_OUTGOING_VARARGS_NAMED} returns @code{true},
5036 then all arguments are treated as named. Otherwise, all named arguments
5037 except the last are treated as named.
5039 You need not define this hook if it always returns zero.
5042 @deftypefn {Target Hook} bool TARGET_PRETEND_OUTGOING_VARARGS_NAMED
5043 If you need to conditionally change ABIs so that one works with
5044 @code{TARGET_SETUP_INCOMING_VARARGS}, but the other works like neither
5045 @code{TARGET_SETUP_INCOMING_VARARGS} nor @code{TARGET_STRICT_ARGUMENT_NAMING} was
5046 defined, then define this hook to return @code{true} if
5047 @code{TARGET_SETUP_INCOMING_VARARGS} is used, @code{false} otherwise.
5048 Otherwise, you should not define this hook.
5052 @section Trampolines for Nested Functions
5053 @cindex trampolines for nested functions
5054 @cindex nested functions, trampolines for
5056 A @dfn{trampoline} is a small piece of code that is created at run time
5057 when the address of a nested function is taken. It normally resides on
5058 the stack, in the stack frame of the containing function. These macros
5059 tell GCC how to generate code to allocate and initialize a
5062 The instructions in the trampoline must do two things: load a constant
5063 address into the static chain register, and jump to the real address of
5064 the nested function. On CISC machines such as the m68k, this requires
5065 two instructions, a move immediate and a jump. Then the two addresses
5066 exist in the trampoline as word-long immediate operands. On RISC
5067 machines, it is often necessary to load each address into a register in
5068 two parts. Then pieces of each address form separate immediate
5071 The code generated to initialize the trampoline must store the variable
5072 parts---the static chain value and the function address---into the
5073 immediate operands of the instructions. On a CISC machine, this is
5074 simply a matter of copying each address to a memory reference at the
5075 proper offset from the start of the trampoline. On a RISC machine, it
5076 may be necessary to take out pieces of the address and store them
5079 @defmac TRAMPOLINE_TEMPLATE (@var{file})
5080 A C statement to output, on the stream @var{file}, assembler code for a
5081 block of data that contains the constant parts of a trampoline. This
5082 code should not include a label---the label is taken care of
5085 If you do not define this macro, it means no template is needed
5086 for the target. Do not define this macro on systems where the block move
5087 code to copy the trampoline into place would be larger than the code
5088 to generate it on the spot.
5091 @defmac TRAMPOLINE_SECTION
5092 Return the section into which the trampoline template is to be placed
5093 (@pxref{Sections}). The default value is @code{readonly_data_section}.
5096 @defmac TRAMPOLINE_SIZE
5097 A C expression for the size in bytes of the trampoline, as an integer.
5100 @defmac TRAMPOLINE_ALIGNMENT
5101 Alignment required for trampolines, in bits.
5103 If you don't define this macro, the value of @code{BIGGEST_ALIGNMENT}
5104 is used for aligning trampolines.
5107 @defmac INITIALIZE_TRAMPOLINE (@var{addr}, @var{fnaddr}, @var{static_chain})
5108 A C statement to initialize the variable parts of a trampoline.
5109 @var{addr} is an RTX for the address of the trampoline; @var{fnaddr} is
5110 an RTX for the address of the nested function; @var{static_chain} is an
5111 RTX for the static chain value that should be passed to the function
5115 @defmac TRAMPOLINE_ADJUST_ADDRESS (@var{addr})
5116 A C statement that should perform any machine-specific adjustment in
5117 the address of the trampoline. Its argument contains the address that
5118 was passed to @code{INITIALIZE_TRAMPOLINE}. In case the address to be
5119 used for a function call should be different from the address in which
5120 the template was stored, the different address should be assigned to
5121 @var{addr}. If this macro is not defined, @var{addr} will be used for
5124 @cindex @code{TARGET_ASM_FUNCTION_EPILOGUE} and trampolines
5125 @cindex @code{TARGET_ASM_FUNCTION_PROLOGUE} and trampolines
5126 If this macro is not defined, by default the trampoline is allocated as
5127 a stack slot. This default is right for most machines. The exceptions
5128 are machines where it is impossible to execute instructions in the stack
5129 area. On such machines, you may have to implement a separate stack,
5130 using this macro in conjunction with @code{TARGET_ASM_FUNCTION_PROLOGUE}
5131 and @code{TARGET_ASM_FUNCTION_EPILOGUE}.
5133 @var{fp} points to a data structure, a @code{struct function}, which
5134 describes the compilation status of the immediate containing function of
5135 the function which the trampoline is for. The stack slot for the
5136 trampoline is in the stack frame of this containing function. Other
5137 allocation strategies probably must do something analogous with this
5141 Implementing trampolines is difficult on many machines because they have
5142 separate instruction and data caches. Writing into a stack location
5143 fails to clear the memory in the instruction cache, so when the program
5144 jumps to that location, it executes the old contents.
5146 Here are two possible solutions. One is to clear the relevant parts of
5147 the instruction cache whenever a trampoline is set up. The other is to
5148 make all trampolines identical, by having them jump to a standard
5149 subroutine. The former technique makes trampoline execution faster; the
5150 latter makes initialization faster.
5152 To clear the instruction cache when a trampoline is initialized, define
5153 the following macro.
5155 @defmac CLEAR_INSN_CACHE (@var{beg}, @var{end})
5156 If defined, expands to a C expression clearing the @emph{instruction
5157 cache} in the specified interval. The definition of this macro would
5158 typically be a series of @code{asm} statements. Both @var{beg} and
5159 @var{end} are both pointer expressions.
5162 The operating system may also require the stack to be made executable
5163 before calling the trampoline. To implement this requirement, define
5164 the following macro.
5166 @defmac ENABLE_EXECUTE_STACK
5167 Define this macro if certain operations must be performed before executing
5168 code located on the stack. The macro should expand to a series of C
5169 file-scope constructs (e.g.@: functions) and provide a unique entry point
5170 named @code{__enable_execute_stack}. The target is responsible for
5171 emitting calls to the entry point in the code, for example from the
5172 @code{INITIALIZE_TRAMPOLINE} macro.
5175 To use a standard subroutine, define the following macro. In addition,
5176 you must make sure that the instructions in a trampoline fill an entire
5177 cache line with identical instructions, or else ensure that the
5178 beginning of the trampoline code is always aligned at the same point in
5179 its cache line. Look in @file{m68k.h} as a guide.
5181 @defmac TRANSFER_FROM_TRAMPOLINE
5182 Define this macro if trampolines need a special subroutine to do their
5183 work. The macro should expand to a series of @code{asm} statements
5184 which will be compiled with GCC@. They go in a library function named
5185 @code{__transfer_from_trampoline}.
5187 If you need to avoid executing the ordinary prologue code of a compiled
5188 C function when you jump to the subroutine, you can do so by placing a
5189 special label of your own in the assembler code. Use one @code{asm}
5190 statement to generate an assembler label, and another to make the label
5191 global. Then trampolines can use that label to jump directly to your
5192 special assembler code.
5196 @section Implicit Calls to Library Routines
5197 @cindex library subroutine names
5198 @cindex @file{libgcc.a}
5200 @c prevent bad page break with this line
5201 Here is an explanation of implicit calls to library routines.
5203 @defmac DECLARE_LIBRARY_RENAMES
5204 This macro, if defined, should expand to a piece of C code that will get
5205 expanded when compiling functions for libgcc.a. It can be used to
5206 provide alternate names for GCC's internal library functions if there
5207 are ABI-mandated names that the compiler should provide.
5210 @findex init_one_libfunc
5211 @findex set_optab_libfunc
5212 @deftypefn {Target Hook} void TARGET_INIT_LIBFUNCS (void)
5213 This hook should declare additional library routines or rename
5214 existing ones, using the functions @code{set_optab_libfunc} and
5215 @code{init_one_libfunc} defined in @file{optabs.c}.
5216 @code{init_optabs} calls this macro after initializing all the normal
5219 The default is to do nothing. Most ports don't need to define this hook.
5222 @defmac FLOAT_LIB_COMPARE_RETURNS_BOOL (@var{mode}, @var{comparison})
5223 This macro should return @code{true} if the library routine that
5224 implements the floating point comparison operator @var{comparison} in
5225 mode @var{mode} will return a boolean, and @var{false} if it will
5228 GCC's own floating point libraries return tristates from the
5229 comparison operators, so the default returns false always. Most ports
5230 don't need to define this macro.
5233 @defmac TARGET_LIB_INT_CMP_BIASED
5234 This macro should evaluate to @code{true} if the integer comparison
5235 functions (like @code{__cmpdi2}) return 0 to indicate that the first
5236 operand is smaller than the second, 1 to indicate that they are equal,
5237 and 2 to indicate that the first operand is greater than the second.
5238 If this macro evaluates to @code{false} the comparison functions return
5239 @minus{}1, 0, and 1 instead of 0, 1, and 2. If the target uses the routines
5240 in @file{libgcc.a}, you do not need to define this macro.
5243 @cindex US Software GOFAST, floating point emulation library
5244 @cindex floating point emulation library, US Software GOFAST
5245 @cindex GOFAST, floating point emulation library
5246 @findex gofast_maybe_init_libfuncs
5247 @defmac US_SOFTWARE_GOFAST
5248 Define this macro if your system C library uses the US Software GOFAST
5249 library to provide floating point emulation.
5251 In addition to defining this macro, your architecture must set
5252 @code{TARGET_INIT_LIBFUNCS} to @code{gofast_maybe_init_libfuncs}, or
5253 else call that function from its version of that hook. It is defined
5254 in @file{config/gofast.h}, which must be included by your
5255 architecture's @file{@var{cpu}.c} file. See @file{sparc/sparc.c} for
5258 If this macro is defined, the
5259 @code{TARGET_FLOAT_LIB_COMPARE_RETURNS_BOOL} target hook must return
5260 false for @code{SFmode} and @code{DFmode} comparisons.
5263 @cindex @code{EDOM}, implicit usage
5266 The value of @code{EDOM} on the target machine, as a C integer constant
5267 expression. If you don't define this macro, GCC does not attempt to
5268 deposit the value of @code{EDOM} into @code{errno} directly. Look in
5269 @file{/usr/include/errno.h} to find the value of @code{EDOM} on your
5272 If you do not define @code{TARGET_EDOM}, then compiled code reports
5273 domain errors by calling the library function and letting it report the
5274 error. If mathematical functions on your system use @code{matherr} when
5275 there is an error, then you should leave @code{TARGET_EDOM} undefined so
5276 that @code{matherr} is used normally.
5279 @cindex @code{errno}, implicit usage
5280 @defmac GEN_ERRNO_RTX
5281 Define this macro as a C expression to create an rtl expression that
5282 refers to the global ``variable'' @code{errno}. (On certain systems,
5283 @code{errno} may not actually be a variable.) If you don't define this
5284 macro, a reasonable default is used.
5287 @cindex C99 math functions, implicit usage
5288 @defmac TARGET_C99_FUNCTIONS
5289 When this macro is nonzero, GCC will implicitly optimize @code{sin} calls into
5290 @code{sinf} and similarly for other functions defined by C99 standard. The
5291 default is zero because a number of existing systems lack support for these
5292 functions in their runtime so this macro needs to be redefined to one on
5293 systems that do support the C99 runtime.
5296 @cindex sincos math function, implicit usage
5297 @defmac TARGET_HAS_SINCOS
5298 When this macro is nonzero, GCC will implicitly optimize calls to @code{sin}
5299 and @code{cos} with the same argument to a call to @code{sincos}. The
5300 default is zero. The target has to provide the following functions:
5302 void sincos(double x, double *sin, double *cos);
5303 void sincosf(float x, float *sin, float *cos);
5304 void sincosl(long double x, long double *sin, long double *cos);
5308 @defmac NEXT_OBJC_RUNTIME
5309 Define this macro to generate code for Objective-C message sending using
5310 the calling convention of the NeXT system. This calling convention
5311 involves passing the object, the selector and the method arguments all
5312 at once to the method-lookup library function.
5314 The default calling convention passes just the object and the selector
5315 to the lookup function, which returns a pointer to the method.
5318 @node Addressing Modes
5319 @section Addressing Modes
5320 @cindex addressing modes
5322 @c prevent bad page break with this line
5323 This is about addressing modes.
5325 @defmac HAVE_PRE_INCREMENT
5326 @defmacx HAVE_PRE_DECREMENT
5327 @defmacx HAVE_POST_INCREMENT
5328 @defmacx HAVE_POST_DECREMENT
5329 A C expression that is nonzero if the machine supports pre-increment,
5330 pre-decrement, post-increment, or post-decrement addressing respectively.
5333 @defmac HAVE_PRE_MODIFY_DISP
5334 @defmacx HAVE_POST_MODIFY_DISP
5335 A C expression that is nonzero if the machine supports pre- or
5336 post-address side-effect generation involving constants other than
5337 the size of the memory operand.
5340 @defmac HAVE_PRE_MODIFY_REG
5341 @defmacx HAVE_POST_MODIFY_REG
5342 A C expression that is nonzero if the machine supports pre- or
5343 post-address side-effect generation involving a register displacement.
5346 @defmac CONSTANT_ADDRESS_P (@var{x})
5347 A C expression that is 1 if the RTX @var{x} is a constant which
5348 is a valid address. On most machines, this can be defined as
5349 @code{CONSTANT_P (@var{x})}, but a few machines are more restrictive
5350 in which constant addresses are supported.
5353 @defmac CONSTANT_P (@var{x})
5354 @code{CONSTANT_P}, which is defined by target-independent code,
5355 accepts integer-values expressions whose values are not explicitly
5356 known, such as @code{symbol_ref}, @code{label_ref}, and @code{high}
5357 expressions and @code{const} arithmetic expressions, in addition to
5358 @code{const_int} and @code{const_double} expressions.
5361 @defmac MAX_REGS_PER_ADDRESS
5362 A number, the maximum number of registers that can appear in a valid
5363 memory address. Note that it is up to you to specify a value equal to
5364 the maximum number that @code{TARGET_LEGITIMATE_ADDRESS_P} would ever
5368 @deftypefn {Target Hook} TARGET_LEGITIMATE_ADDRESS_P (enum machine_mode @var{mode}, rtx @var{x}, bool @var{strict})
5369 A function that returns whether @var{x} (an RTX) is a legitimate memory
5370 address on the target machine for a memory operand of mode @var{mode}.
5372 Legitimate addresses are defined in two variants: a strict variant and a
5373 non-strict one. The @code{strict} parameter chooses which variant is
5374 desired by the caller.
5376 The strict variant is used in the reload pass. It must be defined so
5377 that any pseudo-register that has not been allocated a hard register is
5378 considered a memory reference. This is because in contexts where some
5379 kind of register is required, a pseudo-register with no hard register
5380 must be rejected. For non-hard registers, the strict variant should look
5381 up the @code{reg_renumber} array; it should then proceed using the hard
5382 register number in the array, or treat the pseudo as a memory reference
5383 if the array holds @code{-1}.
5385 The non-strict variant is used in other passes. It must be defined to
5386 accept all pseudo-registers in every context where some kind of
5387 register is required.
5389 Normally, constant addresses which are the sum of a @code{symbol_ref}
5390 and an integer are stored inside a @code{const} RTX to mark them as
5391 constant. Therefore, there is no need to recognize such sums
5392 specifically as legitimate addresses. Normally you would simply
5393 recognize any @code{const} as legitimate.
5395 Usually @code{PRINT_OPERAND_ADDRESS} is not prepared to handle constant
5396 sums that are not marked with @code{const}. It assumes that a naked
5397 @code{plus} indicates indexing. If so, then you @emph{must} reject such
5398 naked constant sums as illegitimate addresses, so that none of them will
5399 be given to @code{PRINT_OPERAND_ADDRESS}.
5401 @cindex @code{TARGET_ENCODE_SECTION_INFO} and address validation
5402 On some machines, whether a symbolic address is legitimate depends on
5403 the section that the address refers to. On these machines, define the
5404 target hook @code{TARGET_ENCODE_SECTION_INFO} to store the information
5405 into the @code{symbol_ref}, and then check for it here. When you see a
5406 @code{const}, you will have to look inside it to find the
5407 @code{symbol_ref} in order to determine the section. @xref{Assembler
5410 @cindex @code{GO_IF_LEGITIMATE_ADDRESS}
5411 Some ports are still using a deprecated legacy substitute for
5412 this hook, the @code{GO_IF_LEGITIMATE_ADDRESS} macro. This macro
5416 #define GO_IF_LEGITIMATE_ADDRESS (@var{mode}, @var{x}, @var{label})
5420 and should @code{goto @var{label}} if the address @var{x} is a valid
5421 address on the target machine for a memory operand of mode @var{mode}.
5422 Whether the strict or non-strict variants are desired is defined by
5423 the @code{REG_OK_STRICT} macro introduced earlier in this section.
5424 Using the hook is usually simpler because it limits the number of
5425 files that are recompiled when changes are made.
5428 @defmac TARGET_MEM_CONSTRAINT
5429 A single character to be used instead of the default @code{'m'}
5430 character for general memory addresses. This defines the constraint
5431 letter which matches the memory addresses accepted by
5432 @code{TARGET_LEGITIMATE_ADDRESS_P}. Define this macro if you want to
5433 support new address formats in your back end without changing the
5434 semantics of the @code{'m'} constraint. This is necessary in order to
5435 preserve functionality of inline assembly constructs using the
5436 @code{'m'} constraint.
5439 @defmac FIND_BASE_TERM (@var{x})
5440 A C expression to determine the base term of address @var{x},
5441 or to provide a simplified version of @var{x} from which @file{alias.c}
5442 can easily find the base term. This macro is used in only two places:
5443 @code{find_base_value} and @code{find_base_term} in @file{alias.c}.
5445 It is always safe for this macro to not be defined. It exists so
5446 that alias analysis can understand machine-dependent addresses.
5448 The typical use of this macro is to handle addresses containing
5449 a label_ref or symbol_ref within an UNSPEC@.
5452 @deftypefn {Target Hook} rtx TARGET_LEGITIMIZE_ADDRESS (rtx @var{x}, rtx @var{oldx}, enum machine_mode @var{mode})
5453 This hook is given an invalid memory address @var{x} for an
5454 operand of mode @var{mode} and should try to return a valid memory
5457 @findex break_out_memory_refs
5458 @var{x} will always be the result of a call to @code{break_out_memory_refs},
5459 and @var{oldx} will be the operand that was given to that function to produce
5462 The code of the hook should not alter the substructure of
5463 @var{x}. If it transforms @var{x} into a more legitimate form, it
5464 should return the new @var{x}.
5466 It is not necessary for this hook to come up with a legitimate address.
5467 The compiler has standard ways of doing so in all cases. In fact, it
5468 is safe to omit this hook or make it return @var{x} if it cannot find
5469 a valid way to legitimize the address. But often a machine-dependent
5470 strategy can generate better code.
5473 @defmac LEGITIMIZE_RELOAD_ADDRESS (@var{x}, @var{mode}, @var{opnum}, @var{type}, @var{ind_levels}, @var{win})
5474 A C compound statement that attempts to replace @var{x}, which is an address
5475 that needs reloading, with a valid memory address for an operand of mode
5476 @var{mode}. @var{win} will be a C statement label elsewhere in the code.
5477 It is not necessary to define this macro, but it might be useful for
5478 performance reasons.
5480 For example, on the i386, it is sometimes possible to use a single
5481 reload register instead of two by reloading a sum of two pseudo
5482 registers into a register. On the other hand, for number of RISC
5483 processors offsets are limited so that often an intermediate address
5484 needs to be generated in order to address a stack slot. By defining
5485 @code{LEGITIMIZE_RELOAD_ADDRESS} appropriately, the intermediate addresses
5486 generated for adjacent some stack slots can be made identical, and thus
5489 @emph{Note}: This macro should be used with caution. It is necessary
5490 to know something of how reload works in order to effectively use this,
5491 and it is quite easy to produce macros that build in too much knowledge
5492 of reload internals.
5494 @emph{Note}: This macro must be able to reload an address created by a
5495 previous invocation of this macro. If it fails to handle such addresses
5496 then the compiler may generate incorrect code or abort.
5499 The macro definition should use @code{push_reload} to indicate parts that
5500 need reloading; @var{opnum}, @var{type} and @var{ind_levels} are usually
5501 suitable to be passed unaltered to @code{push_reload}.
5503 The code generated by this macro must not alter the substructure of
5504 @var{x}. If it transforms @var{x} into a more legitimate form, it
5505 should assign @var{x} (which will always be a C variable) a new value.
5506 This also applies to parts that you change indirectly by calling
5509 @findex strict_memory_address_p
5510 The macro definition may use @code{strict_memory_address_p} to test if
5511 the address has become legitimate.
5514 If you want to change only a part of @var{x}, one standard way of doing
5515 this is to use @code{copy_rtx}. Note, however, that it unshares only a
5516 single level of rtl. Thus, if the part to be changed is not at the
5517 top level, you'll need to replace first the top level.
5518 It is not necessary for this macro to come up with a legitimate
5519 address; but often a machine-dependent strategy can generate better code.
5522 @defmac GO_IF_MODE_DEPENDENT_ADDRESS (@var{addr}, @var{label})
5523 A C statement or compound statement with a conditional @code{goto
5524 @var{label};} executed if memory address @var{x} (an RTX) can have
5525 different meanings depending on the machine mode of the memory
5526 reference it is used for or if the address is valid for some modes
5529 Autoincrement and autodecrement addresses typically have mode-dependent
5530 effects because the amount of the increment or decrement is the size
5531 of the operand being addressed. Some machines have other mode-dependent
5532 addresses. Many RISC machines have no mode-dependent addresses.
5534 You may assume that @var{addr} is a valid address for the machine.
5537 @defmac LEGITIMATE_CONSTANT_P (@var{x})
5538 A C expression that is nonzero if @var{x} is a legitimate constant for
5539 an immediate operand on the target machine. You can assume that
5540 @var{x} satisfies @code{CONSTANT_P}, so you need not check this. In fact,
5541 @samp{1} is a suitable definition for this macro on machines where
5542 anything @code{CONSTANT_P} is valid.
5545 @deftypefn {Target Hook} rtx TARGET_DELEGITIMIZE_ADDRESS (rtx @var{x})
5546 This hook is used to undo the possibly obfuscating effects of the
5547 @code{LEGITIMIZE_ADDRESS} and @code{LEGITIMIZE_RELOAD_ADDRESS} target
5548 macros. Some backend implementations of these macros wrap symbol
5549 references inside an @code{UNSPEC} rtx to represent PIC or similar
5550 addressing modes. This target hook allows GCC's optimizers to understand
5551 the semantics of these opaque @code{UNSPEC}s by converting them back
5552 into their original form.
5555 @deftypefn {Target Hook} bool TARGET_CANNOT_FORCE_CONST_MEM (rtx @var{x})
5556 This hook should return true if @var{x} is of a form that cannot (or
5557 should not) be spilled to the constant pool. The default version of
5558 this hook returns false.
5560 The primary reason to define this hook is to prevent reload from
5561 deciding that a non-legitimate constant would be better reloaded
5562 from the constant pool instead of spilling and reloading a register
5563 holding the constant. This restriction is often true of addresses
5564 of TLS symbols for various targets.
5567 @deftypefn {Target Hook} bool TARGET_USE_BLOCKS_FOR_CONSTANT_P (enum machine_mode @var{mode}, rtx @var{x})
5568 This hook should return true if pool entries for constant @var{x} can
5569 be placed in an @code{object_block} structure. @var{mode} is the mode
5572 The default version returns false for all constants.
5575 @deftypefn {Target Hook} tree TARGET_BUILTIN_RECIPROCAL (enum tree_code @var{fn}, bool @var{tm_fn}, bool @var{sqrt})
5576 This hook should return the DECL of a function that implements reciprocal of
5577 the builtin function with builtin function code @var{fn}, or
5578 @code{NULL_TREE} if such a function is not available. @var{tm_fn} is true
5579 when @var{fn} is a code of a machine-dependent builtin function. When
5580 @var{sqrt} is true, additional optimizations that apply only to the reciprocal
5581 of a square root function are performed, and only reciprocals of @code{sqrt}
5585 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_MASK_FOR_LOAD (void)
5586 This hook should return the DECL of a function @var{f} that given an
5587 address @var{addr} as an argument returns a mask @var{m} that can be
5588 used to extract from two vectors the relevant data that resides in
5589 @var{addr} in case @var{addr} is not properly aligned.
5591 The autovectorizer, when vectorizing a load operation from an address
5592 @var{addr} that may be unaligned, will generate two vector loads from
5593 the two aligned addresses around @var{addr}. It then generates a
5594 @code{REALIGN_LOAD} operation to extract the relevant data from the
5595 two loaded vectors. The first two arguments to @code{REALIGN_LOAD},
5596 @var{v1} and @var{v2}, are the two vectors, each of size @var{VS}, and
5597 the third argument, @var{OFF}, defines how the data will be extracted
5598 from these two vectors: if @var{OFF} is 0, then the returned vector is
5599 @var{v2}; otherwise, the returned vector is composed from the last
5600 @var{VS}-@var{OFF} elements of @var{v1} concatenated to the first
5601 @var{OFF} elements of @var{v2}.
5603 If this hook is defined, the autovectorizer will generate a call
5604 to @var{f} (using the DECL tree that this hook returns) and will
5605 use the return value of @var{f} as the argument @var{OFF} to
5606 @code{REALIGN_LOAD}. Therefore, the mask @var{m} returned by @var{f}
5607 should comply with the semantics expected by @code{REALIGN_LOAD}
5609 If this hook is not defined, then @var{addr} will be used as
5610 the argument @var{OFF} to @code{REALIGN_LOAD}, in which case the low
5611 log2(@var{VS})-1 bits of @var{addr} will be considered.
5614 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_EVEN (tree @var{x})
5615 This hook should return the DECL of a function @var{f} that implements
5616 widening multiplication of the even elements of two input vectors of type @var{x}.
5618 If this hook is defined, the autovectorizer will use it along with the
5619 @code{TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_ODD} target hook when vectorizing
5620 widening multiplication in cases that the order of the results does not have to be
5621 preserved (e.g.@: used only by a reduction computation). Otherwise, the
5622 @code{widen_mult_hi/lo} idioms will be used.
5625 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_ODD (tree @var{x})
5626 This hook should return the DECL of a function @var{f} that implements
5627 widening multiplication of the odd elements of two input vectors of type @var{x}.
5629 If this hook is defined, the autovectorizer will use it along with the
5630 @code{TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_EVEN} target hook when vectorizing
5631 widening multiplication in cases that the order of the results does not have to be
5632 preserved (e.g.@: used only by a reduction computation). Otherwise, the
5633 @code{widen_mult_hi/lo} idioms will be used.
5636 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_CONVERSION (enum tree_code @var{code}, tree @var{type})
5637 This hook should return the DECL of a function that implements conversion of the
5638 input vector of type @var{type}.
5639 If @var{type} is an integral type, the result of the conversion is a vector of
5640 floating-point type of the same size.
5641 If @var{type} is a floating-point type, the result of the conversion is a vector
5642 of integral type of the same size.
5643 @var{code} specifies how the conversion is to be applied
5644 (truncation, rounding, etc.).
5646 If this hook is defined, the autovectorizer will use the
5647 @code{TARGET_VECTORIZE_BUILTIN_CONVERSION} target hook when vectorizing
5648 conversion. Otherwise, it will return @code{NULL_TREE}.
5651 @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})
5652 This hook should return the decl of a function that implements the vectorized
5653 variant of the builtin function with builtin function code @var{code} or
5654 @code{NULL_TREE} if such a function is not available. The return type of
5655 the vectorized function shall be of vector type @var{vec_type_out} and the
5656 argument types should be @var{vec_type_in}.
5659 @node Anchored Addresses
5660 @section Anchored Addresses
5661 @cindex anchored addresses
5662 @cindex @option{-fsection-anchors}
5664 GCC usually addresses every static object as a separate entity.
5665 For example, if we have:
5669 int foo (void) @{ return a + b + c; @}
5672 the code for @code{foo} will usually calculate three separate symbolic
5673 addresses: those of @code{a}, @code{b} and @code{c}. On some targets,
5674 it would be better to calculate just one symbolic address and access
5675 the three variables relative to it. The equivalent pseudocode would
5681 register int *xr = &x;
5682 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
5686 (which isn't valid C). We refer to shared addresses like @code{x} as
5687 ``section anchors''. Their use is controlled by @option{-fsection-anchors}.
5689 The hooks below describe the target properties that GCC needs to know
5690 in order to make effective use of section anchors. It won't use
5691 section anchors at all unless either @code{TARGET_MIN_ANCHOR_OFFSET}
5692 or @code{TARGET_MAX_ANCHOR_OFFSET} is set to a nonzero value.
5694 @deftypevr {Target Hook} HOST_WIDE_INT TARGET_MIN_ANCHOR_OFFSET
5695 The minimum offset that should be applied to a section anchor.
5696 On most targets, it should be the smallest offset that can be
5697 applied to a base register while still giving a legitimate address
5698 for every mode. The default value is 0.
5701 @deftypevr {Target Hook} HOST_WIDE_INT TARGET_MAX_ANCHOR_OFFSET
5702 Like @code{TARGET_MIN_ANCHOR_OFFSET}, but the maximum (inclusive)
5703 offset that should be applied to section anchors. The default
5707 @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_ANCHOR (rtx @var{x})
5708 Write the assembly code to define section anchor @var{x}, which is a
5709 @code{SYMBOL_REF} for which @samp{SYMBOL_REF_ANCHOR_P (@var{x})} is true.
5710 The hook is called with the assembly output position set to the beginning
5711 of @code{SYMBOL_REF_BLOCK (@var{x})}.
5713 If @code{ASM_OUTPUT_DEF} is available, the hook's default definition uses
5714 it to define the symbol as @samp{. + SYMBOL_REF_BLOCK_OFFSET (@var{x})}.
5715 If @code{ASM_OUTPUT_DEF} is not available, the hook's default definition
5716 is @code{NULL}, which disables the use of section anchors altogether.
5719 @deftypefn {Target Hook} bool TARGET_USE_ANCHORS_FOR_SYMBOL_P (rtx @var{x})
5720 Return true if GCC should attempt to use anchors to access @code{SYMBOL_REF}
5721 @var{x}. You can assume @samp{SYMBOL_REF_HAS_BLOCK_INFO_P (@var{x})} and
5722 @samp{!SYMBOL_REF_ANCHOR_P (@var{x})}.
5724 The default version is correct for most targets, but you might need to
5725 intercept this hook to handle things like target-specific attributes
5726 or target-specific sections.
5729 @node Condition Code
5730 @section Condition Code Status
5731 @cindex condition code status
5733 The macros in this section can be split in two families, according to the
5734 two ways of representing condition codes in GCC.
5736 The first representation is the so called @code{(cc0)} representation
5737 (@pxref{Jump Patterns}), where all instructions can have an implicit
5738 clobber of the condition codes. The second is the condition code
5739 register representation, which provides better schedulability for
5740 architectures that do have a condition code register, but on which
5741 most instructions do not affect it. The latter category includes
5744 The implicit clobbering poses a strong restriction on the placement of
5745 the definition and use of the condition code, which need to be in adjacent
5746 insns for machines using @code{(cc0)}. This can prevent important
5747 optimizations on some machines. For example, on the IBM RS/6000, there
5748 is a delay for taken branches unless the condition code register is set
5749 three instructions earlier than the conditional branch. The instruction
5750 scheduler cannot perform this optimization if it is not permitted to
5751 separate the definition and use of the condition code register.
5753 For this reason, it is possible and suggested to use a register to
5754 represent the condition code for new ports. If there is a specific
5755 condition code register in the machine, use a hard register. If the
5756 condition code or comparison result can be placed in any general register,
5757 or if there are multiple condition registers, use a pseudo register.
5758 Registers used to store the condition code value will usually have a mode
5759 that is in class @code{MODE_CC}.
5761 Alternatively, you can use @code{BImode} if the comparison operator is
5762 specified already in the compare instruction. In this case, you are not
5763 interested in most macros in this section.
5766 * CC0 Condition Codes:: Old style representation of condition codes.
5767 * MODE_CC Condition Codes:: Modern representation of condition codes.
5768 * Cond. Exec. Macros:: Macros to control conditional execution.
5771 @node CC0 Condition Codes
5772 @subsection Representation of condition codes using @code{(cc0)}
5776 The file @file{conditions.h} defines a variable @code{cc_status} to
5777 describe how the condition code was computed (in case the interpretation of
5778 the condition code depends on the instruction that it was set by). This
5779 variable contains the RTL expressions on which the condition code is
5780 currently based, and several standard flags.
5782 Sometimes additional machine-specific flags must be defined in the machine
5783 description header file. It can also add additional machine-specific
5784 information by defining @code{CC_STATUS_MDEP}.
5786 @defmac CC_STATUS_MDEP
5787 C code for a data type which is used for declaring the @code{mdep}
5788 component of @code{cc_status}. It defaults to @code{int}.
5790 This macro is not used on machines that do not use @code{cc0}.
5793 @defmac CC_STATUS_MDEP_INIT
5794 A C expression to initialize the @code{mdep} field to ``empty''.
5795 The default definition does nothing, since most machines don't use
5796 the field anyway. If you want to use the field, you should probably
5797 define this macro to initialize it.
5799 This macro is not used on machines that do not use @code{cc0}.
5802 @defmac NOTICE_UPDATE_CC (@var{exp}, @var{insn})
5803 A C compound statement to set the components of @code{cc_status}
5804 appropriately for an insn @var{insn} whose body is @var{exp}. It is
5805 this macro's responsibility to recognize insns that set the condition
5806 code as a byproduct of other activity as well as those that explicitly
5809 This macro is not used on machines that do not use @code{cc0}.
5811 If there are insns that do not set the condition code but do alter
5812 other machine registers, this macro must check to see whether they
5813 invalidate the expressions that the condition code is recorded as
5814 reflecting. For example, on the 68000, insns that store in address
5815 registers do not set the condition code, which means that usually
5816 @code{NOTICE_UPDATE_CC} can leave @code{cc_status} unaltered for such
5817 insns. But suppose that the previous insn set the condition code
5818 based on location @samp{a4@@(102)} and the current insn stores a new
5819 value in @samp{a4}. Although the condition code is not changed by
5820 this, it will no longer be true that it reflects the contents of
5821 @samp{a4@@(102)}. Therefore, @code{NOTICE_UPDATE_CC} must alter
5822 @code{cc_status} in this case to say that nothing is known about the
5823 condition code value.
5825 The definition of @code{NOTICE_UPDATE_CC} must be prepared to deal
5826 with the results of peephole optimization: insns whose patterns are
5827 @code{parallel} RTXs containing various @code{reg}, @code{mem} or
5828 constants which are just the operands. The RTL structure of these
5829 insns is not sufficient to indicate what the insns actually do. What
5830 @code{NOTICE_UPDATE_CC} should do when it sees one is just to run
5831 @code{CC_STATUS_INIT}.
5833 A possible definition of @code{NOTICE_UPDATE_CC} is to call a function
5834 that looks at an attribute (@pxref{Insn Attributes}) named, for example,
5835 @samp{cc}. This avoids having detailed information about patterns in
5836 two places, the @file{md} file and in @code{NOTICE_UPDATE_CC}.
5839 @node MODE_CC Condition Codes
5840 @subsection Representation of condition codes using registers
5844 @defmac SELECT_CC_MODE (@var{op}, @var{x}, @var{y})
5845 On many machines, the condition code may be produced by other instructions
5846 than compares, for example the branch can use directly the condition
5847 code set by a subtract instruction. However, on some machines
5848 when the condition code is set this way some bits (such as the overflow
5849 bit) are not set in the same way as a test instruction, so that a different
5850 branch instruction must be used for some conditional branches. When
5851 this happens, use the machine mode of the condition code register to
5852 record different formats of the condition code register. Modes can
5853 also be used to record which compare instruction (e.g. a signed or an
5854 unsigned comparison) produced the condition codes.
5856 If other modes than @code{CCmode} are required, add them to
5857 @file{@var{machine}-modes.def} and define @code{SELECT_CC_MODE} to choose
5858 a mode given an operand of a compare. This is needed because the modes
5859 have to be chosen not only during RTL generation but also, for example,
5860 by instruction combination. The result of @code{SELECT_CC_MODE} should
5861 be consistent with the mode used in the patterns; for example to support
5862 the case of the add on the SPARC discussed above, we have the pattern
5866 [(set (reg:CC_NOOV 0)
5868 (plus:SI (match_operand:SI 0 "register_operand" "%r")
5869 (match_operand:SI 1 "arith_operand" "rI"))
5876 together with a @code{SELECT_CC_MODE} that returns @code{CC_NOOVmode}
5877 for comparisons whose argument is a @code{plus}:
5880 #define SELECT_CC_MODE(OP,X,Y) \
5881 (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \
5882 ? ((OP == EQ || OP == NE) ? CCFPmode : CCFPEmode) \
5883 : ((GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \
5884 || GET_CODE (X) == NEG) \
5885 ? CC_NOOVmode : CCmode))
5888 Another reason to use modes is to retain information on which operands
5889 were used by the comparison; see @code{REVERSIBLE_CC_MODE} later in
5892 You should define this macro if and only if you define extra CC modes
5893 in @file{@var{machine}-modes.def}.
5896 @defmac CANONICALIZE_COMPARISON (@var{code}, @var{op0}, @var{op1})
5897 On some machines not all possible comparisons are defined, but you can
5898 convert an invalid comparison into a valid one. For example, the Alpha
5899 does not have a @code{GT} comparison, but you can use an @code{LT}
5900 comparison instead and swap the order of the operands.
5902 On such machines, define this macro to be a C statement to do any
5903 required conversions. @var{code} is the initial comparison code
5904 and @var{op0} and @var{op1} are the left and right operands of the
5905 comparison, respectively. You should modify @var{code}, @var{op0}, and
5906 @var{op1} as required.
5908 GCC will not assume that the comparison resulting from this macro is
5909 valid but will see if the resulting insn matches a pattern in the
5912 You need not define this macro if it would never change the comparison
5916 @defmac REVERSIBLE_CC_MODE (@var{mode})
5917 A C expression whose value is one if it is always safe to reverse a
5918 comparison whose mode is @var{mode}. If @code{SELECT_CC_MODE}
5919 can ever return @var{mode} for a floating-point inequality comparison,
5920 then @code{REVERSIBLE_CC_MODE (@var{mode})} must be zero.
5922 You need not define this macro if it would always returns zero or if the
5923 floating-point format is anything other than @code{IEEE_FLOAT_FORMAT}.
5924 For example, here is the definition used on the SPARC, where floating-point
5925 inequality comparisons are always given @code{CCFPEmode}:
5928 #define REVERSIBLE_CC_MODE(MODE) ((MODE) != CCFPEmode)
5932 @defmac REVERSE_CONDITION (@var{code}, @var{mode})
5933 A C expression whose value is reversed condition code of the @var{code} for
5934 comparison done in CC_MODE @var{mode}. The macro is used only in case
5935 @code{REVERSIBLE_CC_MODE (@var{mode})} is nonzero. Define this macro in case
5936 machine has some non-standard way how to reverse certain conditionals. For
5937 instance in case all floating point conditions are non-trapping, compiler may
5938 freely convert unordered compares to ordered one. Then definition may look
5942 #define REVERSE_CONDITION(CODE, MODE) \
5943 ((MODE) != CCFPmode ? reverse_condition (CODE) \
5944 : reverse_condition_maybe_unordered (CODE))
5948 @deftypefn {Target Hook} bool TARGET_FIXED_CONDITION_CODE_REGS (unsigned int *, unsigned int *)
5949 On targets which do not use @code{(cc0)}, and which use a hard
5950 register rather than a pseudo-register to hold condition codes, the
5951 regular CSE passes are often not able to identify cases in which the
5952 hard register is set to a common value. Use this hook to enable a
5953 small pass which optimizes such cases. This hook should return true
5954 to enable this pass, and it should set the integers to which its
5955 arguments point to the hard register numbers used for condition codes.
5956 When there is only one such register, as is true on most systems, the
5957 integer pointed to by the second argument should be set to
5958 @code{INVALID_REGNUM}.
5960 The default version of this hook returns false.
5963 @deftypefn {Target Hook} enum machine_mode TARGET_CC_MODES_COMPATIBLE (enum machine_mode, enum machine_mode)
5964 On targets which use multiple condition code modes in class
5965 @code{MODE_CC}, it is sometimes the case that a comparison can be
5966 validly done in more than one mode. On such a system, define this
5967 target hook to take two mode arguments and to return a mode in which
5968 both comparisons may be validly done. If there is no such mode,
5969 return @code{VOIDmode}.
5971 The default version of this hook checks whether the modes are the
5972 same. If they are, it returns that mode. If they are different, it
5973 returns @code{VOIDmode}.
5976 @node Cond. Exec. Macros
5977 @subsection Macros to control conditional execution
5978 @findex conditional execution
5981 There is one macro that may need to be defined for targets
5982 supporting conditional execution, independent of how they
5983 represent conditional branches.
5985 @defmac REVERSE_CONDEXEC_PREDICATES_P (@var{op1}, @var{op2})
5986 A C expression that returns true if the conditional execution predicate
5987 @var{op1}, a comparison operation, is the inverse of @var{op2} and vice
5988 versa. Define this to return 0 if the target has conditional execution
5989 predicates that cannot be reversed safely. There is no need to validate
5990 that the arguments of op1 and op2 are the same, this is done separately.
5991 If no expansion is specified, this macro is defined as follows:
5994 #define REVERSE_CONDEXEC_PREDICATES_P (x, y) \
5995 (GET_CODE ((x)) == reversed_comparison_code ((y), NULL))
6000 @section Describing Relative Costs of Operations
6001 @cindex costs of instructions
6002 @cindex relative costs
6003 @cindex speed of instructions
6005 These macros let you describe the relative speed of various operations
6006 on the target machine.
6008 @defmac REGISTER_MOVE_COST (@var{mode}, @var{from}, @var{to})
6009 A C expression for the cost of moving data of mode @var{mode} from a
6010 register in class @var{from} to one in class @var{to}. The classes are
6011 expressed using the enumeration values such as @code{GENERAL_REGS}. A
6012 value of 2 is the default; other values are interpreted relative to
6015 It is not required that the cost always equal 2 when @var{from} is the
6016 same as @var{to}; on some machines it is expensive to move between
6017 registers if they are not general registers.
6019 If reload sees an insn consisting of a single @code{set} between two
6020 hard registers, and if @code{REGISTER_MOVE_COST} applied to their
6021 classes returns a value of 2, reload does not check to ensure that the
6022 constraints of the insn are met. Setting a cost of other than 2 will
6023 allow reload to verify that the constraints are met. You should do this
6024 if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
6027 @defmac MEMORY_MOVE_COST (@var{mode}, @var{class}, @var{in})
6028 A C expression for the cost of moving data of mode @var{mode} between a
6029 register of class @var{class} and memory; @var{in} is zero if the value
6030 is to be written to memory, nonzero if it is to be read in. This cost
6031 is relative to those in @code{REGISTER_MOVE_COST}. If moving between
6032 registers and memory is more expensive than between two registers, you
6033 should define this macro to express the relative cost.
6035 If you do not define this macro, GCC uses a default cost of 4 plus
6036 the cost of copying via a secondary reload register, if one is
6037 needed. If your machine requires a secondary reload register to copy
6038 between memory and a register of @var{class} but the reload mechanism is
6039 more complex than copying via an intermediate, define this macro to
6040 reflect the actual cost of the move.
6042 GCC defines the function @code{memory_move_secondary_cost} if
6043 secondary reloads are needed. It computes the costs due to copying via
6044 a secondary register. If your machine copies from memory using a
6045 secondary register in the conventional way but the default base value of
6046 4 is not correct for your machine, define this macro to add some other
6047 value to the result of that function. The arguments to that function
6048 are the same as to this macro.
6051 @defmac BRANCH_COST (@var{speed_p}, @var{predictable_p})
6052 A C expression for the cost of a branch instruction. A value of 1 is the
6053 default; other values are interpreted relative to that. Parameter @var{speed_p}
6054 is true when the branch in question should be optimized for speed. When
6055 it is false, @code{BRANCH_COST} should be returning value optimal for code size
6056 rather then performance considerations. @var{predictable_p} is true for well
6057 predictable branches. On many architectures the @code{BRANCH_COST} can be
6061 Here are additional macros which do not specify precise relative costs,
6062 but only that certain actions are more expensive than GCC would
6065 @defmac SLOW_BYTE_ACCESS
6066 Define this macro as a C expression which is nonzero if accessing less
6067 than a word of memory (i.e.@: a @code{char} or a @code{short}) is no
6068 faster than accessing a word of memory, i.e., if such access
6069 require more than one instruction or if there is no difference in cost
6070 between byte and (aligned) word loads.
6072 When this macro is not defined, the compiler will access a field by
6073 finding the smallest containing object; when it is defined, a fullword
6074 load will be used if alignment permits. Unless bytes accesses are
6075 faster than word accesses, using word accesses is preferable since it
6076 may eliminate subsequent memory access if subsequent accesses occur to
6077 other fields in the same word of the structure, but to different bytes.
6080 @defmac SLOW_UNALIGNED_ACCESS (@var{mode}, @var{alignment})
6081 Define this macro to be the value 1 if memory accesses described by the
6082 @var{mode} and @var{alignment} parameters have a cost many times greater
6083 than aligned accesses, for example if they are emulated in a trap
6086 When this macro is nonzero, the compiler will act as if
6087 @code{STRICT_ALIGNMENT} were nonzero when generating code for block
6088 moves. This can cause significantly more instructions to be produced.
6089 Therefore, do not set this macro nonzero if unaligned accesses only add a
6090 cycle or two to the time for a memory access.
6092 If the value of this macro is always zero, it need not be defined. If
6093 this macro is defined, it should produce a nonzero value when
6094 @code{STRICT_ALIGNMENT} is nonzero.
6098 The threshold of number of scalar memory-to-memory move insns, @emph{below}
6099 which a sequence of insns should be generated instead of a
6100 string move insn or a library call. Increasing the value will always
6101 make code faster, but eventually incurs high cost in increased code size.
6103 Note that on machines where the corresponding move insn is a
6104 @code{define_expand} that emits a sequence of insns, this macro counts
6105 the number of such sequences.
6107 If you don't define this, a reasonable default is used.
6110 @defmac MOVE_BY_PIECES_P (@var{size}, @var{alignment})
6111 A C expression used to determine whether @code{move_by_pieces} will be used to
6112 copy a chunk of memory, or whether some other block move mechanism
6113 will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6114 than @code{MOVE_RATIO}.
6117 @defmac MOVE_MAX_PIECES
6118 A C expression used by @code{move_by_pieces} to determine the largest unit
6119 a load or store used to copy memory is. Defaults to @code{MOVE_MAX}.
6123 The threshold of number of scalar move insns, @emph{below} which a sequence
6124 of insns should be generated to clear memory instead of a string clear insn
6125 or a library call. Increasing the value will always make code faster, but
6126 eventually incurs high cost in increased code size.
6128 If you don't define this, a reasonable default is used.
6131 @defmac CLEAR_BY_PIECES_P (@var{size}, @var{alignment})
6132 A C expression used to determine whether @code{clear_by_pieces} will be used
6133 to clear a chunk of memory, or whether some other block clear mechanism
6134 will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6135 than @code{CLEAR_RATIO}.
6139 The threshold of number of scalar move insns, @emph{below} which a sequence
6140 of insns should be generated to set memory to a constant value, instead of
6141 a block set insn or a library call.
6142 Increasing the value will always make code faster, but
6143 eventually incurs high cost in increased code size.
6145 If you don't define this, it defaults to the value of @code{MOVE_RATIO}.
6148 @defmac SET_BY_PIECES_P (@var{size}, @var{alignment})
6149 A C expression used to determine whether @code{store_by_pieces} will be
6150 used to set a chunk of memory to a constant value, or whether some
6151 other mechanism will be used. Used by @code{__builtin_memset} when
6152 storing values other than constant zero.
6153 Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6154 than @code{SET_RATIO}.
6157 @defmac STORE_BY_PIECES_P (@var{size}, @var{alignment})
6158 A C expression used to determine whether @code{store_by_pieces} will be
6159 used to set a chunk of memory to a constant string value, or whether some
6160 other mechanism will be used. Used by @code{__builtin_strcpy} when
6161 called with a constant source string.
6162 Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6163 than @code{MOVE_RATIO}.
6166 @defmac USE_LOAD_POST_INCREMENT (@var{mode})
6167 A C expression used to determine whether a load postincrement is a good
6168 thing to use for a given mode. Defaults to the value of
6169 @code{HAVE_POST_INCREMENT}.
6172 @defmac USE_LOAD_POST_DECREMENT (@var{mode})
6173 A C expression used to determine whether a load postdecrement is a good
6174 thing to use for a given mode. Defaults to the value of
6175 @code{HAVE_POST_DECREMENT}.
6178 @defmac USE_LOAD_PRE_INCREMENT (@var{mode})
6179 A C expression used to determine whether a load preincrement is a good
6180 thing to use for a given mode. Defaults to the value of
6181 @code{HAVE_PRE_INCREMENT}.
6184 @defmac USE_LOAD_PRE_DECREMENT (@var{mode})
6185 A C expression used to determine whether a load predecrement is a good
6186 thing to use for a given mode. Defaults to the value of
6187 @code{HAVE_PRE_DECREMENT}.
6190 @defmac USE_STORE_POST_INCREMENT (@var{mode})
6191 A C expression used to determine whether a store postincrement is a good
6192 thing to use for a given mode. Defaults to the value of
6193 @code{HAVE_POST_INCREMENT}.
6196 @defmac USE_STORE_POST_DECREMENT (@var{mode})
6197 A C expression used to determine whether a store postdecrement is a good
6198 thing to use for a given mode. Defaults to the value of
6199 @code{HAVE_POST_DECREMENT}.
6202 @defmac USE_STORE_PRE_INCREMENT (@var{mode})
6203 This macro is used to determine whether a store preincrement is a good
6204 thing to use for a given mode. Defaults to the value of
6205 @code{HAVE_PRE_INCREMENT}.
6208 @defmac USE_STORE_PRE_DECREMENT (@var{mode})
6209 This macro is used to determine whether a store predecrement is a good
6210 thing to use for a given mode. Defaults to the value of
6211 @code{HAVE_PRE_DECREMENT}.
6214 @defmac NO_FUNCTION_CSE
6215 Define this macro if it is as good or better to call a constant
6216 function address than to call an address kept in a register.
6219 @defmac RANGE_TEST_NON_SHORT_CIRCUIT
6220 Define this macro if a non-short-circuit operation produced by
6221 @samp{fold_range_test ()} is optimal. This macro defaults to true if
6222 @code{BRANCH_COST} is greater than or equal to the value 2.
6225 @deftypefn {Target Hook} bool TARGET_RTX_COSTS (rtx @var{x}, int @var{code}, int @var{outer_code}, int *@var{total})
6226 This target hook describes the relative costs of RTL expressions.
6228 The cost may depend on the precise form of the expression, which is
6229 available for examination in @var{x}, and the rtx code of the expression
6230 in which it is contained, found in @var{outer_code}. @var{code} is the
6231 expression code---redundant, since it can be obtained with
6232 @code{GET_CODE (@var{x})}.
6234 In implementing this hook, you can use the construct
6235 @code{COSTS_N_INSNS (@var{n})} to specify a cost equal to @var{n} fast
6238 On entry to the hook, @code{*@var{total}} contains a default estimate
6239 for the cost of the expression. The hook should modify this value as
6240 necessary. Traditionally, the default costs are @code{COSTS_N_INSNS (5)}
6241 for multiplications, @code{COSTS_N_INSNS (7)} for division and modulus
6242 operations, and @code{COSTS_N_INSNS (1)} for all other operations.
6244 When optimizing for code size, i.e.@: when @code{optimize_size} is
6245 nonzero, this target hook should be used to estimate the relative
6246 size cost of an expression, again relative to @code{COSTS_N_INSNS}.
6248 The hook returns true when all subexpressions of @var{x} have been
6249 processed, and false when @code{rtx_cost} should recurse.
6252 @deftypefn {Target Hook} int TARGET_ADDRESS_COST (rtx @var{address})
6253 This hook computes the cost of an addressing mode that contains
6254 @var{address}. If not defined, the cost is computed from
6255 the @var{address} expression and the @code{TARGET_RTX_COST} hook.
6257 For most CISC machines, the default cost is a good approximation of the
6258 true cost of the addressing mode. However, on RISC machines, all
6259 instructions normally have the same length and execution time. Hence
6260 all addresses will have equal costs.
6262 In cases where more than one form of an address is known, the form with
6263 the lowest cost will be used. If multiple forms have the same, lowest,
6264 cost, the one that is the most complex will be used.
6266 For example, suppose an address that is equal to the sum of a register
6267 and a constant is used twice in the same basic block. When this macro
6268 is not defined, the address will be computed in a register and memory
6269 references will be indirect through that register. On machines where
6270 the cost of the addressing mode containing the sum is no higher than
6271 that of a simple indirect reference, this will produce an additional
6272 instruction and possibly require an additional register. Proper
6273 specification of this macro eliminates this overhead for such machines.
6275 This hook is never called with an invalid address.
6277 On machines where an address involving more than one register is as
6278 cheap as an address computation involving only one register, defining
6279 @code{TARGET_ADDRESS_COST} to reflect this can cause two registers to
6280 be live over a region of code where only one would have been if
6281 @code{TARGET_ADDRESS_COST} were not defined in that manner. This effect
6282 should be considered in the definition of this macro. Equivalent costs
6283 should probably only be given to addresses with different numbers of
6284 registers on machines with lots of registers.
6288 @section Adjusting the Instruction Scheduler
6290 The instruction scheduler may need a fair amount of machine-specific
6291 adjustment in order to produce good code. GCC provides several target
6292 hooks for this purpose. It is usually enough to define just a few of
6293 them: try the first ones in this list first.
6295 @deftypefn {Target Hook} int TARGET_SCHED_ISSUE_RATE (void)
6296 This hook returns the maximum number of instructions that can ever
6297 issue at the same time on the target machine. The default is one.
6298 Although the insn scheduler can define itself the possibility of issue
6299 an insn on the same cycle, the value can serve as an additional
6300 constraint to issue insns on the same simulated processor cycle (see
6301 hooks @samp{TARGET_SCHED_REORDER} and @samp{TARGET_SCHED_REORDER2}).
6302 This value must be constant over the entire compilation. If you need
6303 it to vary depending on what the instructions are, you must use
6304 @samp{TARGET_SCHED_VARIABLE_ISSUE}.
6307 @deftypefn {Target Hook} int TARGET_SCHED_VARIABLE_ISSUE (FILE *@var{file}, int @var{verbose}, rtx @var{insn}, int @var{more})
6308 This hook is executed by the scheduler after it has scheduled an insn
6309 from the ready list. It should return the number of insns which can
6310 still be issued in the current cycle. The default is
6311 @samp{@w{@var{more} - 1}} for insns other than @code{CLOBBER} and
6312 @code{USE}, which normally are not counted against the issue rate.
6313 You should define this hook if some insns take more machine resources
6314 than others, so that fewer insns can follow them in the same cycle.
6315 @var{file} is either a null pointer, or a stdio stream to write any
6316 debug output to. @var{verbose} is the verbose level provided by
6317 @option{-fsched-verbose-@var{n}}. @var{insn} is the instruction that
6321 @deftypefn {Target Hook} int TARGET_SCHED_ADJUST_COST (rtx @var{insn}, rtx @var{link}, rtx @var{dep_insn}, int @var{cost})
6322 This function corrects the value of @var{cost} based on the
6323 relationship between @var{insn} and @var{dep_insn} through the
6324 dependence @var{link}. It should return the new value. The default
6325 is to make no adjustment to @var{cost}. This can be used for example
6326 to specify to the scheduler using the traditional pipeline description
6327 that an output- or anti-dependence does not incur the same cost as a
6328 data-dependence. If the scheduler using the automaton based pipeline
6329 description, the cost of anti-dependence is zero and the cost of
6330 output-dependence is maximum of one and the difference of latency
6331 times of the first and the second insns. If these values are not
6332 acceptable, you could use the hook to modify them too. See also
6333 @pxref{Processor pipeline description}.
6336 @deftypefn {Target Hook} int TARGET_SCHED_ADJUST_PRIORITY (rtx @var{insn}, int @var{priority})
6337 This hook adjusts the integer scheduling priority @var{priority} of
6338 @var{insn}. It should return the new priority. Increase the priority to
6339 execute @var{insn} earlier, reduce the priority to execute @var{insn}
6340 later. Do not define this hook if you do not need to adjust the
6341 scheduling priorities of insns.
6344 @deftypefn {Target Hook} int TARGET_SCHED_REORDER (FILE *@var{file}, int @var{verbose}, rtx *@var{ready}, int *@var{n_readyp}, int @var{clock})
6345 This hook is executed by the scheduler after it has scheduled the ready
6346 list, to allow the machine description to reorder it (for example to
6347 combine two small instructions together on @samp{VLIW} machines).
6348 @var{file} is either a null pointer, or a stdio stream to write any
6349 debug output to. @var{verbose} is the verbose level provided by
6350 @option{-fsched-verbose-@var{n}}. @var{ready} is a pointer to the ready
6351 list of instructions that are ready to be scheduled. @var{n_readyp} is
6352 a pointer to the number of elements in the ready list. The scheduler
6353 reads the ready list in reverse order, starting with
6354 @var{ready}[@var{*n_readyp}-1] and going to @var{ready}[0]. @var{clock}
6355 is the timer tick of the scheduler. You may modify the ready list and
6356 the number of ready insns. The return value is the number of insns that
6357 can issue this cycle; normally this is just @code{issue_rate}. See also
6358 @samp{TARGET_SCHED_REORDER2}.
6361 @deftypefn {Target Hook} int TARGET_SCHED_REORDER2 (FILE *@var{file}, int @var{verbose}, rtx *@var{ready}, int *@var{n_ready}, @var{clock})
6362 Like @samp{TARGET_SCHED_REORDER}, but called at a different time. That
6363 function is called whenever the scheduler starts a new cycle. This one
6364 is called once per iteration over a cycle, immediately after
6365 @samp{TARGET_SCHED_VARIABLE_ISSUE}; it can reorder the ready list and
6366 return the number of insns to be scheduled in the same cycle. Defining
6367 this hook can be useful if there are frequent situations where
6368 scheduling one insn causes other insns to become ready in the same
6369 cycle. These other insns can then be taken into account properly.
6372 @deftypefn {Target Hook} void TARGET_SCHED_DEPENDENCIES_EVALUATION_HOOK (rtx @var{head}, rtx @var{tail})
6373 This hook is called after evaluation forward dependencies of insns in
6374 chain given by two parameter values (@var{head} and @var{tail}
6375 correspondingly) but before insns scheduling of the insn chain. For
6376 example, it can be used for better insn classification if it requires
6377 analysis of dependencies. This hook can use backward and forward
6378 dependencies of the insn scheduler because they are already
6382 @deftypefn {Target Hook} void TARGET_SCHED_INIT (FILE *@var{file}, int @var{verbose}, int @var{max_ready})
6383 This hook is executed by the scheduler at the beginning of each block of
6384 instructions that are to be scheduled. @var{file} is either a null
6385 pointer, or a stdio stream to write any debug output to. @var{verbose}
6386 is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6387 @var{max_ready} is the maximum number of insns in the current scheduling
6388 region that can be live at the same time. This can be used to allocate
6389 scratch space if it is needed, e.g.@: by @samp{TARGET_SCHED_REORDER}.
6392 @deftypefn {Target Hook} void TARGET_SCHED_FINISH (FILE *@var{file}, int @var{verbose})
6393 This hook is executed by the scheduler at the end of each block of
6394 instructions that are to be scheduled. It can be used to perform
6395 cleanup of any actions done by the other scheduling hooks. @var{file}
6396 is either a null pointer, or a stdio stream to write any debug output
6397 to. @var{verbose} is the verbose level provided by
6398 @option{-fsched-verbose-@var{n}}.
6401 @deftypefn {Target Hook} void TARGET_SCHED_INIT_GLOBAL (FILE *@var{file}, int @var{verbose}, int @var{old_max_uid})
6402 This hook is executed by the scheduler after function level initializations.
6403 @var{file} is either a null pointer, or a stdio stream to write any debug output to.
6404 @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6405 @var{old_max_uid} is the maximum insn uid when scheduling begins.
6408 @deftypefn {Target Hook} void TARGET_SCHED_FINISH_GLOBAL (FILE *@var{file}, int @var{verbose})
6409 This is the cleanup hook corresponding to @code{TARGET_SCHED_INIT_GLOBAL}.
6410 @var{file} is either a null pointer, or a stdio stream to write any debug output to.
6411 @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6414 @deftypefn {Target Hook} int TARGET_SCHED_DFA_PRE_CYCLE_INSN (void)
6415 The hook returns an RTL insn. The automaton state used in the
6416 pipeline hazard recognizer is changed as if the insn were scheduled
6417 when the new simulated processor cycle starts. Usage of the hook may
6418 simplify the automaton pipeline description for some @acronym{VLIW}
6419 processors. If the hook is defined, it is used only for the automaton
6420 based pipeline description. The default is not to change the state
6421 when the new simulated processor cycle starts.
6424 @deftypefn {Target Hook} void TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN (void)
6425 The hook can be used to initialize data used by the previous hook.
6428 @deftypefn {Target Hook} int TARGET_SCHED_DFA_POST_CYCLE_INSN (void)
6429 The hook is analogous to @samp{TARGET_SCHED_DFA_PRE_CYCLE_INSN} but used
6430 to changed the state as if the insn were scheduled when the new
6431 simulated processor cycle finishes.
6434 @deftypefn {Target Hook} void TARGET_SCHED_INIT_DFA_POST_CYCLE_INSN (void)
6435 The hook is analogous to @samp{TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN} but
6436 used to initialize data used by the previous hook.
6439 @deftypefn {Target Hook} void TARGET_SCHED_DFA_PRE_CYCLE_ADVANCE (void)
6440 The hook to notify target that the current simulated cycle is about to finish.
6441 The hook is analogous to @samp{TARGET_SCHED_DFA_PRE_CYCLE_INSN} but used
6442 to change the state in more complicated situations - e.g., when advancing
6443 state on a single insn is not enough.
6446 @deftypefn {Target Hook} void TARGET_SCHED_DFA_POST_CYCLE_ADVANCE (void)
6447 The hook to notify target that new simulated cycle has just started.
6448 The hook is analogous to @samp{TARGET_SCHED_DFA_POST_CYCLE_INSN} but used
6449 to change the state in more complicated situations - e.g., when advancing
6450 state on a single insn is not enough.
6453 @deftypefn {Target Hook} int TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD (void)
6454 This hook controls better choosing an insn from the ready insn queue
6455 for the @acronym{DFA}-based insn scheduler. Usually the scheduler
6456 chooses the first insn from the queue. If the hook returns a positive
6457 value, an additional scheduler code tries all permutations of
6458 @samp{TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD ()}
6459 subsequent ready insns to choose an insn whose issue will result in
6460 maximal number of issued insns on the same cycle. For the
6461 @acronym{VLIW} processor, the code could actually solve the problem of
6462 packing simple insns into the @acronym{VLIW} insn. Of course, if the
6463 rules of @acronym{VLIW} packing are described in the automaton.
6465 This code also could be used for superscalar @acronym{RISC}
6466 processors. Let us consider a superscalar @acronym{RISC} processor
6467 with 3 pipelines. Some insns can be executed in pipelines @var{A} or
6468 @var{B}, some insns can be executed only in pipelines @var{B} or
6469 @var{C}, and one insn can be executed in pipeline @var{B}. The
6470 processor may issue the 1st insn into @var{A} and the 2nd one into
6471 @var{B}. In this case, the 3rd insn will wait for freeing @var{B}
6472 until the next cycle. If the scheduler issues the 3rd insn the first,
6473 the processor could issue all 3 insns per cycle.
6475 Actually this code demonstrates advantages of the automaton based
6476 pipeline hazard recognizer. We try quickly and easy many insn
6477 schedules to choose the best one.
6479 The default is no multipass scheduling.
6482 @deftypefn {Target Hook} int TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD (rtx)
6484 This hook controls what insns from the ready insn queue will be
6485 considered for the multipass insn scheduling. If the hook returns
6486 zero for insn passed as the parameter, the insn will be not chosen to
6489 The default is that any ready insns can be chosen to be issued.
6492 @deftypefn {Target Hook} int TARGET_SCHED_DFA_NEW_CYCLE (FILE *, int, rtx, int, int, int *)
6494 This hook is called by the insn scheduler before issuing insn passed
6495 as the third parameter on given cycle. If the hook returns nonzero,
6496 the insn is not issued on given processors cycle. Instead of that,
6497 the processor cycle is advanced. If the value passed through the last
6498 parameter is zero, the insn ready queue is not sorted on the new cycle
6499 start as usually. The first parameter passes file for debugging
6500 output. The second one passes the scheduler verbose level of the
6501 debugging output. The forth and the fifth parameter values are
6502 correspondingly processor cycle on which the previous insn has been
6503 issued and the current processor cycle.
6506 @deftypefn {Target Hook} bool TARGET_SCHED_IS_COSTLY_DEPENDENCE (struct dep_def *@var{_dep}, int @var{cost}, int @var{distance})
6507 This hook is used to define which dependences are considered costly by
6508 the target, so costly that it is not advisable to schedule the insns that
6509 are involved in the dependence too close to one another. The parameters
6510 to this hook are as follows: The first parameter @var{_dep} is the dependence
6511 being evaluated. The second parameter @var{cost} is the cost of the
6512 dependence, and the third
6513 parameter @var{distance} is the distance in cycles between the two insns.
6514 The hook returns @code{true} if considering the distance between the two
6515 insns the dependence between them is considered costly by the target,
6516 and @code{false} otherwise.
6518 Defining this hook can be useful in multiple-issue out-of-order machines,
6519 where (a) it's practically hopeless to predict the actual data/resource
6520 delays, however: (b) there's a better chance to predict the actual grouping
6521 that will be formed, and (c) correctly emulating the grouping can be very
6522 important. In such targets one may want to allow issuing dependent insns
6523 closer to one another---i.e., closer than the dependence distance; however,
6524 not in cases of "costly dependences", which this hooks allows to define.
6527 @deftypefn {Target Hook} void TARGET_SCHED_H_I_D_EXTENDED (void)
6528 This hook is called by the insn scheduler after emitting a new instruction to
6529 the instruction stream. The hook notifies a target backend to extend its
6530 per instruction data structures.
6533 @deftypefn {Target Hook} void * TARGET_SCHED_ALLOC_SCHED_CONTEXT (void)
6534 Return a pointer to a store large enough to hold target scheduling context.
6537 @deftypefn {Target Hook} void TARGET_SCHED_INIT_SCHED_CONTEXT (void *@var{tc}, bool @var{clean_p})
6538 Initialize store pointed to by @var{tc} to hold target scheduling context.
6539 It @var{clean_p} is true then initialize @var{tc} as if scheduler is at the
6540 beginning of the block. Otherwise, make a copy of the current context in
6544 @deftypefn {Target Hook} void TARGET_SCHED_SET_SCHED_CONTEXT (void *@var{tc})
6545 Copy target scheduling context pointer to by @var{tc} to the current context.
6548 @deftypefn {Target Hook} void TARGET_SCHED_CLEAR_SCHED_CONTEXT (void *@var{tc})
6549 Deallocate internal data in target scheduling context pointed to by @var{tc}.
6552 @deftypefn {Target Hook} void TARGET_SCHED_FREE_SCHED_CONTEXT (void *@var{tc})
6553 Deallocate a store for target scheduling context pointed to by @var{tc}.
6556 @deftypefn {Target Hook} void * TARGET_SCHED_ALLOC_SCHED_CONTEXT (void)
6557 Return a pointer to a store large enough to hold target scheduling context.
6560 @deftypefn {Target Hook} void TARGET_SCHED_INIT_SCHED_CONTEXT (void *@var{tc}, bool @var{clean_p})
6561 Initialize store pointed to by @var{tc} to hold target scheduling context.
6562 It @var{clean_p} is true then initialize @var{tc} as if scheduler is at the
6563 beginning of the block. Otherwise, make a copy of the current context in
6567 @deftypefn {Target Hook} void TARGET_SCHED_SET_SCHED_CONTEXT (void *@var{tc})
6568 Copy target scheduling context pointer to by @var{tc} to the current context.
6571 @deftypefn {Target Hook} void TARGET_SCHED_CLEAR_SCHED_CONTEXT (void *@var{tc})
6572 Deallocate internal data in target scheduling context pointed to by @var{tc}.
6575 @deftypefn {Target Hook} void TARGET_SCHED_FREE_SCHED_CONTEXT (void *@var{tc})
6576 Deallocate a store for target scheduling context pointed to by @var{tc}.
6579 @deftypefn {Target Hook} int TARGET_SCHED_SPECULATE_INSN (rtx @var{insn}, int @var{request}, rtx *@var{new_pat})
6580 This hook is called by the insn scheduler when @var{insn} has only
6581 speculative dependencies and therefore can be scheduled speculatively.
6582 The hook is used to check if the pattern of @var{insn} has a speculative
6583 version and, in case of successful check, to generate that speculative
6584 pattern. The hook should return 1, if the instruction has a speculative form,
6585 or -1, if it doesn't. @var{request} describes the type of requested
6586 speculation. If the return value equals 1 then @var{new_pat} is assigned
6587 the generated speculative pattern.
6590 @deftypefn {Target Hook} int TARGET_SCHED_NEEDS_BLOCK_P (rtx @var{insn})
6591 This hook is called by the insn scheduler during generation of recovery code
6592 for @var{insn}. It should return nonzero, if the corresponding check
6593 instruction should branch to recovery code, or zero otherwise.
6596 @deftypefn {Target Hook} rtx TARGET_SCHED_GEN_CHECK (rtx @var{insn}, rtx @var{label}, int @var{mutate_p})
6597 This hook is called by the insn scheduler to generate a pattern for recovery
6598 check instruction. If @var{mutate_p} is zero, then @var{insn} is a
6599 speculative instruction for which the check should be generated.
6600 @var{label} is either a label of a basic block, where recovery code should
6601 be emitted, or a null pointer, when requested check doesn't branch to
6602 recovery code (a simple check). If @var{mutate_p} is nonzero, then
6603 a pattern for a branchy check corresponding to a simple check denoted by
6604 @var{insn} should be generated. In this case @var{label} can't be null.
6607 @deftypefn {Target Hook} int TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD_SPEC (rtx @var{insn})
6608 This hook is used as a workaround for
6609 @samp{TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD} not being
6610 called on the first instruction of the ready list. The hook is used to
6611 discard speculative instruction that stand first in the ready list from
6612 being scheduled on the current cycle. For non-speculative instructions,
6613 the hook should always return nonzero. For example, in the ia64 backend
6614 the hook is used to cancel data speculative insns when the ALAT table
6618 @deftypefn {Target Hook} void TARGET_SCHED_SET_SCHED_FLAGS (unsigned int *@var{flags}, spec_info_t @var{spec_info})
6619 This hook is used by the insn scheduler to find out what features should be
6620 enabled/used. @var{flags} initially may have either the SCHED_RGN or SCHED_EBB
6621 bit set. This denotes the scheduler pass for which the data should be
6622 provided. The target backend should modify @var{flags} by modifying
6623 the bits corresponding to the following features: USE_DEPS_LIST, USE_GLAT,
6624 DETACH_LIFE_INFO, and DO_SPECULATION@. For the DO_SPECULATION feature
6625 an additional structure @var{spec_info} should be filled by the target.
6626 The structure describes speculation types that can be used in the scheduler.
6629 @deftypefn {Target Hook} int TARGET_SCHED_SMS_RES_MII (struct ddg *@var{g})
6630 This hook is called by the swing modulo scheduler to calculate a
6631 resource-based lower bound which is based on the resources available in
6632 the machine and the resources required by each instruction. The target
6633 backend can use @var{g} to calculate such bound. A very simple lower
6634 bound will be used in case this hook is not implemented: the total number
6635 of instructions divided by the issue rate.
6639 @section Dividing the Output into Sections (Texts, Data, @dots{})
6640 @c the above section title is WAY too long. maybe cut the part between
6641 @c the (...)? --mew 10feb93
6643 An object file is divided into sections containing different types of
6644 data. In the most common case, there are three sections: the @dfn{text
6645 section}, which holds instructions and read-only data; the @dfn{data
6646 section}, which holds initialized writable data; and the @dfn{bss
6647 section}, which holds uninitialized data. Some systems have other kinds
6650 @file{varasm.c} provides several well-known sections, such as
6651 @code{text_section}, @code{data_section} and @code{bss_section}.
6652 The normal way of controlling a @code{@var{foo}_section} variable
6653 is to define the associated @code{@var{FOO}_SECTION_ASM_OP} macro,
6654 as described below. The macros are only read once, when @file{varasm.c}
6655 initializes itself, so their values must be run-time constants.
6656 They may however depend on command-line flags.
6658 @emph{Note:} Some run-time files, such @file{crtstuff.c}, also make
6659 use of the @code{@var{FOO}_SECTION_ASM_OP} macros, and expect them
6660 to be string literals.
6662 Some assemblers require a different string to be written every time a
6663 section is selected. If your assembler falls into this category, you
6664 should define the @code{TARGET_ASM_INIT_SECTIONS} hook and use
6665 @code{get_unnamed_section} to set up the sections.
6667 You must always create a @code{text_section}, either by defining
6668 @code{TEXT_SECTION_ASM_OP} or by initializing @code{text_section}
6669 in @code{TARGET_ASM_INIT_SECTIONS}. The same is true of
6670 @code{data_section} and @code{DATA_SECTION_ASM_OP}. If you do not
6671 create a distinct @code{readonly_data_section}, the default is to
6672 reuse @code{text_section}.
6674 All the other @file{varasm.c} sections are optional, and are null
6675 if the target does not provide them.
6677 @defmac TEXT_SECTION_ASM_OP
6678 A C expression whose value is a string, including spacing, containing the
6679 assembler operation that should precede instructions and read-only data.
6680 Normally @code{"\t.text"} is right.
6683 @defmac HOT_TEXT_SECTION_NAME
6684 If defined, a C string constant for the name of the section containing most
6685 frequently executed functions of the program. If not defined, GCC will provide
6686 a default definition if the target supports named sections.
6689 @defmac UNLIKELY_EXECUTED_TEXT_SECTION_NAME
6690 If defined, a C string constant for the name of the section containing unlikely
6691 executed functions in the program.
6694 @defmac DATA_SECTION_ASM_OP
6695 A C expression whose value is a string, including spacing, containing the
6696 assembler operation to identify the following data as writable initialized
6697 data. Normally @code{"\t.data"} is right.
6700 @defmac SDATA_SECTION_ASM_OP
6701 If defined, a C expression whose value is a string, including spacing,
6702 containing the assembler operation to identify the following data as
6703 initialized, writable small data.
6706 @defmac READONLY_DATA_SECTION_ASM_OP
6707 A C expression whose value is a string, including spacing, containing the
6708 assembler operation to identify the following data as read-only initialized
6712 @defmac BSS_SECTION_ASM_OP
6713 If defined, a C expression whose value is a string, including spacing,
6714 containing the assembler operation to identify the following data as
6715 uninitialized global data. If not defined, and neither
6716 @code{ASM_OUTPUT_BSS} nor @code{ASM_OUTPUT_ALIGNED_BSS} are defined,
6717 uninitialized global data will be output in the data section if
6718 @option{-fno-common} is passed, otherwise @code{ASM_OUTPUT_COMMON} will be
6722 @defmac SBSS_SECTION_ASM_OP
6723 If defined, a C expression whose value is a string, including spacing,
6724 containing the assembler operation to identify the following data as
6725 uninitialized, writable small data.
6728 @defmac INIT_SECTION_ASM_OP
6729 If defined, a C expression whose value is a string, including spacing,
6730 containing the assembler operation to identify the following data as
6731 initialization code. If not defined, GCC will assume such a section does
6732 not exist. This section has no corresponding @code{init_section}
6733 variable; it is used entirely in runtime code.
6736 @defmac FINI_SECTION_ASM_OP
6737 If defined, a C expression whose value is a string, including spacing,
6738 containing the assembler operation to identify the following data as
6739 finalization code. If not defined, GCC will assume such a section does
6740 not exist. This section has no corresponding @code{fini_section}
6741 variable; it is used entirely in runtime code.
6744 @defmac INIT_ARRAY_SECTION_ASM_OP
6745 If defined, a C expression whose value is a string, including spacing,
6746 containing the assembler operation to identify the following data as
6747 part of the @code{.init_array} (or equivalent) section. If not
6748 defined, GCC will assume such a section does not exist. Do not define
6749 both this macro and @code{INIT_SECTION_ASM_OP}.
6752 @defmac FINI_ARRAY_SECTION_ASM_OP
6753 If defined, a C expression whose value is a string, including spacing,
6754 containing the assembler operation to identify the following data as
6755 part of the @code{.fini_array} (or equivalent) section. If not
6756 defined, GCC will assume such a section does not exist. Do not define
6757 both this macro and @code{FINI_SECTION_ASM_OP}.
6760 @defmac CRT_CALL_STATIC_FUNCTION (@var{section_op}, @var{function})
6761 If defined, an ASM statement that switches to a different section
6762 via @var{section_op}, calls @var{function}, and switches back to
6763 the text section. This is used in @file{crtstuff.c} if
6764 @code{INIT_SECTION_ASM_OP} or @code{FINI_SECTION_ASM_OP} to calls
6765 to initialization and finalization functions from the init and fini
6766 sections. By default, this macro uses a simple function call. Some
6767 ports need hand-crafted assembly code to avoid dependencies on
6768 registers initialized in the function prologue or to ensure that
6769 constant pools don't end up too far way in the text section.
6772 @defmac TARGET_LIBGCC_SDATA_SECTION
6773 If defined, a string which names the section into which small
6774 variables defined in crtstuff and libgcc should go. This is useful
6775 when the target has options for optimizing access to small data, and
6776 you want the crtstuff and libgcc routines to be conservative in what
6777 they expect of your application yet liberal in what your application
6778 expects. For example, for targets with a @code{.sdata} section (like
6779 MIPS), you could compile crtstuff with @code{-G 0} so that it doesn't
6780 require small data support from your application, but use this macro
6781 to put small data into @code{.sdata} so that your application can
6782 access these variables whether it uses small data or not.
6785 @defmac FORCE_CODE_SECTION_ALIGN
6786 If defined, an ASM statement that aligns a code section to some
6787 arbitrary boundary. This is used to force all fragments of the
6788 @code{.init} and @code{.fini} sections to have to same alignment
6789 and thus prevent the linker from having to add any padding.
6792 @defmac JUMP_TABLES_IN_TEXT_SECTION
6793 Define this macro to be an expression with a nonzero value if jump
6794 tables (for @code{tablejump} insns) should be output in the text
6795 section, along with the assembler instructions. Otherwise, the
6796 readonly data section is used.
6798 This macro is irrelevant if there is no separate readonly data section.
6801 @deftypefn {Target Hook} void TARGET_ASM_INIT_SECTIONS (void)
6802 Define this hook if you need to do something special to set up the
6803 @file{varasm.c} sections, or if your target has some special sections
6804 of its own that you need to create.
6806 GCC calls this hook after processing the command line, but before writing
6807 any assembly code, and before calling any of the section-returning hooks
6811 @deftypefn {Target Hook} TARGET_ASM_RELOC_RW_MASK (void)
6812 Return a mask describing how relocations should be treated when
6813 selecting sections. Bit 1 should be set if global relocations
6814 should be placed in a read-write section; bit 0 should be set if
6815 local relocations should be placed in a read-write section.
6817 The default version of this function returns 3 when @option{-fpic}
6818 is in effect, and 0 otherwise. The hook is typically redefined
6819 when the target cannot support (some kinds of) dynamic relocations
6820 in read-only sections even in executables.
6823 @deftypefn {Target Hook} {section *} TARGET_ASM_SELECT_SECTION (tree @var{exp}, int @var{reloc}, unsigned HOST_WIDE_INT @var{align})
6824 Return the section into which @var{exp} should be placed. You can
6825 assume that @var{exp} is either a @code{VAR_DECL} node or a constant of
6826 some sort. @var{reloc} indicates whether the initial value of @var{exp}
6827 requires link-time relocations. Bit 0 is set when variable contains
6828 local relocations only, while bit 1 is set for global relocations.
6829 @var{align} is the constant alignment in bits.
6831 The default version of this function takes care of putting read-only
6832 variables in @code{readonly_data_section}.
6834 See also @var{USE_SELECT_SECTION_FOR_FUNCTIONS}.
6837 @defmac USE_SELECT_SECTION_FOR_FUNCTIONS
6838 Define this macro if you wish TARGET_ASM_SELECT_SECTION to be called
6839 for @code{FUNCTION_DECL}s as well as for variables and constants.
6841 In the case of a @code{FUNCTION_DECL}, @var{reloc} will be zero if the
6842 function has been determined to be likely to be called, and nonzero if
6843 it is unlikely to be called.
6846 @deftypefn {Target Hook} void TARGET_ASM_UNIQUE_SECTION (tree @var{decl}, int @var{reloc})
6847 Build up a unique section name, expressed as a @code{STRING_CST} node,
6848 and assign it to @samp{DECL_SECTION_NAME (@var{decl})}.
6849 As with @code{TARGET_ASM_SELECT_SECTION}, @var{reloc} indicates whether
6850 the initial value of @var{exp} requires link-time relocations.
6852 The default version of this function appends the symbol name to the
6853 ELF section name that would normally be used for the symbol. For
6854 example, the function @code{foo} would be placed in @code{.text.foo}.
6855 Whatever the actual target object format, this is often good enough.
6858 @deftypefn {Target Hook} {section *} TARGET_ASM_FUNCTION_RODATA_SECTION (tree @var{decl})
6859 Return the readonly data section associated with
6860 @samp{DECL_SECTION_NAME (@var{decl})}.
6861 The default version of this function selects @code{.gnu.linkonce.r.name} if
6862 the function's section is @code{.gnu.linkonce.t.name}, @code{.rodata.name}
6863 if function is in @code{.text.name}, and the normal readonly-data section
6867 @deftypefn {Target Hook} {section *} TARGET_ASM_SELECT_RTX_SECTION (enum machine_mode @var{mode}, rtx @var{x}, unsigned HOST_WIDE_INT @var{align})
6868 Return the section into which a constant @var{x}, of mode @var{mode},
6869 should be placed. You can assume that @var{x} is some kind of
6870 constant in RTL@. The argument @var{mode} is redundant except in the
6871 case of a @code{const_int} rtx. @var{align} is the constant alignment
6874 The default version of this function takes care of putting symbolic
6875 constants in @code{flag_pic} mode in @code{data_section} and everything
6876 else in @code{readonly_data_section}.
6879 @deftypefn {Target Hook} void TARGET_MANGLE_DECL_ASSEMBLER_NAME (tree @var{decl}, tree @var{id})
6880 Define this hook if you need to postprocess the assembler name generated
6881 by target-independent code. The @var{id} provided to this hook will be
6882 the computed name (e.g., the macro @code{DECL_NAME} of the @var{decl} in C,
6883 or the mangled name of the @var{decl} in C++). The return value of the
6884 hook is an @code{IDENTIFIER_NODE} for the appropriate mangled name on
6885 your target system. The default implementation of this hook just
6886 returns the @var{id} provided.
6889 @deftypefn {Target Hook} void TARGET_ENCODE_SECTION_INFO (tree @var{decl}, rtx @var{rtl}, int @var{new_decl_p})
6890 Define this hook if references to a symbol or a constant must be
6891 treated differently depending on something about the variable or
6892 function named by the symbol (such as what section it is in).
6894 The hook is executed immediately after rtl has been created for
6895 @var{decl}, which may be a variable or function declaration or
6896 an entry in the constant pool. In either case, @var{rtl} is the
6897 rtl in question. Do @emph{not} use @code{DECL_RTL (@var{decl})}
6898 in this hook; that field may not have been initialized yet.
6900 In the case of a constant, it is safe to assume that the rtl is
6901 a @code{mem} whose address is a @code{symbol_ref}. Most decls
6902 will also have this form, but that is not guaranteed. Global
6903 register variables, for instance, will have a @code{reg} for their
6904 rtl. (Normally the right thing to do with such unusual rtl is
6907 The @var{new_decl_p} argument will be true if this is the first time
6908 that @code{TARGET_ENCODE_SECTION_INFO} has been invoked on this decl. It will
6909 be false for subsequent invocations, which will happen for duplicate
6910 declarations. Whether or not anything must be done for the duplicate
6911 declaration depends on whether the hook examines @code{DECL_ATTRIBUTES}.
6912 @var{new_decl_p} is always true when the hook is called for a constant.
6914 @cindex @code{SYMBOL_REF_FLAG}, in @code{TARGET_ENCODE_SECTION_INFO}
6915 The usual thing for this hook to do is to record flags in the
6916 @code{symbol_ref}, using @code{SYMBOL_REF_FLAG} or @code{SYMBOL_REF_FLAGS}.
6917 Historically, the name string was modified if it was necessary to
6918 encode more than one bit of information, but this practice is now
6919 discouraged; use @code{SYMBOL_REF_FLAGS}.
6921 The default definition of this hook, @code{default_encode_section_info}
6922 in @file{varasm.c}, sets a number of commonly-useful bits in
6923 @code{SYMBOL_REF_FLAGS}. Check whether the default does what you need
6924 before overriding it.
6927 @deftypefn {Target Hook} const char *TARGET_STRIP_NAME_ENCODING (const char *name)
6928 Decode @var{name} and return the real name part, sans
6929 the characters that @code{TARGET_ENCODE_SECTION_INFO}
6933 @deftypefn {Target Hook} bool TARGET_IN_SMALL_DATA_P (tree @var{exp})
6934 Returns true if @var{exp} should be placed into a ``small data'' section.
6935 The default version of this hook always returns false.
6938 @deftypevr {Target Hook} bool TARGET_HAVE_SRODATA_SECTION
6939 Contains the value true if the target places read-only
6940 ``small data'' into a separate section. The default value is false.
6943 @deftypefn {Target Hook} bool TARGET_BINDS_LOCAL_P (tree @var{exp})
6944 Returns true if @var{exp} names an object for which name resolution
6945 rules must resolve to the current ``module'' (dynamic shared library
6946 or executable image).
6948 The default version of this hook implements the name resolution rules
6949 for ELF, which has a looser model of global name binding than other
6950 currently supported object file formats.
6953 @deftypevr {Target Hook} bool TARGET_HAVE_TLS
6954 Contains the value true if the target supports thread-local storage.
6955 The default value is false.
6960 @section Position Independent Code
6961 @cindex position independent code
6964 This section describes macros that help implement generation of position
6965 independent code. Simply defining these macros is not enough to
6966 generate valid PIC; you must also add support to the hook
6967 @code{TARGET_LEGITIMATE_ADDRESS_P} and to the macro
6968 @code{PRINT_OPERAND_ADDRESS}, as well as @code{LEGITIMIZE_ADDRESS}. You
6969 must modify the definition of @samp{movsi} to do something appropriate
6970 when the source operand contains a symbolic address. You may also
6971 need to alter the handling of switch statements so that they use
6973 @c i rearranged the order of the macros above to try to force one of
6974 @c them to the next line, to eliminate an overfull hbox. --mew 10feb93
6976 @defmac PIC_OFFSET_TABLE_REGNUM
6977 The register number of the register used to address a table of static
6978 data addresses in memory. In some cases this register is defined by a
6979 processor's ``application binary interface'' (ABI)@. When this macro
6980 is defined, RTL is generated for this register once, as with the stack
6981 pointer and frame pointer registers. If this macro is not defined, it
6982 is up to the machine-dependent files to allocate such a register (if
6983 necessary). Note that this register must be fixed when in use (e.g.@:
6984 when @code{flag_pic} is true).
6987 @defmac PIC_OFFSET_TABLE_REG_CALL_CLOBBERED
6988 Define this macro if the register defined by
6989 @code{PIC_OFFSET_TABLE_REGNUM} is clobbered by calls. Do not define
6990 this macro if @code{PIC_OFFSET_TABLE_REGNUM} is not defined.
6993 @defmac LEGITIMATE_PIC_OPERAND_P (@var{x})
6994 A C expression that is nonzero if @var{x} is a legitimate immediate
6995 operand on the target machine when generating position independent code.
6996 You can assume that @var{x} satisfies @code{CONSTANT_P}, so you need not
6997 check this. You can also assume @var{flag_pic} is true, so you need not
6998 check it either. You need not define this macro if all constants
6999 (including @code{SYMBOL_REF}) can be immediate operands when generating
7000 position independent code.
7003 @node Assembler Format
7004 @section Defining the Output Assembler Language
7006 This section describes macros whose principal purpose is to describe how
7007 to write instructions in assembler language---rather than what the
7011 * File Framework:: Structural information for the assembler file.
7012 * Data Output:: Output of constants (numbers, strings, addresses).
7013 * Uninitialized Data:: Output of uninitialized variables.
7014 * Label Output:: Output and generation of labels.
7015 * Initialization:: General principles of initialization
7016 and termination routines.
7017 * Macros for Initialization::
7018 Specific macros that control the handling of
7019 initialization and termination routines.
7020 * Instruction Output:: Output of actual instructions.
7021 * Dispatch Tables:: Output of jump tables.
7022 * Exception Region Output:: Output of exception region code.
7023 * Alignment Output:: Pseudo ops for alignment and skipping data.
7026 @node File Framework
7027 @subsection The Overall Framework of an Assembler File
7028 @cindex assembler format
7029 @cindex output of assembler code
7031 @c prevent bad page break with this line
7032 This describes the overall framework of an assembly file.
7034 @deftypefn {Target Hook} void TARGET_ASM_FILE_START ()
7035 @findex default_file_start
7036 Output to @code{asm_out_file} any text which the assembler expects to
7037 find at the beginning of a file. The default behavior is controlled
7038 by two flags, documented below. Unless your target's assembler is
7039 quite unusual, if you override the default, you should call
7040 @code{default_file_start} at some point in your target hook. This
7041 lets other target files rely on these variables.
7044 @deftypevr {Target Hook} bool TARGET_ASM_FILE_START_APP_OFF
7045 If this flag is true, the text of the macro @code{ASM_APP_OFF} will be
7046 printed as the very first line in the assembly file, unless
7047 @option{-fverbose-asm} is in effect. (If that macro has been defined
7048 to the empty string, this variable has no effect.) With the normal
7049 definition of @code{ASM_APP_OFF}, the effect is to notify the GNU
7050 assembler that it need not bother stripping comments or extra
7051 whitespace from its input. This allows it to work a bit faster.
7053 The default is false. You should not set it to true unless you have
7054 verified that your port does not generate any extra whitespace or
7055 comments that will cause GAS to issue errors in NO_APP mode.
7058 @deftypevr {Target Hook} bool TARGET_ASM_FILE_START_FILE_DIRECTIVE
7059 If this flag is true, @code{output_file_directive} will be called
7060 for the primary source file, immediately after printing
7061 @code{ASM_APP_OFF} (if that is enabled). Most ELF assemblers expect
7062 this to be done. The default is false.
7065 @deftypefn {Target Hook} void TARGET_ASM_FILE_END ()
7066 Output to @code{asm_out_file} any text which the assembler expects
7067 to find at the end of a file. The default is to output nothing.
7070 @deftypefun void file_end_indicate_exec_stack ()
7071 Some systems use a common convention, the @samp{.note.GNU-stack}
7072 special section, to indicate whether or not an object file relies on
7073 the stack being executable. If your system uses this convention, you
7074 should define @code{TARGET_ASM_FILE_END} to this function. If you
7075 need to do other things in that hook, have your hook function call
7079 @defmac ASM_COMMENT_START
7080 A C string constant describing how to begin a comment in the target
7081 assembler language. The compiler assumes that the comment will end at
7082 the end of the line.
7086 A C string constant for text to be output before each @code{asm}
7087 statement or group of consecutive ones. Normally this is
7088 @code{"#APP"}, which is a comment that has no effect on most
7089 assemblers but tells the GNU assembler that it must check the lines
7090 that follow for all valid assembler constructs.
7094 A C string constant for text to be output after each @code{asm}
7095 statement or group of consecutive ones. Normally this is
7096 @code{"#NO_APP"}, which tells the GNU assembler to resume making the
7097 time-saving assumptions that are valid for ordinary compiler output.
7100 @defmac ASM_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name})
7101 A C statement to output COFF information or DWARF debugging information
7102 which indicates that filename @var{name} is the current source file to
7103 the stdio stream @var{stream}.
7105 This macro need not be defined if the standard form of output
7106 for the file format in use is appropriate.
7109 @defmac OUTPUT_QUOTED_STRING (@var{stream}, @var{string})
7110 A C statement to output the string @var{string} to the stdio stream
7111 @var{stream}. If you do not call the function @code{output_quoted_string}
7112 in your config files, GCC will only call it to output filenames to
7113 the assembler source. So you can use it to canonicalize the format
7114 of the filename using this macro.
7117 @defmac ASM_OUTPUT_IDENT (@var{stream}, @var{string})
7118 A C statement to output something to the assembler file to handle a
7119 @samp{#ident} directive containing the text @var{string}. If this
7120 macro is not defined, nothing is output for a @samp{#ident} directive.
7123 @deftypefn {Target Hook} void TARGET_ASM_NAMED_SECTION (const char *@var{name}, unsigned int @var{flags}, unsigned int @var{align})
7124 Output assembly directives to switch to section @var{name}. The section
7125 should have attributes as specified by @var{flags}, which is a bit mask
7126 of the @code{SECTION_*} flags defined in @file{output.h}. If @var{align}
7127 is nonzero, it contains an alignment in bytes to be used for the section,
7128 otherwise some target default should be used. Only targets that must
7129 specify an alignment within the section directive need pay attention to
7130 @var{align} -- we will still use @code{ASM_OUTPUT_ALIGN}.
7133 @deftypefn {Target Hook} bool TARGET_HAVE_NAMED_SECTIONS
7134 This flag is true if the target supports @code{TARGET_ASM_NAMED_SECTION}.
7137 @anchor{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}
7138 @deftypefn {Target Hook} bool TARGET_HAVE_SWITCHABLE_BSS_SECTIONS
7139 This flag is true if we can create zeroed data by switching to a BSS
7140 section and then using @code{ASM_OUTPUT_SKIP} to allocate the space.
7141 This is true on most ELF targets.
7144 @deftypefn {Target Hook} {unsigned int} TARGET_SECTION_TYPE_FLAGS (tree @var{decl}, const char *@var{name}, int @var{reloc})
7145 Choose a set of section attributes for use by @code{TARGET_ASM_NAMED_SECTION}
7146 based on a variable or function decl, a section name, and whether or not the
7147 declaration's initializer may contain runtime relocations. @var{decl} may be
7148 null, in which case read-write data should be assumed.
7150 The default version of this function handles choosing code vs data,
7151 read-only vs read-write data, and @code{flag_pic}. You should only
7152 need to override this if your target has special flags that might be
7153 set via @code{__attribute__}.
7156 @deftypefn {Target Hook} {int} TARGET_ASM_RECORD_GCC_SWITCHES (print_switch_type @var{type}, const char * @var{text})
7157 Provides the target with the ability to record the gcc command line
7158 switches that have been passed to the compiler, and options that are
7159 enabled. The @var{type} argument specifies what is being recorded.
7160 It can take the following values:
7163 @item SWITCH_TYPE_PASSED
7164 @var{text} is a command line switch that has been set by the user.
7166 @item SWITCH_TYPE_ENABLED
7167 @var{text} is an option which has been enabled. This might be as a
7168 direct result of a command line switch, or because it is enabled by
7169 default or because it has been enabled as a side effect of a different
7170 command line switch. For example, the @option{-O2} switch enables
7171 various different individual optimization passes.
7173 @item SWITCH_TYPE_DESCRIPTIVE
7174 @var{text} is either NULL or some descriptive text which should be
7175 ignored. If @var{text} is NULL then it is being used to warn the
7176 target hook that either recording is starting or ending. The first
7177 time @var{type} is SWITCH_TYPE_DESCRIPTIVE and @var{text} is NULL, the
7178 warning is for start up and the second time the warning is for
7179 wind down. This feature is to allow the target hook to make any
7180 necessary preparations before it starts to record switches and to
7181 perform any necessary tidying up after it has finished recording
7184 @item SWITCH_TYPE_LINE_START
7185 This option can be ignored by this target hook.
7187 @item SWITCH_TYPE_LINE_END
7188 This option can be ignored by this target hook.
7191 The hook's return value must be zero. Other return values may be
7192 supported in the future.
7194 By default this hook is set to NULL, but an example implementation is
7195 provided for ELF based targets. Called @var{elf_record_gcc_switches},
7196 it records the switches as ASCII text inside a new, string mergeable
7197 section in the assembler output file. The name of the new section is
7198 provided by the @code{TARGET_ASM_RECORD_GCC_SWITCHES_SECTION} target
7202 @deftypefn {Target Hook} {const char *} TARGET_ASM_RECORD_GCC_SWITCHES_SECTION
7203 This is the name of the section that will be created by the example
7204 ELF implementation of the @code{TARGET_ASM_RECORD_GCC_SWITCHES} target
7210 @subsection Output of Data
7213 @deftypevr {Target Hook} {const char *} TARGET_ASM_BYTE_OP
7214 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_HI_OP
7215 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_SI_OP
7216 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_DI_OP
7217 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_TI_OP
7218 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_HI_OP
7219 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_SI_OP
7220 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_DI_OP
7221 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_TI_OP
7222 These hooks specify assembly directives for creating certain kinds
7223 of integer object. The @code{TARGET_ASM_BYTE_OP} directive creates a
7224 byte-sized object, the @code{TARGET_ASM_ALIGNED_HI_OP} one creates an
7225 aligned two-byte object, and so on. Any of the hooks may be
7226 @code{NULL}, indicating that no suitable directive is available.
7228 The compiler will print these strings at the start of a new line,
7229 followed immediately by the object's initial value. In most cases,
7230 the string should contain a tab, a pseudo-op, and then another tab.
7233 @deftypefn {Target Hook} bool TARGET_ASM_INTEGER (rtx @var{x}, unsigned int @var{size}, int @var{aligned_p})
7234 The @code{assemble_integer} function uses this hook to output an
7235 integer object. @var{x} is the object's value, @var{size} is its size
7236 in bytes and @var{aligned_p} indicates whether it is aligned. The
7237 function should return @code{true} if it was able to output the
7238 object. If it returns false, @code{assemble_integer} will try to
7239 split the object into smaller parts.
7241 The default implementation of this hook will use the
7242 @code{TARGET_ASM_BYTE_OP} family of strings, returning @code{false}
7243 when the relevant string is @code{NULL}.
7246 @defmac OUTPUT_ADDR_CONST_EXTRA (@var{stream}, @var{x}, @var{fail})
7247 A C statement to recognize @var{rtx} patterns that
7248 @code{output_addr_const} can't deal with, and output assembly code to
7249 @var{stream} corresponding to the pattern @var{x}. This may be used to
7250 allow machine-dependent @code{UNSPEC}s to appear within constants.
7252 If @code{OUTPUT_ADDR_CONST_EXTRA} fails to recognize a pattern, it must
7253 @code{goto fail}, so that a standard error message is printed. If it
7254 prints an error message itself, by calling, for example,
7255 @code{output_operand_lossage}, it may just complete normally.
7258 @defmac ASM_OUTPUT_ASCII (@var{stream}, @var{ptr}, @var{len})
7259 A C statement to output to the stdio stream @var{stream} an assembler
7260 instruction to assemble a string constant containing the @var{len}
7261 bytes at @var{ptr}. @var{ptr} will be a C expression of type
7262 @code{char *} and @var{len} a C expression of type @code{int}.
7264 If the assembler has a @code{.ascii} pseudo-op as found in the
7265 Berkeley Unix assembler, do not define the macro
7266 @code{ASM_OUTPUT_ASCII}.
7269 @defmac ASM_OUTPUT_FDESC (@var{stream}, @var{decl}, @var{n})
7270 A C statement to output word @var{n} of a function descriptor for
7271 @var{decl}. This must be defined if @code{TARGET_VTABLE_USES_DESCRIPTORS}
7272 is defined, and is otherwise unused.
7275 @defmac CONSTANT_POOL_BEFORE_FUNCTION
7276 You may define this macro as a C expression. You should define the
7277 expression to have a nonzero value if GCC should output the constant
7278 pool for a function before the code for the function, or a zero value if
7279 GCC should output the constant pool after the function. If you do
7280 not define this macro, the usual case, GCC will output the constant
7281 pool before the function.
7284 @defmac ASM_OUTPUT_POOL_PROLOGUE (@var{file}, @var{funname}, @var{fundecl}, @var{size})
7285 A C statement to output assembler commands to define the start of the
7286 constant pool for a function. @var{funname} is a string giving
7287 the name of the function. Should the return type of the function
7288 be required, it can be obtained via @var{fundecl}. @var{size}
7289 is the size, in bytes, of the constant pool that will be written
7290 immediately after this call.
7292 If no constant-pool prefix is required, the usual case, this macro need
7296 @defmac ASM_OUTPUT_SPECIAL_POOL_ENTRY (@var{file}, @var{x}, @var{mode}, @var{align}, @var{labelno}, @var{jumpto})
7297 A C statement (with or without semicolon) to output a constant in the
7298 constant pool, if it needs special treatment. (This macro need not do
7299 anything for RTL expressions that can be output normally.)
7301 The argument @var{file} is the standard I/O stream to output the
7302 assembler code on. @var{x} is the RTL expression for the constant to
7303 output, and @var{mode} is the machine mode (in case @var{x} is a
7304 @samp{const_int}). @var{align} is the required alignment for the value
7305 @var{x}; you should output an assembler directive to force this much
7308 The argument @var{labelno} is a number to use in an internal label for
7309 the address of this pool entry. The definition of this macro is
7310 responsible for outputting the label definition at the proper place.
7311 Here is how to do this:
7314 @code{(*targetm.asm_out.internal_label)} (@var{file}, "LC", @var{labelno});
7317 When you output a pool entry specially, you should end with a
7318 @code{goto} to the label @var{jumpto}. This will prevent the same pool
7319 entry from being output a second time in the usual manner.
7321 You need not define this macro if it would do nothing.
7324 @defmac ASM_OUTPUT_POOL_EPILOGUE (@var{file} @var{funname} @var{fundecl} @var{size})
7325 A C statement to output assembler commands to at the end of the constant
7326 pool for a function. @var{funname} is a string giving the name of the
7327 function. Should the return type of the function be required, you can
7328 obtain it via @var{fundecl}. @var{size} is the size, in bytes, of the
7329 constant pool that GCC wrote immediately before this call.
7331 If no constant-pool epilogue is required, the usual case, you need not
7335 @defmac IS_ASM_LOGICAL_LINE_SEPARATOR (@var{C}, @var{STR})
7336 Define this macro as a C expression which is nonzero if @var{C} is
7337 used as a logical line separator by the assembler. @var{STR} points
7338 to the position in the string where @var{C} was found; this can be used if
7339 a line separator uses multiple characters.
7341 If you do not define this macro, the default is that only
7342 the character @samp{;} is treated as a logical line separator.
7345 @deftypevr {Target Hook} {const char *} TARGET_ASM_OPEN_PAREN
7346 @deftypevrx {Target Hook} {const char *} TARGET_ASM_CLOSE_PAREN
7347 These target hooks are C string constants, describing the syntax in the
7348 assembler for grouping arithmetic expressions. If not overridden, they
7349 default to normal parentheses, which is correct for most assemblers.
7352 These macros are provided by @file{real.h} for writing the definitions
7353 of @code{ASM_OUTPUT_DOUBLE} and the like:
7355 @defmac REAL_VALUE_TO_TARGET_SINGLE (@var{x}, @var{l})
7356 @defmacx REAL_VALUE_TO_TARGET_DOUBLE (@var{x}, @var{l})
7357 @defmacx REAL_VALUE_TO_TARGET_LONG_DOUBLE (@var{x}, @var{l})
7358 @defmacx REAL_VALUE_TO_TARGET_DECIMAL32 (@var{x}, @var{l})
7359 @defmacx REAL_VALUE_TO_TARGET_DECIMAL64 (@var{x}, @var{l})
7360 @defmacx REAL_VALUE_TO_TARGET_DECIMAL128 (@var{x}, @var{l})
7361 These translate @var{x}, of type @code{REAL_VALUE_TYPE}, to the
7362 target's floating point representation, and store its bit pattern in
7363 the variable @var{l}. For @code{REAL_VALUE_TO_TARGET_SINGLE} and
7364 @code{REAL_VALUE_TO_TARGET_DECIMAL32}, this variable should be a
7365 simple @code{long int}. For the others, it should be an array of
7366 @code{long int}. The number of elements in this array is determined
7367 by the size of the desired target floating point data type: 32 bits of
7368 it go in each @code{long int} array element. Each array element holds
7369 32 bits of the result, even if @code{long int} is wider than 32 bits
7370 on the host machine.
7372 The array element values are designed so that you can print them out
7373 using @code{fprintf} in the order they should appear in the target
7377 @node Uninitialized Data
7378 @subsection Output of Uninitialized Variables
7380 Each of the macros in this section is used to do the whole job of
7381 outputting a single uninitialized variable.
7383 @defmac ASM_OUTPUT_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
7384 A C statement (sans semicolon) to output to the stdio stream
7385 @var{stream} the assembler definition of a common-label named
7386 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7387 is the size rounded up to whatever alignment the caller wants.
7389 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7390 output the name itself; before and after that, output the additional
7391 assembler syntax for defining the name, and a newline.
7393 This macro controls how the assembler definitions of uninitialized
7394 common global variables are output.
7397 @defmac ASM_OUTPUT_ALIGNED_COMMON (@var{stream}, @var{name}, @var{size}, @var{alignment})
7398 Like @code{ASM_OUTPUT_COMMON} except takes the required alignment as a
7399 separate, explicit argument. If you define this macro, it is used in
7400 place of @code{ASM_OUTPUT_COMMON}, and gives you more flexibility in
7401 handling the required alignment of the variable. The alignment is specified
7402 as the number of bits.
7405 @defmac ASM_OUTPUT_ALIGNED_DECL_COMMON (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7406 Like @code{ASM_OUTPUT_ALIGNED_COMMON} except that @var{decl} of the
7407 variable to be output, if there is one, or @code{NULL_TREE} if there
7408 is no corresponding variable. If you define this macro, GCC will use it
7409 in place of both @code{ASM_OUTPUT_COMMON} and
7410 @code{ASM_OUTPUT_ALIGNED_COMMON}. Define this macro when you need to see
7411 the variable's decl in order to chose what to output.
7414 @defmac ASM_OUTPUT_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{rounded})
7415 A C statement (sans semicolon) to output to the stdio stream
7416 @var{stream} the assembler definition of uninitialized global @var{decl} named
7417 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7418 is the size rounded up to whatever alignment the caller wants.
7420 Try to use function @code{asm_output_bss} defined in @file{varasm.c} when
7421 defining this macro. If unable, use the expression
7422 @code{assemble_name (@var{stream}, @var{name})} to output the name itself;
7423 before and after that, output the additional assembler syntax for defining
7424 the name, and a newline.
7426 There are two ways of handling global BSS@. One is to define either
7427 this macro or its aligned counterpart, @code{ASM_OUTPUT_ALIGNED_BSS}.
7428 The other is to have @code{TARGET_ASM_SELECT_SECTION} return a
7429 switchable BSS section (@pxref{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}).
7430 You do not need to do both.
7432 Some languages do not have @code{common} data, and require a
7433 non-common form of global BSS in order to handle uninitialized globals
7434 efficiently. C++ is one example of this. However, if the target does
7435 not support global BSS, the front end may choose to make globals
7436 common in order to save space in the object file.
7439 @defmac ASM_OUTPUT_ALIGNED_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7440 Like @code{ASM_OUTPUT_BSS} except takes the required alignment as a
7441 separate, explicit argument. If you define this macro, it is used in
7442 place of @code{ASM_OUTPUT_BSS}, and gives you more flexibility in
7443 handling the required alignment of the variable. The alignment is specified
7444 as the number of bits.
7446 Try to use function @code{asm_output_aligned_bss} defined in file
7447 @file{varasm.c} when defining this macro.
7450 @defmac ASM_OUTPUT_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded})
7451 A C statement (sans semicolon) to output to the stdio stream
7452 @var{stream} the assembler definition of a local-common-label named
7453 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7454 is the size rounded up to whatever alignment the caller wants.
7456 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7457 output the name itself; before and after that, output the additional
7458 assembler syntax for defining the name, and a newline.
7460 This macro controls how the assembler definitions of uninitialized
7461 static variables are output.
7464 @defmac ASM_OUTPUT_ALIGNED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{alignment})
7465 Like @code{ASM_OUTPUT_LOCAL} except takes the required alignment as a
7466 separate, explicit argument. If you define this macro, it is used in
7467 place of @code{ASM_OUTPUT_LOCAL}, and gives you more flexibility in
7468 handling the required alignment of the variable. The alignment is specified
7469 as the number of bits.
7472 @defmac ASM_OUTPUT_ALIGNED_DECL_LOCAL (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7473 Like @code{ASM_OUTPUT_ALIGNED_DECL} except that @var{decl} of the
7474 variable to be output, if there is one, or @code{NULL_TREE} if there
7475 is no corresponding variable. If you define this macro, GCC will use it
7476 in place of both @code{ASM_OUTPUT_DECL} and
7477 @code{ASM_OUTPUT_ALIGNED_DECL}. Define this macro when you need to see
7478 the variable's decl in order to chose what to output.
7482 @subsection Output and Generation of Labels
7484 @c prevent bad page break with this line
7485 This is about outputting labels.
7487 @findex assemble_name
7488 @defmac ASM_OUTPUT_LABEL (@var{stream}, @var{name})
7489 A C statement (sans semicolon) to output to the stdio stream
7490 @var{stream} the assembler definition of a label named @var{name}.
7491 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7492 output the name itself; before and after that, output the additional
7493 assembler syntax for defining the name, and a newline. A default
7494 definition of this macro is provided which is correct for most systems.
7497 @findex assemble_name_raw
7498 @defmac ASM_OUTPUT_INTERNAL_LABEL (@var{stream}, @var{name})
7499 Identical to @code{ASM_OUTPUT_LABEL}, except that @var{name} is known
7500 to refer to a compiler-generated label. The default definition uses
7501 @code{assemble_name_raw}, which is like @code{assemble_name} except
7502 that it is more efficient.
7506 A C string containing the appropriate assembler directive to specify the
7507 size of a symbol, without any arguments. On systems that use ELF, the
7508 default (in @file{config/elfos.h}) is @samp{"\t.size\t"}; on other
7509 systems, the default is not to define this macro.
7511 Define this macro only if it is correct to use the default definitions
7512 of @code{ASM_OUTPUT_SIZE_DIRECTIVE} and @code{ASM_OUTPUT_MEASURED_SIZE}
7513 for your system. If you need your own custom definitions of those
7514 macros, or if you do not need explicit symbol sizes at all, do not
7518 @defmac ASM_OUTPUT_SIZE_DIRECTIVE (@var{stream}, @var{name}, @var{size})
7519 A C statement (sans semicolon) to output to the stdio stream
7520 @var{stream} a directive telling the assembler that the size of the
7521 symbol @var{name} is @var{size}. @var{size} is a @code{HOST_WIDE_INT}.
7522 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
7526 @defmac ASM_OUTPUT_MEASURED_SIZE (@var{stream}, @var{name})
7527 A C statement (sans semicolon) to output to the stdio stream
7528 @var{stream} a directive telling the assembler to calculate the size of
7529 the symbol @var{name} by subtracting its address from the current
7532 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
7533 provided. The default assumes that the assembler recognizes a special
7534 @samp{.} symbol as referring to the current address, and can calculate
7535 the difference between this and another symbol. If your assembler does
7536 not recognize @samp{.} or cannot do calculations with it, you will need
7537 to redefine @code{ASM_OUTPUT_MEASURED_SIZE} to use some other technique.
7541 A C string containing the appropriate assembler directive to specify the
7542 type of a symbol, without any arguments. On systems that use ELF, the
7543 default (in @file{config/elfos.h}) is @samp{"\t.type\t"}; on other
7544 systems, the default is not to define this macro.
7546 Define this macro only if it is correct to use the default definition of
7547 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
7548 custom definition of this macro, or if you do not need explicit symbol
7549 types at all, do not define this macro.
7552 @defmac TYPE_OPERAND_FMT
7553 A C string which specifies (using @code{printf} syntax) the format of
7554 the second operand to @code{TYPE_ASM_OP}. On systems that use ELF, the
7555 default (in @file{config/elfos.h}) is @samp{"@@%s"}; on other systems,
7556 the default is not to define this macro.
7558 Define this macro only if it is correct to use the default definition of
7559 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
7560 custom definition of this macro, or if you do not need explicit symbol
7561 types at all, do not define this macro.
7564 @defmac ASM_OUTPUT_TYPE_DIRECTIVE (@var{stream}, @var{type})
7565 A C statement (sans semicolon) to output to the stdio stream
7566 @var{stream} a directive telling the assembler that the type of the
7567 symbol @var{name} is @var{type}. @var{type} is a C string; currently,
7568 that string is always either @samp{"function"} or @samp{"object"}, but
7569 you should not count on this.
7571 If you define @code{TYPE_ASM_OP} and @code{TYPE_OPERAND_FMT}, a default
7572 definition of this macro is provided.
7575 @defmac ASM_DECLARE_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl})
7576 A C statement (sans semicolon) to output to the stdio stream
7577 @var{stream} any text necessary for declaring the name @var{name} of a
7578 function which is being defined. This macro is responsible for
7579 outputting the label definition (perhaps using
7580 @code{ASM_OUTPUT_LABEL}). The argument @var{decl} is the
7581 @code{FUNCTION_DECL} tree node representing the function.
7583 If this macro is not defined, then the function name is defined in the
7584 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
7586 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition
7590 @defmac ASM_DECLARE_FUNCTION_SIZE (@var{stream}, @var{name}, @var{decl})
7591 A C statement (sans semicolon) to output to the stdio stream
7592 @var{stream} any text necessary for declaring the size of a function
7593 which is being defined. The argument @var{name} is the name of the
7594 function. The argument @var{decl} is the @code{FUNCTION_DECL} tree node
7595 representing the function.
7597 If this macro is not defined, then the function size is not defined.
7599 You may wish to use @code{ASM_OUTPUT_MEASURED_SIZE} in the definition
7603 @defmac ASM_DECLARE_OBJECT_NAME (@var{stream}, @var{name}, @var{decl})
7604 A C statement (sans semicolon) to output to the stdio stream
7605 @var{stream} any text necessary for declaring the name @var{name} of an
7606 initialized variable which is being defined. This macro must output the
7607 label definition (perhaps using @code{ASM_OUTPUT_LABEL}). The argument
7608 @var{decl} is the @code{VAR_DECL} tree node representing the variable.
7610 If this macro is not defined, then the variable name is defined in the
7611 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
7613 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} and/or
7614 @code{ASM_OUTPUT_SIZE_DIRECTIVE} in the definition of this macro.
7617 @defmac ASM_DECLARE_CONSTANT_NAME (@var{stream}, @var{name}, @var{exp}, @var{size})
7618 A C statement (sans semicolon) to output to the stdio stream
7619 @var{stream} any text necessary for declaring the name @var{name} of a
7620 constant which is being defined. This macro is responsible for
7621 outputting the label definition (perhaps using
7622 @code{ASM_OUTPUT_LABEL}). The argument @var{exp} is the
7623 value of the constant, and @var{size} is the size of the constant
7624 in bytes. @var{name} will be an internal label.
7626 If this macro is not defined, then the @var{name} is defined in the
7627 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
7629 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition
7633 @defmac ASM_DECLARE_REGISTER_GLOBAL (@var{stream}, @var{decl}, @var{regno}, @var{name})
7634 A C statement (sans semicolon) to output to the stdio stream
7635 @var{stream} any text necessary for claiming a register @var{regno}
7636 for a global variable @var{decl} with name @var{name}.
7638 If you don't define this macro, that is equivalent to defining it to do
7642 @defmac ASM_FINISH_DECLARE_OBJECT (@var{stream}, @var{decl}, @var{toplevel}, @var{atend})
7643 A C statement (sans semicolon) to finish up declaring a variable name
7644 once the compiler has processed its initializer fully and thus has had a
7645 chance to determine the size of an array when controlled by an
7646 initializer. This is used on systems where it's necessary to declare
7647 something about the size of the object.
7649 If you don't define this macro, that is equivalent to defining it to do
7652 You may wish to use @code{ASM_OUTPUT_SIZE_DIRECTIVE} and/or
7653 @code{ASM_OUTPUT_MEASURED_SIZE} in the definition of this macro.
7656 @deftypefn {Target Hook} void TARGET_ASM_GLOBALIZE_LABEL (FILE *@var{stream}, const char *@var{name})
7657 This target hook is a function to output to the stdio stream
7658 @var{stream} some commands that will make the label @var{name} global;
7659 that is, available for reference from other files.
7661 The default implementation relies on a proper definition of
7662 @code{GLOBAL_ASM_OP}.
7665 @deftypefn {Target Hook} void TARGET_ASM_GLOBALIZE_DECL_NAME (FILE *@var{stream}, tree @var{decl})
7666 This target hook is a function to output to the stdio stream
7667 @var{stream} some commands that will make the name associated with @var{decl}
7668 global; that is, available for reference from other files.
7670 The default implementation uses the TARGET_ASM_GLOBALIZE_LABEL target hook.
7673 @defmac ASM_WEAKEN_LABEL (@var{stream}, @var{name})
7674 A C statement (sans semicolon) to output to the stdio stream
7675 @var{stream} some commands that will make the label @var{name} weak;
7676 that is, available for reference from other files but only used if
7677 no other definition is available. Use the expression
7678 @code{assemble_name (@var{stream}, @var{name})} to output the name
7679 itself; before and after that, output the additional assembler syntax
7680 for making that name weak, and a newline.
7682 If you don't define this macro or @code{ASM_WEAKEN_DECL}, GCC will not
7683 support weak symbols and you should not define the @code{SUPPORTS_WEAK}
7687 @defmac ASM_WEAKEN_DECL (@var{stream}, @var{decl}, @var{name}, @var{value})
7688 Combines (and replaces) the function of @code{ASM_WEAKEN_LABEL} and
7689 @code{ASM_OUTPUT_WEAK_ALIAS}, allowing access to the associated function
7690 or variable decl. If @var{value} is not @code{NULL}, this C statement
7691 should output to the stdio stream @var{stream} assembler code which
7692 defines (equates) the weak symbol @var{name} to have the value
7693 @var{value}. If @var{value} is @code{NULL}, it should output commands
7694 to make @var{name} weak.
7697 @defmac ASM_OUTPUT_WEAKREF (@var{stream}, @var{decl}, @var{name}, @var{value})
7698 Outputs a directive that enables @var{name} to be used to refer to
7699 symbol @var{value} with weak-symbol semantics. @code{decl} is the
7700 declaration of @code{name}.
7703 @defmac SUPPORTS_WEAK
7704 A C expression which evaluates to true if the target supports weak symbols.
7706 If you don't define this macro, @file{defaults.h} provides a default
7707 definition. If either @code{ASM_WEAKEN_LABEL} or @code{ASM_WEAKEN_DECL}
7708 is defined, the default definition is @samp{1}; otherwise, it is
7709 @samp{0}. Define this macro if you want to control weak symbol support
7710 with a compiler flag such as @option{-melf}.
7713 @defmac MAKE_DECL_ONE_ONLY (@var{decl})
7714 A C statement (sans semicolon) to mark @var{decl} to be emitted as a
7715 public symbol such that extra copies in multiple translation units will
7716 be discarded by the linker. Define this macro if your object file
7717 format provides support for this concept, such as the @samp{COMDAT}
7718 section flags in the Microsoft Windows PE/COFF format, and this support
7719 requires changes to @var{decl}, such as putting it in a separate section.
7722 @defmac SUPPORTS_ONE_ONLY
7723 A C expression which evaluates to true if the target supports one-only
7726 If you don't define this macro, @file{varasm.c} provides a default
7727 definition. If @code{MAKE_DECL_ONE_ONLY} is defined, the default
7728 definition is @samp{1}; otherwise, it is @samp{0}. Define this macro if
7729 you want to control one-only symbol support with a compiler flag, or if
7730 setting the @code{DECL_ONE_ONLY} flag is enough to mark a declaration to
7731 be emitted as one-only.
7734 @deftypefn {Target Hook} void TARGET_ASM_ASSEMBLE_VISIBILITY (tree @var{decl}, const char *@var{visibility})
7735 This target hook is a function to output to @var{asm_out_file} some
7736 commands that will make the symbol(s) associated with @var{decl} have
7737 hidden, protected or internal visibility as specified by @var{visibility}.
7740 @defmac TARGET_WEAK_NOT_IN_ARCHIVE_TOC
7741 A C expression that evaluates to true if the target's linker expects
7742 that weak symbols do not appear in a static archive's table of contents.
7743 The default is @code{0}.
7745 Leaving weak symbols out of an archive's table of contents means that,
7746 if a symbol will only have a definition in one translation unit and
7747 will have undefined references from other translation units, that
7748 symbol should not be weak. Defining this macro to be nonzero will
7749 thus have the effect that certain symbols that would normally be weak
7750 (explicit template instantiations, and vtables for polymorphic classes
7751 with noninline key methods) will instead be nonweak.
7753 The C++ ABI requires this macro to be zero. Define this macro for
7754 targets where full C++ ABI compliance is impossible and where linker
7755 restrictions require weak symbols to be left out of a static archive's
7759 @defmac ASM_OUTPUT_EXTERNAL (@var{stream}, @var{decl}, @var{name})
7760 A C statement (sans semicolon) to output to the stdio stream
7761 @var{stream} any text necessary for declaring the name of an external
7762 symbol named @var{name} which is referenced in this compilation but
7763 not defined. The value of @var{decl} is the tree node for the
7766 This macro need not be defined if it does not need to output anything.
7767 The GNU assembler and most Unix assemblers don't require anything.
7770 @deftypefn {Target Hook} void TARGET_ASM_EXTERNAL_LIBCALL (rtx @var{symref})
7771 This target hook is a function to output to @var{asm_out_file} an assembler
7772 pseudo-op to declare a library function name external. The name of the
7773 library function is given by @var{symref}, which is a @code{symbol_ref}.
7776 @deftypefn {Target Hook} void TARGET_ASM_MARK_DECL_PRESERVED (tree @var{decl})
7777 This target hook is a function to output to @var{asm_out_file} an assembler
7778 directive to annotate used symbol. Darwin target use .no_dead_code_strip
7782 @defmac ASM_OUTPUT_LABELREF (@var{stream}, @var{name})
7783 A C statement (sans semicolon) to output to the stdio stream
7784 @var{stream} a reference in assembler syntax to a label named
7785 @var{name}. This should add @samp{_} to the front of the name, if that
7786 is customary on your operating system, as it is in most Berkeley Unix
7787 systems. This macro is used in @code{assemble_name}.
7790 @defmac ASM_OUTPUT_SYMBOL_REF (@var{stream}, @var{sym})
7791 A C statement (sans semicolon) to output a reference to
7792 @code{SYMBOL_REF} @var{sym}. If not defined, @code{assemble_name}
7793 will be used to output the name of the symbol. This macro may be used
7794 to modify the way a symbol is referenced depending on information
7795 encoded by @code{TARGET_ENCODE_SECTION_INFO}.
7798 @defmac ASM_OUTPUT_LABEL_REF (@var{stream}, @var{buf})
7799 A C statement (sans semicolon) to output a reference to @var{buf}, the
7800 result of @code{ASM_GENERATE_INTERNAL_LABEL}. If not defined,
7801 @code{assemble_name} will be used to output the name of the symbol.
7802 This macro is not used by @code{output_asm_label}, or the @code{%l}
7803 specifier that calls it; the intention is that this macro should be set
7804 when it is necessary to output a label differently when its address is
7808 @deftypefn {Target Hook} void TARGET_ASM_INTERNAL_LABEL (FILE *@var{stream}, const char *@var{prefix}, unsigned long @var{labelno})
7809 A function to output to the stdio stream @var{stream} a label whose
7810 name is made from the string @var{prefix} and the number @var{labelno}.
7812 It is absolutely essential that these labels be distinct from the labels
7813 used for user-level functions and variables. Otherwise, certain programs
7814 will have name conflicts with internal labels.
7816 It is desirable to exclude internal labels from the symbol table of the
7817 object file. Most assemblers have a naming convention for labels that
7818 should be excluded; on many systems, the letter @samp{L} at the
7819 beginning of a label has this effect. You should find out what
7820 convention your system uses, and follow it.
7822 The default version of this function utilizes @code{ASM_GENERATE_INTERNAL_LABEL}.
7825 @defmac ASM_OUTPUT_DEBUG_LABEL (@var{stream}, @var{prefix}, @var{num})
7826 A C statement to output to the stdio stream @var{stream} a debug info
7827 label whose name is made from the string @var{prefix} and the number
7828 @var{num}. This is useful for VLIW targets, where debug info labels
7829 may need to be treated differently than branch target labels. On some
7830 systems, branch target labels must be at the beginning of instruction
7831 bundles, but debug info labels can occur in the middle of instruction
7834 If this macro is not defined, then @code{(*targetm.asm_out.internal_label)} will be
7838 @defmac ASM_GENERATE_INTERNAL_LABEL (@var{string}, @var{prefix}, @var{num})
7839 A C statement to store into the string @var{string} a label whose name
7840 is made from the string @var{prefix} and the number @var{num}.
7842 This string, when output subsequently by @code{assemble_name}, should
7843 produce the output that @code{(*targetm.asm_out.internal_label)} would produce
7844 with the same @var{prefix} and @var{num}.
7846 If the string begins with @samp{*}, then @code{assemble_name} will
7847 output the rest of the string unchanged. It is often convenient for
7848 @code{ASM_GENERATE_INTERNAL_LABEL} to use @samp{*} in this way. If the
7849 string doesn't start with @samp{*}, then @code{ASM_OUTPUT_LABELREF} gets
7850 to output the string, and may change it. (Of course,
7851 @code{ASM_OUTPUT_LABELREF} is also part of your machine description, so
7852 you should know what it does on your machine.)
7855 @defmac ASM_FORMAT_PRIVATE_NAME (@var{outvar}, @var{name}, @var{number})
7856 A C expression to assign to @var{outvar} (which is a variable of type
7857 @code{char *}) a newly allocated string made from the string
7858 @var{name} and the number @var{number}, with some suitable punctuation
7859 added. Use @code{alloca} to get space for the string.
7861 The string will be used as an argument to @code{ASM_OUTPUT_LABELREF} to
7862 produce an assembler label for an internal static variable whose name is
7863 @var{name}. Therefore, the string must be such as to result in valid
7864 assembler code. The argument @var{number} is different each time this
7865 macro is executed; it prevents conflicts between similarly-named
7866 internal static variables in different scopes.
7868 Ideally this string should not be a valid C identifier, to prevent any
7869 conflict with the user's own symbols. Most assemblers allow periods
7870 or percent signs in assembler symbols; putting at least one of these
7871 between the name and the number will suffice.
7873 If this macro is not defined, a default definition will be provided
7874 which is correct for most systems.
7877 @defmac ASM_OUTPUT_DEF (@var{stream}, @var{name}, @var{value})
7878 A C statement to output to the stdio stream @var{stream} assembler code
7879 which defines (equates) the symbol @var{name} to have the value @var{value}.
7882 If @code{SET_ASM_OP} is defined, a default definition is provided which is
7883 correct for most systems.
7886 @defmac ASM_OUTPUT_DEF_FROM_DECLS (@var{stream}, @var{decl_of_name}, @var{decl_of_value})
7887 A C statement to output to the stdio stream @var{stream} assembler code
7888 which defines (equates) the symbol whose tree node is @var{decl_of_name}
7889 to have the value of the tree node @var{decl_of_value}. This macro will
7890 be used in preference to @samp{ASM_OUTPUT_DEF} if it is defined and if
7891 the tree nodes are available.
7894 If @code{SET_ASM_OP} is defined, a default definition is provided which is
7895 correct for most systems.
7898 @defmac TARGET_DEFERRED_OUTPUT_DEFS (@var{decl_of_name}, @var{decl_of_value})
7899 A C statement that evaluates to true if the assembler code which defines
7900 (equates) the symbol whose tree node is @var{decl_of_name} to have the value
7901 of the tree node @var{decl_of_value} should be emitted near the end of the
7902 current compilation unit. The default is to not defer output of defines.
7903 This macro affects defines output by @samp{ASM_OUTPUT_DEF} and
7904 @samp{ASM_OUTPUT_DEF_FROM_DECLS}.
7907 @defmac ASM_OUTPUT_WEAK_ALIAS (@var{stream}, @var{name}, @var{value})
7908 A C statement to output to the stdio stream @var{stream} assembler code
7909 which defines (equates) the weak symbol @var{name} to have the value
7910 @var{value}. If @var{value} is @code{NULL}, it defines @var{name} as
7911 an undefined weak symbol.
7913 Define this macro if the target only supports weak aliases; define
7914 @code{ASM_OUTPUT_DEF} instead if possible.
7917 @defmac OBJC_GEN_METHOD_LABEL (@var{buf}, @var{is_inst}, @var{class_name}, @var{cat_name}, @var{sel_name})
7918 Define this macro to override the default assembler names used for
7919 Objective-C methods.
7921 The default name is a unique method number followed by the name of the
7922 class (e.g.@: @samp{_1_Foo}). For methods in categories, the name of
7923 the category is also included in the assembler name (e.g.@:
7926 These names are safe on most systems, but make debugging difficult since
7927 the method's selector is not present in the name. Therefore, particular
7928 systems define other ways of computing names.
7930 @var{buf} is an expression of type @code{char *} which gives you a
7931 buffer in which to store the name; its length is as long as
7932 @var{class_name}, @var{cat_name} and @var{sel_name} put together, plus
7933 50 characters extra.
7935 The argument @var{is_inst} specifies whether the method is an instance
7936 method or a class method; @var{class_name} is the name of the class;
7937 @var{cat_name} is the name of the category (or @code{NULL} if the method is not
7938 in a category); and @var{sel_name} is the name of the selector.
7940 On systems where the assembler can handle quoted names, you can use this
7941 macro to provide more human-readable names.
7944 @defmac ASM_DECLARE_CLASS_REFERENCE (@var{stream}, @var{name})
7945 A C statement (sans semicolon) to output to the stdio stream
7946 @var{stream} commands to declare that the label @var{name} is an
7947 Objective-C class reference. This is only needed for targets whose
7948 linkers have special support for NeXT-style runtimes.
7951 @defmac ASM_DECLARE_UNRESOLVED_REFERENCE (@var{stream}, @var{name})
7952 A C statement (sans semicolon) to output to the stdio stream
7953 @var{stream} commands to declare that the label @var{name} is an
7954 unresolved Objective-C class reference. This is only needed for targets
7955 whose linkers have special support for NeXT-style runtimes.
7958 @node Initialization
7959 @subsection How Initialization Functions Are Handled
7960 @cindex initialization routines
7961 @cindex termination routines
7962 @cindex constructors, output of
7963 @cindex destructors, output of
7965 The compiled code for certain languages includes @dfn{constructors}
7966 (also called @dfn{initialization routines})---functions to initialize
7967 data in the program when the program is started. These functions need
7968 to be called before the program is ``started''---that is to say, before
7969 @code{main} is called.
7971 Compiling some languages generates @dfn{destructors} (also called
7972 @dfn{termination routines}) that should be called when the program
7975 To make the initialization and termination functions work, the compiler
7976 must output something in the assembler code to cause those functions to
7977 be called at the appropriate time. When you port the compiler to a new
7978 system, you need to specify how to do this.
7980 There are two major ways that GCC currently supports the execution of
7981 initialization and termination functions. Each way has two variants.
7982 Much of the structure is common to all four variations.
7984 @findex __CTOR_LIST__
7985 @findex __DTOR_LIST__
7986 The linker must build two lists of these functions---a list of
7987 initialization functions, called @code{__CTOR_LIST__}, and a list of
7988 termination functions, called @code{__DTOR_LIST__}.
7990 Each list always begins with an ignored function pointer (which may hold
7991 0, @minus{}1, or a count of the function pointers after it, depending on
7992 the environment). This is followed by a series of zero or more function
7993 pointers to constructors (or destructors), followed by a function
7994 pointer containing zero.
7996 Depending on the operating system and its executable file format, either
7997 @file{crtstuff.c} or @file{libgcc2.c} traverses these lists at startup
7998 time and exit time. Constructors are called in reverse order of the
7999 list; destructors in forward order.
8001 The best way to handle static constructors works only for object file
8002 formats which provide arbitrarily-named sections. A section is set
8003 aside for a list of constructors, and another for a list of destructors.
8004 Traditionally these are called @samp{.ctors} and @samp{.dtors}. Each
8005 object file that defines an initialization function also puts a word in
8006 the constructor section to point to that function. The linker
8007 accumulates all these words into one contiguous @samp{.ctors} section.
8008 Termination functions are handled similarly.
8010 This method will be chosen as the default by @file{target-def.h} if
8011 @code{TARGET_ASM_NAMED_SECTION} is defined. A target that does not
8012 support arbitrary sections, but does support special designated
8013 constructor and destructor sections may define @code{CTORS_SECTION_ASM_OP}
8014 and @code{DTORS_SECTION_ASM_OP} to achieve the same effect.
8016 When arbitrary sections are available, there are two variants, depending
8017 upon how the code in @file{crtstuff.c} is called. On systems that
8018 support a @dfn{.init} section which is executed at program startup,
8019 parts of @file{crtstuff.c} are compiled into that section. The
8020 program is linked by the @command{gcc} driver like this:
8023 ld -o @var{output_file} crti.o crtbegin.o @dots{} -lgcc crtend.o crtn.o
8026 The prologue of a function (@code{__init}) appears in the @code{.init}
8027 section of @file{crti.o}; the epilogue appears in @file{crtn.o}. Likewise
8028 for the function @code{__fini} in the @dfn{.fini} section. Normally these
8029 files are provided by the operating system or by the GNU C library, but
8030 are provided by GCC for a few targets.
8032 The objects @file{crtbegin.o} and @file{crtend.o} are (for most targets)
8033 compiled from @file{crtstuff.c}. They contain, among other things, code
8034 fragments within the @code{.init} and @code{.fini} sections that branch
8035 to routines in the @code{.text} section. The linker will pull all parts
8036 of a section together, which results in a complete @code{__init} function
8037 that invokes the routines we need at startup.
8039 To use this variant, you must define the @code{INIT_SECTION_ASM_OP}
8042 If no init section is available, when GCC compiles any function called
8043 @code{main} (or more accurately, any function designated as a program
8044 entry point by the language front end calling @code{expand_main_function}),
8045 it inserts a procedure call to @code{__main} as the first executable code
8046 after the function prologue. The @code{__main} function is defined
8047 in @file{libgcc2.c} and runs the global constructors.
8049 In file formats that don't support arbitrary sections, there are again
8050 two variants. In the simplest variant, the GNU linker (GNU @code{ld})
8051 and an `a.out' format must be used. In this case,
8052 @code{TARGET_ASM_CONSTRUCTOR} is defined to produce a @code{.stabs}
8053 entry of type @samp{N_SETT}, referencing the name @code{__CTOR_LIST__},
8054 and with the address of the void function containing the initialization
8055 code as its value. The GNU linker recognizes this as a request to add
8056 the value to a @dfn{set}; the values are accumulated, and are eventually
8057 placed in the executable as a vector in the format described above, with
8058 a leading (ignored) count and a trailing zero element.
8059 @code{TARGET_ASM_DESTRUCTOR} is handled similarly. Since no init
8060 section is available, the absence of @code{INIT_SECTION_ASM_OP} causes
8061 the compilation of @code{main} to call @code{__main} as above, starting
8062 the initialization process.
8064 The last variant uses neither arbitrary sections nor the GNU linker.
8065 This is preferable when you want to do dynamic linking and when using
8066 file formats which the GNU linker does not support, such as `ECOFF'@. In
8067 this case, @code{TARGET_HAVE_CTORS_DTORS} is false, initialization and
8068 termination functions are recognized simply by their names. This requires
8069 an extra program in the linkage step, called @command{collect2}. This program
8070 pretends to be the linker, for use with GCC; it does its job by running
8071 the ordinary linker, but also arranges to include the vectors of
8072 initialization and termination functions. These functions are called
8073 via @code{__main} as described above. In order to use this method,
8074 @code{use_collect2} must be defined in the target in @file{config.gcc}.
8077 The following section describes the specific macros that control and
8078 customize the handling of initialization and termination functions.
8081 @node Macros for Initialization
8082 @subsection Macros Controlling Initialization Routines
8084 Here are the macros that control how the compiler handles initialization
8085 and termination functions:
8087 @defmac INIT_SECTION_ASM_OP
8088 If defined, a C string constant, including spacing, for the assembler
8089 operation to identify the following data as initialization code. If not
8090 defined, GCC will assume such a section does not exist. When you are
8091 using special sections for initialization and termination functions, this
8092 macro also controls how @file{crtstuff.c} and @file{libgcc2.c} arrange to
8093 run the initialization functions.
8096 @defmac HAS_INIT_SECTION
8097 If defined, @code{main} will not call @code{__main} as described above.
8098 This macro should be defined for systems that control start-up code
8099 on a symbol-by-symbol basis, such as OSF/1, and should not
8100 be defined explicitly for systems that support @code{INIT_SECTION_ASM_OP}.
8103 @defmac LD_INIT_SWITCH
8104 If defined, a C string constant for a switch that tells the linker that
8105 the following symbol is an initialization routine.
8108 @defmac LD_FINI_SWITCH
8109 If defined, a C string constant for a switch that tells the linker that
8110 the following symbol is a finalization routine.
8113 @defmac COLLECT_SHARED_INIT_FUNC (@var{stream}, @var{func})
8114 If defined, a C statement that will write a function that can be
8115 automatically called when a shared library is loaded. The function
8116 should call @var{func}, which takes no arguments. If not defined, and
8117 the object format requires an explicit initialization function, then a
8118 function called @code{_GLOBAL__DI} will be generated.
8120 This function and the following one are used by collect2 when linking a
8121 shared library that needs constructors or destructors, or has DWARF2
8122 exception tables embedded in the code.
8125 @defmac COLLECT_SHARED_FINI_FUNC (@var{stream}, @var{func})
8126 If defined, a C statement that will write a function that can be
8127 automatically called when a shared library is unloaded. The function
8128 should call @var{func}, which takes no arguments. If not defined, and
8129 the object format requires an explicit finalization function, then a
8130 function called @code{_GLOBAL__DD} will be generated.
8133 @defmac INVOKE__main
8134 If defined, @code{main} will call @code{__main} despite the presence of
8135 @code{INIT_SECTION_ASM_OP}. This macro should be defined for systems
8136 where the init section is not actually run automatically, but is still
8137 useful for collecting the lists of constructors and destructors.
8140 @defmac SUPPORTS_INIT_PRIORITY
8141 If nonzero, the C++ @code{init_priority} attribute is supported and the
8142 compiler should emit instructions to control the order of initialization
8143 of objects. If zero, the compiler will issue an error message upon
8144 encountering an @code{init_priority} attribute.
8147 @deftypefn {Target Hook} bool TARGET_HAVE_CTORS_DTORS
8148 This value is true if the target supports some ``native'' method of
8149 collecting constructors and destructors to be run at startup and exit.
8150 It is false if we must use @command{collect2}.
8153 @deftypefn {Target Hook} void TARGET_ASM_CONSTRUCTOR (rtx @var{symbol}, int @var{priority})
8154 If defined, a function that outputs assembler code to arrange to call
8155 the function referenced by @var{symbol} at initialization time.
8157 Assume that @var{symbol} is a @code{SYMBOL_REF} for a function taking
8158 no arguments and with no return value. If the target supports initialization
8159 priorities, @var{priority} is a value between 0 and @code{MAX_INIT_PRIORITY};
8160 otherwise it must be @code{DEFAULT_INIT_PRIORITY}.
8162 If this macro is not defined by the target, a suitable default will
8163 be chosen if (1) the target supports arbitrary section names, (2) the
8164 target defines @code{CTORS_SECTION_ASM_OP}, or (3) @code{USE_COLLECT2}
8168 @deftypefn {Target Hook} void TARGET_ASM_DESTRUCTOR (rtx @var{symbol}, int @var{priority})
8169 This is like @code{TARGET_ASM_CONSTRUCTOR} but used for termination
8170 functions rather than initialization functions.
8173 If @code{TARGET_HAVE_CTORS_DTORS} is true, the initialization routine
8174 generated for the generated object file will have static linkage.
8176 If your system uses @command{collect2} as the means of processing
8177 constructors, then that program normally uses @command{nm} to scan
8178 an object file for constructor functions to be called.
8180 On certain kinds of systems, you can define this macro to make
8181 @command{collect2} work faster (and, in some cases, make it work at all):
8183 @defmac OBJECT_FORMAT_COFF
8184 Define this macro if the system uses COFF (Common Object File Format)
8185 object files, so that @command{collect2} can assume this format and scan
8186 object files directly for dynamic constructor/destructor functions.
8188 This macro is effective only in a native compiler; @command{collect2} as
8189 part of a cross compiler always uses @command{nm} for the target machine.
8192 @defmac REAL_NM_FILE_NAME
8193 Define this macro as a C string constant containing the file name to use
8194 to execute @command{nm}. The default is to search the path normally for
8197 If your system supports shared libraries and has a program to list the
8198 dynamic dependencies of a given library or executable, you can define
8199 these macros to enable support for running initialization and
8200 termination functions in shared libraries:
8204 Define this macro to a C string constant containing the name of the program
8205 which lists dynamic dependencies, like @command{"ldd"} under SunOS 4.
8208 @defmac PARSE_LDD_OUTPUT (@var{ptr})
8209 Define this macro to be C code that extracts filenames from the output
8210 of the program denoted by @code{LDD_SUFFIX}. @var{ptr} is a variable
8211 of type @code{char *} that points to the beginning of a line of output
8212 from @code{LDD_SUFFIX}. If the line lists a dynamic dependency, the
8213 code must advance @var{ptr} to the beginning of the filename on that
8214 line. Otherwise, it must set @var{ptr} to @code{NULL}.
8217 @defmac SHLIB_SUFFIX
8218 Define this macro to a C string constant containing the default shared
8219 library extension of the target (e.g., @samp{".so"}). @command{collect2}
8220 strips version information after this suffix when generating global
8221 constructor and destructor names. This define is only needed on targets
8222 that use @command{collect2} to process constructors and destructors.
8225 @node Instruction Output
8226 @subsection Output of Assembler Instructions
8228 @c prevent bad page break with this line
8229 This describes assembler instruction output.
8231 @defmac REGISTER_NAMES
8232 A C initializer containing the assembler's names for the machine
8233 registers, each one as a C string constant. This is what translates
8234 register numbers in the compiler into assembler language.
8237 @defmac ADDITIONAL_REGISTER_NAMES
8238 If defined, a C initializer for an array of structures containing a name
8239 and a register number. This macro defines additional names for hard
8240 registers, thus allowing the @code{asm} option in declarations to refer
8241 to registers using alternate names.
8244 @defmac ASM_OUTPUT_OPCODE (@var{stream}, @var{ptr})
8245 Define this macro if you are using an unusual assembler that
8246 requires different names for the machine instructions.
8248 The definition is a C statement or statements which output an
8249 assembler instruction opcode to the stdio stream @var{stream}. The
8250 macro-operand @var{ptr} is a variable of type @code{char *} which
8251 points to the opcode name in its ``internal'' form---the form that is
8252 written in the machine description. The definition should output the
8253 opcode name to @var{stream}, performing any translation you desire, and
8254 increment the variable @var{ptr} to point at the end of the opcode
8255 so that it will not be output twice.
8257 In fact, your macro definition may process less than the entire opcode
8258 name, or more than the opcode name; but if you want to process text
8259 that includes @samp{%}-sequences to substitute operands, you must take
8260 care of the substitution yourself. Just be sure to increment
8261 @var{ptr} over whatever text should not be output normally.
8263 @findex recog_data.operand
8264 If you need to look at the operand values, they can be found as the
8265 elements of @code{recog_data.operand}.
8267 If the macro definition does nothing, the instruction is output
8271 @defmac FINAL_PRESCAN_INSN (@var{insn}, @var{opvec}, @var{noperands})
8272 If defined, a C statement to be executed just prior to the output of
8273 assembler code for @var{insn}, to modify the extracted operands so
8274 they will be output differently.
8276 Here the argument @var{opvec} is the vector containing the operands
8277 extracted from @var{insn}, and @var{noperands} is the number of
8278 elements of the vector which contain meaningful data for this insn.
8279 The contents of this vector are what will be used to convert the insn
8280 template into assembler code, so you can change the assembler output
8281 by changing the contents of the vector.
8283 This macro is useful when various assembler syntaxes share a single
8284 file of instruction patterns; by defining this macro differently, you
8285 can cause a large class of instructions to be output differently (such
8286 as with rearranged operands). Naturally, variations in assembler
8287 syntax affecting individual insn patterns ought to be handled by
8288 writing conditional output routines in those patterns.
8290 If this macro is not defined, it is equivalent to a null statement.
8293 @deftypefn {Target Hook} void TARGET_ASM_FINAL_POSTSCAN_INSN (FILE *@var{FILE}, rtx @var{insn}, rtx *@var{opvec}, int @var{noperands})
8294 If defined, this target hook is a function which is executed just after the
8295 output of assembler code for @var{insn}, to change the mode of the assembler
8298 Here the argument @var{opvec} is the vector containing the operands
8299 extracted from @var{insn}, and @var{noperands} is the number of
8300 elements of the vector which contain meaningful data for this insn.
8301 The contents of this vector are what was used to convert the insn
8302 template into assembler code, so you can change the assembler mode
8303 by checking the contents of the vector.
8306 @defmac PRINT_OPERAND (@var{stream}, @var{x}, @var{code})
8307 A C compound statement to output to stdio stream @var{stream} the
8308 assembler syntax for an instruction operand @var{x}. @var{x} is an
8311 @var{code} is a value that can be used to specify one of several ways
8312 of printing the operand. It is used when identical operands must be
8313 printed differently depending on the context. @var{code} comes from
8314 the @samp{%} specification that was used to request printing of the
8315 operand. If the specification was just @samp{%@var{digit}} then
8316 @var{code} is 0; if the specification was @samp{%@var{ltr}
8317 @var{digit}} then @var{code} is the ASCII code for @var{ltr}.
8320 If @var{x} is a register, this macro should print the register's name.
8321 The names can be found in an array @code{reg_names} whose type is
8322 @code{char *[]}. @code{reg_names} is initialized from
8323 @code{REGISTER_NAMES}.
8325 When the machine description has a specification @samp{%@var{punct}}
8326 (a @samp{%} followed by a punctuation character), this macro is called
8327 with a null pointer for @var{x} and the punctuation character for
8331 @defmac PRINT_OPERAND_PUNCT_VALID_P (@var{code})
8332 A C expression which evaluates to true if @var{code} is a valid
8333 punctuation character for use in the @code{PRINT_OPERAND} macro. If
8334 @code{PRINT_OPERAND_PUNCT_VALID_P} is not defined, it means that no
8335 punctuation characters (except for the standard one, @samp{%}) are used
8339 @defmac PRINT_OPERAND_ADDRESS (@var{stream}, @var{x})
8340 A C compound statement to output to stdio stream @var{stream} the
8341 assembler syntax for an instruction operand that is a memory reference
8342 whose address is @var{x}. @var{x} is an RTL expression.
8344 @cindex @code{TARGET_ENCODE_SECTION_INFO} usage
8345 On some machines, the syntax for a symbolic address depends on the
8346 section that the address refers to. On these machines, define the hook
8347 @code{TARGET_ENCODE_SECTION_INFO} to store the information into the
8348 @code{symbol_ref}, and then check for it here. @xref{Assembler
8352 @findex dbr_sequence_length
8353 @defmac DBR_OUTPUT_SEQEND (@var{file})
8354 A C statement, to be executed after all slot-filler instructions have
8355 been output. If necessary, call @code{dbr_sequence_length} to
8356 determine the number of slots filled in a sequence (zero if not
8357 currently outputting a sequence), to decide how many no-ops to output,
8360 Don't define this macro if it has nothing to do, but it is helpful in
8361 reading assembly output if the extent of the delay sequence is made
8362 explicit (e.g.@: with white space).
8365 @findex final_sequence
8366 Note that output routines for instructions with delay slots must be
8367 prepared to deal with not being output as part of a sequence
8368 (i.e.@: when the scheduling pass is not run, or when no slot fillers could be
8369 found.) The variable @code{final_sequence} is null when not
8370 processing a sequence, otherwise it contains the @code{sequence} rtx
8374 @defmac REGISTER_PREFIX
8375 @defmacx LOCAL_LABEL_PREFIX
8376 @defmacx USER_LABEL_PREFIX
8377 @defmacx IMMEDIATE_PREFIX
8378 If defined, C string expressions to be used for the @samp{%R}, @samp{%L},
8379 @samp{%U}, and @samp{%I} options of @code{asm_fprintf} (see
8380 @file{final.c}). These are useful when a single @file{md} file must
8381 support multiple assembler formats. In that case, the various @file{tm.h}
8382 files can define these macros differently.
8385 @defmac ASM_FPRINTF_EXTENSIONS (@var{file}, @var{argptr}, @var{format})
8386 If defined this macro should expand to a series of @code{case}
8387 statements which will be parsed inside the @code{switch} statement of
8388 the @code{asm_fprintf} function. This allows targets to define extra
8389 printf formats which may useful when generating their assembler
8390 statements. Note that uppercase letters are reserved for future
8391 generic extensions to asm_fprintf, and so are not available to target
8392 specific code. The output file is given by the parameter @var{file}.
8393 The varargs input pointer is @var{argptr} and the rest of the format
8394 string, starting the character after the one that is being switched
8395 upon, is pointed to by @var{format}.
8398 @defmac ASSEMBLER_DIALECT
8399 If your target supports multiple dialects of assembler language (such as
8400 different opcodes), define this macro as a C expression that gives the
8401 numeric index of the assembler language dialect to use, with zero as the
8404 If this macro is defined, you may use constructs of the form
8406 @samp{@{option0|option1|option2@dots{}@}}
8409 in the output templates of patterns (@pxref{Output Template}) or in the
8410 first argument of @code{asm_fprintf}. This construct outputs
8411 @samp{option0}, @samp{option1}, @samp{option2}, etc., if the value of
8412 @code{ASSEMBLER_DIALECT} is zero, one, two, etc. Any special characters
8413 within these strings retain their usual meaning. If there are fewer
8414 alternatives within the braces than the value of
8415 @code{ASSEMBLER_DIALECT}, the construct outputs nothing.
8417 If you do not define this macro, the characters @samp{@{}, @samp{|} and
8418 @samp{@}} do not have any special meaning when used in templates or
8419 operands to @code{asm_fprintf}.
8421 Define the macros @code{REGISTER_PREFIX}, @code{LOCAL_LABEL_PREFIX},
8422 @code{USER_LABEL_PREFIX} and @code{IMMEDIATE_PREFIX} if you can express
8423 the variations in assembler language syntax with that mechanism. Define
8424 @code{ASSEMBLER_DIALECT} and use the @samp{@{option0|option1@}} syntax
8425 if the syntax variant are larger and involve such things as different
8426 opcodes or operand order.
8429 @defmac ASM_OUTPUT_REG_PUSH (@var{stream}, @var{regno})
8430 A C expression to output to @var{stream} some assembler code
8431 which will push hard register number @var{regno} onto the stack.
8432 The code need not be optimal, since this macro is used only when
8436 @defmac ASM_OUTPUT_REG_POP (@var{stream}, @var{regno})
8437 A C expression to output to @var{stream} some assembler code
8438 which will pop hard register number @var{regno} off of the stack.
8439 The code need not be optimal, since this macro is used only when
8443 @node Dispatch Tables
8444 @subsection Output of Dispatch Tables
8446 @c prevent bad page break with this line
8447 This concerns dispatch tables.
8449 @cindex dispatch table
8450 @defmac ASM_OUTPUT_ADDR_DIFF_ELT (@var{stream}, @var{body}, @var{value}, @var{rel})
8451 A C statement to output to the stdio stream @var{stream} an assembler
8452 pseudo-instruction to generate a difference between two labels.
8453 @var{value} and @var{rel} are the numbers of two internal labels. The
8454 definitions of these labels are output using
8455 @code{(*targetm.asm_out.internal_label)}, and they must be printed in the same
8456 way here. For example,
8459 fprintf (@var{stream}, "\t.word L%d-L%d\n",
8460 @var{value}, @var{rel})
8463 You must provide this macro on machines where the addresses in a
8464 dispatch table are relative to the table's own address. If defined, GCC
8465 will also use this macro on all machines when producing PIC@.
8466 @var{body} is the body of the @code{ADDR_DIFF_VEC}; it is provided so that the
8467 mode and flags can be read.
8470 @defmac ASM_OUTPUT_ADDR_VEC_ELT (@var{stream}, @var{value})
8471 This macro should be provided on machines where the addresses
8472 in a dispatch table are absolute.
8474 The definition should be a C statement to output to the stdio stream
8475 @var{stream} an assembler pseudo-instruction to generate a reference to
8476 a label. @var{value} is the number of an internal label whose
8477 definition is output using @code{(*targetm.asm_out.internal_label)}.
8481 fprintf (@var{stream}, "\t.word L%d\n", @var{value})
8485 @defmac ASM_OUTPUT_CASE_LABEL (@var{stream}, @var{prefix}, @var{num}, @var{table})
8486 Define this if the label before a jump-table needs to be output
8487 specially. The first three arguments are the same as for
8488 @code{(*targetm.asm_out.internal_label)}; the fourth argument is the
8489 jump-table which follows (a @code{jump_insn} containing an
8490 @code{addr_vec} or @code{addr_diff_vec}).
8492 This feature is used on system V to output a @code{swbeg} statement
8495 If this macro is not defined, these labels are output with
8496 @code{(*targetm.asm_out.internal_label)}.
8499 @defmac ASM_OUTPUT_CASE_END (@var{stream}, @var{num}, @var{table})
8500 Define this if something special must be output at the end of a
8501 jump-table. The definition should be a C statement to be executed
8502 after the assembler code for the table is written. It should write
8503 the appropriate code to stdio stream @var{stream}. The argument
8504 @var{table} is the jump-table insn, and @var{num} is the label-number
8505 of the preceding label.
8507 If this macro is not defined, nothing special is output at the end of
8511 @deftypefn {Target Hook} void TARGET_ASM_EMIT_UNWIND_LABEL (@var{stream}, @var{decl}, @var{for_eh}, @var{empty})
8512 This target hook emits a label at the beginning of each FDE@. It
8513 should be defined on targets where FDEs need special labels, and it
8514 should write the appropriate label, for the FDE associated with the
8515 function declaration @var{decl}, to the stdio stream @var{stream}.
8516 The third argument, @var{for_eh}, is a boolean: true if this is for an
8517 exception table. The fourth argument, @var{empty}, is a boolean:
8518 true if this is a placeholder label for an omitted FDE@.
8520 The default is that FDEs are not given nonlocal labels.
8523 @deftypefn {Target Hook} void TARGET_ASM_EMIT_EXCEPT_TABLE_LABEL (@var{stream})
8524 This target hook emits a label at the beginning of the exception table.
8525 It should be defined on targets where it is desirable for the table
8526 to be broken up according to function.
8528 The default is that no label is emitted.
8531 @deftypefn {Target Hook} void TARGET_UNWIND_EMIT (FILE * @var{stream}, rtx @var{insn})
8532 This target hook emits and assembly directives required to unwind the
8533 given instruction. This is only used when TARGET_UNWIND_INFO is set.
8536 @node Exception Region Output
8537 @subsection Assembler Commands for Exception Regions
8539 @c prevent bad page break with this line
8541 This describes commands marking the start and the end of an exception
8544 @defmac EH_FRAME_SECTION_NAME
8545 If defined, a C string constant for the name of the section containing
8546 exception handling frame unwind information. If not defined, GCC will
8547 provide a default definition if the target supports named sections.
8548 @file{crtstuff.c} uses this macro to switch to the appropriate section.
8550 You should define this symbol if your target supports DWARF 2 frame
8551 unwind information and the default definition does not work.
8554 @defmac EH_FRAME_IN_DATA_SECTION
8555 If defined, DWARF 2 frame unwind information will be placed in the
8556 data section even though the target supports named sections. This
8557 might be necessary, for instance, if the system linker does garbage
8558 collection and sections cannot be marked as not to be collected.
8560 Do not define this macro unless @code{TARGET_ASM_NAMED_SECTION} is
8564 @defmac EH_TABLES_CAN_BE_READ_ONLY
8565 Define this macro to 1 if your target is such that no frame unwind
8566 information encoding used with non-PIC code will ever require a
8567 runtime relocation, but the linker may not support merging read-only
8568 and read-write sections into a single read-write section.
8571 @defmac MASK_RETURN_ADDR
8572 An rtx used to mask the return address found via @code{RETURN_ADDR_RTX}, so
8573 that it does not contain any extraneous set bits in it.
8576 @defmac DWARF2_UNWIND_INFO
8577 Define this macro to 0 if your target supports DWARF 2 frame unwind
8578 information, but it does not yet work with exception handling.
8579 Otherwise, if your target supports this information (if it defines
8580 @samp{INCOMING_RETURN_ADDR_RTX} and either @samp{UNALIGNED_INT_ASM_OP}
8581 or @samp{OBJECT_FORMAT_ELF}), GCC will provide a default definition of 1.
8583 If @code{TARGET_UNWIND_INFO} is defined, the target specific unwinder
8584 will be used in all cases. Defining this macro will enable the generation
8585 of DWARF 2 frame debugging information.
8587 If @code{TARGET_UNWIND_INFO} is not defined, and this macro is defined to 1,
8588 the DWARF 2 unwinder will be the default exception handling mechanism;
8589 otherwise, the @code{setjmp}/@code{longjmp}-based scheme will be used by
8593 @defmac TARGET_UNWIND_INFO
8594 Define this macro if your target has ABI specified unwind tables. Usually
8595 these will be output by @code{TARGET_UNWIND_EMIT}.
8598 @deftypevr {Target Hook} bool TARGET_UNWIND_TABLES_DEFAULT
8599 This variable should be set to @code{true} if the target ABI requires unwinding
8600 tables even when exceptions are not used.
8603 @defmac MUST_USE_SJLJ_EXCEPTIONS
8604 This macro need only be defined if @code{DWARF2_UNWIND_INFO} is
8605 runtime-variable. In that case, @file{except.h} cannot correctly
8606 determine the corresponding definition of @code{MUST_USE_SJLJ_EXCEPTIONS},
8607 so the target must provide it directly.
8610 @defmac DONT_USE_BUILTIN_SETJMP
8611 Define this macro to 1 if the @code{setjmp}/@code{longjmp}-based scheme
8612 should use the @code{setjmp}/@code{longjmp} functions from the C library
8613 instead of the @code{__builtin_setjmp}/@code{__builtin_longjmp} machinery.
8616 @defmac DWARF_CIE_DATA_ALIGNMENT
8617 This macro need only be defined if the target might save registers in the
8618 function prologue at an offset to the stack pointer that is not aligned to
8619 @code{UNITS_PER_WORD}. The definition should be the negative minimum
8620 alignment if @code{STACK_GROWS_DOWNWARD} is defined, and the positive
8621 minimum alignment otherwise. @xref{SDB and DWARF}. Only applicable if
8622 the target supports DWARF 2 frame unwind information.
8625 @deftypevr {Target Hook} bool TARGET_TERMINATE_DW2_EH_FRAME_INFO
8626 Contains the value true if the target should add a zero word onto the
8627 end of a Dwarf-2 frame info section when used for exception handling.
8628 Default value is false if @code{EH_FRAME_SECTION_NAME} is defined, and
8632 @deftypefn {Target Hook} rtx TARGET_DWARF_REGISTER_SPAN (rtx @var{reg})
8633 Given a register, this hook should return a parallel of registers to
8634 represent where to find the register pieces. Define this hook if the
8635 register and its mode are represented in Dwarf in non-contiguous
8636 locations, or if the register should be represented in more than one
8637 register in Dwarf. Otherwise, this hook should return @code{NULL_RTX}.
8638 If not defined, the default is to return @code{NULL_RTX}.
8641 @deftypefn {Target Hook} void TARGET_INIT_DWARF_REG_SIZES_EXTRA (tree @var{address})
8642 If some registers are represented in Dwarf-2 unwind information in
8643 multiple pieces, define this hook to fill in information about the
8644 sizes of those pieces in the table used by the unwinder at runtime.
8645 It will be called by @code{expand_builtin_init_dwarf_reg_sizes} after
8646 filling in a single size corresponding to each hard register;
8647 @var{address} is the address of the table.
8650 @deftypefn {Target Hook} bool TARGET_ASM_TTYPE (rtx @var{sym})
8651 This hook is used to output a reference from a frame unwinding table to
8652 the type_info object identified by @var{sym}. It should return @code{true}
8653 if the reference was output. Returning @code{false} will cause the
8654 reference to be output using the normal Dwarf2 routines.
8657 @deftypefn {Target Hook} bool TARGET_ARM_EABI_UNWINDER
8658 This hook should be set to @code{true} on targets that use an ARM EABI
8659 based unwinding library, and @code{false} on other targets. This effects
8660 the format of unwinding tables, and how the unwinder in entered after
8661 running a cleanup. The default is @code{false}.
8664 @node Alignment Output
8665 @subsection Assembler Commands for Alignment
8667 @c prevent bad page break with this line
8668 This describes commands for alignment.
8670 @defmac JUMP_ALIGN (@var{label})
8671 The alignment (log base 2) to put in front of @var{label}, which is
8672 a common destination of jumps and has no fallthru incoming edge.
8674 This macro need not be defined if you don't want any special alignment
8675 to be done at such a time. Most machine descriptions do not currently
8678 Unless it's necessary to inspect the @var{label} parameter, it is better
8679 to set the variable @var{align_jumps} in the target's
8680 @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's
8681 selection in @var{align_jumps} in a @code{JUMP_ALIGN} implementation.
8684 @defmac LABEL_ALIGN_AFTER_BARRIER (@var{label})
8685 The alignment (log base 2) to put in front of @var{label}, which follows
8688 This macro need not be defined if you don't want any special alignment
8689 to be done at such a time. Most machine descriptions do not currently
8693 @defmac LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP
8694 The maximum number of bytes to skip when applying
8695 @code{LABEL_ALIGN_AFTER_BARRIER}. This works only if
8696 @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
8699 @defmac LOOP_ALIGN (@var{label})
8700 The alignment (log base 2) to put in front of @var{label}, which follows
8701 a @code{NOTE_INSN_LOOP_BEG} note.
8703 This macro need not be defined if you don't want any special alignment
8704 to be done at such a time. Most machine descriptions do not currently
8707 Unless it's necessary to inspect the @var{label} parameter, it is better
8708 to set the variable @code{align_loops} in the target's
8709 @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's
8710 selection in @code{align_loops} in a @code{LOOP_ALIGN} implementation.
8713 @defmac LOOP_ALIGN_MAX_SKIP
8714 The maximum number of bytes to skip when applying @code{LOOP_ALIGN}.
8715 This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
8718 @defmac LABEL_ALIGN (@var{label})
8719 The alignment (log base 2) to put in front of @var{label}.
8720 If @code{LABEL_ALIGN_AFTER_BARRIER} / @code{LOOP_ALIGN} specify a different alignment,
8721 the maximum of the specified values is used.
8723 Unless it's necessary to inspect the @var{label} parameter, it is better
8724 to set the variable @code{align_labels} in the target's
8725 @code{OVERRIDE_OPTIONS}. Otherwise, you should try to honor the user's
8726 selection in @code{align_labels} in a @code{LABEL_ALIGN} implementation.
8729 @defmac LABEL_ALIGN_MAX_SKIP
8730 The maximum number of bytes to skip when applying @code{LABEL_ALIGN}.
8731 This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
8734 @defmac ASM_OUTPUT_SKIP (@var{stream}, @var{nbytes})
8735 A C statement to output to the stdio stream @var{stream} an assembler
8736 instruction to advance the location counter by @var{nbytes} bytes.
8737 Those bytes should be zero when loaded. @var{nbytes} will be a C
8738 expression of type @code{unsigned HOST_WIDE_INT}.
8741 @defmac ASM_NO_SKIP_IN_TEXT
8742 Define this macro if @code{ASM_OUTPUT_SKIP} should not be used in the
8743 text section because it fails to put zeros in the bytes that are skipped.
8744 This is true on many Unix systems, where the pseudo--op to skip bytes
8745 produces no-op instructions rather than zeros when used in the text
8749 @defmac ASM_OUTPUT_ALIGN (@var{stream}, @var{power})
8750 A C statement to output to the stdio stream @var{stream} an assembler
8751 command to advance the location counter to a multiple of 2 to the
8752 @var{power} bytes. @var{power} will be a C expression of type @code{int}.
8755 @defmac ASM_OUTPUT_ALIGN_WITH_NOP (@var{stream}, @var{power})
8756 Like @code{ASM_OUTPUT_ALIGN}, except that the ``nop'' instruction is used
8757 for padding, if necessary.
8760 @defmac ASM_OUTPUT_MAX_SKIP_ALIGN (@var{stream}, @var{power}, @var{max_skip})
8761 A C statement to output to the stdio stream @var{stream} an assembler
8762 command to advance the location counter to a multiple of 2 to the
8763 @var{power} bytes, but only if @var{max_skip} or fewer bytes are needed to
8764 satisfy the alignment request. @var{power} and @var{max_skip} will be
8765 a C expression of type @code{int}.
8769 @node Debugging Info
8770 @section Controlling Debugging Information Format
8772 @c prevent bad page break with this line
8773 This describes how to specify debugging information.
8776 * All Debuggers:: Macros that affect all debugging formats uniformly.
8777 * DBX Options:: Macros enabling specific options in DBX format.
8778 * DBX Hooks:: Hook macros for varying DBX format.
8779 * File Names and DBX:: Macros controlling output of file names in DBX format.
8780 * SDB and DWARF:: Macros for SDB (COFF) and DWARF formats.
8781 * VMS Debug:: Macros for VMS debug format.
8785 @subsection Macros Affecting All Debugging Formats
8787 @c prevent bad page break with this line
8788 These macros affect all debugging formats.
8790 @defmac DBX_REGISTER_NUMBER (@var{regno})
8791 A C expression that returns the DBX register number for the compiler
8792 register number @var{regno}. In the default macro provided, the value
8793 of this expression will be @var{regno} itself. But sometimes there are
8794 some registers that the compiler knows about and DBX does not, or vice
8795 versa. In such cases, some register may need to have one number in the
8796 compiler and another for DBX@.
8798 If two registers have consecutive numbers inside GCC, and they can be
8799 used as a pair to hold a multiword value, then they @emph{must} have
8800 consecutive numbers after renumbering with @code{DBX_REGISTER_NUMBER}.
8801 Otherwise, debuggers will be unable to access such a pair, because they
8802 expect register pairs to be consecutive in their own numbering scheme.
8804 If you find yourself defining @code{DBX_REGISTER_NUMBER} in way that
8805 does not preserve register pairs, then what you must do instead is
8806 redefine the actual register numbering scheme.
8809 @defmac DEBUGGER_AUTO_OFFSET (@var{x})
8810 A C expression that returns the integer offset value for an automatic
8811 variable having address @var{x} (an RTL expression). The default
8812 computation assumes that @var{x} is based on the frame-pointer and
8813 gives the offset from the frame-pointer. This is required for targets
8814 that produce debugging output for DBX or COFF-style debugging output
8815 for SDB and allow the frame-pointer to be eliminated when the
8816 @option{-g} options is used.
8819 @defmac DEBUGGER_ARG_OFFSET (@var{offset}, @var{x})
8820 A C expression that returns the integer offset value for an argument
8821 having address @var{x} (an RTL expression). The nominal offset is
8825 @defmac PREFERRED_DEBUGGING_TYPE
8826 A C expression that returns the type of debugging output GCC should
8827 produce when the user specifies just @option{-g}. Define
8828 this if you have arranged for GCC to support more than one format of
8829 debugging output. Currently, the allowable values are @code{DBX_DEBUG},
8830 @code{SDB_DEBUG}, @code{DWARF_DEBUG}, @code{DWARF2_DEBUG},
8831 @code{XCOFF_DEBUG}, @code{VMS_DEBUG}, and @code{VMS_AND_DWARF2_DEBUG}.
8833 When the user specifies @option{-ggdb}, GCC normally also uses the
8834 value of this macro to select the debugging output format, but with two
8835 exceptions. If @code{DWARF2_DEBUGGING_INFO} is defined, GCC uses the
8836 value @code{DWARF2_DEBUG}. Otherwise, if @code{DBX_DEBUGGING_INFO} is
8837 defined, GCC uses @code{DBX_DEBUG}.
8839 The value of this macro only affects the default debugging output; the
8840 user can always get a specific type of output by using @option{-gstabs},
8841 @option{-gcoff}, @option{-gdwarf-2}, @option{-gxcoff}, or @option{-gvms}.
8845 @subsection Specific Options for DBX Output
8847 @c prevent bad page break with this line
8848 These are specific options for DBX output.
8850 @defmac DBX_DEBUGGING_INFO
8851 Define this macro if GCC should produce debugging output for DBX
8852 in response to the @option{-g} option.
8855 @defmac XCOFF_DEBUGGING_INFO
8856 Define this macro if GCC should produce XCOFF format debugging output
8857 in response to the @option{-g} option. This is a variant of DBX format.
8860 @defmac DEFAULT_GDB_EXTENSIONS
8861 Define this macro to control whether GCC should by default generate
8862 GDB's extended version of DBX debugging information (assuming DBX-format
8863 debugging information is enabled at all). If you don't define the
8864 macro, the default is 1: always generate the extended information
8865 if there is any occasion to.
8868 @defmac DEBUG_SYMS_TEXT
8869 Define this macro if all @code{.stabs} commands should be output while
8870 in the text section.
8873 @defmac ASM_STABS_OP
8874 A C string constant, including spacing, naming the assembler pseudo op to
8875 use instead of @code{"\t.stabs\t"} to define an ordinary debugging symbol.
8876 If you don't define this macro, @code{"\t.stabs\t"} is used. This macro
8877 applies only to DBX debugging information format.
8880 @defmac ASM_STABD_OP
8881 A C string constant, including spacing, naming the assembler pseudo op to
8882 use instead of @code{"\t.stabd\t"} to define a debugging symbol whose
8883 value is the current location. If you don't define this macro,
8884 @code{"\t.stabd\t"} is used. This macro applies only to DBX debugging
8888 @defmac ASM_STABN_OP
8889 A C string constant, including spacing, naming the assembler pseudo op to
8890 use instead of @code{"\t.stabn\t"} to define a debugging symbol with no
8891 name. If you don't define this macro, @code{"\t.stabn\t"} is used. This
8892 macro applies only to DBX debugging information format.
8895 @defmac DBX_NO_XREFS
8896 Define this macro if DBX on your system does not support the construct
8897 @samp{xs@var{tagname}}. On some systems, this construct is used to
8898 describe a forward reference to a structure named @var{tagname}.
8899 On other systems, this construct is not supported at all.
8902 @defmac DBX_CONTIN_LENGTH
8903 A symbol name in DBX-format debugging information is normally
8904 continued (split into two separate @code{.stabs} directives) when it
8905 exceeds a certain length (by default, 80 characters). On some
8906 operating systems, DBX requires this splitting; on others, splitting
8907 must not be done. You can inhibit splitting by defining this macro
8908 with the value zero. You can override the default splitting-length by
8909 defining this macro as an expression for the length you desire.
8912 @defmac DBX_CONTIN_CHAR
8913 Normally continuation is indicated by adding a @samp{\} character to
8914 the end of a @code{.stabs} string when a continuation follows. To use
8915 a different character instead, define this macro as a character
8916 constant for the character you want to use. Do not define this macro
8917 if backslash is correct for your system.
8920 @defmac DBX_STATIC_STAB_DATA_SECTION
8921 Define this macro if it is necessary to go to the data section before
8922 outputting the @samp{.stabs} pseudo-op for a non-global static
8926 @defmac DBX_TYPE_DECL_STABS_CODE
8927 The value to use in the ``code'' field of the @code{.stabs} directive
8928 for a typedef. The default is @code{N_LSYM}.
8931 @defmac DBX_STATIC_CONST_VAR_CODE
8932 The value to use in the ``code'' field of the @code{.stabs} directive
8933 for a static variable located in the text section. DBX format does not
8934 provide any ``right'' way to do this. The default is @code{N_FUN}.
8937 @defmac DBX_REGPARM_STABS_CODE
8938 The value to use in the ``code'' field of the @code{.stabs} directive
8939 for a parameter passed in registers. DBX format does not provide any
8940 ``right'' way to do this. The default is @code{N_RSYM}.
8943 @defmac DBX_REGPARM_STABS_LETTER
8944 The letter to use in DBX symbol data to identify a symbol as a parameter
8945 passed in registers. DBX format does not customarily provide any way to
8946 do this. The default is @code{'P'}.
8949 @defmac DBX_FUNCTION_FIRST
8950 Define this macro if the DBX information for a function and its
8951 arguments should precede the assembler code for the function. Normally,
8952 in DBX format, the debugging information entirely follows the assembler
8956 @defmac DBX_BLOCKS_FUNCTION_RELATIVE
8957 Define this macro, with value 1, if the value of a symbol describing
8958 the scope of a block (@code{N_LBRAC} or @code{N_RBRAC}) should be
8959 relative to the start of the enclosing function. Normally, GCC uses
8960 an absolute address.
8963 @defmac DBX_LINES_FUNCTION_RELATIVE
8964 Define this macro, with value 1, if the value of a symbol indicating
8965 the current line number (@code{N_SLINE}) should be relative to the
8966 start of the enclosing function. Normally, GCC uses an absolute address.
8969 @defmac DBX_USE_BINCL
8970 Define this macro if GCC should generate @code{N_BINCL} and
8971 @code{N_EINCL} stabs for included header files, as on Sun systems. This
8972 macro also directs GCC to output a type number as a pair of a file
8973 number and a type number within the file. Normally, GCC does not
8974 generate @code{N_BINCL} or @code{N_EINCL} stabs, and it outputs a single
8975 number for a type number.
8979 @subsection Open-Ended Hooks for DBX Format
8981 @c prevent bad page break with this line
8982 These are hooks for DBX format.
8984 @defmac DBX_OUTPUT_LBRAC (@var{stream}, @var{name})
8985 Define this macro to say how to output to @var{stream} the debugging
8986 information for the start of a scope level for variable names. The
8987 argument @var{name} is the name of an assembler symbol (for use with
8988 @code{assemble_name}) whose value is the address where the scope begins.
8991 @defmac DBX_OUTPUT_RBRAC (@var{stream}, @var{name})
8992 Like @code{DBX_OUTPUT_LBRAC}, but for the end of a scope level.
8995 @defmac DBX_OUTPUT_NFUN (@var{stream}, @var{lscope_label}, @var{decl})
8996 Define this macro if the target machine requires special handling to
8997 output an @code{N_FUN} entry for the function @var{decl}.
9000 @defmac DBX_OUTPUT_SOURCE_LINE (@var{stream}, @var{line}, @var{counter})
9001 A C statement to output DBX debugging information before code for line
9002 number @var{line} of the current source file to the stdio stream
9003 @var{stream}. @var{counter} is the number of time the macro was
9004 invoked, including the current invocation; it is intended to generate
9005 unique labels in the assembly output.
9007 This macro should not be defined if the default output is correct, or
9008 if it can be made correct by defining @code{DBX_LINES_FUNCTION_RELATIVE}.
9011 @defmac NO_DBX_FUNCTION_END
9012 Some stabs encapsulation formats (in particular ECOFF), cannot handle the
9013 @code{.stabs "",N_FUN,,0,0,Lscope-function-1} gdb dbx extension construct.
9014 On those machines, define this macro to turn this feature off without
9015 disturbing the rest of the gdb extensions.
9018 @defmac NO_DBX_BNSYM_ENSYM
9019 Some assemblers cannot handle the @code{.stabd BNSYM/ENSYM,0,0} gdb dbx
9020 extension construct. On those machines, define this macro to turn this
9021 feature off without disturbing the rest of the gdb extensions.
9024 @node File Names and DBX
9025 @subsection File Names in DBX Format
9027 @c prevent bad page break with this line
9028 This describes file names in DBX format.
9030 @defmac DBX_OUTPUT_MAIN_SOURCE_FILENAME (@var{stream}, @var{name})
9031 A C statement to output DBX debugging information to the stdio stream
9032 @var{stream}, which indicates that file @var{name} is the main source
9033 file---the file specified as the input file for compilation.
9034 This macro is called only once, at the beginning of compilation.
9036 This macro need not be defined if the standard form of output
9037 for DBX debugging information is appropriate.
9039 It may be necessary to refer to a label equal to the beginning of the
9040 text section. You can use @samp{assemble_name (stream, ltext_label_name)}
9041 to do so. If you do this, you must also set the variable
9042 @var{used_ltext_label_name} to @code{true}.
9045 @defmac NO_DBX_MAIN_SOURCE_DIRECTORY
9046 Define this macro, with value 1, if GCC should not emit an indication
9047 of the current directory for compilation and current source language at
9048 the beginning of the file.
9051 @defmac NO_DBX_GCC_MARKER
9052 Define this macro, with value 1, if GCC should not emit an indication
9053 that this object file was compiled by GCC@. The default is to emit
9054 an @code{N_OPT} stab at the beginning of every source file, with
9055 @samp{gcc2_compiled.} for the string and value 0.
9058 @defmac DBX_OUTPUT_MAIN_SOURCE_FILE_END (@var{stream}, @var{name})
9059 A C statement to output DBX debugging information at the end of
9060 compilation of the main source file @var{name}. Output should be
9061 written to the stdio stream @var{stream}.
9063 If you don't define this macro, nothing special is output at the end
9064 of compilation, which is correct for most machines.
9067 @defmac DBX_OUTPUT_NULL_N_SO_AT_MAIN_SOURCE_FILE_END
9068 Define this macro @emph{instead of} defining
9069 @code{DBX_OUTPUT_MAIN_SOURCE_FILE_END}, if what needs to be output at
9070 the end of compilation is an @code{N_SO} stab with an empty string,
9071 whose value is the highest absolute text address in the file.
9076 @subsection Macros for SDB and DWARF Output
9078 @c prevent bad page break with this line
9079 Here are macros for SDB and DWARF output.
9081 @defmac SDB_DEBUGGING_INFO
9082 Define this macro if GCC should produce COFF-style debugging output
9083 for SDB in response to the @option{-g} option.
9086 @defmac DWARF2_DEBUGGING_INFO
9087 Define this macro if GCC should produce dwarf version 2 format
9088 debugging output in response to the @option{-g} option.
9090 @deftypefn {Target Hook} int TARGET_DWARF_CALLING_CONVENTION (tree @var{function})
9091 Define this to enable the dwarf attribute @code{DW_AT_calling_convention} to
9092 be emitted for each function. Instead of an integer return the enum
9093 value for the @code{DW_CC_} tag.
9096 To support optional call frame debugging information, you must also
9097 define @code{INCOMING_RETURN_ADDR_RTX} and either set
9098 @code{RTX_FRAME_RELATED_P} on the prologue insns if you use RTL for the
9099 prologue, or call @code{dwarf2out_def_cfa} and @code{dwarf2out_reg_save}
9100 as appropriate from @code{TARGET_ASM_FUNCTION_PROLOGUE} if you don't.
9103 @defmac DWARF2_FRAME_INFO
9104 Define this macro to a nonzero value if GCC should always output
9105 Dwarf 2 frame information. If @code{DWARF2_UNWIND_INFO}
9106 (@pxref{Exception Region Output} is nonzero, GCC will output this
9107 information not matter how you define @code{DWARF2_FRAME_INFO}.
9110 @defmac DWARF2_ASM_LINE_DEBUG_INFO
9111 Define this macro to be a nonzero value if the assembler can generate Dwarf 2
9112 line debug info sections. This will result in much more compact line number
9113 tables, and hence is desirable if it works.
9116 @defmac ASM_OUTPUT_DWARF_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
9117 A C statement to issue assembly directives that create a difference
9118 @var{lab1} minus @var{lab2}, using an integer of the given @var{size}.
9121 @defmac ASM_OUTPUT_DWARF_OFFSET (@var{stream}, @var{size}, @var{label}, @var{section})
9122 A C statement to issue assembly directives that create a
9123 section-relative reference to the given @var{label}, using an integer of the
9124 given @var{size}. The label is known to be defined in the given @var{section}.
9127 @defmac ASM_OUTPUT_DWARF_PCREL (@var{stream}, @var{size}, @var{label})
9128 A C statement to issue assembly directives that create a self-relative
9129 reference to the given @var{label}, using an integer of the given @var{size}.
9132 @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_DWARF_DTPREL (FILE *@var{FILE}, int @var{size}, rtx @var{x})
9133 If defined, this target hook is a function which outputs a DTP-relative
9134 reference to the given TLS symbol of the specified size.
9137 @defmac PUT_SDB_@dots{}
9138 Define these macros to override the assembler syntax for the special
9139 SDB assembler directives. See @file{sdbout.c} for a list of these
9140 macros and their arguments. If the standard syntax is used, you need
9141 not define them yourself.
9145 Some assemblers do not support a semicolon as a delimiter, even between
9146 SDB assembler directives. In that case, define this macro to be the
9147 delimiter to use (usually @samp{\n}). It is not necessary to define
9148 a new set of @code{PUT_SDB_@var{op}} macros if this is the only change
9152 @defmac SDB_ALLOW_UNKNOWN_REFERENCES
9153 Define this macro to allow references to unknown structure,
9154 union, or enumeration tags to be emitted. Standard COFF does not
9155 allow handling of unknown references, MIPS ECOFF has support for
9159 @defmac SDB_ALLOW_FORWARD_REFERENCES
9160 Define this macro to allow references to structure, union, or
9161 enumeration tags that have not yet been seen to be handled. Some
9162 assemblers choke if forward tags are used, while some require it.
9165 @defmac SDB_OUTPUT_SOURCE_LINE (@var{stream}, @var{line})
9166 A C statement to output SDB debugging information before code for line
9167 number @var{line} of the current source file to the stdio stream
9168 @var{stream}. The default is to emit an @code{.ln} directive.
9173 @subsection Macros for VMS Debug Format
9175 @c prevent bad page break with this line
9176 Here are macros for VMS debug format.
9178 @defmac VMS_DEBUGGING_INFO
9179 Define this macro if GCC should produce debugging output for VMS
9180 in response to the @option{-g} option. The default behavior for VMS
9181 is to generate minimal debug info for a traceback in the absence of
9182 @option{-g} unless explicitly overridden with @option{-g0}. This
9183 behavior is controlled by @code{OPTIMIZATION_OPTIONS} and
9184 @code{OVERRIDE_OPTIONS}.
9187 @node Floating Point
9188 @section Cross Compilation and Floating Point
9189 @cindex cross compilation and floating point
9190 @cindex floating point and cross compilation
9192 While all modern machines use twos-complement representation for integers,
9193 there are a variety of representations for floating point numbers. This
9194 means that in a cross-compiler the representation of floating point numbers
9195 in the compiled program may be different from that used in the machine
9196 doing the compilation.
9198 Because different representation systems may offer different amounts of
9199 range and precision, all floating point constants must be represented in
9200 the target machine's format. Therefore, the cross compiler cannot
9201 safely use the host machine's floating point arithmetic; it must emulate
9202 the target's arithmetic. To ensure consistency, GCC always uses
9203 emulation to work with floating point values, even when the host and
9204 target floating point formats are identical.
9206 The following macros are provided by @file{real.h} for the compiler to
9207 use. All parts of the compiler which generate or optimize
9208 floating-point calculations must use these macros. They may evaluate
9209 their operands more than once, so operands must not have side effects.
9211 @defmac REAL_VALUE_TYPE
9212 The C data type to be used to hold a floating point value in the target
9213 machine's format. Typically this is a @code{struct} containing an
9214 array of @code{HOST_WIDE_INT}, but all code should treat it as an opaque
9218 @deftypefn Macro int REAL_VALUES_EQUAL (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
9219 Compares for equality the two values, @var{x} and @var{y}. If the target
9220 floating point format supports negative zeroes and/or NaNs,
9221 @samp{REAL_VALUES_EQUAL (-0.0, 0.0)} is true, and
9222 @samp{REAL_VALUES_EQUAL (NaN, NaN)} is false.
9225 @deftypefn Macro int REAL_VALUES_LESS (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
9226 Tests whether @var{x} is less than @var{y}.
9229 @deftypefn Macro HOST_WIDE_INT REAL_VALUE_FIX (REAL_VALUE_TYPE @var{x})
9230 Truncates @var{x} to a signed integer, rounding toward zero.
9233 @deftypefn Macro {unsigned HOST_WIDE_INT} REAL_VALUE_UNSIGNED_FIX (REAL_VALUE_TYPE @var{x})
9234 Truncates @var{x} to an unsigned integer, rounding toward zero. If
9235 @var{x} is negative, returns zero.
9238 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ATOF (const char *@var{string}, enum machine_mode @var{mode})
9239 Converts @var{string} into a floating point number in the target machine's
9240 representation for mode @var{mode}. This routine can handle both
9241 decimal and hexadecimal floating point constants, using the syntax
9242 defined by the C language for both.
9245 @deftypefn Macro int REAL_VALUE_NEGATIVE (REAL_VALUE_TYPE @var{x})
9246 Returns 1 if @var{x} is negative (including negative zero), 0 otherwise.
9249 @deftypefn Macro int REAL_VALUE_ISINF (REAL_VALUE_TYPE @var{x})
9250 Determines whether @var{x} represents infinity (positive or negative).
9253 @deftypefn Macro int REAL_VALUE_ISNAN (REAL_VALUE_TYPE @var{x})
9254 Determines whether @var{x} represents a ``NaN'' (not-a-number).
9257 @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})
9258 Calculates an arithmetic operation on the two floating point values
9259 @var{x} and @var{y}, storing the result in @var{output} (which must be a
9262 The operation to be performed is specified by @var{code}. Only the
9263 following codes are supported: @code{PLUS_EXPR}, @code{MINUS_EXPR},
9264 @code{MULT_EXPR}, @code{RDIV_EXPR}, @code{MAX_EXPR}, @code{MIN_EXPR}.
9266 If @code{REAL_ARITHMETIC} is asked to evaluate division by zero and the
9267 target's floating point format cannot represent infinity, it will call
9268 @code{abort}. Callers should check for this situation first, using
9269 @code{MODE_HAS_INFINITIES}. @xref{Storage Layout}.
9272 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_NEGATE (REAL_VALUE_TYPE @var{x})
9273 Returns the negative of the floating point value @var{x}.
9276 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ABS (REAL_VALUE_TYPE @var{x})
9277 Returns the absolute value of @var{x}.
9280 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_TRUNCATE (REAL_VALUE_TYPE @var{mode}, enum machine_mode @var{x})
9281 Truncates the floating point value @var{x} to fit in @var{mode}. The
9282 return value is still a full-size @code{REAL_VALUE_TYPE}, but it has an
9283 appropriate bit pattern to be output as a floating constant whose
9284 precision accords with mode @var{mode}.
9287 @deftypefn Macro void REAL_VALUE_TO_INT (HOST_WIDE_INT @var{low}, HOST_WIDE_INT @var{high}, REAL_VALUE_TYPE @var{x})
9288 Converts a floating point value @var{x} into a double-precision integer
9289 which is then stored into @var{low} and @var{high}. If the value is not
9290 integral, it is truncated.
9293 @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})
9294 Converts a double-precision integer found in @var{low} and @var{high},
9295 into a floating point value which is then stored into @var{x}. The
9296 value is truncated to fit in mode @var{mode}.
9299 @node Mode Switching
9300 @section Mode Switching Instructions
9301 @cindex mode switching
9302 The following macros control mode switching optimizations:
9304 @defmac OPTIMIZE_MODE_SWITCHING (@var{entity})
9305 Define this macro if the port needs extra instructions inserted for mode
9306 switching in an optimizing compilation.
9308 For an example, the SH4 can perform both single and double precision
9309 floating point operations, but to perform a single precision operation,
9310 the FPSCR PR bit has to be cleared, while for a double precision
9311 operation, this bit has to be set. Changing the PR bit requires a general
9312 purpose register as a scratch register, hence these FPSCR sets have to
9313 be inserted before reload, i.e.@: you can't put this into instruction emitting
9314 or @code{TARGET_MACHINE_DEPENDENT_REORG}.
9316 You can have multiple entities that are mode-switched, and select at run time
9317 which entities actually need it. @code{OPTIMIZE_MODE_SWITCHING} should
9318 return nonzero for any @var{entity} that needs mode-switching.
9319 If you define this macro, you also have to define
9320 @code{NUM_MODES_FOR_MODE_SWITCHING}, @code{MODE_NEEDED},
9321 @code{MODE_PRIORITY_TO_MODE} and @code{EMIT_MODE_SET}.
9322 @code{MODE_AFTER}, @code{MODE_ENTRY}, and @code{MODE_EXIT}
9326 @defmac NUM_MODES_FOR_MODE_SWITCHING
9327 If you define @code{OPTIMIZE_MODE_SWITCHING}, you have to define this as
9328 initializer for an array of integers. Each initializer element
9329 N refers to an entity that needs mode switching, and specifies the number
9330 of different modes that might need to be set for this entity.
9331 The position of the initializer in the initializer---starting counting at
9332 zero---determines the integer that is used to refer to the mode-switched
9334 In macros that take mode arguments / yield a mode result, modes are
9335 represented as numbers 0 @dots{} N @minus{} 1. N is used to specify that no mode
9336 switch is needed / supplied.
9339 @defmac MODE_NEEDED (@var{entity}, @var{insn})
9340 @var{entity} is an integer specifying a mode-switched entity. If
9341 @code{OPTIMIZE_MODE_SWITCHING} is defined, you must define this macro to
9342 return an integer value not larger than the corresponding element in
9343 @code{NUM_MODES_FOR_MODE_SWITCHING}, to denote the mode that @var{entity} must
9344 be switched into prior to the execution of @var{insn}.
9347 @defmac MODE_AFTER (@var{mode}, @var{insn})
9348 If this macro is defined, it is evaluated for every @var{insn} during
9349 mode switching. It determines the mode that an insn results in (if
9350 different from the incoming mode).
9353 @defmac MODE_ENTRY (@var{entity})
9354 If this macro is defined, it is evaluated for every @var{entity} that needs
9355 mode switching. It should evaluate to an integer, which is a mode that
9356 @var{entity} is assumed to be switched to at function entry. If @code{MODE_ENTRY}
9357 is defined then @code{MODE_EXIT} must be defined.
9360 @defmac MODE_EXIT (@var{entity})
9361 If this macro is defined, it is evaluated for every @var{entity} that needs
9362 mode switching. It should evaluate to an integer, which is a mode that
9363 @var{entity} is assumed to be switched to at function exit. If @code{MODE_EXIT}
9364 is defined then @code{MODE_ENTRY} must be defined.
9367 @defmac MODE_PRIORITY_TO_MODE (@var{entity}, @var{n})
9368 This macro specifies the order in which modes for @var{entity} are processed.
9369 0 is the highest priority, @code{NUM_MODES_FOR_MODE_SWITCHING[@var{entity}] - 1} the
9370 lowest. The value of the macro should be an integer designating a mode
9371 for @var{entity}. For any fixed @var{entity}, @code{mode_priority_to_mode}
9372 (@var{entity}, @var{n}) shall be a bijection in 0 @dots{}
9373 @code{num_modes_for_mode_switching[@var{entity}] - 1}.
9376 @defmac EMIT_MODE_SET (@var{entity}, @var{mode}, @var{hard_regs_live})
9377 Generate one or more insns to set @var{entity} to @var{mode}.
9378 @var{hard_reg_live} is the set of hard registers live at the point where
9379 the insn(s) are to be inserted.
9382 @node Target Attributes
9383 @section Defining target-specific uses of @code{__attribute__}
9384 @cindex target attributes
9385 @cindex machine attributes
9386 @cindex attributes, target-specific
9388 Target-specific attributes may be defined for functions, data and types.
9389 These are described using the following target hooks; they also need to
9390 be documented in @file{extend.texi}.
9392 @deftypevr {Target Hook} {const struct attribute_spec *} TARGET_ATTRIBUTE_TABLE
9393 If defined, this target hook points to an array of @samp{struct
9394 attribute_spec} (defined in @file{tree.h}) specifying the machine
9395 specific attributes for this target and some of the restrictions on the
9396 entities to which these attributes are applied and the arguments they
9400 @deftypefn {Target Hook} int TARGET_COMP_TYPE_ATTRIBUTES (tree @var{type1}, tree @var{type2})
9401 If defined, this target hook is a function which returns zero if the attributes on
9402 @var{type1} and @var{type2} are incompatible, one if they are compatible,
9403 and two if they are nearly compatible (which causes a warning to be
9404 generated). If this is not defined, machine-specific attributes are
9405 supposed always to be compatible.
9408 @deftypefn {Target Hook} void TARGET_SET_DEFAULT_TYPE_ATTRIBUTES (tree @var{type})
9409 If defined, this target hook is a function which assigns default attributes to
9410 newly defined @var{type}.
9413 @deftypefn {Target Hook} tree TARGET_MERGE_TYPE_ATTRIBUTES (tree @var{type1}, tree @var{type2})
9414 Define this target hook if the merging of type attributes needs special
9415 handling. If defined, the result is a list of the combined
9416 @code{TYPE_ATTRIBUTES} of @var{type1} and @var{type2}. It is assumed
9417 that @code{comptypes} has already been called and returned 1. This
9418 function may call @code{merge_attributes} to handle machine-independent
9422 @deftypefn {Target Hook} tree TARGET_MERGE_DECL_ATTRIBUTES (tree @var{olddecl}, tree @var{newdecl})
9423 Define this target hook if the merging of decl attributes needs special
9424 handling. If defined, the result is a list of the combined
9425 @code{DECL_ATTRIBUTES} of @var{olddecl} and @var{newdecl}.
9426 @var{newdecl} is a duplicate declaration of @var{olddecl}. Examples of
9427 when this is needed are when one attribute overrides another, or when an
9428 attribute is nullified by a subsequent definition. This function may
9429 call @code{merge_attributes} to handle machine-independent merging.
9431 @findex TARGET_DLLIMPORT_DECL_ATTRIBUTES
9432 If the only target-specific handling you require is @samp{dllimport}
9433 for Microsoft Windows targets, you should define the macro
9434 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES} to @code{1}. The compiler
9435 will then define a function called
9436 @code{merge_dllimport_decl_attributes} which can then be defined as
9437 the expansion of @code{TARGET_MERGE_DECL_ATTRIBUTES}. You can also
9438 add @code{handle_dll_attribute} in the attribute table for your port
9439 to perform initial processing of the @samp{dllimport} and
9440 @samp{dllexport} attributes. This is done in @file{i386/cygwin.h} and
9441 @file{i386/i386.c}, for example.
9444 @deftypefn {Target Hook} bool TARGET_VALID_DLLIMPORT_ATTRIBUTE_P (tree @var{decl})
9445 @var{decl} is a variable or function with @code{__attribute__((dllimport))}
9446 specified. Use this hook if the target needs to add extra validation
9447 checks to @code{handle_dll_attribute}.
9450 @defmac TARGET_DECLSPEC
9451 Define this macro to a nonzero value if you want to treat
9452 @code{__declspec(X)} as equivalent to @code{__attribute((X))}. By
9453 default, this behavior is enabled only for targets that define
9454 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES}. The current implementation
9455 of @code{__declspec} is via a built-in macro, but you should not rely
9456 on this implementation detail.
9459 @deftypefn {Target Hook} void TARGET_INSERT_ATTRIBUTES (tree @var{node}, tree *@var{attr_ptr})
9460 Define this target hook if you want to be able to add attributes to a decl
9461 when it is being created. This is normally useful for back ends which
9462 wish to implement a pragma by using the attributes which correspond to
9463 the pragma's effect. The @var{node} argument is the decl which is being
9464 created. The @var{attr_ptr} argument is a pointer to the attribute list
9465 for this decl. The list itself should not be modified, since it may be
9466 shared with other decls, but attributes may be chained on the head of
9467 the list and @code{*@var{attr_ptr}} modified to point to the new
9468 attributes, or a copy of the list may be made if further changes are
9472 @deftypefn {Target Hook} bool TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P (tree @var{fndecl})
9474 This target hook returns @code{true} if it is ok to inline @var{fndecl}
9475 into the current function, despite its having target-specific
9476 attributes, @code{false} otherwise. By default, if a function has a
9477 target specific attribute attached to it, it will not be inlined.
9480 @deftypefn {Target Hook} bool TARGET_VALID_OPTION_ATTRIBUTE_P (tree @var{fndecl}, tree @var{name}, tree @var{args}, int @var{flags})
9481 This hook is called to parse the @code{attribute(option("..."))}, and
9482 it allows the function to set different target machine compile time
9483 options for the current function that might be different than the
9484 options specified on the command line. The hook should return
9485 @code{true} if the options are valid.
9487 The hook should set the @var{DECL_FUNCTION_SPECIFIC_TARGET} field in
9488 the function declaration to hold a pointer to a target specific
9489 @var{struct cl_target_option} structure.
9492 @deftypefn {Target Hook} void TARGET_OPTION_SAVE (struct cl_target_option *@var{ptr})
9493 This hook is called to save any additional target specific information
9494 in the @var{struct cl_target_option} structure for function specific
9496 @xref{Option file format}.
9499 @deftypefn {Target Hook} void TARGET_OPTION_RESTORE (struct cl_target_option *@var{ptr})
9500 This hook is called to restore any additional target specific
9501 information in the @var{struct cl_target_option} structure for
9502 function specific options.
9505 @deftypefn {Target Hook} void TARGET_OPTION_PRINT (struct cl_target_option *@var{ptr})
9506 This hook is called to print any additional target specific
9507 information in the @var{struct cl_target_option} structure for
9508 function specific options.
9511 @deftypefn {Target Hook} bool TARGET_OPTION_PRAGMA_PARSE (target @var{args})
9512 This target hook parses the options for @code{#pragma GCC option} to
9513 set the machine specific options for functions that occur later in the
9514 input stream. The options should be the same as handled by the
9515 @code{TARGET_VALID_OPTION_ATTRIBUTE_P} hook.
9518 @deftypefn {Target Hook} bool TARGET_CAN_INLINE_P (tree @var{caller}, tree @var{callee})
9519 This target hook returns @code{false} if the @var{caller} function
9520 cannot inline @var{callee}, based on target specific information. By
9521 default, inlining is not allowed if the callee function has function
9522 specific target options and the caller does not use the same options.
9526 @section Emulating TLS
9527 @cindex Emulated TLS
9529 For targets whose psABI does not provide Thread Local Storage via
9530 specific relocations and instruction sequences, an emulation layer is
9531 used. A set of target hooks allows this emulation layer to be
9532 configured for the requirements of a particular target. For instance
9533 the psABI may in fact specify TLS support in terms of an emulation
9536 The emulation layer works by creating a control object for every TLS
9537 object. To access the TLS object, a lookup function is provided
9538 which, when given the address of the control object, will return the
9539 address of the current thread's instance of the TLS object.
9541 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_GET_ADDRESS
9542 Contains the name of the helper function that uses a TLS control
9543 object to locate a TLS instance. The default causes libgcc's
9544 emulated TLS helper function to be used.
9547 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_REGISTER_COMMON
9548 Contains the name of the helper function that should be used at
9549 program startup to register TLS objects that are implicitly
9550 initialized to zero. If this is @code{NULL}, all TLS objects will
9551 have explicit initializers. The default causes libgcc's emulated TLS
9552 registration function to be used.
9555 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_VAR_SECTION
9556 Contains the name of the section in which TLS control variables should
9557 be placed. The default of @code{NULL} allows these to be placed in
9561 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_TMPL_SECTION
9562 Contains the name of the section in which TLS initializers should be
9563 placed. The default of @code{NULL} allows these to be placed in any
9567 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_VAR_PREFIX
9568 Contains the prefix to be prepended to TLS control variable names.
9569 The default of @code{NULL} uses a target-specific prefix.
9572 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_TMPL_PREFIX
9573 Contains the prefix to be prepended to TLS initializer objects. The
9574 default of @code{NULL} uses a target-specific prefix.
9577 @deftypefn {Target Hook} tree TARGET_EMUTLS_VAR_FIELDS (tree @var{type}, tree *@var{name})
9578 Specifies a function that generates the FIELD_DECLs for a TLS control
9579 object type. @var{type} is the RECORD_TYPE the fields are for and
9580 @var{name} should be filled with the structure tag, if the default of
9581 @code{__emutls_object} is unsuitable. The default creates a type suitable
9582 for libgcc's emulated TLS function.
9585 @deftypefn {Target Hook} tree TARGET_EMUTLS_VAR_INIT (tree @var{var}, tree @var{decl}, tree @var{tmpl_addr})
9586 Specifies a function that generates the CONSTRUCTOR to initialize a
9587 TLS control object. @var{var} is the TLS control object, @var{decl}
9588 is the TLS object and @var{tmpl_addr} is the address of the
9589 initializer. The default initializes libgcc's emulated TLS control object.
9592 @deftypevr {Target Hook} {bool} TARGET_EMUTLS_VAR_ALIGN_FIXED
9593 Specifies whether the alignment of TLS control variable objects is
9594 fixed and should not be increased as some backends may do to optimize
9595 single objects. The default is false.
9598 @deftypevr {Target Hook} {bool} TARGET_EMUTLS_DEBUG_FORM_TLS_ADDRESS
9599 Specifies whether a DWARF @code{DW_OP_form_tls_address} location descriptor
9600 may be used to describe emulated TLS control objects.
9603 @node MIPS Coprocessors
9604 @section Defining coprocessor specifics for MIPS targets.
9605 @cindex MIPS coprocessor-definition macros
9607 The MIPS specification allows MIPS implementations to have as many as 4
9608 coprocessors, each with as many as 32 private registers. GCC supports
9609 accessing these registers and transferring values between the registers
9610 and memory using asm-ized variables. For example:
9613 register unsigned int cp0count asm ("c0r1");
9619 (``c0r1'' is the default name of register 1 in coprocessor 0; alternate
9620 names may be added as described below, or the default names may be
9621 overridden entirely in @code{SUBTARGET_CONDITIONAL_REGISTER_USAGE}.)
9623 Coprocessor registers are assumed to be epilogue-used; sets to them will
9624 be preserved even if it does not appear that the register is used again
9625 later in the function.
9627 Another note: according to the MIPS spec, coprocessor 1 (if present) is
9628 the FPU@. One accesses COP1 registers through standard mips
9629 floating-point support; they are not included in this mechanism.
9631 There is one macro used in defining the MIPS coprocessor interface which
9632 you may want to override in subtargets; it is described below.
9634 @defmac ALL_COP_ADDITIONAL_REGISTER_NAMES
9635 A comma-separated list (with leading comma) of pairs describing the
9636 alternate names of coprocessor registers. The format of each entry should be
9638 @{ @var{alternatename}, @var{register_number}@}
9644 @section Parameters for Precompiled Header Validity Checking
9645 @cindex parameters, precompiled headers
9647 @deftypefn {Target Hook} void *TARGET_GET_PCH_VALIDITY (size_t *@var{sz})
9648 This hook returns the data needed by @code{TARGET_PCH_VALID_P} and sets
9649 @samp{*@var{sz}} to the size of the data in bytes.
9652 @deftypefn {Target Hook} const char *TARGET_PCH_VALID_P (const void *@var{data}, size_t @var{sz})
9653 This hook checks whether the options used to create a PCH file are
9654 compatible with the current settings. It returns @code{NULL}
9655 if so and a suitable error message if not. Error messages will
9656 be presented to the user and must be localized using @samp{_(@var{msg})}.
9658 @var{data} is the data that was returned by @code{TARGET_GET_PCH_VALIDITY}
9659 when the PCH file was created and @var{sz} is the size of that data in bytes.
9660 It's safe to assume that the data was created by the same version of the
9661 compiler, so no format checking is needed.
9663 The default definition of @code{default_pch_valid_p} should be
9664 suitable for most targets.
9667 @deftypefn {Target Hook} const char *TARGET_CHECK_PCH_TARGET_FLAGS (int @var{pch_flags})
9668 If this hook is nonnull, the default implementation of
9669 @code{TARGET_PCH_VALID_P} will use it to check for compatible values
9670 of @code{target_flags}. @var{pch_flags} specifies the value that
9671 @code{target_flags} had when the PCH file was created. The return
9672 value is the same as for @code{TARGET_PCH_VALID_P}.
9676 @section C++ ABI parameters
9677 @cindex parameters, c++ abi
9679 @deftypefn {Target Hook} tree TARGET_CXX_GUARD_TYPE (void)
9680 Define this hook to override the integer type used for guard variables.
9681 These are used to implement one-time construction of static objects. The
9682 default is long_long_integer_type_node.
9685 @deftypefn {Target Hook} bool TARGET_CXX_GUARD_MASK_BIT (void)
9686 This hook determines how guard variables are used. It should return
9687 @code{false} (the default) if first byte should be used. A return value of
9688 @code{true} indicates the least significant bit should be used.
9691 @deftypefn {Target Hook} tree TARGET_CXX_GET_COOKIE_SIZE (tree @var{type})
9692 This hook returns the size of the cookie to use when allocating an array
9693 whose elements have the indicated @var{type}. Assumes that it is already
9694 known that a cookie is needed. The default is
9695 @code{max(sizeof (size_t), alignof(type))}, as defined in section 2.7 of the
9696 IA64/Generic C++ ABI@.
9699 @deftypefn {Target Hook} bool TARGET_CXX_COOKIE_HAS_SIZE (void)
9700 This hook should return @code{true} if the element size should be stored in
9701 array cookies. The default is to return @code{false}.
9704 @deftypefn {Target Hook} int TARGET_CXX_IMPORT_EXPORT_CLASS (tree @var{type}, int @var{import_export})
9705 If defined by a backend this hook allows the decision made to export
9706 class @var{type} to be overruled. Upon entry @var{import_export}
9707 will contain 1 if the class is going to be exported, @minus{}1 if it is going
9708 to be imported and 0 otherwise. This function should return the
9709 modified value and perform any other actions necessary to support the
9710 backend's targeted operating system.
9713 @deftypefn {Target Hook} bool TARGET_CXX_CDTOR_RETURNS_THIS (void)
9714 This hook should return @code{true} if constructors and destructors return
9715 the address of the object created/destroyed. The default is to return
9719 @deftypefn {Target Hook} bool TARGET_CXX_KEY_METHOD_MAY_BE_INLINE (void)
9720 This hook returns true if the key method for a class (i.e., the method
9721 which, if defined in the current translation unit, causes the virtual
9722 table to be emitted) may be an inline function. Under the standard
9723 Itanium C++ ABI the key method may be an inline function so long as
9724 the function is not declared inline in the class definition. Under
9725 some variants of the ABI, an inline function can never be the key
9726 method. The default is to return @code{true}.
9729 @deftypefn {Target Hook} void TARGET_CXX_DETERMINE_CLASS_DATA_VISIBILITY (tree @var{decl})
9730 @var{decl} is a virtual table, virtual table table, typeinfo object,
9731 or other similar implicit class data object that will be emitted with
9732 external linkage in this translation unit. No ELF visibility has been
9733 explicitly specified. If the target needs to specify a visibility
9734 other than that of the containing class, use this hook to set
9735 @code{DECL_VISIBILITY} and @code{DECL_VISIBILITY_SPECIFIED}.
9738 @deftypefn {Target Hook} bool TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT (void)
9739 This hook returns true (the default) if virtual tables and other
9740 similar implicit class data objects are always COMDAT if they have
9741 external linkage. If this hook returns false, then class data for
9742 classes whose virtual table will be emitted in only one translation
9743 unit will not be COMDAT.
9746 @deftypefn {Target Hook} bool TARGET_CXX_LIBRARY_RTTI_COMDAT (void)
9747 This hook returns true (the default) if the RTTI information for
9748 the basic types which is defined in the C++ runtime should always
9749 be COMDAT, false if it should not be COMDAT.
9752 @deftypefn {Target Hook} bool TARGET_CXX_USE_AEABI_ATEXIT (void)
9753 This hook returns true if @code{__aeabi_atexit} (as defined by the ARM EABI)
9754 should be used to register static destructors when @option{-fuse-cxa-atexit}
9755 is in effect. The default is to return false to use @code{__cxa_atexit}.
9758 @deftypefn {Target Hook} bool TARGET_CXX_USE_ATEXIT_FOR_CXA_ATEXIT (void)
9759 This hook returns true if the target @code{atexit} function can be used
9760 in the same manner as @code{__cxa_atexit} to register C++ static
9761 destructors. This requires that @code{atexit}-registered functions in
9762 shared libraries are run in the correct order when the libraries are
9763 unloaded. The default is to return false.
9766 @deftypefn {Target Hook} void TARGET_CXX_ADJUST_CLASS_AT_DEFINITION (tree @var{type})
9767 @var{type} is a C++ class (i.e., RECORD_TYPE or UNION_TYPE) that has just been
9768 defined. Use this hook to make adjustments to the class (eg, tweak
9769 visibility or perform any other required target modifications).
9773 @section Miscellaneous Parameters
9774 @cindex parameters, miscellaneous
9776 @c prevent bad page break with this line
9777 Here are several miscellaneous parameters.
9779 @defmac HAS_LONG_COND_BRANCH
9780 Define this boolean macro to indicate whether or not your architecture
9781 has conditional branches that can span all of memory. It is used in
9782 conjunction with an optimization that partitions hot and cold basic
9783 blocks into separate sections of the executable. If this macro is
9784 set to false, gcc will convert any conditional branches that attempt
9785 to cross between sections into unconditional branches or indirect jumps.
9788 @defmac HAS_LONG_UNCOND_BRANCH
9789 Define this boolean macro to indicate whether or not your architecture
9790 has unconditional branches that can span all of memory. It is used in
9791 conjunction with an optimization that partitions hot and cold basic
9792 blocks into separate sections of the executable. If this macro is
9793 set to false, gcc will convert any unconditional branches that attempt
9794 to cross between sections into indirect jumps.
9797 @defmac CASE_VECTOR_MODE
9798 An alias for a machine mode name. This is the machine mode that
9799 elements of a jump-table should have.
9802 @defmac CASE_VECTOR_SHORTEN_MODE (@var{min_offset}, @var{max_offset}, @var{body})
9803 Optional: return the preferred mode for an @code{addr_diff_vec}
9804 when the minimum and maximum offset are known. If you define this,
9805 it enables extra code in branch shortening to deal with @code{addr_diff_vec}.
9806 To make this work, you also have to define @code{INSN_ALIGN} and
9807 make the alignment for @code{addr_diff_vec} explicit.
9808 The @var{body} argument is provided so that the offset_unsigned and scale
9809 flags can be updated.
9812 @defmac CASE_VECTOR_PC_RELATIVE
9813 Define this macro to be a C expression to indicate when jump-tables
9814 should contain relative addresses. You need not define this macro if
9815 jump-tables never contain relative addresses, or jump-tables should
9816 contain relative addresses only when @option{-fPIC} or @option{-fPIC}
9820 @deftypefn {Target Hook} unsigned int TARGET_CASE_VALUES_THRESHOLD (void)
9821 This function return the smallest number of different values for which it
9822 is best to use a jump-table instead of a tree of conditional branches.
9823 The default is four for machines with a @code{casesi} instruction and
9824 five otherwise. This is best for most machines.
9827 @defmac CASE_USE_BIT_TESTS
9828 Define this macro to be a C expression to indicate whether C switch
9829 statements may be implemented by a sequence of bit tests. This is
9830 advantageous on processors that can efficiently implement left shift
9831 of 1 by the number of bits held in a register, but inappropriate on
9832 targets that would require a loop. By default, this macro returns
9833 @code{true} if the target defines an @code{ashlsi3} pattern, and
9834 @code{false} otherwise.
9837 @defmac WORD_REGISTER_OPERATIONS
9838 Define this macro if operations between registers with integral mode
9839 smaller than a word are always performed on the entire register.
9840 Most RISC machines have this property and most CISC machines do not.
9843 @defmac LOAD_EXTEND_OP (@var{mem_mode})
9844 Define this macro to be a C expression indicating when insns that read
9845 memory in @var{mem_mode}, an integral mode narrower than a word, set the
9846 bits outside of @var{mem_mode} to be either the sign-extension or the
9847 zero-extension of the data read. Return @code{SIGN_EXTEND} for values
9848 of @var{mem_mode} for which the
9849 insn sign-extends, @code{ZERO_EXTEND} for which it zero-extends, and
9850 @code{UNKNOWN} for other modes.
9852 This macro is not called with @var{mem_mode} non-integral or with a width
9853 greater than or equal to @code{BITS_PER_WORD}, so you may return any
9854 value in this case. Do not define this macro if it would always return
9855 @code{UNKNOWN}. On machines where this macro is defined, you will normally
9856 define it as the constant @code{SIGN_EXTEND} or @code{ZERO_EXTEND}.
9858 You may return a non-@code{UNKNOWN} value even if for some hard registers
9859 the sign extension is not performed, if for the @code{REGNO_REG_CLASS}
9860 of these hard registers @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero
9861 when the @var{from} mode is @var{mem_mode} and the @var{to} mode is any
9862 integral mode larger than this but not larger than @code{word_mode}.
9864 You must return @code{UNKNOWN} if for some hard registers that allow this
9865 mode, @code{CANNOT_CHANGE_MODE_CLASS} says that they cannot change to
9866 @code{word_mode}, but that they can change to another integral mode that
9867 is larger then @var{mem_mode} but still smaller than @code{word_mode}.
9870 @defmac SHORT_IMMEDIATES_SIGN_EXTEND
9871 Define this macro if loading short immediate values into registers sign
9875 @defmac FIXUNS_TRUNC_LIKE_FIX_TRUNC
9876 Define this macro if the same instructions that convert a floating
9877 point number to a signed fixed point number also convert validly to an
9881 @deftypefn {Target Hook} int TARGET_MIN_DIVISIONS_FOR_RECIP_MUL (enum machine_mode @var{mode})
9882 When @option{-ffast-math} is in effect, GCC tries to optimize
9883 divisions by the same divisor, by turning them into multiplications by
9884 the reciprocal. This target hook specifies the minimum number of divisions
9885 that should be there for GCC to perform the optimization for a variable
9886 of mode @var{mode}. The default implementation returns 3 if the machine
9887 has an instruction for the division, and 2 if it does not.
9891 The maximum number of bytes that a single instruction can move quickly
9892 between memory and registers or between two memory locations.
9895 @defmac MAX_MOVE_MAX
9896 The maximum number of bytes that a single instruction can move quickly
9897 between memory and registers or between two memory locations. If this
9898 is undefined, the default is @code{MOVE_MAX}. Otherwise, it is the
9899 constant value that is the largest value that @code{MOVE_MAX} can have
9903 @defmac SHIFT_COUNT_TRUNCATED
9904 A C expression that is nonzero if on this machine the number of bits
9905 actually used for the count of a shift operation is equal to the number
9906 of bits needed to represent the size of the object being shifted. When
9907 this macro is nonzero, the compiler will assume that it is safe to omit
9908 a sign-extend, zero-extend, and certain bitwise `and' instructions that
9909 truncates the count of a shift operation. On machines that have
9910 instructions that act on bit-fields at variable positions, which may
9911 include `bit test' instructions, a nonzero @code{SHIFT_COUNT_TRUNCATED}
9912 also enables deletion of truncations of the values that serve as
9913 arguments to bit-field instructions.
9915 If both types of instructions truncate the count (for shifts) and
9916 position (for bit-field operations), or if no variable-position bit-field
9917 instructions exist, you should define this macro.
9919 However, on some machines, such as the 80386 and the 680x0, truncation
9920 only applies to shift operations and not the (real or pretended)
9921 bit-field operations. Define @code{SHIFT_COUNT_TRUNCATED} to be zero on
9922 such machines. Instead, add patterns to the @file{md} file that include
9923 the implied truncation of the shift instructions.
9925 You need not define this macro if it would always have the value of zero.
9928 @anchor{TARGET_SHIFT_TRUNCATION_MASK}
9929 @deftypefn {Target Hook} int TARGET_SHIFT_TRUNCATION_MASK (enum machine_mode @var{mode})
9930 This function describes how the standard shift patterns for @var{mode}
9931 deal with shifts by negative amounts or by more than the width of the mode.
9932 @xref{shift patterns}.
9934 On many machines, the shift patterns will apply a mask @var{m} to the
9935 shift count, meaning that a fixed-width shift of @var{x} by @var{y} is
9936 equivalent to an arbitrary-width shift of @var{x} by @var{y & m}. If
9937 this is true for mode @var{mode}, the function should return @var{m},
9938 otherwise it should return 0. A return value of 0 indicates that no
9939 particular behavior is guaranteed.
9941 Note that, unlike @code{SHIFT_COUNT_TRUNCATED}, this function does
9942 @emph{not} apply to general shift rtxes; it applies only to instructions
9943 that are generated by the named shift patterns.
9945 The default implementation of this function returns
9946 @code{GET_MODE_BITSIZE (@var{mode}) - 1} if @code{SHIFT_COUNT_TRUNCATED}
9947 and 0 otherwise. This definition is always safe, but if
9948 @code{SHIFT_COUNT_TRUNCATED} is false, and some shift patterns
9949 nevertheless truncate the shift count, you may get better code
9953 @defmac TRULY_NOOP_TRUNCATION (@var{outprec}, @var{inprec})
9954 A C expression which is nonzero if on this machine it is safe to
9955 ``convert'' an integer of @var{inprec} bits to one of @var{outprec}
9956 bits (where @var{outprec} is smaller than @var{inprec}) by merely
9957 operating on it as if it had only @var{outprec} bits.
9959 On many machines, this expression can be 1.
9961 @c rearranged this, removed the phrase "it is reported that". this was
9962 @c to fix an overfull hbox. --mew 10feb93
9963 When @code{TRULY_NOOP_TRUNCATION} returns 1 for a pair of sizes for
9964 modes for which @code{MODES_TIEABLE_P} is 0, suboptimal code can result.
9965 If this is the case, making @code{TRULY_NOOP_TRUNCATION} return 0 in
9966 such cases may improve things.
9969 @deftypefn {Target Hook} int TARGET_MODE_REP_EXTENDED (enum machine_mode @var{mode}, enum machine_mode @var{rep_mode})
9970 The representation of an integral mode can be such that the values
9971 are always extended to a wider integral mode. Return
9972 @code{SIGN_EXTEND} if values of @var{mode} are represented in
9973 sign-extended form to @var{rep_mode}. Return @code{UNKNOWN}
9974 otherwise. (Currently, none of the targets use zero-extended
9975 representation this way so unlike @code{LOAD_EXTEND_OP},
9976 @code{TARGET_MODE_REP_EXTENDED} is expected to return either
9977 @code{SIGN_EXTEND} or @code{UNKNOWN}. Also no target extends
9978 @var{mode} to @var{mode_rep} so that @var{mode_rep} is not the next
9979 widest integral mode and currently we take advantage of this fact.)
9981 Similarly to @code{LOAD_EXTEND_OP} you may return a non-@code{UNKNOWN}
9982 value even if the extension is not performed on certain hard registers
9983 as long as for the @code{REGNO_REG_CLASS} of these hard registers
9984 @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero.
9986 Note that @code{TARGET_MODE_REP_EXTENDED} and @code{LOAD_EXTEND_OP}
9987 describe two related properties. If you define
9988 @code{TARGET_MODE_REP_EXTENDED (mode, word_mode)} you probably also want
9989 to define @code{LOAD_EXTEND_OP (mode)} to return the same type of
9992 In order to enforce the representation of @code{mode},
9993 @code{TRULY_NOOP_TRUNCATION} should return false when truncating to
9997 @defmac STORE_FLAG_VALUE
9998 A C expression describing the value returned by a comparison operator
9999 with an integral mode and stored by a store-flag instruction
10000 (@samp{s@var{cond}}) when the condition is true. This description must
10001 apply to @emph{all} the @samp{s@var{cond}} patterns and all the
10002 comparison operators whose results have a @code{MODE_INT} mode.
10004 A value of 1 or @minus{}1 means that the instruction implementing the
10005 comparison operator returns exactly 1 or @minus{}1 when the comparison is true
10006 and 0 when the comparison is false. Otherwise, the value indicates
10007 which bits of the result are guaranteed to be 1 when the comparison is
10008 true. This value is interpreted in the mode of the comparison
10009 operation, which is given by the mode of the first operand in the
10010 @samp{s@var{cond}} pattern. Either the low bit or the sign bit of
10011 @code{STORE_FLAG_VALUE} be on. Presently, only those bits are used by
10014 If @code{STORE_FLAG_VALUE} is neither 1 or @minus{}1, the compiler will
10015 generate code that depends only on the specified bits. It can also
10016 replace comparison operators with equivalent operations if they cause
10017 the required bits to be set, even if the remaining bits are undefined.
10018 For example, on a machine whose comparison operators return an
10019 @code{SImode} value and where @code{STORE_FLAG_VALUE} is defined as
10020 @samp{0x80000000}, saying that just the sign bit is relevant, the
10024 (ne:SI (and:SI @var{x} (const_int @var{power-of-2})) (const_int 0))
10028 can be converted to
10031 (ashift:SI @var{x} (const_int @var{n}))
10035 where @var{n} is the appropriate shift count to move the bit being
10036 tested into the sign bit.
10038 There is no way to describe a machine that always sets the low-order bit
10039 for a true value, but does not guarantee the value of any other bits,
10040 but we do not know of any machine that has such an instruction. If you
10041 are trying to port GCC to such a machine, include an instruction to
10042 perform a logical-and of the result with 1 in the pattern for the
10043 comparison operators and let us know at @email{gcc@@gcc.gnu.org}.
10045 Often, a machine will have multiple instructions that obtain a value
10046 from a comparison (or the condition codes). Here are rules to guide the
10047 choice of value for @code{STORE_FLAG_VALUE}, and hence the instructions
10052 Use the shortest sequence that yields a valid definition for
10053 @code{STORE_FLAG_VALUE}. It is more efficient for the compiler to
10054 ``normalize'' the value (convert it to, e.g., 1 or 0) than for the
10055 comparison operators to do so because there may be opportunities to
10056 combine the normalization with other operations.
10059 For equal-length sequences, use a value of 1 or @minus{}1, with @minus{}1 being
10060 slightly preferred on machines with expensive jumps and 1 preferred on
10064 As a second choice, choose a value of @samp{0x80000001} if instructions
10065 exist that set both the sign and low-order bits but do not define the
10069 Otherwise, use a value of @samp{0x80000000}.
10072 Many machines can produce both the value chosen for
10073 @code{STORE_FLAG_VALUE} and its negation in the same number of
10074 instructions. On those machines, you should also define a pattern for
10075 those cases, e.g., one matching
10078 (set @var{A} (neg:@var{m} (ne:@var{m} @var{B} @var{C})))
10081 Some machines can also perform @code{and} or @code{plus} operations on
10082 condition code values with less instructions than the corresponding
10083 @samp{s@var{cond}} insn followed by @code{and} or @code{plus}. On those
10084 machines, define the appropriate patterns. Use the names @code{incscc}
10085 and @code{decscc}, respectively, for the patterns which perform
10086 @code{plus} or @code{minus} operations on condition code values. See
10087 @file{rs6000.md} for some examples. The GNU Superoptizer can be used to
10088 find such instruction sequences on other machines.
10090 If this macro is not defined, the default value, 1, is used. You need
10091 not define @code{STORE_FLAG_VALUE} if the machine has no store-flag
10092 instructions, or if the value generated by these instructions is 1.
10095 @defmac FLOAT_STORE_FLAG_VALUE (@var{mode})
10096 A C expression that gives a nonzero @code{REAL_VALUE_TYPE} value that is
10097 returned when comparison operators with floating-point results are true.
10098 Define this macro on machines that have comparison operations that return
10099 floating-point values. If there are no such operations, do not define
10103 @defmac VECTOR_STORE_FLAG_VALUE (@var{mode})
10104 A C expression that gives a rtx representing the nonzero true element
10105 for vector comparisons. The returned rtx should be valid for the inner
10106 mode of @var{mode} which is guaranteed to be a vector mode. Define
10107 this macro on machines that have vector comparison operations that
10108 return a vector result. If there are no such operations, do not define
10109 this macro. Typically, this macro is defined as @code{const1_rtx} or
10110 @code{constm1_rtx}. This macro may return @code{NULL_RTX} to prevent
10111 the compiler optimizing such vector comparison operations for the
10115 @defmac CLZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
10116 @defmacx CTZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
10117 A C expression that indicates whether the architecture defines a value
10118 for @code{clz} or @code{ctz} with a zero operand.
10119 A result of @code{0} indicates the value is undefined.
10120 If the value is defined for only the RTL expression, the macro should
10121 evaluate to @code{1}; if the value applies also to the corresponding optab
10122 entry (which is normally the case if it expands directly into
10123 the corresponding RTL), then the macro should evaluate to @code{2}.
10124 In the cases where the value is defined, @var{value} should be set to
10127 If this macro is not defined, the value of @code{clz} or
10128 @code{ctz} at zero is assumed to be undefined.
10130 This macro must be defined if the target's expansion for @code{ffs}
10131 relies on a particular value to get correct results. Otherwise it
10132 is not necessary, though it may be used to optimize some corner cases, and
10133 to provide a default expansion for the @code{ffs} optab.
10135 Note that regardless of this macro the ``definedness'' of @code{clz}
10136 and @code{ctz} at zero do @emph{not} extend to the builtin functions
10137 visible to the user. Thus one may be free to adjust the value at will
10138 to match the target expansion of these operations without fear of
10143 An alias for the machine mode for pointers. On most machines, define
10144 this to be the integer mode corresponding to the width of a hardware
10145 pointer; @code{SImode} on 32-bit machine or @code{DImode} on 64-bit machines.
10146 On some machines you must define this to be one of the partial integer
10147 modes, such as @code{PSImode}.
10149 The width of @code{Pmode} must be at least as large as the value of
10150 @code{POINTER_SIZE}. If it is not equal, you must define the macro
10151 @code{POINTERS_EXTEND_UNSIGNED} to specify how pointers are extended
10155 @defmac FUNCTION_MODE
10156 An alias for the machine mode used for memory references to functions
10157 being called, in @code{call} RTL expressions. On most CISC machines,
10158 where an instruction can begin at any byte address, this should be
10159 @code{QImode}. On most RISC machines, where all instructions have fixed
10160 size and alignment, this should be a mode with the same size and alignment
10161 as the machine instruction words - typically @code{SImode} or @code{HImode}.
10164 @defmac STDC_0_IN_SYSTEM_HEADERS
10165 In normal operation, the preprocessor expands @code{__STDC__} to the
10166 constant 1, to signify that GCC conforms to ISO Standard C@. On some
10167 hosts, like Solaris, the system compiler uses a different convention,
10168 where @code{__STDC__} is normally 0, but is 1 if the user specifies
10169 strict conformance to the C Standard.
10171 Defining @code{STDC_0_IN_SYSTEM_HEADERS} makes GNU CPP follows the host
10172 convention when processing system header files, but when processing user
10173 files @code{__STDC__} will always expand to 1.
10176 @defmac NO_IMPLICIT_EXTERN_C
10177 Define this macro if the system header files support C++ as well as C@.
10178 This macro inhibits the usual method of using system header files in
10179 C++, which is to pretend that the file's contents are enclosed in
10180 @samp{extern "C" @{@dots{}@}}.
10185 @defmac REGISTER_TARGET_PRAGMAS ()
10186 Define this macro if you want to implement any target-specific pragmas.
10187 If defined, it is a C expression which makes a series of calls to
10188 @code{c_register_pragma} or @code{c_register_pragma_with_expansion}
10189 for each pragma. The macro may also do any
10190 setup required for the pragmas.
10192 The primary reason to define this macro is to provide compatibility with
10193 other compilers for the same target. In general, we discourage
10194 definition of target-specific pragmas for GCC@.
10196 If the pragma can be implemented by attributes then you should consider
10197 defining the target hook @samp{TARGET_INSERT_ATTRIBUTES} as well.
10199 Preprocessor macros that appear on pragma lines are not expanded. All
10200 @samp{#pragma} directives that do not match any registered pragma are
10201 silently ignored, unless the user specifies @option{-Wunknown-pragmas}.
10204 @deftypefun void c_register_pragma (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
10205 @deftypefunx void c_register_pragma_with_expansion (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
10207 Each call to @code{c_register_pragma} or
10208 @code{c_register_pragma_with_expansion} establishes one pragma. The
10209 @var{callback} routine will be called when the preprocessor encounters a
10213 #pragma [@var{space}] @var{name} @dots{}
10216 @var{space} is the case-sensitive namespace of the pragma, or
10217 @code{NULL} to put the pragma in the global namespace. The callback
10218 routine receives @var{pfile} as its first argument, which can be passed
10219 on to cpplib's functions if necessary. You can lex tokens after the
10220 @var{name} by calling @code{pragma_lex}. Tokens that are not read by the
10221 callback will be silently ignored. The end of the line is indicated by
10222 a token of type @code{CPP_EOF}. Macro expansion occurs on the
10223 arguments of pragmas registered with
10224 @code{c_register_pragma_with_expansion} but not on the arguments of
10225 pragmas registered with @code{c_register_pragma}.
10227 Note that the use of @code{pragma_lex} is specific to the C and C++
10228 compilers. It will not work in the Java or Fortran compilers, or any
10229 other language compilers for that matter. Thus if @code{pragma_lex} is going
10230 to be called from target-specific code, it must only be done so when
10231 building the C and C++ compilers. This can be done by defining the
10232 variables @code{c_target_objs} and @code{cxx_target_objs} in the
10233 target entry in the @file{config.gcc} file. These variables should name
10234 the target-specific, language-specific object file which contains the
10235 code that uses @code{pragma_lex}. Note it will also be necessary to add a
10236 rule to the makefile fragment pointed to by @code{tmake_file} that shows
10237 how to build this object file.
10242 @defmac HANDLE_SYSV_PRAGMA
10243 Define this macro (to a value of 1) if you want the System V style
10244 pragmas @samp{#pragma pack(<n>)} and @samp{#pragma weak <name>
10245 [=<value>]} to be supported by gcc.
10247 The pack pragma specifies the maximum alignment (in bytes) of fields
10248 within a structure, in much the same way as the @samp{__aligned__} and
10249 @samp{__packed__} @code{__attribute__}s do. A pack value of zero resets
10250 the behavior to the default.
10252 A subtlety for Microsoft Visual C/C++ style bit-field packing
10253 (e.g.@: -mms-bitfields) for targets that support it:
10254 When a bit-field is inserted into a packed record, the whole size
10255 of the underlying type is used by one or more same-size adjacent
10256 bit-fields (that is, if its long:3, 32 bits is used in the record,
10257 and any additional adjacent long bit-fields are packed into the same
10258 chunk of 32 bits. However, if the size changes, a new field of that
10259 size is allocated).
10261 If both MS bit-fields and @samp{__attribute__((packed))} are used,
10262 the latter will take precedence. If @samp{__attribute__((packed))} is
10263 used on a single field when MS bit-fields are in use, it will take
10264 precedence for that field, but the alignment of the rest of the structure
10265 may affect its placement.
10267 The weak pragma only works if @code{SUPPORTS_WEAK} and
10268 @code{ASM_WEAKEN_LABEL} are defined. If enabled it allows the creation
10269 of specifically named weak labels, optionally with a value.
10274 @defmac HANDLE_PRAGMA_PACK_PUSH_POP
10275 Define this macro (to a value of 1) if you want to support the Win32
10276 style pragmas @samp{#pragma pack(push[,@var{n}])} and @samp{#pragma
10277 pack(pop)}. The @samp{pack(push,[@var{n}])} pragma specifies the maximum
10278 alignment (in bytes) of fields within a structure, in much the same way as
10279 the @samp{__aligned__} and @samp{__packed__} @code{__attribute__}s do. A
10280 pack value of zero resets the behavior to the default. Successive
10281 invocations of this pragma cause the previous values to be stacked, so
10282 that invocations of @samp{#pragma pack(pop)} will return to the previous
10286 @defmac HANDLE_PRAGMA_PACK_WITH_EXPANSION
10287 Define this macro, as well as
10288 @code{HANDLE_SYSV_PRAGMA}, if macros should be expanded in the
10289 arguments of @samp{#pragma pack}.
10292 @defmac TARGET_DEFAULT_PACK_STRUCT
10293 If your target requires a structure packing default other than 0 (meaning
10294 the machine default), define this macro to the necessary value (in bytes).
10295 This must be a value that would also be valid to use with
10296 @samp{#pragma pack()} (that is, a small power of two).
10301 @defmac HANDLE_PRAGMA_PUSH_POP_MACRO
10302 Define this macro if you want to support the Win32 style pragmas
10303 @samp{#pragma push_macro(macro-name-as-string)} and @samp{#pragma
10304 pop_macro(macro-name-as-string)}. The @samp{#pragma push_macro(
10305 macro-name-as-string)} pragma saves the named macro and via
10306 @samp{#pragma pop_macro(macro-name-as-string)} it will return to the
10311 @defmac DOLLARS_IN_IDENTIFIERS
10312 Define this macro to control use of the character @samp{$} in
10313 identifier names for the C family of languages. 0 means @samp{$} is
10314 not allowed by default; 1 means it is allowed. 1 is the default;
10315 there is no need to define this macro in that case.
10318 @defmac NO_DOLLAR_IN_LABEL
10319 Define this macro if the assembler does not accept the character
10320 @samp{$} in label names. By default constructors and destructors in
10321 G++ have @samp{$} in the identifiers. If this macro is defined,
10322 @samp{.} is used instead.
10325 @defmac NO_DOT_IN_LABEL
10326 Define this macro if the assembler does not accept the character
10327 @samp{.} in label names. By default constructors and destructors in G++
10328 have names that use @samp{.}. If this macro is defined, these names
10329 are rewritten to avoid @samp{.}.
10332 @defmac INSN_SETS_ARE_DELAYED (@var{insn})
10333 Define this macro as a C expression that is nonzero if it is safe for the
10334 delay slot scheduler to place instructions in the delay slot of @var{insn},
10335 even if they appear to use a resource set or clobbered in @var{insn}.
10336 @var{insn} is always a @code{jump_insn} or an @code{insn}; GCC knows that
10337 every @code{call_insn} has this behavior. On machines where some @code{insn}
10338 or @code{jump_insn} is really a function call and hence has this behavior,
10339 you should define this macro.
10341 You need not define this macro if it would always return zero.
10344 @defmac INSN_REFERENCES_ARE_DELAYED (@var{insn})
10345 Define this macro as a C expression that is nonzero if it is safe for the
10346 delay slot scheduler to place instructions in the delay slot of @var{insn},
10347 even if they appear to set or clobber a resource referenced in @var{insn}.
10348 @var{insn} is always a @code{jump_insn} or an @code{insn}. On machines where
10349 some @code{insn} or @code{jump_insn} is really a function call and its operands
10350 are registers whose use is actually in the subroutine it calls, you should
10351 define this macro. Doing so allows the delay slot scheduler to move
10352 instructions which copy arguments into the argument registers into the delay
10353 slot of @var{insn}.
10355 You need not define this macro if it would always return zero.
10358 @defmac MULTIPLE_SYMBOL_SPACES
10359 Define this macro as a C expression that is nonzero if, in some cases,
10360 global symbols from one translation unit may not be bound to undefined
10361 symbols in another translation unit without user intervention. For
10362 instance, under Microsoft Windows symbols must be explicitly imported
10363 from shared libraries (DLLs).
10365 You need not define this macro if it would always evaluate to zero.
10368 @deftypefn {Target Hook} tree TARGET_MD_ASM_CLOBBERS (tree @var{outputs}, tree @var{inputs}, tree @var{clobbers})
10369 This target hook should add to @var{clobbers} @code{STRING_CST} trees for
10370 any hard regs the port wishes to automatically clobber for an asm.
10371 It should return the result of the last @code{tree_cons} used to add a
10372 clobber. The @var{outputs}, @var{inputs} and @var{clobber} lists are the
10373 corresponding parameters to the asm and may be inspected to avoid
10374 clobbering a register that is an input or output of the asm. You can use
10375 @code{tree_overlaps_hard_reg_set}, declared in @file{tree.h}, to test
10376 for overlap with regards to asm-declared registers.
10379 @defmac MATH_LIBRARY
10380 Define this macro as a C string constant for the linker argument to link
10381 in the system math library, or @samp{""} if the target does not have a
10382 separate math library.
10384 You need only define this macro if the default of @samp{"-lm"} is wrong.
10387 @defmac LIBRARY_PATH_ENV
10388 Define this macro as a C string constant for the environment variable that
10389 specifies where the linker should look for libraries.
10391 You need only define this macro if the default of @samp{"LIBRARY_PATH"}
10395 @defmac TARGET_POSIX_IO
10396 Define this macro if the target supports the following POSIX@ file
10397 functions, access, mkdir and file locking with fcntl / F_SETLKW@.
10398 Defining @code{TARGET_POSIX_IO} will enable the test coverage code
10399 to use file locking when exiting a program, which avoids race conditions
10400 if the program has forked. It will also create directories at run-time
10401 for cross-profiling.
10404 @defmac MAX_CONDITIONAL_EXECUTE
10406 A C expression for the maximum number of instructions to execute via
10407 conditional execution instructions instead of a branch. A value of
10408 @code{BRANCH_COST}+1 is the default if the machine does not use cc0, and
10409 1 if it does use cc0.
10412 @defmac IFCVT_MODIFY_TESTS (@var{ce_info}, @var{true_expr}, @var{false_expr})
10413 Used if the target needs to perform machine-dependent modifications on the
10414 conditionals used for turning basic blocks into conditionally executed code.
10415 @var{ce_info} points to a data structure, @code{struct ce_if_block}, which
10416 contains information about the currently processed blocks. @var{true_expr}
10417 and @var{false_expr} are the tests that are used for converting the
10418 then-block and the else-block, respectively. Set either @var{true_expr} or
10419 @var{false_expr} to a null pointer if the tests cannot be converted.
10422 @defmac IFCVT_MODIFY_MULTIPLE_TESTS (@var{ce_info}, @var{bb}, @var{true_expr}, @var{false_expr})
10423 Like @code{IFCVT_MODIFY_TESTS}, but used when converting more complicated
10424 if-statements into conditions combined by @code{and} and @code{or} operations.
10425 @var{bb} contains the basic block that contains the test that is currently
10426 being processed and about to be turned into a condition.
10429 @defmac IFCVT_MODIFY_INSN (@var{ce_info}, @var{pattern}, @var{insn})
10430 A C expression to modify the @var{PATTERN} of an @var{INSN} that is to
10431 be converted to conditional execution format. @var{ce_info} points to
10432 a data structure, @code{struct ce_if_block}, which contains information
10433 about the currently processed blocks.
10436 @defmac IFCVT_MODIFY_FINAL (@var{ce_info})
10437 A C expression to perform any final machine dependent modifications in
10438 converting code to conditional execution. The involved basic blocks
10439 can be found in the @code{struct ce_if_block} structure that is pointed
10440 to by @var{ce_info}.
10443 @defmac IFCVT_MODIFY_CANCEL (@var{ce_info})
10444 A C expression to cancel any machine dependent modifications in
10445 converting code to conditional execution. The involved basic blocks
10446 can be found in the @code{struct ce_if_block} structure that is pointed
10447 to by @var{ce_info}.
10450 @defmac IFCVT_INIT_EXTRA_FIELDS (@var{ce_info})
10451 A C expression to initialize any extra fields in a @code{struct ce_if_block}
10452 structure, which are defined by the @code{IFCVT_EXTRA_FIELDS} macro.
10455 @defmac IFCVT_EXTRA_FIELDS
10456 If defined, it should expand to a set of field declarations that will be
10457 added to the @code{struct ce_if_block} structure. These should be initialized
10458 by the @code{IFCVT_INIT_EXTRA_FIELDS} macro.
10461 @deftypefn {Target Hook} void TARGET_MACHINE_DEPENDENT_REORG ()
10462 If non-null, this hook performs a target-specific pass over the
10463 instruction stream. The compiler will run it at all optimization levels,
10464 just before the point at which it normally does delayed-branch scheduling.
10466 The exact purpose of the hook varies from target to target. Some use
10467 it to do transformations that are necessary for correctness, such as
10468 laying out in-function constant pools or avoiding hardware hazards.
10469 Others use it as an opportunity to do some machine-dependent optimizations.
10471 You need not implement the hook if it has nothing to do. The default
10472 definition is null.
10475 @deftypefn {Target Hook} void TARGET_INIT_BUILTINS ()
10476 Define this hook if you have any machine-specific built-in functions
10477 that need to be defined. It should be a function that performs the
10480 Machine specific built-in functions can be useful to expand special machine
10481 instructions that would otherwise not normally be generated because
10482 they have no equivalent in the source language (for example, SIMD vector
10483 instructions or prefetch instructions).
10485 To create a built-in function, call the function
10486 @code{lang_hooks.builtin_function}
10487 which is defined by the language front end. You can use any type nodes set
10488 up by @code{build_common_tree_nodes} and @code{build_common_tree_nodes_2};
10489 only language front ends that use those two functions will call
10490 @samp{TARGET_INIT_BUILTINS}.
10493 @deftypefn {Target Hook} rtx TARGET_EXPAND_BUILTIN (tree @var{exp}, rtx @var{target}, rtx @var{subtarget}, enum machine_mode @var{mode}, int @var{ignore})
10495 Expand a call to a machine specific built-in function that was set up by
10496 @samp{TARGET_INIT_BUILTINS}. @var{exp} is the expression for the
10497 function call; the result should go to @var{target} if that is
10498 convenient, and have mode @var{mode} if that is convenient.
10499 @var{subtarget} may be used as the target for computing one of
10500 @var{exp}'s operands. @var{ignore} is nonzero if the value is to be
10501 ignored. This function should return the result of the call to the
10505 @deftypefn {Target Hook} tree TARGET_RESOLVE_OVERLOADED_BUILTIN (tree @var{fndecl}, tree @var{arglist})
10507 Select a replacement for a machine specific built-in function that
10508 was set up by @samp{TARGET_INIT_BUILTINS}. This is done
10509 @emph{before} regular type checking, and so allows the target to
10510 implement a crude form of function overloading. @var{fndecl} is the
10511 declaration of the built-in function. @var{arglist} is the list of
10512 arguments passed to the built-in function. The result is a
10513 complete expression that implements the operation, usually
10514 another @code{CALL_EXPR}.
10517 @deftypefn {Target Hook} tree TARGET_FOLD_BUILTIN (tree @var{fndecl}, tree @var{arglist}, bool @var{ignore})
10519 Fold a call to a machine specific built-in function that was set up by
10520 @samp{TARGET_INIT_BUILTINS}. @var{fndecl} is the declaration of the
10521 built-in function. @var{arglist} is the list of arguments passed to
10522 the built-in function. The result is another tree containing a
10523 simplified expression for the call's result. If @var{ignore} is true
10524 the value will be ignored.
10527 @deftypefn {Target Hook} const char * TARGET_INVALID_WITHIN_DOLOOP (rtx @var{insn})
10529 Take an instruction in @var{insn} and return NULL if it is valid within a
10530 low-overhead loop, otherwise return a string why doloop could not be applied.
10532 Many targets use special registers for low-overhead looping. For any
10533 instruction that clobbers these this function should return a string indicating
10534 the reason why the doloop could not be applied.
10535 By default, the RTL loop optimizer does not use a present doloop pattern for
10536 loops containing function calls or branch on table instructions.
10539 @defmac MD_CAN_REDIRECT_BRANCH (@var{branch1}, @var{branch2})
10541 Take a branch insn in @var{branch1} and another in @var{branch2}.
10542 Return true if redirecting @var{branch1} to the destination of
10543 @var{branch2} is possible.
10545 On some targets, branches may have a limited range. Optimizing the
10546 filling of delay slots can result in branches being redirected, and this
10547 may in turn cause a branch offset to overflow.
10550 @deftypefn {Target Hook} bool TARGET_COMMUTATIVE_P (rtx @var{x}, @var{outer_code})
10551 This target hook returns @code{true} if @var{x} is considered to be commutative.
10552 Usually, this is just COMMUTATIVE_P (@var{x}), but the HP PA doesn't consider
10553 PLUS to be commutative inside a MEM@. @var{outer_code} is the rtx code
10554 of the enclosing rtl, if known, otherwise it is UNKNOWN.
10557 @deftypefn {Target Hook} rtx TARGET_ALLOCATE_INITIAL_VALUE (rtx @var{hard_reg})
10559 When the initial value of a hard register has been copied in a pseudo
10560 register, it is often not necessary to actually allocate another register
10561 to this pseudo register, because the original hard register or a stack slot
10562 it has been saved into can be used. @code{TARGET_ALLOCATE_INITIAL_VALUE}
10563 is called at the start of register allocation once for each hard register
10564 that had its initial value copied by using
10565 @code{get_func_hard_reg_initial_val} or @code{get_hard_reg_initial_val}.
10566 Possible values are @code{NULL_RTX}, if you don't want
10567 to do any special allocation, a @code{REG} rtx---that would typically be
10568 the hard register itself, if it is known not to be clobbered---or a
10570 If you are returning a @code{MEM}, this is only a hint for the allocator;
10571 it might decide to use another register anyways.
10572 You may use @code{current_function_leaf_function} in the hook, functions
10573 that use @code{REG_N_SETS}, to determine if the hard
10574 register in question will not be clobbered.
10575 The default value of this hook is @code{NULL}, which disables any special
10579 @deftypefn {Target Hook} int TARGET_UNSPEC_MAY_TRAP_P (const_rtx @var{x}, unsigned @var{flags})
10580 This target hook returns nonzero if @var{x}, an @code{unspec} or
10581 @code{unspec_volatile} operation, might cause a trap. Targets can use
10582 this hook to enhance precision of analysis for @code{unspec} and
10583 @code{unspec_volatile} operations. You may call @code{may_trap_p_1}
10584 to analyze inner elements of @var{x} in which case @var{flags} should be
10588 @deftypefn {Target Hook} void TARGET_SET_CURRENT_FUNCTION (tree @var{decl})
10589 The compiler invokes this hook whenever it changes its current function
10590 context (@code{cfun}). You can define this function if
10591 the back end needs to perform any initialization or reset actions on a
10592 per-function basis. For example, it may be used to implement function
10593 attributes that affect register usage or code generation patterns.
10594 The argument @var{decl} is the declaration for the new function context,
10595 and may be null to indicate that the compiler has left a function context
10596 and is returning to processing at the top level.
10597 The default hook function does nothing.
10599 GCC sets @code{cfun} to a dummy function context during initialization of
10600 some parts of the back end. The hook function is not invoked in this
10601 situation; you need not worry about the hook being invoked recursively,
10602 or when the back end is in a partially-initialized state.
10605 @defmac TARGET_OBJECT_SUFFIX
10606 Define this macro to be a C string representing the suffix for object
10607 files on your target machine. If you do not define this macro, GCC will
10608 use @samp{.o} as the suffix for object files.
10611 @defmac TARGET_EXECUTABLE_SUFFIX
10612 Define this macro to be a C string representing the suffix to be
10613 automatically added to executable files on your target machine. If you
10614 do not define this macro, GCC will use the null string as the suffix for
10618 @defmac COLLECT_EXPORT_LIST
10619 If defined, @code{collect2} will scan the individual object files
10620 specified on its command line and create an export list for the linker.
10621 Define this macro for systems like AIX, where the linker discards
10622 object files that are not referenced from @code{main} and uses export
10626 @defmac MODIFY_JNI_METHOD_CALL (@var{mdecl})
10627 Define this macro to a C expression representing a variant of the
10628 method call @var{mdecl}, if Java Native Interface (JNI) methods
10629 must be invoked differently from other methods on your target.
10630 For example, on 32-bit Microsoft Windows, JNI methods must be invoked using
10631 the @code{stdcall} calling convention and this macro is then
10632 defined as this expression:
10635 build_type_attribute_variant (@var{mdecl},
10637 (get_identifier ("stdcall"),
10642 @deftypefn {Target Hook} bool TARGET_CANNOT_MODIFY_JUMPS_P (void)
10643 This target hook returns @code{true} past the point in which new jump
10644 instructions could be created. On machines that require a register for
10645 every jump such as the SHmedia ISA of SH5, this point would typically be
10646 reload, so this target hook should be defined to a function such as:
10650 cannot_modify_jumps_past_reload_p ()
10652 return (reload_completed || reload_in_progress);
10657 @deftypefn {Target Hook} int TARGET_BRANCH_TARGET_REGISTER_CLASS (void)
10658 This target hook returns a register class for which branch target register
10659 optimizations should be applied. All registers in this class should be
10660 usable interchangeably. After reload, registers in this class will be
10661 re-allocated and loads will be hoisted out of loops and be subjected
10662 to inter-block scheduling.
10665 @deftypefn {Target Hook} bool TARGET_BRANCH_TARGET_REGISTER_CALLEE_SAVED (bool @var{after_prologue_epilogue_gen})
10666 Branch target register optimization will by default exclude callee-saved
10668 that are not already live during the current function; if this target hook
10669 returns true, they will be included. The target code must than make sure
10670 that all target registers in the class returned by
10671 @samp{TARGET_BRANCH_TARGET_REGISTER_CLASS} that might need saving are
10672 saved. @var{after_prologue_epilogue_gen} indicates if prologues and
10673 epilogues have already been generated. Note, even if you only return
10674 true when @var{after_prologue_epilogue_gen} is false, you still are likely
10675 to have to make special provisions in @code{INITIAL_ELIMINATION_OFFSET}
10676 to reserve space for caller-saved target registers.
10679 @defmac POWI_MAX_MULTS
10680 If defined, this macro is interpreted as a signed integer C expression
10681 that specifies the maximum number of floating point multiplications
10682 that should be emitted when expanding exponentiation by an integer
10683 constant inline. When this value is defined, exponentiation requiring
10684 more than this number of multiplications is implemented by calling the
10685 system library's @code{pow}, @code{powf} or @code{powl} routines.
10686 The default value places no upper bound on the multiplication count.
10689 @deftypefn Macro void TARGET_EXTRA_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
10690 This target hook should register any extra include files for the
10691 target. The parameter @var{stdinc} indicates if normal include files
10692 are present. The parameter @var{sysroot} is the system root directory.
10693 The parameter @var{iprefix} is the prefix for the gcc directory.
10696 @deftypefn Macro void TARGET_EXTRA_PRE_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
10697 This target hook should register any extra include files for the
10698 target before any standard headers. The parameter @var{stdinc}
10699 indicates if normal include files are present. The parameter
10700 @var{sysroot} is the system root directory. The parameter
10701 @var{iprefix} is the prefix for the gcc directory.
10704 @deftypefn Macro void TARGET_OPTF (char *@var{path})
10705 This target hook should register special include paths for the target.
10706 The parameter @var{path} is the include to register. On Darwin
10707 systems, this is used for Framework includes, which have semantics
10708 that are different from @option{-I}.
10711 @deftypefn {Target Hook} bool TARGET_USE_LOCAL_THUNK_ALIAS_P (tree @var{fndecl})
10712 This target hook returns @code{true} if it is safe to use a local alias
10713 for a virtual function @var{fndecl} when constructing thunks,
10714 @code{false} otherwise. By default, the hook returns @code{true} for all
10715 functions, if a target supports aliases (i.e.@: defines
10716 @code{ASM_OUTPUT_DEF}), @code{false} otherwise,
10719 @defmac TARGET_FORMAT_TYPES
10720 If defined, this macro is the name of a global variable containing
10721 target-specific format checking information for the @option{-Wformat}
10722 option. The default is to have no target-specific format checks.
10725 @defmac TARGET_N_FORMAT_TYPES
10726 If defined, this macro is the number of entries in
10727 @code{TARGET_FORMAT_TYPES}.
10730 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES
10731 If defined, this macro is the name of a global variable containing
10732 target-specific format overrides for the @option{-Wformat} option. The
10733 default is to have no target-specific format overrides. If defined,
10734 @code{TARGET_FORMAT_TYPES} must be defined, too.
10737 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES_COUNT
10738 If defined, this macro specifies the number of entries in
10739 @code{TARGET_OVERRIDES_FORMAT_ATTRIBUTES}.
10742 @defmac TARGET_OVERRIDES_FORMAT_INIT
10743 If defined, this macro specifies the optional initialization
10744 routine for target specific customizations of the system printf
10745 and scanf formatter settings.
10748 @deftypefn {Target Hook} bool TARGET_RELAXED_ORDERING
10749 If set to @code{true}, means that the target's memory model does not
10750 guarantee that loads which do not depend on one another will access
10751 main memory in the order of the instruction stream; if ordering is
10752 important, an explicit memory barrier must be used. This is true of
10753 many recent processors which implement a policy of ``relaxed,''
10754 ``weak,'' or ``release'' memory consistency, such as Alpha, PowerPC,
10755 and ia64. The default is @code{false}.
10758 @deftypefn {Target Hook} const char *TARGET_INVALID_ARG_FOR_UNPROTOTYPED_FN (tree @var{typelist}, tree @var{funcdecl}, tree @var{val})
10759 If defined, this macro returns the diagnostic message when it is
10760 illegal to pass argument @var{val} to function @var{funcdecl}
10761 with prototype @var{typelist}.
10764 @deftypefn {Target Hook} {const char *} TARGET_INVALID_CONVERSION (tree @var{fromtype}, tree @var{totype})
10765 If defined, this macro returns the diagnostic message when it is
10766 invalid to convert from @var{fromtype} to @var{totype}, or @code{NULL}
10767 if validity should be determined by the front end.
10770 @deftypefn {Target Hook} {const char *} TARGET_INVALID_UNARY_OP (int @var{op}, tree @var{type})
10771 If defined, this macro returns the diagnostic message when it is
10772 invalid to apply operation @var{op} (where unary plus is denoted by
10773 @code{CONVERT_EXPR}) to an operand of type @var{type}, or @code{NULL}
10774 if validity should be determined by the front end.
10777 @deftypefn {Target Hook} {const char *} TARGET_INVALID_BINARY_OP (int @var{op}, tree @var{type1}, tree @var{type2})
10778 If defined, this macro returns the diagnostic message when it is
10779 invalid to apply operation @var{op} to operands of types @var{type1}
10780 and @var{type2}, or @code{NULL} if validity should be determined by
10784 @defmac TARGET_USE_JCR_SECTION
10785 This macro determines whether to use the JCR section to register Java
10786 classes. By default, TARGET_USE_JCR_SECTION is defined to 1 if both
10787 SUPPORTS_WEAK and TARGET_HAVE_NAMED_SECTIONS are true, else 0.
10791 This macro determines the size of the objective C jump buffer for the
10792 NeXT runtime. By default, OBJC_JBLEN is defined to an innocuous value.
10795 @defmac LIBGCC2_UNWIND_ATTRIBUTE
10796 Define this macro if any target-specific attributes need to be attached
10797 to the functions in @file{libgcc} that provide low-level support for
10798 call stack unwinding. It is used in declarations in @file{unwind-generic.h}
10799 and the associated definitions of those functions.
10802 @deftypefn {Target Hook} void TARGET_UPDATE_STACK_BOUNDARY (void)
10803 Define this macro to update the current function stack boundary if
10807 @deftypefn {Target Hook} rtx TARGET_GET_DRAP_RTX (void)
10808 Define this macro to an rtx for Dynamic Realign Argument Pointer if a
10809 different argument pointer register is needed to access the function's
10810 argument list when stack is aligned.
10813 @deftypefn {Target Hook} {bool} TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS (void)
10814 When optimization is disabled, this hook indicates whether or not
10815 arguments should be allocated to stack slots. Normally, GCC allocates
10816 stacks slots for arguments when not optimizing in order to make
10817 debugging easier. However, when a function is declared with
10818 @code{__attribute__((naked))}, there is no stack frame, and the compiler
10819 cannot safely move arguments from the registers in which they are passed
10820 to the stack. Therefore, this hook should return true in general, but
10821 false for naked functions. The default implementation always returns true.