1 @c Copyright (C) 1988-2016 Free Software Foundation, Inc.
2 @c This is part of the GCC manual.
3 @c For copying conditions, see the file gcc.texi.
6 @chapter Target Description Macros and Functions
7 @cindex machine description macros
8 @cindex target description macros
9 @cindex macros, target description
10 @cindex @file{tm.h} macros
12 In addition to the file @file{@var{machine}.md}, a machine description
13 includes a C header file conventionally given the name
14 @file{@var{machine}.h} and a C source file named @file{@var{machine}.c}.
15 The header file defines numerous macros that convey the information
16 about the target machine that does not fit into the scheme of the
17 @file{.md} file. The file @file{tm.h} should be a link to
18 @file{@var{machine}.h}. The header file @file{config.h} includes
19 @file{tm.h} and most compiler source files include @file{config.h}. The
20 source file defines a variable @code{targetm}, which is a structure
21 containing pointers to functions and data relating to the target
22 machine. @file{@var{machine}.c} should also contain their definitions,
23 if they are not defined elsewhere in GCC, and other functions called
24 through the macros defined in the @file{.h} file.
27 * Target Structure:: The @code{targetm} variable.
28 * Driver:: Controlling how the driver runs the compilation passes.
29 * Run-time Target:: Defining @samp{-m} options like @option{-m68000} and @option{-m68020}.
30 * Per-Function Data:: Defining data structures for per-function information.
31 * Storage Layout:: Defining sizes and alignments of data.
32 * Type Layout:: Defining sizes and properties of basic user data types.
33 * Registers:: Naming and describing the hardware registers.
34 * Register Classes:: Defining the classes of hardware registers.
35 * Stack and Calling:: Defining which way the stack grows and by how much.
36 * Varargs:: Defining the varargs macros.
37 * Trampolines:: Code set up at run time to enter a nested function.
38 * Library Calls:: Controlling how library routines are implicitly called.
39 * Addressing Modes:: Defining addressing modes valid for memory operands.
40 * Anchored Addresses:: Defining how @option{-fsection-anchors} should work.
41 * Condition Code:: Defining how insns update the condition code.
42 * Costs:: Defining relative costs of different operations.
43 * Scheduling:: Adjusting the behavior of the instruction scheduler.
44 * Sections:: Dividing storage into text, data, and other sections.
45 * PIC:: Macros for position independent code.
46 * Assembler Format:: Defining how to write insns and pseudo-ops to output.
47 * Debugging Info:: Defining the format of debugging output.
48 * Floating Point:: Handling floating point for cross-compilers.
49 * Mode Switching:: Insertion of mode-switching instructions.
50 * Target Attributes:: Defining target-specific uses of @code{__attribute__}.
51 * Emulated TLS:: Emulated TLS support.
52 * MIPS Coprocessors:: MIPS coprocessor support and how to customize it.
53 * PCH Target:: Validity checking for precompiled headers.
54 * C++ ABI:: Controlling C++ ABI changes.
55 * Named Address Spaces:: Adding support for named address spaces
56 * Misc:: Everything else.
59 @node Target Structure
60 @section The Global @code{targetm} Variable
62 @cindex target functions
64 @deftypevar {struct gcc_target} targetm
65 The target @file{.c} file must define the global @code{targetm} variable
66 which contains pointers to functions and data relating to the target
67 machine. The variable is declared in @file{target.h};
68 @file{target-def.h} defines the macro @code{TARGET_INITIALIZER} which is
69 used to initialize the variable, and macros for the default initializers
70 for elements of the structure. The @file{.c} file should override those
71 macros for which the default definition is inappropriate. For example:
74 #include "target-def.h"
76 /* @r{Initialize the GCC target structure.} */
78 #undef TARGET_COMP_TYPE_ATTRIBUTES
79 #define TARGET_COMP_TYPE_ATTRIBUTES @var{machine}_comp_type_attributes
81 struct gcc_target targetm = TARGET_INITIALIZER;
85 Where a macro should be defined in the @file{.c} file in this manner to
86 form part of the @code{targetm} structure, it is documented below as a
87 ``Target Hook'' with a prototype. Many macros will change in future
88 from being defined in the @file{.h} file to being part of the
89 @code{targetm} structure.
91 Similarly, there is a @code{targetcm} variable for hooks that are
92 specific to front ends for C-family languages, documented as ``C
93 Target Hook''. This is declared in @file{c-family/c-target.h}, the
94 initializer @code{TARGETCM_INITIALIZER} in
95 @file{c-family/c-target-def.h}. If targets initialize @code{targetcm}
96 themselves, they should set @code{target_has_targetcm=yes} in
97 @file{config.gcc}; otherwise a default definition is used.
99 Similarly, there is a @code{targetm_common} variable for hooks that
100 are shared between the compiler driver and the compilers proper,
101 documented as ``Common Target Hook''. This is declared in
102 @file{common/common-target.h}, the initializer
103 @code{TARGETM_COMMON_INITIALIZER} in
104 @file{common/common-target-def.h}. If targets initialize
105 @code{targetm_common} themselves, they should set
106 @code{target_has_targetm_common=yes} in @file{config.gcc}; otherwise a
107 default definition is used.
110 @section Controlling the Compilation Driver, @file{gcc}
112 @cindex controlling the compilation driver
114 @c prevent bad page break with this line
115 You can control the compilation driver.
117 @defmac DRIVER_SELF_SPECS
118 A list of specs for the driver itself. It should be a suitable
119 initializer for an array of strings, with no surrounding braces.
121 The driver applies these specs to its own command line between loading
122 default @file{specs} files (but not command-line specified ones) and
123 choosing the multilib directory or running any subcommands. It
124 applies them in the order given, so each spec can depend on the
125 options added by earlier ones. It is also possible to remove options
126 using @samp{%<@var{option}} in the usual way.
128 This macro can be useful when a port has several interdependent target
129 options. It provides a way of standardizing the command line so
130 that the other specs are easier to write.
132 Do not define this macro if it does not need to do anything.
135 @defmac OPTION_DEFAULT_SPECS
136 A list of specs used to support configure-time default options (i.e.@:
137 @option{--with} options) in the driver. It should be a suitable initializer
138 for an array of structures, each containing two strings, without the
139 outermost pair of surrounding braces.
141 The first item in the pair is the name of the default. This must match
142 the code in @file{config.gcc} for the target. The second item is a spec
143 to apply if a default with this name was specified. The string
144 @samp{%(VALUE)} in the spec will be replaced by the value of the default
145 everywhere it occurs.
147 The driver will apply these specs to its own command line between loading
148 default @file{specs} files and processing @code{DRIVER_SELF_SPECS}, using
149 the same mechanism as @code{DRIVER_SELF_SPECS}.
151 Do not define this macro if it does not need to do anything.
155 A C string constant that tells the GCC driver program options to
156 pass to CPP@. It can also specify how to translate options you
157 give to GCC into options for GCC to pass to the CPP@.
159 Do not define this macro if it does not need to do anything.
162 @defmac CPLUSPLUS_CPP_SPEC
163 This macro is just like @code{CPP_SPEC}, but is used for C++, rather
164 than C@. If you do not define this macro, then the value of
165 @code{CPP_SPEC} (if any) will be used instead.
169 A C string constant that tells the GCC driver program options to
170 pass to @code{cc1}, @code{cc1plus}, @code{f771}, and the other language
172 It can also specify how to translate options you give to GCC into options
173 for GCC to pass to front ends.
175 Do not define this macro if it does not need to do anything.
179 A C string constant that tells the GCC driver program options to
180 pass to @code{cc1plus}. It can also specify how to translate options you
181 give to GCC into options for GCC to pass to the @code{cc1plus}.
183 Do not define this macro if it does not need to do anything.
184 Note that everything defined in CC1_SPEC is already passed to
185 @code{cc1plus} so there is no need to duplicate the contents of
186 CC1_SPEC in CC1PLUS_SPEC@.
190 A C string constant that tells the GCC driver program options to
191 pass to the assembler. It can also specify how to translate options
192 you give to GCC into options for GCC to pass to the assembler.
193 See the file @file{sun3.h} for an example of this.
195 Do not define this macro if it does not need to do anything.
198 @defmac ASM_FINAL_SPEC
199 A C string constant that tells the GCC driver program how to
200 run any programs which cleanup after the normal assembler.
201 Normally, this is not needed. See the file @file{mips.h} for
204 Do not define this macro if it does not need to do anything.
207 @defmac AS_NEEDS_DASH_FOR_PIPED_INPUT
208 Define this macro, with no value, if the driver should give the assembler
209 an argument consisting of a single dash, @option{-}, to instruct it to
210 read from its standard input (which will be a pipe connected to the
211 output of the compiler proper). This argument is given after any
212 @option{-o} option specifying the name of the output file.
214 If you do not define this macro, the assembler is assumed to read its
215 standard input if given no non-option arguments. If your assembler
216 cannot read standard input at all, use a @samp{%@{pipe:%e@}} construct;
217 see @file{mips.h} for instance.
221 A C string constant that tells the GCC driver program options to
222 pass to the linker. It can also specify how to translate options you
223 give to GCC into options for GCC to pass to the linker.
225 Do not define this macro if it does not need to do anything.
229 Another C string constant used much like @code{LINK_SPEC}. The difference
230 between the two is that @code{LIB_SPEC} is used at the end of the
231 command given to the linker.
233 If this macro is not defined, a default is provided that
234 loads the standard C library from the usual place. See @file{gcc.c}.
238 Another C string constant that tells the GCC driver program
239 how and when to place a reference to @file{libgcc.a} into the
240 linker command line. This constant is placed both before and after
241 the value of @code{LIB_SPEC}.
243 If this macro is not defined, the GCC driver provides a default that
244 passes the string @option{-lgcc} to the linker.
247 @defmac REAL_LIBGCC_SPEC
248 By default, if @code{ENABLE_SHARED_LIBGCC} is defined, the
249 @code{LIBGCC_SPEC} is not directly used by the driver program but is
250 instead modified to refer to different versions of @file{libgcc.a}
251 depending on the values of the command line flags @option{-static},
252 @option{-shared}, @option{-static-libgcc}, and @option{-shared-libgcc}. On
253 targets where these modifications are inappropriate, define
254 @code{REAL_LIBGCC_SPEC} instead. @code{REAL_LIBGCC_SPEC} tells the
255 driver how to place a reference to @file{libgcc} on the link command
256 line, but, unlike @code{LIBGCC_SPEC}, it is used unmodified.
259 @defmac USE_LD_AS_NEEDED
260 A macro that controls the modifications to @code{LIBGCC_SPEC}
261 mentioned in @code{REAL_LIBGCC_SPEC}. If nonzero, a spec will be
262 generated that uses @option{--as-needed} or equivalent options and the
263 shared @file{libgcc} in place of the
264 static exception handler library, when linking without any of
265 @code{-static}, @code{-static-libgcc}, or @code{-shared-libgcc}.
269 If defined, this C string constant is added to @code{LINK_SPEC}.
270 When @code{USE_LD_AS_NEEDED} is zero or undefined, it also affects
271 the modifications to @code{LIBGCC_SPEC} mentioned in
272 @code{REAL_LIBGCC_SPEC}.
275 @defmac STARTFILE_SPEC
276 Another C string constant used much like @code{LINK_SPEC}. The
277 difference between the two is that @code{STARTFILE_SPEC} is used at
278 the very beginning of the command given to the linker.
280 If this macro is not defined, a default is provided that loads the
281 standard C startup file from the usual place. See @file{gcc.c}.
285 Another C string constant used much like @code{LINK_SPEC}. The
286 difference between the two is that @code{ENDFILE_SPEC} is used at
287 the very end of the command given to the linker.
289 Do not define this macro if it does not need to do anything.
292 @defmac THREAD_MODEL_SPEC
293 GCC @code{-v} will print the thread model GCC was configured to use.
294 However, this doesn't work on platforms that are multilibbed on thread
295 models, such as AIX 4.3. On such platforms, define
296 @code{THREAD_MODEL_SPEC} such that it evaluates to a string without
297 blanks that names one of the recognized thread models. @code{%*}, the
298 default value of this macro, will expand to the value of
299 @code{thread_file} set in @file{config.gcc}.
302 @defmac SYSROOT_SUFFIX_SPEC
303 Define this macro to add a suffix to the target sysroot when GCC is
304 configured with a sysroot. This will cause GCC to search for usr/lib,
305 et al, within sysroot+suffix.
308 @defmac SYSROOT_HEADERS_SUFFIX_SPEC
309 Define this macro to add a headers_suffix to the target sysroot when
310 GCC is configured with a sysroot. This will cause GCC to pass the
311 updated sysroot+headers_suffix to CPP, causing it to search for
312 usr/include, et al, within sysroot+headers_suffix.
316 Define this macro to provide additional specifications to put in the
317 @file{specs} file that can be used in various specifications like
320 The definition should be an initializer for an array of structures,
321 containing a string constant, that defines the specification name, and a
322 string constant that provides the specification.
324 Do not define this macro if it does not need to do anything.
326 @code{EXTRA_SPECS} is useful when an architecture contains several
327 related targets, which have various @code{@dots{}_SPECS} which are similar
328 to each other, and the maintainer would like one central place to keep
331 For example, the PowerPC System V.4 targets use @code{EXTRA_SPECS} to
332 define either @code{_CALL_SYSV} when the System V calling sequence is
333 used or @code{_CALL_AIX} when the older AIX-based calling sequence is
336 The @file{config/rs6000/rs6000.h} target file defines:
339 #define EXTRA_SPECS \
340 @{ "cpp_sysv_default", CPP_SYSV_DEFAULT @},
342 #define CPP_SYS_DEFAULT ""
345 The @file{config/rs6000/sysv.h} target file defines:
349 "%@{posix: -D_POSIX_SOURCE @} \
350 %@{mcall-sysv: -D_CALL_SYSV @} \
351 %@{!mcall-sysv: %(cpp_sysv_default) @} \
352 %@{msoft-float: -D_SOFT_FLOAT@} %@{mcpu=403: -D_SOFT_FLOAT@}"
354 #undef CPP_SYSV_DEFAULT
355 #define CPP_SYSV_DEFAULT "-D_CALL_SYSV"
358 while the @file{config/rs6000/eabiaix.h} target file defines
359 @code{CPP_SYSV_DEFAULT} as:
362 #undef CPP_SYSV_DEFAULT
363 #define CPP_SYSV_DEFAULT "-D_CALL_AIX"
367 @defmac LINK_LIBGCC_SPECIAL_1
368 Define this macro if the driver program should find the library
369 @file{libgcc.a}. If you do not define this macro, the driver program will pass
370 the argument @option{-lgcc} to tell the linker to do the search.
373 @defmac LINK_GCC_C_SEQUENCE_SPEC
374 The sequence in which libgcc and libc are specified to the linker.
375 By default this is @code{%G %L %G}.
378 @defmac POST_LINK_SPEC
379 Define this macro to add additional steps to be executed after linker.
380 The default value of this macro is empty string.
383 @defmac LINK_COMMAND_SPEC
384 A C string constant giving the complete command line need to execute the
385 linker. When you do this, you will need to update your port each time a
386 change is made to the link command line within @file{gcc.c}. Therefore,
387 define this macro only if you need to completely redefine the command
388 line for invoking the linker and there is no other way to accomplish
389 the effect you need. Overriding this macro may be avoidable by overriding
390 @code{LINK_GCC_C_SEQUENCE_SPEC} instead.
393 @hook TARGET_ALWAYS_STRIP_DOTDOT
395 @defmac MULTILIB_DEFAULTS
396 Define this macro as a C expression for the initializer of an array of
397 string to tell the driver program which options are defaults for this
398 target and thus do not need to be handled specially when using
399 @code{MULTILIB_OPTIONS}.
401 Do not define this macro if @code{MULTILIB_OPTIONS} is not defined in
402 the target makefile fragment or if none of the options listed in
403 @code{MULTILIB_OPTIONS} are set by default.
404 @xref{Target Fragment}.
407 @defmac RELATIVE_PREFIX_NOT_LINKDIR
408 Define this macro to tell @command{gcc} that it should only translate
409 a @option{-B} prefix into a @option{-L} linker option if the prefix
410 indicates an absolute file name.
413 @defmac MD_EXEC_PREFIX
414 If defined, this macro is an additional prefix to try after
415 @code{STANDARD_EXEC_PREFIX}. @code{MD_EXEC_PREFIX} is not searched
416 when the compiler is built as a cross
417 compiler. If you define @code{MD_EXEC_PREFIX}, then be sure to add it
418 to the list of directories used to find the assembler in @file{configure.ac}.
421 @defmac STANDARD_STARTFILE_PREFIX
422 Define this macro as a C string constant if you wish to override the
423 standard choice of @code{libdir} as the default prefix to
424 try when searching for startup files such as @file{crt0.o}.
425 @code{STANDARD_STARTFILE_PREFIX} is not searched when the compiler
426 is built as a cross compiler.
429 @defmac STANDARD_STARTFILE_PREFIX_1
430 Define this macro as a C string constant if you wish to override the
431 standard choice of @code{/lib} as a prefix to try after the default prefix
432 when searching for startup files such as @file{crt0.o}.
433 @code{STANDARD_STARTFILE_PREFIX_1} is not searched when the compiler
434 is built as a cross compiler.
437 @defmac STANDARD_STARTFILE_PREFIX_2
438 Define this macro as a C string constant if you wish to override the
439 standard choice of @code{/lib} as yet another prefix to try after the
440 default prefix when searching for startup files such as @file{crt0.o}.
441 @code{STANDARD_STARTFILE_PREFIX_2} is not searched when the compiler
442 is built as a cross compiler.
445 @defmac MD_STARTFILE_PREFIX
446 If defined, this macro supplies an additional prefix to try after the
447 standard prefixes. @code{MD_EXEC_PREFIX} is not searched when the
448 compiler is built as a cross compiler.
451 @defmac MD_STARTFILE_PREFIX_1
452 If defined, this macro supplies yet another prefix to try after the
453 standard prefixes. It is not searched when the compiler is built as a
457 @defmac INIT_ENVIRONMENT
458 Define this macro as a C string constant if you wish to set environment
459 variables for programs called by the driver, such as the assembler and
460 loader. The driver passes the value of this macro to @code{putenv} to
461 initialize the necessary environment variables.
464 @defmac LOCAL_INCLUDE_DIR
465 Define this macro as a C string constant if you wish to override the
466 standard choice of @file{/usr/local/include} as the default prefix to
467 try when searching for local header files. @code{LOCAL_INCLUDE_DIR}
468 comes before @code{NATIVE_SYSTEM_HEADER_DIR} (set in
469 @file{config.gcc}, normally @file{/usr/include}) in the search order.
471 Cross compilers do not search either @file{/usr/local/include} or its
475 @defmac NATIVE_SYSTEM_HEADER_COMPONENT
476 The ``component'' corresponding to @code{NATIVE_SYSTEM_HEADER_DIR}.
477 See @code{INCLUDE_DEFAULTS}, below, for the description of components.
478 If you do not define this macro, no component is used.
481 @defmac INCLUDE_DEFAULTS
482 Define this macro if you wish to override the entire default search path
483 for include files. For a native compiler, the default search path
484 usually consists of @code{GCC_INCLUDE_DIR}, @code{LOCAL_INCLUDE_DIR},
485 @code{GPLUSPLUS_INCLUDE_DIR}, and
486 @code{NATIVE_SYSTEM_HEADER_DIR}. In addition, @code{GPLUSPLUS_INCLUDE_DIR}
487 and @code{GCC_INCLUDE_DIR} are defined automatically by @file{Makefile},
488 and specify private search areas for GCC@. The directory
489 @code{GPLUSPLUS_INCLUDE_DIR} is used only for C++ programs.
491 The definition should be an initializer for an array of structures.
492 Each array element should have four elements: the directory name (a
493 string constant), the component name (also a string constant), a flag
494 for C++-only directories,
495 and a flag showing that the includes in the directory don't need to be
496 wrapped in @code{extern @samp{C}} when compiling C++. Mark the end of
497 the array with a null element.
499 The component name denotes what GNU package the include file is part of,
500 if any, in all uppercase letters. For example, it might be @samp{GCC}
501 or @samp{BINUTILS}. If the package is part of a vendor-supplied
502 operating system, code the component name as @samp{0}.
504 For example, here is the definition used for VAX/VMS:
507 #define INCLUDE_DEFAULTS \
509 @{ "GNU_GXX_INCLUDE:", "G++", 1, 1@}, \
510 @{ "GNU_CC_INCLUDE:", "GCC", 0, 0@}, \
511 @{ "SYS$SYSROOT:[SYSLIB.]", 0, 0, 0@}, \
518 Here is the order of prefixes tried for exec files:
522 Any prefixes specified by the user with @option{-B}.
525 The environment variable @code{GCC_EXEC_PREFIX} or, if @code{GCC_EXEC_PREFIX}
526 is not set and the compiler has not been installed in the configure-time
527 @var{prefix}, the location in which the compiler has actually been installed.
530 The directories specified by the environment variable @code{COMPILER_PATH}.
533 The macro @code{STANDARD_EXEC_PREFIX}, if the compiler has been installed
534 in the configured-time @var{prefix}.
537 The location @file{/usr/libexec/gcc/}, but only if this is a native compiler.
540 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
543 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
547 Here is the order of prefixes tried for startfiles:
551 Any prefixes specified by the user with @option{-B}.
554 The environment variable @code{GCC_EXEC_PREFIX} or its automatically determined
555 value based on the installed toolchain location.
558 The directories specified by the environment variable @code{LIBRARY_PATH}
559 (or port-specific name; native only, cross compilers do not use this).
562 The macro @code{STANDARD_EXEC_PREFIX}, but only if the toolchain is installed
563 in the configured @var{prefix} or this is a native compiler.
566 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
569 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
573 The macro @code{MD_STARTFILE_PREFIX}, if defined, but only if this is a
574 native compiler, or we have a target system root.
577 The macro @code{MD_STARTFILE_PREFIX_1}, if defined, but only if this is a
578 native compiler, or we have a target system root.
581 The macro @code{STANDARD_STARTFILE_PREFIX}, with any sysroot modifications.
582 If this path is relative it will be prefixed by @code{GCC_EXEC_PREFIX} and
583 the machine suffix or @code{STANDARD_EXEC_PREFIX} and the machine suffix.
586 The macro @code{STANDARD_STARTFILE_PREFIX_1}, but only if this is a native
587 compiler, or we have a target system root. The default for this macro is
591 The macro @code{STANDARD_STARTFILE_PREFIX_2}, but only if this is a native
592 compiler, or we have a target system root. The default for this macro is
596 @node Run-time Target
597 @section Run-time Target Specification
598 @cindex run-time target specification
599 @cindex predefined macros
600 @cindex target specifications
602 @c prevent bad page break with this line
603 Here are run-time target specifications.
605 @defmac TARGET_CPU_CPP_BUILTINS ()
606 This function-like macro expands to a block of code that defines
607 built-in preprocessor macros and assertions for the target CPU, using
608 the functions @code{builtin_define}, @code{builtin_define_std} and
609 @code{builtin_assert}. When the front end
610 calls this macro it provides a trailing semicolon, and since it has
611 finished command line option processing your code can use those
614 @code{builtin_assert} takes a string in the form you pass to the
615 command-line option @option{-A}, such as @code{cpu=mips}, and creates
616 the assertion. @code{builtin_define} takes a string in the form
617 accepted by option @option{-D} and unconditionally defines the macro.
619 @code{builtin_define_std} takes a string representing the name of an
620 object-like macro. If it doesn't lie in the user's namespace,
621 @code{builtin_define_std} defines it unconditionally. Otherwise, it
622 defines a version with two leading underscores, and another version
623 with two leading and trailing underscores, and defines the original
624 only if an ISO standard was not requested on the command line. For
625 example, passing @code{unix} defines @code{__unix}, @code{__unix__}
626 and possibly @code{unix}; passing @code{_mips} defines @code{__mips},
627 @code{__mips__} and possibly @code{_mips}, and passing @code{_ABI64}
628 defines only @code{_ABI64}.
630 You can also test for the C dialect being compiled. The variable
631 @code{c_language} is set to one of @code{clk_c}, @code{clk_cplusplus}
632 or @code{clk_objective_c}. Note that if we are preprocessing
633 assembler, this variable will be @code{clk_c} but the function-like
634 macro @code{preprocessing_asm_p()} will return true, so you might want
635 to check for that first. If you need to check for strict ANSI, the
636 variable @code{flag_iso} can be used. The function-like macro
637 @code{preprocessing_trad_p()} can be used to check for traditional
641 @defmac TARGET_OS_CPP_BUILTINS ()
642 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
643 and is used for the target operating system instead.
646 @defmac TARGET_OBJFMT_CPP_BUILTINS ()
647 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
648 and is used for the target object format. @file{elfos.h} uses this
649 macro to define @code{__ELF__}, so you probably do not need to define
653 @deftypevar {extern int} target_flags
654 This variable is declared in @file{options.h}, which is included before
655 any target-specific headers.
658 @hook TARGET_DEFAULT_TARGET_FLAGS
659 This variable specifies the initial value of @code{target_flags}.
660 Its default setting is 0.
663 @cindex optional hardware or system features
664 @cindex features, optional, in system conventions
666 @hook TARGET_HANDLE_OPTION
667 This hook is called whenever the user specifies one of the
668 target-specific options described by the @file{.opt} definition files
669 (@pxref{Options}). It has the opportunity to do some option-specific
670 processing and should return true if the option is valid. The default
671 definition does nothing but return true.
673 @var{decoded} specifies the option and its arguments. @var{opts} and
674 @var{opts_set} are the @code{gcc_options} structures to be used for
675 storing option state, and @var{loc} is the location at which the
676 option was passed (@code{UNKNOWN_LOCATION} except for options passed
680 @hook TARGET_HANDLE_C_OPTION
681 This target hook is called whenever the user specifies one of the
682 target-specific C language family options described by the @file{.opt}
683 definition files(@pxref{Options}). It has the opportunity to do some
684 option-specific processing and should return true if the option is
685 valid. The arguments are like for @code{TARGET_HANDLE_OPTION}. The
686 default definition does nothing but return false.
688 In general, you should use @code{TARGET_HANDLE_OPTION} to handle
689 options. However, if processing an option requires routines that are
690 only available in the C (and related language) front ends, then you
691 should use @code{TARGET_HANDLE_C_OPTION} instead.
694 @hook TARGET_OBJC_CONSTRUCT_STRING_OBJECT
696 @hook TARGET_OBJC_DECLARE_UNRESOLVED_CLASS_REFERENCE
698 @hook TARGET_OBJC_DECLARE_CLASS_DEFINITION
700 @hook TARGET_STRING_OBJECT_REF_TYPE_P
702 @hook TARGET_CHECK_STRING_OBJECT_FORMAT_ARG
704 @hook TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE
706 @defmac C_COMMON_OVERRIDE_OPTIONS
707 This is similar to the @code{TARGET_OPTION_OVERRIDE} hook
708 but is only used in the C
709 language frontends (C, Objective-C, C++, Objective-C++) and so can be
710 used to alter option flag variables which only exist in those
714 @hook TARGET_OPTION_OPTIMIZATION_TABLE
715 Some machines may desire to change what optimizations are performed for
716 various optimization levels. This variable, if defined, describes
717 options to enable at particular sets of optimization levels. These
718 options are processed once
719 just after the optimization level is determined and before the remainder
720 of the command options have been parsed, so may be overridden by other
721 options passed explicitly.
723 This processing is run once at program startup and when the optimization
724 options are changed via @code{#pragma GCC optimize} or by using the
725 @code{optimize} attribute.
728 @hook TARGET_OPTION_INIT_STRUCT
730 @hook TARGET_OPTION_DEFAULT_PARAMS
732 @defmac SWITCHABLE_TARGET
733 Some targets need to switch between substantially different subtargets
734 during compilation. For example, the MIPS target has one subtarget for
735 the traditional MIPS architecture and another for MIPS16. Source code
736 can switch between these two subarchitectures using the @code{mips16}
737 and @code{nomips16} attributes.
739 Such subtargets can differ in things like the set of available
740 registers, the set of available instructions, the costs of various
741 operations, and so on. GCC caches a lot of this type of information
742 in global variables, and recomputing them for each subtarget takes a
743 significant amount of time. The compiler therefore provides a facility
744 for maintaining several versions of the global variables and quickly
745 switching between them; see @file{target-globals.h} for details.
747 Define this macro to 1 if your target needs this facility. The default
751 @hook TARGET_FLOAT_EXCEPTIONS_ROUNDING_SUPPORTED_P
753 @node Per-Function Data
754 @section Defining data structures for per-function information.
755 @cindex per-function data
756 @cindex data structures
758 If the target needs to store information on a per-function basis, GCC
759 provides a macro and a couple of variables to allow this. Note, just
760 using statics to store the information is a bad idea, since GCC supports
761 nested functions, so you can be halfway through encoding one function
762 when another one comes along.
764 GCC defines a data structure called @code{struct function} which
765 contains all of the data specific to an individual function. This
766 structure contains a field called @code{machine} whose type is
767 @code{struct machine_function *}, which can be used by targets to point
768 to their own specific data.
770 If a target needs per-function specific data it should define the type
771 @code{struct machine_function} and also the macro @code{INIT_EXPANDERS}.
772 This macro should be used to initialize the function pointer
773 @code{init_machine_status}. This pointer is explained below.
775 One typical use of per-function, target specific data is to create an
776 RTX to hold the register containing the function's return address. This
777 RTX can then be used to implement the @code{__builtin_return_address}
778 function, for level 0.
780 Note---earlier implementations of GCC used a single data area to hold
781 all of the per-function information. Thus when processing of a nested
782 function began the old per-function data had to be pushed onto a
783 stack, and when the processing was finished, it had to be popped off the
784 stack. GCC used to provide function pointers called
785 @code{save_machine_status} and @code{restore_machine_status} to handle
786 the saving and restoring of the target specific information. Since the
787 single data area approach is no longer used, these pointers are no
790 @defmac INIT_EXPANDERS
791 Macro called to initialize any target specific information. This macro
792 is called once per function, before generation of any RTL has begun.
793 The intention of this macro is to allow the initialization of the
794 function pointer @code{init_machine_status}.
797 @deftypevar {void (*)(struct function *)} init_machine_status
798 If this function pointer is non-@code{NULL} it will be called once per
799 function, before function compilation starts, in order to allow the
800 target to perform any target specific initialization of the
801 @code{struct function} structure. It is intended that this would be
802 used to initialize the @code{machine} of that structure.
804 @code{struct machine_function} structures are expected to be freed by GC@.
805 Generally, any memory that they reference must be allocated by using
806 GC allocation, including the structure itself.
810 @section Storage Layout
811 @cindex storage layout
813 Note that the definitions of the macros in this table which are sizes or
814 alignments measured in bits do not need to be constant. They can be C
815 expressions that refer to static variables, such as the @code{target_flags}.
816 @xref{Run-time Target}.
818 @defmac BITS_BIG_ENDIAN
819 Define this macro to have the value 1 if the most significant bit in a
820 byte has the lowest number; otherwise define it to have the value zero.
821 This means that bit-field instructions count from the most significant
822 bit. If the machine has no bit-field instructions, then this must still
823 be defined, but it doesn't matter which value it is defined to. This
824 macro need not be a constant.
826 This macro does not affect the way structure fields are packed into
827 bytes or words; that is controlled by @code{BYTES_BIG_ENDIAN}.
830 @defmac BYTES_BIG_ENDIAN
831 Define this macro to have the value 1 if the most significant byte in a
832 word has the lowest number. This macro need not be a constant.
835 @defmac WORDS_BIG_ENDIAN
836 Define this macro to have the value 1 if, in a multiword object, the
837 most significant word has the lowest number. This applies to both
838 memory locations and registers; see @code{REG_WORDS_BIG_ENDIAN} if the
839 order of words in memory is not the same as the order in registers. This
840 macro need not be a constant.
843 @defmac REG_WORDS_BIG_ENDIAN
844 On some machines, the order of words in a multiword object differs between
845 registers in memory. In such a situation, define this macro to describe
846 the order of words in a register. The macro @code{WORDS_BIG_ENDIAN} controls
847 the order of words in memory.
850 @defmac FLOAT_WORDS_BIG_ENDIAN
851 Define this macro to have the value 1 if @code{DFmode}, @code{XFmode} or
852 @code{TFmode} floating point numbers are stored in memory with the word
853 containing the sign bit at the lowest address; otherwise define it to
854 have the value 0. This macro need not be a constant.
856 You need not define this macro if the ordering is the same as for
860 @defmac BITS_PER_WORD
861 Number of bits in a word. If you do not define this macro, the default
862 is @code{BITS_PER_UNIT * UNITS_PER_WORD}.
865 @defmac MAX_BITS_PER_WORD
866 Maximum number of bits in a word. If this is undefined, the default is
867 @code{BITS_PER_WORD}. Otherwise, it is the constant value that is the
868 largest value that @code{BITS_PER_WORD} can have at run-time.
871 @defmac UNITS_PER_WORD
872 Number of storage units in a word; normally the size of a general-purpose
873 register, a power of two from 1 or 8.
876 @defmac MIN_UNITS_PER_WORD
877 Minimum number of units in a word. If this is undefined, the default is
878 @code{UNITS_PER_WORD}. Otherwise, it is the constant value that is the
879 smallest value that @code{UNITS_PER_WORD} can have at run-time.
883 Width of a pointer, in bits. You must specify a value no wider than the
884 width of @code{Pmode}. If it is not equal to the width of @code{Pmode},
885 you must define @code{POINTERS_EXTEND_UNSIGNED}. If you do not specify
886 a value the default is @code{BITS_PER_WORD}.
889 @defmac POINTERS_EXTEND_UNSIGNED
890 A C expression that determines how pointers should be extended from
891 @code{ptr_mode} to either @code{Pmode} or @code{word_mode}. It is
892 greater than zero if pointers should be zero-extended, zero if they
893 should be sign-extended, and negative if some other sort of conversion
894 is needed. In the last case, the extension is done by the target's
895 @code{ptr_extend} instruction.
897 You need not define this macro if the @code{ptr_mode}, @code{Pmode}
898 and @code{word_mode} are all the same width.
901 @defmac PROMOTE_MODE (@var{m}, @var{unsignedp}, @var{type})
902 A macro to update @var{m} and @var{unsignedp} when an object whose type
903 is @var{type} and which has the specified mode and signedness is to be
904 stored in a register. This macro is only called when @var{type} is a
907 On most RISC machines, which only have operations that operate on a full
908 register, define this macro to set @var{m} to @code{word_mode} if
909 @var{m} is an integer mode narrower than @code{BITS_PER_WORD}. In most
910 cases, only integer modes should be widened because wider-precision
911 floating-point operations are usually more expensive than their narrower
914 For most machines, the macro definition does not change @var{unsignedp}.
915 However, some machines, have instructions that preferentially handle
916 either signed or unsigned quantities of certain modes. For example, on
917 the DEC Alpha, 32-bit loads from memory and 32-bit add instructions
918 sign-extend the result to 64 bits. On such machines, set
919 @var{unsignedp} according to which kind of extension is more efficient.
921 Do not define this macro if it would never modify @var{m}.
924 @hook TARGET_C_EXCESS_PRECISION
926 @hook TARGET_PROMOTE_FUNCTION_MODE
928 @defmac PARM_BOUNDARY
929 Normal alignment required for function parameters on the stack, in
930 bits. All stack parameters receive at least this much alignment
931 regardless of data type. On most machines, this is the same as the
935 @defmac STACK_BOUNDARY
936 Define this macro to the minimum alignment enforced by hardware for the
937 stack pointer on this machine. The definition is a C expression for the
938 desired alignment (measured in bits). This value is used as a default
939 if @code{PREFERRED_STACK_BOUNDARY} is not defined. On most machines,
940 this should be the same as @code{PARM_BOUNDARY}.
943 @defmac PREFERRED_STACK_BOUNDARY
944 Define this macro if you wish to preserve a certain alignment for the
945 stack pointer, greater than what the hardware enforces. The definition
946 is a C expression for the desired alignment (measured in bits). This
947 macro must evaluate to a value equal to or larger than
948 @code{STACK_BOUNDARY}.
951 @defmac INCOMING_STACK_BOUNDARY
952 Define this macro if the incoming stack boundary may be different
953 from @code{PREFERRED_STACK_BOUNDARY}. This macro must evaluate
954 to a value equal to or larger than @code{STACK_BOUNDARY}.
957 @defmac FUNCTION_BOUNDARY
958 Alignment required for a function entry point, in bits.
961 @defmac BIGGEST_ALIGNMENT
962 Biggest alignment that any data type can require on this machine, in
963 bits. Note that this is not the biggest alignment that is supported,
964 just the biggest alignment that, when violated, may cause a fault.
967 @hook TARGET_ABSOLUTE_BIGGEST_ALIGNMENT
969 @defmac MALLOC_ABI_ALIGNMENT
970 Alignment, in bits, a C conformant malloc implementation has to
971 provide. If not defined, the default value is @code{BITS_PER_WORD}.
974 @defmac ATTRIBUTE_ALIGNED_VALUE
975 Alignment used by the @code{__attribute__ ((aligned))} construct. If
976 not defined, the default value is @code{BIGGEST_ALIGNMENT}.
979 @defmac MINIMUM_ATOMIC_ALIGNMENT
980 If defined, the smallest alignment, in bits, that can be given to an
981 object that can be referenced in one operation, without disturbing any
982 nearby object. Normally, this is @code{BITS_PER_UNIT}, but may be larger
983 on machines that don't have byte or half-word store operations.
986 @defmac BIGGEST_FIELD_ALIGNMENT
987 Biggest alignment that any structure or union field can require on this
988 machine, in bits. If defined, this overrides @code{BIGGEST_ALIGNMENT} for
989 structure and union fields only, unless the field alignment has been set
990 by the @code{__attribute__ ((aligned (@var{n})))} construct.
993 @defmac ADJUST_FIELD_ALIGN (@var{field}, @var{computed})
994 An expression for the alignment of a structure field @var{field} if the
995 alignment computed in the usual way (including applying of
996 @code{BIGGEST_ALIGNMENT} and @code{BIGGEST_FIELD_ALIGNMENT} to the
997 alignment) is @var{computed}. It overrides alignment only if the
998 field alignment has not been set by the
999 @code{__attribute__ ((aligned (@var{n})))} construct.
1002 @defmac MAX_STACK_ALIGNMENT
1003 Biggest stack alignment guaranteed by the backend. Use this macro
1004 to specify the maximum alignment of a variable on stack.
1006 If not defined, the default value is @code{STACK_BOUNDARY}.
1008 @c FIXME: The default should be @code{PREFERRED_STACK_BOUNDARY}.
1009 @c But the fix for PR 32893 indicates that we can only guarantee
1010 @c maximum stack alignment on stack up to @code{STACK_BOUNDARY}, not
1011 @c @code{PREFERRED_STACK_BOUNDARY}, if stack alignment isn't supported.
1014 @defmac MAX_OFILE_ALIGNMENT
1015 Biggest alignment supported by the object file format of this machine.
1016 Use this macro to limit the alignment which can be specified using the
1017 @code{__attribute__ ((aligned (@var{n})))} construct. If not defined,
1018 the default value is @code{BIGGEST_ALIGNMENT}.
1020 On systems that use ELF, the default (in @file{config/elfos.h}) is
1021 the largest supported 32-bit ELF section alignment representable on
1022 a 32-bit host e.g. @samp{(((uint64_t) 1 << 28) * 8)}.
1023 On 32-bit ELF the largest supported section alignment in bits is
1024 @samp{(0x80000000 * 8)}, but this is not representable on 32-bit hosts.
1027 @defmac DATA_ALIGNMENT (@var{type}, @var{basic-align})
1028 If defined, a C expression to compute the alignment for a variable in
1029 the static store. @var{type} is the data type, and @var{basic-align} is
1030 the alignment that the object would ordinarily have. The value of this
1031 macro is used instead of that alignment to align the object.
1033 If this macro is not defined, then @var{basic-align} is used.
1036 One use of this macro is to increase alignment of medium-size data to
1037 make it all fit in fewer cache lines. Another is to cause character
1038 arrays to be word-aligned so that @code{strcpy} calls that copy
1039 constants to character arrays can be done inline.
1042 @defmac DATA_ABI_ALIGNMENT (@var{type}, @var{basic-align})
1043 Similar to @code{DATA_ALIGNMENT}, but for the cases where the ABI mandates
1044 some alignment increase, instead of optimization only purposes. E.g.@
1045 AMD x86-64 psABI says that variables with array type larger than 15 bytes
1046 must be aligned to 16 byte boundaries.
1048 If this macro is not defined, then @var{basic-align} is used.
1051 @defmac CONSTANT_ALIGNMENT (@var{constant}, @var{basic-align})
1052 If defined, a C expression to compute the alignment given to a constant
1053 that is being placed in memory. @var{constant} is the constant and
1054 @var{basic-align} is the alignment that the object would ordinarily
1055 have. The value of this macro is used instead of that alignment to
1058 The default definition just returns @var{basic-align}.
1060 The typical use of this macro is to increase alignment for string
1061 constants to be word aligned so that @code{strcpy} calls that copy
1062 constants can be done inline.
1065 @defmac LOCAL_ALIGNMENT (@var{type}, @var{basic-align})
1066 If defined, a C expression to compute the alignment for a variable in
1067 the local store. @var{type} is the data type, and @var{basic-align} is
1068 the alignment that the object would ordinarily have. The value of this
1069 macro is used instead of that alignment to align the object.
1071 If this macro is not defined, then @var{basic-align} is used.
1073 One use of this macro is to increase alignment of medium-size data to
1074 make it all fit in fewer cache lines.
1076 If the value of this macro has a type, it should be an unsigned type.
1079 @hook TARGET_VECTOR_ALIGNMENT
1081 @defmac STACK_SLOT_ALIGNMENT (@var{type}, @var{mode}, @var{basic-align})
1082 If defined, a C expression to compute the alignment for stack slot.
1083 @var{type} is the data type, @var{mode} is the widest mode available,
1084 and @var{basic-align} is the alignment that the slot would ordinarily
1085 have. The value of this macro is used instead of that alignment to
1088 If this macro is not defined, then @var{basic-align} is used when
1089 @var{type} is @code{NULL}. Otherwise, @code{LOCAL_ALIGNMENT} will
1092 This macro is to set alignment of stack slot to the maximum alignment
1093 of all possible modes which the slot may have.
1095 If the value of this macro has a type, it should be an unsigned type.
1098 @defmac LOCAL_DECL_ALIGNMENT (@var{decl})
1099 If defined, a C expression to compute the alignment for a local
1100 variable @var{decl}.
1102 If this macro is not defined, then
1103 @code{LOCAL_ALIGNMENT (TREE_TYPE (@var{decl}), DECL_ALIGN (@var{decl}))}
1106 One use of this macro is to increase alignment of medium-size data to
1107 make it all fit in fewer cache lines.
1109 If the value of this macro has a type, it should be an unsigned type.
1112 @defmac MINIMUM_ALIGNMENT (@var{exp}, @var{mode}, @var{align})
1113 If defined, a C expression to compute the minimum required alignment
1114 for dynamic stack realignment purposes for @var{exp} (a type or decl),
1115 @var{mode}, assuming normal alignment @var{align}.
1117 If this macro is not defined, then @var{align} will be used.
1120 @defmac EMPTY_FIELD_BOUNDARY
1121 Alignment in bits to be given to a structure bit-field that follows an
1122 empty field such as @code{int : 0;}.
1124 If @code{PCC_BITFIELD_TYPE_MATTERS} is true, it overrides this macro.
1127 @defmac STRUCTURE_SIZE_BOUNDARY
1128 Number of bits which any structure or union's size must be a multiple of.
1129 Each structure or union's size is rounded up to a multiple of this.
1131 If you do not define this macro, the default is the same as
1132 @code{BITS_PER_UNIT}.
1135 @defmac STRICT_ALIGNMENT
1136 Define this macro to be the value 1 if instructions will fail to work
1137 if given data not on the nominal alignment. If instructions will merely
1138 go slower in that case, define this macro as 0.
1141 @defmac PCC_BITFIELD_TYPE_MATTERS
1142 Define this if you wish to imitate the way many other C compilers handle
1143 alignment of bit-fields and the structures that contain them.
1145 The behavior is that the type written for a named bit-field (@code{int},
1146 @code{short}, or other integer type) imposes an alignment for the entire
1147 structure, as if the structure really did contain an ordinary field of
1148 that type. In addition, the bit-field is placed within the structure so
1149 that it would fit within such a field, not crossing a boundary for it.
1151 Thus, on most machines, a named bit-field whose type is written as
1152 @code{int} would not cross a four-byte boundary, and would force
1153 four-byte alignment for the whole structure. (The alignment used may
1154 not be four bytes; it is controlled by the other alignment parameters.)
1156 An unnamed bit-field will not affect the alignment of the containing
1159 If the macro is defined, its definition should be a C expression;
1160 a nonzero value for the expression enables this behavior.
1162 Note that if this macro is not defined, or its value is zero, some
1163 bit-fields may cross more than one alignment boundary. The compiler can
1164 support such references if there are @samp{insv}, @samp{extv}, and
1165 @samp{extzv} insns that can directly reference memory.
1167 The other known way of making bit-fields work is to define
1168 @code{STRUCTURE_SIZE_BOUNDARY} as large as @code{BIGGEST_ALIGNMENT}.
1169 Then every structure can be accessed with fullwords.
1171 Unless the machine has bit-field instructions or you define
1172 @code{STRUCTURE_SIZE_BOUNDARY} that way, you must define
1173 @code{PCC_BITFIELD_TYPE_MATTERS} to have a nonzero value.
1175 If your aim is to make GCC use the same conventions for laying out
1176 bit-fields as are used by another compiler, here is how to investigate
1177 what the other compiler does. Compile and run this program:
1196 printf ("Size of foo1 is %d\n",
1197 sizeof (struct foo1));
1198 printf ("Size of foo2 is %d\n",
1199 sizeof (struct foo2));
1204 If this prints 2 and 5, then the compiler's behavior is what you would
1205 get from @code{PCC_BITFIELD_TYPE_MATTERS}.
1208 @defmac BITFIELD_NBYTES_LIMITED
1209 Like @code{PCC_BITFIELD_TYPE_MATTERS} except that its effect is limited
1210 to aligning a bit-field within the structure.
1213 @hook TARGET_ALIGN_ANON_BITFIELD
1215 @hook TARGET_NARROW_VOLATILE_BITFIELD
1217 @hook TARGET_MEMBER_TYPE_FORCES_BLK
1219 @defmac ROUND_TYPE_ALIGN (@var{type}, @var{computed}, @var{specified})
1220 Define this macro as an expression for the alignment of a type (given
1221 by @var{type} as a tree node) if the alignment computed in the usual
1222 way is @var{computed} and the alignment explicitly specified was
1225 The default is to use @var{specified} if it is larger; otherwise, use
1226 the smaller of @var{computed} and @code{BIGGEST_ALIGNMENT}
1229 @defmac MAX_FIXED_MODE_SIZE
1230 An integer expression for the size in bits of the largest integer
1231 machine mode that should actually be used. All integer machine modes of
1232 this size or smaller can be used for structures and unions with the
1233 appropriate sizes. If this macro is undefined, @code{GET_MODE_BITSIZE
1234 (DImode)} is assumed.
1237 @defmac STACK_SAVEAREA_MODE (@var{save_level})
1238 If defined, an expression of type @code{machine_mode} that
1239 specifies the mode of the save area operand of a
1240 @code{save_stack_@var{level}} named pattern (@pxref{Standard Names}).
1241 @var{save_level} is one of @code{SAVE_BLOCK}, @code{SAVE_FUNCTION}, or
1242 @code{SAVE_NONLOCAL} and selects which of the three named patterns is
1243 having its mode specified.
1245 You need not define this macro if it always returns @code{Pmode}. You
1246 would most commonly define this macro if the
1247 @code{save_stack_@var{level}} patterns need to support both a 32- and a
1251 @defmac STACK_SIZE_MODE
1252 If defined, an expression of type @code{machine_mode} that
1253 specifies the mode of the size increment operand of an
1254 @code{allocate_stack} named pattern (@pxref{Standard Names}).
1256 You need not define this macro if it always returns @code{word_mode}.
1257 You would most commonly define this macro if the @code{allocate_stack}
1258 pattern needs to support both a 32- and a 64-bit mode.
1261 @hook TARGET_LIBGCC_CMP_RETURN_MODE
1263 @hook TARGET_LIBGCC_SHIFT_COUNT_MODE
1265 @hook TARGET_UNWIND_WORD_MODE
1267 @hook TARGET_MS_BITFIELD_LAYOUT_P
1269 @hook TARGET_DECIMAL_FLOAT_SUPPORTED_P
1271 @hook TARGET_FIXED_POINT_SUPPORTED_P
1273 @hook TARGET_EXPAND_TO_RTL_HOOK
1275 @hook TARGET_INSTANTIATE_DECLS
1277 @hook TARGET_MANGLE_TYPE
1280 @section Layout of Source Language Data Types
1282 These macros define the sizes and other characteristics of the standard
1283 basic data types used in programs being compiled. Unlike the macros in
1284 the previous section, these apply to specific features of C and related
1285 languages, rather than to fundamental aspects of storage layout.
1287 @defmac INT_TYPE_SIZE
1288 A C expression for the size in bits of the type @code{int} on the
1289 target machine. If you don't define this, the default is one word.
1292 @defmac SHORT_TYPE_SIZE
1293 A C expression for the size in bits of the type @code{short} on the
1294 target machine. If you don't define this, the default is half a word.
1295 (If this would be less than one storage unit, it is rounded up to one
1299 @defmac LONG_TYPE_SIZE
1300 A C expression for the size in bits of the type @code{long} on the
1301 target machine. If you don't define this, the default is one word.
1304 @defmac ADA_LONG_TYPE_SIZE
1305 On some machines, the size used for the Ada equivalent of the type
1306 @code{long} by a native Ada compiler differs from that used by C@. In
1307 that situation, define this macro to be a C expression to be used for
1308 the size of that type. If you don't define this, the default is the
1309 value of @code{LONG_TYPE_SIZE}.
1312 @defmac LONG_LONG_TYPE_SIZE
1313 A C expression for the size in bits of the type @code{long long} on the
1314 target machine. If you don't define this, the default is two
1315 words. If you want to support GNU Ada on your machine, the value of this
1316 macro must be at least 64.
1319 @defmac CHAR_TYPE_SIZE
1320 A C expression for the size in bits of the type @code{char} on the
1321 target machine. If you don't define this, the default is
1322 @code{BITS_PER_UNIT}.
1325 @defmac BOOL_TYPE_SIZE
1326 A C expression for the size in bits of the C++ type @code{bool} and
1327 C99 type @code{_Bool} on the target machine. If you don't define
1328 this, and you probably shouldn't, the default is @code{CHAR_TYPE_SIZE}.
1331 @defmac FLOAT_TYPE_SIZE
1332 A C expression for the size in bits of the type @code{float} on the
1333 target machine. If you don't define this, the default is one word.
1336 @defmac DOUBLE_TYPE_SIZE
1337 A C expression for the size in bits of the type @code{double} on the
1338 target machine. If you don't define this, the default is two
1342 @defmac LONG_DOUBLE_TYPE_SIZE
1343 A C expression for the size in bits of the type @code{long double} on
1344 the target machine. If you don't define this, the default is two
1348 @defmac SHORT_FRACT_TYPE_SIZE
1349 A C expression for the size in bits of the type @code{short _Fract} on
1350 the target machine. If you don't define this, the default is
1351 @code{BITS_PER_UNIT}.
1354 @defmac FRACT_TYPE_SIZE
1355 A C expression for the size in bits of the type @code{_Fract} on
1356 the target machine. If you don't define this, the default is
1357 @code{BITS_PER_UNIT * 2}.
1360 @defmac LONG_FRACT_TYPE_SIZE
1361 A C expression for the size in bits of the type @code{long _Fract} on
1362 the target machine. If you don't define this, the default is
1363 @code{BITS_PER_UNIT * 4}.
1366 @defmac LONG_LONG_FRACT_TYPE_SIZE
1367 A C expression for the size in bits of the type @code{long long _Fract} on
1368 the target machine. If you don't define this, the default is
1369 @code{BITS_PER_UNIT * 8}.
1372 @defmac SHORT_ACCUM_TYPE_SIZE
1373 A C expression for the size in bits of the type @code{short _Accum} on
1374 the target machine. If you don't define this, the default is
1375 @code{BITS_PER_UNIT * 2}.
1378 @defmac ACCUM_TYPE_SIZE
1379 A C expression for the size in bits of the type @code{_Accum} on
1380 the target machine. If you don't define this, the default is
1381 @code{BITS_PER_UNIT * 4}.
1384 @defmac LONG_ACCUM_TYPE_SIZE
1385 A C expression for the size in bits of the type @code{long _Accum} on
1386 the target machine. If you don't define this, the default is
1387 @code{BITS_PER_UNIT * 8}.
1390 @defmac LONG_LONG_ACCUM_TYPE_SIZE
1391 A C expression for the size in bits of the type @code{long long _Accum} on
1392 the target machine. If you don't define this, the default is
1393 @code{BITS_PER_UNIT * 16}.
1396 @defmac LIBGCC2_GNU_PREFIX
1397 This macro corresponds to the @code{TARGET_LIBFUNC_GNU_PREFIX} target
1398 hook and should be defined if that hook is overriden to be true. It
1399 causes function names in libgcc to be changed to use a @code{__gnu_}
1400 prefix for their name rather than the default @code{__}. A port which
1401 uses this macro should also arrange to use @file{t-gnu-prefix} in
1402 the libgcc @file{config.host}.
1405 @defmac TARGET_FLT_EVAL_METHOD
1406 A C expression for the value for @code{FLT_EVAL_METHOD} in @file{float.h},
1407 assuming, if applicable, that the floating-point control word is in its
1408 default state. If you do not define this macro the value of
1409 @code{FLT_EVAL_METHOD} will be zero.
1412 @defmac WIDEST_HARDWARE_FP_SIZE
1413 A C expression for the size in bits of the widest floating-point format
1414 supported by the hardware. If you define this macro, you must specify a
1415 value less than or equal to the value of @code{LONG_DOUBLE_TYPE_SIZE}.
1416 If you do not define this macro, the value of @code{LONG_DOUBLE_TYPE_SIZE}
1420 @defmac DEFAULT_SIGNED_CHAR
1421 An expression whose value is 1 or 0, according to whether the type
1422 @code{char} should be signed or unsigned by default. The user can
1423 always override this default with the options @option{-fsigned-char}
1424 and @option{-funsigned-char}.
1427 @hook TARGET_DEFAULT_SHORT_ENUMS
1430 A C expression for a string describing the name of the data type to use
1431 for size values. The typedef name @code{size_t} is defined using the
1432 contents of the string.
1434 The string can contain more than one keyword. If so, separate them with
1435 spaces, and write first any length keyword, then @code{unsigned} if
1436 appropriate, and finally @code{int}. The string must exactly match one
1437 of the data type names defined in the function
1438 @code{c_common_nodes_and_builtins} in the file @file{c-family/c-common.c}.
1439 You may not omit @code{int} or change the order---that would cause the
1440 compiler to crash on startup.
1442 If you don't define this macro, the default is @code{"long unsigned
1447 GCC defines internal types (@code{sizetype}, @code{ssizetype},
1448 @code{bitsizetype} and @code{sbitsizetype}) for expressions
1449 dealing with size. This macro is a C expression for a string describing
1450 the name of the data type from which the precision of @code{sizetype}
1453 The string has the same restrictions as @code{SIZE_TYPE} string.
1455 If you don't define this macro, the default is @code{SIZE_TYPE}.
1458 @defmac PTRDIFF_TYPE
1459 A C expression for a string describing the name of the data type to use
1460 for the result of subtracting two pointers. The typedef name
1461 @code{ptrdiff_t} is defined using the contents of the string. See
1462 @code{SIZE_TYPE} above for more information.
1464 If you don't define this macro, the default is @code{"long int"}.
1468 A C expression for a string describing the name of the data type to use
1469 for wide characters. The typedef name @code{wchar_t} is defined using
1470 the contents of the string. See @code{SIZE_TYPE} above for more
1473 If you don't define this macro, the default is @code{"int"}.
1476 @defmac WCHAR_TYPE_SIZE
1477 A C expression for the size in bits of the data type for wide
1478 characters. This is used in @code{cpp}, which cannot make use of
1483 A C expression for a string describing the name of the data type to
1484 use for wide characters passed to @code{printf} and returned from
1485 @code{getwc}. The typedef name @code{wint_t} is defined using the
1486 contents of the string. See @code{SIZE_TYPE} above for more
1489 If you don't define this macro, the default is @code{"unsigned int"}.
1493 A C expression for a string describing the name of the data type that
1494 can represent any value of any standard or extended signed integer type.
1495 The typedef name @code{intmax_t} is defined using the contents of the
1496 string. See @code{SIZE_TYPE} above for more information.
1498 If you don't define this macro, the default is the first of
1499 @code{"int"}, @code{"long int"}, or @code{"long long int"} that has as
1500 much precision as @code{long long int}.
1503 @defmac UINTMAX_TYPE
1504 A C expression for a string describing the name of the data type that
1505 can represent any value of any standard or extended unsigned integer
1506 type. The typedef name @code{uintmax_t} is defined using the contents
1507 of the string. See @code{SIZE_TYPE} above for more information.
1509 If you don't define this macro, the default is the first of
1510 @code{"unsigned int"}, @code{"long unsigned int"}, or @code{"long long
1511 unsigned int"} that has as much precision as @code{long long unsigned
1515 @defmac SIG_ATOMIC_TYPE
1521 @defmacx UINT16_TYPE
1522 @defmacx UINT32_TYPE
1523 @defmacx UINT64_TYPE
1524 @defmacx INT_LEAST8_TYPE
1525 @defmacx INT_LEAST16_TYPE
1526 @defmacx INT_LEAST32_TYPE
1527 @defmacx INT_LEAST64_TYPE
1528 @defmacx UINT_LEAST8_TYPE
1529 @defmacx UINT_LEAST16_TYPE
1530 @defmacx UINT_LEAST32_TYPE
1531 @defmacx UINT_LEAST64_TYPE
1532 @defmacx INT_FAST8_TYPE
1533 @defmacx INT_FAST16_TYPE
1534 @defmacx INT_FAST32_TYPE
1535 @defmacx INT_FAST64_TYPE
1536 @defmacx UINT_FAST8_TYPE
1537 @defmacx UINT_FAST16_TYPE
1538 @defmacx UINT_FAST32_TYPE
1539 @defmacx UINT_FAST64_TYPE
1540 @defmacx INTPTR_TYPE
1541 @defmacx UINTPTR_TYPE
1542 C expressions for the standard types @code{sig_atomic_t},
1543 @code{int8_t}, @code{int16_t}, @code{int32_t}, @code{int64_t},
1544 @code{uint8_t}, @code{uint16_t}, @code{uint32_t}, @code{uint64_t},
1545 @code{int_least8_t}, @code{int_least16_t}, @code{int_least32_t},
1546 @code{int_least64_t}, @code{uint_least8_t}, @code{uint_least16_t},
1547 @code{uint_least32_t}, @code{uint_least64_t}, @code{int_fast8_t},
1548 @code{int_fast16_t}, @code{int_fast32_t}, @code{int_fast64_t},
1549 @code{uint_fast8_t}, @code{uint_fast16_t}, @code{uint_fast32_t},
1550 @code{uint_fast64_t}, @code{intptr_t}, and @code{uintptr_t}. See
1551 @code{SIZE_TYPE} above for more information.
1553 If any of these macros evaluates to a null pointer, the corresponding
1554 type is not supported; if GCC is configured to provide
1555 @code{<stdint.h>} in such a case, the header provided may not conform
1556 to C99, depending on the type in question. The defaults for all of
1557 these macros are null pointers.
1560 @defmac TARGET_PTRMEMFUNC_VBIT_LOCATION
1561 The C++ compiler represents a pointer-to-member-function with a struct
1568 ptrdiff_t vtable_index;
1575 The C++ compiler must use one bit to indicate whether the function that
1576 will be called through a pointer-to-member-function is virtual.
1577 Normally, we assume that the low-order bit of a function pointer must
1578 always be zero. Then, by ensuring that the vtable_index is odd, we can
1579 distinguish which variant of the union is in use. But, on some
1580 platforms function pointers can be odd, and so this doesn't work. In
1581 that case, we use the low-order bit of the @code{delta} field, and shift
1582 the remainder of the @code{delta} field to the left.
1584 GCC will automatically make the right selection about where to store
1585 this bit using the @code{FUNCTION_BOUNDARY} setting for your platform.
1586 However, some platforms such as ARM/Thumb have @code{FUNCTION_BOUNDARY}
1587 set such that functions always start at even addresses, but the lowest
1588 bit of pointers to functions indicate whether the function at that
1589 address is in ARM or Thumb mode. If this is the case of your
1590 architecture, you should define this macro to
1591 @code{ptrmemfunc_vbit_in_delta}.
1593 In general, you should not have to define this macro. On architectures
1594 in which function addresses are always even, according to
1595 @code{FUNCTION_BOUNDARY}, GCC will automatically define this macro to
1596 @code{ptrmemfunc_vbit_in_pfn}.
1599 @defmac TARGET_VTABLE_USES_DESCRIPTORS
1600 Normally, the C++ compiler uses function pointers in vtables. This
1601 macro allows the target to change to use ``function descriptors''
1602 instead. Function descriptors are found on targets for whom a
1603 function pointer is actually a small data structure. Normally the
1604 data structure consists of the actual code address plus a data
1605 pointer to which the function's data is relative.
1607 If vtables are used, the value of this macro should be the number
1608 of words that the function descriptor occupies.
1611 @defmac TARGET_VTABLE_ENTRY_ALIGN
1612 By default, the vtable entries are void pointers, the so the alignment
1613 is the same as pointer alignment. The value of this macro specifies
1614 the alignment of the vtable entry in bits. It should be defined only
1615 when special alignment is necessary. */
1618 @defmac TARGET_VTABLE_DATA_ENTRY_DISTANCE
1619 There are a few non-descriptor entries in the vtable at offsets below
1620 zero. If these entries must be padded (say, to preserve the alignment
1621 specified by @code{TARGET_VTABLE_ENTRY_ALIGN}), set this to the number
1622 of words in each data entry.
1626 @section Register Usage
1627 @cindex register usage
1629 This section explains how to describe what registers the target machine
1630 has, and how (in general) they can be used.
1632 The description of which registers a specific instruction can use is
1633 done with register classes; see @ref{Register Classes}. For information
1634 on using registers to access a stack frame, see @ref{Frame Registers}.
1635 For passing values in registers, see @ref{Register Arguments}.
1636 For returning values in registers, see @ref{Scalar Return}.
1639 * Register Basics:: Number and kinds of registers.
1640 * Allocation Order:: Order in which registers are allocated.
1641 * Values in Registers:: What kinds of values each reg can hold.
1642 * Leaf Functions:: Renumbering registers for leaf functions.
1643 * Stack Registers:: Handling a register stack such as 80387.
1646 @node Register Basics
1647 @subsection Basic Characteristics of Registers
1649 @c prevent bad page break with this line
1650 Registers have various characteristics.
1652 @defmac FIRST_PSEUDO_REGISTER
1653 Number of hardware registers known to the compiler. They receive
1654 numbers 0 through @code{FIRST_PSEUDO_REGISTER-1}; thus, the first
1655 pseudo register's number really is assigned the number
1656 @code{FIRST_PSEUDO_REGISTER}.
1659 @defmac FIXED_REGISTERS
1660 @cindex fixed register
1661 An initializer that says which registers are used for fixed purposes
1662 all throughout the compiled code and are therefore not available for
1663 general allocation. These would include the stack pointer, the frame
1664 pointer (except on machines where that can be used as a general
1665 register when no frame pointer is needed), the program counter on
1666 machines where that is considered one of the addressable registers,
1667 and any other numbered register with a standard use.
1669 This information is expressed as a sequence of numbers, separated by
1670 commas and surrounded by braces. The @var{n}th number is 1 if
1671 register @var{n} is fixed, 0 otherwise.
1673 The table initialized from this macro, and the table initialized by
1674 the following one, may be overridden at run time either automatically,
1675 by the actions of the macro @code{CONDITIONAL_REGISTER_USAGE}, or by
1676 the user with the command options @option{-ffixed-@var{reg}},
1677 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}.
1680 @defmac CALL_USED_REGISTERS
1681 @cindex call-used register
1682 @cindex call-clobbered register
1683 @cindex call-saved register
1684 Like @code{FIXED_REGISTERS} but has 1 for each register that is
1685 clobbered (in general) by function calls as well as for fixed
1686 registers. This macro therefore identifies the registers that are not
1687 available for general allocation of values that must live across
1690 If a register has 0 in @code{CALL_USED_REGISTERS}, the compiler
1691 automatically saves it on function entry and restores it on function
1692 exit, if the register is used within the function.
1695 @defmac CALL_REALLY_USED_REGISTERS
1696 @cindex call-used register
1697 @cindex call-clobbered register
1698 @cindex call-saved register
1699 Like @code{CALL_USED_REGISTERS} except this macro doesn't require
1700 that the entire set of @code{FIXED_REGISTERS} be included.
1701 (@code{CALL_USED_REGISTERS} must be a superset of @code{FIXED_REGISTERS}).
1702 This macro is optional. If not specified, it defaults to the value
1703 of @code{CALL_USED_REGISTERS}.
1706 @defmac HARD_REGNO_CALL_PART_CLOBBERED (@var{regno}, @var{mode})
1707 @cindex call-used register
1708 @cindex call-clobbered register
1709 @cindex call-saved register
1710 A C expression that is nonzero if it is not permissible to store a
1711 value of mode @var{mode} in hard register number @var{regno} across a
1712 call without some part of it being clobbered. For most machines this
1713 macro need not be defined. It is only required for machines that do not
1714 preserve the entire contents of a register across a call.
1718 @findex call_used_regs
1721 @findex reg_class_contents
1722 @hook TARGET_CONDITIONAL_REGISTER_USAGE
1724 @defmac INCOMING_REGNO (@var{out})
1725 Define this macro if the target machine has register windows. This C
1726 expression returns the register number as seen by the called function
1727 corresponding to the register number @var{out} as seen by the calling
1728 function. Return @var{out} if register number @var{out} is not an
1732 @defmac OUTGOING_REGNO (@var{in})
1733 Define this macro if the target machine has register windows. This C
1734 expression returns the register number as seen by the calling function
1735 corresponding to the register number @var{in} as seen by the called
1736 function. Return @var{in} if register number @var{in} is not an inbound
1740 @defmac LOCAL_REGNO (@var{regno})
1741 Define this macro if the target machine has register windows. This C
1742 expression returns true if the register is call-saved but is in the
1743 register window. Unlike most call-saved registers, such registers
1744 need not be explicitly restored on function exit or during non-local
1749 If the program counter has a register number, define this as that
1750 register number. Otherwise, do not define it.
1753 @node Allocation Order
1754 @subsection Order of Allocation of Registers
1755 @cindex order of register allocation
1756 @cindex register allocation order
1758 @c prevent bad page break with this line
1759 Registers are allocated in order.
1761 @defmac REG_ALLOC_ORDER
1762 If defined, an initializer for a vector of integers, containing the
1763 numbers of hard registers in the order in which GCC should prefer
1764 to use them (from most preferred to least).
1766 If this macro is not defined, registers are used lowest numbered first
1767 (all else being equal).
1769 One use of this macro is on machines where the highest numbered
1770 registers must always be saved and the save-multiple-registers
1771 instruction supports only sequences of consecutive registers. On such
1772 machines, define @code{REG_ALLOC_ORDER} to be an initializer that lists
1773 the highest numbered allocable register first.
1776 @defmac ADJUST_REG_ALLOC_ORDER
1777 A C statement (sans semicolon) to choose the order in which to allocate
1778 hard registers for pseudo-registers local to a basic block.
1780 Store the desired register order in the array @code{reg_alloc_order}.
1781 Element 0 should be the register to allocate first; element 1, the next
1782 register; and so on.
1784 The macro body should not assume anything about the contents of
1785 @code{reg_alloc_order} before execution of the macro.
1787 On most machines, it is not necessary to define this macro.
1790 @defmac HONOR_REG_ALLOC_ORDER
1791 Normally, IRA tries to estimate the costs for saving a register in the
1792 prologue and restoring it in the epilogue. This discourages it from
1793 using call-saved registers. If a machine wants to ensure that IRA
1794 allocates registers in the order given by REG_ALLOC_ORDER even if some
1795 call-saved registers appear earlier than call-used ones, then define this
1796 macro as a C expression to nonzero. Default is 0.
1799 @defmac IRA_HARD_REGNO_ADD_COST_MULTIPLIER (@var{regno})
1800 In some case register allocation order is not enough for the
1801 Integrated Register Allocator (@acronym{IRA}) to generate a good code.
1802 If this macro is defined, it should return a floating point value
1803 based on @var{regno}. The cost of using @var{regno} for a pseudo will
1804 be increased by approximately the pseudo's usage frequency times the
1805 value returned by this macro. Not defining this macro is equivalent
1806 to having it always return @code{0.0}.
1808 On most machines, it is not necessary to define this macro.
1811 @node Values in Registers
1812 @subsection How Values Fit in Registers
1814 This section discusses the macros that describe which kinds of values
1815 (specifically, which machine modes) each register can hold, and how many
1816 consecutive registers are needed for a given mode.
1818 @defmac HARD_REGNO_NREGS (@var{regno}, @var{mode})
1819 A C expression for the number of consecutive hard registers, starting
1820 at register number @var{regno}, required to hold a value of mode
1821 @var{mode}. This macro must never return zero, even if a register
1822 cannot hold the requested mode - indicate that with HARD_REGNO_MODE_OK
1823 and/or CANNOT_CHANGE_MODE_CLASS instead.
1825 On a machine where all registers are exactly one word, a suitable
1826 definition of this macro is
1829 #define HARD_REGNO_NREGS(REGNO, MODE) \
1830 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) \
1835 @defmac HARD_REGNO_NREGS_HAS_PADDING (@var{regno}, @var{mode})
1836 A C expression that is nonzero if a value of mode @var{mode}, stored
1837 in memory, ends with padding that causes it to take up more space than
1838 in registers starting at register number @var{regno} (as determined by
1839 multiplying GCC's notion of the size of the register when containing
1840 this mode by the number of registers returned by
1841 @code{HARD_REGNO_NREGS}). By default this is zero.
1843 For example, if a floating-point value is stored in three 32-bit
1844 registers but takes up 128 bits in memory, then this would be
1847 This macros only needs to be defined if there are cases where
1848 @code{subreg_get_info}
1849 would otherwise wrongly determine that a @code{subreg} can be
1850 represented by an offset to the register number, when in fact such a
1851 @code{subreg} would contain some of the padding not stored in
1852 registers and so not be representable.
1855 @defmac HARD_REGNO_NREGS_WITH_PADDING (@var{regno}, @var{mode})
1856 For values of @var{regno} and @var{mode} for which
1857 @code{HARD_REGNO_NREGS_HAS_PADDING} returns nonzero, a C expression
1858 returning the greater number of registers required to hold the value
1859 including any padding. In the example above, the value would be four.
1862 @defmac REGMODE_NATURAL_SIZE (@var{mode})
1863 Define this macro if the natural size of registers that hold values
1864 of mode @var{mode} is not the word size. It is a C expression that
1865 should give the natural size in bytes for the specified mode. It is
1866 used by the register allocator to try to optimize its results. This
1867 happens for example on SPARC 64-bit where the natural size of
1868 floating-point registers is still 32-bit.
1871 @defmac HARD_REGNO_MODE_OK (@var{regno}, @var{mode})
1872 A C expression that is nonzero if it is permissible to store a value
1873 of mode @var{mode} in hard register number @var{regno} (or in several
1874 registers starting with that one). For a machine where all registers
1875 are equivalent, a suitable definition is
1878 #define HARD_REGNO_MODE_OK(REGNO, MODE) 1
1881 You need not include code to check for the numbers of fixed registers,
1882 because the allocation mechanism considers them to be always occupied.
1884 @cindex register pairs
1885 On some machines, double-precision values must be kept in even/odd
1886 register pairs. You can implement that by defining this macro to reject
1887 odd register numbers for such modes.
1889 The minimum requirement for a mode to be OK in a register is that the
1890 @samp{mov@var{mode}} instruction pattern support moves between the
1891 register and other hard register in the same class and that moving a
1892 value into the register and back out not alter it.
1894 Since the same instruction used to move @code{word_mode} will work for
1895 all narrower integer modes, it is not necessary on any machine for
1896 @code{HARD_REGNO_MODE_OK} to distinguish between these modes, provided
1897 you define patterns @samp{movhi}, etc., to take advantage of this. This
1898 is useful because of the interaction between @code{HARD_REGNO_MODE_OK}
1899 and @code{MODES_TIEABLE_P}; it is very desirable for all integer modes
1902 Many machines have special registers for floating point arithmetic.
1903 Often people assume that floating point machine modes are allowed only
1904 in floating point registers. This is not true. Any registers that
1905 can hold integers can safely @emph{hold} a floating point machine
1906 mode, whether or not floating arithmetic can be done on it in those
1907 registers. Integer move instructions can be used to move the values.
1909 On some machines, though, the converse is true: fixed-point machine
1910 modes may not go in floating registers. This is true if the floating
1911 registers normalize any value stored in them, because storing a
1912 non-floating value there would garble it. In this case,
1913 @code{HARD_REGNO_MODE_OK} should reject fixed-point machine modes in
1914 floating registers. But if the floating registers do not automatically
1915 normalize, if you can store any bit pattern in one and retrieve it
1916 unchanged without a trap, then any machine mode may go in a floating
1917 register, so you can define this macro to say so.
1919 The primary significance of special floating registers is rather that
1920 they are the registers acceptable in floating point arithmetic
1921 instructions. However, this is of no concern to
1922 @code{HARD_REGNO_MODE_OK}. You handle it by writing the proper
1923 constraints for those instructions.
1925 On some machines, the floating registers are especially slow to access,
1926 so that it is better to store a value in a stack frame than in such a
1927 register if floating point arithmetic is not being done. As long as the
1928 floating registers are not in class @code{GENERAL_REGS}, they will not
1929 be used unless some pattern's constraint asks for one.
1932 @defmac HARD_REGNO_RENAME_OK (@var{from}, @var{to})
1933 A C expression that is nonzero if it is OK to rename a hard register
1934 @var{from} to another hard register @var{to}.
1936 One common use of this macro is to prevent renaming of a register to
1937 another register that is not saved by a prologue in an interrupt
1940 The default is always nonzero.
1943 @defmac MODES_TIEABLE_P (@var{mode1}, @var{mode2})
1944 A C expression that is nonzero if a value of mode
1945 @var{mode1} is accessible in mode @var{mode2} without copying.
1947 If @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode1})} and
1948 @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode2})} are always the same for
1949 any @var{r}, then @code{MODES_TIEABLE_P (@var{mode1}, @var{mode2})}
1950 should be nonzero. If they differ for any @var{r}, you should define
1951 this macro to return zero unless some other mechanism ensures the
1952 accessibility of the value in a narrower mode.
1954 You should define this macro to return nonzero in as many cases as
1955 possible since doing so will allow GCC to perform better register
1959 @hook TARGET_HARD_REGNO_SCRATCH_OK
1961 @defmac AVOID_CCMODE_COPIES
1962 Define this macro if the compiler should avoid copies to/from @code{CCmode}
1963 registers. You should only define this macro if support for copying to/from
1964 @code{CCmode} is incomplete.
1967 @node Leaf Functions
1968 @subsection Handling Leaf Functions
1970 @cindex leaf functions
1971 @cindex functions, leaf
1972 On some machines, a leaf function (i.e., one which makes no calls) can run
1973 more efficiently if it does not make its own register window. Often this
1974 means it is required to receive its arguments in the registers where they
1975 are passed by the caller, instead of the registers where they would
1978 The special treatment for leaf functions generally applies only when
1979 other conditions are met; for example, often they may use only those
1980 registers for its own variables and temporaries. We use the term ``leaf
1981 function'' to mean a function that is suitable for this special
1982 handling, so that functions with no calls are not necessarily ``leaf
1985 GCC assigns register numbers before it knows whether the function is
1986 suitable for leaf function treatment. So it needs to renumber the
1987 registers in order to output a leaf function. The following macros
1990 @defmac LEAF_REGISTERS
1991 Name of a char vector, indexed by hard register number, which
1992 contains 1 for a register that is allowable in a candidate for leaf
1995 If leaf function treatment involves renumbering the registers, then the
1996 registers marked here should be the ones before renumbering---those that
1997 GCC would ordinarily allocate. The registers which will actually be
1998 used in the assembler code, after renumbering, should not be marked with 1
2001 Define this macro only if the target machine offers a way to optimize
2002 the treatment of leaf functions.
2005 @defmac LEAF_REG_REMAP (@var{regno})
2006 A C expression whose value is the register number to which @var{regno}
2007 should be renumbered, when a function is treated as a leaf function.
2009 If @var{regno} is a register number which should not appear in a leaf
2010 function before renumbering, then the expression should yield @minus{}1, which
2011 will cause the compiler to abort.
2013 Define this macro only if the target machine offers a way to optimize the
2014 treatment of leaf functions, and registers need to be renumbered to do
2018 @findex current_function_is_leaf
2019 @findex current_function_uses_only_leaf_regs
2020 @code{TARGET_ASM_FUNCTION_PROLOGUE} and
2021 @code{TARGET_ASM_FUNCTION_EPILOGUE} must usually treat leaf functions
2022 specially. They can test the C variable @code{current_function_is_leaf}
2023 which is nonzero for leaf functions. @code{current_function_is_leaf} is
2024 set prior to local register allocation and is valid for the remaining
2025 compiler passes. They can also test the C variable
2026 @code{current_function_uses_only_leaf_regs} which is nonzero for leaf
2027 functions which only use leaf registers.
2028 @code{current_function_uses_only_leaf_regs} is valid after all passes
2029 that modify the instructions have been run and is only useful if
2030 @code{LEAF_REGISTERS} is defined.
2031 @c changed this to fix overfull. ALSO: why the "it" at the beginning
2032 @c of the next paragraph?! --mew 2feb93
2034 @node Stack Registers
2035 @subsection Registers That Form a Stack
2037 There are special features to handle computers where some of the
2038 ``registers'' form a stack. Stack registers are normally written by
2039 pushing onto the stack, and are numbered relative to the top of the
2042 Currently, GCC can only handle one group of stack-like registers, and
2043 they must be consecutively numbered. Furthermore, the existing
2044 support for stack-like registers is specific to the 80387 floating
2045 point coprocessor. If you have a new architecture that uses
2046 stack-like registers, you will need to do substantial work on
2047 @file{reg-stack.c} and write your machine description to cooperate
2048 with it, as well as defining these macros.
2051 Define this if the machine has any stack-like registers.
2054 @defmac STACK_REG_COVER_CLASS
2055 This is a cover class containing the stack registers. Define this if
2056 the machine has any stack-like registers.
2059 @defmac FIRST_STACK_REG
2060 The number of the first stack-like register. This one is the top
2064 @defmac LAST_STACK_REG
2065 The number of the last stack-like register. This one is the bottom of
2069 @node Register Classes
2070 @section Register Classes
2071 @cindex register class definitions
2072 @cindex class definitions, register
2074 On many machines, the numbered registers are not all equivalent.
2075 For example, certain registers may not be allowed for indexed addressing;
2076 certain registers may not be allowed in some instructions. These machine
2077 restrictions are described to the compiler using @dfn{register classes}.
2079 You define a number of register classes, giving each one a name and saying
2080 which of the registers belong to it. Then you can specify register classes
2081 that are allowed as operands to particular instruction patterns.
2085 In general, each register will belong to several classes. In fact, one
2086 class must be named @code{ALL_REGS} and contain all the registers. Another
2087 class must be named @code{NO_REGS} and contain no registers. Often the
2088 union of two classes will be another class; however, this is not required.
2090 @findex GENERAL_REGS
2091 One of the classes must be named @code{GENERAL_REGS}. There is nothing
2092 terribly special about the name, but the operand constraint letters
2093 @samp{r} and @samp{g} specify this class. If @code{GENERAL_REGS} is
2094 the same as @code{ALL_REGS}, just define it as a macro which expands
2097 Order the classes so that if class @var{x} is contained in class @var{y}
2098 then @var{x} has a lower class number than @var{y}.
2100 The way classes other than @code{GENERAL_REGS} are specified in operand
2101 constraints is through machine-dependent operand constraint letters.
2102 You can define such letters to correspond to various classes, then use
2103 them in operand constraints.
2105 You must define the narrowest register classes for allocatable
2106 registers, so that each class either has no subclasses, or that for
2107 some mode, the move cost between registers within the class is
2108 cheaper than moving a register in the class to or from memory
2111 You should define a class for the union of two classes whenever some
2112 instruction allows both classes. For example, if an instruction allows
2113 either a floating point (coprocessor) register or a general register for a
2114 certain operand, you should define a class @code{FLOAT_OR_GENERAL_REGS}
2115 which includes both of them. Otherwise you will get suboptimal code,
2116 or even internal compiler errors when reload cannot find a register in the
2117 class computed via @code{reg_class_subunion}.
2119 You must also specify certain redundant information about the register
2120 classes: for each class, which classes contain it and which ones are
2121 contained in it; for each pair of classes, the largest class contained
2124 When a value occupying several consecutive registers is expected in a
2125 certain class, all the registers used must belong to that class.
2126 Therefore, register classes cannot be used to enforce a requirement for
2127 a register pair to start with an even-numbered register. The way to
2128 specify this requirement is with @code{HARD_REGNO_MODE_OK}.
2130 Register classes used for input-operands of bitwise-and or shift
2131 instructions have a special requirement: each such class must have, for
2132 each fixed-point machine mode, a subclass whose registers can transfer that
2133 mode to or from memory. For example, on some machines, the operations for
2134 single-byte values (@code{QImode}) are limited to certain registers. When
2135 this is so, each register class that is used in a bitwise-and or shift
2136 instruction must have a subclass consisting of registers from which
2137 single-byte values can be loaded or stored. This is so that
2138 @code{PREFERRED_RELOAD_CLASS} can always have a possible value to return.
2140 @deftp {Data type} {enum reg_class}
2141 An enumerated type that must be defined with all the register class names
2142 as enumerated values. @code{NO_REGS} must be first. @code{ALL_REGS}
2143 must be the last register class, followed by one more enumerated value,
2144 @code{LIM_REG_CLASSES}, which is not a register class but rather
2145 tells how many classes there are.
2147 Each register class has a number, which is the value of casting
2148 the class name to type @code{int}. The number serves as an index
2149 in many of the tables described below.
2152 @defmac N_REG_CLASSES
2153 The number of distinct register classes, defined as follows:
2156 #define N_REG_CLASSES (int) LIM_REG_CLASSES
2160 @defmac REG_CLASS_NAMES
2161 An initializer containing the names of the register classes as C string
2162 constants. These names are used in writing some of the debugging dumps.
2165 @defmac REG_CLASS_CONTENTS
2166 An initializer containing the contents of the register classes, as integers
2167 which are bit masks. The @var{n}th integer specifies the contents of class
2168 @var{n}. The way the integer @var{mask} is interpreted is that
2169 register @var{r} is in the class if @code{@var{mask} & (1 << @var{r})} is 1.
2171 When the machine has more than 32 registers, an integer does not suffice.
2172 Then the integers are replaced by sub-initializers, braced groupings containing
2173 several integers. Each sub-initializer must be suitable as an initializer
2174 for the type @code{HARD_REG_SET} which is defined in @file{hard-reg-set.h}.
2175 In this situation, the first integer in each sub-initializer corresponds to
2176 registers 0 through 31, the second integer to registers 32 through 63, and
2180 @defmac REGNO_REG_CLASS (@var{regno})
2181 A C expression whose value is a register class containing hard register
2182 @var{regno}. In general there is more than one such class; choose a class
2183 which is @dfn{minimal}, meaning that no smaller class also contains the
2187 @defmac BASE_REG_CLASS
2188 A macro whose definition is the name of the class to which a valid
2189 base register must belong. A base register is one used in an address
2190 which is the register value plus a displacement.
2193 @defmac MODE_BASE_REG_CLASS (@var{mode})
2194 This is a variation of the @code{BASE_REG_CLASS} macro which allows
2195 the selection of a base register in a mode dependent manner. If
2196 @var{mode} is VOIDmode then it should return the same value as
2197 @code{BASE_REG_CLASS}.
2200 @defmac MODE_BASE_REG_REG_CLASS (@var{mode})
2201 A C expression whose value is the register class to which a valid
2202 base register must belong in order to be used in a base plus index
2203 register address. You should define this macro if base plus index
2204 addresses have different requirements than other base register uses.
2207 @defmac MODE_CODE_BASE_REG_CLASS (@var{mode}, @var{address_space}, @var{outer_code}, @var{index_code})
2208 A C expression whose value is the register class to which a valid
2209 base register for a memory reference in mode @var{mode} to address
2210 space @var{address_space} must belong. @var{outer_code} and @var{index_code}
2211 define the context in which the base register occurs. @var{outer_code} is
2212 the code of the immediately enclosing expression (@code{MEM} for the top level
2213 of an address, @code{ADDRESS} for something that occurs in an
2214 @code{address_operand}). @var{index_code} is the code of the corresponding
2215 index expression if @var{outer_code} is @code{PLUS}; @code{SCRATCH} otherwise.
2218 @defmac INDEX_REG_CLASS
2219 A macro whose definition is the name of the class to which a valid
2220 index register must belong. An index register is one used in an
2221 address where its value is either multiplied by a scale factor or
2222 added to another register (as well as added to a displacement).
2225 @defmac REGNO_OK_FOR_BASE_P (@var{num})
2226 A C expression which is nonzero if register number @var{num} is
2227 suitable for use as a base register in operand addresses.
2230 @defmac REGNO_MODE_OK_FOR_BASE_P (@var{num}, @var{mode})
2231 A C expression that is just like @code{REGNO_OK_FOR_BASE_P}, except that
2232 that expression may examine the mode of the memory reference in
2233 @var{mode}. You should define this macro if the mode of the memory
2234 reference affects whether a register may be used as a base register. If
2235 you define this macro, the compiler will use it instead of
2236 @code{REGNO_OK_FOR_BASE_P}. The mode may be @code{VOIDmode} for
2237 addresses that appear outside a @code{MEM}, i.e., as an
2238 @code{address_operand}.
2241 @defmac REGNO_MODE_OK_FOR_REG_BASE_P (@var{num}, @var{mode})
2242 A C expression which is nonzero if register number @var{num} is suitable for
2243 use as a base register in base plus index operand addresses, accessing
2244 memory in mode @var{mode}. It may be either a suitable hard register or a
2245 pseudo register that has been allocated such a hard register. You should
2246 define this macro if base plus index addresses have different requirements
2247 than other base register uses.
2249 Use of this macro is deprecated; please use the more general
2250 @code{REGNO_MODE_CODE_OK_FOR_BASE_P}.
2253 @defmac REGNO_MODE_CODE_OK_FOR_BASE_P (@var{num}, @var{mode}, @var{address_space}, @var{outer_code}, @var{index_code})
2254 A C expression which is nonzero if register number @var{num} is
2255 suitable for use as a base register in operand addresses, accessing
2256 memory in mode @var{mode} in address space @var{address_space}.
2257 This is similar to @code{REGNO_MODE_OK_FOR_BASE_P}, except
2258 that that expression may examine the context in which the register
2259 appears in the memory reference. @var{outer_code} is the code of the
2260 immediately enclosing expression (@code{MEM} if at the top level of the
2261 address, @code{ADDRESS} for something that occurs in an
2262 @code{address_operand}). @var{index_code} is the code of the
2263 corresponding index expression if @var{outer_code} is @code{PLUS};
2264 @code{SCRATCH} otherwise. The mode may be @code{VOIDmode} for addresses
2265 that appear outside a @code{MEM}, i.e., as an @code{address_operand}.
2268 @defmac REGNO_OK_FOR_INDEX_P (@var{num})
2269 A C expression which is nonzero if register number @var{num} is
2270 suitable for use as an index register in operand addresses. It may be
2271 either a suitable hard register or a pseudo register that has been
2272 allocated such a hard register.
2274 The difference between an index register and a base register is that
2275 the index register may be scaled. If an address involves the sum of
2276 two registers, neither one of them scaled, then either one may be
2277 labeled the ``base'' and the other the ``index''; but whichever
2278 labeling is used must fit the machine's constraints of which registers
2279 may serve in each capacity. The compiler will try both labelings,
2280 looking for one that is valid, and will reload one or both registers
2281 only if neither labeling works.
2284 @hook TARGET_PREFERRED_RENAME_CLASS
2286 @hook TARGET_PREFERRED_RELOAD_CLASS
2288 @defmac PREFERRED_RELOAD_CLASS (@var{x}, @var{class})
2289 A C expression that places additional restrictions on the register class
2290 to use when it is necessary to copy value @var{x} into a register in class
2291 @var{class}. The value is a register class; perhaps @var{class}, or perhaps
2292 another, smaller class. On many machines, the following definition is
2296 #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS
2299 Sometimes returning a more restrictive class makes better code. For
2300 example, on the 68000, when @var{x} is an integer constant that is in range
2301 for a @samp{moveq} instruction, the value of this macro is always
2302 @code{DATA_REGS} as long as @var{class} includes the data registers.
2303 Requiring a data register guarantees that a @samp{moveq} will be used.
2305 One case where @code{PREFERRED_RELOAD_CLASS} must not return
2306 @var{class} is if @var{x} is a legitimate constant which cannot be
2307 loaded into some register class. By returning @code{NO_REGS} you can
2308 force @var{x} into a memory location. For example, rs6000 can load
2309 immediate values into general-purpose registers, but does not have an
2310 instruction for loading an immediate value into a floating-point
2311 register, so @code{PREFERRED_RELOAD_CLASS} returns @code{NO_REGS} when
2312 @var{x} is a floating-point constant. If the constant can't be loaded
2313 into any kind of register, code generation will be better if
2314 @code{TARGET_LEGITIMATE_CONSTANT_P} makes the constant illegitimate instead
2315 of using @code{TARGET_PREFERRED_RELOAD_CLASS}.
2317 If an insn has pseudos in it after register allocation, reload will go
2318 through the alternatives and call repeatedly @code{PREFERRED_RELOAD_CLASS}
2319 to find the best one. Returning @code{NO_REGS}, in this case, makes
2320 reload add a @code{!} in front of the constraint: the x86 back-end uses
2321 this feature to discourage usage of 387 registers when math is done in
2322 the SSE registers (and vice versa).
2325 @hook TARGET_PREFERRED_OUTPUT_RELOAD_CLASS
2327 @defmac LIMIT_RELOAD_CLASS (@var{mode}, @var{class})
2328 A C expression that places additional restrictions on the register class
2329 to use when it is necessary to be able to hold a value of mode
2330 @var{mode} in a reload register for which class @var{class} would
2333 Unlike @code{PREFERRED_RELOAD_CLASS}, this macro should be used when
2334 there are certain modes that simply can't go in certain reload classes.
2336 The value is a register class; perhaps @var{class}, or perhaps another,
2339 Don't define this macro unless the target machine has limitations which
2340 require the macro to do something nontrivial.
2343 @hook TARGET_SECONDARY_RELOAD
2345 @defmac SECONDARY_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2346 @defmacx SECONDARY_INPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2347 @defmacx SECONDARY_OUTPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2348 These macros are obsolete, new ports should use the target hook
2349 @code{TARGET_SECONDARY_RELOAD} instead.
2351 These are obsolete macros, replaced by the @code{TARGET_SECONDARY_RELOAD}
2352 target hook. Older ports still define these macros to indicate to the
2353 reload phase that it may
2354 need to allocate at least one register for a reload in addition to the
2355 register to contain the data. Specifically, if copying @var{x} to a
2356 register @var{class} in @var{mode} requires an intermediate register,
2357 you were supposed to define @code{SECONDARY_INPUT_RELOAD_CLASS} to return the
2358 largest register class all of whose registers can be used as
2359 intermediate registers or scratch registers.
2361 If copying a register @var{class} in @var{mode} to @var{x} requires an
2362 intermediate or scratch register, @code{SECONDARY_OUTPUT_RELOAD_CLASS}
2363 was supposed to be defined be defined to return the largest register
2364 class required. If the
2365 requirements for input and output reloads were the same, the macro
2366 @code{SECONDARY_RELOAD_CLASS} should have been used instead of defining both
2369 The values returned by these macros are often @code{GENERAL_REGS}.
2370 Return @code{NO_REGS} if no spare register is needed; i.e., if @var{x}
2371 can be directly copied to or from a register of @var{class} in
2372 @var{mode} without requiring a scratch register. Do not define this
2373 macro if it would always return @code{NO_REGS}.
2375 If a scratch register is required (either with or without an
2376 intermediate register), you were supposed to define patterns for
2377 @samp{reload_in@var{m}} or @samp{reload_out@var{m}}, as required
2378 (@pxref{Standard Names}. These patterns, which were normally
2379 implemented with a @code{define_expand}, should be similar to the
2380 @samp{mov@var{m}} patterns, except that operand 2 is the scratch
2383 These patterns need constraints for the reload register and scratch
2385 contain a single register class. If the original reload register (whose
2386 class is @var{class}) can meet the constraint given in the pattern, the
2387 value returned by these macros is used for the class of the scratch
2388 register. Otherwise, two additional reload registers are required.
2389 Their classes are obtained from the constraints in the insn pattern.
2391 @var{x} might be a pseudo-register or a @code{subreg} of a
2392 pseudo-register, which could either be in a hard register or in memory.
2393 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2394 in memory and the hard register number if it is in a register.
2396 These macros should not be used in the case where a particular class of
2397 registers can only be copied to memory and not to another class of
2398 registers. In that case, secondary reload registers are not needed and
2399 would not be helpful. Instead, a stack location must be used to perform
2400 the copy and the @code{mov@var{m}} pattern should use memory as an
2401 intermediate storage. This case often occurs between floating-point and
2405 @defmac SECONDARY_MEMORY_NEEDED (@var{class1}, @var{class2}, @var{m})
2406 Certain machines have the property that some registers cannot be copied
2407 to some other registers without using memory. Define this macro on
2408 those machines to be a C expression that is nonzero if objects of mode
2409 @var{m} in registers of @var{class1} can only be copied to registers of
2410 class @var{class2} by storing a register of @var{class1} into memory
2411 and loading that memory location into a register of @var{class2}.
2413 Do not define this macro if its value would always be zero.
2416 @defmac SECONDARY_MEMORY_NEEDED_RTX (@var{mode})
2417 Normally when @code{SECONDARY_MEMORY_NEEDED} is defined, the compiler
2418 allocates a stack slot for a memory location needed for register copies.
2419 If this macro is defined, the compiler instead uses the memory location
2420 defined by this macro.
2422 Do not define this macro if you do not define
2423 @code{SECONDARY_MEMORY_NEEDED}.
2426 @defmac SECONDARY_MEMORY_NEEDED_MODE (@var{mode})
2427 When the compiler needs a secondary memory location to copy between two
2428 registers of mode @var{mode}, it normally allocates sufficient memory to
2429 hold a quantity of @code{BITS_PER_WORD} bits and performs the store and
2430 load operations in a mode that many bits wide and whose class is the
2431 same as that of @var{mode}.
2433 This is right thing to do on most machines because it ensures that all
2434 bits of the register are copied and prevents accesses to the registers
2435 in a narrower mode, which some machines prohibit for floating-point
2438 However, this default behavior is not correct on some machines, such as
2439 the DEC Alpha, that store short integers in floating-point registers
2440 differently than in integer registers. On those machines, the default
2441 widening will not work correctly and you must define this macro to
2442 suppress that widening in some cases. See the file @file{alpha.h} for
2445 Do not define this macro if you do not define
2446 @code{SECONDARY_MEMORY_NEEDED} or if widening @var{mode} to a mode that
2447 is @code{BITS_PER_WORD} bits wide is correct for your machine.
2450 @hook TARGET_CLASS_LIKELY_SPILLED_P
2452 @hook TARGET_CLASS_MAX_NREGS
2454 @defmac CLASS_MAX_NREGS (@var{class}, @var{mode})
2455 A C expression for the maximum number of consecutive registers
2456 of class @var{class} needed to hold a value of mode @var{mode}.
2458 This is closely related to the macro @code{HARD_REGNO_NREGS}. In fact,
2459 the value of the macro @code{CLASS_MAX_NREGS (@var{class}, @var{mode})}
2460 should be the maximum value of @code{HARD_REGNO_NREGS (@var{regno},
2461 @var{mode})} for all @var{regno} values in the class @var{class}.
2463 This macro helps control the handling of multiple-word values
2467 @defmac CANNOT_CHANGE_MODE_CLASS (@var{from}, @var{to}, @var{class})
2468 If defined, a C expression that returns nonzero for a @var{class} for which
2469 a change from mode @var{from} to mode @var{to} is invalid.
2471 For example, loading 32-bit integer or floating-point objects into
2472 floating-point registers on Alpha extends them to 64 bits.
2473 Therefore loading a 64-bit object and then storing it as a 32-bit object
2474 does not store the low-order 32 bits, as would be the case for a normal
2475 register. Therefore, @file{alpha.h} defines @code{CANNOT_CHANGE_MODE_CLASS}
2479 #define CANNOT_CHANGE_MODE_CLASS(FROM, TO, CLASS) \
2480 (GET_MODE_SIZE (FROM) != GET_MODE_SIZE (TO) \
2481 ? reg_classes_intersect_p (FLOAT_REGS, (CLASS)) : 0)
2484 Even if storing from a register in mode @var{to} would be valid,
2485 if both @var{from} and @code{raw_reg_mode} for @var{class} are wider
2486 than @code{word_mode}, then we must prevent @var{to} narrowing the
2487 mode. This happens when the middle-end assumes that it can load
2488 or store pieces of an @var{N}-word pseudo, and that the pseudo will
2489 eventually be allocated to @var{N} @code{word_mode} hard registers.
2490 Failure to prevent this kind of mode change will result in the
2491 entire @code{raw_reg_mode} being modified instead of the partial
2492 value that the middle-end intended.
2496 @hook TARGET_IRA_CHANGE_PSEUDO_ALLOCNO_CLASS
2500 @hook TARGET_REGISTER_PRIORITY
2502 @hook TARGET_REGISTER_USAGE_LEVELING_P
2504 @hook TARGET_DIFFERENT_ADDR_DISPLACEMENT_P
2506 @hook TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P
2508 @hook TARGET_LEGITIMIZE_ADDRESS_DISPLACEMENT
2510 @hook TARGET_SPILL_CLASS
2512 @hook TARGET_ADDITIONAL_ALLOCNO_CLASS_P
2514 @hook TARGET_CSTORE_MODE
2516 @hook TARGET_COMPUTE_PRESSURE_CLASSES
2518 @node Stack and Calling
2519 @section Stack Layout and Calling Conventions
2520 @cindex calling conventions
2522 @c prevent bad page break with this line
2523 This describes the stack layout and calling conventions.
2527 * Exception Handling::
2532 * Register Arguments::
2534 * Aggregate Return::
2539 * Shrink-wrapping separate components::
2540 * Stack Smashing Protection::
2541 * Miscellaneous Register Hooks::
2545 @subsection Basic Stack Layout
2546 @cindex stack frame layout
2547 @cindex frame layout
2549 @c prevent bad page break with this line
2550 Here is the basic stack layout.
2552 @defmac STACK_GROWS_DOWNWARD
2553 Define this macro to be true if pushing a word onto the stack moves the stack
2554 pointer to a smaller address, and false otherwise.
2557 @defmac STACK_PUSH_CODE
2558 This macro defines the operation used when something is pushed
2559 on the stack. In RTL, a push operation will be
2560 @code{(set (mem (STACK_PUSH_CODE (reg sp))) @dots{})}
2562 The choices are @code{PRE_DEC}, @code{POST_DEC}, @code{PRE_INC},
2563 and @code{POST_INC}. Which of these is correct depends on
2564 the stack direction and on whether the stack pointer points
2565 to the last item on the stack or whether it points to the
2566 space for the next item on the stack.
2568 The default is @code{PRE_DEC} when @code{STACK_GROWS_DOWNWARD} is
2569 true, which is almost always right, and @code{PRE_INC} otherwise,
2570 which is often wrong.
2573 @defmac FRAME_GROWS_DOWNWARD
2574 Define this macro to nonzero value if the addresses of local variable slots
2575 are at negative offsets from the frame pointer.
2578 @defmac ARGS_GROW_DOWNWARD
2579 Define this macro if successive arguments to a function occupy decreasing
2580 addresses on the stack.
2583 @defmac STARTING_FRAME_OFFSET
2584 Offset from the frame pointer to the first local variable slot to be allocated.
2586 If @code{FRAME_GROWS_DOWNWARD}, find the next slot's offset by
2587 subtracting the first slot's length from @code{STARTING_FRAME_OFFSET}.
2588 Otherwise, it is found by adding the length of the first slot to the
2589 value @code{STARTING_FRAME_OFFSET}.
2590 @c i'm not sure if the above is still correct.. had to change it to get
2591 @c rid of an overfull. --mew 2feb93
2594 @defmac STACK_ALIGNMENT_NEEDED
2595 Define to zero to disable final alignment of the stack during reload.
2596 The nonzero default for this macro is suitable for most ports.
2598 On ports where @code{STARTING_FRAME_OFFSET} is nonzero or where there
2599 is a register save block following the local block that doesn't require
2600 alignment to @code{STACK_BOUNDARY}, it may be beneficial to disable
2601 stack alignment and do it in the backend.
2604 @defmac STACK_POINTER_OFFSET
2605 Offset from the stack pointer register to the first location at which
2606 outgoing arguments are placed. If not specified, the default value of
2607 zero is used. This is the proper value for most machines.
2609 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
2610 the first location at which outgoing arguments are placed.
2613 @defmac FIRST_PARM_OFFSET (@var{fundecl})
2614 Offset from the argument pointer register to the first argument's
2615 address. On some machines it may depend on the data type of the
2618 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
2619 the first argument's address.
2622 @defmac STACK_DYNAMIC_OFFSET (@var{fundecl})
2623 Offset from the stack pointer register to an item dynamically allocated
2624 on the stack, e.g., by @code{alloca}.
2626 The default value for this macro is @code{STACK_POINTER_OFFSET} plus the
2627 length of the outgoing arguments. The default is correct for most
2628 machines. See @file{function.c} for details.
2631 @defmac INITIAL_FRAME_ADDRESS_RTX
2632 A C expression whose value is RTL representing the address of the initial
2633 stack frame. This address is passed to @code{RETURN_ADDR_RTX} and
2634 @code{DYNAMIC_CHAIN_ADDRESS}. If you don't define this macro, a reasonable
2635 default value will be used. Define this macro in order to make frame pointer
2636 elimination work in the presence of @code{__builtin_frame_address (count)} and
2637 @code{__builtin_return_address (count)} for @code{count} not equal to zero.
2640 @defmac DYNAMIC_CHAIN_ADDRESS (@var{frameaddr})
2641 A C expression whose value is RTL representing the address in a stack
2642 frame where the pointer to the caller's frame is stored. Assume that
2643 @var{frameaddr} is an RTL expression for the address of the stack frame
2646 If you don't define this macro, the default is to return the value
2647 of @var{frameaddr}---that is, the stack frame address is also the
2648 address of the stack word that points to the previous frame.
2651 @defmac SETUP_FRAME_ADDRESSES
2652 A C expression that produces the machine-specific code to
2653 setup the stack so that arbitrary frames can be accessed. For example,
2654 on the SPARC, we must flush all of the register windows to the stack
2655 before we can access arbitrary stack frames. You will seldom need to
2656 define this macro. The default is to do nothing.
2659 @hook TARGET_BUILTIN_SETJMP_FRAME_VALUE
2661 @defmac FRAME_ADDR_RTX (@var{frameaddr})
2662 A C expression whose value is RTL representing the value of the frame
2663 address for the current frame. @var{frameaddr} is the frame pointer
2664 of the current frame. This is used for __builtin_frame_address.
2665 You need only define this macro if the frame address is not the same
2666 as the frame pointer. Most machines do not need to define it.
2669 @defmac RETURN_ADDR_RTX (@var{count}, @var{frameaddr})
2670 A C expression whose value is RTL representing the value of the return
2671 address for the frame @var{count} steps up from the current frame, after
2672 the prologue. @var{frameaddr} is the frame pointer of the @var{count}
2673 frame, or the frame pointer of the @var{count} @minus{} 1 frame if
2674 @code{RETURN_ADDR_IN_PREVIOUS_FRAME} is nonzero.
2676 The value of the expression must always be the correct address when
2677 @var{count} is zero, but may be @code{NULL_RTX} if there is no way to
2678 determine the return address of other frames.
2681 @defmac RETURN_ADDR_IN_PREVIOUS_FRAME
2682 Define this macro to nonzero value if the return address of a particular
2683 stack frame is accessed from the frame pointer of the previous stack
2684 frame. The zero default for this macro is suitable for most ports.
2687 @defmac INCOMING_RETURN_ADDR_RTX
2688 A C expression whose value is RTL representing the location of the
2689 incoming return address at the beginning of any function, before the
2690 prologue. This RTL is either a @code{REG}, indicating that the return
2691 value is saved in @samp{REG}, or a @code{MEM} representing a location in
2694 You only need to define this macro if you want to support call frame
2695 debugging information like that provided by DWARF 2.
2697 If this RTL is a @code{REG}, you should also define
2698 @code{DWARF_FRAME_RETURN_COLUMN} to @code{DWARF_FRAME_REGNUM (REGNO)}.
2701 @defmac DWARF_ALT_FRAME_RETURN_COLUMN
2702 A C expression whose value is an integer giving a DWARF 2 column
2703 number that may be used as an alternative return column. The column
2704 must not correspond to any gcc hard register (that is, it must not
2705 be in the range of @code{DWARF_FRAME_REGNUM}).
2707 This macro can be useful if @code{DWARF_FRAME_RETURN_COLUMN} is set to a
2708 general register, but an alternative column needs to be used for signal
2709 frames. Some targets have also used different frame return columns
2713 @defmac DWARF_ZERO_REG
2714 A C expression whose value is an integer giving a DWARF 2 register
2715 number that is considered to always have the value zero. This should
2716 only be defined if the target has an architected zero register, and
2717 someone decided it was a good idea to use that register number to
2718 terminate the stack backtrace. New ports should avoid this.
2721 @hook TARGET_DWARF_HANDLE_FRAME_UNSPEC
2723 @defmac INCOMING_FRAME_SP_OFFSET
2724 A C expression whose value is an integer giving the offset, in bytes,
2725 from the value of the stack pointer register to the top of the stack
2726 frame at the beginning of any function, before the prologue. The top of
2727 the frame is defined to be the value of the stack pointer in the
2728 previous frame, just before the call instruction.
2730 You only need to define this macro if you want to support call frame
2731 debugging information like that provided by DWARF 2.
2734 @defmac ARG_POINTER_CFA_OFFSET (@var{fundecl})
2735 A C expression whose value is an integer giving the offset, in bytes,
2736 from the argument pointer to the canonical frame address (cfa). The
2737 final value should coincide with that calculated by
2738 @code{INCOMING_FRAME_SP_OFFSET}. Which is unfortunately not usable
2739 during virtual register instantiation.
2741 The default value for this macro is
2742 @code{FIRST_PARM_OFFSET (fundecl) + crtl->args.pretend_args_size},
2743 which is correct for most machines; in general, the arguments are found
2744 immediately before the stack frame. Note that this is not the case on
2745 some targets that save registers into the caller's frame, such as SPARC
2746 and rs6000, and so such targets need to define this macro.
2748 You only need to define this macro if the default is incorrect, and you
2749 want to support call frame debugging information like that provided by
2753 @defmac FRAME_POINTER_CFA_OFFSET (@var{fundecl})
2754 If defined, a C expression whose value is an integer giving the offset
2755 in bytes from the frame pointer to the canonical frame address (cfa).
2756 The final value should coincide with that calculated by
2757 @code{INCOMING_FRAME_SP_OFFSET}.
2759 Normally the CFA is calculated as an offset from the argument pointer,
2760 via @code{ARG_POINTER_CFA_OFFSET}, but if the argument pointer is
2761 variable due to the ABI, this may not be possible. If this macro is
2762 defined, it implies that the virtual register instantiation should be
2763 based on the frame pointer instead of the argument pointer. Only one
2764 of @code{FRAME_POINTER_CFA_OFFSET} and @code{ARG_POINTER_CFA_OFFSET}
2768 @defmac CFA_FRAME_BASE_OFFSET (@var{fundecl})
2769 If defined, a C expression whose value is an integer giving the offset
2770 in bytes from the canonical frame address (cfa) to the frame base used
2771 in DWARF 2 debug information. The default is zero. A different value
2772 may reduce the size of debug information on some ports.
2775 @node Exception Handling
2776 @subsection Exception Handling Support
2777 @cindex exception handling
2779 @defmac EH_RETURN_DATA_REGNO (@var{N})
2780 A C expression whose value is the @var{N}th register number used for
2781 data by exception handlers, or @code{INVALID_REGNUM} if fewer than
2782 @var{N} registers are usable.
2784 The exception handling library routines communicate with the exception
2785 handlers via a set of agreed upon registers. Ideally these registers
2786 should be call-clobbered; it is possible to use call-saved registers,
2787 but may negatively impact code size. The target must support at least
2788 2 data registers, but should define 4 if there are enough free registers.
2790 You must define this macro if you want to support call frame exception
2791 handling like that provided by DWARF 2.
2794 @defmac EH_RETURN_STACKADJ_RTX
2795 A C expression whose value is RTL representing a location in which
2796 to store a stack adjustment to be applied before function return.
2797 This is used to unwind the stack to an exception handler's call frame.
2798 It will be assigned zero on code paths that return normally.
2800 Typically this is a call-clobbered hard register that is otherwise
2801 untouched by the epilogue, but could also be a stack slot.
2803 Do not define this macro if the stack pointer is saved and restored
2804 by the regular prolog and epilog code in the call frame itself; in
2805 this case, the exception handling library routines will update the
2806 stack location to be restored in place. Otherwise, you must define
2807 this macro if you want to support call frame exception handling like
2808 that provided by DWARF 2.
2811 @defmac EH_RETURN_HANDLER_RTX
2812 A C expression whose value is RTL representing a location in which
2813 to store the address of an exception handler to which we should
2814 return. It will not be assigned on code paths that return normally.
2816 Typically this is the location in the call frame at which the normal
2817 return address is stored. For targets that return by popping an
2818 address off the stack, this might be a memory address just below
2819 the @emph{target} call frame rather than inside the current call
2820 frame. If defined, @code{EH_RETURN_STACKADJ_RTX} will have already
2821 been assigned, so it may be used to calculate the location of the
2824 Some targets have more complex requirements than storing to an
2825 address calculable during initial code generation. In that case
2826 the @code{eh_return} instruction pattern should be used instead.
2828 If you want to support call frame exception handling, you must
2829 define either this macro or the @code{eh_return} instruction pattern.
2832 @defmac RETURN_ADDR_OFFSET
2833 If defined, an integer-valued C expression for which rtl will be generated
2834 to add it to the exception handler address before it is searched in the
2835 exception handling tables, and to subtract it again from the address before
2836 using it to return to the exception handler.
2839 @defmac ASM_PREFERRED_EH_DATA_FORMAT (@var{code}, @var{global})
2840 This macro chooses the encoding of pointers embedded in the exception
2841 handling sections. If at all possible, this should be defined such
2842 that the exception handling section will not require dynamic relocations,
2843 and so may be read-only.
2845 @var{code} is 0 for data, 1 for code labels, 2 for function pointers.
2846 @var{global} is true if the symbol may be affected by dynamic relocations.
2847 The macro should return a combination of the @code{DW_EH_PE_*} defines
2848 as found in @file{dwarf2.h}.
2850 If this macro is not defined, pointers will not be encoded but
2851 represented directly.
2854 @defmac ASM_MAYBE_OUTPUT_ENCODED_ADDR_RTX (@var{file}, @var{encoding}, @var{size}, @var{addr}, @var{done})
2855 This macro allows the target to emit whatever special magic is required
2856 to represent the encoding chosen by @code{ASM_PREFERRED_EH_DATA_FORMAT}.
2857 Generic code takes care of pc-relative and indirect encodings; this must
2858 be defined if the target uses text-relative or data-relative encodings.
2860 This is a C statement that branches to @var{done} if the format was
2861 handled. @var{encoding} is the format chosen, @var{size} is the number
2862 of bytes that the format occupies, @var{addr} is the @code{SYMBOL_REF}
2866 @defmac MD_FALLBACK_FRAME_STATE_FOR (@var{context}, @var{fs})
2867 This macro allows the target to add CPU and operating system specific
2868 code to the call-frame unwinder for use when there is no unwind data
2869 available. The most common reason to implement this macro is to unwind
2870 through signal frames.
2872 This macro is called from @code{uw_frame_state_for} in
2873 @file{unwind-dw2.c}, @file{unwind-dw2-xtensa.c} and
2874 @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context};
2875 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{context->ra}
2876 for the address of the code being executed and @code{context->cfa} for
2877 the stack pointer value. If the frame can be decoded, the register
2878 save addresses should be updated in @var{fs} and the macro should
2879 evaluate to @code{_URC_NO_REASON}. If the frame cannot be decoded,
2880 the macro should evaluate to @code{_URC_END_OF_STACK}.
2882 For proper signal handling in Java this macro is accompanied by
2883 @code{MAKE_THROW_FRAME}, defined in @file{libjava/include/*-signal.h} headers.
2886 @defmac MD_HANDLE_UNWABI (@var{context}, @var{fs})
2887 This macro allows the target to add operating system specific code to the
2888 call-frame unwinder to handle the IA-64 @code{.unwabi} unwinding directive,
2889 usually used for signal or interrupt frames.
2891 This macro is called from @code{uw_update_context} in libgcc's
2892 @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context};
2893 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{fs->unwabi}
2894 for the abi and context in the @code{.unwabi} directive. If the
2895 @code{.unwabi} directive can be handled, the register save addresses should
2896 be updated in @var{fs}.
2899 @defmac TARGET_USES_WEAK_UNWIND_INFO
2900 A C expression that evaluates to true if the target requires unwind
2901 info to be given comdat linkage. Define it to be @code{1} if comdat
2902 linkage is necessary. The default is @code{0}.
2905 @node Stack Checking
2906 @subsection Specifying How Stack Checking is Done
2908 GCC will check that stack references are within the boundaries of the
2909 stack, if the option @option{-fstack-check} is specified, in one of
2914 If the value of the @code{STACK_CHECK_BUILTIN} macro is nonzero, GCC
2915 will assume that you have arranged for full stack checking to be done
2916 at appropriate places in the configuration files. GCC will not do
2917 other special processing.
2920 If @code{STACK_CHECK_BUILTIN} is zero and the value of the
2921 @code{STACK_CHECK_STATIC_BUILTIN} macro is nonzero, GCC will assume
2922 that you have arranged for static stack checking (checking of the
2923 static stack frame of functions) to be done at appropriate places
2924 in the configuration files. GCC will only emit code to do dynamic
2925 stack checking (checking on dynamic stack allocations) using the third
2929 If neither of the above are true, GCC will generate code to periodically
2930 ``probe'' the stack pointer using the values of the macros defined below.
2933 If neither STACK_CHECK_BUILTIN nor STACK_CHECK_STATIC_BUILTIN is defined,
2934 GCC will change its allocation strategy for large objects if the option
2935 @option{-fstack-check} is specified: they will always be allocated
2936 dynamically if their size exceeds @code{STACK_CHECK_MAX_VAR_SIZE} bytes.
2938 @defmac STACK_CHECK_BUILTIN
2939 A nonzero value if stack checking is done by the configuration files in a
2940 machine-dependent manner. You should define this macro if stack checking
2941 is required by the ABI of your machine or if you would like to do stack
2942 checking in some more efficient way than the generic approach. The default
2943 value of this macro is zero.
2946 @defmac STACK_CHECK_STATIC_BUILTIN
2947 A nonzero value if static stack checking is done by the configuration files
2948 in a machine-dependent manner. You should define this macro if you would
2949 like to do static stack checking in some more efficient way than the generic
2950 approach. The default value of this macro is zero.
2953 @defmac STACK_CHECK_PROBE_INTERVAL_EXP
2954 An integer specifying the interval at which GCC must generate stack probe
2955 instructions, defined as 2 raised to this integer. You will normally
2956 define this macro so that the interval be no larger than the size of
2957 the ``guard pages'' at the end of a stack area. The default value
2958 of 12 (4096-byte interval) is suitable for most systems.
2961 @defmac STACK_CHECK_MOVING_SP
2962 An integer which is nonzero if GCC should move the stack pointer page by page
2963 when doing probes. This can be necessary on systems where the stack pointer
2964 contains the bottom address of the memory area accessible to the executing
2965 thread at any point in time. In this situation an alternate signal stack
2966 is required in order to be able to recover from a stack overflow. The
2967 default value of this macro is zero.
2970 @defmac STACK_CHECK_PROTECT
2971 The number of bytes of stack needed to recover from a stack overflow, for
2972 languages where such a recovery is supported. The default value of 4KB/8KB
2973 with the @code{setjmp}/@code{longjmp}-based exception handling mechanism and
2974 8KB/12KB with other exception handling mechanisms should be adequate for most
2975 architectures and operating systems.
2978 The following macros are relevant only if neither STACK_CHECK_BUILTIN
2979 nor STACK_CHECK_STATIC_BUILTIN is defined; you can omit them altogether
2980 in the opposite case.
2982 @defmac STACK_CHECK_MAX_FRAME_SIZE
2983 The maximum size of a stack frame, in bytes. GCC will generate probe
2984 instructions in non-leaf functions to ensure at least this many bytes of
2985 stack are available. If a stack frame is larger than this size, stack
2986 checking will not be reliable and GCC will issue a warning. The
2987 default is chosen so that GCC only generates one instruction on most
2988 systems. You should normally not change the default value of this macro.
2991 @defmac STACK_CHECK_FIXED_FRAME_SIZE
2992 GCC uses this value to generate the above warning message. It
2993 represents the amount of fixed frame used by a function, not including
2994 space for any callee-saved registers, temporaries and user variables.
2995 You need only specify an upper bound for this amount and will normally
2996 use the default of four words.
2999 @defmac STACK_CHECK_MAX_VAR_SIZE
3000 The maximum size, in bytes, of an object that GCC will place in the
3001 fixed area of the stack frame when the user specifies
3002 @option{-fstack-check}.
3003 GCC computed the default from the values of the above macros and you will
3004 normally not need to override that default.
3008 @node Frame Registers
3009 @subsection Registers That Address the Stack Frame
3011 @c prevent bad page break with this line
3012 This discusses registers that address the stack frame.
3014 @defmac STACK_POINTER_REGNUM
3015 The register number of the stack pointer register, which must also be a
3016 fixed register according to @code{FIXED_REGISTERS}. On most machines,
3017 the hardware determines which register this is.
3020 @defmac FRAME_POINTER_REGNUM
3021 The register number of the frame pointer register, which is used to
3022 access automatic variables in the stack frame. On some machines, the
3023 hardware determines which register this is. On other machines, you can
3024 choose any register you wish for this purpose.
3027 @defmac HARD_FRAME_POINTER_REGNUM
3028 On some machines the offset between the frame pointer and starting
3029 offset of the automatic variables is not known until after register
3030 allocation has been done (for example, because the saved registers are
3031 between these two locations). On those machines, define
3032 @code{FRAME_POINTER_REGNUM} the number of a special, fixed register to
3033 be used internally until the offset is known, and define
3034 @code{HARD_FRAME_POINTER_REGNUM} to be the actual hard register number
3035 used for the frame pointer.
3037 You should define this macro only in the very rare circumstances when it
3038 is not possible to calculate the offset between the frame pointer and
3039 the automatic variables until after register allocation has been
3040 completed. When this macro is defined, you must also indicate in your
3041 definition of @code{ELIMINABLE_REGS} how to eliminate
3042 @code{FRAME_POINTER_REGNUM} into either @code{HARD_FRAME_POINTER_REGNUM}
3043 or @code{STACK_POINTER_REGNUM}.
3045 Do not define this macro if it would be the same as
3046 @code{FRAME_POINTER_REGNUM}.
3049 @defmac ARG_POINTER_REGNUM
3050 The register number of the arg pointer register, which is used to access
3051 the function's argument list. On some machines, this is the same as the
3052 frame pointer register. On some machines, the hardware determines which
3053 register this is. On other machines, you can choose any register you
3054 wish for this purpose. If this is not the same register as the frame
3055 pointer register, then you must mark it as a fixed register according to
3056 @code{FIXED_REGISTERS}, or arrange to be able to eliminate it
3057 (@pxref{Elimination}).
3060 @defmac HARD_FRAME_POINTER_IS_FRAME_POINTER
3061 Define this to a preprocessor constant that is nonzero if
3062 @code{hard_frame_pointer_rtx} and @code{frame_pointer_rtx} should be
3063 the same. The default definition is @samp{(HARD_FRAME_POINTER_REGNUM
3064 == FRAME_POINTER_REGNUM)}; you only need to define this macro if that
3065 definition is not suitable for use in preprocessor conditionals.
3068 @defmac HARD_FRAME_POINTER_IS_ARG_POINTER
3069 Define this to a preprocessor constant that is nonzero if
3070 @code{hard_frame_pointer_rtx} and @code{arg_pointer_rtx} should be the
3071 same. The default definition is @samp{(HARD_FRAME_POINTER_REGNUM ==
3072 ARG_POINTER_REGNUM)}; you only need to define this macro if that
3073 definition is not suitable for use in preprocessor conditionals.
3076 @defmac RETURN_ADDRESS_POINTER_REGNUM
3077 The register number of the return address pointer register, which is used to
3078 access the current function's return address from the stack. On some
3079 machines, the return address is not at a fixed offset from the frame
3080 pointer or stack pointer or argument pointer. This register can be defined
3081 to point to the return address on the stack, and then be converted by
3082 @code{ELIMINABLE_REGS} into either the frame pointer or stack pointer.
3084 Do not define this macro unless there is no other way to get the return
3085 address from the stack.
3088 @defmac STATIC_CHAIN_REGNUM
3089 @defmacx STATIC_CHAIN_INCOMING_REGNUM
3090 Register numbers used for passing a function's static chain pointer. If
3091 register windows are used, the register number as seen by the called
3092 function is @code{STATIC_CHAIN_INCOMING_REGNUM}, while the register
3093 number as seen by the calling function is @code{STATIC_CHAIN_REGNUM}. If
3094 these registers are the same, @code{STATIC_CHAIN_INCOMING_REGNUM} need
3097 The static chain register need not be a fixed register.
3099 If the static chain is passed in memory, these macros should not be
3100 defined; instead, the @code{TARGET_STATIC_CHAIN} hook should be used.
3103 @hook TARGET_STATIC_CHAIN
3105 @defmac DWARF_FRAME_REGISTERS
3106 This macro specifies the maximum number of hard registers that can be
3107 saved in a call frame. This is used to size data structures used in
3108 DWARF2 exception handling.
3110 Prior to GCC 3.0, this macro was needed in order to establish a stable
3111 exception handling ABI in the face of adding new hard registers for ISA
3112 extensions. In GCC 3.0 and later, the EH ABI is insulated from changes
3113 in the number of hard registers. Nevertheless, this macro can still be
3114 used to reduce the runtime memory requirements of the exception handling
3115 routines, which can be substantial if the ISA contains a lot of
3116 registers that are not call-saved.
3118 If this macro is not defined, it defaults to
3119 @code{FIRST_PSEUDO_REGISTER}.
3122 @defmac PRE_GCC3_DWARF_FRAME_REGISTERS
3124 This macro is similar to @code{DWARF_FRAME_REGISTERS}, but is provided
3125 for backward compatibility in pre GCC 3.0 compiled code.
3127 If this macro is not defined, it defaults to
3128 @code{DWARF_FRAME_REGISTERS}.
3131 @defmac DWARF_REG_TO_UNWIND_COLUMN (@var{regno})
3133 Define this macro if the target's representation for dwarf registers
3134 is different than the internal representation for unwind column.
3135 Given a dwarf register, this macro should return the internal unwind
3136 column number to use instead.
3138 See the PowerPC's SPE target for an example.
3141 @defmac DWARF_FRAME_REGNUM (@var{regno})
3143 Define this macro if the target's representation for dwarf registers
3144 used in .eh_frame or .debug_frame is different from that used in other
3145 debug info sections. Given a GCC hard register number, this macro
3146 should return the .eh_frame register number. The default is
3147 @code{DBX_REGISTER_NUMBER (@var{regno})}.
3151 @defmac DWARF2_FRAME_REG_OUT (@var{regno}, @var{for_eh})
3153 Define this macro to map register numbers held in the call frame info
3154 that GCC has collected using @code{DWARF_FRAME_REGNUM} to those that
3155 should be output in .debug_frame (@code{@var{for_eh}} is zero) and
3156 .eh_frame (@code{@var{for_eh}} is nonzero). The default is to
3157 return @code{@var{regno}}.
3161 @defmac REG_VALUE_IN_UNWIND_CONTEXT
3163 Define this macro if the target stores register values as
3164 @code{_Unwind_Word} type in unwind context. It should be defined if
3165 target register size is larger than the size of @code{void *}. The
3166 default is to store register values as @code{void *} type.
3170 @defmac ASSUME_EXTENDED_UNWIND_CONTEXT
3172 Define this macro to be 1 if the target always uses extended unwind
3173 context with version, args_size and by_value fields. If it is undefined,
3174 it will be defined to 1 when @code{REG_VALUE_IN_UNWIND_CONTEXT} is
3175 defined and 0 otherwise.
3180 @subsection Eliminating Frame Pointer and Arg Pointer
3182 @c prevent bad page break with this line
3183 This is about eliminating the frame pointer and arg pointer.
3185 @hook TARGET_FRAME_POINTER_REQUIRED
3187 @defmac ELIMINABLE_REGS
3188 This macro specifies a table of register pairs used to eliminate
3189 unneeded registers that point into the stack frame.
3191 The definition of this macro is a list of structure initializations, each
3192 of which specifies an original and replacement register.
3194 On some machines, the position of the argument pointer is not known until
3195 the compilation is completed. In such a case, a separate hard register
3196 must be used for the argument pointer. This register can be eliminated by
3197 replacing it with either the frame pointer or the argument pointer,
3198 depending on whether or not the frame pointer has been eliminated.
3200 In this case, you might specify:
3202 #define ELIMINABLE_REGS \
3203 @{@{ARG_POINTER_REGNUM, STACK_POINTER_REGNUM@}, \
3204 @{ARG_POINTER_REGNUM, FRAME_POINTER_REGNUM@}, \
3205 @{FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM@}@}
3208 Note that the elimination of the argument pointer with the stack pointer is
3209 specified first since that is the preferred elimination.
3212 @hook TARGET_CAN_ELIMINATE
3214 @defmac INITIAL_ELIMINATION_OFFSET (@var{from-reg}, @var{to-reg}, @var{offset-var})
3215 This macro returns the initial difference between the specified pair
3216 of registers. The value would be computed from information
3217 such as the result of @code{get_frame_size ()} and the tables of
3218 registers @code{df_regs_ever_live_p} and @code{call_used_regs}.
3221 @node Stack Arguments
3222 @subsection Passing Function Arguments on the Stack
3223 @cindex arguments on stack
3224 @cindex stack arguments
3226 The macros in this section control how arguments are passed
3227 on the stack. See the following section for other macros that
3228 control passing certain arguments in registers.
3230 @hook TARGET_PROMOTE_PROTOTYPES
3233 A C expression. If nonzero, push insns will be used to pass
3235 If the target machine does not have a push instruction, set it to zero.
3236 That directs GCC to use an alternate strategy: to
3237 allocate the entire argument block and then store the arguments into
3238 it. When @code{PUSH_ARGS} is nonzero, @code{PUSH_ROUNDING} must be defined too.
3241 @defmac PUSH_ARGS_REVERSED
3242 A C expression. If nonzero, function arguments will be evaluated from
3243 last to first, rather than from first to last. If this macro is not
3244 defined, it defaults to @code{PUSH_ARGS} on targets where the stack
3245 and args grow in opposite directions, and 0 otherwise.
3248 @defmac PUSH_ROUNDING (@var{npushed})
3249 A C expression that is the number of bytes actually pushed onto the
3250 stack when an instruction attempts to push @var{npushed} bytes.
3252 On some machines, the definition
3255 #define PUSH_ROUNDING(BYTES) (BYTES)
3259 will suffice. But on other machines, instructions that appear
3260 to push one byte actually push two bytes in an attempt to maintain
3261 alignment. Then the definition should be
3264 #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1)
3267 If the value of this macro has a type, it should be an unsigned type.
3270 @findex outgoing_args_size
3271 @findex crtl->outgoing_args_size
3272 @defmac ACCUMULATE_OUTGOING_ARGS
3273 A C expression. If nonzero, the maximum amount of space required for outgoing arguments
3274 will be computed and placed into
3275 @code{crtl->outgoing_args_size}. No space will be pushed
3276 onto the stack for each call; instead, the function prologue should
3277 increase the stack frame size by this amount.
3279 Setting both @code{PUSH_ARGS} and @code{ACCUMULATE_OUTGOING_ARGS}
3283 @defmac REG_PARM_STACK_SPACE (@var{fndecl})
3284 Define this macro if functions should assume that stack space has been
3285 allocated for arguments even when their values are passed in
3288 The value of this macro is the size, in bytes, of the area reserved for
3289 arguments passed in registers for the function represented by @var{fndecl},
3290 which can be zero if GCC is calling a library function.
3291 The argument @var{fndecl} can be the FUNCTION_DECL, or the type itself
3294 This space can be allocated by the caller, or be a part of the
3295 machine-dependent stack frame: @code{OUTGOING_REG_PARM_STACK_SPACE} says
3298 @c above is overfull. not sure what to do. --mew 5feb93 did
3299 @c something, not sure if it looks good. --mew 10feb93
3301 @defmac INCOMING_REG_PARM_STACK_SPACE (@var{fndecl})
3302 Like @code{REG_PARM_STACK_SPACE}, but for incoming register arguments.
3303 Define this macro if space guaranteed when compiling a function body
3304 is different to space required when making a call, a situation that
3305 can arise with K&R style function definitions.
3308 @defmac OUTGOING_REG_PARM_STACK_SPACE (@var{fntype})
3309 Define this to a nonzero value if it is the responsibility of the
3310 caller to allocate the area reserved for arguments passed in registers
3311 when calling a function of @var{fntype}. @var{fntype} may be NULL
3312 if the function called is a library function.
3314 If @code{ACCUMULATE_OUTGOING_ARGS} is defined, this macro controls
3315 whether the space for these arguments counts in the value of
3316 @code{crtl->outgoing_args_size}.
3319 @defmac STACK_PARMS_IN_REG_PARM_AREA
3320 Define this macro if @code{REG_PARM_STACK_SPACE} is defined, but the
3321 stack parameters don't skip the area specified by it.
3322 @c i changed this, makes more sens and it should have taken care of the
3323 @c overfull.. not as specific, tho. --mew 5feb93
3325 Normally, when a parameter is not passed in registers, it is placed on the
3326 stack beyond the @code{REG_PARM_STACK_SPACE} area. Defining this macro
3327 suppresses this behavior and causes the parameter to be passed on the
3328 stack in its natural location.
3331 @hook TARGET_RETURN_POPS_ARGS
3333 @defmac CALL_POPS_ARGS (@var{cum})
3334 A C expression that should indicate the number of bytes a call sequence
3335 pops off the stack. It is added to the value of @code{RETURN_POPS_ARGS}
3336 when compiling a function call.
3338 @var{cum} is the variable in which all arguments to the called function
3339 have been accumulated.
3341 On certain architectures, such as the SH5, a call trampoline is used
3342 that pops certain registers off the stack, depending on the arguments
3343 that have been passed to the function. Since this is a property of the
3344 call site, not of the called function, @code{RETURN_POPS_ARGS} is not
3348 @node Register Arguments
3349 @subsection Passing Arguments in Registers
3350 @cindex arguments in registers
3351 @cindex registers arguments
3353 This section describes the macros which let you control how various
3354 types of arguments are passed in registers or how they are arranged in
3357 @hook TARGET_FUNCTION_ARG
3359 @hook TARGET_MUST_PASS_IN_STACK
3361 @hook TARGET_FUNCTION_INCOMING_ARG
3363 @hook TARGET_USE_PSEUDO_PIC_REG
3365 @hook TARGET_INIT_PIC_REG
3367 @hook TARGET_ARG_PARTIAL_BYTES
3369 @hook TARGET_PASS_BY_REFERENCE
3371 @hook TARGET_CALLEE_COPIES
3373 @defmac CUMULATIVE_ARGS
3374 A C type for declaring a variable that is used as the first argument
3375 of @code{TARGET_FUNCTION_ARG} and other related values. For some
3376 target machines, the type @code{int} suffices and can hold the number
3377 of bytes of argument so far.
3379 There is no need to record in @code{CUMULATIVE_ARGS} anything about the
3380 arguments that have been passed on the stack. The compiler has other
3381 variables to keep track of that. For target machines on which all
3382 arguments are passed on the stack, there is no need to store anything in
3383 @code{CUMULATIVE_ARGS}; however, the data structure must exist and
3384 should not be empty, so use @code{int}.
3387 @defmac OVERRIDE_ABI_FORMAT (@var{fndecl})
3388 If defined, this macro is called before generating any code for a
3389 function, but after the @var{cfun} descriptor for the function has been
3390 created. The back end may use this macro to update @var{cfun} to
3391 reflect an ABI other than that which would normally be used by default.
3392 If the compiler is generating code for a compiler-generated function,
3393 @var{fndecl} may be @code{NULL}.
3396 @defmac INIT_CUMULATIVE_ARGS (@var{cum}, @var{fntype}, @var{libname}, @var{fndecl}, @var{n_named_args})
3397 A C statement (sans semicolon) for initializing the variable
3398 @var{cum} for the state at the beginning of the argument list. The
3399 variable has type @code{CUMULATIVE_ARGS}. The value of @var{fntype}
3400 is the tree node for the data type of the function which will receive
3401 the args, or 0 if the args are to a compiler support library function.
3402 For direct calls that are not libcalls, @var{fndecl} contain the
3403 declaration node of the function. @var{fndecl} is also set when
3404 @code{INIT_CUMULATIVE_ARGS} is used to find arguments for the function
3405 being compiled. @var{n_named_args} is set to the number of named
3406 arguments, including a structure return address if it is passed as a
3407 parameter, when making a call. When processing incoming arguments,
3408 @var{n_named_args} is set to @minus{}1.
3410 When processing a call to a compiler support library function,
3411 @var{libname} identifies which one. It is a @code{symbol_ref} rtx which
3412 contains the name of the function, as a string. @var{libname} is 0 when
3413 an ordinary C function call is being processed. Thus, each time this
3414 macro is called, either @var{libname} or @var{fntype} is nonzero, but
3415 never both of them at once.
3418 @defmac INIT_CUMULATIVE_LIBCALL_ARGS (@var{cum}, @var{mode}, @var{libname})
3419 Like @code{INIT_CUMULATIVE_ARGS} but only used for outgoing libcalls,
3420 it gets a @code{MODE} argument instead of @var{fntype}, that would be
3421 @code{NULL}. @var{indirect} would always be zero, too. If this macro
3422 is not defined, @code{INIT_CUMULATIVE_ARGS (cum, NULL_RTX, libname,
3423 0)} is used instead.
3426 @defmac INIT_CUMULATIVE_INCOMING_ARGS (@var{cum}, @var{fntype}, @var{libname})
3427 Like @code{INIT_CUMULATIVE_ARGS} but overrides it for the purposes of
3428 finding the arguments for the function being compiled. If this macro is
3429 undefined, @code{INIT_CUMULATIVE_ARGS} is used instead.
3431 The value passed for @var{libname} is always 0, since library routines
3432 with special calling conventions are never compiled with GCC@. The
3433 argument @var{libname} exists for symmetry with
3434 @code{INIT_CUMULATIVE_ARGS}.
3435 @c could use "this macro" in place of @code{INIT_CUMULATIVE_ARGS}, maybe.
3436 @c --mew 5feb93 i switched the order of the sentences. --mew 10feb93
3439 @hook TARGET_FUNCTION_ARG_ADVANCE
3441 @defmac FUNCTION_ARG_OFFSET (@var{mode}, @var{type})
3442 If defined, a C expression that is the number of bytes to add to the
3443 offset of the argument passed in memory. This is needed for the SPU,
3444 which passes @code{char} and @code{short} arguments in the preferred
3445 slot that is in the middle of the quad word instead of starting at the
3449 @defmac FUNCTION_ARG_PADDING (@var{mode}, @var{type})
3450 If defined, a C expression which determines whether, and in which direction,
3451 to pad out an argument with extra space. The value should be of type
3452 @code{enum direction}: either @code{upward} to pad above the argument,
3453 @code{downward} to pad below, or @code{none} to inhibit padding.
3455 The @emph{amount} of padding is not controlled by this macro, but by the
3456 target hook @code{TARGET_FUNCTION_ARG_ROUND_BOUNDARY}. It is
3457 always just enough to reach the next multiple of that boundary.
3459 This macro has a default definition which is right for most systems.
3460 For little-endian machines, the default is to pad upward. For
3461 big-endian machines, the default is to pad downward for an argument of
3462 constant size shorter than an @code{int}, and upward otherwise.
3465 @defmac PAD_VARARGS_DOWN
3466 If defined, a C expression which determines whether the default
3467 implementation of va_arg will attempt to pad down before reading the
3468 next argument, if that argument is smaller than its aligned space as
3469 controlled by @code{PARM_BOUNDARY}. If this macro is not defined, all such
3470 arguments are padded down if @code{BYTES_BIG_ENDIAN} is true.
3473 @defmac BLOCK_REG_PADDING (@var{mode}, @var{type}, @var{first})
3474 Specify padding for the last element of a block move between registers and
3475 memory. @var{first} is nonzero if this is the only element. Defining this
3476 macro allows better control of register function parameters on big-endian
3477 machines, without using @code{PARALLEL} rtl. In particular,
3478 @code{MUST_PASS_IN_STACK} need not test padding and mode of types in
3479 registers, as there is no longer a "wrong" part of a register; For example,
3480 a three byte aggregate may be passed in the high part of a register if so
3484 @hook TARGET_FUNCTION_ARG_BOUNDARY
3486 @hook TARGET_FUNCTION_ARG_ROUND_BOUNDARY
3488 @defmac FUNCTION_ARG_REGNO_P (@var{regno})
3489 A C expression that is nonzero if @var{regno} is the number of a hard
3490 register in which function arguments are sometimes passed. This does
3491 @emph{not} include implicit arguments such as the static chain and
3492 the structure-value address. On many machines, no registers can be
3493 used for this purpose since all function arguments are pushed on the
3497 @hook TARGET_SPLIT_COMPLEX_ARG
3499 @hook TARGET_BUILD_BUILTIN_VA_LIST
3501 @hook TARGET_ENUM_VA_LIST_P
3503 @hook TARGET_FN_ABI_VA_LIST
3505 @hook TARGET_CANONICAL_VA_LIST_TYPE
3507 @hook TARGET_GIMPLIFY_VA_ARG_EXPR
3509 @hook TARGET_VALID_POINTER_MODE
3511 @hook TARGET_REF_MAY_ALIAS_ERRNO
3513 @hook TARGET_SCALAR_MODE_SUPPORTED_P
3515 @hook TARGET_VECTOR_MODE_SUPPORTED_P
3517 @hook TARGET_ARRAY_MODE_SUPPORTED_P
3519 @hook TARGET_LIBGCC_FLOATING_MODE_SUPPORTED_P
3521 @hook TARGET_FLOATN_MODE
3523 @hook TARGET_SMALL_REGISTER_CLASSES_FOR_MODE_P
3526 @subsection How Scalar Function Values Are Returned
3527 @cindex return values in registers
3528 @cindex values, returned by functions
3529 @cindex scalars, returned as values
3531 This section discusses the macros that control returning scalars as
3532 values---values that can fit in registers.
3534 @hook TARGET_FUNCTION_VALUE
3536 @defmac FUNCTION_VALUE (@var{valtype}, @var{func})
3537 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE} for
3538 a new target instead.
3541 @defmac LIBCALL_VALUE (@var{mode})
3542 A C expression to create an RTX representing the place where a library
3543 function returns a value of mode @var{mode}.
3545 Note that ``library function'' in this context means a compiler
3546 support routine, used to perform arithmetic, whose name is known
3547 specially by the compiler and was not mentioned in the C code being
3551 @hook TARGET_LIBCALL_VALUE
3553 @defmac FUNCTION_VALUE_REGNO_P (@var{regno})
3554 A C expression that is nonzero if @var{regno} is the number of a hard
3555 register in which the values of called function may come back.
3557 A register whose use for returning values is limited to serving as the
3558 second of a pair (for a value of type @code{double}, say) need not be
3559 recognized by this macro. So for most machines, this definition
3563 #define FUNCTION_VALUE_REGNO_P(N) ((N) == 0)
3566 If the machine has register windows, so that the caller and the called
3567 function use different registers for the return value, this macro
3568 should recognize only the caller's register numbers.
3570 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE_REGNO_P}
3571 for a new target instead.
3574 @hook TARGET_FUNCTION_VALUE_REGNO_P
3576 @defmac APPLY_RESULT_SIZE
3577 Define this macro if @samp{untyped_call} and @samp{untyped_return}
3578 need more space than is implied by @code{FUNCTION_VALUE_REGNO_P} for
3579 saving and restoring an arbitrary return value.
3582 @hook TARGET_OMIT_STRUCT_RETURN_REG
3584 @hook TARGET_RETURN_IN_MSB
3586 @node Aggregate Return
3587 @subsection How Large Values Are Returned
3588 @cindex aggregates as return values
3589 @cindex large return values
3590 @cindex returning aggregate values
3591 @cindex structure value address
3593 When a function value's mode is @code{BLKmode} (and in some other
3594 cases), the value is not returned according to
3595 @code{TARGET_FUNCTION_VALUE} (@pxref{Scalar Return}). Instead, the
3596 caller passes the address of a block of memory in which the value
3597 should be stored. This address is called the @dfn{structure value
3600 This section describes how to control returning structure values in
3603 @hook TARGET_RETURN_IN_MEMORY
3605 @defmac DEFAULT_PCC_STRUCT_RETURN
3606 Define this macro to be 1 if all structure and union return values must be
3607 in memory. Since this results in slower code, this should be defined
3608 only if needed for compatibility with other compilers or with an ABI@.
3609 If you define this macro to be 0, then the conventions used for structure
3610 and union return values are decided by the @code{TARGET_RETURN_IN_MEMORY}
3613 If not defined, this defaults to the value 1.
3616 @hook TARGET_STRUCT_VALUE_RTX
3618 @defmac PCC_STATIC_STRUCT_RETURN
3619 Define this macro if the usual system convention on the target machine
3620 for returning structures and unions is for the called function to return
3621 the address of a static variable containing the value.
3623 Do not define this if the usual system convention is for the caller to
3624 pass an address to the subroutine.
3626 This macro has effect in @option{-fpcc-struct-return} mode, but it does
3627 nothing when you use @option{-freg-struct-return} mode.
3630 @hook TARGET_GET_RAW_RESULT_MODE
3632 @hook TARGET_GET_RAW_ARG_MODE
3635 @subsection Caller-Saves Register Allocation
3637 If you enable it, GCC can save registers around function calls. This
3638 makes it possible to use call-clobbered registers to hold variables that
3639 must live across calls.
3641 @defmac HARD_REGNO_CALLER_SAVE_MODE (@var{regno}, @var{nregs})
3642 A C expression specifying which mode is required for saving @var{nregs}
3643 of a pseudo-register in call-clobbered hard register @var{regno}. If
3644 @var{regno} is unsuitable for caller save, @code{VOIDmode} should be
3645 returned. For most machines this macro need not be defined since GCC
3646 will select the smallest suitable mode.
3649 @node Function Entry
3650 @subsection Function Entry and Exit
3651 @cindex function entry and exit
3655 This section describes the macros that output function entry
3656 (@dfn{prologue}) and exit (@dfn{epilogue}) code.
3658 @hook TARGET_ASM_FUNCTION_PROLOGUE
3660 @hook TARGET_ASM_FUNCTION_END_PROLOGUE
3662 @hook TARGET_ASM_FUNCTION_BEGIN_EPILOGUE
3664 @hook TARGET_ASM_FUNCTION_EPILOGUE
3668 @findex pretend_args_size
3669 @findex crtl->args.pretend_args_size
3670 A region of @code{crtl->args.pretend_args_size} bytes of
3671 uninitialized space just underneath the first argument arriving on the
3672 stack. (This may not be at the very start of the allocated stack region
3673 if the calling sequence has pushed anything else since pushing the stack
3674 arguments. But usually, on such machines, nothing else has been pushed
3675 yet, because the function prologue itself does all the pushing.) This
3676 region is used on machines where an argument may be passed partly in
3677 registers and partly in memory, and, in some cases to support the
3678 features in @code{<stdarg.h>}.
3681 An area of memory used to save certain registers used by the function.
3682 The size of this area, which may also include space for such things as
3683 the return address and pointers to previous stack frames, is
3684 machine-specific and usually depends on which registers have been used
3685 in the function. Machines with register windows often do not require
3689 A region of at least @var{size} bytes, possibly rounded up to an allocation
3690 boundary, to contain the local variables of the function. On some machines,
3691 this region and the save area may occur in the opposite order, with the
3692 save area closer to the top of the stack.
3695 @cindex @code{ACCUMULATE_OUTGOING_ARGS} and stack frames
3696 Optionally, when @code{ACCUMULATE_OUTGOING_ARGS} is defined, a region of
3697 @code{crtl->outgoing_args_size} bytes to be used for outgoing
3698 argument lists of the function. @xref{Stack Arguments}.
3701 @defmac EXIT_IGNORE_STACK
3702 Define this macro as a C expression that is nonzero if the return
3703 instruction or the function epilogue ignores the value of the stack
3704 pointer; in other words, if it is safe to delete an instruction to
3705 adjust the stack pointer before a return from the function. The
3708 Note that this macro's value is relevant only for functions for which
3709 frame pointers are maintained. It is never safe to delete a final
3710 stack adjustment in a function that has no frame pointer, and the
3711 compiler knows this regardless of @code{EXIT_IGNORE_STACK}.
3714 @defmac EPILOGUE_USES (@var{regno})
3715 Define this macro as a C expression that is nonzero for registers that are
3716 used by the epilogue or the @samp{return} pattern. The stack and frame
3717 pointer registers are already assumed to be used as needed.
3720 @defmac EH_USES (@var{regno})
3721 Define this macro as a C expression that is nonzero for registers that are
3722 used by the exception handling mechanism, and so should be considered live
3723 on entry to an exception edge.
3726 @hook TARGET_ASM_OUTPUT_MI_THUNK
3728 @hook TARGET_ASM_CAN_OUTPUT_MI_THUNK
3731 @subsection Generating Code for Profiling
3732 @cindex profiling, code generation
3734 These macros will help you generate code for profiling.
3736 @defmac FUNCTION_PROFILER (@var{file}, @var{labelno})
3737 A C statement or compound statement to output to @var{file} some
3738 assembler code to call the profiling subroutine @code{mcount}.
3741 The details of how @code{mcount} expects to be called are determined by
3742 your operating system environment, not by GCC@. To figure them out,
3743 compile a small program for profiling using the system's installed C
3744 compiler and look at the assembler code that results.
3746 Older implementations of @code{mcount} expect the address of a counter
3747 variable to be loaded into some register. The name of this variable is
3748 @samp{LP} followed by the number @var{labelno}, so you would generate
3749 the name using @samp{LP%d} in a @code{fprintf}.
3752 @defmac PROFILE_HOOK
3753 A C statement or compound statement to output to @var{file} some assembly
3754 code to call the profiling subroutine @code{mcount} even the target does
3755 not support profiling.
3758 @defmac NO_PROFILE_COUNTERS
3759 Define this macro to be an expression with a nonzero value if the
3760 @code{mcount} subroutine on your system does not need a counter variable
3761 allocated for each function. This is true for almost all modern
3762 implementations. If you define this macro, you must not use the
3763 @var{labelno} argument to @code{FUNCTION_PROFILER}.
3766 @defmac PROFILE_BEFORE_PROLOGUE
3767 Define this macro if the code for function profiling should come before
3768 the function prologue. Normally, the profiling code comes after.
3771 @hook TARGET_KEEP_LEAF_WHEN_PROFILED
3774 @subsection Permitting tail calls
3777 @hook TARGET_FUNCTION_OK_FOR_SIBCALL
3779 @hook TARGET_EXTRA_LIVE_ON_ENTRY
3781 @hook TARGET_SET_UP_BY_PROLOGUE
3783 @hook TARGET_WARN_FUNC_RETURN
3785 @node Shrink-wrapping separate components
3786 @subsection Shrink-wrapping separate components
3787 @cindex shrink-wrapping separate components
3789 The prologue may perform a variety of target dependent tasks such as
3790 saving callee-saved registers, saving the return address, aligning the
3791 stack, creating a stack frame, initializing the PIC register, setting
3792 up the static chain, etc.
3794 On some targets some of these tasks may be independent of others and
3795 thus may be shrink-wrapped separately. These independent tasks are
3796 referred to as components and are handled generically by the target
3797 independent parts of GCC.
3799 Using the following hooks those prologue or epilogue components can be
3800 shrink-wrapped separately, so that the initialization (and possibly
3801 teardown) those components do is not done as frequently on execution
3802 paths where this would unnecessary.
3804 What exactly those components are is up to the target code; the generic
3805 code treats them abstractly, as a bit in an @code{sbitmap}. These
3806 @code{sbitmap}s are allocated by the @code{shrink_wrap.get_separate_components}
3807 and @code{shrink_wrap.components_for_bb} hooks, and deallocated by the
3810 @hook TARGET_SHRINK_WRAP_GET_SEPARATE_COMPONENTS
3812 @hook TARGET_SHRINK_WRAP_COMPONENTS_FOR_BB
3814 @hook TARGET_SHRINK_WRAP_DISQUALIFY_COMPONENTS
3816 @hook TARGET_SHRINK_WRAP_EMIT_PROLOGUE_COMPONENTS
3818 @hook TARGET_SHRINK_WRAP_EMIT_EPILOGUE_COMPONENTS
3820 @hook TARGET_SHRINK_WRAP_SET_HANDLED_COMPONENTS
3822 @node Stack Smashing Protection
3823 @subsection Stack smashing protection
3824 @cindex stack smashing protection
3826 @hook TARGET_STACK_PROTECT_GUARD
3828 @hook TARGET_STACK_PROTECT_FAIL
3830 @hook TARGET_SUPPORTS_SPLIT_STACK
3832 @node Miscellaneous Register Hooks
3833 @subsection Miscellaneous register hooks
3834 @cindex miscellaneous register hooks
3836 @hook TARGET_CALL_FUSAGE_CONTAINS_NON_CALLEE_CLOBBERS
3839 @section Implementing the Varargs Macros
3840 @cindex varargs implementation
3842 GCC comes with an implementation of @code{<varargs.h>} and
3843 @code{<stdarg.h>} that work without change on machines that pass arguments
3844 on the stack. Other machines require their own implementations of
3845 varargs, and the two machine independent header files must have
3846 conditionals to include it.
3848 ISO @code{<stdarg.h>} differs from traditional @code{<varargs.h>} mainly in
3849 the calling convention for @code{va_start}. The traditional
3850 implementation takes just one argument, which is the variable in which
3851 to store the argument pointer. The ISO implementation of
3852 @code{va_start} takes an additional second argument. The user is
3853 supposed to write the last named argument of the function here.
3855 However, @code{va_start} should not use this argument. The way to find
3856 the end of the named arguments is with the built-in functions described
3859 @defmac __builtin_saveregs ()
3860 Use this built-in function to save the argument registers in memory so
3861 that the varargs mechanism can access them. Both ISO and traditional
3862 versions of @code{va_start} must use @code{__builtin_saveregs}, unless
3863 you use @code{TARGET_SETUP_INCOMING_VARARGS} (see below) instead.
3865 On some machines, @code{__builtin_saveregs} is open-coded under the
3866 control of the target hook @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. On
3867 other machines, it calls a routine written in assembler language,
3868 found in @file{libgcc2.c}.
3870 Code generated for the call to @code{__builtin_saveregs} appears at the
3871 beginning of the function, as opposed to where the call to
3872 @code{__builtin_saveregs} is written, regardless of what the code is.
3873 This is because the registers must be saved before the function starts
3874 to use them for its own purposes.
3875 @c i rewrote the first sentence above to fix an overfull hbox. --mew
3879 @defmac __builtin_next_arg (@var{lastarg})
3880 This builtin returns the address of the first anonymous stack
3881 argument, as type @code{void *}. If @code{ARGS_GROW_DOWNWARD}, it
3882 returns the address of the location above the first anonymous stack
3883 argument. Use it in @code{va_start} to initialize the pointer for
3884 fetching arguments from the stack. Also use it in @code{va_start} to
3885 verify that the second parameter @var{lastarg} is the last named argument
3886 of the current function.
3889 @defmac __builtin_classify_type (@var{object})
3890 Since each machine has its own conventions for which data types are
3891 passed in which kind of register, your implementation of @code{va_arg}
3892 has to embody these conventions. The easiest way to categorize the
3893 specified data type is to use @code{__builtin_classify_type} together
3894 with @code{sizeof} and @code{__alignof__}.
3896 @code{__builtin_classify_type} ignores the value of @var{object},
3897 considering only its data type. It returns an integer describing what
3898 kind of type that is---integer, floating, pointer, structure, and so on.
3900 The file @file{typeclass.h} defines an enumeration that you can use to
3901 interpret the values of @code{__builtin_classify_type}.
3904 These machine description macros help implement varargs:
3906 @hook TARGET_EXPAND_BUILTIN_SAVEREGS
3908 @hook TARGET_SETUP_INCOMING_VARARGS
3910 @hook TARGET_STRICT_ARGUMENT_NAMING
3912 @hook TARGET_CALL_ARGS
3914 @hook TARGET_END_CALL_ARGS
3916 @hook TARGET_PRETEND_OUTGOING_VARARGS_NAMED
3918 @hook TARGET_LOAD_BOUNDS_FOR_ARG
3920 @hook TARGET_STORE_BOUNDS_FOR_ARG
3922 @hook TARGET_LOAD_RETURNED_BOUNDS
3924 @hook TARGET_STORE_RETURNED_BOUNDS
3926 @hook TARGET_CHKP_FUNCTION_VALUE_BOUNDS
3928 @hook TARGET_SETUP_INCOMING_VARARG_BOUNDS
3931 @section Trampolines for Nested Functions
3932 @cindex trampolines for nested functions
3933 @cindex nested functions, trampolines for
3935 A @dfn{trampoline} is a small piece of code that is created at run time
3936 when the address of a nested function is taken. It normally resides on
3937 the stack, in the stack frame of the containing function. These macros
3938 tell GCC how to generate code to allocate and initialize a
3941 The instructions in the trampoline must do two things: load a constant
3942 address into the static chain register, and jump to the real address of
3943 the nested function. On CISC machines such as the m68k, this requires
3944 two instructions, a move immediate and a jump. Then the two addresses
3945 exist in the trampoline as word-long immediate operands. On RISC
3946 machines, it is often necessary to load each address into a register in
3947 two parts. Then pieces of each address form separate immediate
3950 The code generated to initialize the trampoline must store the variable
3951 parts---the static chain value and the function address---into the
3952 immediate operands of the instructions. On a CISC machine, this is
3953 simply a matter of copying each address to a memory reference at the
3954 proper offset from the start of the trampoline. On a RISC machine, it
3955 may be necessary to take out pieces of the address and store them
3958 @hook TARGET_ASM_TRAMPOLINE_TEMPLATE
3960 @defmac TRAMPOLINE_SECTION
3961 Return the section into which the trampoline template is to be placed
3962 (@pxref{Sections}). The default value is @code{readonly_data_section}.
3965 @defmac TRAMPOLINE_SIZE
3966 A C expression for the size in bytes of the trampoline, as an integer.
3969 @defmac TRAMPOLINE_ALIGNMENT
3970 Alignment required for trampolines, in bits.
3972 If you don't define this macro, the value of @code{FUNCTION_ALIGNMENT}
3973 is used for aligning trampolines.
3976 @hook TARGET_TRAMPOLINE_INIT
3978 @hook TARGET_TRAMPOLINE_ADJUST_ADDRESS
3980 @hook TARGET_CUSTOM_FUNCTION_DESCRIPTORS
3982 Implementing trampolines is difficult on many machines because they have
3983 separate instruction and data caches. Writing into a stack location
3984 fails to clear the memory in the instruction cache, so when the program
3985 jumps to that location, it executes the old contents.
3987 Here are two possible solutions. One is to clear the relevant parts of
3988 the instruction cache whenever a trampoline is set up. The other is to
3989 make all trampolines identical, by having them jump to a standard
3990 subroutine. The former technique makes trampoline execution faster; the
3991 latter makes initialization faster.
3993 To clear the instruction cache when a trampoline is initialized, define
3994 the following macro.
3996 @defmac CLEAR_INSN_CACHE (@var{beg}, @var{end})
3997 If defined, expands to a C expression clearing the @emph{instruction
3998 cache} in the specified interval. The definition of this macro would
3999 typically be a series of @code{asm} statements. Both @var{beg} and
4000 @var{end} are both pointer expressions.
4003 To use a standard subroutine, define the following macro. In addition,
4004 you must make sure that the instructions in a trampoline fill an entire
4005 cache line with identical instructions, or else ensure that the
4006 beginning of the trampoline code is always aligned at the same point in
4007 its cache line. Look in @file{m68k.h} as a guide.
4009 @defmac TRANSFER_FROM_TRAMPOLINE
4010 Define this macro if trampolines need a special subroutine to do their
4011 work. The macro should expand to a series of @code{asm} statements
4012 which will be compiled with GCC@. They go in a library function named
4013 @code{__transfer_from_trampoline}.
4015 If you need to avoid executing the ordinary prologue code of a compiled
4016 C function when you jump to the subroutine, you can do so by placing a
4017 special label of your own in the assembler code. Use one @code{asm}
4018 statement to generate an assembler label, and another to make the label
4019 global. Then trampolines can use that label to jump directly to your
4020 special assembler code.
4024 @section Implicit Calls to Library Routines
4025 @cindex library subroutine names
4026 @cindex @file{libgcc.a}
4028 @c prevent bad page break with this line
4029 Here is an explanation of implicit calls to library routines.
4031 @defmac DECLARE_LIBRARY_RENAMES
4032 This macro, if defined, should expand to a piece of C code that will get
4033 expanded when compiling functions for libgcc.a. It can be used to
4034 provide alternate names for GCC's internal library functions if there
4035 are ABI-mandated names that the compiler should provide.
4038 @findex set_optab_libfunc
4039 @findex init_one_libfunc
4040 @hook TARGET_INIT_LIBFUNCS
4042 @hook TARGET_LIBFUNC_GNU_PREFIX
4044 @defmac FLOAT_LIB_COMPARE_RETURNS_BOOL (@var{mode}, @var{comparison})
4045 This macro should return @code{true} if the library routine that
4046 implements the floating point comparison operator @var{comparison} in
4047 mode @var{mode} will return a boolean, and @var{false} if it will
4050 GCC's own floating point libraries return tristates from the
4051 comparison operators, so the default returns false always. Most ports
4052 don't need to define this macro.
4055 @defmac TARGET_LIB_INT_CMP_BIASED
4056 This macro should evaluate to @code{true} if the integer comparison
4057 functions (like @code{__cmpdi2}) return 0 to indicate that the first
4058 operand is smaller than the second, 1 to indicate that they are equal,
4059 and 2 to indicate that the first operand is greater than the second.
4060 If this macro evaluates to @code{false} the comparison functions return
4061 @minus{}1, 0, and 1 instead of 0, 1, and 2. If the target uses the routines
4062 in @file{libgcc.a}, you do not need to define this macro.
4065 @defmac TARGET_HAS_NO_HW_DIVIDE
4066 This macro should be defined if the target has no hardware divide
4067 instructions. If this macro is defined, GCC will use an algorithm which
4068 make use of simple logical and arithmetic operations for 64-bit
4069 division. If the macro is not defined, GCC will use an algorithm which
4070 make use of a 64-bit by 32-bit divide primitive.
4073 @cindex @code{EDOM}, implicit usage
4076 The value of @code{EDOM} on the target machine, as a C integer constant
4077 expression. If you don't define this macro, GCC does not attempt to
4078 deposit the value of @code{EDOM} into @code{errno} directly. Look in
4079 @file{/usr/include/errno.h} to find the value of @code{EDOM} on your
4082 If you do not define @code{TARGET_EDOM}, then compiled code reports
4083 domain errors by calling the library function and letting it report the
4084 error. If mathematical functions on your system use @code{matherr} when
4085 there is an error, then you should leave @code{TARGET_EDOM} undefined so
4086 that @code{matherr} is used normally.
4089 @cindex @code{errno}, implicit usage
4090 @defmac GEN_ERRNO_RTX
4091 Define this macro as a C expression to create an rtl expression that
4092 refers to the global ``variable'' @code{errno}. (On certain systems,
4093 @code{errno} may not actually be a variable.) If you don't define this
4094 macro, a reasonable default is used.
4097 @hook TARGET_LIBC_HAS_FUNCTION
4099 @defmac NEXT_OBJC_RUNTIME
4100 Set this macro to 1 to use the "NeXT" Objective-C message sending conventions
4101 by default. This calling convention involves passing the object, the selector
4102 and the method arguments all at once to the method-lookup library function.
4103 This is the usual setting when targeting Darwin/Mac OS X systems, which have
4104 the NeXT runtime installed.
4106 If the macro is set to 0, the "GNU" Objective-C message sending convention
4107 will be used by default. This convention passes just the object and the
4108 selector to the method-lookup function, which returns a pointer to the method.
4110 In either case, it remains possible to select code-generation for the alternate
4111 scheme, by means of compiler command line switches.
4114 @hook TARGET_PRINTF_POINTER_FORMAT
4116 @node Addressing Modes
4117 @section Addressing Modes
4118 @cindex addressing modes
4120 @c prevent bad page break with this line
4121 This is about addressing modes.
4123 @defmac HAVE_PRE_INCREMENT
4124 @defmacx HAVE_PRE_DECREMENT
4125 @defmacx HAVE_POST_INCREMENT
4126 @defmacx HAVE_POST_DECREMENT
4127 A C expression that is nonzero if the machine supports pre-increment,
4128 pre-decrement, post-increment, or post-decrement addressing respectively.
4131 @defmac HAVE_PRE_MODIFY_DISP
4132 @defmacx HAVE_POST_MODIFY_DISP
4133 A C expression that is nonzero if the machine supports pre- or
4134 post-address side-effect generation involving constants other than
4135 the size of the memory operand.
4138 @defmac HAVE_PRE_MODIFY_REG
4139 @defmacx HAVE_POST_MODIFY_REG
4140 A C expression that is nonzero if the machine supports pre- or
4141 post-address side-effect generation involving a register displacement.
4144 @defmac CONSTANT_ADDRESS_P (@var{x})
4145 A C expression that is 1 if the RTX @var{x} is a constant which
4146 is a valid address. On most machines the default definition of
4147 @code{(CONSTANT_P (@var{x}) && GET_CODE (@var{x}) != CONST_DOUBLE)}
4148 is acceptable, but a few machines are more restrictive as to which
4149 constant addresses are supported.
4152 @defmac CONSTANT_P (@var{x})
4153 @code{CONSTANT_P}, which is defined by target-independent code,
4154 accepts integer-values expressions whose values are not explicitly
4155 known, such as @code{symbol_ref}, @code{label_ref}, and @code{high}
4156 expressions and @code{const} arithmetic expressions, in addition to
4157 @code{const_int} and @code{const_double} expressions.
4160 @defmac MAX_REGS_PER_ADDRESS
4161 A number, the maximum number of registers that can appear in a valid
4162 memory address. Note that it is up to you to specify a value equal to
4163 the maximum number that @code{TARGET_LEGITIMATE_ADDRESS_P} would ever
4167 @hook TARGET_LEGITIMATE_ADDRESS_P
4169 @defmac TARGET_MEM_CONSTRAINT
4170 A single character to be used instead of the default @code{'m'}
4171 character for general memory addresses. This defines the constraint
4172 letter which matches the memory addresses accepted by
4173 @code{TARGET_LEGITIMATE_ADDRESS_P}. Define this macro if you want to
4174 support new address formats in your back end without changing the
4175 semantics of the @code{'m'} constraint. This is necessary in order to
4176 preserve functionality of inline assembly constructs using the
4177 @code{'m'} constraint.
4180 @defmac FIND_BASE_TERM (@var{x})
4181 A C expression to determine the base term of address @var{x},
4182 or to provide a simplified version of @var{x} from which @file{alias.c}
4183 can easily find the base term. This macro is used in only two places:
4184 @code{find_base_value} and @code{find_base_term} in @file{alias.c}.
4186 It is always safe for this macro to not be defined. It exists so
4187 that alias analysis can understand machine-dependent addresses.
4189 The typical use of this macro is to handle addresses containing
4190 a label_ref or symbol_ref within an UNSPEC@.
4193 @hook TARGET_LEGITIMIZE_ADDRESS
4195 @defmac LEGITIMIZE_RELOAD_ADDRESS (@var{x}, @var{mode}, @var{opnum}, @var{type}, @var{ind_levels}, @var{win})
4196 A C compound statement that attempts to replace @var{x}, which is an address
4197 that needs reloading, with a valid memory address for an operand of mode
4198 @var{mode}. @var{win} will be a C statement label elsewhere in the code.
4199 It is not necessary to define this macro, but it might be useful for
4200 performance reasons.
4202 For example, on the i386, it is sometimes possible to use a single
4203 reload register instead of two by reloading a sum of two pseudo
4204 registers into a register. On the other hand, for number of RISC
4205 processors offsets are limited so that often an intermediate address
4206 needs to be generated in order to address a stack slot. By defining
4207 @code{LEGITIMIZE_RELOAD_ADDRESS} appropriately, the intermediate addresses
4208 generated for adjacent some stack slots can be made identical, and thus
4211 @emph{Note}: This macro should be used with caution. It is necessary
4212 to know something of how reload works in order to effectively use this,
4213 and it is quite easy to produce macros that build in too much knowledge
4214 of reload internals.
4216 @emph{Note}: This macro must be able to reload an address created by a
4217 previous invocation of this macro. If it fails to handle such addresses
4218 then the compiler may generate incorrect code or abort.
4221 The macro definition should use @code{push_reload} to indicate parts that
4222 need reloading; @var{opnum}, @var{type} and @var{ind_levels} are usually
4223 suitable to be passed unaltered to @code{push_reload}.
4225 The code generated by this macro must not alter the substructure of
4226 @var{x}. If it transforms @var{x} into a more legitimate form, it
4227 should assign @var{x} (which will always be a C variable) a new value.
4228 This also applies to parts that you change indirectly by calling
4231 @findex strict_memory_address_p
4232 The macro definition may use @code{strict_memory_address_p} to test if
4233 the address has become legitimate.
4236 If you want to change only a part of @var{x}, one standard way of doing
4237 this is to use @code{copy_rtx}. Note, however, that it unshares only a
4238 single level of rtl. Thus, if the part to be changed is not at the
4239 top level, you'll need to replace first the top level.
4240 It is not necessary for this macro to come up with a legitimate
4241 address; but often a machine-dependent strategy can generate better code.
4244 @hook TARGET_MODE_DEPENDENT_ADDRESS_P
4246 @hook TARGET_LEGITIMATE_CONSTANT_P
4248 @hook TARGET_DELEGITIMIZE_ADDRESS
4250 @hook TARGET_CONST_NOT_OK_FOR_DEBUG_P
4252 @hook TARGET_CANNOT_FORCE_CONST_MEM
4254 @hook TARGET_USE_BLOCKS_FOR_CONSTANT_P
4256 @hook TARGET_USE_BLOCKS_FOR_DECL_P
4258 @hook TARGET_BUILTIN_RECIPROCAL
4260 @hook TARGET_VECTORIZE_BUILTIN_MASK_FOR_LOAD
4262 @hook TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST
4264 @hook TARGET_VECTORIZE_VECTOR_ALIGNMENT_REACHABLE
4266 @hook TARGET_VECTORIZE_VEC_PERM_CONST_OK
4268 @hook TARGET_VECTORIZE_BUILTIN_CONVERSION
4270 @hook TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION
4272 @hook TARGET_VECTORIZE_BUILTIN_MD_VECTORIZED_FUNCTION
4274 @hook TARGET_VECTORIZE_SUPPORT_VECTOR_MISALIGNMENT
4276 @hook TARGET_VECTORIZE_PREFERRED_SIMD_MODE
4278 @hook TARGET_VECTORIZE_AUTOVECTORIZE_VECTOR_SIZES
4280 @hook TARGET_VECTORIZE_GET_MASK_MODE
4282 @hook TARGET_VECTORIZE_INIT_COST
4284 @hook TARGET_VECTORIZE_ADD_STMT_COST
4286 @hook TARGET_VECTORIZE_FINISH_COST
4288 @hook TARGET_VECTORIZE_DESTROY_COST_DATA
4290 @hook TARGET_VECTORIZE_BUILTIN_GATHER
4292 @hook TARGET_VECTORIZE_BUILTIN_SCATTER
4294 @hook TARGET_SIMD_CLONE_COMPUTE_VECSIZE_AND_SIMDLEN
4296 @hook TARGET_SIMD_CLONE_ADJUST
4298 @hook TARGET_SIMD_CLONE_USABLE
4300 @hook TARGET_SIMT_VF
4302 @hook TARGET_GOACC_VALIDATE_DIMS
4304 @hook TARGET_GOACC_DIM_LIMIT
4306 @hook TARGET_GOACC_FORK_JOIN
4308 @hook TARGET_GOACC_REDUCTION
4310 @node Anchored Addresses
4311 @section Anchored Addresses
4312 @cindex anchored addresses
4313 @cindex @option{-fsection-anchors}
4315 GCC usually addresses every static object as a separate entity.
4316 For example, if we have:
4320 int foo (void) @{ return a + b + c; @}
4323 the code for @code{foo} will usually calculate three separate symbolic
4324 addresses: those of @code{a}, @code{b} and @code{c}. On some targets,
4325 it would be better to calculate just one symbolic address and access
4326 the three variables relative to it. The equivalent pseudocode would
4332 register int *xr = &x;
4333 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
4337 (which isn't valid C). We refer to shared addresses like @code{x} as
4338 ``section anchors''. Their use is controlled by @option{-fsection-anchors}.
4340 The hooks below describe the target properties that GCC needs to know
4341 in order to make effective use of section anchors. It won't use
4342 section anchors at all unless either @code{TARGET_MIN_ANCHOR_OFFSET}
4343 or @code{TARGET_MAX_ANCHOR_OFFSET} is set to a nonzero value.
4345 @hook TARGET_MIN_ANCHOR_OFFSET
4347 @hook TARGET_MAX_ANCHOR_OFFSET
4349 @hook TARGET_ASM_OUTPUT_ANCHOR
4351 @hook TARGET_USE_ANCHORS_FOR_SYMBOL_P
4353 @node Condition Code
4354 @section Condition Code Status
4355 @cindex condition code status
4357 The macros in this section can be split in two families, according to the
4358 two ways of representing condition codes in GCC.
4360 The first representation is the so called @code{(cc0)} representation
4361 (@pxref{Jump Patterns}), where all instructions can have an implicit
4362 clobber of the condition codes. The second is the condition code
4363 register representation, which provides better schedulability for
4364 architectures that do have a condition code register, but on which
4365 most instructions do not affect it. The latter category includes
4368 The implicit clobbering poses a strong restriction on the placement of
4369 the definition and use of the condition code. In the past the definition
4370 and use were always adjacent. However, recent changes to support trapping
4371 arithmatic may result in the definition and user being in different blocks.
4372 Thus, there may be a @code{NOTE_INSN_BASIC_BLOCK} between them. Additionally,
4373 the definition may be the source of exception handling edges.
4375 These restrictions can prevent important
4376 optimizations on some machines. For example, on the IBM RS/6000, there
4377 is a delay for taken branches unless the condition code register is set
4378 three instructions earlier than the conditional branch. The instruction
4379 scheduler cannot perform this optimization if it is not permitted to
4380 separate the definition and use of the condition code register.
4382 For this reason, it is possible and suggested to use a register to
4383 represent the condition code for new ports. If there is a specific
4384 condition code register in the machine, use a hard register. If the
4385 condition code or comparison result can be placed in any general register,
4386 or if there are multiple condition registers, use a pseudo register.
4387 Registers used to store the condition code value will usually have a mode
4388 that is in class @code{MODE_CC}.
4390 Alternatively, you can use @code{BImode} if the comparison operator is
4391 specified already in the compare instruction. In this case, you are not
4392 interested in most macros in this section.
4395 * CC0 Condition Codes:: Old style representation of condition codes.
4396 * MODE_CC Condition Codes:: Modern representation of condition codes.
4399 @node CC0 Condition Codes
4400 @subsection Representation of condition codes using @code{(cc0)}
4404 The file @file{conditions.h} defines a variable @code{cc_status} to
4405 describe how the condition code was computed (in case the interpretation of
4406 the condition code depends on the instruction that it was set by). This
4407 variable contains the RTL expressions on which the condition code is
4408 currently based, and several standard flags.
4410 Sometimes additional machine-specific flags must be defined in the machine
4411 description header file. It can also add additional machine-specific
4412 information by defining @code{CC_STATUS_MDEP}.
4414 @defmac CC_STATUS_MDEP
4415 C code for a data type which is used for declaring the @code{mdep}
4416 component of @code{cc_status}. It defaults to @code{int}.
4418 This macro is not used on machines that do not use @code{cc0}.
4421 @defmac CC_STATUS_MDEP_INIT
4422 A C expression to initialize the @code{mdep} field to ``empty''.
4423 The default definition does nothing, since most machines don't use
4424 the field anyway. If you want to use the field, you should probably
4425 define this macro to initialize it.
4427 This macro is not used on machines that do not use @code{cc0}.
4430 @defmac NOTICE_UPDATE_CC (@var{exp}, @var{insn})
4431 A C compound statement to set the components of @code{cc_status}
4432 appropriately for an insn @var{insn} whose body is @var{exp}. It is
4433 this macro's responsibility to recognize insns that set the condition
4434 code as a byproduct of other activity as well as those that explicitly
4437 This macro is not used on machines that do not use @code{cc0}.
4439 If there are insns that do not set the condition code but do alter
4440 other machine registers, this macro must check to see whether they
4441 invalidate the expressions that the condition code is recorded as
4442 reflecting. For example, on the 68000, insns that store in address
4443 registers do not set the condition code, which means that usually
4444 @code{NOTICE_UPDATE_CC} can leave @code{cc_status} unaltered for such
4445 insns. But suppose that the previous insn set the condition code
4446 based on location @samp{a4@@(102)} and the current insn stores a new
4447 value in @samp{a4}. Although the condition code is not changed by
4448 this, it will no longer be true that it reflects the contents of
4449 @samp{a4@@(102)}. Therefore, @code{NOTICE_UPDATE_CC} must alter
4450 @code{cc_status} in this case to say that nothing is known about the
4451 condition code value.
4453 The definition of @code{NOTICE_UPDATE_CC} must be prepared to deal
4454 with the results of peephole optimization: insns whose patterns are
4455 @code{parallel} RTXs containing various @code{reg}, @code{mem} or
4456 constants which are just the operands. The RTL structure of these
4457 insns is not sufficient to indicate what the insns actually do. What
4458 @code{NOTICE_UPDATE_CC} should do when it sees one is just to run
4459 @code{CC_STATUS_INIT}.
4461 A possible definition of @code{NOTICE_UPDATE_CC} is to call a function
4462 that looks at an attribute (@pxref{Insn Attributes}) named, for example,
4463 @samp{cc}. This avoids having detailed information about patterns in
4464 two places, the @file{md} file and in @code{NOTICE_UPDATE_CC}.
4467 @node MODE_CC Condition Codes
4468 @subsection Representation of condition codes using registers
4472 @defmac SELECT_CC_MODE (@var{op}, @var{x}, @var{y})
4473 On many machines, the condition code may be produced by other instructions
4474 than compares, for example the branch can use directly the condition
4475 code set by a subtract instruction. However, on some machines
4476 when the condition code is set this way some bits (such as the overflow
4477 bit) are not set in the same way as a test instruction, so that a different
4478 branch instruction must be used for some conditional branches. When
4479 this happens, use the machine mode of the condition code register to
4480 record different formats of the condition code register. Modes can
4481 also be used to record which compare instruction (e.g. a signed or an
4482 unsigned comparison) produced the condition codes.
4484 If other modes than @code{CCmode} are required, add them to
4485 @file{@var{machine}-modes.def} and define @code{SELECT_CC_MODE} to choose
4486 a mode given an operand of a compare. This is needed because the modes
4487 have to be chosen not only during RTL generation but also, for example,
4488 by instruction combination. The result of @code{SELECT_CC_MODE} should
4489 be consistent with the mode used in the patterns; for example to support
4490 the case of the add on the SPARC discussed above, we have the pattern
4496 (plus:SI (match_operand:SI 0 "register_operand" "%r")
4497 (match_operand:SI 1 "arith_operand" "rI"))
4504 together with a @code{SELECT_CC_MODE} that returns @code{CCNZmode}
4505 for comparisons whose argument is a @code{plus}:
4508 #define SELECT_CC_MODE(OP,X,Y) \
4509 (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \
4510 ? ((OP == LT || OP == LE || OP == GT || OP == GE) \
4511 ? CCFPEmode : CCFPmode) \
4512 : ((GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \
4513 || GET_CODE (X) == NEG || GET_CODE (x) == ASHIFT) \
4514 ? CCNZmode : CCmode))
4517 Another reason to use modes is to retain information on which operands
4518 were used by the comparison; see @code{REVERSIBLE_CC_MODE} later in
4521 You should define this macro if and only if you define extra CC modes
4522 in @file{@var{machine}-modes.def}.
4525 @hook TARGET_CANONICALIZE_COMPARISON
4527 @defmac REVERSIBLE_CC_MODE (@var{mode})
4528 A C expression whose value is one if it is always safe to reverse a
4529 comparison whose mode is @var{mode}. If @code{SELECT_CC_MODE}
4530 can ever return @var{mode} for a floating-point inequality comparison,
4531 then @code{REVERSIBLE_CC_MODE (@var{mode})} must be zero.
4533 You need not define this macro if it would always returns zero or if the
4534 floating-point format is anything other than @code{IEEE_FLOAT_FORMAT}.
4535 For example, here is the definition used on the SPARC, where floating-point
4536 inequality comparisons are given either @code{CCFPEmode} or @code{CCFPmode}:
4539 #define REVERSIBLE_CC_MODE(MODE) \
4540 ((MODE) != CCFPEmode && (MODE) != CCFPmode)
4544 @defmac REVERSE_CONDITION (@var{code}, @var{mode})
4545 A C expression whose value is reversed condition code of the @var{code} for
4546 comparison done in CC_MODE @var{mode}. The macro is used only in case
4547 @code{REVERSIBLE_CC_MODE (@var{mode})} is nonzero. Define this macro in case
4548 machine has some non-standard way how to reverse certain conditionals. For
4549 instance in case all floating point conditions are non-trapping, compiler may
4550 freely convert unordered compares to ordered ones. Then definition may look
4554 #define REVERSE_CONDITION(CODE, MODE) \
4555 ((MODE) != CCFPmode ? reverse_condition (CODE) \
4556 : reverse_condition_maybe_unordered (CODE))
4560 @hook TARGET_FIXED_CONDITION_CODE_REGS
4562 @hook TARGET_CC_MODES_COMPATIBLE
4564 @hook TARGET_FLAGS_REGNUM
4567 @section Describing Relative Costs of Operations
4568 @cindex costs of instructions
4569 @cindex relative costs
4570 @cindex speed of instructions
4572 These macros let you describe the relative speed of various operations
4573 on the target machine.
4575 @defmac REGISTER_MOVE_COST (@var{mode}, @var{from}, @var{to})
4576 A C expression for the cost of moving data of mode @var{mode} from a
4577 register in class @var{from} to one in class @var{to}. The classes are
4578 expressed using the enumeration values such as @code{GENERAL_REGS}. A
4579 value of 2 is the default; other values are interpreted relative to
4582 It is not required that the cost always equal 2 when @var{from} is the
4583 same as @var{to}; on some machines it is expensive to move between
4584 registers if they are not general registers.
4586 If reload sees an insn consisting of a single @code{set} between two
4587 hard registers, and if @code{REGISTER_MOVE_COST} applied to their
4588 classes returns a value of 2, reload does not check to ensure that the
4589 constraints of the insn are met. Setting a cost of other than 2 will
4590 allow reload to verify that the constraints are met. You should do this
4591 if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
4593 These macros are obsolete, new ports should use the target hook
4594 @code{TARGET_REGISTER_MOVE_COST} instead.
4597 @hook TARGET_REGISTER_MOVE_COST
4599 @defmac MEMORY_MOVE_COST (@var{mode}, @var{class}, @var{in})
4600 A C expression for the cost of moving data of mode @var{mode} between a
4601 register of class @var{class} and memory; @var{in} is zero if the value
4602 is to be written to memory, nonzero if it is to be read in. This cost
4603 is relative to those in @code{REGISTER_MOVE_COST}. If moving between
4604 registers and memory is more expensive than between two registers, you
4605 should define this macro to express the relative cost.
4607 If you do not define this macro, GCC uses a default cost of 4 plus
4608 the cost of copying via a secondary reload register, if one is
4609 needed. If your machine requires a secondary reload register to copy
4610 between memory and a register of @var{class} but the reload mechanism is
4611 more complex than copying via an intermediate, define this macro to
4612 reflect the actual cost of the move.
4614 GCC defines the function @code{memory_move_secondary_cost} if
4615 secondary reloads are needed. It computes the costs due to copying via
4616 a secondary register. If your machine copies from memory using a
4617 secondary register in the conventional way but the default base value of
4618 4 is not correct for your machine, define this macro to add some other
4619 value to the result of that function. The arguments to that function
4620 are the same as to this macro.
4622 These macros are obsolete, new ports should use the target hook
4623 @code{TARGET_MEMORY_MOVE_COST} instead.
4626 @hook TARGET_MEMORY_MOVE_COST
4628 @defmac BRANCH_COST (@var{speed_p}, @var{predictable_p})
4629 A C expression for the cost of a branch instruction. A value of 1 is
4630 the default; other values are interpreted relative to that. Parameter
4631 @var{speed_p} is true when the branch in question should be optimized
4632 for speed. When it is false, @code{BRANCH_COST} should return a value
4633 optimal for code size rather than performance. @var{predictable_p} is
4634 true for well-predicted branches. On many architectures the
4635 @code{BRANCH_COST} can be reduced then.
4638 Here are additional macros which do not specify precise relative costs,
4639 but only that certain actions are more expensive than GCC would
4642 @defmac SLOW_BYTE_ACCESS
4643 Define this macro as a C expression which is nonzero if accessing less
4644 than a word of memory (i.e.@: a @code{char} or a @code{short}) is no
4645 faster than accessing a word of memory, i.e., if such access
4646 require more than one instruction or if there is no difference in cost
4647 between byte and (aligned) word loads.
4649 When this macro is not defined, the compiler will access a field by
4650 finding the smallest containing object; when it is defined, a fullword
4651 load will be used if alignment permits. Unless bytes accesses are
4652 faster than word accesses, using word accesses is preferable since it
4653 may eliminate subsequent memory access if subsequent accesses occur to
4654 other fields in the same word of the structure, but to different bytes.
4657 @defmac SLOW_UNALIGNED_ACCESS (@var{mode}, @var{alignment})
4658 Define this macro to be the value 1 if memory accesses described by the
4659 @var{mode} and @var{alignment} parameters have a cost many times greater
4660 than aligned accesses, for example if they are emulated in a trap
4661 handler. This macro is invoked only for unaligned accesses, i.e. when
4662 @code{@var{alignment} < GET_MODE_ALIGNMENT (@var{mode})}.
4664 When this macro is nonzero, the compiler will act as if
4665 @code{STRICT_ALIGNMENT} were nonzero when generating code for block
4666 moves. This can cause significantly more instructions to be produced.
4667 Therefore, do not set this macro nonzero if unaligned accesses only add a
4668 cycle or two to the time for a memory access.
4670 If the value of this macro is always zero, it need not be defined. If
4671 this macro is defined, it should produce a nonzero value when
4672 @code{STRICT_ALIGNMENT} is nonzero.
4675 @defmac MOVE_RATIO (@var{speed})
4676 The threshold of number of scalar memory-to-memory move insns, @emph{below}
4677 which a sequence of insns should be generated instead of a
4678 string move insn or a library call. Increasing the value will always
4679 make code faster, but eventually incurs high cost in increased code size.
4681 Note that on machines where the corresponding move insn is a
4682 @code{define_expand} that emits a sequence of insns, this macro counts
4683 the number of such sequences.
4685 The parameter @var{speed} is true if the code is currently being
4686 optimized for speed rather than size.
4688 If you don't define this, a reasonable default is used.
4691 @hook TARGET_USE_BY_PIECES_INFRASTRUCTURE_P
4693 @hook TARGET_COMPARE_BY_PIECES_BRANCH_RATIO
4695 @defmac MOVE_MAX_PIECES
4696 A C expression used by @code{move_by_pieces} to determine the largest unit
4697 a load or store used to copy memory is. Defaults to @code{MOVE_MAX}.
4700 @defmac STORE_MAX_PIECES
4701 A C expression used by @code{store_by_pieces} to determine the largest unit
4702 a store used to memory is. Defaults to @code{MOVE_MAX_PIECES}, or two times
4703 the size of @code{HOST_WIDE_INT}, whichever is smaller.
4706 @defmac COMPARE_MAX_PIECES
4707 A C expression used by @code{compare_by_pieces} to determine the largest unit
4708 a load or store used to compare memory is. Defaults to
4709 @code{MOVE_MAX_PIECES}.
4712 @defmac CLEAR_RATIO (@var{speed})
4713 The threshold of number of scalar move insns, @emph{below} which a sequence
4714 of insns should be generated to clear memory instead of a string clear insn
4715 or a library call. Increasing the value will always make code faster, but
4716 eventually incurs high cost in increased code size.
4718 The parameter @var{speed} is true if the code is currently being
4719 optimized for speed rather than size.
4721 If you don't define this, a reasonable default is used.
4724 @defmac SET_RATIO (@var{speed})
4725 The threshold of number of scalar move insns, @emph{below} which a sequence
4726 of insns should be generated to set memory to a constant value, instead of
4727 a block set insn or a library call.
4728 Increasing the value will always make code faster, but
4729 eventually incurs high cost in increased code size.
4731 The parameter @var{speed} is true if the code is currently being
4732 optimized for speed rather than size.
4734 If you don't define this, it defaults to the value of @code{MOVE_RATIO}.
4737 @defmac USE_LOAD_POST_INCREMENT (@var{mode})
4738 A C expression used to determine whether a load postincrement is a good
4739 thing to use for a given mode. Defaults to the value of
4740 @code{HAVE_POST_INCREMENT}.
4743 @defmac USE_LOAD_POST_DECREMENT (@var{mode})
4744 A C expression used to determine whether a load postdecrement is a good
4745 thing to use for a given mode. Defaults to the value of
4746 @code{HAVE_POST_DECREMENT}.
4749 @defmac USE_LOAD_PRE_INCREMENT (@var{mode})
4750 A C expression used to determine whether a load preincrement is a good
4751 thing to use for a given mode. Defaults to the value of
4752 @code{HAVE_PRE_INCREMENT}.
4755 @defmac USE_LOAD_PRE_DECREMENT (@var{mode})
4756 A C expression used to determine whether a load predecrement is a good
4757 thing to use for a given mode. Defaults to the value of
4758 @code{HAVE_PRE_DECREMENT}.
4761 @defmac USE_STORE_POST_INCREMENT (@var{mode})
4762 A C expression used to determine whether a store postincrement is a good
4763 thing to use for a given mode. Defaults to the value of
4764 @code{HAVE_POST_INCREMENT}.
4767 @defmac USE_STORE_POST_DECREMENT (@var{mode})
4768 A C expression used to determine whether a store postdecrement is a good
4769 thing to use for a given mode. Defaults to the value of
4770 @code{HAVE_POST_DECREMENT}.
4773 @defmac USE_STORE_PRE_INCREMENT (@var{mode})
4774 This macro is used to determine whether a store preincrement is a good
4775 thing to use for a given mode. Defaults to the value of
4776 @code{HAVE_PRE_INCREMENT}.
4779 @defmac USE_STORE_PRE_DECREMENT (@var{mode})
4780 This macro is used to determine whether a store predecrement is a good
4781 thing to use for a given mode. Defaults to the value of
4782 @code{HAVE_PRE_DECREMENT}.
4785 @defmac NO_FUNCTION_CSE
4786 Define this macro to be true if it is as good or better to call a constant
4787 function address than to call an address kept in a register.
4790 @defmac LOGICAL_OP_NON_SHORT_CIRCUIT
4791 Define this macro if a non-short-circuit operation produced by
4792 @samp{fold_range_test ()} is optimal. This macro defaults to true if
4793 @code{BRANCH_COST} is greater than or equal to the value 2.
4796 @hook TARGET_OPTAB_SUPPORTED_P
4798 @hook TARGET_RTX_COSTS
4800 @hook TARGET_ADDRESS_COST
4802 @hook TARGET_MAX_NOCE_IFCVT_SEQ_COST
4804 @hook TARGET_NO_SPECULATION_IN_DELAY_SLOTS_P
4807 @section Adjusting the Instruction Scheduler
4809 The instruction scheduler may need a fair amount of machine-specific
4810 adjustment in order to produce good code. GCC provides several target
4811 hooks for this purpose. It is usually enough to define just a few of
4812 them: try the first ones in this list first.
4814 @hook TARGET_SCHED_ISSUE_RATE
4816 @hook TARGET_SCHED_VARIABLE_ISSUE
4818 @hook TARGET_SCHED_ADJUST_COST
4820 @hook TARGET_SCHED_ADJUST_PRIORITY
4822 @hook TARGET_SCHED_REORDER
4824 @hook TARGET_SCHED_REORDER2
4826 @hook TARGET_SCHED_MACRO_FUSION_P
4828 @hook TARGET_SCHED_MACRO_FUSION_PAIR_P
4830 @hook TARGET_SCHED_DEPENDENCIES_EVALUATION_HOOK
4832 @hook TARGET_SCHED_INIT
4834 @hook TARGET_SCHED_FINISH
4836 @hook TARGET_SCHED_INIT_GLOBAL
4838 @hook TARGET_SCHED_FINISH_GLOBAL
4840 @hook TARGET_SCHED_DFA_PRE_CYCLE_INSN
4842 @hook TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN
4844 @hook TARGET_SCHED_DFA_POST_CYCLE_INSN
4846 @hook TARGET_SCHED_INIT_DFA_POST_CYCLE_INSN
4848 @hook TARGET_SCHED_DFA_PRE_ADVANCE_CYCLE
4850 @hook TARGET_SCHED_DFA_POST_ADVANCE_CYCLE
4852 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD
4854 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD
4856 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_BEGIN
4858 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_ISSUE
4860 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_BACKTRACK
4862 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_END
4864 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_INIT
4866 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_FINI
4868 @hook TARGET_SCHED_DFA_NEW_CYCLE
4870 @hook TARGET_SCHED_IS_COSTLY_DEPENDENCE
4872 @hook TARGET_SCHED_H_I_D_EXTENDED
4874 @hook TARGET_SCHED_ALLOC_SCHED_CONTEXT
4876 @hook TARGET_SCHED_INIT_SCHED_CONTEXT
4878 @hook TARGET_SCHED_SET_SCHED_CONTEXT
4880 @hook TARGET_SCHED_CLEAR_SCHED_CONTEXT
4882 @hook TARGET_SCHED_FREE_SCHED_CONTEXT
4884 @hook TARGET_SCHED_SPECULATE_INSN
4886 @hook TARGET_SCHED_NEEDS_BLOCK_P
4888 @hook TARGET_SCHED_GEN_SPEC_CHECK
4890 @hook TARGET_SCHED_SET_SCHED_FLAGS
4892 @hook TARGET_SCHED_SMS_RES_MII
4894 @hook TARGET_SCHED_DISPATCH
4896 @hook TARGET_SCHED_DISPATCH_DO
4898 @hook TARGET_SCHED_EXPOSED_PIPELINE
4900 @hook TARGET_SCHED_REASSOCIATION_WIDTH
4902 @hook TARGET_SCHED_FUSION_PRIORITY
4904 @hook TARGET_EXPAND_DIVMOD_LIBFUNC
4907 @section Dividing the Output into Sections (Texts, Data, @dots{})
4908 @c the above section title is WAY too long. maybe cut the part between
4909 @c the (...)? --mew 10feb93
4911 An object file is divided into sections containing different types of
4912 data. In the most common case, there are three sections: the @dfn{text
4913 section}, which holds instructions and read-only data; the @dfn{data
4914 section}, which holds initialized writable data; and the @dfn{bss
4915 section}, which holds uninitialized data. Some systems have other kinds
4918 @file{varasm.c} provides several well-known sections, such as
4919 @code{text_section}, @code{data_section} and @code{bss_section}.
4920 The normal way of controlling a @code{@var{foo}_section} variable
4921 is to define the associated @code{@var{FOO}_SECTION_ASM_OP} macro,
4922 as described below. The macros are only read once, when @file{varasm.c}
4923 initializes itself, so their values must be run-time constants.
4924 They may however depend on command-line flags.
4926 @emph{Note:} Some run-time files, such @file{crtstuff.c}, also make
4927 use of the @code{@var{FOO}_SECTION_ASM_OP} macros, and expect them
4928 to be string literals.
4930 Some assemblers require a different string to be written every time a
4931 section is selected. If your assembler falls into this category, you
4932 should define the @code{TARGET_ASM_INIT_SECTIONS} hook and use
4933 @code{get_unnamed_section} to set up the sections.
4935 You must always create a @code{text_section}, either by defining
4936 @code{TEXT_SECTION_ASM_OP} or by initializing @code{text_section}
4937 in @code{TARGET_ASM_INIT_SECTIONS}. The same is true of
4938 @code{data_section} and @code{DATA_SECTION_ASM_OP}. If you do not
4939 create a distinct @code{readonly_data_section}, the default is to
4940 reuse @code{text_section}.
4942 All the other @file{varasm.c} sections are optional, and are null
4943 if the target does not provide them.
4945 @defmac TEXT_SECTION_ASM_OP
4946 A C expression whose value is a string, including spacing, containing the
4947 assembler operation that should precede instructions and read-only data.
4948 Normally @code{"\t.text"} is right.
4951 @defmac HOT_TEXT_SECTION_NAME
4952 If defined, a C string constant for the name of the section containing most
4953 frequently executed functions of the program. If not defined, GCC will provide
4954 a default definition if the target supports named sections.
4957 @defmac UNLIKELY_EXECUTED_TEXT_SECTION_NAME
4958 If defined, a C string constant for the name of the section containing unlikely
4959 executed functions in the program.
4962 @defmac DATA_SECTION_ASM_OP
4963 A C expression whose value is a string, including spacing, containing the
4964 assembler operation to identify the following data as writable initialized
4965 data. Normally @code{"\t.data"} is right.
4968 @defmac SDATA_SECTION_ASM_OP
4969 If defined, a C expression whose value is a string, including spacing,
4970 containing the assembler operation to identify the following data as
4971 initialized, writable small data.
4974 @defmac READONLY_DATA_SECTION_ASM_OP
4975 A C expression whose value is a string, including spacing, containing the
4976 assembler operation to identify the following data as read-only initialized
4980 @defmac BSS_SECTION_ASM_OP
4981 If defined, a C expression whose value is a string, including spacing,
4982 containing the assembler operation to identify the following data as
4983 uninitialized global data. If not defined, and
4984 @code{ASM_OUTPUT_ALIGNED_BSS} not defined,
4985 uninitialized global data will be output in the data section if
4986 @option{-fno-common} is passed, otherwise @code{ASM_OUTPUT_COMMON} will be
4990 @defmac SBSS_SECTION_ASM_OP
4991 If defined, a C expression whose value is a string, including spacing,
4992 containing the assembler operation to identify the following data as
4993 uninitialized, writable small data.
4996 @defmac TLS_COMMON_ASM_OP
4997 If defined, a C expression whose value is a string containing the
4998 assembler operation to identify the following data as thread-local
4999 common data. The default is @code{".tls_common"}.
5002 @defmac TLS_SECTION_ASM_FLAG
5003 If defined, a C expression whose value is a character constant
5004 containing the flag used to mark a section as a TLS section. The
5005 default is @code{'T'}.
5008 @defmac INIT_SECTION_ASM_OP
5009 If defined, a C expression whose value is a string, including spacing,
5010 containing the assembler operation to identify the following data as
5011 initialization code. If not defined, GCC will assume such a section does
5012 not exist. This section has no corresponding @code{init_section}
5013 variable; it is used entirely in runtime code.
5016 @defmac FINI_SECTION_ASM_OP
5017 If defined, a C expression whose value is a string, including spacing,
5018 containing the assembler operation to identify the following data as
5019 finalization code. If not defined, GCC will assume such a section does
5020 not exist. This section has no corresponding @code{fini_section}
5021 variable; it is used entirely in runtime code.
5024 @defmac INIT_ARRAY_SECTION_ASM_OP
5025 If defined, a C expression whose value is a string, including spacing,
5026 containing the assembler operation to identify the following data as
5027 part of the @code{.init_array} (or equivalent) section. If not
5028 defined, GCC will assume such a section does not exist. Do not define
5029 both this macro and @code{INIT_SECTION_ASM_OP}.
5032 @defmac FINI_ARRAY_SECTION_ASM_OP
5033 If defined, a C expression whose value is a string, including spacing,
5034 containing the assembler operation to identify the following data as
5035 part of the @code{.fini_array} (or equivalent) section. If not
5036 defined, GCC will assume such a section does not exist. Do not define
5037 both this macro and @code{FINI_SECTION_ASM_OP}.
5040 @defmac MACH_DEP_SECTION_ASM_FLAG
5041 If defined, a C expression whose value is a character constant
5042 containing the flag used to mark a machine-dependent section. This
5043 corresponds to the @code{SECTION_MACH_DEP} section flag.
5046 @defmac CRT_CALL_STATIC_FUNCTION (@var{section_op}, @var{function})
5047 If defined, an ASM statement that switches to a different section
5048 via @var{section_op}, calls @var{function}, and switches back to
5049 the text section. This is used in @file{crtstuff.c} if
5050 @code{INIT_SECTION_ASM_OP} or @code{FINI_SECTION_ASM_OP} to calls
5051 to initialization and finalization functions from the init and fini
5052 sections. By default, this macro uses a simple function call. Some
5053 ports need hand-crafted assembly code to avoid dependencies on
5054 registers initialized in the function prologue or to ensure that
5055 constant pools don't end up too far way in the text section.
5058 @defmac TARGET_LIBGCC_SDATA_SECTION
5059 If defined, a string which names the section into which small
5060 variables defined in crtstuff and libgcc should go. This is useful
5061 when the target has options for optimizing access to small data, and
5062 you want the crtstuff and libgcc routines to be conservative in what
5063 they expect of your application yet liberal in what your application
5064 expects. For example, for targets with a @code{.sdata} section (like
5065 MIPS), you could compile crtstuff with @code{-G 0} so that it doesn't
5066 require small data support from your application, but use this macro
5067 to put small data into @code{.sdata} so that your application can
5068 access these variables whether it uses small data or not.
5071 @defmac FORCE_CODE_SECTION_ALIGN
5072 If defined, an ASM statement that aligns a code section to some
5073 arbitrary boundary. This is used to force all fragments of the
5074 @code{.init} and @code{.fini} sections to have to same alignment
5075 and thus prevent the linker from having to add any padding.
5078 @defmac JUMP_TABLES_IN_TEXT_SECTION
5079 Define this macro to be an expression with a nonzero value if jump
5080 tables (for @code{tablejump} insns) should be output in the text
5081 section, along with the assembler instructions. Otherwise, the
5082 readonly data section is used.
5084 This macro is irrelevant if there is no separate readonly data section.
5087 @hook TARGET_ASM_INIT_SECTIONS
5089 @hook TARGET_ASM_RELOC_RW_MASK
5091 @hook TARGET_ASM_SELECT_SECTION
5093 @defmac USE_SELECT_SECTION_FOR_FUNCTIONS
5094 Define this macro if you wish TARGET_ASM_SELECT_SECTION to be called
5095 for @code{FUNCTION_DECL}s as well as for variables and constants.
5097 In the case of a @code{FUNCTION_DECL}, @var{reloc} will be zero if the
5098 function has been determined to be likely to be called, and nonzero if
5099 it is unlikely to be called.
5102 @hook TARGET_ASM_UNIQUE_SECTION
5104 @hook TARGET_ASM_FUNCTION_RODATA_SECTION
5106 @hook TARGET_ASM_MERGEABLE_RODATA_PREFIX
5108 @hook TARGET_ASM_TM_CLONE_TABLE_SECTION
5110 @hook TARGET_ASM_SELECT_RTX_SECTION
5112 @hook TARGET_MANGLE_DECL_ASSEMBLER_NAME
5114 @hook TARGET_ENCODE_SECTION_INFO
5116 @hook TARGET_STRIP_NAME_ENCODING
5118 @hook TARGET_IN_SMALL_DATA_P
5120 @hook TARGET_HAVE_SRODATA_SECTION
5122 @hook TARGET_PROFILE_BEFORE_PROLOGUE
5124 @hook TARGET_BINDS_LOCAL_P
5126 @hook TARGET_HAVE_TLS
5130 @section Position Independent Code
5131 @cindex position independent code
5134 This section describes macros that help implement generation of position
5135 independent code. Simply defining these macros is not enough to
5136 generate valid PIC; you must also add support to the hook
5137 @code{TARGET_LEGITIMATE_ADDRESS_P} and to the macro
5138 @code{PRINT_OPERAND_ADDRESS}, as well as @code{LEGITIMIZE_ADDRESS}. You
5139 must modify the definition of @samp{movsi} to do something appropriate
5140 when the source operand contains a symbolic address. You may also
5141 need to alter the handling of switch statements so that they use
5143 @c i rearranged the order of the macros above to try to force one of
5144 @c them to the next line, to eliminate an overfull hbox. --mew 10feb93
5146 @defmac PIC_OFFSET_TABLE_REGNUM
5147 The register number of the register used to address a table of static
5148 data addresses in memory. In some cases this register is defined by a
5149 processor's ``application binary interface'' (ABI)@. When this macro
5150 is defined, RTL is generated for this register once, as with the stack
5151 pointer and frame pointer registers. If this macro is not defined, it
5152 is up to the machine-dependent files to allocate such a register (if
5153 necessary). Note that this register must be fixed when in use (e.g.@:
5154 when @code{flag_pic} is true).
5157 @defmac PIC_OFFSET_TABLE_REG_CALL_CLOBBERED
5158 A C expression that is nonzero if the register defined by
5159 @code{PIC_OFFSET_TABLE_REGNUM} is clobbered by calls. If not defined,
5160 the default is zero. Do not define
5161 this macro if @code{PIC_OFFSET_TABLE_REGNUM} is not defined.
5164 @defmac LEGITIMATE_PIC_OPERAND_P (@var{x})
5165 A C expression that is nonzero if @var{x} is a legitimate immediate
5166 operand on the target machine when generating position independent code.
5167 You can assume that @var{x} satisfies @code{CONSTANT_P}, so you need not
5168 check this. You can also assume @var{flag_pic} is true, so you need not
5169 check it either. You need not define this macro if all constants
5170 (including @code{SYMBOL_REF}) can be immediate operands when generating
5171 position independent code.
5174 @node Assembler Format
5175 @section Defining the Output Assembler Language
5177 This section describes macros whose principal purpose is to describe how
5178 to write instructions in assembler language---rather than what the
5182 * File Framework:: Structural information for the assembler file.
5183 * Data Output:: Output of constants (numbers, strings, addresses).
5184 * Uninitialized Data:: Output of uninitialized variables.
5185 * Label Output:: Output and generation of labels.
5186 * Initialization:: General principles of initialization
5187 and termination routines.
5188 * Macros for Initialization::
5189 Specific macros that control the handling of
5190 initialization and termination routines.
5191 * Instruction Output:: Output of actual instructions.
5192 * Dispatch Tables:: Output of jump tables.
5193 * Exception Region Output:: Output of exception region code.
5194 * Alignment Output:: Pseudo ops for alignment and skipping data.
5197 @node File Framework
5198 @subsection The Overall Framework of an Assembler File
5199 @cindex assembler format
5200 @cindex output of assembler code
5202 @c prevent bad page break with this line
5203 This describes the overall framework of an assembly file.
5205 @findex default_file_start
5206 @hook TARGET_ASM_FILE_START
5208 @hook TARGET_ASM_FILE_START_APP_OFF
5210 @hook TARGET_ASM_FILE_START_FILE_DIRECTIVE
5212 @hook TARGET_ASM_FILE_END
5214 @deftypefun void file_end_indicate_exec_stack ()
5215 Some systems use a common convention, the @samp{.note.GNU-stack}
5216 special section, to indicate whether or not an object file relies on
5217 the stack being executable. If your system uses this convention, you
5218 should define @code{TARGET_ASM_FILE_END} to this function. If you
5219 need to do other things in that hook, have your hook function call
5223 @hook TARGET_ASM_LTO_START
5225 @hook TARGET_ASM_LTO_END
5227 @hook TARGET_ASM_CODE_END
5229 @defmac ASM_COMMENT_START
5230 A C string constant describing how to begin a comment in the target
5231 assembler language. The compiler assumes that the comment will end at
5232 the end of the line.
5236 A C string constant for text to be output before each @code{asm}
5237 statement or group of consecutive ones. Normally this is
5238 @code{"#APP"}, which is a comment that has no effect on most
5239 assemblers but tells the GNU assembler that it must check the lines
5240 that follow for all valid assembler constructs.
5244 A C string constant for text to be output after each @code{asm}
5245 statement or group of consecutive ones. Normally this is
5246 @code{"#NO_APP"}, which tells the GNU assembler to resume making the
5247 time-saving assumptions that are valid for ordinary compiler output.
5250 @defmac ASM_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name})
5251 A C statement to output COFF information or DWARF debugging information
5252 which indicates that filename @var{name} is the current source file to
5253 the stdio stream @var{stream}.
5255 This macro need not be defined if the standard form of output
5256 for the file format in use is appropriate.
5259 @hook TARGET_ASM_OUTPUT_SOURCE_FILENAME
5261 @hook TARGET_ASM_OUTPUT_IDENT
5263 @defmac OUTPUT_QUOTED_STRING (@var{stream}, @var{string})
5264 A C statement to output the string @var{string} to the stdio stream
5265 @var{stream}. If you do not call the function @code{output_quoted_string}
5266 in your config files, GCC will only call it to output filenames to
5267 the assembler source. So you can use it to canonicalize the format
5268 of the filename using this macro.
5271 @hook TARGET_ASM_NAMED_SECTION
5273 @hook TARGET_ASM_ELF_FLAGS_NUMERIC
5275 @hook TARGET_ASM_FUNCTION_SECTION
5277 @hook TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS
5279 @hook TARGET_HAVE_NAMED_SECTIONS
5280 This flag is true if the target supports @code{TARGET_ASM_NAMED_SECTION}.
5281 It must not be modified by command-line option processing.
5284 @anchor{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}
5285 @hook TARGET_HAVE_SWITCHABLE_BSS_SECTIONS
5287 @hook TARGET_SECTION_TYPE_FLAGS
5289 @hook TARGET_ASM_RECORD_GCC_SWITCHES
5291 @hook TARGET_ASM_RECORD_GCC_SWITCHES_SECTION
5295 @subsection Output of Data
5298 @hook TARGET_ASM_BYTE_OP
5300 @hook TARGET_ASM_INTEGER
5302 @hook TARGET_ASM_DECL_END
5304 @hook TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA
5306 @defmac ASM_OUTPUT_ASCII (@var{stream}, @var{ptr}, @var{len})
5307 A C statement to output to the stdio stream @var{stream} an assembler
5308 instruction to assemble a string constant containing the @var{len}
5309 bytes at @var{ptr}. @var{ptr} will be a C expression of type
5310 @code{char *} and @var{len} a C expression of type @code{int}.
5312 If the assembler has a @code{.ascii} pseudo-op as found in the
5313 Berkeley Unix assembler, do not define the macro
5314 @code{ASM_OUTPUT_ASCII}.
5317 @defmac ASM_OUTPUT_FDESC (@var{stream}, @var{decl}, @var{n})
5318 A C statement to output word @var{n} of a function descriptor for
5319 @var{decl}. This must be defined if @code{TARGET_VTABLE_USES_DESCRIPTORS}
5320 is defined, and is otherwise unused.
5323 @defmac CONSTANT_POOL_BEFORE_FUNCTION
5324 You may define this macro as a C expression. You should define the
5325 expression to have a nonzero value if GCC should output the constant
5326 pool for a function before the code for the function, or a zero value if
5327 GCC should output the constant pool after the function. If you do
5328 not define this macro, the usual case, GCC will output the constant
5329 pool before the function.
5332 @defmac ASM_OUTPUT_POOL_PROLOGUE (@var{file}, @var{funname}, @var{fundecl}, @var{size})
5333 A C statement to output assembler commands to define the start of the
5334 constant pool for a function. @var{funname} is a string giving
5335 the name of the function. Should the return type of the function
5336 be required, it can be obtained via @var{fundecl}. @var{size}
5337 is the size, in bytes, of the constant pool that will be written
5338 immediately after this call.
5340 If no constant-pool prefix is required, the usual case, this macro need
5344 @defmac ASM_OUTPUT_SPECIAL_POOL_ENTRY (@var{file}, @var{x}, @var{mode}, @var{align}, @var{labelno}, @var{jumpto})
5345 A C statement (with or without semicolon) to output a constant in the
5346 constant pool, if it needs special treatment. (This macro need not do
5347 anything for RTL expressions that can be output normally.)
5349 The argument @var{file} is the standard I/O stream to output the
5350 assembler code on. @var{x} is the RTL expression for the constant to
5351 output, and @var{mode} is the machine mode (in case @var{x} is a
5352 @samp{const_int}). @var{align} is the required alignment for the value
5353 @var{x}; you should output an assembler directive to force this much
5356 The argument @var{labelno} is a number to use in an internal label for
5357 the address of this pool entry. The definition of this macro is
5358 responsible for outputting the label definition at the proper place.
5359 Here is how to do this:
5362 @code{(*targetm.asm_out.internal_label)} (@var{file}, "LC", @var{labelno});
5365 When you output a pool entry specially, you should end with a
5366 @code{goto} to the label @var{jumpto}. This will prevent the same pool
5367 entry from being output a second time in the usual manner.
5369 You need not define this macro if it would do nothing.
5372 @defmac ASM_OUTPUT_POOL_EPILOGUE (@var{file} @var{funname} @var{fundecl} @var{size})
5373 A C statement to output assembler commands to at the end of the constant
5374 pool for a function. @var{funname} is a string giving the name of the
5375 function. Should the return type of the function be required, you can
5376 obtain it via @var{fundecl}. @var{size} is the size, in bytes, of the
5377 constant pool that GCC wrote immediately before this call.
5379 If no constant-pool epilogue is required, the usual case, you need not
5383 @defmac IS_ASM_LOGICAL_LINE_SEPARATOR (@var{C}, @var{STR})
5384 Define this macro as a C expression which is nonzero if @var{C} is
5385 used as a logical line separator by the assembler. @var{STR} points
5386 to the position in the string where @var{C} was found; this can be used if
5387 a line separator uses multiple characters.
5389 If you do not define this macro, the default is that only
5390 the character @samp{;} is treated as a logical line separator.
5393 @hook TARGET_ASM_OPEN_PAREN
5395 These macros are provided by @file{real.h} for writing the definitions
5396 of @code{ASM_OUTPUT_DOUBLE} and the like:
5398 @defmac REAL_VALUE_TO_TARGET_SINGLE (@var{x}, @var{l})
5399 @defmacx REAL_VALUE_TO_TARGET_DOUBLE (@var{x}, @var{l})
5400 @defmacx REAL_VALUE_TO_TARGET_LONG_DOUBLE (@var{x}, @var{l})
5401 @defmacx REAL_VALUE_TO_TARGET_DECIMAL32 (@var{x}, @var{l})
5402 @defmacx REAL_VALUE_TO_TARGET_DECIMAL64 (@var{x}, @var{l})
5403 @defmacx REAL_VALUE_TO_TARGET_DECIMAL128 (@var{x}, @var{l})
5404 These translate @var{x}, of type @code{REAL_VALUE_TYPE}, to the
5405 target's floating point representation, and store its bit pattern in
5406 the variable @var{l}. For @code{REAL_VALUE_TO_TARGET_SINGLE} and
5407 @code{REAL_VALUE_TO_TARGET_DECIMAL32}, this variable should be a
5408 simple @code{long int}. For the others, it should be an array of
5409 @code{long int}. The number of elements in this array is determined
5410 by the size of the desired target floating point data type: 32 bits of
5411 it go in each @code{long int} array element. Each array element holds
5412 32 bits of the result, even if @code{long int} is wider than 32 bits
5413 on the host machine.
5415 The array element values are designed so that you can print them out
5416 using @code{fprintf} in the order they should appear in the target
5420 @node Uninitialized Data
5421 @subsection Output of Uninitialized Variables
5423 Each of the macros in this section is used to do the whole job of
5424 outputting a single uninitialized variable.
5426 @defmac ASM_OUTPUT_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
5427 A C statement (sans semicolon) to output to the stdio stream
5428 @var{stream} the assembler definition of a common-label named
5429 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
5430 is the size rounded up to whatever alignment the caller wants. It is
5431 possible that @var{size} may be zero, for instance if a struct with no
5432 other member than a zero-length array is defined. In this case, the
5433 backend must output a symbol definition that allocates at least one
5434 byte, both so that the address of the resulting object does not compare
5435 equal to any other, and because some object formats cannot even express
5436 the concept of a zero-sized common symbol, as that is how they represent
5437 an ordinary undefined external.
5439 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
5440 output the name itself; before and after that, output the additional
5441 assembler syntax for defining the name, and a newline.
5443 This macro controls how the assembler definitions of uninitialized
5444 common global variables are output.
5447 @defmac ASM_OUTPUT_ALIGNED_COMMON (@var{stream}, @var{name}, @var{size}, @var{alignment})
5448 Like @code{ASM_OUTPUT_COMMON} except takes the required alignment as a
5449 separate, explicit argument. If you define this macro, it is used in
5450 place of @code{ASM_OUTPUT_COMMON}, and gives you more flexibility in
5451 handling the required alignment of the variable. The alignment is specified
5452 as the number of bits.
5455 @defmac ASM_OUTPUT_ALIGNED_DECL_COMMON (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
5456 Like @code{ASM_OUTPUT_ALIGNED_COMMON} except that @var{decl} of the
5457 variable to be output, if there is one, or @code{NULL_TREE} if there
5458 is no corresponding variable. If you define this macro, GCC will use it
5459 in place of both @code{ASM_OUTPUT_COMMON} and
5460 @code{ASM_OUTPUT_ALIGNED_COMMON}. Define this macro when you need to see
5461 the variable's decl in order to chose what to output.
5464 @defmac ASM_OUTPUT_ALIGNED_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
5465 A C statement (sans semicolon) to output to the stdio stream
5466 @var{stream} the assembler definition of uninitialized global @var{decl} named
5467 @var{name} whose size is @var{size} bytes. The variable @var{alignment}
5468 is the alignment specified as the number of bits.
5470 Try to use function @code{asm_output_aligned_bss} defined in file
5471 @file{varasm.c} when defining this macro. If unable, use the expression
5472 @code{assemble_name (@var{stream}, @var{name})} to output the name itself;
5473 before and after that, output the additional assembler syntax for defining
5474 the name, and a newline.
5476 There are two ways of handling global BSS@. One is to define this macro.
5477 The other is to have @code{TARGET_ASM_SELECT_SECTION} return a
5478 switchable BSS section (@pxref{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}).
5479 You do not need to do both.
5481 Some languages do not have @code{common} data, and require a
5482 non-common form of global BSS in order to handle uninitialized globals
5483 efficiently. C++ is one example of this. However, if the target does
5484 not support global BSS, the front end may choose to make globals
5485 common in order to save space in the object file.
5488 @defmac ASM_OUTPUT_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded})
5489 A C statement (sans semicolon) to output to the stdio stream
5490 @var{stream} the assembler definition of a local-common-label named
5491 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
5492 is the size rounded up to whatever alignment the caller wants.
5494 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
5495 output the name itself; before and after that, output the additional
5496 assembler syntax for defining the name, and a newline.
5498 This macro controls how the assembler definitions of uninitialized
5499 static variables are output.
5502 @defmac ASM_OUTPUT_ALIGNED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{alignment})
5503 Like @code{ASM_OUTPUT_LOCAL} except takes the required alignment as a
5504 separate, explicit argument. If you define this macro, it is used in
5505 place of @code{ASM_OUTPUT_LOCAL}, and gives you more flexibility in
5506 handling the required alignment of the variable. The alignment is specified
5507 as the number of bits.
5510 @defmac ASM_OUTPUT_ALIGNED_DECL_LOCAL (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
5511 Like @code{ASM_OUTPUT_ALIGNED_DECL} except that @var{decl} of the
5512 variable to be output, if there is one, or @code{NULL_TREE} if there
5513 is no corresponding variable. If you define this macro, GCC will use it
5514 in place of both @code{ASM_OUTPUT_DECL} and
5515 @code{ASM_OUTPUT_ALIGNED_DECL}. Define this macro when you need to see
5516 the variable's decl in order to chose what to output.
5520 @subsection Output and Generation of Labels
5522 @c prevent bad page break with this line
5523 This is about outputting labels.
5525 @findex assemble_name
5526 @defmac ASM_OUTPUT_LABEL (@var{stream}, @var{name})
5527 A C statement (sans semicolon) to output to the stdio stream
5528 @var{stream} the assembler definition of a label named @var{name}.
5529 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
5530 output the name itself; before and after that, output the additional
5531 assembler syntax for defining the name, and a newline. A default
5532 definition of this macro is provided which is correct for most systems.
5535 @defmac ASM_OUTPUT_FUNCTION_LABEL (@var{stream}, @var{name}, @var{decl})
5536 A C statement (sans semicolon) to output to the stdio stream
5537 @var{stream} the assembler definition of a label named @var{name} of
5539 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
5540 output the name itself; before and after that, output the additional
5541 assembler syntax for defining the name, and a newline. A default
5542 definition of this macro is provided which is correct for most systems.
5544 If this macro is not defined, then the function name is defined in the
5545 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
5548 @findex assemble_name_raw
5549 @defmac ASM_OUTPUT_INTERNAL_LABEL (@var{stream}, @var{name})
5550 Identical to @code{ASM_OUTPUT_LABEL}, except that @var{name} is known
5551 to refer to a compiler-generated label. The default definition uses
5552 @code{assemble_name_raw}, which is like @code{assemble_name} except
5553 that it is more efficient.
5557 A C string containing the appropriate assembler directive to specify the
5558 size of a symbol, without any arguments. On systems that use ELF, the
5559 default (in @file{config/elfos.h}) is @samp{"\t.size\t"}; on other
5560 systems, the default is not to define this macro.
5562 Define this macro only if it is correct to use the default definitions
5563 of @code{ASM_OUTPUT_SIZE_DIRECTIVE} and @code{ASM_OUTPUT_MEASURED_SIZE}
5564 for your system. If you need your own custom definitions of those
5565 macros, or if you do not need explicit symbol sizes at all, do not
5569 @defmac ASM_OUTPUT_SIZE_DIRECTIVE (@var{stream}, @var{name}, @var{size})
5570 A C statement (sans semicolon) to output to the stdio stream
5571 @var{stream} a directive telling the assembler that the size of the
5572 symbol @var{name} is @var{size}. @var{size} is a @code{HOST_WIDE_INT}.
5573 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
5577 @defmac ASM_OUTPUT_MEASURED_SIZE (@var{stream}, @var{name})
5578 A C statement (sans semicolon) to output to the stdio stream
5579 @var{stream} a directive telling the assembler to calculate the size of
5580 the symbol @var{name} by subtracting its address from the current
5583 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
5584 provided. The default assumes that the assembler recognizes a special
5585 @samp{.} symbol as referring to the current address, and can calculate
5586 the difference between this and another symbol. If your assembler does
5587 not recognize @samp{.} or cannot do calculations with it, you will need
5588 to redefine @code{ASM_OUTPUT_MEASURED_SIZE} to use some other technique.
5591 @defmac NO_DOLLAR_IN_LABEL
5592 Define this macro if the assembler does not accept the character
5593 @samp{$} in label names. By default constructors and destructors in
5594 G++ have @samp{$} in the identifiers. If this macro is defined,
5595 @samp{.} is used instead.
5598 @defmac NO_DOT_IN_LABEL
5599 Define this macro if the assembler does not accept the character
5600 @samp{.} in label names. By default constructors and destructors in G++
5601 have names that use @samp{.}. If this macro is defined, these names
5602 are rewritten to avoid @samp{.}.
5606 A C string containing the appropriate assembler directive to specify the
5607 type of a symbol, without any arguments. On systems that use ELF, the
5608 default (in @file{config/elfos.h}) is @samp{"\t.type\t"}; on other
5609 systems, the default is not to define this macro.
5611 Define this macro only if it is correct to use the default definition of
5612 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
5613 custom definition of this macro, or if you do not need explicit symbol
5614 types at all, do not define this macro.
5617 @defmac TYPE_OPERAND_FMT
5618 A C string which specifies (using @code{printf} syntax) the format of
5619 the second operand to @code{TYPE_ASM_OP}. On systems that use ELF, the
5620 default (in @file{config/elfos.h}) is @samp{"@@%s"}; on other systems,
5621 the default is not to define this macro.
5623 Define this macro only if it is correct to use the default definition of
5624 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
5625 custom definition of this macro, or if you do not need explicit symbol
5626 types at all, do not define this macro.
5629 @defmac ASM_OUTPUT_TYPE_DIRECTIVE (@var{stream}, @var{type})
5630 A C statement (sans semicolon) to output to the stdio stream
5631 @var{stream} a directive telling the assembler that the type of the
5632 symbol @var{name} is @var{type}. @var{type} is a C string; currently,
5633 that string is always either @samp{"function"} or @samp{"object"}, but
5634 you should not count on this.
5636 If you define @code{TYPE_ASM_OP} and @code{TYPE_OPERAND_FMT}, a default
5637 definition of this macro is provided.
5640 @defmac ASM_DECLARE_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl})
5641 A C statement (sans semicolon) to output to the stdio stream
5642 @var{stream} any text necessary for declaring the name @var{name} of a
5643 function which is being defined. This macro is responsible for
5644 outputting the label definition (perhaps using
5645 @code{ASM_OUTPUT_FUNCTION_LABEL}). The argument @var{decl} is the
5646 @code{FUNCTION_DECL} tree node representing the function.
5648 If this macro is not defined, then the function name is defined in the
5649 usual manner as a label (by means of @code{ASM_OUTPUT_FUNCTION_LABEL}).
5651 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition
5655 @defmac ASM_DECLARE_FUNCTION_SIZE (@var{stream}, @var{name}, @var{decl})
5656 A C statement (sans semicolon) to output to the stdio stream
5657 @var{stream} any text necessary for declaring the size of a function
5658 which is being defined. The argument @var{name} is the name of the
5659 function. The argument @var{decl} is the @code{FUNCTION_DECL} tree node
5660 representing the function.
5662 If this macro is not defined, then the function size is not defined.
5664 You may wish to use @code{ASM_OUTPUT_MEASURED_SIZE} in the definition
5668 @defmac ASM_DECLARE_COLD_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl})
5669 A C statement (sans semicolon) to output to the stdio stream
5670 @var{stream} any text necessary for declaring the name @var{name} of a
5671 cold function partition which is being defined. This macro is responsible
5672 for outputting the label definition (perhaps using
5673 @code{ASM_OUTPUT_FUNCTION_LABEL}). The argument @var{decl} is the
5674 @code{FUNCTION_DECL} tree node representing the function.
5676 If this macro is not defined, then the cold partition name is defined in the
5677 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
5679 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition
5683 @defmac ASM_DECLARE_COLD_FUNCTION_SIZE (@var{stream}, @var{name}, @var{decl})
5684 A C statement (sans semicolon) to output to the stdio stream
5685 @var{stream} any text necessary for declaring the size of a cold function
5686 partition which is being defined. The argument @var{name} is the name of the
5687 cold partition of the function. The argument @var{decl} is the
5688 @code{FUNCTION_DECL} tree node representing the function.
5690 If this macro is not defined, then the partition size is not defined.
5692 You may wish to use @code{ASM_OUTPUT_MEASURED_SIZE} in the definition
5696 @defmac ASM_DECLARE_OBJECT_NAME (@var{stream}, @var{name}, @var{decl})
5697 A C statement (sans semicolon) to output to the stdio stream
5698 @var{stream} any text necessary for declaring the name @var{name} of an
5699 initialized variable which is being defined. This macro must output the
5700 label definition (perhaps using @code{ASM_OUTPUT_LABEL}). The argument
5701 @var{decl} is the @code{VAR_DECL} tree node representing the variable.
5703 If this macro is not defined, then the variable name is defined in the
5704 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
5706 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} and/or
5707 @code{ASM_OUTPUT_SIZE_DIRECTIVE} in the definition of this macro.
5710 @hook TARGET_ASM_DECLARE_CONSTANT_NAME
5712 @defmac ASM_DECLARE_REGISTER_GLOBAL (@var{stream}, @var{decl}, @var{regno}, @var{name})
5713 A C statement (sans semicolon) to output to the stdio stream
5714 @var{stream} any text necessary for claiming a register @var{regno}
5715 for a global variable @var{decl} with name @var{name}.
5717 If you don't define this macro, that is equivalent to defining it to do
5721 @defmac ASM_FINISH_DECLARE_OBJECT (@var{stream}, @var{decl}, @var{toplevel}, @var{atend})
5722 A C statement (sans semicolon) to finish up declaring a variable name
5723 once the compiler has processed its initializer fully and thus has had a
5724 chance to determine the size of an array when controlled by an
5725 initializer. This is used on systems where it's necessary to declare
5726 something about the size of the object.
5728 If you don't define this macro, that is equivalent to defining it to do
5731 You may wish to use @code{ASM_OUTPUT_SIZE_DIRECTIVE} and/or
5732 @code{ASM_OUTPUT_MEASURED_SIZE} in the definition of this macro.
5735 @hook TARGET_ASM_GLOBALIZE_LABEL
5737 @hook TARGET_ASM_GLOBALIZE_DECL_NAME
5739 @hook TARGET_ASM_ASSEMBLE_UNDEFINED_DECL
5741 @defmac ASM_WEAKEN_LABEL (@var{stream}, @var{name})
5742 A C statement (sans semicolon) to output to the stdio stream
5743 @var{stream} some commands that will make the label @var{name} weak;
5744 that is, available for reference from other files but only used if
5745 no other definition is available. Use the expression
5746 @code{assemble_name (@var{stream}, @var{name})} to output the name
5747 itself; before and after that, output the additional assembler syntax
5748 for making that name weak, and a newline.
5750 If you don't define this macro or @code{ASM_WEAKEN_DECL}, GCC will not
5751 support weak symbols and you should not define the @code{SUPPORTS_WEAK}
5755 @defmac ASM_WEAKEN_DECL (@var{stream}, @var{decl}, @var{name}, @var{value})
5756 Combines (and replaces) the function of @code{ASM_WEAKEN_LABEL} and
5757 @code{ASM_OUTPUT_WEAK_ALIAS}, allowing access to the associated function
5758 or variable decl. If @var{value} is not @code{NULL}, this C statement
5759 should output to the stdio stream @var{stream} assembler code which
5760 defines (equates) the weak symbol @var{name} to have the value
5761 @var{value}. If @var{value} is @code{NULL}, it should output commands
5762 to make @var{name} weak.
5765 @defmac ASM_OUTPUT_WEAKREF (@var{stream}, @var{decl}, @var{name}, @var{value})
5766 Outputs a directive that enables @var{name} to be used to refer to
5767 symbol @var{value} with weak-symbol semantics. @code{decl} is the
5768 declaration of @code{name}.
5771 @defmac SUPPORTS_WEAK
5772 A preprocessor constant expression which evaluates to true if the target
5773 supports weak symbols.
5775 If you don't define this macro, @file{defaults.h} provides a default
5776 definition. If either @code{ASM_WEAKEN_LABEL} or @code{ASM_WEAKEN_DECL}
5777 is defined, the default definition is @samp{1}; otherwise, it is @samp{0}.
5780 @defmac TARGET_SUPPORTS_WEAK
5781 A C expression which evaluates to true if the target supports weak symbols.
5783 If you don't define this macro, @file{defaults.h} provides a default
5784 definition. The default definition is @samp{(SUPPORTS_WEAK)}. Define
5785 this macro if you want to control weak symbol support with a compiler
5786 flag such as @option{-melf}.
5789 @defmac MAKE_DECL_ONE_ONLY (@var{decl})
5790 A C statement (sans semicolon) to mark @var{decl} to be emitted as a
5791 public symbol such that extra copies in multiple translation units will
5792 be discarded by the linker. Define this macro if your object file
5793 format provides support for this concept, such as the @samp{COMDAT}
5794 section flags in the Microsoft Windows PE/COFF format, and this support
5795 requires changes to @var{decl}, such as putting it in a separate section.
5798 @defmac SUPPORTS_ONE_ONLY
5799 A C expression which evaluates to true if the target supports one-only
5802 If you don't define this macro, @file{varasm.c} provides a default
5803 definition. If @code{MAKE_DECL_ONE_ONLY} is defined, the default
5804 definition is @samp{1}; otherwise, it is @samp{0}. Define this macro if
5805 you want to control one-only symbol support with a compiler flag, or if
5806 setting the @code{DECL_ONE_ONLY} flag is enough to mark a declaration to
5807 be emitted as one-only.
5810 @hook TARGET_ASM_ASSEMBLE_VISIBILITY
5812 @defmac TARGET_WEAK_NOT_IN_ARCHIVE_TOC
5813 A C expression that evaluates to true if the target's linker expects
5814 that weak symbols do not appear in a static archive's table of contents.
5815 The default is @code{0}.
5817 Leaving weak symbols out of an archive's table of contents means that,
5818 if a symbol will only have a definition in one translation unit and
5819 will have undefined references from other translation units, that
5820 symbol should not be weak. Defining this macro to be nonzero will
5821 thus have the effect that certain symbols that would normally be weak
5822 (explicit template instantiations, and vtables for polymorphic classes
5823 with noninline key methods) will instead be nonweak.
5825 The C++ ABI requires this macro to be zero. Define this macro for
5826 targets where full C++ ABI compliance is impossible and where linker
5827 restrictions require weak symbols to be left out of a static archive's
5831 @defmac ASM_OUTPUT_EXTERNAL (@var{stream}, @var{decl}, @var{name})
5832 A C statement (sans semicolon) to output to the stdio stream
5833 @var{stream} any text necessary for declaring the name of an external
5834 symbol named @var{name} which is referenced in this compilation but
5835 not defined. The value of @var{decl} is the tree node for the
5838 This macro need not be defined if it does not need to output anything.
5839 The GNU assembler and most Unix assemblers don't require anything.
5842 @hook TARGET_ASM_EXTERNAL_LIBCALL
5844 @hook TARGET_ASM_MARK_DECL_PRESERVED
5846 @defmac ASM_OUTPUT_LABELREF (@var{stream}, @var{name})
5847 A C statement (sans semicolon) to output to the stdio stream
5848 @var{stream} a reference in assembler syntax to a label named
5849 @var{name}. This should add @samp{_} to the front of the name, if that
5850 is customary on your operating system, as it is in most Berkeley Unix
5851 systems. This macro is used in @code{assemble_name}.
5854 @hook TARGET_MANGLE_ASSEMBLER_NAME
5856 @defmac ASM_OUTPUT_SYMBOL_REF (@var{stream}, @var{sym})
5857 A C statement (sans semicolon) to output a reference to
5858 @code{SYMBOL_REF} @var{sym}. If not defined, @code{assemble_name}
5859 will be used to output the name of the symbol. This macro may be used
5860 to modify the way a symbol is referenced depending on information
5861 encoded by @code{TARGET_ENCODE_SECTION_INFO}.
5864 @defmac ASM_OUTPUT_LABEL_REF (@var{stream}, @var{buf})
5865 A C statement (sans semicolon) to output a reference to @var{buf}, the
5866 result of @code{ASM_GENERATE_INTERNAL_LABEL}. If not defined,
5867 @code{assemble_name} will be used to output the name of the symbol.
5868 This macro is not used by @code{output_asm_label}, or the @code{%l}
5869 specifier that calls it; the intention is that this macro should be set
5870 when it is necessary to output a label differently when its address is
5874 @hook TARGET_ASM_INTERNAL_LABEL
5876 @defmac ASM_OUTPUT_DEBUG_LABEL (@var{stream}, @var{prefix}, @var{num})
5877 A C statement to output to the stdio stream @var{stream} a debug info
5878 label whose name is made from the string @var{prefix} and the number
5879 @var{num}. This is useful for VLIW targets, where debug info labels
5880 may need to be treated differently than branch target labels. On some
5881 systems, branch target labels must be at the beginning of instruction
5882 bundles, but debug info labels can occur in the middle of instruction
5885 If this macro is not defined, then @code{(*targetm.asm_out.internal_label)} will be
5889 @defmac ASM_GENERATE_INTERNAL_LABEL (@var{string}, @var{prefix}, @var{num})
5890 A C statement to store into the string @var{string} a label whose name
5891 is made from the string @var{prefix} and the number @var{num}.
5893 This string, when output subsequently by @code{assemble_name}, should
5894 produce the output that @code{(*targetm.asm_out.internal_label)} would produce
5895 with the same @var{prefix} and @var{num}.
5897 If the string begins with @samp{*}, then @code{assemble_name} will
5898 output the rest of the string unchanged. It is often convenient for
5899 @code{ASM_GENERATE_INTERNAL_LABEL} to use @samp{*} in this way. If the
5900 string doesn't start with @samp{*}, then @code{ASM_OUTPUT_LABELREF} gets
5901 to output the string, and may change it. (Of course,
5902 @code{ASM_OUTPUT_LABELREF} is also part of your machine description, so
5903 you should know what it does on your machine.)
5906 @defmac ASM_FORMAT_PRIVATE_NAME (@var{outvar}, @var{name}, @var{number})
5907 A C expression to assign to @var{outvar} (which is a variable of type
5908 @code{char *}) a newly allocated string made from the string
5909 @var{name} and the number @var{number}, with some suitable punctuation
5910 added. Use @code{alloca} to get space for the string.
5912 The string will be used as an argument to @code{ASM_OUTPUT_LABELREF} to
5913 produce an assembler label for an internal static variable whose name is
5914 @var{name}. Therefore, the string must be such as to result in valid
5915 assembler code. The argument @var{number} is different each time this
5916 macro is executed; it prevents conflicts between similarly-named
5917 internal static variables in different scopes.
5919 Ideally this string should not be a valid C identifier, to prevent any
5920 conflict with the user's own symbols. Most assemblers allow periods
5921 or percent signs in assembler symbols; putting at least one of these
5922 between the name and the number will suffice.
5924 If this macro is not defined, a default definition will be provided
5925 which is correct for most systems.
5928 @defmac ASM_OUTPUT_DEF (@var{stream}, @var{name}, @var{value})
5929 A C statement to output to the stdio stream @var{stream} assembler code
5930 which defines (equates) the symbol @var{name} to have the value @var{value}.
5933 If @code{SET_ASM_OP} is defined, a default definition is provided which is
5934 correct for most systems.
5937 @defmac ASM_OUTPUT_DEF_FROM_DECLS (@var{stream}, @var{decl_of_name}, @var{decl_of_value})
5938 A C statement to output to the stdio stream @var{stream} assembler code
5939 which defines (equates) the symbol whose tree node is @var{decl_of_name}
5940 to have the value of the tree node @var{decl_of_value}. This macro will
5941 be used in preference to @samp{ASM_OUTPUT_DEF} if it is defined and if
5942 the tree nodes are available.
5945 If @code{SET_ASM_OP} is defined, a default definition is provided which is
5946 correct for most systems.
5949 @defmac TARGET_DEFERRED_OUTPUT_DEFS (@var{decl_of_name}, @var{decl_of_value})
5950 A C statement that evaluates to true if the assembler code which defines
5951 (equates) the symbol whose tree node is @var{decl_of_name} to have the value
5952 of the tree node @var{decl_of_value} should be emitted near the end of the
5953 current compilation unit. The default is to not defer output of defines.
5954 This macro affects defines output by @samp{ASM_OUTPUT_DEF} and
5955 @samp{ASM_OUTPUT_DEF_FROM_DECLS}.
5958 @defmac ASM_OUTPUT_WEAK_ALIAS (@var{stream}, @var{name}, @var{value})
5959 A C statement to output to the stdio stream @var{stream} assembler code
5960 which defines (equates) the weak symbol @var{name} to have the value
5961 @var{value}. If @var{value} is @code{NULL}, it defines @var{name} as
5962 an undefined weak symbol.
5964 Define this macro if the target only supports weak aliases; define
5965 @code{ASM_OUTPUT_DEF} instead if possible.
5968 @defmac OBJC_GEN_METHOD_LABEL (@var{buf}, @var{is_inst}, @var{class_name}, @var{cat_name}, @var{sel_name})
5969 Define this macro to override the default assembler names used for
5970 Objective-C methods.
5972 The default name is a unique method number followed by the name of the
5973 class (e.g.@: @samp{_1_Foo}). For methods in categories, the name of
5974 the category is also included in the assembler name (e.g.@:
5977 These names are safe on most systems, but make debugging difficult since
5978 the method's selector is not present in the name. Therefore, particular
5979 systems define other ways of computing names.
5981 @var{buf} is an expression of type @code{char *} which gives you a
5982 buffer in which to store the name; its length is as long as
5983 @var{class_name}, @var{cat_name} and @var{sel_name} put together, plus
5984 50 characters extra.
5986 The argument @var{is_inst} specifies whether the method is an instance
5987 method or a class method; @var{class_name} is the name of the class;
5988 @var{cat_name} is the name of the category (or @code{NULL} if the method is not
5989 in a category); and @var{sel_name} is the name of the selector.
5991 On systems where the assembler can handle quoted names, you can use this
5992 macro to provide more human-readable names.
5995 @node Initialization
5996 @subsection How Initialization Functions Are Handled
5997 @cindex initialization routines
5998 @cindex termination routines
5999 @cindex constructors, output of
6000 @cindex destructors, output of
6002 The compiled code for certain languages includes @dfn{constructors}
6003 (also called @dfn{initialization routines})---functions to initialize
6004 data in the program when the program is started. These functions need
6005 to be called before the program is ``started''---that is to say, before
6006 @code{main} is called.
6008 Compiling some languages generates @dfn{destructors} (also called
6009 @dfn{termination routines}) that should be called when the program
6012 To make the initialization and termination functions work, the compiler
6013 must output something in the assembler code to cause those functions to
6014 be called at the appropriate time. When you port the compiler to a new
6015 system, you need to specify how to do this.
6017 There are two major ways that GCC currently supports the execution of
6018 initialization and termination functions. Each way has two variants.
6019 Much of the structure is common to all four variations.
6021 @findex __CTOR_LIST__
6022 @findex __DTOR_LIST__
6023 The linker must build two lists of these functions---a list of
6024 initialization functions, called @code{__CTOR_LIST__}, and a list of
6025 termination functions, called @code{__DTOR_LIST__}.
6027 Each list always begins with an ignored function pointer (which may hold
6028 0, @minus{}1, or a count of the function pointers after it, depending on
6029 the environment). This is followed by a series of zero or more function
6030 pointers to constructors (or destructors), followed by a function
6031 pointer containing zero.
6033 Depending on the operating system and its executable file format, either
6034 @file{crtstuff.c} or @file{libgcc2.c} traverses these lists at startup
6035 time and exit time. Constructors are called in reverse order of the
6036 list; destructors in forward order.
6038 The best way to handle static constructors works only for object file
6039 formats which provide arbitrarily-named sections. A section is set
6040 aside for a list of constructors, and another for a list of destructors.
6041 Traditionally these are called @samp{.ctors} and @samp{.dtors}. Each
6042 object file that defines an initialization function also puts a word in
6043 the constructor section to point to that function. The linker
6044 accumulates all these words into one contiguous @samp{.ctors} section.
6045 Termination functions are handled similarly.
6047 This method will be chosen as the default by @file{target-def.h} if
6048 @code{TARGET_ASM_NAMED_SECTION} is defined. A target that does not
6049 support arbitrary sections, but does support special designated
6050 constructor and destructor sections may define @code{CTORS_SECTION_ASM_OP}
6051 and @code{DTORS_SECTION_ASM_OP} to achieve the same effect.
6053 When arbitrary sections are available, there are two variants, depending
6054 upon how the code in @file{crtstuff.c} is called. On systems that
6055 support a @dfn{.init} section which is executed at program startup,
6056 parts of @file{crtstuff.c} are compiled into that section. The
6057 program is linked by the @command{gcc} driver like this:
6060 ld -o @var{output_file} crti.o crtbegin.o @dots{} -lgcc crtend.o crtn.o
6063 The prologue of a function (@code{__init}) appears in the @code{.init}
6064 section of @file{crti.o}; the epilogue appears in @file{crtn.o}. Likewise
6065 for the function @code{__fini} in the @dfn{.fini} section. Normally these
6066 files are provided by the operating system or by the GNU C library, but
6067 are provided by GCC for a few targets.
6069 The objects @file{crtbegin.o} and @file{crtend.o} are (for most targets)
6070 compiled from @file{crtstuff.c}. They contain, among other things, code
6071 fragments within the @code{.init} and @code{.fini} sections that branch
6072 to routines in the @code{.text} section. The linker will pull all parts
6073 of a section together, which results in a complete @code{__init} function
6074 that invokes the routines we need at startup.
6076 To use this variant, you must define the @code{INIT_SECTION_ASM_OP}
6079 If no init section is available, when GCC compiles any function called
6080 @code{main} (or more accurately, any function designated as a program
6081 entry point by the language front end calling @code{expand_main_function}),
6082 it inserts a procedure call to @code{__main} as the first executable code
6083 after the function prologue. The @code{__main} function is defined
6084 in @file{libgcc2.c} and runs the global constructors.
6086 In file formats that don't support arbitrary sections, there are again
6087 two variants. In the simplest variant, the GNU linker (GNU @code{ld})
6088 and an `a.out' format must be used. In this case,
6089 @code{TARGET_ASM_CONSTRUCTOR} is defined to produce a @code{.stabs}
6090 entry of type @samp{N_SETT}, referencing the name @code{__CTOR_LIST__},
6091 and with the address of the void function containing the initialization
6092 code as its value. The GNU linker recognizes this as a request to add
6093 the value to a @dfn{set}; the values are accumulated, and are eventually
6094 placed in the executable as a vector in the format described above, with
6095 a leading (ignored) count and a trailing zero element.
6096 @code{TARGET_ASM_DESTRUCTOR} is handled similarly. Since no init
6097 section is available, the absence of @code{INIT_SECTION_ASM_OP} causes
6098 the compilation of @code{main} to call @code{__main} as above, starting
6099 the initialization process.
6101 The last variant uses neither arbitrary sections nor the GNU linker.
6102 This is preferable when you want to do dynamic linking and when using
6103 file formats which the GNU linker does not support, such as `ECOFF'@. In
6104 this case, @code{TARGET_HAVE_CTORS_DTORS} is false, initialization and
6105 termination functions are recognized simply by their names. This requires
6106 an extra program in the linkage step, called @command{collect2}. This program
6107 pretends to be the linker, for use with GCC; it does its job by running
6108 the ordinary linker, but also arranges to include the vectors of
6109 initialization and termination functions. These functions are called
6110 via @code{__main} as described above. In order to use this method,
6111 @code{use_collect2} must be defined in the target in @file{config.gcc}.
6114 The following section describes the specific macros that control and
6115 customize the handling of initialization and termination functions.
6118 @node Macros for Initialization
6119 @subsection Macros Controlling Initialization Routines
6121 Here are the macros that control how the compiler handles initialization
6122 and termination functions:
6124 @defmac INIT_SECTION_ASM_OP
6125 If defined, a C string constant, including spacing, for the assembler
6126 operation to identify the following data as initialization code. If not
6127 defined, GCC will assume such a section does not exist. When you are
6128 using special sections for initialization and termination functions, this
6129 macro also controls how @file{crtstuff.c} and @file{libgcc2.c} arrange to
6130 run the initialization functions.
6133 @defmac HAS_INIT_SECTION
6134 If defined, @code{main} will not call @code{__main} as described above.
6135 This macro should be defined for systems that control start-up code
6136 on a symbol-by-symbol basis, such as OSF/1, and should not
6137 be defined explicitly for systems that support @code{INIT_SECTION_ASM_OP}.
6140 @defmac LD_INIT_SWITCH
6141 If defined, a C string constant for a switch that tells the linker that
6142 the following symbol is an initialization routine.
6145 @defmac LD_FINI_SWITCH
6146 If defined, a C string constant for a switch that tells the linker that
6147 the following symbol is a finalization routine.
6150 @defmac COLLECT_SHARED_INIT_FUNC (@var{stream}, @var{func})
6151 If defined, a C statement that will write a function that can be
6152 automatically called when a shared library is loaded. The function
6153 should call @var{func}, which takes no arguments. If not defined, and
6154 the object format requires an explicit initialization function, then a
6155 function called @code{_GLOBAL__DI} will be generated.
6157 This function and the following one are used by collect2 when linking a
6158 shared library that needs constructors or destructors, or has DWARF2
6159 exception tables embedded in the code.
6162 @defmac COLLECT_SHARED_FINI_FUNC (@var{stream}, @var{func})
6163 If defined, a C statement that will write a function that can be
6164 automatically called when a shared library is unloaded. The function
6165 should call @var{func}, which takes no arguments. If not defined, and
6166 the object format requires an explicit finalization function, then a
6167 function called @code{_GLOBAL__DD} will be generated.
6170 @defmac INVOKE__main
6171 If defined, @code{main} will call @code{__main} despite the presence of
6172 @code{INIT_SECTION_ASM_OP}. This macro should be defined for systems
6173 where the init section is not actually run automatically, but is still
6174 useful for collecting the lists of constructors and destructors.
6177 @defmac SUPPORTS_INIT_PRIORITY
6178 If nonzero, the C++ @code{init_priority} attribute is supported and the
6179 compiler should emit instructions to control the order of initialization
6180 of objects. If zero, the compiler will issue an error message upon
6181 encountering an @code{init_priority} attribute.
6184 @hook TARGET_HAVE_CTORS_DTORS
6186 @hook TARGET_ASM_CONSTRUCTOR
6188 @hook TARGET_ASM_DESTRUCTOR
6190 If @code{TARGET_HAVE_CTORS_DTORS} is true, the initialization routine
6191 generated for the generated object file will have static linkage.
6193 If your system uses @command{collect2} as the means of processing
6194 constructors, then that program normally uses @command{nm} to scan
6195 an object file for constructor functions to be called.
6197 On certain kinds of systems, you can define this macro to make
6198 @command{collect2} work faster (and, in some cases, make it work at all):
6200 @defmac OBJECT_FORMAT_COFF
6201 Define this macro if the system uses COFF (Common Object File Format)
6202 object files, so that @command{collect2} can assume this format and scan
6203 object files directly for dynamic constructor/destructor functions.
6205 This macro is effective only in a native compiler; @command{collect2} as
6206 part of a cross compiler always uses @command{nm} for the target machine.
6209 @defmac REAL_NM_FILE_NAME
6210 Define this macro as a C string constant containing the file name to use
6211 to execute @command{nm}. The default is to search the path normally for
6216 @command{collect2} calls @command{nm} to scan object files for static
6217 constructors and destructors and LTO info. By default, @option{-n} is
6218 passed. Define @code{NM_FLAGS} to a C string constant if other options
6219 are needed to get the same output format as GNU @command{nm -n}
6223 If your system supports shared libraries and has a program to list the
6224 dynamic dependencies of a given library or executable, you can define
6225 these macros to enable support for running initialization and
6226 termination functions in shared libraries:
6229 Define this macro to a C string constant containing the name of the program
6230 which lists dynamic dependencies, like @command{ldd} under SunOS 4.
6233 @defmac PARSE_LDD_OUTPUT (@var{ptr})
6234 Define this macro to be C code that extracts filenames from the output
6235 of the program denoted by @code{LDD_SUFFIX}. @var{ptr} is a variable
6236 of type @code{char *} that points to the beginning of a line of output
6237 from @code{LDD_SUFFIX}. If the line lists a dynamic dependency, the
6238 code must advance @var{ptr} to the beginning of the filename on that
6239 line. Otherwise, it must set @var{ptr} to @code{NULL}.
6242 @defmac SHLIB_SUFFIX
6243 Define this macro to a C string constant containing the default shared
6244 library extension of the target (e.g., @samp{".so"}). @command{collect2}
6245 strips version information after this suffix when generating global
6246 constructor and destructor names. This define is only needed on targets
6247 that use @command{collect2} to process constructors and destructors.
6250 @node Instruction Output
6251 @subsection Output of Assembler Instructions
6253 @c prevent bad page break with this line
6254 This describes assembler instruction output.
6256 @defmac REGISTER_NAMES
6257 A C initializer containing the assembler's names for the machine
6258 registers, each one as a C string constant. This is what translates
6259 register numbers in the compiler into assembler language.
6262 @defmac ADDITIONAL_REGISTER_NAMES
6263 If defined, a C initializer for an array of structures containing a name
6264 and a register number. This macro defines additional names for hard
6265 registers, thus allowing the @code{asm} option in declarations to refer
6266 to registers using alternate names.
6269 @defmac OVERLAPPING_REGISTER_NAMES
6270 If defined, a C initializer for an array of structures containing a
6271 name, a register number and a count of the number of consecutive
6272 machine registers the name overlaps. This macro defines additional
6273 names for hard registers, thus allowing the @code{asm} option in
6274 declarations to refer to registers using alternate names. Unlike
6275 @code{ADDITIONAL_REGISTER_NAMES}, this macro should be used when the
6276 register name implies multiple underlying registers.
6278 This macro should be used when it is important that a clobber in an
6279 @code{asm} statement clobbers all the underlying values implied by the
6280 register name. For example, on ARM, clobbering the double-precision
6281 VFP register ``d0'' implies clobbering both single-precision registers
6285 @defmac ASM_OUTPUT_OPCODE (@var{stream}, @var{ptr})
6286 Define this macro if you are using an unusual assembler that
6287 requires different names for the machine instructions.
6289 The definition is a C statement or statements which output an
6290 assembler instruction opcode to the stdio stream @var{stream}. The
6291 macro-operand @var{ptr} is a variable of type @code{char *} which
6292 points to the opcode name in its ``internal'' form---the form that is
6293 written in the machine description. The definition should output the
6294 opcode name to @var{stream}, performing any translation you desire, and
6295 increment the variable @var{ptr} to point at the end of the opcode
6296 so that it will not be output twice.
6298 In fact, your macro definition may process less than the entire opcode
6299 name, or more than the opcode name; but if you want to process text
6300 that includes @samp{%}-sequences to substitute operands, you must take
6301 care of the substitution yourself. Just be sure to increment
6302 @var{ptr} over whatever text should not be output normally.
6304 @findex recog_data.operand
6305 If you need to look at the operand values, they can be found as the
6306 elements of @code{recog_data.operand}.
6308 If the macro definition does nothing, the instruction is output
6312 @defmac FINAL_PRESCAN_INSN (@var{insn}, @var{opvec}, @var{noperands})
6313 If defined, a C statement to be executed just prior to the output of
6314 assembler code for @var{insn}, to modify the extracted operands so
6315 they will be output differently.
6317 Here the argument @var{opvec} is the vector containing the operands
6318 extracted from @var{insn}, and @var{noperands} is the number of
6319 elements of the vector which contain meaningful data for this insn.
6320 The contents of this vector are what will be used to convert the insn
6321 template into assembler code, so you can change the assembler output
6322 by changing the contents of the vector.
6324 This macro is useful when various assembler syntaxes share a single
6325 file of instruction patterns; by defining this macro differently, you
6326 can cause a large class of instructions to be output differently (such
6327 as with rearranged operands). Naturally, variations in assembler
6328 syntax affecting individual insn patterns ought to be handled by
6329 writing conditional output routines in those patterns.
6331 If this macro is not defined, it is equivalent to a null statement.
6334 @hook TARGET_ASM_FINAL_POSTSCAN_INSN
6336 @defmac PRINT_OPERAND (@var{stream}, @var{x}, @var{code})
6337 A C compound statement to output to stdio stream @var{stream} the
6338 assembler syntax for an instruction operand @var{x}. @var{x} is an
6341 @var{code} is a value that can be used to specify one of several ways
6342 of printing the operand. It is used when identical operands must be
6343 printed differently depending on the context. @var{code} comes from
6344 the @samp{%} specification that was used to request printing of the
6345 operand. If the specification was just @samp{%@var{digit}} then
6346 @var{code} is 0; if the specification was @samp{%@var{ltr}
6347 @var{digit}} then @var{code} is the ASCII code for @var{ltr}.
6350 If @var{x} is a register, this macro should print the register's name.
6351 The names can be found in an array @code{reg_names} whose type is
6352 @code{char *[]}. @code{reg_names} is initialized from
6353 @code{REGISTER_NAMES}.
6355 When the machine description has a specification @samp{%@var{punct}}
6356 (a @samp{%} followed by a punctuation character), this macro is called
6357 with a null pointer for @var{x} and the punctuation character for
6361 @defmac PRINT_OPERAND_PUNCT_VALID_P (@var{code})
6362 A C expression which evaluates to true if @var{code} is a valid
6363 punctuation character for use in the @code{PRINT_OPERAND} macro. If
6364 @code{PRINT_OPERAND_PUNCT_VALID_P} is not defined, it means that no
6365 punctuation characters (except for the standard one, @samp{%}) are used
6369 @defmac PRINT_OPERAND_ADDRESS (@var{stream}, @var{x})
6370 A C compound statement to output to stdio stream @var{stream} the
6371 assembler syntax for an instruction operand that is a memory reference
6372 whose address is @var{x}. @var{x} is an RTL expression.
6374 @cindex @code{TARGET_ENCODE_SECTION_INFO} usage
6375 On some machines, the syntax for a symbolic address depends on the
6376 section that the address refers to. On these machines, define the hook
6377 @code{TARGET_ENCODE_SECTION_INFO} to store the information into the
6378 @code{symbol_ref}, and then check for it here. @xref{Assembler
6382 @findex dbr_sequence_length
6383 @defmac DBR_OUTPUT_SEQEND (@var{file})
6384 A C statement, to be executed after all slot-filler instructions have
6385 been output. If necessary, call @code{dbr_sequence_length} to
6386 determine the number of slots filled in a sequence (zero if not
6387 currently outputting a sequence), to decide how many no-ops to output,
6390 Don't define this macro if it has nothing to do, but it is helpful in
6391 reading assembly output if the extent of the delay sequence is made
6392 explicit (e.g.@: with white space).
6395 @findex final_sequence
6396 Note that output routines for instructions with delay slots must be
6397 prepared to deal with not being output as part of a sequence
6398 (i.e.@: when the scheduling pass is not run, or when no slot fillers could be
6399 found.) The variable @code{final_sequence} is null when not
6400 processing a sequence, otherwise it contains the @code{sequence} rtx
6404 @defmac REGISTER_PREFIX
6405 @defmacx LOCAL_LABEL_PREFIX
6406 @defmacx USER_LABEL_PREFIX
6407 @defmacx IMMEDIATE_PREFIX
6408 If defined, C string expressions to be used for the @samp{%R}, @samp{%L},
6409 @samp{%U}, and @samp{%I} options of @code{asm_fprintf} (see
6410 @file{final.c}). These are useful when a single @file{md} file must
6411 support multiple assembler formats. In that case, the various @file{tm.h}
6412 files can define these macros differently.
6415 @defmac ASM_FPRINTF_EXTENSIONS (@var{file}, @var{argptr}, @var{format})
6416 If defined this macro should expand to a series of @code{case}
6417 statements which will be parsed inside the @code{switch} statement of
6418 the @code{asm_fprintf} function. This allows targets to define extra
6419 printf formats which may useful when generating their assembler
6420 statements. Note that uppercase letters are reserved for future
6421 generic extensions to asm_fprintf, and so are not available to target
6422 specific code. The output file is given by the parameter @var{file}.
6423 The varargs input pointer is @var{argptr} and the rest of the format
6424 string, starting the character after the one that is being switched
6425 upon, is pointed to by @var{format}.
6428 @defmac ASSEMBLER_DIALECT
6429 If your target supports multiple dialects of assembler language (such as
6430 different opcodes), define this macro as a C expression that gives the
6431 numeric index of the assembler language dialect to use, with zero as the
6434 If this macro is defined, you may use constructs of the form
6436 @samp{@{option0|option1|option2@dots{}@}}
6439 in the output templates of patterns (@pxref{Output Template}) or in the
6440 first argument of @code{asm_fprintf}. This construct outputs
6441 @samp{option0}, @samp{option1}, @samp{option2}, etc., if the value of
6442 @code{ASSEMBLER_DIALECT} is zero, one, two, etc. Any special characters
6443 within these strings retain their usual meaning. If there are fewer
6444 alternatives within the braces than the value of
6445 @code{ASSEMBLER_DIALECT}, the construct outputs nothing. If it's needed
6446 to print curly braces or @samp{|} character in assembler output directly,
6447 @samp{%@{}, @samp{%@}} and @samp{%|} can be used.
6449 If you do not define this macro, the characters @samp{@{}, @samp{|} and
6450 @samp{@}} do not have any special meaning when used in templates or
6451 operands to @code{asm_fprintf}.
6453 Define the macros @code{REGISTER_PREFIX}, @code{LOCAL_LABEL_PREFIX},
6454 @code{USER_LABEL_PREFIX} and @code{IMMEDIATE_PREFIX} if you can express
6455 the variations in assembler language syntax with that mechanism. Define
6456 @code{ASSEMBLER_DIALECT} and use the @samp{@{option0|option1@}} syntax
6457 if the syntax variant are larger and involve such things as different
6458 opcodes or operand order.
6461 @defmac ASM_OUTPUT_REG_PUSH (@var{stream}, @var{regno})
6462 A C expression to output to @var{stream} some assembler code
6463 which will push hard register number @var{regno} onto the stack.
6464 The code need not be optimal, since this macro is used only when
6468 @defmac ASM_OUTPUT_REG_POP (@var{stream}, @var{regno})
6469 A C expression to output to @var{stream} some assembler code
6470 which will pop hard register number @var{regno} off of the stack.
6471 The code need not be optimal, since this macro is used only when
6475 @node Dispatch Tables
6476 @subsection Output of Dispatch Tables
6478 @c prevent bad page break with this line
6479 This concerns dispatch tables.
6481 @cindex dispatch table
6482 @defmac ASM_OUTPUT_ADDR_DIFF_ELT (@var{stream}, @var{body}, @var{value}, @var{rel})
6483 A C statement to output to the stdio stream @var{stream} an assembler
6484 pseudo-instruction to generate a difference between two labels.
6485 @var{value} and @var{rel} are the numbers of two internal labels. The
6486 definitions of these labels are output using
6487 @code{(*targetm.asm_out.internal_label)}, and they must be printed in the same
6488 way here. For example,
6491 fprintf (@var{stream}, "\t.word L%d-L%d\n",
6492 @var{value}, @var{rel})
6495 You must provide this macro on machines where the addresses in a
6496 dispatch table are relative to the table's own address. If defined, GCC
6497 will also use this macro on all machines when producing PIC@.
6498 @var{body} is the body of the @code{ADDR_DIFF_VEC}; it is provided so that the
6499 mode and flags can be read.
6502 @defmac ASM_OUTPUT_ADDR_VEC_ELT (@var{stream}, @var{value})
6503 This macro should be provided on machines where the addresses
6504 in a dispatch table are absolute.
6506 The definition should be a C statement to output to the stdio stream
6507 @var{stream} an assembler pseudo-instruction to generate a reference to
6508 a label. @var{value} is the number of an internal label whose
6509 definition is output using @code{(*targetm.asm_out.internal_label)}.
6513 fprintf (@var{stream}, "\t.word L%d\n", @var{value})
6517 @defmac ASM_OUTPUT_CASE_LABEL (@var{stream}, @var{prefix}, @var{num}, @var{table})
6518 Define this if the label before a jump-table needs to be output
6519 specially. The first three arguments are the same as for
6520 @code{(*targetm.asm_out.internal_label)}; the fourth argument is the
6521 jump-table which follows (a @code{jump_table_data} containing an
6522 @code{addr_vec} or @code{addr_diff_vec}).
6524 This feature is used on system V to output a @code{swbeg} statement
6527 If this macro is not defined, these labels are output with
6528 @code{(*targetm.asm_out.internal_label)}.
6531 @defmac ASM_OUTPUT_CASE_END (@var{stream}, @var{num}, @var{table})
6532 Define this if something special must be output at the end of a
6533 jump-table. The definition should be a C statement to be executed
6534 after the assembler code for the table is written. It should write
6535 the appropriate code to stdio stream @var{stream}. The argument
6536 @var{table} is the jump-table insn, and @var{num} is the label-number
6537 of the preceding label.
6539 If this macro is not defined, nothing special is output at the end of
6543 @hook TARGET_ASM_EMIT_UNWIND_LABEL
6545 @hook TARGET_ASM_EMIT_EXCEPT_TABLE_LABEL
6547 @hook TARGET_ASM_EMIT_EXCEPT_PERSONALITY
6549 @hook TARGET_ASM_UNWIND_EMIT
6551 @hook TARGET_ASM_UNWIND_EMIT_BEFORE_INSN
6553 @node Exception Region Output
6554 @subsection Assembler Commands for Exception Regions
6556 @c prevent bad page break with this line
6558 This describes commands marking the start and the end of an exception
6561 @defmac EH_FRAME_SECTION_NAME
6562 If defined, a C string constant for the name of the section containing
6563 exception handling frame unwind information. If not defined, GCC will
6564 provide a default definition if the target supports named sections.
6565 @file{crtstuff.c} uses this macro to switch to the appropriate section.
6567 You should define this symbol if your target supports DWARF 2 frame
6568 unwind information and the default definition does not work.
6571 @defmac EH_FRAME_THROUGH_COLLECT2
6572 If defined, DWARF 2 frame unwind information will identified by
6573 specially named labels. The collect2 process will locate these
6574 labels and generate code to register the frames.
6576 This might be necessary, for instance, if the system linker will not
6577 place the eh_frames in-between the sentinals from @file{crtstuff.c},
6578 or if the system linker does garbage collection and sections cannot
6579 be marked as not to be collected.
6582 @defmac EH_TABLES_CAN_BE_READ_ONLY
6583 Define this macro to 1 if your target is such that no frame unwind
6584 information encoding used with non-PIC code will ever require a
6585 runtime relocation, but the linker may not support merging read-only
6586 and read-write sections into a single read-write section.
6589 @defmac MASK_RETURN_ADDR
6590 An rtx used to mask the return address found via @code{RETURN_ADDR_RTX}, so
6591 that it does not contain any extraneous set bits in it.
6594 @defmac DWARF2_UNWIND_INFO
6595 Define this macro to 0 if your target supports DWARF 2 frame unwind
6596 information, but it does not yet work with exception handling.
6597 Otherwise, if your target supports this information (if it defines
6598 @code{INCOMING_RETURN_ADDR_RTX} and @code{OBJECT_FORMAT_ELF}),
6599 GCC will provide a default definition of 1.
6602 @hook TARGET_EXCEPT_UNWIND_INFO
6603 This hook defines the mechanism that will be used for exception handling
6604 by the target. If the target has ABI specified unwind tables, the hook
6605 should return @code{UI_TARGET}. If the target is to use the
6606 @code{setjmp}/@code{longjmp}-based exception handling scheme, the hook
6607 should return @code{UI_SJLJ}. If the target supports DWARF 2 frame unwind
6608 information, the hook should return @code{UI_DWARF2}.
6610 A target may, if exceptions are disabled, choose to return @code{UI_NONE}.
6611 This may end up simplifying other parts of target-specific code. The
6612 default implementation of this hook never returns @code{UI_NONE}.
6614 Note that the value returned by this hook should be constant. It should
6615 not depend on anything except the command-line switches described by
6616 @var{opts}. In particular, the
6617 setting @code{UI_SJLJ} must be fixed at compiler start-up as C pre-processor
6618 macros and builtin functions related to exception handling are set up
6619 depending on this setting.
6621 The default implementation of the hook first honors the
6622 @option{--enable-sjlj-exceptions} configure option, then
6623 @code{DWARF2_UNWIND_INFO}, and finally defaults to @code{UI_SJLJ}. If
6624 @code{DWARF2_UNWIND_INFO} depends on command-line options, the target
6625 must define this hook so that @var{opts} is used correctly.
6628 @hook TARGET_UNWIND_TABLES_DEFAULT
6629 This variable should be set to @code{true} if the target ABI requires unwinding
6630 tables even when exceptions are not used. It must not be modified by
6631 command-line option processing.
6634 @defmac DONT_USE_BUILTIN_SETJMP
6635 Define this macro to 1 if the @code{setjmp}/@code{longjmp}-based scheme
6636 should use the @code{setjmp}/@code{longjmp} functions from the C library
6637 instead of the @code{__builtin_setjmp}/@code{__builtin_longjmp} machinery.
6640 @defmac JMP_BUF_SIZE
6641 This macro has no effect unless @code{DONT_USE_BUILTIN_SETJMP} is also
6642 defined. Define this macro if the default size of @code{jmp_buf} buffer
6643 for the @code{setjmp}/@code{longjmp}-based exception handling mechanism
6644 is not large enough, or if it is much too large.
6645 The default size is @code{FIRST_PSEUDO_REGISTER * sizeof(void *)}.
6648 @defmac DWARF_CIE_DATA_ALIGNMENT
6649 This macro need only be defined if the target might save registers in the
6650 function prologue at an offset to the stack pointer that is not aligned to
6651 @code{UNITS_PER_WORD}. The definition should be the negative minimum
6652 alignment if @code{STACK_GROWS_DOWNWARD} is true, and the positive
6653 minimum alignment otherwise. @xref{SDB and DWARF}. Only applicable if
6654 the target supports DWARF 2 frame unwind information.
6657 @hook TARGET_TERMINATE_DW2_EH_FRAME_INFO
6659 @hook TARGET_DWARF_REGISTER_SPAN
6661 @hook TARGET_DWARF_FRAME_REG_MODE
6663 @hook TARGET_INIT_DWARF_REG_SIZES_EXTRA
6665 @hook TARGET_ASM_TTYPE
6667 @hook TARGET_ARM_EABI_UNWINDER
6669 @node Alignment Output
6670 @subsection Assembler Commands for Alignment
6672 @c prevent bad page break with this line
6673 This describes commands for alignment.
6675 @defmac JUMP_ALIGN (@var{label})
6676 The alignment (log base 2) to put in front of @var{label}, which is
6677 a common destination of jumps and has no fallthru incoming edge.
6679 This macro need not be defined if you don't want any special alignment
6680 to be done at such a time. Most machine descriptions do not currently
6683 Unless it's necessary to inspect the @var{label} parameter, it is better
6684 to set the variable @var{align_jumps} in the target's
6685 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
6686 selection in @var{align_jumps} in a @code{JUMP_ALIGN} implementation.
6689 @hook TARGET_ASM_JUMP_ALIGN_MAX_SKIP
6691 @defmac LABEL_ALIGN_AFTER_BARRIER (@var{label})
6692 The alignment (log base 2) to put in front of @var{label}, which follows
6695 This macro need not be defined if you don't want any special alignment
6696 to be done at such a time. Most machine descriptions do not currently
6700 @hook TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP
6702 @defmac LOOP_ALIGN (@var{label})
6703 The alignment (log base 2) to put in front of @var{label} that heads
6704 a frequently executed basic block (usually the header of a loop).
6706 This macro need not be defined if you don't want any special alignment
6707 to be done at such a time. Most machine descriptions do not currently
6710 Unless it's necessary to inspect the @var{label} parameter, it is better
6711 to set the variable @code{align_loops} in the target's
6712 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
6713 selection in @code{align_loops} in a @code{LOOP_ALIGN} implementation.
6716 @hook TARGET_ASM_LOOP_ALIGN_MAX_SKIP
6718 @defmac LABEL_ALIGN (@var{label})
6719 The alignment (log base 2) to put in front of @var{label}.
6720 If @code{LABEL_ALIGN_AFTER_BARRIER} / @code{LOOP_ALIGN} specify a different alignment,
6721 the maximum of the specified values is used.
6723 Unless it's necessary to inspect the @var{label} parameter, it is better
6724 to set the variable @code{align_labels} in the target's
6725 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
6726 selection in @code{align_labels} in a @code{LABEL_ALIGN} implementation.
6729 @hook TARGET_ASM_LABEL_ALIGN_MAX_SKIP
6731 @defmac ASM_OUTPUT_SKIP (@var{stream}, @var{nbytes})
6732 A C statement to output to the stdio stream @var{stream} an assembler
6733 instruction to advance the location counter by @var{nbytes} bytes.
6734 Those bytes should be zero when loaded. @var{nbytes} will be a C
6735 expression of type @code{unsigned HOST_WIDE_INT}.
6738 @defmac ASM_NO_SKIP_IN_TEXT
6739 Define this macro if @code{ASM_OUTPUT_SKIP} should not be used in the
6740 text section because it fails to put zeros in the bytes that are skipped.
6741 This is true on many Unix systems, where the pseudo--op to skip bytes
6742 produces no-op instructions rather than zeros when used in the text
6746 @defmac ASM_OUTPUT_ALIGN (@var{stream}, @var{power})
6747 A C statement to output to the stdio stream @var{stream} an assembler
6748 command to advance the location counter to a multiple of 2 to the
6749 @var{power} bytes. @var{power} will be a C expression of type @code{int}.
6752 @defmac ASM_OUTPUT_ALIGN_WITH_NOP (@var{stream}, @var{power})
6753 Like @code{ASM_OUTPUT_ALIGN}, except that the ``nop'' instruction is used
6754 for padding, if necessary.
6757 @defmac ASM_OUTPUT_MAX_SKIP_ALIGN (@var{stream}, @var{power}, @var{max_skip})
6758 A C statement to output to the stdio stream @var{stream} an assembler
6759 command to advance the location counter to a multiple of 2 to the
6760 @var{power} bytes, but only if @var{max_skip} or fewer bytes are needed to
6761 satisfy the alignment request. @var{power} and @var{max_skip} will be
6762 a C expression of type @code{int}.
6766 @node Debugging Info
6767 @section Controlling Debugging Information Format
6769 @c prevent bad page break with this line
6770 This describes how to specify debugging information.
6773 * All Debuggers:: Macros that affect all debugging formats uniformly.
6774 * DBX Options:: Macros enabling specific options in DBX format.
6775 * DBX Hooks:: Hook macros for varying DBX format.
6776 * File Names and DBX:: Macros controlling output of file names in DBX format.
6777 * SDB and DWARF:: Macros for SDB (COFF) and DWARF formats.
6778 * VMS Debug:: Macros for VMS debug format.
6782 @subsection Macros Affecting All Debugging Formats
6784 @c prevent bad page break with this line
6785 These macros affect all debugging formats.
6787 @defmac DBX_REGISTER_NUMBER (@var{regno})
6788 A C expression that returns the DBX register number for the compiler
6789 register number @var{regno}. In the default macro provided, the value
6790 of this expression will be @var{regno} itself. But sometimes there are
6791 some registers that the compiler knows about and DBX does not, or vice
6792 versa. In such cases, some register may need to have one number in the
6793 compiler and another for DBX@.
6795 If two registers have consecutive numbers inside GCC, and they can be
6796 used as a pair to hold a multiword value, then they @emph{must} have
6797 consecutive numbers after renumbering with @code{DBX_REGISTER_NUMBER}.
6798 Otherwise, debuggers will be unable to access such a pair, because they
6799 expect register pairs to be consecutive in their own numbering scheme.
6801 If you find yourself defining @code{DBX_REGISTER_NUMBER} in way that
6802 does not preserve register pairs, then what you must do instead is
6803 redefine the actual register numbering scheme.
6806 @defmac DEBUGGER_AUTO_OFFSET (@var{x})
6807 A C expression that returns the integer offset value for an automatic
6808 variable having address @var{x} (an RTL expression). The default
6809 computation assumes that @var{x} is based on the frame-pointer and
6810 gives the offset from the frame-pointer. This is required for targets
6811 that produce debugging output for DBX or COFF-style debugging output
6812 for SDB and allow the frame-pointer to be eliminated when the
6813 @option{-g} options is used.
6816 @defmac DEBUGGER_ARG_OFFSET (@var{offset}, @var{x})
6817 A C expression that returns the integer offset value for an argument
6818 having address @var{x} (an RTL expression). The nominal offset is
6822 @defmac PREFERRED_DEBUGGING_TYPE
6823 A C expression that returns the type of debugging output GCC should
6824 produce when the user specifies just @option{-g}. Define
6825 this if you have arranged for GCC to support more than one format of
6826 debugging output. Currently, the allowable values are @code{DBX_DEBUG},
6827 @code{SDB_DEBUG}, @code{DWARF_DEBUG}, @code{DWARF2_DEBUG},
6828 @code{XCOFF_DEBUG}, @code{VMS_DEBUG}, and @code{VMS_AND_DWARF2_DEBUG}.
6830 When the user specifies @option{-ggdb}, GCC normally also uses the
6831 value of this macro to select the debugging output format, but with two
6832 exceptions. If @code{DWARF2_DEBUGGING_INFO} is defined, GCC uses the
6833 value @code{DWARF2_DEBUG}. Otherwise, if @code{DBX_DEBUGGING_INFO} is
6834 defined, GCC uses @code{DBX_DEBUG}.
6836 The value of this macro only affects the default debugging output; the
6837 user can always get a specific type of output by using @option{-gstabs},
6838 @option{-gcoff}, @option{-gdwarf-2}, @option{-gxcoff}, or @option{-gvms}.
6842 @subsection Specific Options for DBX Output
6844 @c prevent bad page break with this line
6845 These are specific options for DBX output.
6847 @defmac DBX_DEBUGGING_INFO
6848 Define this macro if GCC should produce debugging output for DBX
6849 in response to the @option{-g} option.
6852 @defmac XCOFF_DEBUGGING_INFO
6853 Define this macro if GCC should produce XCOFF format debugging output
6854 in response to the @option{-g} option. This is a variant of DBX format.
6857 @defmac DEFAULT_GDB_EXTENSIONS
6858 Define this macro to control whether GCC should by default generate
6859 GDB's extended version of DBX debugging information (assuming DBX-format
6860 debugging information is enabled at all). If you don't define the
6861 macro, the default is 1: always generate the extended information
6862 if there is any occasion to.
6865 @defmac DEBUG_SYMS_TEXT
6866 Define this macro if all @code{.stabs} commands should be output while
6867 in the text section.
6870 @defmac ASM_STABS_OP
6871 A C string constant, including spacing, naming the assembler pseudo op to
6872 use instead of @code{"\t.stabs\t"} to define an ordinary debugging symbol.
6873 If you don't define this macro, @code{"\t.stabs\t"} is used. This macro
6874 applies only to DBX debugging information format.
6877 @defmac ASM_STABD_OP
6878 A C string constant, including spacing, naming the assembler pseudo op to
6879 use instead of @code{"\t.stabd\t"} to define a debugging symbol whose
6880 value is the current location. If you don't define this macro,
6881 @code{"\t.stabd\t"} is used. This macro applies only to DBX debugging
6885 @defmac ASM_STABN_OP
6886 A C string constant, including spacing, naming the assembler pseudo op to
6887 use instead of @code{"\t.stabn\t"} to define a debugging symbol with no
6888 name. If you don't define this macro, @code{"\t.stabn\t"} is used. This
6889 macro applies only to DBX debugging information format.
6892 @defmac DBX_NO_XREFS
6893 Define this macro if DBX on your system does not support the construct
6894 @samp{xs@var{tagname}}. On some systems, this construct is used to
6895 describe a forward reference to a structure named @var{tagname}.
6896 On other systems, this construct is not supported at all.
6899 @defmac DBX_CONTIN_LENGTH
6900 A symbol name in DBX-format debugging information is normally
6901 continued (split into two separate @code{.stabs} directives) when it
6902 exceeds a certain length (by default, 80 characters). On some
6903 operating systems, DBX requires this splitting; on others, splitting
6904 must not be done. You can inhibit splitting by defining this macro
6905 with the value zero. You can override the default splitting-length by
6906 defining this macro as an expression for the length you desire.
6909 @defmac DBX_CONTIN_CHAR
6910 Normally continuation is indicated by adding a @samp{\} character to
6911 the end of a @code{.stabs} string when a continuation follows. To use
6912 a different character instead, define this macro as a character
6913 constant for the character you want to use. Do not define this macro
6914 if backslash is correct for your system.
6917 @defmac DBX_STATIC_STAB_DATA_SECTION
6918 Define this macro if it is necessary to go to the data section before
6919 outputting the @samp{.stabs} pseudo-op for a non-global static
6923 @defmac DBX_TYPE_DECL_STABS_CODE
6924 The value to use in the ``code'' field of the @code{.stabs} directive
6925 for a typedef. The default is @code{N_LSYM}.
6928 @defmac DBX_STATIC_CONST_VAR_CODE
6929 The value to use in the ``code'' field of the @code{.stabs} directive
6930 for a static variable located in the text section. DBX format does not
6931 provide any ``right'' way to do this. The default is @code{N_FUN}.
6934 @defmac DBX_REGPARM_STABS_CODE
6935 The value to use in the ``code'' field of the @code{.stabs} directive
6936 for a parameter passed in registers. DBX format does not provide any
6937 ``right'' way to do this. The default is @code{N_RSYM}.
6940 @defmac DBX_REGPARM_STABS_LETTER
6941 The letter to use in DBX symbol data to identify a symbol as a parameter
6942 passed in registers. DBX format does not customarily provide any way to
6943 do this. The default is @code{'P'}.
6946 @defmac DBX_FUNCTION_FIRST
6947 Define this macro if the DBX information for a function and its
6948 arguments should precede the assembler code for the function. Normally,
6949 in DBX format, the debugging information entirely follows the assembler
6953 @defmac DBX_BLOCKS_FUNCTION_RELATIVE
6954 Define this macro, with value 1, if the value of a symbol describing
6955 the scope of a block (@code{N_LBRAC} or @code{N_RBRAC}) should be
6956 relative to the start of the enclosing function. Normally, GCC uses
6957 an absolute address.
6960 @defmac DBX_LINES_FUNCTION_RELATIVE
6961 Define this macro, with value 1, if the value of a symbol indicating
6962 the current line number (@code{N_SLINE}) should be relative to the
6963 start of the enclosing function. Normally, GCC uses an absolute address.
6966 @defmac DBX_USE_BINCL
6967 Define this macro if GCC should generate @code{N_BINCL} and
6968 @code{N_EINCL} stabs for included header files, as on Sun systems. This
6969 macro also directs GCC to output a type number as a pair of a file
6970 number and a type number within the file. Normally, GCC does not
6971 generate @code{N_BINCL} or @code{N_EINCL} stabs, and it outputs a single
6972 number for a type number.
6976 @subsection Open-Ended Hooks for DBX Format
6978 @c prevent bad page break with this line
6979 These are hooks for DBX format.
6981 @defmac DBX_OUTPUT_SOURCE_LINE (@var{stream}, @var{line}, @var{counter})
6982 A C statement to output DBX debugging information before code for line
6983 number @var{line} of the current source file to the stdio stream
6984 @var{stream}. @var{counter} is the number of time the macro was
6985 invoked, including the current invocation; it is intended to generate
6986 unique labels in the assembly output.
6988 This macro should not be defined if the default output is correct, or
6989 if it can be made correct by defining @code{DBX_LINES_FUNCTION_RELATIVE}.
6992 @defmac NO_DBX_FUNCTION_END
6993 Some stabs encapsulation formats (in particular ECOFF), cannot handle the
6994 @code{.stabs "",N_FUN,,0,0,Lscope-function-1} gdb dbx extension construct.
6995 On those machines, define this macro to turn this feature off without
6996 disturbing the rest of the gdb extensions.
6999 @defmac NO_DBX_BNSYM_ENSYM
7000 Some assemblers cannot handle the @code{.stabd BNSYM/ENSYM,0,0} gdb dbx
7001 extension construct. On those machines, define this macro to turn this
7002 feature off without disturbing the rest of the gdb extensions.
7005 @node File Names and DBX
7006 @subsection File Names in DBX Format
7008 @c prevent bad page break with this line
7009 This describes file names in DBX format.
7011 @defmac DBX_OUTPUT_MAIN_SOURCE_FILENAME (@var{stream}, @var{name})
7012 A C statement to output DBX debugging information to the stdio stream
7013 @var{stream}, which indicates that file @var{name} is the main source
7014 file---the file specified as the input file for compilation.
7015 This macro is called only once, at the beginning of compilation.
7017 This macro need not be defined if the standard form of output
7018 for DBX debugging information is appropriate.
7020 It may be necessary to refer to a label equal to the beginning of the
7021 text section. You can use @samp{assemble_name (stream, ltext_label_name)}
7022 to do so. If you do this, you must also set the variable
7023 @var{used_ltext_label_name} to @code{true}.
7026 @defmac NO_DBX_MAIN_SOURCE_DIRECTORY
7027 Define this macro, with value 1, if GCC should not emit an indication
7028 of the current directory for compilation and current source language at
7029 the beginning of the file.
7032 @defmac NO_DBX_GCC_MARKER
7033 Define this macro, with value 1, if GCC should not emit an indication
7034 that this object file was compiled by GCC@. The default is to emit
7035 an @code{N_OPT} stab at the beginning of every source file, with
7036 @samp{gcc2_compiled.} for the string and value 0.
7039 @defmac DBX_OUTPUT_MAIN_SOURCE_FILE_END (@var{stream}, @var{name})
7040 A C statement to output DBX debugging information at the end of
7041 compilation of the main source file @var{name}. Output should be
7042 written to the stdio stream @var{stream}.
7044 If you don't define this macro, nothing special is output at the end
7045 of compilation, which is correct for most machines.
7048 @defmac DBX_OUTPUT_NULL_N_SO_AT_MAIN_SOURCE_FILE_END
7049 Define this macro @emph{instead of} defining
7050 @code{DBX_OUTPUT_MAIN_SOURCE_FILE_END}, if what needs to be output at
7051 the end of compilation is an @code{N_SO} stab with an empty string,
7052 whose value is the highest absolute text address in the file.
7057 @subsection Macros for SDB and DWARF Output
7059 @c prevent bad page break with this line
7060 Here are macros for SDB and DWARF output.
7062 @defmac SDB_DEBUGGING_INFO
7063 Define this macro to 1 if GCC should produce COFF-style debugging output
7064 for SDB in response to the @option{-g} option.
7067 @defmac DWARF2_DEBUGGING_INFO
7068 Define this macro if GCC should produce dwarf version 2 format
7069 debugging output in response to the @option{-g} option.
7071 @hook TARGET_DWARF_CALLING_CONVENTION
7073 To support optional call frame debugging information, you must also
7074 define @code{INCOMING_RETURN_ADDR_RTX} and either set
7075 @code{RTX_FRAME_RELATED_P} on the prologue insns if you use RTL for the
7076 prologue, or call @code{dwarf2out_def_cfa} and @code{dwarf2out_reg_save}
7077 as appropriate from @code{TARGET_ASM_FUNCTION_PROLOGUE} if you don't.
7080 @defmac DWARF2_FRAME_INFO
7081 Define this macro to a nonzero value if GCC should always output
7082 Dwarf 2 frame information. If @code{TARGET_EXCEPT_UNWIND_INFO}
7083 (@pxref{Exception Region Output}) returns @code{UI_DWARF2}, and
7084 exceptions are enabled, GCC will output this information not matter
7085 how you define @code{DWARF2_FRAME_INFO}.
7088 @hook TARGET_DEBUG_UNWIND_INFO
7090 @defmac DWARF2_ASM_LINE_DEBUG_INFO
7091 Define this macro to be a nonzero value if the assembler can generate Dwarf 2
7092 line debug info sections. This will result in much more compact line number
7093 tables, and hence is desirable if it works.
7096 @hook TARGET_WANT_DEBUG_PUB_SECTIONS
7098 @hook TARGET_DELAY_SCHED2
7100 @hook TARGET_DELAY_VARTRACK
7102 @hook TARGET_NO_REGISTER_ALLOCATION
7104 @defmac ASM_OUTPUT_DWARF_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
7105 A C statement to issue assembly directives that create a difference
7106 @var{lab1} minus @var{lab2}, using an integer of the given @var{size}.
7109 @defmac ASM_OUTPUT_DWARF_VMS_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
7110 A C statement to issue assembly directives that create a difference
7111 between the two given labels in system defined units, e.g. instruction
7112 slots on IA64 VMS, using an integer of the given size.
7115 @defmac ASM_OUTPUT_DWARF_OFFSET (@var{stream}, @var{size}, @var{label}, @var{offset}, @var{section})
7116 A C statement to issue assembly directives that create a
7117 section-relative reference to the given @var{label} plus @var{offset}, using
7118 an integer of the given @var{size}. The label is known to be defined in the
7119 given @var{section}.
7122 @defmac ASM_OUTPUT_DWARF_PCREL (@var{stream}, @var{size}, @var{label})
7123 A C statement to issue assembly directives that create a self-relative
7124 reference to the given @var{label}, using an integer of the given @var{size}.
7127 @defmac ASM_OUTPUT_DWARF_DATAREL (@var{stream}, @var{size}, @var{label})
7128 A C statement to issue assembly directives that create a reference to the
7129 given @var{label} relative to the dbase, using an integer of the given @var{size}.
7132 @defmac ASM_OUTPUT_DWARF_TABLE_REF (@var{label})
7133 A C statement to issue assembly directives that create a reference to
7134 the DWARF table identifier @var{label} from the current section. This
7135 is used on some systems to avoid garbage collecting a DWARF table which
7136 is referenced by a function.
7139 @hook TARGET_ASM_OUTPUT_DWARF_DTPREL
7141 @defmac PUT_SDB_@dots{}
7142 Define these macros to override the assembler syntax for the special
7143 SDB assembler directives. See @file{sdbout.c} for a list of these
7144 macros and their arguments. If the standard syntax is used, you need
7145 not define them yourself.
7149 Some assemblers do not support a semicolon as a delimiter, even between
7150 SDB assembler directives. In that case, define this macro to be the
7151 delimiter to use (usually @samp{\n}). It is not necessary to define
7152 a new set of @code{PUT_SDB_@var{op}} macros if this is the only change
7156 @defmac SDB_ALLOW_UNKNOWN_REFERENCES
7157 Define this macro to allow references to unknown structure,
7158 union, or enumeration tags to be emitted. Standard COFF does not
7159 allow handling of unknown references, MIPS ECOFF has support for
7163 @defmac SDB_ALLOW_FORWARD_REFERENCES
7164 Define this macro to allow references to structure, union, or
7165 enumeration tags that have not yet been seen to be handled. Some
7166 assemblers choke if forward tags are used, while some require it.
7169 @defmac SDB_OUTPUT_SOURCE_LINE (@var{stream}, @var{line})
7170 A C statement to output SDB debugging information before code for line
7171 number @var{line} of the current source file to the stdio stream
7172 @var{stream}. The default is to emit an @code{.ln} directive.
7177 @subsection Macros for VMS Debug Format
7179 @c prevent bad page break with this line
7180 Here are macros for VMS debug format.
7182 @defmac VMS_DEBUGGING_INFO
7183 Define this macro if GCC should produce debugging output for VMS
7184 in response to the @option{-g} option. The default behavior for VMS
7185 is to generate minimal debug info for a traceback in the absence of
7186 @option{-g} unless explicitly overridden with @option{-g0}. This
7187 behavior is controlled by @code{TARGET_OPTION_OPTIMIZATION} and
7188 @code{TARGET_OPTION_OVERRIDE}.
7191 @node Floating Point
7192 @section Cross Compilation and Floating Point
7193 @cindex cross compilation and floating point
7194 @cindex floating point and cross compilation
7196 While all modern machines use twos-complement representation for integers,
7197 there are a variety of representations for floating point numbers. This
7198 means that in a cross-compiler the representation of floating point numbers
7199 in the compiled program may be different from that used in the machine
7200 doing the compilation.
7202 Because different representation systems may offer different amounts of
7203 range and precision, all floating point constants must be represented in
7204 the target machine's format. Therefore, the cross compiler cannot
7205 safely use the host machine's floating point arithmetic; it must emulate
7206 the target's arithmetic. To ensure consistency, GCC always uses
7207 emulation to work with floating point values, even when the host and
7208 target floating point formats are identical.
7210 The following macros are provided by @file{real.h} for the compiler to
7211 use. All parts of the compiler which generate or optimize
7212 floating-point calculations must use these macros. They may evaluate
7213 their operands more than once, so operands must not have side effects.
7215 @defmac REAL_VALUE_TYPE
7216 The C data type to be used to hold a floating point value in the target
7217 machine's format. Typically this is a @code{struct} containing an
7218 array of @code{HOST_WIDE_INT}, but all code should treat it as an opaque
7222 @deftypefn Macro HOST_WIDE_INT REAL_VALUE_FIX (REAL_VALUE_TYPE @var{x})
7223 Truncates @var{x} to a signed integer, rounding toward zero.
7226 @deftypefn Macro {unsigned HOST_WIDE_INT} REAL_VALUE_UNSIGNED_FIX (REAL_VALUE_TYPE @var{x})
7227 Truncates @var{x} to an unsigned integer, rounding toward zero. If
7228 @var{x} is negative, returns zero.
7231 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ATOF (const char *@var{string}, machine_mode @var{mode})
7232 Converts @var{string} into a floating point number in the target machine's
7233 representation for mode @var{mode}. This routine can handle both
7234 decimal and hexadecimal floating point constants, using the syntax
7235 defined by the C language for both.
7238 @deftypefn Macro int REAL_VALUE_NEGATIVE (REAL_VALUE_TYPE @var{x})
7239 Returns 1 if @var{x} is negative (including negative zero), 0 otherwise.
7242 @deftypefn Macro int REAL_VALUE_ISINF (REAL_VALUE_TYPE @var{x})
7243 Determines whether @var{x} represents infinity (positive or negative).
7246 @deftypefn Macro int REAL_VALUE_ISNAN (REAL_VALUE_TYPE @var{x})
7247 Determines whether @var{x} represents a ``NaN'' (not-a-number).
7250 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_NEGATE (REAL_VALUE_TYPE @var{x})
7251 Returns the negative of the floating point value @var{x}.
7254 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ABS (REAL_VALUE_TYPE @var{x})
7255 Returns the absolute value of @var{x}.
7258 @node Mode Switching
7259 @section Mode Switching Instructions
7260 @cindex mode switching
7261 The following macros control mode switching optimizations:
7263 @defmac OPTIMIZE_MODE_SWITCHING (@var{entity})
7264 Define this macro if the port needs extra instructions inserted for mode
7265 switching in an optimizing compilation.
7267 For an example, the SH4 can perform both single and double precision
7268 floating point operations, but to perform a single precision operation,
7269 the FPSCR PR bit has to be cleared, while for a double precision
7270 operation, this bit has to be set. Changing the PR bit requires a general
7271 purpose register as a scratch register, hence these FPSCR sets have to
7272 be inserted before reload, i.e.@: you can't put this into instruction emitting
7273 or @code{TARGET_MACHINE_DEPENDENT_REORG}.
7275 You can have multiple entities that are mode-switched, and select at run time
7276 which entities actually need it. @code{OPTIMIZE_MODE_SWITCHING} should
7277 return nonzero for any @var{entity} that needs mode-switching.
7278 If you define this macro, you also have to define
7279 @code{NUM_MODES_FOR_MODE_SWITCHING}, @code{TARGET_MODE_NEEDED},
7280 @code{TARGET_MODE_PRIORITY} and @code{TARGET_MODE_EMIT}.
7281 @code{TARGET_MODE_AFTER}, @code{TARGET_MODE_ENTRY}, and @code{TARGET_MODE_EXIT}
7285 @defmac NUM_MODES_FOR_MODE_SWITCHING
7286 If you define @code{OPTIMIZE_MODE_SWITCHING}, you have to define this as
7287 initializer for an array of integers. Each initializer element
7288 N refers to an entity that needs mode switching, and specifies the number
7289 of different modes that might need to be set for this entity.
7290 The position of the initializer in the initializer---starting counting at
7291 zero---determines the integer that is used to refer to the mode-switched
7293 In macros that take mode arguments / yield a mode result, modes are
7294 represented as numbers 0 @dots{} N @minus{} 1. N is used to specify that no mode
7295 switch is needed / supplied.
7298 @hook TARGET_MODE_EMIT
7300 @hook TARGET_MODE_NEEDED
7302 @hook TARGET_MODE_AFTER
7304 @hook TARGET_MODE_ENTRY
7306 @hook TARGET_MODE_EXIT
7308 @hook TARGET_MODE_PRIORITY
7310 @node Target Attributes
7311 @section Defining target-specific uses of @code{__attribute__}
7312 @cindex target attributes
7313 @cindex machine attributes
7314 @cindex attributes, target-specific
7316 Target-specific attributes may be defined for functions, data and types.
7317 These are described using the following target hooks; they also need to
7318 be documented in @file{extend.texi}.
7320 @hook TARGET_ATTRIBUTE_TABLE
7322 @hook TARGET_ATTRIBUTE_TAKES_IDENTIFIER_P
7324 @hook TARGET_COMP_TYPE_ATTRIBUTES
7326 @hook TARGET_SET_DEFAULT_TYPE_ATTRIBUTES
7328 @hook TARGET_MERGE_TYPE_ATTRIBUTES
7330 @hook TARGET_MERGE_DECL_ATTRIBUTES
7332 @hook TARGET_VALID_DLLIMPORT_ATTRIBUTE_P
7334 @defmac TARGET_DECLSPEC
7335 Define this macro to a nonzero value if you want to treat
7336 @code{__declspec(X)} as equivalent to @code{__attribute((X))}. By
7337 default, this behavior is enabled only for targets that define
7338 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES}. The current implementation
7339 of @code{__declspec} is via a built-in macro, but you should not rely
7340 on this implementation detail.
7343 @hook TARGET_INSERT_ATTRIBUTES
7345 @hook TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P
7347 @hook TARGET_OPTION_VALID_ATTRIBUTE_P
7349 @hook TARGET_OPTION_SAVE
7351 @hook TARGET_OPTION_RESTORE
7353 @hook TARGET_OPTION_POST_STREAM_IN
7355 @hook TARGET_OPTION_PRINT
7357 @hook TARGET_OPTION_PRAGMA_PARSE
7359 @hook TARGET_OPTION_OVERRIDE
7361 @hook TARGET_OPTION_FUNCTION_VERSIONS
7363 @hook TARGET_CAN_INLINE_P
7365 @hook TARGET_RELAYOUT_FUNCTION
7368 @section Emulating TLS
7369 @cindex Emulated TLS
7371 For targets whose psABI does not provide Thread Local Storage via
7372 specific relocations and instruction sequences, an emulation layer is
7373 used. A set of target hooks allows this emulation layer to be
7374 configured for the requirements of a particular target. For instance
7375 the psABI may in fact specify TLS support in terms of an emulation
7378 The emulation layer works by creating a control object for every TLS
7379 object. To access the TLS object, a lookup function is provided
7380 which, when given the address of the control object, will return the
7381 address of the current thread's instance of the TLS object.
7383 @hook TARGET_EMUTLS_GET_ADDRESS
7385 @hook TARGET_EMUTLS_REGISTER_COMMON
7387 @hook TARGET_EMUTLS_VAR_SECTION
7389 @hook TARGET_EMUTLS_TMPL_SECTION
7391 @hook TARGET_EMUTLS_VAR_PREFIX
7393 @hook TARGET_EMUTLS_TMPL_PREFIX
7395 @hook TARGET_EMUTLS_VAR_FIELDS
7397 @hook TARGET_EMUTLS_VAR_INIT
7399 @hook TARGET_EMUTLS_VAR_ALIGN_FIXED
7401 @hook TARGET_EMUTLS_DEBUG_FORM_TLS_ADDRESS
7403 @node MIPS Coprocessors
7404 @section Defining coprocessor specifics for MIPS targets.
7405 @cindex MIPS coprocessor-definition macros
7407 The MIPS specification allows MIPS implementations to have as many as 4
7408 coprocessors, each with as many as 32 private registers. GCC supports
7409 accessing these registers and transferring values between the registers
7410 and memory using asm-ized variables. For example:
7413 register unsigned int cp0count asm ("c0r1");
7419 (``c0r1'' is the default name of register 1 in coprocessor 0; alternate
7420 names may be added as described below, or the default names may be
7421 overridden entirely in @code{SUBTARGET_CONDITIONAL_REGISTER_USAGE}.)
7423 Coprocessor registers are assumed to be epilogue-used; sets to them will
7424 be preserved even if it does not appear that the register is used again
7425 later in the function.
7427 Another note: according to the MIPS spec, coprocessor 1 (if present) is
7428 the FPU@. One accesses COP1 registers through standard mips
7429 floating-point support; they are not included in this mechanism.
7432 @section Parameters for Precompiled Header Validity Checking
7433 @cindex parameters, precompiled headers
7435 @hook TARGET_GET_PCH_VALIDITY
7437 @hook TARGET_PCH_VALID_P
7439 @hook TARGET_CHECK_PCH_TARGET_FLAGS
7441 @hook TARGET_PREPARE_PCH_SAVE
7444 @section C++ ABI parameters
7445 @cindex parameters, c++ abi
7447 @hook TARGET_CXX_GUARD_TYPE
7449 @hook TARGET_CXX_GUARD_MASK_BIT
7451 @hook TARGET_CXX_GET_COOKIE_SIZE
7453 @hook TARGET_CXX_COOKIE_HAS_SIZE
7455 @hook TARGET_CXX_IMPORT_EXPORT_CLASS
7457 @hook TARGET_CXX_CDTOR_RETURNS_THIS
7459 @hook TARGET_CXX_KEY_METHOD_MAY_BE_INLINE
7461 @hook TARGET_CXX_DETERMINE_CLASS_DATA_VISIBILITY
7463 @hook TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT
7465 @hook TARGET_CXX_LIBRARY_RTTI_COMDAT
7467 @hook TARGET_CXX_USE_AEABI_ATEXIT
7469 @hook TARGET_CXX_USE_ATEXIT_FOR_CXA_ATEXIT
7471 @hook TARGET_CXX_ADJUST_CLASS_AT_DEFINITION
7473 @hook TARGET_CXX_DECL_MANGLING_CONTEXT
7475 @node Named Address Spaces
7476 @section Adding support for named address spaces
7477 @cindex named address spaces
7479 The draft technical report of the ISO/IEC JTC1 S22 WG14 N1275
7480 standards committee, @cite{Programming Languages - C - Extensions to
7481 support embedded processors}, specifies a syntax for embedded
7482 processors to specify alternate address spaces. You can configure a
7483 GCC port to support section 5.1 of the draft report to add support for
7484 address spaces other than the default address space. These address
7485 spaces are new keywords that are similar to the @code{volatile} and
7486 @code{const} type attributes.
7488 Pointers to named address spaces can have a different size than
7489 pointers to the generic address space.
7491 For example, the SPU port uses the @code{__ea} address space to refer
7492 to memory in the host processor, rather than memory local to the SPU
7493 processor. Access to memory in the @code{__ea} address space involves
7494 issuing DMA operations to move data between the host processor and the
7495 local processor memory address space. Pointers in the @code{__ea}
7496 address space are either 32 bits or 64 bits based on the
7497 @option{-mea32} or @option{-mea64} switches (native SPU pointers are
7500 Internally, address spaces are represented as a small integer in the
7501 range 0 to 15 with address space 0 being reserved for the generic
7504 To register a named address space qualifier keyword with the C front end,
7505 the target may call the @code{c_register_addr_space} routine. For example,
7506 the SPU port uses the following to declare @code{__ea} as the keyword for
7507 named address space #1:
7509 #define ADDR_SPACE_EA 1
7510 c_register_addr_space ("__ea", ADDR_SPACE_EA);
7513 @hook TARGET_ADDR_SPACE_POINTER_MODE
7515 @hook TARGET_ADDR_SPACE_ADDRESS_MODE
7517 @hook TARGET_ADDR_SPACE_VALID_POINTER_MODE
7519 @hook TARGET_ADDR_SPACE_LEGITIMATE_ADDRESS_P
7521 @hook TARGET_ADDR_SPACE_LEGITIMIZE_ADDRESS
7523 @hook TARGET_ADDR_SPACE_SUBSET_P
7525 @hook TARGET_ADDR_SPACE_ZERO_ADDRESS_VALID
7527 @hook TARGET_ADDR_SPACE_CONVERT
7529 @hook TARGET_ADDR_SPACE_DEBUG
7531 @hook TARGET_ADDR_SPACE_DIAGNOSE_USAGE
7534 @section Miscellaneous Parameters
7535 @cindex parameters, miscellaneous
7537 @c prevent bad page break with this line
7538 Here are several miscellaneous parameters.
7540 @defmac HAS_LONG_COND_BRANCH
7541 Define this boolean macro to indicate whether or not your architecture
7542 has conditional branches that can span all of memory. It is used in
7543 conjunction with an optimization that partitions hot and cold basic
7544 blocks into separate sections of the executable. If this macro is
7545 set to false, gcc will convert any conditional branches that attempt
7546 to cross between sections into unconditional branches or indirect jumps.
7549 @defmac HAS_LONG_UNCOND_BRANCH
7550 Define this boolean macro to indicate whether or not your architecture
7551 has unconditional branches that can span all of memory. It is used in
7552 conjunction with an optimization that partitions hot and cold basic
7553 blocks into separate sections of the executable. If this macro is
7554 set to false, gcc will convert any unconditional branches that attempt
7555 to cross between sections into indirect jumps.
7558 @defmac CASE_VECTOR_MODE
7559 An alias for a machine mode name. This is the machine mode that
7560 elements of a jump-table should have.
7563 @defmac CASE_VECTOR_SHORTEN_MODE (@var{min_offset}, @var{max_offset}, @var{body})
7564 Optional: return the preferred mode for an @code{addr_diff_vec}
7565 when the minimum and maximum offset are known. If you define this,
7566 it enables extra code in branch shortening to deal with @code{addr_diff_vec}.
7567 To make this work, you also have to define @code{INSN_ALIGN} and
7568 make the alignment for @code{addr_diff_vec} explicit.
7569 The @var{body} argument is provided so that the offset_unsigned and scale
7570 flags can be updated.
7573 @defmac CASE_VECTOR_PC_RELATIVE
7574 Define this macro to be a C expression to indicate when jump-tables
7575 should contain relative addresses. You need not define this macro if
7576 jump-tables never contain relative addresses, or jump-tables should
7577 contain relative addresses only when @option{-fPIC} or @option{-fPIC}
7581 @hook TARGET_CASE_VALUES_THRESHOLD
7583 @defmac WORD_REGISTER_OPERATIONS
7584 Define this macro to 1 if operations between registers with integral mode
7585 smaller than a word are always performed on the entire register.
7586 Most RISC machines have this property and most CISC machines do not.
7589 @hook TARGET_MIN_ARITHMETIC_PRECISION
7591 @defmac LOAD_EXTEND_OP (@var{mem_mode})
7592 Define this macro to be a C expression indicating when insns that read
7593 memory in @var{mem_mode}, an integral mode narrower than a word, set the
7594 bits outside of @var{mem_mode} to be either the sign-extension or the
7595 zero-extension of the data read. Return @code{SIGN_EXTEND} for values
7596 of @var{mem_mode} for which the
7597 insn sign-extends, @code{ZERO_EXTEND} for which it zero-extends, and
7598 @code{UNKNOWN} for other modes.
7600 This macro is not called with @var{mem_mode} non-integral or with a width
7601 greater than or equal to @code{BITS_PER_WORD}, so you may return any
7602 value in this case. Do not define this macro if it would always return
7603 @code{UNKNOWN}. On machines where this macro is defined, you will normally
7604 define it as the constant @code{SIGN_EXTEND} or @code{ZERO_EXTEND}.
7606 You may return a non-@code{UNKNOWN} value even if for some hard registers
7607 the sign extension is not performed, if for the @code{REGNO_REG_CLASS}
7608 of these hard registers @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero
7609 when the @var{from} mode is @var{mem_mode} and the @var{to} mode is any
7610 integral mode larger than this but not larger than @code{word_mode}.
7612 You must return @code{UNKNOWN} if for some hard registers that allow this
7613 mode, @code{CANNOT_CHANGE_MODE_CLASS} says that they cannot change to
7614 @code{word_mode}, but that they can change to another integral mode that
7615 is larger then @var{mem_mode} but still smaller than @code{word_mode}.
7618 @defmac SHORT_IMMEDIATES_SIGN_EXTEND
7619 Define this macro to 1 if loading short immediate values into registers sign
7623 @hook TARGET_MIN_DIVISIONS_FOR_RECIP_MUL
7626 The maximum number of bytes that a single instruction can move quickly
7627 between memory and registers or between two memory locations.
7630 @defmac MAX_MOVE_MAX
7631 The maximum number of bytes that a single instruction can move quickly
7632 between memory and registers or between two memory locations. If this
7633 is undefined, the default is @code{MOVE_MAX}. Otherwise, it is the
7634 constant value that is the largest value that @code{MOVE_MAX} can have
7638 @defmac SHIFT_COUNT_TRUNCATED
7639 A C expression that is nonzero if on this machine the number of bits
7640 actually used for the count of a shift operation is equal to the number
7641 of bits needed to represent the size of the object being shifted. When
7642 this macro is nonzero, the compiler will assume that it is safe to omit
7643 a sign-extend, zero-extend, and certain bitwise `and' instructions that
7644 truncates the count of a shift operation. On machines that have
7645 instructions that act on bit-fields at variable positions, which may
7646 include `bit test' instructions, a nonzero @code{SHIFT_COUNT_TRUNCATED}
7647 also enables deletion of truncations of the values that serve as
7648 arguments to bit-field instructions.
7650 If both types of instructions truncate the count (for shifts) and
7651 position (for bit-field operations), or if no variable-position bit-field
7652 instructions exist, you should define this macro.
7654 However, on some machines, such as the 80386 and the 680x0, truncation
7655 only applies to shift operations and not the (real or pretended)
7656 bit-field operations. Define @code{SHIFT_COUNT_TRUNCATED} to be zero on
7657 such machines. Instead, add patterns to the @file{md} file that include
7658 the implied truncation of the shift instructions.
7660 You need not define this macro if it would always have the value of zero.
7663 @anchor{TARGET_SHIFT_TRUNCATION_MASK}
7664 @hook TARGET_SHIFT_TRUNCATION_MASK
7666 @defmac TRULY_NOOP_TRUNCATION (@var{outprec}, @var{inprec})
7667 A C expression which is nonzero if on this machine it is safe to
7668 ``convert'' an integer of @var{inprec} bits to one of @var{outprec}
7669 bits (where @var{outprec} is smaller than @var{inprec}) by merely
7670 operating on it as if it had only @var{outprec} bits.
7672 On many machines, this expression can be 1.
7674 @c rearranged this, removed the phrase "it is reported that". this was
7675 @c to fix an overfull hbox. --mew 10feb93
7676 When @code{TRULY_NOOP_TRUNCATION} returns 1 for a pair of sizes for
7677 modes for which @code{MODES_TIEABLE_P} is 0, suboptimal code can result.
7678 If this is the case, making @code{TRULY_NOOP_TRUNCATION} return 0 in
7679 such cases may improve things.
7682 @hook TARGET_MODE_REP_EXTENDED
7684 @defmac STORE_FLAG_VALUE
7685 A C expression describing the value returned by a comparison operator
7686 with an integral mode and stored by a store-flag instruction
7687 (@samp{cstore@var{mode}4}) when the condition is true. This description must
7688 apply to @emph{all} the @samp{cstore@var{mode}4} patterns and all the
7689 comparison operators whose results have a @code{MODE_INT} mode.
7691 A value of 1 or @minus{}1 means that the instruction implementing the
7692 comparison operator returns exactly 1 or @minus{}1 when the comparison is true
7693 and 0 when the comparison is false. Otherwise, the value indicates
7694 which bits of the result are guaranteed to be 1 when the comparison is
7695 true. This value is interpreted in the mode of the comparison
7696 operation, which is given by the mode of the first operand in the
7697 @samp{cstore@var{mode}4} pattern. Either the low bit or the sign bit of
7698 @code{STORE_FLAG_VALUE} be on. Presently, only those bits are used by
7701 If @code{STORE_FLAG_VALUE} is neither 1 or @minus{}1, the compiler will
7702 generate code that depends only on the specified bits. It can also
7703 replace comparison operators with equivalent operations if they cause
7704 the required bits to be set, even if the remaining bits are undefined.
7705 For example, on a machine whose comparison operators return an
7706 @code{SImode} value and where @code{STORE_FLAG_VALUE} is defined as
7707 @samp{0x80000000}, saying that just the sign bit is relevant, the
7711 (ne:SI (and:SI @var{x} (const_int @var{power-of-2})) (const_int 0))
7718 (ashift:SI @var{x} (const_int @var{n}))
7722 where @var{n} is the appropriate shift count to move the bit being
7723 tested into the sign bit.
7725 There is no way to describe a machine that always sets the low-order bit
7726 for a true value, but does not guarantee the value of any other bits,
7727 but we do not know of any machine that has such an instruction. If you
7728 are trying to port GCC to such a machine, include an instruction to
7729 perform a logical-and of the result with 1 in the pattern for the
7730 comparison operators and let us know at @email{gcc@@gcc.gnu.org}.
7732 Often, a machine will have multiple instructions that obtain a value
7733 from a comparison (or the condition codes). Here are rules to guide the
7734 choice of value for @code{STORE_FLAG_VALUE}, and hence the instructions
7739 Use the shortest sequence that yields a valid definition for
7740 @code{STORE_FLAG_VALUE}. It is more efficient for the compiler to
7741 ``normalize'' the value (convert it to, e.g., 1 or 0) than for the
7742 comparison operators to do so because there may be opportunities to
7743 combine the normalization with other operations.
7746 For equal-length sequences, use a value of 1 or @minus{}1, with @minus{}1 being
7747 slightly preferred on machines with expensive jumps and 1 preferred on
7751 As a second choice, choose a value of @samp{0x80000001} if instructions
7752 exist that set both the sign and low-order bits but do not define the
7756 Otherwise, use a value of @samp{0x80000000}.
7759 Many machines can produce both the value chosen for
7760 @code{STORE_FLAG_VALUE} and its negation in the same number of
7761 instructions. On those machines, you should also define a pattern for
7762 those cases, e.g., one matching
7765 (set @var{A} (neg:@var{m} (ne:@var{m} @var{B} @var{C})))
7768 Some machines can also perform @code{and} or @code{plus} operations on
7769 condition code values with less instructions than the corresponding
7770 @samp{cstore@var{mode}4} insn followed by @code{and} or @code{plus}. On those
7771 machines, define the appropriate patterns. Use the names @code{incscc}
7772 and @code{decscc}, respectively, for the patterns which perform
7773 @code{plus} or @code{minus} operations on condition code values. See
7774 @file{rs6000.md} for some examples. The GNU Superoptimizer can be used to
7775 find such instruction sequences on other machines.
7777 If this macro is not defined, the default value, 1, is used. You need
7778 not define @code{STORE_FLAG_VALUE} if the machine has no store-flag
7779 instructions, or if the value generated by these instructions is 1.
7782 @defmac FLOAT_STORE_FLAG_VALUE (@var{mode})
7783 A C expression that gives a nonzero @code{REAL_VALUE_TYPE} value that is
7784 returned when comparison operators with floating-point results are true.
7785 Define this macro on machines that have comparison operations that return
7786 floating-point values. If there are no such operations, do not define
7790 @defmac VECTOR_STORE_FLAG_VALUE (@var{mode})
7791 A C expression that gives a rtx representing the nonzero true element
7792 for vector comparisons. The returned rtx should be valid for the inner
7793 mode of @var{mode} which is guaranteed to be a vector mode. Define
7794 this macro on machines that have vector comparison operations that
7795 return a vector result. If there are no such operations, do not define
7796 this macro. Typically, this macro is defined as @code{const1_rtx} or
7797 @code{constm1_rtx}. This macro may return @code{NULL_RTX} to prevent
7798 the compiler optimizing such vector comparison operations for the
7802 @defmac CLZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
7803 @defmacx CTZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
7804 A C expression that indicates whether the architecture defines a value
7805 for @code{clz} or @code{ctz} with a zero operand.
7806 A result of @code{0} indicates the value is undefined.
7807 If the value is defined for only the RTL expression, the macro should
7808 evaluate to @code{1}; if the value applies also to the corresponding optab
7809 entry (which is normally the case if it expands directly into
7810 the corresponding RTL), then the macro should evaluate to @code{2}.
7811 In the cases where the value is defined, @var{value} should be set to
7814 If this macro is not defined, the value of @code{clz} or
7815 @code{ctz} at zero is assumed to be undefined.
7817 This macro must be defined if the target's expansion for @code{ffs}
7818 relies on a particular value to get correct results. Otherwise it
7819 is not necessary, though it may be used to optimize some corner cases, and
7820 to provide a default expansion for the @code{ffs} optab.
7822 Note that regardless of this macro the ``definedness'' of @code{clz}
7823 and @code{ctz} at zero do @emph{not} extend to the builtin functions
7824 visible to the user. Thus one may be free to adjust the value at will
7825 to match the target expansion of these operations without fear of
7830 An alias for the machine mode for pointers. On most machines, define
7831 this to be the integer mode corresponding to the width of a hardware
7832 pointer; @code{SImode} on 32-bit machine or @code{DImode} on 64-bit machines.
7833 On some machines you must define this to be one of the partial integer
7834 modes, such as @code{PSImode}.
7836 The width of @code{Pmode} must be at least as large as the value of
7837 @code{POINTER_SIZE}. If it is not equal, you must define the macro
7838 @code{POINTERS_EXTEND_UNSIGNED} to specify how pointers are extended
7842 @defmac FUNCTION_MODE
7843 An alias for the machine mode used for memory references to functions
7844 being called, in @code{call} RTL expressions. On most CISC machines,
7845 where an instruction can begin at any byte address, this should be
7846 @code{QImode}. On most RISC machines, where all instructions have fixed
7847 size and alignment, this should be a mode with the same size and alignment
7848 as the machine instruction words - typically @code{SImode} or @code{HImode}.
7851 @defmac STDC_0_IN_SYSTEM_HEADERS
7852 In normal operation, the preprocessor expands @code{__STDC__} to the
7853 constant 1, to signify that GCC conforms to ISO Standard C@. On some
7854 hosts, like Solaris, the system compiler uses a different convention,
7855 where @code{__STDC__} is normally 0, but is 1 if the user specifies
7856 strict conformance to the C Standard.
7858 Defining @code{STDC_0_IN_SYSTEM_HEADERS} makes GNU CPP follows the host
7859 convention when processing system header files, but when processing user
7860 files @code{__STDC__} will always expand to 1.
7863 @hook TARGET_C_PREINCLUDE
7865 @hook TARGET_CXX_IMPLICIT_EXTERN_C
7867 @defmac NO_IMPLICIT_EXTERN_C
7868 Define this macro if the system header files support C++ as well as C@.
7869 This macro inhibits the usual method of using system header files in
7870 C++, which is to pretend that the file's contents are enclosed in
7871 @samp{extern "C" @{@dots{}@}}.
7876 @defmac REGISTER_TARGET_PRAGMAS ()
7877 Define this macro if you want to implement any target-specific pragmas.
7878 If defined, it is a C expression which makes a series of calls to
7879 @code{c_register_pragma} or @code{c_register_pragma_with_expansion}
7880 for each pragma. The macro may also do any
7881 setup required for the pragmas.
7883 The primary reason to define this macro is to provide compatibility with
7884 other compilers for the same target. In general, we discourage
7885 definition of target-specific pragmas for GCC@.
7887 If the pragma can be implemented by attributes then you should consider
7888 defining the target hook @samp{TARGET_INSERT_ATTRIBUTES} as well.
7890 Preprocessor macros that appear on pragma lines are not expanded. All
7891 @samp{#pragma} directives that do not match any registered pragma are
7892 silently ignored, unless the user specifies @option{-Wunknown-pragmas}.
7895 @deftypefun void c_register_pragma (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
7896 @deftypefunx void c_register_pragma_with_expansion (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
7898 Each call to @code{c_register_pragma} or
7899 @code{c_register_pragma_with_expansion} establishes one pragma. The
7900 @var{callback} routine will be called when the preprocessor encounters a
7904 #pragma [@var{space}] @var{name} @dots{}
7907 @var{space} is the case-sensitive namespace of the pragma, or
7908 @code{NULL} to put the pragma in the global namespace. The callback
7909 routine receives @var{pfile} as its first argument, which can be passed
7910 on to cpplib's functions if necessary. You can lex tokens after the
7911 @var{name} by calling @code{pragma_lex}. Tokens that are not read by the
7912 callback will be silently ignored. The end of the line is indicated by
7913 a token of type @code{CPP_EOF}. Macro expansion occurs on the
7914 arguments of pragmas registered with
7915 @code{c_register_pragma_with_expansion} but not on the arguments of
7916 pragmas registered with @code{c_register_pragma}.
7918 Note that the use of @code{pragma_lex} is specific to the C and C++
7919 compilers. It will not work in the Java or Fortran compilers, or any
7920 other language compilers for that matter. Thus if @code{pragma_lex} is going
7921 to be called from target-specific code, it must only be done so when
7922 building the C and C++ compilers. This can be done by defining the
7923 variables @code{c_target_objs} and @code{cxx_target_objs} in the
7924 target entry in the @file{config.gcc} file. These variables should name
7925 the target-specific, language-specific object file which contains the
7926 code that uses @code{pragma_lex}. Note it will also be necessary to add a
7927 rule to the makefile fragment pointed to by @code{tmake_file} that shows
7928 how to build this object file.
7931 @defmac HANDLE_PRAGMA_PACK_WITH_EXPANSION
7932 Define this macro if macros should be expanded in the
7933 arguments of @samp{#pragma pack}.
7936 @defmac TARGET_DEFAULT_PACK_STRUCT
7937 If your target requires a structure packing default other than 0 (meaning
7938 the machine default), define this macro to the necessary value (in bytes).
7939 This must be a value that would also be valid to use with
7940 @samp{#pragma pack()} (that is, a small power of two).
7943 @defmac DOLLARS_IN_IDENTIFIERS
7944 Define this macro to control use of the character @samp{$} in
7945 identifier names for the C family of languages. 0 means @samp{$} is
7946 not allowed by default; 1 means it is allowed. 1 is the default;
7947 there is no need to define this macro in that case.
7950 @defmac INSN_SETS_ARE_DELAYED (@var{insn})
7951 Define this macro as a C expression that is nonzero if it is safe for the
7952 delay slot scheduler to place instructions in the delay slot of @var{insn},
7953 even if they appear to use a resource set or clobbered in @var{insn}.
7954 @var{insn} is always a @code{jump_insn} or an @code{insn}; GCC knows that
7955 every @code{call_insn} has this behavior. On machines where some @code{insn}
7956 or @code{jump_insn} is really a function call and hence has this behavior,
7957 you should define this macro.
7959 You need not define this macro if it would always return zero.
7962 @defmac INSN_REFERENCES_ARE_DELAYED (@var{insn})
7963 Define this macro as a C expression that is nonzero if it is safe for the
7964 delay slot scheduler to place instructions in the delay slot of @var{insn},
7965 even if they appear to set or clobber a resource referenced in @var{insn}.
7966 @var{insn} is always a @code{jump_insn} or an @code{insn}. On machines where
7967 some @code{insn} or @code{jump_insn} is really a function call and its operands
7968 are registers whose use is actually in the subroutine it calls, you should
7969 define this macro. Doing so allows the delay slot scheduler to move
7970 instructions which copy arguments into the argument registers into the delay
7973 You need not define this macro if it would always return zero.
7976 @defmac MULTIPLE_SYMBOL_SPACES
7977 Define this macro as a C expression that is nonzero if, in some cases,
7978 global symbols from one translation unit may not be bound to undefined
7979 symbols in another translation unit without user intervention. For
7980 instance, under Microsoft Windows symbols must be explicitly imported
7981 from shared libraries (DLLs).
7983 You need not define this macro if it would always evaluate to zero.
7986 @hook TARGET_MD_ASM_ADJUST
7988 @defmac MATH_LIBRARY
7989 Define this macro as a C string constant for the linker argument to link
7990 in the system math library, minus the initial @samp{"-l"}, or
7991 @samp{""} if the target does not have a
7992 separate math library.
7994 You need only define this macro if the default of @samp{"m"} is wrong.
7997 @defmac LIBRARY_PATH_ENV
7998 Define this macro as a C string constant for the environment variable that
7999 specifies where the linker should look for libraries.
8001 You need only define this macro if the default of @samp{"LIBRARY_PATH"}
8005 @defmac TARGET_POSIX_IO
8006 Define this macro if the target supports the following POSIX@ file
8007 functions, access, mkdir and file locking with fcntl / F_SETLKW@.
8008 Defining @code{TARGET_POSIX_IO} will enable the test coverage code
8009 to use file locking when exiting a program, which avoids race conditions
8010 if the program has forked. It will also create directories at run-time
8011 for cross-profiling.
8014 @defmac MAX_CONDITIONAL_EXECUTE
8016 A C expression for the maximum number of instructions to execute via
8017 conditional execution instructions instead of a branch. A value of
8018 @code{BRANCH_COST}+1 is the default if the machine does not use cc0, and
8019 1 if it does use cc0.
8022 @defmac IFCVT_MODIFY_TESTS (@var{ce_info}, @var{true_expr}, @var{false_expr})
8023 Used if the target needs to perform machine-dependent modifications on the
8024 conditionals used for turning basic blocks into conditionally executed code.
8025 @var{ce_info} points to a data structure, @code{struct ce_if_block}, which
8026 contains information about the currently processed blocks. @var{true_expr}
8027 and @var{false_expr} are the tests that are used for converting the
8028 then-block and the else-block, respectively. Set either @var{true_expr} or
8029 @var{false_expr} to a null pointer if the tests cannot be converted.
8032 @defmac IFCVT_MODIFY_MULTIPLE_TESTS (@var{ce_info}, @var{bb}, @var{true_expr}, @var{false_expr})
8033 Like @code{IFCVT_MODIFY_TESTS}, but used when converting more complicated
8034 if-statements into conditions combined by @code{and} and @code{or} operations.
8035 @var{bb} contains the basic block that contains the test that is currently
8036 being processed and about to be turned into a condition.
8039 @defmac IFCVT_MODIFY_INSN (@var{ce_info}, @var{pattern}, @var{insn})
8040 A C expression to modify the @var{PATTERN} of an @var{INSN} that is to
8041 be converted to conditional execution format. @var{ce_info} points to
8042 a data structure, @code{struct ce_if_block}, which contains information
8043 about the currently processed blocks.
8046 @defmac IFCVT_MODIFY_FINAL (@var{ce_info})
8047 A C expression to perform any final machine dependent modifications in
8048 converting code to conditional execution. The involved basic blocks
8049 can be found in the @code{struct ce_if_block} structure that is pointed
8050 to by @var{ce_info}.
8053 @defmac IFCVT_MODIFY_CANCEL (@var{ce_info})
8054 A C expression to cancel any machine dependent modifications in
8055 converting code to conditional execution. The involved basic blocks
8056 can be found in the @code{struct ce_if_block} structure that is pointed
8057 to by @var{ce_info}.
8060 @defmac IFCVT_MACHDEP_INIT (@var{ce_info})
8061 A C expression to initialize any machine specific data for if-conversion
8062 of the if-block in the @code{struct ce_if_block} structure that is pointed
8063 to by @var{ce_info}.
8066 @hook TARGET_MACHINE_DEPENDENT_REORG
8068 @hook TARGET_INIT_BUILTINS
8070 @hook TARGET_BUILTIN_DECL
8072 @hook TARGET_EXPAND_BUILTIN
8074 @hook TARGET_BUILTIN_CHKP_FUNCTION
8075 @hook TARGET_CHKP_BOUND_TYPE
8076 @hook TARGET_CHKP_BOUND_MODE
8077 @hook TARGET_CHKP_MAKE_BOUNDS_CONSTANT
8078 @hook TARGET_CHKP_INITIALIZE_BOUNDS
8080 @hook TARGET_RESOLVE_OVERLOADED_BUILTIN
8082 @hook TARGET_FOLD_BUILTIN
8084 @hook TARGET_GIMPLE_FOLD_BUILTIN
8086 @hook TARGET_COMPARE_VERSION_PRIORITY
8088 @hook TARGET_GET_FUNCTION_VERSIONS_DISPATCHER
8090 @hook TARGET_GENERATE_VERSION_DISPATCHER_BODY
8092 @hook TARGET_CAN_USE_DOLOOP_P
8094 @hook TARGET_INVALID_WITHIN_DOLOOP
8096 @hook TARGET_LEGITIMATE_COMBINED_INSN
8098 @hook TARGET_CAN_FOLLOW_JUMP
8100 @hook TARGET_COMMUTATIVE_P
8102 @hook TARGET_ALLOCATE_INITIAL_VALUE
8104 @hook TARGET_UNSPEC_MAY_TRAP_P
8106 @hook TARGET_SET_CURRENT_FUNCTION
8108 @defmac TARGET_OBJECT_SUFFIX
8109 Define this macro to be a C string representing the suffix for object
8110 files on your target machine. If you do not define this macro, GCC will
8111 use @samp{.o} as the suffix for object files.
8114 @defmac TARGET_EXECUTABLE_SUFFIX
8115 Define this macro to be a C string representing the suffix to be
8116 automatically added to executable files on your target machine. If you
8117 do not define this macro, GCC will use the null string as the suffix for
8121 @defmac COLLECT_EXPORT_LIST
8122 If defined, @code{collect2} will scan the individual object files
8123 specified on its command line and create an export list for the linker.
8124 Define this macro for systems like AIX, where the linker discards
8125 object files that are not referenced from @code{main} and uses export
8129 @defmac MODIFY_JNI_METHOD_CALL (@var{mdecl})
8130 Define this macro to a C expression representing a variant of the
8131 method call @var{mdecl}, if Java Native Interface (JNI) methods
8132 must be invoked differently from other methods on your target.
8133 For example, on 32-bit Microsoft Windows, JNI methods must be invoked using
8134 the @code{stdcall} calling convention and this macro is then
8135 defined as this expression:
8138 build_type_attribute_variant (@var{mdecl},
8140 (get_identifier ("stdcall"),
8145 @hook TARGET_CANNOT_MODIFY_JUMPS_P
8147 @hook TARGET_BRANCH_TARGET_REGISTER_CLASS
8149 @hook TARGET_BRANCH_TARGET_REGISTER_CALLEE_SAVED
8151 @hook TARGET_HAVE_CONDITIONAL_EXECUTION
8153 @hook TARGET_GEN_CCMP_FIRST
8155 @hook TARGET_GEN_CCMP_NEXT
8157 @hook TARGET_LOOP_UNROLL_ADJUST
8159 @defmac POWI_MAX_MULTS
8160 If defined, this macro is interpreted as a signed integer C expression
8161 that specifies the maximum number of floating point multiplications
8162 that should be emitted when expanding exponentiation by an integer
8163 constant inline. When this value is defined, exponentiation requiring
8164 more than this number of multiplications is implemented by calling the
8165 system library's @code{pow}, @code{powf} or @code{powl} routines.
8166 The default value places no upper bound on the multiplication count.
8169 @deftypefn Macro void TARGET_EXTRA_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
8170 This target hook should register any extra include files for the
8171 target. The parameter @var{stdinc} indicates if normal include files
8172 are present. The parameter @var{sysroot} is the system root directory.
8173 The parameter @var{iprefix} is the prefix for the gcc directory.
8176 @deftypefn Macro void TARGET_EXTRA_PRE_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
8177 This target hook should register any extra include files for the
8178 target before any standard headers. The parameter @var{stdinc}
8179 indicates if normal include files are present. The parameter
8180 @var{sysroot} is the system root directory. The parameter
8181 @var{iprefix} is the prefix for the gcc directory.
8184 @deftypefn Macro void TARGET_OPTF (char *@var{path})
8185 This target hook should register special include paths for the target.
8186 The parameter @var{path} is the include to register. On Darwin
8187 systems, this is used for Framework includes, which have semantics
8188 that are different from @option{-I}.
8191 @defmac bool TARGET_USE_LOCAL_THUNK_ALIAS_P (tree @var{fndecl})
8192 This target macro returns @code{true} if it is safe to use a local alias
8193 for a virtual function @var{fndecl} when constructing thunks,
8194 @code{false} otherwise. By default, the macro returns @code{true} for all
8195 functions, if a target supports aliases (i.e.@: defines
8196 @code{ASM_OUTPUT_DEF}), @code{false} otherwise,
8199 @defmac TARGET_FORMAT_TYPES
8200 If defined, this macro is the name of a global variable containing
8201 target-specific format checking information for the @option{-Wformat}
8202 option. The default is to have no target-specific format checks.
8205 @defmac TARGET_N_FORMAT_TYPES
8206 If defined, this macro is the number of entries in
8207 @code{TARGET_FORMAT_TYPES}.
8210 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES
8211 If defined, this macro is the name of a global variable containing
8212 target-specific format overrides for the @option{-Wformat} option. The
8213 default is to have no target-specific format overrides. If defined,
8214 @code{TARGET_FORMAT_TYPES} must be defined, too.
8217 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES_COUNT
8218 If defined, this macro specifies the number of entries in
8219 @code{TARGET_OVERRIDES_FORMAT_ATTRIBUTES}.
8222 @defmac TARGET_OVERRIDES_FORMAT_INIT
8223 If defined, this macro specifies the optional initialization
8224 routine for target specific customizations of the system printf
8225 and scanf formatter settings.
8228 @hook TARGET_INVALID_ARG_FOR_UNPROTOTYPED_FN
8230 @hook TARGET_INVALID_CONVERSION
8232 @hook TARGET_INVALID_UNARY_OP
8234 @hook TARGET_INVALID_BINARY_OP
8236 @hook TARGET_PROMOTED_TYPE
8238 @hook TARGET_CONVERT_TO_TYPE
8241 This macro determines the size of the objective C jump buffer for the
8242 NeXT runtime. By default, OBJC_JBLEN is defined to an innocuous value.
8245 @defmac LIBGCC2_UNWIND_ATTRIBUTE
8246 Define this macro if any target-specific attributes need to be attached
8247 to the functions in @file{libgcc} that provide low-level support for
8248 call stack unwinding. It is used in declarations in @file{unwind-generic.h}
8249 and the associated definitions of those functions.
8252 @hook TARGET_UPDATE_STACK_BOUNDARY
8254 @hook TARGET_GET_DRAP_RTX
8256 @hook TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS
8258 @hook TARGET_CONST_ANCHOR
8260 @hook TARGET_ASAN_SHADOW_OFFSET
8262 @hook TARGET_MEMMODEL_CHECK
8264 @hook TARGET_ATOMIC_TEST_AND_SET_TRUEVAL
8266 @hook TARGET_HAS_IFUNC_P
8268 @hook TARGET_ATOMIC_ALIGN_FOR_MODE
8270 @hook TARGET_ATOMIC_ASSIGN_EXPAND_FENV
8272 @hook TARGET_RECORD_OFFLOAD_SYMBOL
8274 @hook TARGET_OFFLOAD_OPTIONS
8276 @defmac TARGET_SUPPORTS_WIDE_INT
8278 On older ports, large integers are stored in @code{CONST_DOUBLE} rtl
8279 objects. Newer ports define @code{TARGET_SUPPORTS_WIDE_INT} to be nonzero
8280 to indicate that large integers are stored in
8281 @code{CONST_WIDE_INT} rtl objects. The @code{CONST_WIDE_INT} allows
8282 very large integer constants to be represented. @code{CONST_DOUBLE}
8283 is limited to twice the size of the host's @code{HOST_WIDE_INT}
8286 Converting a port mostly requires looking for the places where
8287 @code{CONST_DOUBLE}s are used with @code{VOIDmode} and replacing that
8288 code with code that accesses @code{CONST_WIDE_INT}s. @samp{"grep -i
8289 const_double"} at the port level gets you to 95% of the changes that
8290 need to be made. There are a few places that require a deeper look.
8294 There is no equivalent to @code{hval} and @code{lval} for
8295 @code{CONST_WIDE_INT}s. This would be difficult to express in the md
8296 language since there are a variable number of elements.
8298 Most ports only check that @code{hval} is either 0 or -1 to see if the
8299 value is small. As mentioned above, this will no longer be necessary
8300 since small constants are always @code{CONST_INT}. Of course there
8301 are still a few exceptions, the alpha's constraint used by the zap
8302 instruction certainly requires careful examination by C code.
8303 However, all the current code does is pass the hval and lval to C
8304 code, so evolving the c code to look at the @code{CONST_WIDE_INT} is
8305 not really a large change.
8308 Because there is no standard template that ports use to materialize
8309 constants, there is likely to be some futzing that is unique to each
8313 The rtx costs may have to be adjusted to properly account for larger
8314 constants that are represented as @code{CONST_WIDE_INT}.
8317 All and all it does not take long to convert ports that the
8318 maintainer is familiar with.
8322 @hook TARGET_RUN_TARGET_SELFTESTS