1 @c Copyright (C) 1988-2015 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 LINK_COMMAND_SPEC
379 A C string constant giving the complete command line need to execute the
380 linker. When you do this, you will need to update your port each time a
381 change is made to the link command line within @file{gcc.c}. Therefore,
382 define this macro only if you need to completely redefine the command
383 line for invoking the linker and there is no other way to accomplish
384 the effect you need. Overriding this macro may be avoidable by overriding
385 @code{LINK_GCC_C_SEQUENCE_SPEC} instead.
388 @hook TARGET_ALWAYS_STRIP_DOTDOT
390 @defmac MULTILIB_DEFAULTS
391 Define this macro as a C expression for the initializer of an array of
392 string to tell the driver program which options are defaults for this
393 target and thus do not need to be handled specially when using
394 @code{MULTILIB_OPTIONS}.
396 Do not define this macro if @code{MULTILIB_OPTIONS} is not defined in
397 the target makefile fragment or if none of the options listed in
398 @code{MULTILIB_OPTIONS} are set by default.
399 @xref{Target Fragment}.
402 @defmac RELATIVE_PREFIX_NOT_LINKDIR
403 Define this macro to tell @command{gcc} that it should only translate
404 a @option{-B} prefix into a @option{-L} linker option if the prefix
405 indicates an absolute file name.
408 @defmac MD_EXEC_PREFIX
409 If defined, this macro is an additional prefix to try after
410 @code{STANDARD_EXEC_PREFIX}. @code{MD_EXEC_PREFIX} is not searched
411 when the compiler is built as a cross
412 compiler. If you define @code{MD_EXEC_PREFIX}, then be sure to add it
413 to the list of directories used to find the assembler in @file{configure.in}.
416 @defmac STANDARD_STARTFILE_PREFIX
417 Define this macro as a C string constant if you wish to override the
418 standard choice of @code{libdir} as the default prefix to
419 try when searching for startup files such as @file{crt0.o}.
420 @code{STANDARD_STARTFILE_PREFIX} is not searched when the compiler
421 is built as a cross compiler.
424 @defmac STANDARD_STARTFILE_PREFIX_1
425 Define this macro as a C string constant if you wish to override the
426 standard choice of @code{/lib} as a prefix to try after the default prefix
427 when searching for startup files such as @file{crt0.o}.
428 @code{STANDARD_STARTFILE_PREFIX_1} is not searched when the compiler
429 is built as a cross compiler.
432 @defmac STANDARD_STARTFILE_PREFIX_2
433 Define this macro as a C string constant if you wish to override the
434 standard choice of @code{/lib} as yet another prefix to try after the
435 default prefix when searching for startup files such as @file{crt0.o}.
436 @code{STANDARD_STARTFILE_PREFIX_2} is not searched when the compiler
437 is built as a cross compiler.
440 @defmac MD_STARTFILE_PREFIX
441 If defined, this macro supplies an additional prefix to try after the
442 standard prefixes. @code{MD_EXEC_PREFIX} is not searched when the
443 compiler is built as a cross compiler.
446 @defmac MD_STARTFILE_PREFIX_1
447 If defined, this macro supplies yet another prefix to try after the
448 standard prefixes. It is not searched when the compiler is built as a
452 @defmac INIT_ENVIRONMENT
453 Define this macro as a C string constant if you wish to set environment
454 variables for programs called by the driver, such as the assembler and
455 loader. The driver passes the value of this macro to @code{putenv} to
456 initialize the necessary environment variables.
459 @defmac LOCAL_INCLUDE_DIR
460 Define this macro as a C string constant if you wish to override the
461 standard choice of @file{/usr/local/include} as the default prefix to
462 try when searching for local header files. @code{LOCAL_INCLUDE_DIR}
463 comes before @code{NATIVE_SYSTEM_HEADER_DIR} (set in
464 @file{config.gcc}, normally @file{/usr/include}) in the search order.
466 Cross compilers do not search either @file{/usr/local/include} or its
470 @defmac NATIVE_SYSTEM_HEADER_COMPONENT
471 The ``component'' corresponding to @code{NATIVE_SYSTEM_HEADER_DIR}.
472 See @code{INCLUDE_DEFAULTS}, below, for the description of components.
473 If you do not define this macro, no component is used.
476 @defmac INCLUDE_DEFAULTS
477 Define this macro if you wish to override the entire default search path
478 for include files. For a native compiler, the default search path
479 usually consists of @code{GCC_INCLUDE_DIR}, @code{LOCAL_INCLUDE_DIR},
480 @code{GPLUSPLUS_INCLUDE_DIR}, and
481 @code{NATIVE_SYSTEM_HEADER_DIR}. In addition, @code{GPLUSPLUS_INCLUDE_DIR}
482 and @code{GCC_INCLUDE_DIR} are defined automatically by @file{Makefile},
483 and specify private search areas for GCC@. The directory
484 @code{GPLUSPLUS_INCLUDE_DIR} is used only for C++ programs.
486 The definition should be an initializer for an array of structures.
487 Each array element should have four elements: the directory name (a
488 string constant), the component name (also a string constant), a flag
489 for C++-only directories,
490 and a flag showing that the includes in the directory don't need to be
491 wrapped in @code{extern @samp{C}} when compiling C++. Mark the end of
492 the array with a null element.
494 The component name denotes what GNU package the include file is part of,
495 if any, in all uppercase letters. For example, it might be @samp{GCC}
496 or @samp{BINUTILS}. If the package is part of a vendor-supplied
497 operating system, code the component name as @samp{0}.
499 For example, here is the definition used for VAX/VMS:
502 #define INCLUDE_DEFAULTS \
504 @{ "GNU_GXX_INCLUDE:", "G++", 1, 1@}, \
505 @{ "GNU_CC_INCLUDE:", "GCC", 0, 0@}, \
506 @{ "SYS$SYSROOT:[SYSLIB.]", 0, 0, 0@}, \
513 Here is the order of prefixes tried for exec files:
517 Any prefixes specified by the user with @option{-B}.
520 The environment variable @code{GCC_EXEC_PREFIX} or, if @code{GCC_EXEC_PREFIX}
521 is not set and the compiler has not been installed in the configure-time
522 @var{prefix}, the location in which the compiler has actually been installed.
525 The directories specified by the environment variable @code{COMPILER_PATH}.
528 The macro @code{STANDARD_EXEC_PREFIX}, if the compiler has been installed
529 in the configured-time @var{prefix}.
532 The location @file{/usr/libexec/gcc/}, but only if this is a native compiler.
535 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
538 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
542 Here is the order of prefixes tried for startfiles:
546 Any prefixes specified by the user with @option{-B}.
549 The environment variable @code{GCC_EXEC_PREFIX} or its automatically determined
550 value based on the installed toolchain location.
553 The directories specified by the environment variable @code{LIBRARY_PATH}
554 (or port-specific name; native only, cross compilers do not use this).
557 The macro @code{STANDARD_EXEC_PREFIX}, but only if the toolchain is installed
558 in the configured @var{prefix} or this is a native compiler.
561 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
564 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
568 The macro @code{MD_STARTFILE_PREFIX}, if defined, but only if this is a
569 native compiler, or we have a target system root.
572 The macro @code{MD_STARTFILE_PREFIX_1}, if defined, but only if this is a
573 native compiler, or we have a target system root.
576 The macro @code{STANDARD_STARTFILE_PREFIX}, with any sysroot modifications.
577 If this path is relative it will be prefixed by @code{GCC_EXEC_PREFIX} and
578 the machine suffix or @code{STANDARD_EXEC_PREFIX} and the machine suffix.
581 The macro @code{STANDARD_STARTFILE_PREFIX_1}, but only if this is a native
582 compiler, or we have a target system root. The default for this macro is
586 The macro @code{STANDARD_STARTFILE_PREFIX_2}, but only if this is a native
587 compiler, or we have a target system root. The default for this macro is
591 @node Run-time Target
592 @section Run-time Target Specification
593 @cindex run-time target specification
594 @cindex predefined macros
595 @cindex target specifications
597 @c prevent bad page break with this line
598 Here are run-time target specifications.
600 @defmac TARGET_CPU_CPP_BUILTINS ()
601 This function-like macro expands to a block of code that defines
602 built-in preprocessor macros and assertions for the target CPU, using
603 the functions @code{builtin_define}, @code{builtin_define_std} and
604 @code{builtin_assert}. When the front end
605 calls this macro it provides a trailing semicolon, and since it has
606 finished command line option processing your code can use those
609 @code{builtin_assert} takes a string in the form you pass to the
610 command-line option @option{-A}, such as @code{cpu=mips}, and creates
611 the assertion. @code{builtin_define} takes a string in the form
612 accepted by option @option{-D} and unconditionally defines the macro.
614 @code{builtin_define_std} takes a string representing the name of an
615 object-like macro. If it doesn't lie in the user's namespace,
616 @code{builtin_define_std} defines it unconditionally. Otherwise, it
617 defines a version with two leading underscores, and another version
618 with two leading and trailing underscores, and defines the original
619 only if an ISO standard was not requested on the command line. For
620 example, passing @code{unix} defines @code{__unix}, @code{__unix__}
621 and possibly @code{unix}; passing @code{_mips} defines @code{__mips},
622 @code{__mips__} and possibly @code{_mips}, and passing @code{_ABI64}
623 defines only @code{_ABI64}.
625 You can also test for the C dialect being compiled. The variable
626 @code{c_language} is set to one of @code{clk_c}, @code{clk_cplusplus}
627 or @code{clk_objective_c}. Note that if we are preprocessing
628 assembler, this variable will be @code{clk_c} but the function-like
629 macro @code{preprocessing_asm_p()} will return true, so you might want
630 to check for that first. If you need to check for strict ANSI, the
631 variable @code{flag_iso} can be used. The function-like macro
632 @code{preprocessing_trad_p()} can be used to check for traditional
636 @defmac TARGET_OS_CPP_BUILTINS ()
637 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
638 and is used for the target operating system instead.
641 @defmac TARGET_OBJFMT_CPP_BUILTINS ()
642 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
643 and is used for the target object format. @file{elfos.h} uses this
644 macro to define @code{__ELF__}, so you probably do not need to define
648 @deftypevar {extern int} target_flags
649 This variable is declared in @file{options.h}, which is included before
650 any target-specific headers.
653 @hook TARGET_DEFAULT_TARGET_FLAGS
654 This variable specifies the initial value of @code{target_flags}.
655 Its default setting is 0.
658 @cindex optional hardware or system features
659 @cindex features, optional, in system conventions
661 @hook TARGET_HANDLE_OPTION
662 This hook is called whenever the user specifies one of the
663 target-specific options described by the @file{.opt} definition files
664 (@pxref{Options}). It has the opportunity to do some option-specific
665 processing and should return true if the option is valid. The default
666 definition does nothing but return true.
668 @var{decoded} specifies the option and its arguments. @var{opts} and
669 @var{opts_set} are the @code{gcc_options} structures to be used for
670 storing option state, and @var{loc} is the location at which the
671 option was passed (@code{UNKNOWN_LOCATION} except for options passed
675 @hook TARGET_HANDLE_C_OPTION
676 This target hook is called whenever the user specifies one of the
677 target-specific C language family options described by the @file{.opt}
678 definition files(@pxref{Options}). It has the opportunity to do some
679 option-specific processing and should return true if the option is
680 valid. The arguments are like for @code{TARGET_HANDLE_OPTION}. The
681 default definition does nothing but return false.
683 In general, you should use @code{TARGET_HANDLE_OPTION} to handle
684 options. However, if processing an option requires routines that are
685 only available in the C (and related language) front ends, then you
686 should use @code{TARGET_HANDLE_C_OPTION} instead.
689 @hook TARGET_OBJC_CONSTRUCT_STRING_OBJECT
691 @hook TARGET_OBJC_DECLARE_UNRESOLVED_CLASS_REFERENCE
693 @hook TARGET_OBJC_DECLARE_CLASS_DEFINITION
695 @hook TARGET_STRING_OBJECT_REF_TYPE_P
697 @hook TARGET_CHECK_STRING_OBJECT_FORMAT_ARG
699 @hook TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE
701 @defmac C_COMMON_OVERRIDE_OPTIONS
702 This is similar to the @code{TARGET_OPTION_OVERRIDE} hook
703 but is only used in the C
704 language frontends (C, Objective-C, C++, Objective-C++) and so can be
705 used to alter option flag variables which only exist in those
709 @hook TARGET_OPTION_OPTIMIZATION_TABLE
710 Some machines may desire to change what optimizations are performed for
711 various optimization levels. This variable, if defined, describes
712 options to enable at particular sets of optimization levels. These
713 options are processed once
714 just after the optimization level is determined and before the remainder
715 of the command options have been parsed, so may be overridden by other
716 options passed explicitly.
718 This processing is run once at program startup and when the optimization
719 options are changed via @code{#pragma GCC optimize} or by using the
720 @code{optimize} attribute.
723 @hook TARGET_OPTION_INIT_STRUCT
725 @hook TARGET_OPTION_DEFAULT_PARAMS
727 @defmac SWITCHABLE_TARGET
728 Some targets need to switch between substantially different subtargets
729 during compilation. For example, the MIPS target has one subtarget for
730 the traditional MIPS architecture and another for MIPS16. Source code
731 can switch between these two subarchitectures using the @code{mips16}
732 and @code{nomips16} attributes.
734 Such subtargets can differ in things like the set of available
735 registers, the set of available instructions, the costs of various
736 operations, and so on. GCC caches a lot of this type of information
737 in global variables, and recomputing them for each subtarget takes a
738 significant amount of time. The compiler therefore provides a facility
739 for maintaining several versions of the global variables and quickly
740 switching between them; see @file{target-globals.h} for details.
742 Define this macro to 1 if your target needs this facility. The default
746 @hook TARGET_FLOAT_EXCEPTIONS_ROUNDING_SUPPORTED_P
748 @node Per-Function Data
749 @section Defining data structures for per-function information.
750 @cindex per-function data
751 @cindex data structures
753 If the target needs to store information on a per-function basis, GCC
754 provides a macro and a couple of variables to allow this. Note, just
755 using statics to store the information is a bad idea, since GCC supports
756 nested functions, so you can be halfway through encoding one function
757 when another one comes along.
759 GCC defines a data structure called @code{struct function} which
760 contains all of the data specific to an individual function. This
761 structure contains a field called @code{machine} whose type is
762 @code{struct machine_function *}, which can be used by targets to point
763 to their own specific data.
765 If a target needs per-function specific data it should define the type
766 @code{struct machine_function} and also the macro @code{INIT_EXPANDERS}.
767 This macro should be used to initialize the function pointer
768 @code{init_machine_status}. This pointer is explained below.
770 One typical use of per-function, target specific data is to create an
771 RTX to hold the register containing the function's return address. This
772 RTX can then be used to implement the @code{__builtin_return_address}
773 function, for level 0.
775 Note---earlier implementations of GCC used a single data area to hold
776 all of the per-function information. Thus when processing of a nested
777 function began the old per-function data had to be pushed onto a
778 stack, and when the processing was finished, it had to be popped off the
779 stack. GCC used to provide function pointers called
780 @code{save_machine_status} and @code{restore_machine_status} to handle
781 the saving and restoring of the target specific information. Since the
782 single data area approach is no longer used, these pointers are no
785 @defmac INIT_EXPANDERS
786 Macro called to initialize any target specific information. This macro
787 is called once per function, before generation of any RTL has begun.
788 The intention of this macro is to allow the initialization of the
789 function pointer @code{init_machine_status}.
792 @deftypevar {void (*)(struct function *)} init_machine_status
793 If this function pointer is non-@code{NULL} it will be called once per
794 function, before function compilation starts, in order to allow the
795 target to perform any target specific initialization of the
796 @code{struct function} structure. It is intended that this would be
797 used to initialize the @code{machine} of that structure.
799 @code{struct machine_function} structures are expected to be freed by GC@.
800 Generally, any memory that they reference must be allocated by using
801 GC allocation, including the structure itself.
805 @section Storage Layout
806 @cindex storage layout
808 Note that the definitions of the macros in this table which are sizes or
809 alignments measured in bits do not need to be constant. They can be C
810 expressions that refer to static variables, such as the @code{target_flags}.
811 @xref{Run-time Target}.
813 @defmac BITS_BIG_ENDIAN
814 Define this macro to have the value 1 if the most significant bit in a
815 byte has the lowest number; otherwise define it to have the value zero.
816 This means that bit-field instructions count from the most significant
817 bit. If the machine has no bit-field instructions, then this must still
818 be defined, but it doesn't matter which value it is defined to. This
819 macro need not be a constant.
821 This macro does not affect the way structure fields are packed into
822 bytes or words; that is controlled by @code{BYTES_BIG_ENDIAN}.
825 @defmac BYTES_BIG_ENDIAN
826 Define this macro to have the value 1 if the most significant byte in a
827 word has the lowest number. This macro need not be a constant.
830 @defmac WORDS_BIG_ENDIAN
831 Define this macro to have the value 1 if, in a multiword object, the
832 most significant word has the lowest number. This applies to both
833 memory locations and registers; see @code{REG_WORDS_BIG_ENDIAN} if the
834 order of words in memory is not the same as the order in registers. This
835 macro need not be a constant.
838 @defmac REG_WORDS_BIG_ENDIAN
839 On some machines, the order of words in a multiword object differs between
840 registers in memory. In such a situation, define this macro to describe
841 the order of words in a register. The macro @code{WORDS_BIG_ENDIAN} controls
842 the order of words in memory.
845 @defmac FLOAT_WORDS_BIG_ENDIAN
846 Define this macro to have the value 1 if @code{DFmode}, @code{XFmode} or
847 @code{TFmode} floating point numbers are stored in memory with the word
848 containing the sign bit at the lowest address; otherwise define it to
849 have the value 0. This macro need not be a constant.
851 You need not define this macro if the ordering is the same as for
855 @defmac BITS_PER_WORD
856 Number of bits in a word. If you do not define this macro, the default
857 is @code{BITS_PER_UNIT * UNITS_PER_WORD}.
860 @defmac MAX_BITS_PER_WORD
861 Maximum number of bits in a word. If this is undefined, the default is
862 @code{BITS_PER_WORD}. Otherwise, it is the constant value that is the
863 largest value that @code{BITS_PER_WORD} can have at run-time.
866 @defmac UNITS_PER_WORD
867 Number of storage units in a word; normally the size of a general-purpose
868 register, a power of two from 1 or 8.
871 @defmac MIN_UNITS_PER_WORD
872 Minimum number of units in a word. If this is undefined, the default is
873 @code{UNITS_PER_WORD}. Otherwise, it is the constant value that is the
874 smallest value that @code{UNITS_PER_WORD} can have at run-time.
878 Width of a pointer, in bits. You must specify a value no wider than the
879 width of @code{Pmode}. If it is not equal to the width of @code{Pmode},
880 you must define @code{POINTERS_EXTEND_UNSIGNED}. If you do not specify
881 a value the default is @code{BITS_PER_WORD}.
884 @defmac POINTERS_EXTEND_UNSIGNED
885 A C expression that determines how pointers should be extended from
886 @code{ptr_mode} to either @code{Pmode} or @code{word_mode}. It is
887 greater than zero if pointers should be zero-extended, zero if they
888 should be sign-extended, and negative if some other sort of conversion
889 is needed. In the last case, the extension is done by the target's
890 @code{ptr_extend} instruction.
892 You need not define this macro if the @code{ptr_mode}, @code{Pmode}
893 and @code{word_mode} are all the same width.
896 @defmac PROMOTE_MODE (@var{m}, @var{unsignedp}, @var{type})
897 A macro to update @var{m} and @var{unsignedp} when an object whose type
898 is @var{type} and which has the specified mode and signedness is to be
899 stored in a register. This macro is only called when @var{type} is a
902 On most RISC machines, which only have operations that operate on a full
903 register, define this macro to set @var{m} to @code{word_mode} if
904 @var{m} is an integer mode narrower than @code{BITS_PER_WORD}. In most
905 cases, only integer modes should be widened because wider-precision
906 floating-point operations are usually more expensive than their narrower
909 For most machines, the macro definition does not change @var{unsignedp}.
910 However, some machines, have instructions that preferentially handle
911 either signed or unsigned quantities of certain modes. For example, on
912 the DEC Alpha, 32-bit loads from memory and 32-bit add instructions
913 sign-extend the result to 64 bits. On such machines, set
914 @var{unsignedp} according to which kind of extension is more efficient.
916 Do not define this macro if it would never modify @var{m}.
919 @hook TARGET_PROMOTE_FUNCTION_MODE
921 @defmac PARM_BOUNDARY
922 Normal alignment required for function parameters on the stack, in
923 bits. All stack parameters receive at least this much alignment
924 regardless of data type. On most machines, this is the same as the
928 @defmac STACK_BOUNDARY
929 Define this macro to the minimum alignment enforced by hardware for the
930 stack pointer on this machine. The definition is a C expression for the
931 desired alignment (measured in bits). This value is used as a default
932 if @code{PREFERRED_STACK_BOUNDARY} is not defined. On most machines,
933 this should be the same as @code{PARM_BOUNDARY}.
936 @defmac PREFERRED_STACK_BOUNDARY
937 Define this macro if you wish to preserve a certain alignment for the
938 stack pointer, greater than what the hardware enforces. The definition
939 is a C expression for the desired alignment (measured in bits). This
940 macro must evaluate to a value equal to or larger than
941 @code{STACK_BOUNDARY}.
944 @defmac INCOMING_STACK_BOUNDARY
945 Define this macro if the incoming stack boundary may be different
946 from @code{PREFERRED_STACK_BOUNDARY}. This macro must evaluate
947 to a value equal to or larger than @code{STACK_BOUNDARY}.
950 @defmac FUNCTION_BOUNDARY
951 Alignment required for a function entry point, in bits.
954 @defmac BIGGEST_ALIGNMENT
955 Biggest alignment that any data type can require on this machine, in
956 bits. Note that this is not the biggest alignment that is supported,
957 just the biggest alignment that, when violated, may cause a fault.
960 @hook TARGET_ABSOLUTE_BIGGEST_ALIGNMENT
962 @defmac MALLOC_ABI_ALIGNMENT
963 Alignment, in bits, a C conformant malloc implementation has to
964 provide. If not defined, the default value is @code{BITS_PER_WORD}.
967 @defmac ATTRIBUTE_ALIGNED_VALUE
968 Alignment used by the @code{__attribute__ ((aligned))} construct. If
969 not defined, the default value is @code{BIGGEST_ALIGNMENT}.
972 @defmac MINIMUM_ATOMIC_ALIGNMENT
973 If defined, the smallest alignment, in bits, that can be given to an
974 object that can be referenced in one operation, without disturbing any
975 nearby object. Normally, this is @code{BITS_PER_UNIT}, but may be larger
976 on machines that don't have byte or half-word store operations.
979 @defmac BIGGEST_FIELD_ALIGNMENT
980 Biggest alignment that any structure or union field can require on this
981 machine, in bits. If defined, this overrides @code{BIGGEST_ALIGNMENT} for
982 structure and union fields only, unless the field alignment has been set
983 by the @code{__attribute__ ((aligned (@var{n})))} construct.
986 @defmac ADJUST_FIELD_ALIGN (@var{field}, @var{computed})
987 An expression for the alignment of a structure field @var{field} if the
988 alignment computed in the usual way (including applying of
989 @code{BIGGEST_ALIGNMENT} and @code{BIGGEST_FIELD_ALIGNMENT} to the
990 alignment) is @var{computed}. It overrides alignment only if the
991 field alignment has not been set by the
992 @code{__attribute__ ((aligned (@var{n})))} construct.
995 @defmac MAX_STACK_ALIGNMENT
996 Biggest stack alignment guaranteed by the backend. Use this macro
997 to specify the maximum alignment of a variable on stack.
999 If not defined, the default value is @code{STACK_BOUNDARY}.
1001 @c FIXME: The default should be @code{PREFERRED_STACK_BOUNDARY}.
1002 @c But the fix for PR 32893 indicates that we can only guarantee
1003 @c maximum stack alignment on stack up to @code{STACK_BOUNDARY}, not
1004 @c @code{PREFERRED_STACK_BOUNDARY}, if stack alignment isn't supported.
1007 @defmac MAX_OFILE_ALIGNMENT
1008 Biggest alignment supported by the object file format of this machine.
1009 Use this macro to limit the alignment which can be specified using the
1010 @code{__attribute__ ((aligned (@var{n})))} construct. If not defined,
1011 the default value is @code{BIGGEST_ALIGNMENT}.
1013 On systems that use ELF, the default (in @file{config/elfos.h}) is
1014 the largest supported 32-bit ELF section alignment representable on
1015 a 32-bit host e.g. @samp{(((uint64_t) 1 << 28) * 8)}.
1016 On 32-bit ELF the largest supported section alignment in bits is
1017 @samp{(0x80000000 * 8)}, but this is not representable on 32-bit hosts.
1020 @defmac DATA_ALIGNMENT (@var{type}, @var{basic-align})
1021 If defined, a C expression to compute the alignment for a variable in
1022 the static store. @var{type} is the data type, and @var{basic-align} is
1023 the alignment that the object would ordinarily have. The value of this
1024 macro is used instead of that alignment to align the object.
1026 If this macro is not defined, then @var{basic-align} is used.
1029 One use of this macro is to increase alignment of medium-size data to
1030 make it all fit in fewer cache lines. Another is to cause character
1031 arrays to be word-aligned so that @code{strcpy} calls that copy
1032 constants to character arrays can be done inline.
1035 @defmac DATA_ABI_ALIGNMENT (@var{type}, @var{basic-align})
1036 Similar to @code{DATA_ALIGNMENT}, but for the cases where the ABI mandates
1037 some alignment increase, instead of optimization only purposes. E.g.@
1038 AMD x86-64 psABI says that variables with array type larger than 15 bytes
1039 must be aligned to 16 byte boundaries.
1041 If this macro is not defined, then @var{basic-align} is used.
1044 @defmac CONSTANT_ALIGNMENT (@var{constant}, @var{basic-align})
1045 If defined, a C expression to compute the alignment given to a constant
1046 that is being placed in memory. @var{constant} is the constant and
1047 @var{basic-align} is the alignment that the object would ordinarily
1048 have. The value of this macro is used instead of that alignment to
1051 If this macro is not defined, then @var{basic-align} is used.
1053 The typical use of this macro is to increase alignment for string
1054 constants to be word aligned so that @code{strcpy} calls that copy
1055 constants can be done inline.
1058 @defmac LOCAL_ALIGNMENT (@var{type}, @var{basic-align})
1059 If defined, a C expression to compute the alignment for a variable in
1060 the local store. @var{type} is the data type, and @var{basic-align} is
1061 the alignment that the object would ordinarily have. The value of this
1062 macro is used instead of that alignment to align the object.
1064 If this macro is not defined, then @var{basic-align} is used.
1066 One use of this macro is to increase alignment of medium-size data to
1067 make it all fit in fewer cache lines.
1069 If the value of this macro has a type, it should be an unsigned type.
1072 @hook TARGET_VECTOR_ALIGNMENT
1074 @defmac STACK_SLOT_ALIGNMENT (@var{type}, @var{mode}, @var{basic-align})
1075 If defined, a C expression to compute the alignment for stack slot.
1076 @var{type} is the data type, @var{mode} is the widest mode available,
1077 and @var{basic-align} is the alignment that the slot would ordinarily
1078 have. The value of this macro is used instead of that alignment to
1081 If this macro is not defined, then @var{basic-align} is used when
1082 @var{type} is @code{NULL}. Otherwise, @code{LOCAL_ALIGNMENT} will
1085 This macro is to set alignment of stack slot to the maximum alignment
1086 of all possible modes which the slot may have.
1088 If the value of this macro has a type, it should be an unsigned type.
1091 @defmac LOCAL_DECL_ALIGNMENT (@var{decl})
1092 If defined, a C expression to compute the alignment for a local
1093 variable @var{decl}.
1095 If this macro is not defined, then
1096 @code{LOCAL_ALIGNMENT (TREE_TYPE (@var{decl}), DECL_ALIGN (@var{decl}))}
1099 One use of this macro is to increase alignment of medium-size data to
1100 make it all fit in fewer cache lines.
1102 If the value of this macro has a type, it should be an unsigned type.
1105 @defmac MINIMUM_ALIGNMENT (@var{exp}, @var{mode}, @var{align})
1106 If defined, a C expression to compute the minimum required alignment
1107 for dynamic stack realignment purposes for @var{exp} (a type or decl),
1108 @var{mode}, assuming normal alignment @var{align}.
1110 If this macro is not defined, then @var{align} will be used.
1113 @defmac EMPTY_FIELD_BOUNDARY
1114 Alignment in bits to be given to a structure bit-field that follows an
1115 empty field such as @code{int : 0;}.
1117 If @code{PCC_BITFIELD_TYPE_MATTERS} is true, it overrides this macro.
1120 @defmac STRUCTURE_SIZE_BOUNDARY
1121 Number of bits which any structure or union's size must be a multiple of.
1122 Each structure or union's size is rounded up to a multiple of this.
1124 If you do not define this macro, the default is the same as
1125 @code{BITS_PER_UNIT}.
1128 @defmac STRICT_ALIGNMENT
1129 Define this macro to be the value 1 if instructions will fail to work
1130 if given data not on the nominal alignment. If instructions will merely
1131 go slower in that case, define this macro as 0.
1134 @defmac PCC_BITFIELD_TYPE_MATTERS
1135 Define this if you wish to imitate the way many other C compilers handle
1136 alignment of bit-fields and the structures that contain them.
1138 The behavior is that the type written for a named bit-field (@code{int},
1139 @code{short}, or other integer type) imposes an alignment for the entire
1140 structure, as if the structure really did contain an ordinary field of
1141 that type. In addition, the bit-field is placed within the structure so
1142 that it would fit within such a field, not crossing a boundary for it.
1144 Thus, on most machines, a named bit-field whose type is written as
1145 @code{int} would not cross a four-byte boundary, and would force
1146 four-byte alignment for the whole structure. (The alignment used may
1147 not be four bytes; it is controlled by the other alignment parameters.)
1149 An unnamed bit-field will not affect the alignment of the containing
1152 If the macro is defined, its definition should be a C expression;
1153 a nonzero value for the expression enables this behavior.
1155 Note that if this macro is not defined, or its value is zero, some
1156 bit-fields may cross more than one alignment boundary. The compiler can
1157 support such references if there are @samp{insv}, @samp{extv}, and
1158 @samp{extzv} insns that can directly reference memory.
1160 The other known way of making bit-fields work is to define
1161 @code{STRUCTURE_SIZE_BOUNDARY} as large as @code{BIGGEST_ALIGNMENT}.
1162 Then every structure can be accessed with fullwords.
1164 Unless the machine has bit-field instructions or you define
1165 @code{STRUCTURE_SIZE_BOUNDARY} that way, you must define
1166 @code{PCC_BITFIELD_TYPE_MATTERS} to have a nonzero value.
1168 If your aim is to make GCC use the same conventions for laying out
1169 bit-fields as are used by another compiler, here is how to investigate
1170 what the other compiler does. Compile and run this program:
1189 printf ("Size of foo1 is %d\n",
1190 sizeof (struct foo1));
1191 printf ("Size of foo2 is %d\n",
1192 sizeof (struct foo2));
1197 If this prints 2 and 5, then the compiler's behavior is what you would
1198 get from @code{PCC_BITFIELD_TYPE_MATTERS}.
1201 @defmac BITFIELD_NBYTES_LIMITED
1202 Like @code{PCC_BITFIELD_TYPE_MATTERS} except that its effect is limited
1203 to aligning a bit-field within the structure.
1206 @hook TARGET_ALIGN_ANON_BITFIELD
1208 @hook TARGET_NARROW_VOLATILE_BITFIELD
1210 @hook TARGET_MEMBER_TYPE_FORCES_BLK
1212 @defmac ROUND_TYPE_ALIGN (@var{type}, @var{computed}, @var{specified})
1213 Define this macro as an expression for the alignment of a type (given
1214 by @var{type} as a tree node) if the alignment computed in the usual
1215 way is @var{computed} and the alignment explicitly specified was
1218 The default is to use @var{specified} if it is larger; otherwise, use
1219 the smaller of @var{computed} and @code{BIGGEST_ALIGNMENT}
1222 @defmac MAX_FIXED_MODE_SIZE
1223 An integer expression for the size in bits of the largest integer
1224 machine mode that should actually be used. All integer machine modes of
1225 this size or smaller can be used for structures and unions with the
1226 appropriate sizes. If this macro is undefined, @code{GET_MODE_BITSIZE
1227 (DImode)} is assumed.
1230 @defmac STACK_SAVEAREA_MODE (@var{save_level})
1231 If defined, an expression of type @code{machine_mode} that
1232 specifies the mode of the save area operand of a
1233 @code{save_stack_@var{level}} named pattern (@pxref{Standard Names}).
1234 @var{save_level} is one of @code{SAVE_BLOCK}, @code{SAVE_FUNCTION}, or
1235 @code{SAVE_NONLOCAL} and selects which of the three named patterns is
1236 having its mode specified.
1238 You need not define this macro if it always returns @code{Pmode}. You
1239 would most commonly define this macro if the
1240 @code{save_stack_@var{level}} patterns need to support both a 32- and a
1244 @defmac STACK_SIZE_MODE
1245 If defined, an expression of type @code{machine_mode} that
1246 specifies the mode of the size increment operand of an
1247 @code{allocate_stack} named pattern (@pxref{Standard Names}).
1249 You need not define this macro if it always returns @code{word_mode}.
1250 You would most commonly define this macro if the @code{allocate_stack}
1251 pattern needs to support both a 32- and a 64-bit mode.
1254 @hook TARGET_LIBGCC_CMP_RETURN_MODE
1256 @hook TARGET_LIBGCC_SHIFT_COUNT_MODE
1258 @hook TARGET_UNWIND_WORD_MODE
1260 @hook TARGET_MS_BITFIELD_LAYOUT_P
1262 @hook TARGET_DECIMAL_FLOAT_SUPPORTED_P
1264 @hook TARGET_FIXED_POINT_SUPPORTED_P
1266 @hook TARGET_EXPAND_TO_RTL_HOOK
1268 @hook TARGET_INSTANTIATE_DECLS
1270 @hook TARGET_MANGLE_TYPE
1273 @section Layout of Source Language Data Types
1275 These macros define the sizes and other characteristics of the standard
1276 basic data types used in programs being compiled. Unlike the macros in
1277 the previous section, these apply to specific features of C and related
1278 languages, rather than to fundamental aspects of storage layout.
1280 @defmac INT_TYPE_SIZE
1281 A C expression for the size in bits of the type @code{int} on the
1282 target machine. If you don't define this, the default is one word.
1285 @defmac SHORT_TYPE_SIZE
1286 A C expression for the size in bits of the type @code{short} on the
1287 target machine. If you don't define this, the default is half a word.
1288 (If this would be less than one storage unit, it is rounded up to one
1292 @defmac LONG_TYPE_SIZE
1293 A C expression for the size in bits of the type @code{long} on the
1294 target machine. If you don't define this, the default is one word.
1297 @defmac ADA_LONG_TYPE_SIZE
1298 On some machines, the size used for the Ada equivalent of the type
1299 @code{long} by a native Ada compiler differs from that used by C@. In
1300 that situation, define this macro to be a C expression to be used for
1301 the size of that type. If you don't define this, the default is the
1302 value of @code{LONG_TYPE_SIZE}.
1305 @defmac LONG_LONG_TYPE_SIZE
1306 A C expression for the size in bits of the type @code{long long} on the
1307 target machine. If you don't define this, the default is two
1308 words. If you want to support GNU Ada on your machine, the value of this
1309 macro must be at least 64.
1312 @defmac CHAR_TYPE_SIZE
1313 A C expression for the size in bits of the type @code{char} on the
1314 target machine. If you don't define this, the default is
1315 @code{BITS_PER_UNIT}.
1318 @defmac BOOL_TYPE_SIZE
1319 A C expression for the size in bits of the C++ type @code{bool} and
1320 C99 type @code{_Bool} on the target machine. If you don't define
1321 this, and you probably shouldn't, the default is @code{CHAR_TYPE_SIZE}.
1324 @defmac FLOAT_TYPE_SIZE
1325 A C expression for the size in bits of the type @code{float} on the
1326 target machine. If you don't define this, the default is one word.
1329 @defmac DOUBLE_TYPE_SIZE
1330 A C expression for the size in bits of the type @code{double} on the
1331 target machine. If you don't define this, the default is two
1335 @defmac LONG_DOUBLE_TYPE_SIZE
1336 A C expression for the size in bits of the type @code{long double} on
1337 the target machine. If you don't define this, the default is two
1341 @defmac SHORT_FRACT_TYPE_SIZE
1342 A C expression for the size in bits of the type @code{short _Fract} on
1343 the target machine. If you don't define this, the default is
1344 @code{BITS_PER_UNIT}.
1347 @defmac FRACT_TYPE_SIZE
1348 A C expression for the size in bits of the type @code{_Fract} on
1349 the target machine. If you don't define this, the default is
1350 @code{BITS_PER_UNIT * 2}.
1353 @defmac LONG_FRACT_TYPE_SIZE
1354 A C expression for the size in bits of the type @code{long _Fract} on
1355 the target machine. If you don't define this, the default is
1356 @code{BITS_PER_UNIT * 4}.
1359 @defmac LONG_LONG_FRACT_TYPE_SIZE
1360 A C expression for the size in bits of the type @code{long long _Fract} on
1361 the target machine. If you don't define this, the default is
1362 @code{BITS_PER_UNIT * 8}.
1365 @defmac SHORT_ACCUM_TYPE_SIZE
1366 A C expression for the size in bits of the type @code{short _Accum} on
1367 the target machine. If you don't define this, the default is
1368 @code{BITS_PER_UNIT * 2}.
1371 @defmac ACCUM_TYPE_SIZE
1372 A C expression for the size in bits of the type @code{_Accum} on
1373 the target machine. If you don't define this, the default is
1374 @code{BITS_PER_UNIT * 4}.
1377 @defmac LONG_ACCUM_TYPE_SIZE
1378 A C expression for the size in bits of the type @code{long _Accum} on
1379 the target machine. If you don't define this, the default is
1380 @code{BITS_PER_UNIT * 8}.
1383 @defmac LONG_LONG_ACCUM_TYPE_SIZE
1384 A C expression for the size in bits of the type @code{long long _Accum} on
1385 the target machine. If you don't define this, the default is
1386 @code{BITS_PER_UNIT * 16}.
1389 @defmac LIBGCC2_GNU_PREFIX
1390 This macro corresponds to the @code{TARGET_LIBFUNC_GNU_PREFIX} target
1391 hook and should be defined if that hook is overriden to be true. It
1392 causes function names in libgcc to be changed to use a @code{__gnu_}
1393 prefix for their name rather than the default @code{__}. A port which
1394 uses this macro should also arrange to use @file{t-gnu-prefix} in
1395 the libgcc @file{config.host}.
1398 @defmac TARGET_FLT_EVAL_METHOD
1399 A C expression for the value for @code{FLT_EVAL_METHOD} in @file{float.h},
1400 assuming, if applicable, that the floating-point control word is in its
1401 default state. If you do not define this macro the value of
1402 @code{FLT_EVAL_METHOD} will be zero.
1405 @defmac WIDEST_HARDWARE_FP_SIZE
1406 A C expression for the size in bits of the widest floating-point format
1407 supported by the hardware. If you define this macro, you must specify a
1408 value less than or equal to the value of @code{LONG_DOUBLE_TYPE_SIZE}.
1409 If you do not define this macro, the value of @code{LONG_DOUBLE_TYPE_SIZE}
1413 @defmac DEFAULT_SIGNED_CHAR
1414 An expression whose value is 1 or 0, according to whether the type
1415 @code{char} should be signed or unsigned by default. The user can
1416 always override this default with the options @option{-fsigned-char}
1417 and @option{-funsigned-char}.
1420 @hook TARGET_DEFAULT_SHORT_ENUMS
1423 A C expression for a string describing the name of the data type to use
1424 for size values. The typedef name @code{size_t} is defined using the
1425 contents of the string.
1427 The string can contain more than one keyword. If so, separate them with
1428 spaces, and write first any length keyword, then @code{unsigned} if
1429 appropriate, and finally @code{int}. The string must exactly match one
1430 of the data type names defined in the function
1431 @code{c_common_nodes_and_builtins} in the file @file{c-family/c-common.c}.
1432 You may not omit @code{int} or change the order---that would cause the
1433 compiler to crash on startup.
1435 If you don't define this macro, the default is @code{"long unsigned
1440 GCC defines internal types (@code{sizetype}, @code{ssizetype},
1441 @code{bitsizetype} and @code{sbitsizetype}) for expressions
1442 dealing with size. This macro is a C expression for a string describing
1443 the name of the data type from which the precision of @code{sizetype}
1446 The string has the same restrictions as @code{SIZE_TYPE} string.
1448 If you don't define this macro, the default is @code{SIZE_TYPE}.
1451 @defmac PTRDIFF_TYPE
1452 A C expression for a string describing the name of the data type to use
1453 for the result of subtracting two pointers. The typedef name
1454 @code{ptrdiff_t} is defined using the contents of the string. See
1455 @code{SIZE_TYPE} above for more information.
1457 If you don't define this macro, the default is @code{"long int"}.
1461 A C expression for a string describing the name of the data type to use
1462 for wide characters. The typedef name @code{wchar_t} is defined using
1463 the contents of the string. See @code{SIZE_TYPE} above for more
1466 If you don't define this macro, the default is @code{"int"}.
1469 @defmac WCHAR_TYPE_SIZE
1470 A C expression for the size in bits of the data type for wide
1471 characters. This is used in @code{cpp}, which cannot make use of
1476 A C expression for a string describing the name of the data type to
1477 use for wide characters passed to @code{printf} and returned from
1478 @code{getwc}. The typedef name @code{wint_t} is defined using the
1479 contents of the string. See @code{SIZE_TYPE} above for more
1482 If you don't define this macro, the default is @code{"unsigned int"}.
1486 A C expression for a string describing the name of the data type that
1487 can represent any value of any standard or extended signed integer type.
1488 The typedef name @code{intmax_t} is defined using the contents of the
1489 string. See @code{SIZE_TYPE} above for more information.
1491 If you don't define this macro, the default is the first of
1492 @code{"int"}, @code{"long int"}, or @code{"long long int"} that has as
1493 much precision as @code{long long int}.
1496 @defmac UINTMAX_TYPE
1497 A C expression for a string describing the name of the data type that
1498 can represent any value of any standard or extended unsigned integer
1499 type. The typedef name @code{uintmax_t} is defined using the contents
1500 of the string. See @code{SIZE_TYPE} above for more information.
1502 If you don't define this macro, the default is the first of
1503 @code{"unsigned int"}, @code{"long unsigned int"}, or @code{"long long
1504 unsigned int"} that has as much precision as @code{long long unsigned
1508 @defmac SIG_ATOMIC_TYPE
1514 @defmacx UINT16_TYPE
1515 @defmacx UINT32_TYPE
1516 @defmacx UINT64_TYPE
1517 @defmacx INT_LEAST8_TYPE
1518 @defmacx INT_LEAST16_TYPE
1519 @defmacx INT_LEAST32_TYPE
1520 @defmacx INT_LEAST64_TYPE
1521 @defmacx UINT_LEAST8_TYPE
1522 @defmacx UINT_LEAST16_TYPE
1523 @defmacx UINT_LEAST32_TYPE
1524 @defmacx UINT_LEAST64_TYPE
1525 @defmacx INT_FAST8_TYPE
1526 @defmacx INT_FAST16_TYPE
1527 @defmacx INT_FAST32_TYPE
1528 @defmacx INT_FAST64_TYPE
1529 @defmacx UINT_FAST8_TYPE
1530 @defmacx UINT_FAST16_TYPE
1531 @defmacx UINT_FAST32_TYPE
1532 @defmacx UINT_FAST64_TYPE
1533 @defmacx INTPTR_TYPE
1534 @defmacx UINTPTR_TYPE
1535 C expressions for the standard types @code{sig_atomic_t},
1536 @code{int8_t}, @code{int16_t}, @code{int32_t}, @code{int64_t},
1537 @code{uint8_t}, @code{uint16_t}, @code{uint32_t}, @code{uint64_t},
1538 @code{int_least8_t}, @code{int_least16_t}, @code{int_least32_t},
1539 @code{int_least64_t}, @code{uint_least8_t}, @code{uint_least16_t},
1540 @code{uint_least32_t}, @code{uint_least64_t}, @code{int_fast8_t},
1541 @code{int_fast16_t}, @code{int_fast32_t}, @code{int_fast64_t},
1542 @code{uint_fast8_t}, @code{uint_fast16_t}, @code{uint_fast32_t},
1543 @code{uint_fast64_t}, @code{intptr_t}, and @code{uintptr_t}. See
1544 @code{SIZE_TYPE} above for more information.
1546 If any of these macros evaluates to a null pointer, the corresponding
1547 type is not supported; if GCC is configured to provide
1548 @code{<stdint.h>} in such a case, the header provided may not conform
1549 to C99, depending on the type in question. The defaults for all of
1550 these macros are null pointers.
1553 @defmac TARGET_PTRMEMFUNC_VBIT_LOCATION
1554 The C++ compiler represents a pointer-to-member-function with a struct
1561 ptrdiff_t vtable_index;
1568 The C++ compiler must use one bit to indicate whether the function that
1569 will be called through a pointer-to-member-function is virtual.
1570 Normally, we assume that the low-order bit of a function pointer must
1571 always be zero. Then, by ensuring that the vtable_index is odd, we can
1572 distinguish which variant of the union is in use. But, on some
1573 platforms function pointers can be odd, and so this doesn't work. In
1574 that case, we use the low-order bit of the @code{delta} field, and shift
1575 the remainder of the @code{delta} field to the left.
1577 GCC will automatically make the right selection about where to store
1578 this bit using the @code{FUNCTION_BOUNDARY} setting for your platform.
1579 However, some platforms such as ARM/Thumb have @code{FUNCTION_BOUNDARY}
1580 set such that functions always start at even addresses, but the lowest
1581 bit of pointers to functions indicate whether the function at that
1582 address is in ARM or Thumb mode. If this is the case of your
1583 architecture, you should define this macro to
1584 @code{ptrmemfunc_vbit_in_delta}.
1586 In general, you should not have to define this macro. On architectures
1587 in which function addresses are always even, according to
1588 @code{FUNCTION_BOUNDARY}, GCC will automatically define this macro to
1589 @code{ptrmemfunc_vbit_in_pfn}.
1592 @defmac TARGET_VTABLE_USES_DESCRIPTORS
1593 Normally, the C++ compiler uses function pointers in vtables. This
1594 macro allows the target to change to use ``function descriptors''
1595 instead. Function descriptors are found on targets for whom a
1596 function pointer is actually a small data structure. Normally the
1597 data structure consists of the actual code address plus a data
1598 pointer to which the function's data is relative.
1600 If vtables are used, the value of this macro should be the number
1601 of words that the function descriptor occupies.
1604 @defmac TARGET_VTABLE_ENTRY_ALIGN
1605 By default, the vtable entries are void pointers, the so the alignment
1606 is the same as pointer alignment. The value of this macro specifies
1607 the alignment of the vtable entry in bits. It should be defined only
1608 when special alignment is necessary. */
1611 @defmac TARGET_VTABLE_DATA_ENTRY_DISTANCE
1612 There are a few non-descriptor entries in the vtable at offsets below
1613 zero. If these entries must be padded (say, to preserve the alignment
1614 specified by @code{TARGET_VTABLE_ENTRY_ALIGN}), set this to the number
1615 of words in each data entry.
1619 @section Register Usage
1620 @cindex register usage
1622 This section explains how to describe what registers the target machine
1623 has, and how (in general) they can be used.
1625 The description of which registers a specific instruction can use is
1626 done with register classes; see @ref{Register Classes}. For information
1627 on using registers to access a stack frame, see @ref{Frame Registers}.
1628 For passing values in registers, see @ref{Register Arguments}.
1629 For returning values in registers, see @ref{Scalar Return}.
1632 * Register Basics:: Number and kinds of registers.
1633 * Allocation Order:: Order in which registers are allocated.
1634 * Values in Registers:: What kinds of values each reg can hold.
1635 * Leaf Functions:: Renumbering registers for leaf functions.
1636 * Stack Registers:: Handling a register stack such as 80387.
1639 @node Register Basics
1640 @subsection Basic Characteristics of Registers
1642 @c prevent bad page break with this line
1643 Registers have various characteristics.
1645 @defmac FIRST_PSEUDO_REGISTER
1646 Number of hardware registers known to the compiler. They receive
1647 numbers 0 through @code{FIRST_PSEUDO_REGISTER-1}; thus, the first
1648 pseudo register's number really is assigned the number
1649 @code{FIRST_PSEUDO_REGISTER}.
1652 @defmac FIXED_REGISTERS
1653 @cindex fixed register
1654 An initializer that says which registers are used for fixed purposes
1655 all throughout the compiled code and are therefore not available for
1656 general allocation. These would include the stack pointer, the frame
1657 pointer (except on machines where that can be used as a general
1658 register when no frame pointer is needed), the program counter on
1659 machines where that is considered one of the addressable registers,
1660 and any other numbered register with a standard use.
1662 This information is expressed as a sequence of numbers, separated by
1663 commas and surrounded by braces. The @var{n}th number is 1 if
1664 register @var{n} is fixed, 0 otherwise.
1666 The table initialized from this macro, and the table initialized by
1667 the following one, may be overridden at run time either automatically,
1668 by the actions of the macro @code{CONDITIONAL_REGISTER_USAGE}, or by
1669 the user with the command options @option{-ffixed-@var{reg}},
1670 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}.
1673 @defmac CALL_USED_REGISTERS
1674 @cindex call-used register
1675 @cindex call-clobbered register
1676 @cindex call-saved register
1677 Like @code{FIXED_REGISTERS} but has 1 for each register that is
1678 clobbered (in general) by function calls as well as for fixed
1679 registers. This macro therefore identifies the registers that are not
1680 available for general allocation of values that must live across
1683 If a register has 0 in @code{CALL_USED_REGISTERS}, the compiler
1684 automatically saves it on function entry and restores it on function
1685 exit, if the register is used within the function.
1688 @defmac CALL_REALLY_USED_REGISTERS
1689 @cindex call-used register
1690 @cindex call-clobbered register
1691 @cindex call-saved register
1692 Like @code{CALL_USED_REGISTERS} except this macro doesn't require
1693 that the entire set of @code{FIXED_REGISTERS} be included.
1694 (@code{CALL_USED_REGISTERS} must be a superset of @code{FIXED_REGISTERS}).
1695 This macro is optional. If not specified, it defaults to the value
1696 of @code{CALL_USED_REGISTERS}.
1699 @defmac HARD_REGNO_CALL_PART_CLOBBERED (@var{regno}, @var{mode})
1700 @cindex call-used register
1701 @cindex call-clobbered register
1702 @cindex call-saved register
1703 A C expression that is nonzero if it is not permissible to store a
1704 value of mode @var{mode} in hard register number @var{regno} across a
1705 call without some part of it being clobbered. For most machines this
1706 macro need not be defined. It is only required for machines that do not
1707 preserve the entire contents of a register across a call.
1711 @findex call_used_regs
1714 @findex reg_class_contents
1715 @hook TARGET_CONDITIONAL_REGISTER_USAGE
1717 @defmac INCOMING_REGNO (@var{out})
1718 Define this macro if the target machine has register windows. This C
1719 expression returns the register number as seen by the called function
1720 corresponding to the register number @var{out} as seen by the calling
1721 function. Return @var{out} if register number @var{out} is not an
1725 @defmac OUTGOING_REGNO (@var{in})
1726 Define this macro if the target machine has register windows. This C
1727 expression returns the register number as seen by the calling function
1728 corresponding to the register number @var{in} as seen by the called
1729 function. Return @var{in} if register number @var{in} is not an inbound
1733 @defmac LOCAL_REGNO (@var{regno})
1734 Define this macro if the target machine has register windows. This C
1735 expression returns true if the register is call-saved but is in the
1736 register window. Unlike most call-saved registers, such registers
1737 need not be explicitly restored on function exit or during non-local
1742 If the program counter has a register number, define this as that
1743 register number. Otherwise, do not define it.
1746 @node Allocation Order
1747 @subsection Order of Allocation of Registers
1748 @cindex order of register allocation
1749 @cindex register allocation order
1751 @c prevent bad page break with this line
1752 Registers are allocated in order.
1754 @defmac REG_ALLOC_ORDER
1755 If defined, an initializer for a vector of integers, containing the
1756 numbers of hard registers in the order in which GCC should prefer
1757 to use them (from most preferred to least).
1759 If this macro is not defined, registers are used lowest numbered first
1760 (all else being equal).
1762 One use of this macro is on machines where the highest numbered
1763 registers must always be saved and the save-multiple-registers
1764 instruction supports only sequences of consecutive registers. On such
1765 machines, define @code{REG_ALLOC_ORDER} to be an initializer that lists
1766 the highest numbered allocable register first.
1769 @defmac ADJUST_REG_ALLOC_ORDER
1770 A C statement (sans semicolon) to choose the order in which to allocate
1771 hard registers for pseudo-registers local to a basic block.
1773 Store the desired register order in the array @code{reg_alloc_order}.
1774 Element 0 should be the register to allocate first; element 1, the next
1775 register; and so on.
1777 The macro body should not assume anything about the contents of
1778 @code{reg_alloc_order} before execution of the macro.
1780 On most machines, it is not necessary to define this macro.
1783 @defmac HONOR_REG_ALLOC_ORDER
1784 Normally, IRA tries to estimate the costs for saving a register in the
1785 prologue and restoring it in the epilogue. This discourages it from
1786 using call-saved registers. If a machine wants to ensure that IRA
1787 allocates registers in the order given by REG_ALLOC_ORDER even if some
1788 call-saved registers appear earlier than call-used ones, then define this
1789 macro as a C expression to nonzero. Default is 0.
1792 @defmac IRA_HARD_REGNO_ADD_COST_MULTIPLIER (@var{regno})
1793 In some case register allocation order is not enough for the
1794 Integrated Register Allocator (@acronym{IRA}) to generate a good code.
1795 If this macro is defined, it should return a floating point value
1796 based on @var{regno}. The cost of using @var{regno} for a pseudo will
1797 be increased by approximately the pseudo's usage frequency times the
1798 value returned by this macro. Not defining this macro is equivalent
1799 to having it always return @code{0.0}.
1801 On most machines, it is not necessary to define this macro.
1804 @node Values in Registers
1805 @subsection How Values Fit in Registers
1807 This section discusses the macros that describe which kinds of values
1808 (specifically, which machine modes) each register can hold, and how many
1809 consecutive registers are needed for a given mode.
1811 @defmac HARD_REGNO_NREGS (@var{regno}, @var{mode})
1812 A C expression for the number of consecutive hard registers, starting
1813 at register number @var{regno}, required to hold a value of mode
1814 @var{mode}. This macro must never return zero, even if a register
1815 cannot hold the requested mode - indicate that with HARD_REGNO_MODE_OK
1816 and/or CANNOT_CHANGE_MODE_CLASS instead.
1818 On a machine where all registers are exactly one word, a suitable
1819 definition of this macro is
1822 #define HARD_REGNO_NREGS(REGNO, MODE) \
1823 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) \
1828 @defmac HARD_REGNO_NREGS_HAS_PADDING (@var{regno}, @var{mode})
1829 A C expression that is nonzero if a value of mode @var{mode}, stored
1830 in memory, ends with padding that causes it to take up more space than
1831 in registers starting at register number @var{regno} (as determined by
1832 multiplying GCC's notion of the size of the register when containing
1833 this mode by the number of registers returned by
1834 @code{HARD_REGNO_NREGS}). By default this is zero.
1836 For example, if a floating-point value is stored in three 32-bit
1837 registers but takes up 128 bits in memory, then this would be
1840 This macros only needs to be defined if there are cases where
1841 @code{subreg_get_info}
1842 would otherwise wrongly determine that a @code{subreg} can be
1843 represented by an offset to the register number, when in fact such a
1844 @code{subreg} would contain some of the padding not stored in
1845 registers and so not be representable.
1848 @defmac HARD_REGNO_NREGS_WITH_PADDING (@var{regno}, @var{mode})
1849 For values of @var{regno} and @var{mode} for which
1850 @code{HARD_REGNO_NREGS_HAS_PADDING} returns nonzero, a C expression
1851 returning the greater number of registers required to hold the value
1852 including any padding. In the example above, the value would be four.
1855 @defmac REGMODE_NATURAL_SIZE (@var{mode})
1856 Define this macro if the natural size of registers that hold values
1857 of mode @var{mode} is not the word size. It is a C expression that
1858 should give the natural size in bytes for the specified mode. It is
1859 used by the register allocator to try to optimize its results. This
1860 happens for example on SPARC 64-bit where the natural size of
1861 floating-point registers is still 32-bit.
1864 @defmac HARD_REGNO_MODE_OK (@var{regno}, @var{mode})
1865 A C expression that is nonzero if it is permissible to store a value
1866 of mode @var{mode} in hard register number @var{regno} (or in several
1867 registers starting with that one). For a machine where all registers
1868 are equivalent, a suitable definition is
1871 #define HARD_REGNO_MODE_OK(REGNO, MODE) 1
1874 You need not include code to check for the numbers of fixed registers,
1875 because the allocation mechanism considers them to be always occupied.
1877 @cindex register pairs
1878 On some machines, double-precision values must be kept in even/odd
1879 register pairs. You can implement that by defining this macro to reject
1880 odd register numbers for such modes.
1882 The minimum requirement for a mode to be OK in a register is that the
1883 @samp{mov@var{mode}} instruction pattern support moves between the
1884 register and other hard register in the same class and that moving a
1885 value into the register and back out not alter it.
1887 Since the same instruction used to move @code{word_mode} will work for
1888 all narrower integer modes, it is not necessary on any machine for
1889 @code{HARD_REGNO_MODE_OK} to distinguish between these modes, provided
1890 you define patterns @samp{movhi}, etc., to take advantage of this. This
1891 is useful because of the interaction between @code{HARD_REGNO_MODE_OK}
1892 and @code{MODES_TIEABLE_P}; it is very desirable for all integer modes
1895 Many machines have special registers for floating point arithmetic.
1896 Often people assume that floating point machine modes are allowed only
1897 in floating point registers. This is not true. Any registers that
1898 can hold integers can safely @emph{hold} a floating point machine
1899 mode, whether or not floating arithmetic can be done on it in those
1900 registers. Integer move instructions can be used to move the values.
1902 On some machines, though, the converse is true: fixed-point machine
1903 modes may not go in floating registers. This is true if the floating
1904 registers normalize any value stored in them, because storing a
1905 non-floating value there would garble it. In this case,
1906 @code{HARD_REGNO_MODE_OK} should reject fixed-point machine modes in
1907 floating registers. But if the floating registers do not automatically
1908 normalize, if you can store any bit pattern in one and retrieve it
1909 unchanged without a trap, then any machine mode may go in a floating
1910 register, so you can define this macro to say so.
1912 The primary significance of special floating registers is rather that
1913 they are the registers acceptable in floating point arithmetic
1914 instructions. However, this is of no concern to
1915 @code{HARD_REGNO_MODE_OK}. You handle it by writing the proper
1916 constraints for those instructions.
1918 On some machines, the floating registers are especially slow to access,
1919 so that it is better to store a value in a stack frame than in such a
1920 register if floating point arithmetic is not being done. As long as the
1921 floating registers are not in class @code{GENERAL_REGS}, they will not
1922 be used unless some pattern's constraint asks for one.
1925 @defmac HARD_REGNO_RENAME_OK (@var{from}, @var{to})
1926 A C expression that is nonzero if it is OK to rename a hard register
1927 @var{from} to another hard register @var{to}.
1929 One common use of this macro is to prevent renaming of a register to
1930 another register that is not saved by a prologue in an interrupt
1933 The default is always nonzero.
1936 @defmac MODES_TIEABLE_P (@var{mode1}, @var{mode2})
1937 A C expression that is nonzero if a value of mode
1938 @var{mode1} is accessible in mode @var{mode2} without copying.
1940 If @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode1})} and
1941 @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode2})} are always the same for
1942 any @var{r}, then @code{MODES_TIEABLE_P (@var{mode1}, @var{mode2})}
1943 should be nonzero. If they differ for any @var{r}, you should define
1944 this macro to return zero unless some other mechanism ensures the
1945 accessibility of the value in a narrower mode.
1947 You should define this macro to return nonzero in as many cases as
1948 possible since doing so will allow GCC to perform better register
1952 @hook TARGET_HARD_REGNO_SCRATCH_OK
1954 @defmac AVOID_CCMODE_COPIES
1955 Define this macro if the compiler should avoid copies to/from @code{CCmode}
1956 registers. You should only define this macro if support for copying to/from
1957 @code{CCmode} is incomplete.
1960 @node Leaf Functions
1961 @subsection Handling Leaf Functions
1963 @cindex leaf functions
1964 @cindex functions, leaf
1965 On some machines, a leaf function (i.e., one which makes no calls) can run
1966 more efficiently if it does not make its own register window. Often this
1967 means it is required to receive its arguments in the registers where they
1968 are passed by the caller, instead of the registers where they would
1971 The special treatment for leaf functions generally applies only when
1972 other conditions are met; for example, often they may use only those
1973 registers for its own variables and temporaries. We use the term ``leaf
1974 function'' to mean a function that is suitable for this special
1975 handling, so that functions with no calls are not necessarily ``leaf
1978 GCC assigns register numbers before it knows whether the function is
1979 suitable for leaf function treatment. So it needs to renumber the
1980 registers in order to output a leaf function. The following macros
1983 @defmac LEAF_REGISTERS
1984 Name of a char vector, indexed by hard register number, which
1985 contains 1 for a register that is allowable in a candidate for leaf
1988 If leaf function treatment involves renumbering the registers, then the
1989 registers marked here should be the ones before renumbering---those that
1990 GCC would ordinarily allocate. The registers which will actually be
1991 used in the assembler code, after renumbering, should not be marked with 1
1994 Define this macro only if the target machine offers a way to optimize
1995 the treatment of leaf functions.
1998 @defmac LEAF_REG_REMAP (@var{regno})
1999 A C expression whose value is the register number to which @var{regno}
2000 should be renumbered, when a function is treated as a leaf function.
2002 If @var{regno} is a register number which should not appear in a leaf
2003 function before renumbering, then the expression should yield @minus{}1, which
2004 will cause the compiler to abort.
2006 Define this macro only if the target machine offers a way to optimize the
2007 treatment of leaf functions, and registers need to be renumbered to do
2011 @findex current_function_is_leaf
2012 @findex current_function_uses_only_leaf_regs
2013 @code{TARGET_ASM_FUNCTION_PROLOGUE} and
2014 @code{TARGET_ASM_FUNCTION_EPILOGUE} must usually treat leaf functions
2015 specially. They can test the C variable @code{current_function_is_leaf}
2016 which is nonzero for leaf functions. @code{current_function_is_leaf} is
2017 set prior to local register allocation and is valid for the remaining
2018 compiler passes. They can also test the C variable
2019 @code{current_function_uses_only_leaf_regs} which is nonzero for leaf
2020 functions which only use leaf registers.
2021 @code{current_function_uses_only_leaf_regs} is valid after all passes
2022 that modify the instructions have been run and is only useful if
2023 @code{LEAF_REGISTERS} is defined.
2024 @c changed this to fix overfull. ALSO: why the "it" at the beginning
2025 @c of the next paragraph?! --mew 2feb93
2027 @node Stack Registers
2028 @subsection Registers That Form a Stack
2030 There are special features to handle computers where some of the
2031 ``registers'' form a stack. Stack registers are normally written by
2032 pushing onto the stack, and are numbered relative to the top of the
2035 Currently, GCC can only handle one group of stack-like registers, and
2036 they must be consecutively numbered. Furthermore, the existing
2037 support for stack-like registers is specific to the 80387 floating
2038 point coprocessor. If you have a new architecture that uses
2039 stack-like registers, you will need to do substantial work on
2040 @file{reg-stack.c} and write your machine description to cooperate
2041 with it, as well as defining these macros.
2044 Define this if the machine has any stack-like registers.
2047 @defmac STACK_REG_COVER_CLASS
2048 This is a cover class containing the stack registers. Define this if
2049 the machine has any stack-like registers.
2052 @defmac FIRST_STACK_REG
2053 The number of the first stack-like register. This one is the top
2057 @defmac LAST_STACK_REG
2058 The number of the last stack-like register. This one is the bottom of
2062 @node Register Classes
2063 @section Register Classes
2064 @cindex register class definitions
2065 @cindex class definitions, register
2067 On many machines, the numbered registers are not all equivalent.
2068 For example, certain registers may not be allowed for indexed addressing;
2069 certain registers may not be allowed in some instructions. These machine
2070 restrictions are described to the compiler using @dfn{register classes}.
2072 You define a number of register classes, giving each one a name and saying
2073 which of the registers belong to it. Then you can specify register classes
2074 that are allowed as operands to particular instruction patterns.
2078 In general, each register will belong to several classes. In fact, one
2079 class must be named @code{ALL_REGS} and contain all the registers. Another
2080 class must be named @code{NO_REGS} and contain no registers. Often the
2081 union of two classes will be another class; however, this is not required.
2083 @findex GENERAL_REGS
2084 One of the classes must be named @code{GENERAL_REGS}. There is nothing
2085 terribly special about the name, but the operand constraint letters
2086 @samp{r} and @samp{g} specify this class. If @code{GENERAL_REGS} is
2087 the same as @code{ALL_REGS}, just define it as a macro which expands
2090 Order the classes so that if class @var{x} is contained in class @var{y}
2091 then @var{x} has a lower class number than @var{y}.
2093 The way classes other than @code{GENERAL_REGS} are specified in operand
2094 constraints is through machine-dependent operand constraint letters.
2095 You can define such letters to correspond to various classes, then use
2096 them in operand constraints.
2098 You must define the narrowest register classes for allocatable
2099 registers, so that each class either has no subclasses, or that for
2100 some mode, the move cost between registers within the class is
2101 cheaper than moving a register in the class to or from memory
2104 You should define a class for the union of two classes whenever some
2105 instruction allows both classes. For example, if an instruction allows
2106 either a floating point (coprocessor) register or a general register for a
2107 certain operand, you should define a class @code{FLOAT_OR_GENERAL_REGS}
2108 which includes both of them. Otherwise you will get suboptimal code,
2109 or even internal compiler errors when reload cannot find a register in the
2110 class computed via @code{reg_class_subunion}.
2112 You must also specify certain redundant information about the register
2113 classes: for each class, which classes contain it and which ones are
2114 contained in it; for each pair of classes, the largest class contained
2117 When a value occupying several consecutive registers is expected in a
2118 certain class, all the registers used must belong to that class.
2119 Therefore, register classes cannot be used to enforce a requirement for
2120 a register pair to start with an even-numbered register. The way to
2121 specify this requirement is with @code{HARD_REGNO_MODE_OK}.
2123 Register classes used for input-operands of bitwise-and or shift
2124 instructions have a special requirement: each such class must have, for
2125 each fixed-point machine mode, a subclass whose registers can transfer that
2126 mode to or from memory. For example, on some machines, the operations for
2127 single-byte values (@code{QImode}) are limited to certain registers. When
2128 this is so, each register class that is used in a bitwise-and or shift
2129 instruction must have a subclass consisting of registers from which
2130 single-byte values can be loaded or stored. This is so that
2131 @code{PREFERRED_RELOAD_CLASS} can always have a possible value to return.
2133 @deftp {Data type} {enum reg_class}
2134 An enumerated type that must be defined with all the register class names
2135 as enumerated values. @code{NO_REGS} must be first. @code{ALL_REGS}
2136 must be the last register class, followed by one more enumerated value,
2137 @code{LIM_REG_CLASSES}, which is not a register class but rather
2138 tells how many classes there are.
2140 Each register class has a number, which is the value of casting
2141 the class name to type @code{int}. The number serves as an index
2142 in many of the tables described below.
2145 @defmac N_REG_CLASSES
2146 The number of distinct register classes, defined as follows:
2149 #define N_REG_CLASSES (int) LIM_REG_CLASSES
2153 @defmac REG_CLASS_NAMES
2154 An initializer containing the names of the register classes as C string
2155 constants. These names are used in writing some of the debugging dumps.
2158 @defmac REG_CLASS_CONTENTS
2159 An initializer containing the contents of the register classes, as integers
2160 which are bit masks. The @var{n}th integer specifies the contents of class
2161 @var{n}. The way the integer @var{mask} is interpreted is that
2162 register @var{r} is in the class if @code{@var{mask} & (1 << @var{r})} is 1.
2164 When the machine has more than 32 registers, an integer does not suffice.
2165 Then the integers are replaced by sub-initializers, braced groupings containing
2166 several integers. Each sub-initializer must be suitable as an initializer
2167 for the type @code{HARD_REG_SET} which is defined in @file{hard-reg-set.h}.
2168 In this situation, the first integer in each sub-initializer corresponds to
2169 registers 0 through 31, the second integer to registers 32 through 63, and
2173 @defmac REGNO_REG_CLASS (@var{regno})
2174 A C expression whose value is a register class containing hard register
2175 @var{regno}. In general there is more than one such class; choose a class
2176 which is @dfn{minimal}, meaning that no smaller class also contains the
2180 @defmac BASE_REG_CLASS
2181 A macro whose definition is the name of the class to which a valid
2182 base register must belong. A base register is one used in an address
2183 which is the register value plus a displacement.
2186 @defmac MODE_BASE_REG_CLASS (@var{mode})
2187 This is a variation of the @code{BASE_REG_CLASS} macro which allows
2188 the selection of a base register in a mode dependent manner. If
2189 @var{mode} is VOIDmode then it should return the same value as
2190 @code{BASE_REG_CLASS}.
2193 @defmac MODE_BASE_REG_REG_CLASS (@var{mode})
2194 A C expression whose value is the register class to which a valid
2195 base register must belong in order to be used in a base plus index
2196 register address. You should define this macro if base plus index
2197 addresses have different requirements than other base register uses.
2200 @defmac MODE_CODE_BASE_REG_CLASS (@var{mode}, @var{address_space}, @var{outer_code}, @var{index_code})
2201 A C expression whose value is the register class to which a valid
2202 base register for a memory reference in mode @var{mode} to address
2203 space @var{address_space} must belong. @var{outer_code} and @var{index_code}
2204 define the context in which the base register occurs. @var{outer_code} is
2205 the code of the immediately enclosing expression (@code{MEM} for the top level
2206 of an address, @code{ADDRESS} for something that occurs in an
2207 @code{address_operand}). @var{index_code} is the code of the corresponding
2208 index expression if @var{outer_code} is @code{PLUS}; @code{SCRATCH} otherwise.
2211 @defmac INDEX_REG_CLASS
2212 A macro whose definition is the name of the class to which a valid
2213 index register must belong. An index register is one used in an
2214 address where its value is either multiplied by a scale factor or
2215 added to another register (as well as added to a displacement).
2218 @defmac REGNO_OK_FOR_BASE_P (@var{num})
2219 A C expression which is nonzero if register number @var{num} is
2220 suitable for use as a base register in operand addresses.
2223 @defmac REGNO_MODE_OK_FOR_BASE_P (@var{num}, @var{mode})
2224 A C expression that is just like @code{REGNO_OK_FOR_BASE_P}, except that
2225 that expression may examine the mode of the memory reference in
2226 @var{mode}. You should define this macro if the mode of the memory
2227 reference affects whether a register may be used as a base register. If
2228 you define this macro, the compiler will use it instead of
2229 @code{REGNO_OK_FOR_BASE_P}. The mode may be @code{VOIDmode} for
2230 addresses that appear outside a @code{MEM}, i.e., as an
2231 @code{address_operand}.
2234 @defmac REGNO_MODE_OK_FOR_REG_BASE_P (@var{num}, @var{mode})
2235 A C expression which is nonzero if register number @var{num} is suitable for
2236 use as a base register in base plus index operand addresses, accessing
2237 memory in mode @var{mode}. It may be either a suitable hard register or a
2238 pseudo register that has been allocated such a hard register. You should
2239 define this macro if base plus index addresses have different requirements
2240 than other base register uses.
2242 Use of this macro is deprecated; please use the more general
2243 @code{REGNO_MODE_CODE_OK_FOR_BASE_P}.
2246 @defmac REGNO_MODE_CODE_OK_FOR_BASE_P (@var{num}, @var{mode}, @var{address_space}, @var{outer_code}, @var{index_code})
2247 A C expression which is nonzero if register number @var{num} is
2248 suitable for use as a base register in operand addresses, accessing
2249 memory in mode @var{mode} in address space @var{address_space}.
2250 This is similar to @code{REGNO_MODE_OK_FOR_BASE_P}, except
2251 that that expression may examine the context in which the register
2252 appears in the memory reference. @var{outer_code} is the code of the
2253 immediately enclosing expression (@code{MEM} if at the top level of the
2254 address, @code{ADDRESS} for something that occurs in an
2255 @code{address_operand}). @var{index_code} is the code of the
2256 corresponding index expression if @var{outer_code} is @code{PLUS};
2257 @code{SCRATCH} otherwise. The mode may be @code{VOIDmode} for addresses
2258 that appear outside a @code{MEM}, i.e., as an @code{address_operand}.
2261 @defmac REGNO_OK_FOR_INDEX_P (@var{num})
2262 A C expression which is nonzero if register number @var{num} is
2263 suitable for use as an index register in operand addresses. It may be
2264 either a suitable hard register or a pseudo register that has been
2265 allocated such a hard register.
2267 The difference between an index register and a base register is that
2268 the index register may be scaled. If an address involves the sum of
2269 two registers, neither one of them scaled, then either one may be
2270 labeled the ``base'' and the other the ``index''; but whichever
2271 labeling is used must fit the machine's constraints of which registers
2272 may serve in each capacity. The compiler will try both labelings,
2273 looking for one that is valid, and will reload one or both registers
2274 only if neither labeling works.
2277 @hook TARGET_PREFERRED_RENAME_CLASS
2279 @hook TARGET_PREFERRED_RELOAD_CLASS
2281 @defmac PREFERRED_RELOAD_CLASS (@var{x}, @var{class})
2282 A C expression that places additional restrictions on the register class
2283 to use when it is necessary to copy value @var{x} into a register in class
2284 @var{class}. The value is a register class; perhaps @var{class}, or perhaps
2285 another, smaller class. On many machines, the following definition is
2289 #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS
2292 Sometimes returning a more restrictive class makes better code. For
2293 example, on the 68000, when @var{x} is an integer constant that is in range
2294 for a @samp{moveq} instruction, the value of this macro is always
2295 @code{DATA_REGS} as long as @var{class} includes the data registers.
2296 Requiring a data register guarantees that a @samp{moveq} will be used.
2298 One case where @code{PREFERRED_RELOAD_CLASS} must not return
2299 @var{class} is if @var{x} is a legitimate constant which cannot be
2300 loaded into some register class. By returning @code{NO_REGS} you can
2301 force @var{x} into a memory location. For example, rs6000 can load
2302 immediate values into general-purpose registers, but does not have an
2303 instruction for loading an immediate value into a floating-point
2304 register, so @code{PREFERRED_RELOAD_CLASS} returns @code{NO_REGS} when
2305 @var{x} is a floating-point constant. If the constant can't be loaded
2306 into any kind of register, code generation will be better if
2307 @code{TARGET_LEGITIMATE_CONSTANT_P} makes the constant illegitimate instead
2308 of using @code{TARGET_PREFERRED_RELOAD_CLASS}.
2310 If an insn has pseudos in it after register allocation, reload will go
2311 through the alternatives and call repeatedly @code{PREFERRED_RELOAD_CLASS}
2312 to find the best one. Returning @code{NO_REGS}, in this case, makes
2313 reload add a @code{!} in front of the constraint: the x86 back-end uses
2314 this feature to discourage usage of 387 registers when math is done in
2315 the SSE registers (and vice versa).
2318 @hook TARGET_PREFERRED_OUTPUT_RELOAD_CLASS
2320 @defmac LIMIT_RELOAD_CLASS (@var{mode}, @var{class})
2321 A C expression that places additional restrictions on the register class
2322 to use when it is necessary to be able to hold a value of mode
2323 @var{mode} in a reload register for which class @var{class} would
2326 Unlike @code{PREFERRED_RELOAD_CLASS}, this macro should be used when
2327 there are certain modes that simply can't go in certain reload classes.
2329 The value is a register class; perhaps @var{class}, or perhaps another,
2332 Don't define this macro unless the target machine has limitations which
2333 require the macro to do something nontrivial.
2336 @hook TARGET_SECONDARY_RELOAD
2338 @defmac SECONDARY_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2339 @defmacx SECONDARY_INPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2340 @defmacx SECONDARY_OUTPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2341 These macros are obsolete, new ports should use the target hook
2342 @code{TARGET_SECONDARY_RELOAD} instead.
2344 These are obsolete macros, replaced by the @code{TARGET_SECONDARY_RELOAD}
2345 target hook. Older ports still define these macros to indicate to the
2346 reload phase that it may
2347 need to allocate at least one register for a reload in addition to the
2348 register to contain the data. Specifically, if copying @var{x} to a
2349 register @var{class} in @var{mode} requires an intermediate register,
2350 you were supposed to define @code{SECONDARY_INPUT_RELOAD_CLASS} to return the
2351 largest register class all of whose registers can be used as
2352 intermediate registers or scratch registers.
2354 If copying a register @var{class} in @var{mode} to @var{x} requires an
2355 intermediate or scratch register, @code{SECONDARY_OUTPUT_RELOAD_CLASS}
2356 was supposed to be defined be defined to return the largest register
2357 class required. If the
2358 requirements for input and output reloads were the same, the macro
2359 @code{SECONDARY_RELOAD_CLASS} should have been used instead of defining both
2362 The values returned by these macros are often @code{GENERAL_REGS}.
2363 Return @code{NO_REGS} if no spare register is needed; i.e., if @var{x}
2364 can be directly copied to or from a register of @var{class} in
2365 @var{mode} without requiring a scratch register. Do not define this
2366 macro if it would always return @code{NO_REGS}.
2368 If a scratch register is required (either with or without an
2369 intermediate register), you were supposed to define patterns for
2370 @samp{reload_in@var{m}} or @samp{reload_out@var{m}}, as required
2371 (@pxref{Standard Names}. These patterns, which were normally
2372 implemented with a @code{define_expand}, should be similar to the
2373 @samp{mov@var{m}} patterns, except that operand 2 is the scratch
2376 These patterns need constraints for the reload register and scratch
2378 contain a single register class. If the original reload register (whose
2379 class is @var{class}) can meet the constraint given in the pattern, the
2380 value returned by these macros is used for the class of the scratch
2381 register. Otherwise, two additional reload registers are required.
2382 Their classes are obtained from the constraints in the insn pattern.
2384 @var{x} might be a pseudo-register or a @code{subreg} of a
2385 pseudo-register, which could either be in a hard register or in memory.
2386 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2387 in memory and the hard register number if it is in a register.
2389 These macros should not be used in the case where a particular class of
2390 registers can only be copied to memory and not to another class of
2391 registers. In that case, secondary reload registers are not needed and
2392 would not be helpful. Instead, a stack location must be used to perform
2393 the copy and the @code{mov@var{m}} pattern should use memory as an
2394 intermediate storage. This case often occurs between floating-point and
2398 @defmac SECONDARY_MEMORY_NEEDED (@var{class1}, @var{class2}, @var{m})
2399 Certain machines have the property that some registers cannot be copied
2400 to some other registers without using memory. Define this macro on
2401 those machines to be a C expression that is nonzero if objects of mode
2402 @var{m} in registers of @var{class1} can only be copied to registers of
2403 class @var{class2} by storing a register of @var{class1} into memory
2404 and loading that memory location into a register of @var{class2}.
2406 Do not define this macro if its value would always be zero.
2409 @defmac SECONDARY_MEMORY_NEEDED_RTX (@var{mode})
2410 Normally when @code{SECONDARY_MEMORY_NEEDED} is defined, the compiler
2411 allocates a stack slot for a memory location needed for register copies.
2412 If this macro is defined, the compiler instead uses the memory location
2413 defined by this macro.
2415 Do not define this macro if you do not define
2416 @code{SECONDARY_MEMORY_NEEDED}.
2419 @defmac SECONDARY_MEMORY_NEEDED_MODE (@var{mode})
2420 When the compiler needs a secondary memory location to copy between two
2421 registers of mode @var{mode}, it normally allocates sufficient memory to
2422 hold a quantity of @code{BITS_PER_WORD} bits and performs the store and
2423 load operations in a mode that many bits wide and whose class is the
2424 same as that of @var{mode}.
2426 This is right thing to do on most machines because it ensures that all
2427 bits of the register are copied and prevents accesses to the registers
2428 in a narrower mode, which some machines prohibit for floating-point
2431 However, this default behavior is not correct on some machines, such as
2432 the DEC Alpha, that store short integers in floating-point registers
2433 differently than in integer registers. On those machines, the default
2434 widening will not work correctly and you must define this macro to
2435 suppress that widening in some cases. See the file @file{alpha.h} for
2438 Do not define this macro if you do not define
2439 @code{SECONDARY_MEMORY_NEEDED} or if widening @var{mode} to a mode that
2440 is @code{BITS_PER_WORD} bits wide is correct for your machine.
2443 @hook TARGET_CLASS_LIKELY_SPILLED_P
2445 @hook TARGET_CLASS_MAX_NREGS
2447 @defmac CLASS_MAX_NREGS (@var{class}, @var{mode})
2448 A C expression for the maximum number of consecutive registers
2449 of class @var{class} needed to hold a value of mode @var{mode}.
2451 This is closely related to the macro @code{HARD_REGNO_NREGS}. In fact,
2452 the value of the macro @code{CLASS_MAX_NREGS (@var{class}, @var{mode})}
2453 should be the maximum value of @code{HARD_REGNO_NREGS (@var{regno},
2454 @var{mode})} for all @var{regno} values in the class @var{class}.
2456 This macro helps control the handling of multiple-word values
2460 @defmac CANNOT_CHANGE_MODE_CLASS (@var{from}, @var{to}, @var{class})
2461 If defined, a C expression that returns nonzero for a @var{class} for which
2462 a change from mode @var{from} to mode @var{to} is invalid.
2464 For the example, loading 32-bit integer or floating-point objects into
2465 floating-point registers on the Alpha extends them to 64 bits.
2466 Therefore loading a 64-bit object and then storing it as a 32-bit object
2467 does not store the low-order 32 bits, as would be the case for a normal
2468 register. Therefore, @file{alpha.h} defines @code{CANNOT_CHANGE_MODE_CLASS}
2472 #define CANNOT_CHANGE_MODE_CLASS(FROM, TO, CLASS) \
2473 (GET_MODE_SIZE (FROM) != GET_MODE_SIZE (TO) \
2474 ? reg_classes_intersect_p (FLOAT_REGS, (CLASS)) : 0)
2480 @hook TARGET_REGISTER_PRIORITY
2482 @hook TARGET_REGISTER_USAGE_LEVELING_P
2484 @hook TARGET_DIFFERENT_ADDR_DISPLACEMENT_P
2486 @hook TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P
2488 @hook TARGET_LEGITIMIZE_ADDRESS_DISPLACEMENT
2490 @hook TARGET_SPILL_CLASS
2492 @hook TARGET_CSTORE_MODE
2494 @node Stack and Calling
2495 @section Stack Layout and Calling Conventions
2496 @cindex calling conventions
2498 @c prevent bad page break with this line
2499 This describes the stack layout and calling conventions.
2503 * Exception Handling::
2508 * Register Arguments::
2510 * Aggregate Return::
2515 * Stack Smashing Protection::
2516 * Miscellaneous Register Hooks::
2520 @subsection Basic Stack Layout
2521 @cindex stack frame layout
2522 @cindex frame layout
2524 @c prevent bad page break with this line
2525 Here is the basic stack layout.
2527 @defmac STACK_GROWS_DOWNWARD
2528 Define this macro if pushing a word onto the stack moves the stack
2529 pointer to a smaller address.
2531 When we say, ``define this macro if @dots{}'', it means that the
2532 compiler checks this macro only with @code{#ifdef} so the precise
2533 definition used does not matter.
2536 @defmac STACK_PUSH_CODE
2537 This macro defines the operation used when something is pushed
2538 on the stack. In RTL, a push operation will be
2539 @code{(set (mem (STACK_PUSH_CODE (reg sp))) @dots{})}
2541 The choices are @code{PRE_DEC}, @code{POST_DEC}, @code{PRE_INC},
2542 and @code{POST_INC}. Which of these is correct depends on
2543 the stack direction and on whether the stack pointer points
2544 to the last item on the stack or whether it points to the
2545 space for the next item on the stack.
2547 The default is @code{PRE_DEC} when @code{STACK_GROWS_DOWNWARD} is
2548 defined, which is almost always right, and @code{PRE_INC} otherwise,
2549 which is often wrong.
2552 @defmac FRAME_GROWS_DOWNWARD
2553 Define this macro to nonzero value if the addresses of local variable slots
2554 are at negative offsets from the frame pointer.
2557 @defmac ARGS_GROW_DOWNWARD
2558 Define this macro if successive arguments to a function occupy decreasing
2559 addresses on the stack.
2562 @defmac STARTING_FRAME_OFFSET
2563 Offset from the frame pointer to the first local variable slot to be allocated.
2565 If @code{FRAME_GROWS_DOWNWARD}, find the next slot's offset by
2566 subtracting the first slot's length from @code{STARTING_FRAME_OFFSET}.
2567 Otherwise, it is found by adding the length of the first slot to the
2568 value @code{STARTING_FRAME_OFFSET}.
2569 @c i'm not sure if the above is still correct.. had to change it to get
2570 @c rid of an overfull. --mew 2feb93
2573 @defmac STACK_ALIGNMENT_NEEDED
2574 Define to zero to disable final alignment of the stack during reload.
2575 The nonzero default for this macro is suitable for most ports.
2577 On ports where @code{STARTING_FRAME_OFFSET} is nonzero or where there
2578 is a register save block following the local block that doesn't require
2579 alignment to @code{STACK_BOUNDARY}, it may be beneficial to disable
2580 stack alignment and do it in the backend.
2583 @defmac STACK_POINTER_OFFSET
2584 Offset from the stack pointer register to the first location at which
2585 outgoing arguments are placed. If not specified, the default value of
2586 zero is used. This is the proper value for most machines.
2588 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
2589 the first location at which outgoing arguments are placed.
2592 @defmac FIRST_PARM_OFFSET (@var{fundecl})
2593 Offset from the argument pointer register to the first argument's
2594 address. On some machines it may depend on the data type of the
2597 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
2598 the first argument's address.
2601 @defmac STACK_DYNAMIC_OFFSET (@var{fundecl})
2602 Offset from the stack pointer register to an item dynamically allocated
2603 on the stack, e.g., by @code{alloca}.
2605 The default value for this macro is @code{STACK_POINTER_OFFSET} plus the
2606 length of the outgoing arguments. The default is correct for most
2607 machines. See @file{function.c} for details.
2610 @defmac INITIAL_FRAME_ADDRESS_RTX
2611 A C expression whose value is RTL representing the address of the initial
2612 stack frame. This address is passed to @code{RETURN_ADDR_RTX} and
2613 @code{DYNAMIC_CHAIN_ADDRESS}. If you don't define this macro, a reasonable
2614 default value will be used. Define this macro in order to make frame pointer
2615 elimination work in the presence of @code{__builtin_frame_address (count)} and
2616 @code{__builtin_return_address (count)} for @code{count} not equal to zero.
2619 @defmac DYNAMIC_CHAIN_ADDRESS (@var{frameaddr})
2620 A C expression whose value is RTL representing the address in a stack
2621 frame where the pointer to the caller's frame is stored. Assume that
2622 @var{frameaddr} is an RTL expression for the address of the stack frame
2625 If you don't define this macro, the default is to return the value
2626 of @var{frameaddr}---that is, the stack frame address is also the
2627 address of the stack word that points to the previous frame.
2630 @defmac SETUP_FRAME_ADDRESSES
2631 If defined, a C expression that produces the machine-specific code to
2632 setup the stack so that arbitrary frames can be accessed. For example,
2633 on the SPARC, we must flush all of the register windows to the stack
2634 before we can access arbitrary stack frames. You will seldom need to
2638 @hook TARGET_BUILTIN_SETJMP_FRAME_VALUE
2640 @defmac FRAME_ADDR_RTX (@var{frameaddr})
2641 A C expression whose value is RTL representing the value of the frame
2642 address for the current frame. @var{frameaddr} is the frame pointer
2643 of the current frame. This is used for __builtin_frame_address.
2644 You need only define this macro if the frame address is not the same
2645 as the frame pointer. Most machines do not need to define it.
2648 @defmac RETURN_ADDR_RTX (@var{count}, @var{frameaddr})
2649 A C expression whose value is RTL representing the value of the return
2650 address for the frame @var{count} steps up from the current frame, after
2651 the prologue. @var{frameaddr} is the frame pointer of the @var{count}
2652 frame, or the frame pointer of the @var{count} @minus{} 1 frame if
2653 @code{RETURN_ADDR_IN_PREVIOUS_FRAME} is nonzero.
2655 The value of the expression must always be the correct address when
2656 @var{count} is zero, but may be @code{NULL_RTX} if there is no way to
2657 determine the return address of other frames.
2660 @defmac RETURN_ADDR_IN_PREVIOUS_FRAME
2661 Define this macro to nonzero value if the return address of a particular
2662 stack frame is accessed from the frame pointer of the previous stack
2663 frame. The zero default for this macro is suitable for most ports.
2666 @defmac INCOMING_RETURN_ADDR_RTX
2667 A C expression whose value is RTL representing the location of the
2668 incoming return address at the beginning of any function, before the
2669 prologue. This RTL is either a @code{REG}, indicating that the return
2670 value is saved in @samp{REG}, or a @code{MEM} representing a location in
2673 You only need to define this macro if you want to support call frame
2674 debugging information like that provided by DWARF 2.
2676 If this RTL is a @code{REG}, you should also define
2677 @code{DWARF_FRAME_RETURN_COLUMN} to @code{DWARF_FRAME_REGNUM (REGNO)}.
2680 @defmac DWARF_ALT_FRAME_RETURN_COLUMN
2681 A C expression whose value is an integer giving a DWARF 2 column
2682 number that may be used as an alternative return column. The column
2683 must not correspond to any gcc hard register (that is, it must not
2684 be in the range of @code{DWARF_FRAME_REGNUM}).
2686 This macro can be useful if @code{DWARF_FRAME_RETURN_COLUMN} is set to a
2687 general register, but an alternative column needs to be used for signal
2688 frames. Some targets have also used different frame return columns
2692 @defmac DWARF_ZERO_REG
2693 A C expression whose value is an integer giving a DWARF 2 register
2694 number that is considered to always have the value zero. This should
2695 only be defined if the target has an architected zero register, and
2696 someone decided it was a good idea to use that register number to
2697 terminate the stack backtrace. New ports should avoid this.
2700 @hook TARGET_DWARF_HANDLE_FRAME_UNSPEC
2702 @defmac INCOMING_FRAME_SP_OFFSET
2703 A C expression whose value is an integer giving the offset, in bytes,
2704 from the value of the stack pointer register to the top of the stack
2705 frame at the beginning of any function, before the prologue. The top of
2706 the frame is defined to be the value of the stack pointer in the
2707 previous frame, just before the call instruction.
2709 You only need to define this macro if you want to support call frame
2710 debugging information like that provided by DWARF 2.
2713 @defmac ARG_POINTER_CFA_OFFSET (@var{fundecl})
2714 A C expression whose value is an integer giving the offset, in bytes,
2715 from the argument pointer to the canonical frame address (cfa). The
2716 final value should coincide with that calculated by
2717 @code{INCOMING_FRAME_SP_OFFSET}. Which is unfortunately not usable
2718 during virtual register instantiation.
2720 The default value for this macro is
2721 @code{FIRST_PARM_OFFSET (fundecl) + crtl->args.pretend_args_size},
2722 which is correct for most machines; in general, the arguments are found
2723 immediately before the stack frame. Note that this is not the case on
2724 some targets that save registers into the caller's frame, such as SPARC
2725 and rs6000, and so such targets need to define this macro.
2727 You only need to define this macro if the default is incorrect, and you
2728 want to support call frame debugging information like that provided by
2732 @defmac FRAME_POINTER_CFA_OFFSET (@var{fundecl})
2733 If defined, a C expression whose value is an integer giving the offset
2734 in bytes from the frame pointer to the canonical frame address (cfa).
2735 The final value should coincide with that calculated by
2736 @code{INCOMING_FRAME_SP_OFFSET}.
2738 Normally the CFA is calculated as an offset from the argument pointer,
2739 via @code{ARG_POINTER_CFA_OFFSET}, but if the argument pointer is
2740 variable due to the ABI, this may not be possible. If this macro is
2741 defined, it implies that the virtual register instantiation should be
2742 based on the frame pointer instead of the argument pointer. Only one
2743 of @code{FRAME_POINTER_CFA_OFFSET} and @code{ARG_POINTER_CFA_OFFSET}
2747 @defmac CFA_FRAME_BASE_OFFSET (@var{fundecl})
2748 If defined, a C expression whose value is an integer giving the offset
2749 in bytes from the canonical frame address (cfa) to the frame base used
2750 in DWARF 2 debug information. The default is zero. A different value
2751 may reduce the size of debug information on some ports.
2754 @node Exception Handling
2755 @subsection Exception Handling Support
2756 @cindex exception handling
2758 @defmac EH_RETURN_DATA_REGNO (@var{N})
2759 A C expression whose value is the @var{N}th register number used for
2760 data by exception handlers, or @code{INVALID_REGNUM} if fewer than
2761 @var{N} registers are usable.
2763 The exception handling library routines communicate with the exception
2764 handlers via a set of agreed upon registers. Ideally these registers
2765 should be call-clobbered; it is possible to use call-saved registers,
2766 but may negatively impact code size. The target must support at least
2767 2 data registers, but should define 4 if there are enough free registers.
2769 You must define this macro if you want to support call frame exception
2770 handling like that provided by DWARF 2.
2773 @defmac EH_RETURN_STACKADJ_RTX
2774 A C expression whose value is RTL representing a location in which
2775 to store a stack adjustment to be applied before function return.
2776 This is used to unwind the stack to an exception handler's call frame.
2777 It will be assigned zero on code paths that return normally.
2779 Typically this is a call-clobbered hard register that is otherwise
2780 untouched by the epilogue, but could also be a stack slot.
2782 Do not define this macro if the stack pointer is saved and restored
2783 by the regular prolog and epilog code in the call frame itself; in
2784 this case, the exception handling library routines will update the
2785 stack location to be restored in place. Otherwise, you must define
2786 this macro if you want to support call frame exception handling like
2787 that provided by DWARF 2.
2790 @defmac EH_RETURN_HANDLER_RTX
2791 A C expression whose value is RTL representing a location in which
2792 to store the address of an exception handler to which we should
2793 return. It will not be assigned on code paths that return normally.
2795 Typically this is the location in the call frame at which the normal
2796 return address is stored. For targets that return by popping an
2797 address off the stack, this might be a memory address just below
2798 the @emph{target} call frame rather than inside the current call
2799 frame. If defined, @code{EH_RETURN_STACKADJ_RTX} will have already
2800 been assigned, so it may be used to calculate the location of the
2803 Some targets have more complex requirements than storing to an
2804 address calculable during initial code generation. In that case
2805 the @code{eh_return} instruction pattern should be used instead.
2807 If you want to support call frame exception handling, you must
2808 define either this macro or the @code{eh_return} instruction pattern.
2811 @defmac RETURN_ADDR_OFFSET
2812 If defined, an integer-valued C expression for which rtl will be generated
2813 to add it to the exception handler address before it is searched in the
2814 exception handling tables, and to subtract it again from the address before
2815 using it to return to the exception handler.
2818 @defmac ASM_PREFERRED_EH_DATA_FORMAT (@var{code}, @var{global})
2819 This macro chooses the encoding of pointers embedded in the exception
2820 handling sections. If at all possible, this should be defined such
2821 that the exception handling section will not require dynamic relocations,
2822 and so may be read-only.
2824 @var{code} is 0 for data, 1 for code labels, 2 for function pointers.
2825 @var{global} is true if the symbol may be affected by dynamic relocations.
2826 The macro should return a combination of the @code{DW_EH_PE_*} defines
2827 as found in @file{dwarf2.h}.
2829 If this macro is not defined, pointers will not be encoded but
2830 represented directly.
2833 @defmac ASM_MAYBE_OUTPUT_ENCODED_ADDR_RTX (@var{file}, @var{encoding}, @var{size}, @var{addr}, @var{done})
2834 This macro allows the target to emit whatever special magic is required
2835 to represent the encoding chosen by @code{ASM_PREFERRED_EH_DATA_FORMAT}.
2836 Generic code takes care of pc-relative and indirect encodings; this must
2837 be defined if the target uses text-relative or data-relative encodings.
2839 This is a C statement that branches to @var{done} if the format was
2840 handled. @var{encoding} is the format chosen, @var{size} is the number
2841 of bytes that the format occupies, @var{addr} is the @code{SYMBOL_REF}
2845 @defmac MD_FALLBACK_FRAME_STATE_FOR (@var{context}, @var{fs})
2846 This macro allows the target to add CPU and operating system specific
2847 code to the call-frame unwinder for use when there is no unwind data
2848 available. The most common reason to implement this macro is to unwind
2849 through signal frames.
2851 This macro is called from @code{uw_frame_state_for} in
2852 @file{unwind-dw2.c}, @file{unwind-dw2-xtensa.c} and
2853 @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context};
2854 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{context->ra}
2855 for the address of the code being executed and @code{context->cfa} for
2856 the stack pointer value. If the frame can be decoded, the register
2857 save addresses should be updated in @var{fs} and the macro should
2858 evaluate to @code{_URC_NO_REASON}. If the frame cannot be decoded,
2859 the macro should evaluate to @code{_URC_END_OF_STACK}.
2861 For proper signal handling in Java this macro is accompanied by
2862 @code{MAKE_THROW_FRAME}, defined in @file{libjava/include/*-signal.h} headers.
2865 @defmac MD_HANDLE_UNWABI (@var{context}, @var{fs})
2866 This macro allows the target to add operating system specific code to the
2867 call-frame unwinder to handle the IA-64 @code{.unwabi} unwinding directive,
2868 usually used for signal or interrupt frames.
2870 This macro is called from @code{uw_update_context} in libgcc's
2871 @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context};
2872 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{fs->unwabi}
2873 for the abi and context in the @code{.unwabi} directive. If the
2874 @code{.unwabi} directive can be handled, the register save addresses should
2875 be updated in @var{fs}.
2878 @defmac TARGET_USES_WEAK_UNWIND_INFO
2879 A C expression that evaluates to true if the target requires unwind
2880 info to be given comdat linkage. Define it to be @code{1} if comdat
2881 linkage is necessary. The default is @code{0}.
2884 @node Stack Checking
2885 @subsection Specifying How Stack Checking is Done
2887 GCC will check that stack references are within the boundaries of the
2888 stack, if the option @option{-fstack-check} is specified, in one of
2893 If the value of the @code{STACK_CHECK_BUILTIN} macro is nonzero, GCC
2894 will assume that you have arranged for full stack checking to be done
2895 at appropriate places in the configuration files. GCC will not do
2896 other special processing.
2899 If @code{STACK_CHECK_BUILTIN} is zero and the value of the
2900 @code{STACK_CHECK_STATIC_BUILTIN} macro is nonzero, GCC will assume
2901 that you have arranged for static stack checking (checking of the
2902 static stack frame of functions) to be done at appropriate places
2903 in the configuration files. GCC will only emit code to do dynamic
2904 stack checking (checking on dynamic stack allocations) using the third
2908 If neither of the above are true, GCC will generate code to periodically
2909 ``probe'' the stack pointer using the values of the macros defined below.
2912 If neither STACK_CHECK_BUILTIN nor STACK_CHECK_STATIC_BUILTIN is defined,
2913 GCC will change its allocation strategy for large objects if the option
2914 @option{-fstack-check} is specified: they will always be allocated
2915 dynamically if their size exceeds @code{STACK_CHECK_MAX_VAR_SIZE} bytes.
2917 @defmac STACK_CHECK_BUILTIN
2918 A nonzero value if stack checking is done by the configuration files in a
2919 machine-dependent manner. You should define this macro if stack checking
2920 is required by the ABI of your machine or if you would like to do stack
2921 checking in some more efficient way than the generic approach. The default
2922 value of this macro is zero.
2925 @defmac STACK_CHECK_STATIC_BUILTIN
2926 A nonzero value if static stack checking is done by the configuration files
2927 in a machine-dependent manner. You should define this macro if you would
2928 like to do static stack checking in some more efficient way than the generic
2929 approach. The default value of this macro is zero.
2932 @defmac STACK_CHECK_PROBE_INTERVAL_EXP
2933 An integer specifying the interval at which GCC must generate stack probe
2934 instructions, defined as 2 raised to this integer. You will normally
2935 define this macro so that the interval be no larger than the size of
2936 the ``guard pages'' at the end of a stack area. The default value
2937 of 12 (4096-byte interval) is suitable for most systems.
2940 @defmac STACK_CHECK_MOVING_SP
2941 An integer which is nonzero if GCC should move the stack pointer page by page
2942 when doing probes. This can be necessary on systems where the stack pointer
2943 contains the bottom address of the memory area accessible to the executing
2944 thread at any point in time. In this situation an alternate signal stack
2945 is required in order to be able to recover from a stack overflow. The
2946 default value of this macro is zero.
2949 @defmac STACK_CHECK_PROTECT
2950 The number of bytes of stack needed to recover from a stack overflow, for
2951 languages where such a recovery is supported. The default value of 75 words
2952 with the @code{setjmp}/@code{longjmp}-based exception handling mechanism and
2953 8192 bytes with other exception handling mechanisms should be adequate for
2957 The following macros are relevant only if neither STACK_CHECK_BUILTIN
2958 nor STACK_CHECK_STATIC_BUILTIN is defined; you can omit them altogether
2959 in the opposite case.
2961 @defmac STACK_CHECK_MAX_FRAME_SIZE
2962 The maximum size of a stack frame, in bytes. GCC will generate probe
2963 instructions in non-leaf functions to ensure at least this many bytes of
2964 stack are available. If a stack frame is larger than this size, stack
2965 checking will not be reliable and GCC will issue a warning. The
2966 default is chosen so that GCC only generates one instruction on most
2967 systems. You should normally not change the default value of this macro.
2970 @defmac STACK_CHECK_FIXED_FRAME_SIZE
2971 GCC uses this value to generate the above warning message. It
2972 represents the amount of fixed frame used by a function, not including
2973 space for any callee-saved registers, temporaries and user variables.
2974 You need only specify an upper bound for this amount and will normally
2975 use the default of four words.
2978 @defmac STACK_CHECK_MAX_VAR_SIZE
2979 The maximum size, in bytes, of an object that GCC will place in the
2980 fixed area of the stack frame when the user specifies
2981 @option{-fstack-check}.
2982 GCC computed the default from the values of the above macros and you will
2983 normally not need to override that default.
2987 @node Frame Registers
2988 @subsection Registers That Address the Stack Frame
2990 @c prevent bad page break with this line
2991 This discusses registers that address the stack frame.
2993 @defmac STACK_POINTER_REGNUM
2994 The register number of the stack pointer register, which must also be a
2995 fixed register according to @code{FIXED_REGISTERS}. On most machines,
2996 the hardware determines which register this is.
2999 @defmac FRAME_POINTER_REGNUM
3000 The register number of the frame pointer register, which is used to
3001 access automatic variables in the stack frame. On some machines, the
3002 hardware determines which register this is. On other machines, you can
3003 choose any register you wish for this purpose.
3006 @defmac HARD_FRAME_POINTER_REGNUM
3007 On some machines the offset between the frame pointer and starting
3008 offset of the automatic variables is not known until after register
3009 allocation has been done (for example, because the saved registers are
3010 between these two locations). On those machines, define
3011 @code{FRAME_POINTER_REGNUM} the number of a special, fixed register to
3012 be used internally until the offset is known, and define
3013 @code{HARD_FRAME_POINTER_REGNUM} to be the actual hard register number
3014 used for the frame pointer.
3016 You should define this macro only in the very rare circumstances when it
3017 is not possible to calculate the offset between the frame pointer and
3018 the automatic variables until after register allocation has been
3019 completed. When this macro is defined, you must also indicate in your
3020 definition of @code{ELIMINABLE_REGS} how to eliminate
3021 @code{FRAME_POINTER_REGNUM} into either @code{HARD_FRAME_POINTER_REGNUM}
3022 or @code{STACK_POINTER_REGNUM}.
3024 Do not define this macro if it would be the same as
3025 @code{FRAME_POINTER_REGNUM}.
3028 @defmac ARG_POINTER_REGNUM
3029 The register number of the arg pointer register, which is used to access
3030 the function's argument list. On some machines, this is the same as the
3031 frame pointer register. On some machines, the hardware determines which
3032 register this is. On other machines, you can choose any register you
3033 wish for this purpose. If this is not the same register as the frame
3034 pointer register, then you must mark it as a fixed register according to
3035 @code{FIXED_REGISTERS}, or arrange to be able to eliminate it
3036 (@pxref{Elimination}).
3039 @defmac HARD_FRAME_POINTER_IS_FRAME_POINTER
3040 Define this to a preprocessor constant that is nonzero if
3041 @code{hard_frame_pointer_rtx} and @code{frame_pointer_rtx} should be
3042 the same. The default definition is @samp{(HARD_FRAME_POINTER_REGNUM
3043 == FRAME_POINTER_REGNUM)}; you only need to define this macro if that
3044 definition is not suitable for use in preprocessor conditionals.
3047 @defmac HARD_FRAME_POINTER_IS_ARG_POINTER
3048 Define this to a preprocessor constant that is nonzero if
3049 @code{hard_frame_pointer_rtx} and @code{arg_pointer_rtx} should be the
3050 same. The default definition is @samp{(HARD_FRAME_POINTER_REGNUM ==
3051 ARG_POINTER_REGNUM)}; you only need to define this macro if that
3052 definition is not suitable for use in preprocessor conditionals.
3055 @defmac RETURN_ADDRESS_POINTER_REGNUM
3056 The register number of the return address pointer register, which is used to
3057 access the current function's return address from the stack. On some
3058 machines, the return address is not at a fixed offset from the frame
3059 pointer or stack pointer or argument pointer. This register can be defined
3060 to point to the return address on the stack, and then be converted by
3061 @code{ELIMINABLE_REGS} into either the frame pointer or stack pointer.
3063 Do not define this macro unless there is no other way to get the return
3064 address from the stack.
3067 @defmac STATIC_CHAIN_REGNUM
3068 @defmacx STATIC_CHAIN_INCOMING_REGNUM
3069 Register numbers used for passing a function's static chain pointer. If
3070 register windows are used, the register number as seen by the called
3071 function is @code{STATIC_CHAIN_INCOMING_REGNUM}, while the register
3072 number as seen by the calling function is @code{STATIC_CHAIN_REGNUM}. If
3073 these registers are the same, @code{STATIC_CHAIN_INCOMING_REGNUM} need
3076 The static chain register need not be a fixed register.
3078 If the static chain is passed in memory, these macros should not be
3079 defined; instead, the @code{TARGET_STATIC_CHAIN} hook should be used.
3082 @hook TARGET_STATIC_CHAIN
3084 @defmac DWARF_FRAME_REGISTERS
3085 This macro specifies the maximum number of hard registers that can be
3086 saved in a call frame. This is used to size data structures used in
3087 DWARF2 exception handling.
3089 Prior to GCC 3.0, this macro was needed in order to establish a stable
3090 exception handling ABI in the face of adding new hard registers for ISA
3091 extensions. In GCC 3.0 and later, the EH ABI is insulated from changes
3092 in the number of hard registers. Nevertheless, this macro can still be
3093 used to reduce the runtime memory requirements of the exception handling
3094 routines, which can be substantial if the ISA contains a lot of
3095 registers that are not call-saved.
3097 If this macro is not defined, it defaults to
3098 @code{FIRST_PSEUDO_REGISTER}.
3101 @defmac PRE_GCC3_DWARF_FRAME_REGISTERS
3103 This macro is similar to @code{DWARF_FRAME_REGISTERS}, but is provided
3104 for backward compatibility in pre GCC 3.0 compiled code.
3106 If this macro is not defined, it defaults to
3107 @code{DWARF_FRAME_REGISTERS}.
3110 @defmac DWARF_REG_TO_UNWIND_COLUMN (@var{regno})
3112 Define this macro if the target's representation for dwarf registers
3113 is different than the internal representation for unwind column.
3114 Given a dwarf register, this macro should return the internal unwind
3115 column number to use instead.
3117 See the PowerPC's SPE target for an example.
3120 @defmac DWARF_FRAME_REGNUM (@var{regno})
3122 Define this macro if the target's representation for dwarf registers
3123 used in .eh_frame or .debug_frame is different from that used in other
3124 debug info sections. Given a GCC hard register number, this macro
3125 should return the .eh_frame register number. The default is
3126 @code{DBX_REGISTER_NUMBER (@var{regno})}.
3130 @defmac DWARF2_FRAME_REG_OUT (@var{regno}, @var{for_eh})
3132 Define this macro to map register numbers held in the call frame info
3133 that GCC has collected using @code{DWARF_FRAME_REGNUM} to those that
3134 should be output in .debug_frame (@code{@var{for_eh}} is zero) and
3135 .eh_frame (@code{@var{for_eh}} is nonzero). The default is to
3136 return @code{@var{regno}}.
3140 @defmac REG_VALUE_IN_UNWIND_CONTEXT
3142 Define this macro if the target stores register values as
3143 @code{_Unwind_Word} type in unwind context. It should be defined if
3144 target register size is larger than the size of @code{void *}. The
3145 default is to store register values as @code{void *} type.
3149 @defmac ASSUME_EXTENDED_UNWIND_CONTEXT
3151 Define this macro to be 1 if the target always uses extended unwind
3152 context with version, args_size and by_value fields. If it is undefined,
3153 it will be defined to 1 when @code{REG_VALUE_IN_UNWIND_CONTEXT} is
3154 defined and 0 otherwise.
3159 @subsection Eliminating Frame Pointer and Arg Pointer
3161 @c prevent bad page break with this line
3162 This is about eliminating the frame pointer and arg pointer.
3164 @hook TARGET_FRAME_POINTER_REQUIRED
3166 @findex get_frame_size
3167 @defmac INITIAL_FRAME_POINTER_OFFSET (@var{depth-var})
3168 A C statement to store in the variable @var{depth-var} the difference
3169 between the frame pointer and the stack pointer values immediately after
3170 the function prologue. The value would be computed from information
3171 such as the result of @code{get_frame_size ()} and the tables of
3172 registers @code{regs_ever_live} and @code{call_used_regs}.
3174 If @code{ELIMINABLE_REGS} is defined, this macro will be not be used and
3175 need not be defined. Otherwise, it must be defined even if
3176 @code{TARGET_FRAME_POINTER_REQUIRED} always returns true; in that
3177 case, you may set @var{depth-var} to anything.
3180 @defmac ELIMINABLE_REGS
3181 If defined, this macro specifies a table of register pairs used to
3182 eliminate unneeded registers that point into the stack frame. If it is not
3183 defined, the only elimination attempted by the compiler is to replace
3184 references to the frame pointer with references to the stack pointer.
3186 The definition of this macro is a list of structure initializations, each
3187 of which specifies an original and replacement register.
3189 On some machines, the position of the argument pointer is not known until
3190 the compilation is completed. In such a case, a separate hard register
3191 must be used for the argument pointer. This register can be eliminated by
3192 replacing it with either the frame pointer or the argument pointer,
3193 depending on whether or not the frame pointer has been eliminated.
3195 In this case, you might specify:
3197 #define ELIMINABLE_REGS \
3198 @{@{ARG_POINTER_REGNUM, STACK_POINTER_REGNUM@}, \
3199 @{ARG_POINTER_REGNUM, FRAME_POINTER_REGNUM@}, \
3200 @{FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM@}@}
3203 Note that the elimination of the argument pointer with the stack pointer is
3204 specified first since that is the preferred elimination.
3207 @hook TARGET_CAN_ELIMINATE
3209 @defmac INITIAL_ELIMINATION_OFFSET (@var{from-reg}, @var{to-reg}, @var{offset-var})
3210 This macro is similar to @code{INITIAL_FRAME_POINTER_OFFSET}. It
3211 specifies the initial difference between the specified pair of
3212 registers. This macro must be defined if @code{ELIMINABLE_REGS} is
3216 @node Stack Arguments
3217 @subsection Passing Function Arguments on the Stack
3218 @cindex arguments on stack
3219 @cindex stack arguments
3221 The macros in this section control how arguments are passed
3222 on the stack. See the following section for other macros that
3223 control passing certain arguments in registers.
3225 @hook TARGET_PROMOTE_PROTOTYPES
3228 A C expression. If nonzero, push insns will be used to pass
3230 If the target machine does not have a push instruction, set it to zero.
3231 That directs GCC to use an alternate strategy: to
3232 allocate the entire argument block and then store the arguments into
3233 it. When @code{PUSH_ARGS} is nonzero, @code{PUSH_ROUNDING} must be defined too.
3236 @defmac PUSH_ARGS_REVERSED
3237 A C expression. If nonzero, function arguments will be evaluated from
3238 last to first, rather than from first to last. If this macro is not
3239 defined, it defaults to @code{PUSH_ARGS} on targets where the stack
3240 and args grow in opposite directions, and 0 otherwise.
3243 @defmac PUSH_ROUNDING (@var{npushed})
3244 A C expression that is the number of bytes actually pushed onto the
3245 stack when an instruction attempts to push @var{npushed} bytes.
3247 On some machines, the definition
3250 #define PUSH_ROUNDING(BYTES) (BYTES)
3254 will suffice. But on other machines, instructions that appear
3255 to push one byte actually push two bytes in an attempt to maintain
3256 alignment. Then the definition should be
3259 #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1)
3262 If the value of this macro has a type, it should be an unsigned type.
3265 @findex outgoing_args_size
3266 @findex crtl->outgoing_args_size
3267 @defmac ACCUMULATE_OUTGOING_ARGS
3268 A C expression. If nonzero, the maximum amount of space required for outgoing arguments
3269 will be computed and placed into
3270 @code{crtl->outgoing_args_size}. No space will be pushed
3271 onto the stack for each call; instead, the function prologue should
3272 increase the stack frame size by this amount.
3274 Setting both @code{PUSH_ARGS} and @code{ACCUMULATE_OUTGOING_ARGS}
3278 @defmac REG_PARM_STACK_SPACE (@var{fndecl})
3279 Define this macro if functions should assume that stack space has been
3280 allocated for arguments even when their values are passed in
3283 The value of this macro is the size, in bytes, of the area reserved for
3284 arguments passed in registers for the function represented by @var{fndecl},
3285 which can be zero if GCC is calling a library function.
3286 The argument @var{fndecl} can be the FUNCTION_DECL, or the type itself
3289 This space can be allocated by the caller, or be a part of the
3290 machine-dependent stack frame: @code{OUTGOING_REG_PARM_STACK_SPACE} says
3293 @c above is overfull. not sure what to do. --mew 5feb93 did
3294 @c something, not sure if it looks good. --mew 10feb93
3296 @defmac INCOMING_REG_PARM_STACK_SPACE (@var{fndecl})
3297 Like @code{REG_PARM_STACK_SPACE}, but for incoming register arguments.
3298 Define this macro if space guaranteed when compiling a function body
3299 is different to space required when making a call, a situation that
3300 can arise with K&R style function definitions.
3303 @defmac OUTGOING_REG_PARM_STACK_SPACE (@var{fntype})
3304 Define this to a nonzero value if it is the responsibility of the
3305 caller to allocate the area reserved for arguments passed in registers
3306 when calling a function of @var{fntype}. @var{fntype} may be NULL
3307 if the function called is a library function.
3309 If @code{ACCUMULATE_OUTGOING_ARGS} is defined, this macro controls
3310 whether the space for these arguments counts in the value of
3311 @code{crtl->outgoing_args_size}.
3314 @defmac STACK_PARMS_IN_REG_PARM_AREA
3315 Define this macro if @code{REG_PARM_STACK_SPACE} is defined, but the
3316 stack parameters don't skip the area specified by it.
3317 @c i changed this, makes more sens and it should have taken care of the
3318 @c overfull.. not as specific, tho. --mew 5feb93
3320 Normally, when a parameter is not passed in registers, it is placed on the
3321 stack beyond the @code{REG_PARM_STACK_SPACE} area. Defining this macro
3322 suppresses this behavior and causes the parameter to be passed on the
3323 stack in its natural location.
3326 @hook TARGET_RETURN_POPS_ARGS
3328 @defmac CALL_POPS_ARGS (@var{cum})
3329 A C expression that should indicate the number of bytes a call sequence
3330 pops off the stack. It is added to the value of @code{RETURN_POPS_ARGS}
3331 when compiling a function call.
3333 @var{cum} is the variable in which all arguments to the called function
3334 have been accumulated.
3336 On certain architectures, such as the SH5, a call trampoline is used
3337 that pops certain registers off the stack, depending on the arguments
3338 that have been passed to the function. Since this is a property of the
3339 call site, not of the called function, @code{RETURN_POPS_ARGS} is not
3343 @node Register Arguments
3344 @subsection Passing Arguments in Registers
3345 @cindex arguments in registers
3346 @cindex registers arguments
3348 This section describes the macros which let you control how various
3349 types of arguments are passed in registers or how they are arranged in
3352 @hook TARGET_FUNCTION_ARG
3354 @hook TARGET_MUST_PASS_IN_STACK
3356 @hook TARGET_FUNCTION_INCOMING_ARG
3358 @hook TARGET_USE_PSEUDO_PIC_REG
3360 @hook TARGET_INIT_PIC_REG
3362 @hook TARGET_ARG_PARTIAL_BYTES
3364 @hook TARGET_PASS_BY_REFERENCE
3366 @hook TARGET_CALLEE_COPIES
3368 @defmac CUMULATIVE_ARGS
3369 A C type for declaring a variable that is used as the first argument
3370 of @code{TARGET_FUNCTION_ARG} and other related values. For some
3371 target machines, the type @code{int} suffices and can hold the number
3372 of bytes of argument so far.
3374 There is no need to record in @code{CUMULATIVE_ARGS} anything about the
3375 arguments that have been passed on the stack. The compiler has other
3376 variables to keep track of that. For target machines on which all
3377 arguments are passed on the stack, there is no need to store anything in
3378 @code{CUMULATIVE_ARGS}; however, the data structure must exist and
3379 should not be empty, so use @code{int}.
3382 @defmac OVERRIDE_ABI_FORMAT (@var{fndecl})
3383 If defined, this macro is called before generating any code for a
3384 function, but after the @var{cfun} descriptor for the function has been
3385 created. The back end may use this macro to update @var{cfun} to
3386 reflect an ABI other than that which would normally be used by default.
3387 If the compiler is generating code for a compiler-generated function,
3388 @var{fndecl} may be @code{NULL}.
3391 @defmac INIT_CUMULATIVE_ARGS (@var{cum}, @var{fntype}, @var{libname}, @var{fndecl}, @var{n_named_args})
3392 A C statement (sans semicolon) for initializing the variable
3393 @var{cum} for the state at the beginning of the argument list. The
3394 variable has type @code{CUMULATIVE_ARGS}. The value of @var{fntype}
3395 is the tree node for the data type of the function which will receive
3396 the args, or 0 if the args are to a compiler support library function.
3397 For direct calls that are not libcalls, @var{fndecl} contain the
3398 declaration node of the function. @var{fndecl} is also set when
3399 @code{INIT_CUMULATIVE_ARGS} is used to find arguments for the function
3400 being compiled. @var{n_named_args} is set to the number of named
3401 arguments, including a structure return address if it is passed as a
3402 parameter, when making a call. When processing incoming arguments,
3403 @var{n_named_args} is set to @minus{}1.
3405 When processing a call to a compiler support library function,
3406 @var{libname} identifies which one. It is a @code{symbol_ref} rtx which
3407 contains the name of the function, as a string. @var{libname} is 0 when
3408 an ordinary C function call is being processed. Thus, each time this
3409 macro is called, either @var{libname} or @var{fntype} is nonzero, but
3410 never both of them at once.
3413 @defmac INIT_CUMULATIVE_LIBCALL_ARGS (@var{cum}, @var{mode}, @var{libname})
3414 Like @code{INIT_CUMULATIVE_ARGS} but only used for outgoing libcalls,
3415 it gets a @code{MODE} argument instead of @var{fntype}, that would be
3416 @code{NULL}. @var{indirect} would always be zero, too. If this macro
3417 is not defined, @code{INIT_CUMULATIVE_ARGS (cum, NULL_RTX, libname,
3418 0)} is used instead.
3421 @defmac INIT_CUMULATIVE_INCOMING_ARGS (@var{cum}, @var{fntype}, @var{libname})
3422 Like @code{INIT_CUMULATIVE_ARGS} but overrides it for the purposes of
3423 finding the arguments for the function being compiled. If this macro is
3424 undefined, @code{INIT_CUMULATIVE_ARGS} is used instead.
3426 The value passed for @var{libname} is always 0, since library routines
3427 with special calling conventions are never compiled with GCC@. The
3428 argument @var{libname} exists for symmetry with
3429 @code{INIT_CUMULATIVE_ARGS}.
3430 @c could use "this macro" in place of @code{INIT_CUMULATIVE_ARGS}, maybe.
3431 @c --mew 5feb93 i switched the order of the sentences. --mew 10feb93
3434 @hook TARGET_FUNCTION_ARG_ADVANCE
3436 @defmac FUNCTION_ARG_OFFSET (@var{mode}, @var{type})
3437 If defined, a C expression that is the number of bytes to add to the
3438 offset of the argument passed in memory. This is needed for the SPU,
3439 which passes @code{char} and @code{short} arguments in the preferred
3440 slot that is in the middle of the quad word instead of starting at the
3444 @defmac FUNCTION_ARG_PADDING (@var{mode}, @var{type})
3445 If defined, a C expression which determines whether, and in which direction,
3446 to pad out an argument with extra space. The value should be of type
3447 @code{enum direction}: either @code{upward} to pad above the argument,
3448 @code{downward} to pad below, or @code{none} to inhibit padding.
3450 The @emph{amount} of padding is not controlled by this macro, but by the
3451 target hook @code{TARGET_FUNCTION_ARG_ROUND_BOUNDARY}. It is
3452 always just enough to reach the next multiple of that boundary.
3454 This macro has a default definition which is right for most systems.
3455 For little-endian machines, the default is to pad upward. For
3456 big-endian machines, the default is to pad downward for an argument of
3457 constant size shorter than an @code{int}, and upward otherwise.
3460 @defmac PAD_VARARGS_DOWN
3461 If defined, a C expression which determines whether the default
3462 implementation of va_arg will attempt to pad down before reading the
3463 next argument, if that argument is smaller than its aligned space as
3464 controlled by @code{PARM_BOUNDARY}. If this macro is not defined, all such
3465 arguments are padded down if @code{BYTES_BIG_ENDIAN} is true.
3468 @defmac BLOCK_REG_PADDING (@var{mode}, @var{type}, @var{first})
3469 Specify padding for the last element of a block move between registers and
3470 memory. @var{first} is nonzero if this is the only element. Defining this
3471 macro allows better control of register function parameters on big-endian
3472 machines, without using @code{PARALLEL} rtl. In particular,
3473 @code{MUST_PASS_IN_STACK} need not test padding and mode of types in
3474 registers, as there is no longer a "wrong" part of a register; For example,
3475 a three byte aggregate may be passed in the high part of a register if so
3479 @hook TARGET_FUNCTION_ARG_BOUNDARY
3481 @hook TARGET_FUNCTION_ARG_ROUND_BOUNDARY
3483 @defmac FUNCTION_ARG_REGNO_P (@var{regno})
3484 A C expression that is nonzero if @var{regno} is the number of a hard
3485 register in which function arguments are sometimes passed. This does
3486 @emph{not} include implicit arguments such as the static chain and
3487 the structure-value address. On many machines, no registers can be
3488 used for this purpose since all function arguments are pushed on the
3492 @hook TARGET_SPLIT_COMPLEX_ARG
3494 @hook TARGET_BUILD_BUILTIN_VA_LIST
3496 @hook TARGET_ENUM_VA_LIST_P
3498 @hook TARGET_FN_ABI_VA_LIST
3500 @hook TARGET_CANONICAL_VA_LIST_TYPE
3502 @hook TARGET_GIMPLIFY_VA_ARG_EXPR
3504 @hook TARGET_VALID_POINTER_MODE
3506 @hook TARGET_REF_MAY_ALIAS_ERRNO
3508 @hook TARGET_SCALAR_MODE_SUPPORTED_P
3510 @hook TARGET_VECTOR_MODE_SUPPORTED_P
3512 @hook TARGET_ARRAY_MODE_SUPPORTED_P
3514 @hook TARGET_LIBGCC_FLOATING_MODE_SUPPORTED_P
3516 @hook TARGET_SMALL_REGISTER_CLASSES_FOR_MODE_P
3519 @subsection How Scalar Function Values Are Returned
3520 @cindex return values in registers
3521 @cindex values, returned by functions
3522 @cindex scalars, returned as values
3524 This section discusses the macros that control returning scalars as
3525 values---values that can fit in registers.
3527 @hook TARGET_FUNCTION_VALUE
3529 @defmac FUNCTION_VALUE (@var{valtype}, @var{func})
3530 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE} for
3531 a new target instead.
3534 @defmac LIBCALL_VALUE (@var{mode})
3535 A C expression to create an RTX representing the place where a library
3536 function returns a value of mode @var{mode}.
3538 Note that ``library function'' in this context means a compiler
3539 support routine, used to perform arithmetic, whose name is known
3540 specially by the compiler and was not mentioned in the C code being
3544 @hook TARGET_LIBCALL_VALUE
3546 @defmac FUNCTION_VALUE_REGNO_P (@var{regno})
3547 A C expression that is nonzero if @var{regno} is the number of a hard
3548 register in which the values of called function may come back.
3550 A register whose use for returning values is limited to serving as the
3551 second of a pair (for a value of type @code{double}, say) need not be
3552 recognized by this macro. So for most machines, this definition
3556 #define FUNCTION_VALUE_REGNO_P(N) ((N) == 0)
3559 If the machine has register windows, so that the caller and the called
3560 function use different registers for the return value, this macro
3561 should recognize only the caller's register numbers.
3563 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE_REGNO_P}
3564 for a new target instead.
3567 @hook TARGET_FUNCTION_VALUE_REGNO_P
3569 @defmac APPLY_RESULT_SIZE
3570 Define this macro if @samp{untyped_call} and @samp{untyped_return}
3571 need more space than is implied by @code{FUNCTION_VALUE_REGNO_P} for
3572 saving and restoring an arbitrary return value.
3575 @hook TARGET_OMIT_STRUCT_RETURN_REG
3577 @hook TARGET_RETURN_IN_MSB
3579 @node Aggregate Return
3580 @subsection How Large Values Are Returned
3581 @cindex aggregates as return values
3582 @cindex large return values
3583 @cindex returning aggregate values
3584 @cindex structure value address
3586 When a function value's mode is @code{BLKmode} (and in some other
3587 cases), the value is not returned according to
3588 @code{TARGET_FUNCTION_VALUE} (@pxref{Scalar Return}). Instead, the
3589 caller passes the address of a block of memory in which the value
3590 should be stored. This address is called the @dfn{structure value
3593 This section describes how to control returning structure values in
3596 @hook TARGET_RETURN_IN_MEMORY
3598 @defmac DEFAULT_PCC_STRUCT_RETURN
3599 Define this macro to be 1 if all structure and union return values must be
3600 in memory. Since this results in slower code, this should be defined
3601 only if needed for compatibility with other compilers or with an ABI@.
3602 If you define this macro to be 0, then the conventions used for structure
3603 and union return values are decided by the @code{TARGET_RETURN_IN_MEMORY}
3606 If not defined, this defaults to the value 1.
3609 @hook TARGET_STRUCT_VALUE_RTX
3611 @defmac PCC_STATIC_STRUCT_RETURN
3612 Define this macro if the usual system convention on the target machine
3613 for returning structures and unions is for the called function to return
3614 the address of a static variable containing the value.
3616 Do not define this if the usual system convention is for the caller to
3617 pass an address to the subroutine.
3619 This macro has effect in @option{-fpcc-struct-return} mode, but it does
3620 nothing when you use @option{-freg-struct-return} mode.
3623 @hook TARGET_GET_RAW_RESULT_MODE
3625 @hook TARGET_GET_RAW_ARG_MODE
3628 @subsection Caller-Saves Register Allocation
3630 If you enable it, GCC can save registers around function calls. This
3631 makes it possible to use call-clobbered registers to hold variables that
3632 must live across calls.
3634 @defmac HARD_REGNO_CALLER_SAVE_MODE (@var{regno}, @var{nregs})
3635 A C expression specifying which mode is required for saving @var{nregs}
3636 of a pseudo-register in call-clobbered hard register @var{regno}. If
3637 @var{regno} is unsuitable for caller save, @code{VOIDmode} should be
3638 returned. For most machines this macro need not be defined since GCC
3639 will select the smallest suitable mode.
3642 @node Function Entry
3643 @subsection Function Entry and Exit
3644 @cindex function entry and exit
3648 This section describes the macros that output function entry
3649 (@dfn{prologue}) and exit (@dfn{epilogue}) code.
3651 @hook TARGET_ASM_FUNCTION_PROLOGUE
3653 @hook TARGET_ASM_FUNCTION_END_PROLOGUE
3655 @hook TARGET_ASM_FUNCTION_BEGIN_EPILOGUE
3657 @hook TARGET_ASM_FUNCTION_EPILOGUE
3661 @findex pretend_args_size
3662 @findex crtl->args.pretend_args_size
3663 A region of @code{crtl->args.pretend_args_size} bytes of
3664 uninitialized space just underneath the first argument arriving on the
3665 stack. (This may not be at the very start of the allocated stack region
3666 if the calling sequence has pushed anything else since pushing the stack
3667 arguments. But usually, on such machines, nothing else has been pushed
3668 yet, because the function prologue itself does all the pushing.) This
3669 region is used on machines where an argument may be passed partly in
3670 registers and partly in memory, and, in some cases to support the
3671 features in @code{<stdarg.h>}.
3674 An area of memory used to save certain registers used by the function.
3675 The size of this area, which may also include space for such things as
3676 the return address and pointers to previous stack frames, is
3677 machine-specific and usually depends on which registers have been used
3678 in the function. Machines with register windows often do not require
3682 A region of at least @var{size} bytes, possibly rounded up to an allocation
3683 boundary, to contain the local variables of the function. On some machines,
3684 this region and the save area may occur in the opposite order, with the
3685 save area closer to the top of the stack.
3688 @cindex @code{ACCUMULATE_OUTGOING_ARGS} and stack frames
3689 Optionally, when @code{ACCUMULATE_OUTGOING_ARGS} is defined, a region of
3690 @code{crtl->outgoing_args_size} bytes to be used for outgoing
3691 argument lists of the function. @xref{Stack Arguments}.
3694 @defmac EXIT_IGNORE_STACK
3695 Define this macro as a C expression that is nonzero if the return
3696 instruction or the function epilogue ignores the value of the stack
3697 pointer; in other words, if it is safe to delete an instruction to
3698 adjust the stack pointer before a return from the function. The
3701 Note that this macro's value is relevant only for functions for which
3702 frame pointers are maintained. It is never safe to delete a final
3703 stack adjustment in a function that has no frame pointer, and the
3704 compiler knows this regardless of @code{EXIT_IGNORE_STACK}.
3707 @defmac EPILOGUE_USES (@var{regno})
3708 Define this macro as a C expression that is nonzero for registers that are
3709 used by the epilogue or the @samp{return} pattern. The stack and frame
3710 pointer registers are already assumed to be used as needed.
3713 @defmac EH_USES (@var{regno})
3714 Define this macro as a C expression that is nonzero for registers that are
3715 used by the exception handling mechanism, and so should be considered live
3716 on entry to an exception edge.
3719 @hook TARGET_ASM_OUTPUT_MI_THUNK
3721 @hook TARGET_ASM_CAN_OUTPUT_MI_THUNK
3724 @subsection Generating Code for Profiling
3725 @cindex profiling, code generation
3727 These macros will help you generate code for profiling.
3729 @defmac FUNCTION_PROFILER (@var{file}, @var{labelno})
3730 A C statement or compound statement to output to @var{file} some
3731 assembler code to call the profiling subroutine @code{mcount}.
3734 The details of how @code{mcount} expects to be called are determined by
3735 your operating system environment, not by GCC@. To figure them out,
3736 compile a small program for profiling using the system's installed C
3737 compiler and look at the assembler code that results.
3739 Older implementations of @code{mcount} expect the address of a counter
3740 variable to be loaded into some register. The name of this variable is
3741 @samp{LP} followed by the number @var{labelno}, so you would generate
3742 the name using @samp{LP%d} in a @code{fprintf}.
3745 @defmac PROFILE_HOOK
3746 A C statement or compound statement to output to @var{file} some assembly
3747 code to call the profiling subroutine @code{mcount} even the target does
3748 not support profiling.
3751 @defmac NO_PROFILE_COUNTERS
3752 Define this macro to be an expression with a nonzero value if the
3753 @code{mcount} subroutine on your system does not need a counter variable
3754 allocated for each function. This is true for almost all modern
3755 implementations. If you define this macro, you must not use the
3756 @var{labelno} argument to @code{FUNCTION_PROFILER}.
3759 @defmac PROFILE_BEFORE_PROLOGUE
3760 Define this macro if the code for function profiling should come before
3761 the function prologue. Normally, the profiling code comes after.
3764 @hook TARGET_KEEP_LEAF_WHEN_PROFILED
3767 @subsection Permitting tail calls
3770 @hook TARGET_FUNCTION_OK_FOR_SIBCALL
3772 @hook TARGET_EXTRA_LIVE_ON_ENTRY
3774 @hook TARGET_SET_UP_BY_PROLOGUE
3776 @hook TARGET_WARN_FUNC_RETURN
3778 @node Stack Smashing Protection
3779 @subsection Stack smashing protection
3780 @cindex stack smashing protection
3782 @hook TARGET_STACK_PROTECT_GUARD
3784 @hook TARGET_STACK_PROTECT_FAIL
3786 @hook TARGET_SUPPORTS_SPLIT_STACK
3788 @node Miscellaneous Register Hooks
3789 @subsection Miscellaneous register hooks
3790 @cindex miscellaneous register hooks
3792 @hook TARGET_CALL_FUSAGE_CONTAINS_NON_CALLEE_CLOBBERS
3795 @section Implementing the Varargs Macros
3796 @cindex varargs implementation
3798 GCC comes with an implementation of @code{<varargs.h>} and
3799 @code{<stdarg.h>} that work without change on machines that pass arguments
3800 on the stack. Other machines require their own implementations of
3801 varargs, and the two machine independent header files must have
3802 conditionals to include it.
3804 ISO @code{<stdarg.h>} differs from traditional @code{<varargs.h>} mainly in
3805 the calling convention for @code{va_start}. The traditional
3806 implementation takes just one argument, which is the variable in which
3807 to store the argument pointer. The ISO implementation of
3808 @code{va_start} takes an additional second argument. The user is
3809 supposed to write the last named argument of the function here.
3811 However, @code{va_start} should not use this argument. The way to find
3812 the end of the named arguments is with the built-in functions described
3815 @defmac __builtin_saveregs ()
3816 Use this built-in function to save the argument registers in memory so
3817 that the varargs mechanism can access them. Both ISO and traditional
3818 versions of @code{va_start} must use @code{__builtin_saveregs}, unless
3819 you use @code{TARGET_SETUP_INCOMING_VARARGS} (see below) instead.
3821 On some machines, @code{__builtin_saveregs} is open-coded under the
3822 control of the target hook @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. On
3823 other machines, it calls a routine written in assembler language,
3824 found in @file{libgcc2.c}.
3826 Code generated for the call to @code{__builtin_saveregs} appears at the
3827 beginning of the function, as opposed to where the call to
3828 @code{__builtin_saveregs} is written, regardless of what the code is.
3829 This is because the registers must be saved before the function starts
3830 to use them for its own purposes.
3831 @c i rewrote the first sentence above to fix an overfull hbox. --mew
3835 @defmac __builtin_next_arg (@var{lastarg})
3836 This builtin returns the address of the first anonymous stack
3837 argument, as type @code{void *}. If @code{ARGS_GROW_DOWNWARD}, it
3838 returns the address of the location above the first anonymous stack
3839 argument. Use it in @code{va_start} to initialize the pointer for
3840 fetching arguments from the stack. Also use it in @code{va_start} to
3841 verify that the second parameter @var{lastarg} is the last named argument
3842 of the current function.
3845 @defmac __builtin_classify_type (@var{object})
3846 Since each machine has its own conventions for which data types are
3847 passed in which kind of register, your implementation of @code{va_arg}
3848 has to embody these conventions. The easiest way to categorize the
3849 specified data type is to use @code{__builtin_classify_type} together
3850 with @code{sizeof} and @code{__alignof__}.
3852 @code{__builtin_classify_type} ignores the value of @var{object},
3853 considering only its data type. It returns an integer describing what
3854 kind of type that is---integer, floating, pointer, structure, and so on.
3856 The file @file{typeclass.h} defines an enumeration that you can use to
3857 interpret the values of @code{__builtin_classify_type}.
3860 These machine description macros help implement varargs:
3862 @hook TARGET_EXPAND_BUILTIN_SAVEREGS
3864 @hook TARGET_SETUP_INCOMING_VARARGS
3866 @hook TARGET_STRICT_ARGUMENT_NAMING
3868 @hook TARGET_CALL_ARGS
3870 @hook TARGET_END_CALL_ARGS
3872 @hook TARGET_PRETEND_OUTGOING_VARARGS_NAMED
3874 @hook TARGET_LOAD_BOUNDS_FOR_ARG
3876 @hook TARGET_STORE_BOUNDS_FOR_ARG
3878 @hook TARGET_LOAD_RETURNED_BOUNDS
3880 @hook TARGET_STORE_RETURNED_BOUNDS
3882 @hook TARGET_CHKP_FUNCTION_VALUE_BOUNDS
3884 @hook TARGET_SETUP_INCOMING_VARARG_BOUNDS
3887 @section Trampolines for Nested Functions
3888 @cindex trampolines for nested functions
3889 @cindex nested functions, trampolines for
3891 A @dfn{trampoline} is a small piece of code that is created at run time
3892 when the address of a nested function is taken. It normally resides on
3893 the stack, in the stack frame of the containing function. These macros
3894 tell GCC how to generate code to allocate and initialize a
3897 The instructions in the trampoline must do two things: load a constant
3898 address into the static chain register, and jump to the real address of
3899 the nested function. On CISC machines such as the m68k, this requires
3900 two instructions, a move immediate and a jump. Then the two addresses
3901 exist in the trampoline as word-long immediate operands. On RISC
3902 machines, it is often necessary to load each address into a register in
3903 two parts. Then pieces of each address form separate immediate
3906 The code generated to initialize the trampoline must store the variable
3907 parts---the static chain value and the function address---into the
3908 immediate operands of the instructions. On a CISC machine, this is
3909 simply a matter of copying each address to a memory reference at the
3910 proper offset from the start of the trampoline. On a RISC machine, it
3911 may be necessary to take out pieces of the address and store them
3914 @hook TARGET_ASM_TRAMPOLINE_TEMPLATE
3916 @defmac TRAMPOLINE_SECTION
3917 Return the section into which the trampoline template is to be placed
3918 (@pxref{Sections}). The default value is @code{readonly_data_section}.
3921 @defmac TRAMPOLINE_SIZE
3922 A C expression for the size in bytes of the trampoline, as an integer.
3925 @defmac TRAMPOLINE_ALIGNMENT
3926 Alignment required for trampolines, in bits.
3928 If you don't define this macro, the value of @code{FUNCTION_ALIGNMENT}
3929 is used for aligning trampolines.
3932 @hook TARGET_TRAMPOLINE_INIT
3934 @hook TARGET_TRAMPOLINE_ADJUST_ADDRESS
3936 Implementing trampolines is difficult on many machines because they have
3937 separate instruction and data caches. Writing into a stack location
3938 fails to clear the memory in the instruction cache, so when the program
3939 jumps to that location, it executes the old contents.
3941 Here are two possible solutions. One is to clear the relevant parts of
3942 the instruction cache whenever a trampoline is set up. The other is to
3943 make all trampolines identical, by having them jump to a standard
3944 subroutine. The former technique makes trampoline execution faster; the
3945 latter makes initialization faster.
3947 To clear the instruction cache when a trampoline is initialized, define
3948 the following macro.
3950 @defmac CLEAR_INSN_CACHE (@var{beg}, @var{end})
3951 If defined, expands to a C expression clearing the @emph{instruction
3952 cache} in the specified interval. The definition of this macro would
3953 typically be a series of @code{asm} statements. Both @var{beg} and
3954 @var{end} are both pointer expressions.
3957 To use a standard subroutine, define the following macro. In addition,
3958 you must make sure that the instructions in a trampoline fill an entire
3959 cache line with identical instructions, or else ensure that the
3960 beginning of the trampoline code is always aligned at the same point in
3961 its cache line. Look in @file{m68k.h} as a guide.
3963 @defmac TRANSFER_FROM_TRAMPOLINE
3964 Define this macro if trampolines need a special subroutine to do their
3965 work. The macro should expand to a series of @code{asm} statements
3966 which will be compiled with GCC@. They go in a library function named
3967 @code{__transfer_from_trampoline}.
3969 If you need to avoid executing the ordinary prologue code of a compiled
3970 C function when you jump to the subroutine, you can do so by placing a
3971 special label of your own in the assembler code. Use one @code{asm}
3972 statement to generate an assembler label, and another to make the label
3973 global. Then trampolines can use that label to jump directly to your
3974 special assembler code.
3978 @section Implicit Calls to Library Routines
3979 @cindex library subroutine names
3980 @cindex @file{libgcc.a}
3982 @c prevent bad page break with this line
3983 Here is an explanation of implicit calls to library routines.
3985 @defmac DECLARE_LIBRARY_RENAMES
3986 This macro, if defined, should expand to a piece of C code that will get
3987 expanded when compiling functions for libgcc.a. It can be used to
3988 provide alternate names for GCC's internal library functions if there
3989 are ABI-mandated names that the compiler should provide.
3992 @findex set_optab_libfunc
3993 @findex init_one_libfunc
3994 @hook TARGET_INIT_LIBFUNCS
3996 @hook TARGET_LIBFUNC_GNU_PREFIX
3998 @defmac FLOAT_LIB_COMPARE_RETURNS_BOOL (@var{mode}, @var{comparison})
3999 This macro should return @code{true} if the library routine that
4000 implements the floating point comparison operator @var{comparison} in
4001 mode @var{mode} will return a boolean, and @var{false} if it will
4004 GCC's own floating point libraries return tristates from the
4005 comparison operators, so the default returns false always. Most ports
4006 don't need to define this macro.
4009 @defmac TARGET_LIB_INT_CMP_BIASED
4010 This macro should evaluate to @code{true} if the integer comparison
4011 functions (like @code{__cmpdi2}) return 0 to indicate that the first
4012 operand is smaller than the second, 1 to indicate that they are equal,
4013 and 2 to indicate that the first operand is greater than the second.
4014 If this macro evaluates to @code{false} the comparison functions return
4015 @minus{}1, 0, and 1 instead of 0, 1, and 2. If the target uses the routines
4016 in @file{libgcc.a}, you do not need to define this macro.
4019 @defmac TARGET_HAS_NO_HW_DIVIDE
4020 This macro should be defined if the target has no hardware divide
4021 instructions. If this macro is defined, GCC will use an algorithm which
4022 make use of simple logical and arithmetic operations for 64-bit
4023 division. If the macro is not defined, GCC will use an algorithm which
4024 make use of a 64-bit by 32-bit divide primitive.
4027 @cindex @code{EDOM}, implicit usage
4030 The value of @code{EDOM} on the target machine, as a C integer constant
4031 expression. If you don't define this macro, GCC does not attempt to
4032 deposit the value of @code{EDOM} into @code{errno} directly. Look in
4033 @file{/usr/include/errno.h} to find the value of @code{EDOM} on your
4036 If you do not define @code{TARGET_EDOM}, then compiled code reports
4037 domain errors by calling the library function and letting it report the
4038 error. If mathematical functions on your system use @code{matherr} when
4039 there is an error, then you should leave @code{TARGET_EDOM} undefined so
4040 that @code{matherr} is used normally.
4043 @cindex @code{errno}, implicit usage
4044 @defmac GEN_ERRNO_RTX
4045 Define this macro as a C expression to create an rtl expression that
4046 refers to the global ``variable'' @code{errno}. (On certain systems,
4047 @code{errno} may not actually be a variable.) If you don't define this
4048 macro, a reasonable default is used.
4051 @hook TARGET_LIBC_HAS_FUNCTION
4053 @defmac NEXT_OBJC_RUNTIME
4054 Set this macro to 1 to use the "NeXT" Objective-C message sending conventions
4055 by default. This calling convention involves passing the object, the selector
4056 and the method arguments all at once to the method-lookup library function.
4057 This is the usual setting when targeting Darwin/Mac OS X systems, which have
4058 the NeXT runtime installed.
4060 If the macro is set to 0, the "GNU" Objective-C message sending convention
4061 will be used by default. This convention passes just the object and the
4062 selector to the method-lookup function, which returns a pointer to the method.
4064 In either case, it remains possible to select code-generation for the alternate
4065 scheme, by means of compiler command line switches.
4068 @node Addressing Modes
4069 @section Addressing Modes
4070 @cindex addressing modes
4072 @c prevent bad page break with this line
4073 This is about addressing modes.
4075 @defmac HAVE_PRE_INCREMENT
4076 @defmacx HAVE_PRE_DECREMENT
4077 @defmacx HAVE_POST_INCREMENT
4078 @defmacx HAVE_POST_DECREMENT
4079 A C expression that is nonzero if the machine supports pre-increment,
4080 pre-decrement, post-increment, or post-decrement addressing respectively.
4083 @defmac HAVE_PRE_MODIFY_DISP
4084 @defmacx HAVE_POST_MODIFY_DISP
4085 A C expression that is nonzero if the machine supports pre- or
4086 post-address side-effect generation involving constants other than
4087 the size of the memory operand.
4090 @defmac HAVE_PRE_MODIFY_REG
4091 @defmacx HAVE_POST_MODIFY_REG
4092 A C expression that is nonzero if the machine supports pre- or
4093 post-address side-effect generation involving a register displacement.
4096 @defmac CONSTANT_ADDRESS_P (@var{x})
4097 A C expression that is 1 if the RTX @var{x} is a constant which
4098 is a valid address. On most machines the default definition of
4099 @code{(CONSTANT_P (@var{x}) && GET_CODE (@var{x}) != CONST_DOUBLE)}
4100 is acceptable, but a few machines are more restrictive as to which
4101 constant addresses are supported.
4104 @defmac CONSTANT_P (@var{x})
4105 @code{CONSTANT_P}, which is defined by target-independent code,
4106 accepts integer-values expressions whose values are not explicitly
4107 known, such as @code{symbol_ref}, @code{label_ref}, and @code{high}
4108 expressions and @code{const} arithmetic expressions, in addition to
4109 @code{const_int} and @code{const_double} expressions.
4112 @defmac MAX_REGS_PER_ADDRESS
4113 A number, the maximum number of registers that can appear in a valid
4114 memory address. Note that it is up to you to specify a value equal to
4115 the maximum number that @code{TARGET_LEGITIMATE_ADDRESS_P} would ever
4119 @hook TARGET_LEGITIMATE_ADDRESS_P
4121 @defmac TARGET_MEM_CONSTRAINT
4122 A single character to be used instead of the default @code{'m'}
4123 character for general memory addresses. This defines the constraint
4124 letter which matches the memory addresses accepted by
4125 @code{TARGET_LEGITIMATE_ADDRESS_P}. Define this macro if you want to
4126 support new address formats in your back end without changing the
4127 semantics of the @code{'m'} constraint. This is necessary in order to
4128 preserve functionality of inline assembly constructs using the
4129 @code{'m'} constraint.
4132 @defmac FIND_BASE_TERM (@var{x})
4133 A C expression to determine the base term of address @var{x},
4134 or to provide a simplified version of @var{x} from which @file{alias.c}
4135 can easily find the base term. This macro is used in only two places:
4136 @code{find_base_value} and @code{find_base_term} in @file{alias.c}.
4138 It is always safe for this macro to not be defined. It exists so
4139 that alias analysis can understand machine-dependent addresses.
4141 The typical use of this macro is to handle addresses containing
4142 a label_ref or symbol_ref within an UNSPEC@.
4145 @hook TARGET_LEGITIMIZE_ADDRESS
4147 @defmac LEGITIMIZE_RELOAD_ADDRESS (@var{x}, @var{mode}, @var{opnum}, @var{type}, @var{ind_levels}, @var{win})
4148 A C compound statement that attempts to replace @var{x}, which is an address
4149 that needs reloading, with a valid memory address for an operand of mode
4150 @var{mode}. @var{win} will be a C statement label elsewhere in the code.
4151 It is not necessary to define this macro, but it might be useful for
4152 performance reasons.
4154 For example, on the i386, it is sometimes possible to use a single
4155 reload register instead of two by reloading a sum of two pseudo
4156 registers into a register. On the other hand, for number of RISC
4157 processors offsets are limited so that often an intermediate address
4158 needs to be generated in order to address a stack slot. By defining
4159 @code{LEGITIMIZE_RELOAD_ADDRESS} appropriately, the intermediate addresses
4160 generated for adjacent some stack slots can be made identical, and thus
4163 @emph{Note}: This macro should be used with caution. It is necessary
4164 to know something of how reload works in order to effectively use this,
4165 and it is quite easy to produce macros that build in too much knowledge
4166 of reload internals.
4168 @emph{Note}: This macro must be able to reload an address created by a
4169 previous invocation of this macro. If it fails to handle such addresses
4170 then the compiler may generate incorrect code or abort.
4173 The macro definition should use @code{push_reload} to indicate parts that
4174 need reloading; @var{opnum}, @var{type} and @var{ind_levels} are usually
4175 suitable to be passed unaltered to @code{push_reload}.
4177 The code generated by this macro must not alter the substructure of
4178 @var{x}. If it transforms @var{x} into a more legitimate form, it
4179 should assign @var{x} (which will always be a C variable) a new value.
4180 This also applies to parts that you change indirectly by calling
4183 @findex strict_memory_address_p
4184 The macro definition may use @code{strict_memory_address_p} to test if
4185 the address has become legitimate.
4188 If you want to change only a part of @var{x}, one standard way of doing
4189 this is to use @code{copy_rtx}. Note, however, that it unshares only a
4190 single level of rtl. Thus, if the part to be changed is not at the
4191 top level, you'll need to replace first the top level.
4192 It is not necessary for this macro to come up with a legitimate
4193 address; but often a machine-dependent strategy can generate better code.
4196 @hook TARGET_MODE_DEPENDENT_ADDRESS_P
4198 @hook TARGET_LEGITIMATE_CONSTANT_P
4200 @hook TARGET_DELEGITIMIZE_ADDRESS
4202 @hook TARGET_CONST_NOT_OK_FOR_DEBUG_P
4204 @hook TARGET_CANNOT_FORCE_CONST_MEM
4206 @hook TARGET_USE_BLOCKS_FOR_CONSTANT_P
4208 @hook TARGET_USE_BLOCKS_FOR_DECL_P
4210 @hook TARGET_BUILTIN_RECIPROCAL
4212 @hook TARGET_VECTORIZE_BUILTIN_MASK_FOR_LOAD
4214 @hook TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST
4216 @hook TARGET_VECTORIZE_VECTOR_ALIGNMENT_REACHABLE
4218 @hook TARGET_VECTORIZE_VEC_PERM_CONST_OK
4220 @hook TARGET_VECTORIZE_BUILTIN_CONVERSION
4222 @hook TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION
4224 @hook TARGET_VECTORIZE_SUPPORT_VECTOR_MISALIGNMENT
4226 @hook TARGET_VECTORIZE_PREFERRED_SIMD_MODE
4228 @hook TARGET_VECTORIZE_AUTOVECTORIZE_VECTOR_SIZES
4230 @hook TARGET_VECTORIZE_INIT_COST
4232 @hook TARGET_VECTORIZE_ADD_STMT_COST
4234 @hook TARGET_VECTORIZE_FINISH_COST
4236 @hook TARGET_VECTORIZE_DESTROY_COST_DATA
4238 @hook TARGET_VECTORIZE_BUILTIN_TM_LOAD
4240 @hook TARGET_VECTORIZE_BUILTIN_TM_STORE
4242 @hook TARGET_VECTORIZE_BUILTIN_GATHER
4244 @hook TARGET_SIMD_CLONE_COMPUTE_VECSIZE_AND_SIMDLEN
4246 @hook TARGET_SIMD_CLONE_ADJUST
4248 @hook TARGET_SIMD_CLONE_USABLE
4250 @node Anchored Addresses
4251 @section Anchored Addresses
4252 @cindex anchored addresses
4253 @cindex @option{-fsection-anchors}
4255 GCC usually addresses every static object as a separate entity.
4256 For example, if we have:
4260 int foo (void) @{ return a + b + c; @}
4263 the code for @code{foo} will usually calculate three separate symbolic
4264 addresses: those of @code{a}, @code{b} and @code{c}. On some targets,
4265 it would be better to calculate just one symbolic address and access
4266 the three variables relative to it. The equivalent pseudocode would
4272 register int *xr = &x;
4273 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
4277 (which isn't valid C). We refer to shared addresses like @code{x} as
4278 ``section anchors''. Their use is controlled by @option{-fsection-anchors}.
4280 The hooks below describe the target properties that GCC needs to know
4281 in order to make effective use of section anchors. It won't use
4282 section anchors at all unless either @code{TARGET_MIN_ANCHOR_OFFSET}
4283 or @code{TARGET_MAX_ANCHOR_OFFSET} is set to a nonzero value.
4285 @hook TARGET_MIN_ANCHOR_OFFSET
4287 @hook TARGET_MAX_ANCHOR_OFFSET
4289 @hook TARGET_ASM_OUTPUT_ANCHOR
4291 @hook TARGET_USE_ANCHORS_FOR_SYMBOL_P
4293 @node Condition Code
4294 @section Condition Code Status
4295 @cindex condition code status
4297 The macros in this section can be split in two families, according to the
4298 two ways of representing condition codes in GCC.
4300 The first representation is the so called @code{(cc0)} representation
4301 (@pxref{Jump Patterns}), where all instructions can have an implicit
4302 clobber of the condition codes. The second is the condition code
4303 register representation, which provides better schedulability for
4304 architectures that do have a condition code register, but on which
4305 most instructions do not affect it. The latter category includes
4308 The implicit clobbering poses a strong restriction on the placement of
4309 the definition and use of the condition code. In the past the definition
4310 and use were always adjacent. However, recent changes to support trapping
4311 arithmatic may result in the definition and user being in different blocks.
4312 Thus, there may be a @code{NOTE_INSN_BASIC_BLOCK} between them. Additionally,
4313 the definition may be the source of exception handling edges.
4315 These restrictions can prevent important
4316 optimizations on some machines. For example, on the IBM RS/6000, there
4317 is a delay for taken branches unless the condition code register is set
4318 three instructions earlier than the conditional branch. The instruction
4319 scheduler cannot perform this optimization if it is not permitted to
4320 separate the definition and use of the condition code register.
4322 For this reason, it is possible and suggested to use a register to
4323 represent the condition code for new ports. If there is a specific
4324 condition code register in the machine, use a hard register. If the
4325 condition code or comparison result can be placed in any general register,
4326 or if there are multiple condition registers, use a pseudo register.
4327 Registers used to store the condition code value will usually have a mode
4328 that is in class @code{MODE_CC}.
4330 Alternatively, you can use @code{BImode} if the comparison operator is
4331 specified already in the compare instruction. In this case, you are not
4332 interested in most macros in this section.
4335 * CC0 Condition Codes:: Old style representation of condition codes.
4336 * MODE_CC Condition Codes:: Modern representation of condition codes.
4339 @node CC0 Condition Codes
4340 @subsection Representation of condition codes using @code{(cc0)}
4344 The file @file{conditions.h} defines a variable @code{cc_status} to
4345 describe how the condition code was computed (in case the interpretation of
4346 the condition code depends on the instruction that it was set by). This
4347 variable contains the RTL expressions on which the condition code is
4348 currently based, and several standard flags.
4350 Sometimes additional machine-specific flags must be defined in the machine
4351 description header file. It can also add additional machine-specific
4352 information by defining @code{CC_STATUS_MDEP}.
4354 @defmac CC_STATUS_MDEP
4355 C code for a data type which is used for declaring the @code{mdep}
4356 component of @code{cc_status}. It defaults to @code{int}.
4358 This macro is not used on machines that do not use @code{cc0}.
4361 @defmac CC_STATUS_MDEP_INIT
4362 A C expression to initialize the @code{mdep} field to ``empty''.
4363 The default definition does nothing, since most machines don't use
4364 the field anyway. If you want to use the field, you should probably
4365 define this macro to initialize it.
4367 This macro is not used on machines that do not use @code{cc0}.
4370 @defmac NOTICE_UPDATE_CC (@var{exp}, @var{insn})
4371 A C compound statement to set the components of @code{cc_status}
4372 appropriately for an insn @var{insn} whose body is @var{exp}. It is
4373 this macro's responsibility to recognize insns that set the condition
4374 code as a byproduct of other activity as well as those that explicitly
4377 This macro is not used on machines that do not use @code{cc0}.
4379 If there are insns that do not set the condition code but do alter
4380 other machine registers, this macro must check to see whether they
4381 invalidate the expressions that the condition code is recorded as
4382 reflecting. For example, on the 68000, insns that store in address
4383 registers do not set the condition code, which means that usually
4384 @code{NOTICE_UPDATE_CC} can leave @code{cc_status} unaltered for such
4385 insns. But suppose that the previous insn set the condition code
4386 based on location @samp{a4@@(102)} and the current insn stores a new
4387 value in @samp{a4}. Although the condition code is not changed by
4388 this, it will no longer be true that it reflects the contents of
4389 @samp{a4@@(102)}. Therefore, @code{NOTICE_UPDATE_CC} must alter
4390 @code{cc_status} in this case to say that nothing is known about the
4391 condition code value.
4393 The definition of @code{NOTICE_UPDATE_CC} must be prepared to deal
4394 with the results of peephole optimization: insns whose patterns are
4395 @code{parallel} RTXs containing various @code{reg}, @code{mem} or
4396 constants which are just the operands. The RTL structure of these
4397 insns is not sufficient to indicate what the insns actually do. What
4398 @code{NOTICE_UPDATE_CC} should do when it sees one is just to run
4399 @code{CC_STATUS_INIT}.
4401 A possible definition of @code{NOTICE_UPDATE_CC} is to call a function
4402 that looks at an attribute (@pxref{Insn Attributes}) named, for example,
4403 @samp{cc}. This avoids having detailed information about patterns in
4404 two places, the @file{md} file and in @code{NOTICE_UPDATE_CC}.
4407 @node MODE_CC Condition Codes
4408 @subsection Representation of condition codes using registers
4412 @defmac SELECT_CC_MODE (@var{op}, @var{x}, @var{y})
4413 On many machines, the condition code may be produced by other instructions
4414 than compares, for example the branch can use directly the condition
4415 code set by a subtract instruction. However, on some machines
4416 when the condition code is set this way some bits (such as the overflow
4417 bit) are not set in the same way as a test instruction, so that a different
4418 branch instruction must be used for some conditional branches. When
4419 this happens, use the machine mode of the condition code register to
4420 record different formats of the condition code register. Modes can
4421 also be used to record which compare instruction (e.g. a signed or an
4422 unsigned comparison) produced the condition codes.
4424 If other modes than @code{CCmode} are required, add them to
4425 @file{@var{machine}-modes.def} and define @code{SELECT_CC_MODE} to choose
4426 a mode given an operand of a compare. This is needed because the modes
4427 have to be chosen not only during RTL generation but also, for example,
4428 by instruction combination. The result of @code{SELECT_CC_MODE} should
4429 be consistent with the mode used in the patterns; for example to support
4430 the case of the add on the SPARC discussed above, we have the pattern
4434 [(set (reg:CC_NOOV 0)
4436 (plus:SI (match_operand:SI 0 "register_operand" "%r")
4437 (match_operand:SI 1 "arith_operand" "rI"))
4444 together with a @code{SELECT_CC_MODE} that returns @code{CC_NOOVmode}
4445 for comparisons whose argument is a @code{plus}:
4448 #define SELECT_CC_MODE(OP,X,Y) \
4449 (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \
4450 ? ((OP == LT || OP == LE || OP == GT || OP == GE) \
4451 ? CCFPEmode : CCFPmode) \
4452 : ((GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \
4453 || GET_CODE (X) == NEG || GET_CODE (x) == ASHIFT) \
4454 ? CC_NOOVmode : CCmode))
4457 Another reason to use modes is to retain information on which operands
4458 were used by the comparison; see @code{REVERSIBLE_CC_MODE} later in
4461 You should define this macro if and only if you define extra CC modes
4462 in @file{@var{machine}-modes.def}.
4465 @hook TARGET_CANONICALIZE_COMPARISON
4467 @defmac REVERSIBLE_CC_MODE (@var{mode})
4468 A C expression whose value is one if it is always safe to reverse a
4469 comparison whose mode is @var{mode}. If @code{SELECT_CC_MODE}
4470 can ever return @var{mode} for a floating-point inequality comparison,
4471 then @code{REVERSIBLE_CC_MODE (@var{mode})} must be zero.
4473 You need not define this macro if it would always returns zero or if the
4474 floating-point format is anything other than @code{IEEE_FLOAT_FORMAT}.
4475 For example, here is the definition used on the SPARC, where floating-point
4476 inequality comparisons are given either @code{CCFPEmode} or @code{CCFPmode}:
4479 #define REVERSIBLE_CC_MODE(MODE) \
4480 ((MODE) != CCFPEmode && (MODE) != CCFPmode)
4484 @defmac REVERSE_CONDITION (@var{code}, @var{mode})
4485 A C expression whose value is reversed condition code of the @var{code} for
4486 comparison done in CC_MODE @var{mode}. The macro is used only in case
4487 @code{REVERSIBLE_CC_MODE (@var{mode})} is nonzero. Define this macro in case
4488 machine has some non-standard way how to reverse certain conditionals. For
4489 instance in case all floating point conditions are non-trapping, compiler may
4490 freely convert unordered compares to ordered ones. Then definition may look
4494 #define REVERSE_CONDITION(CODE, MODE) \
4495 ((MODE) != CCFPmode ? reverse_condition (CODE) \
4496 : reverse_condition_maybe_unordered (CODE))
4500 @hook TARGET_FIXED_CONDITION_CODE_REGS
4502 @hook TARGET_CC_MODES_COMPATIBLE
4504 @hook TARGET_FLAGS_REGNUM
4507 @section Describing Relative Costs of Operations
4508 @cindex costs of instructions
4509 @cindex relative costs
4510 @cindex speed of instructions
4512 These macros let you describe the relative speed of various operations
4513 on the target machine.
4515 @defmac REGISTER_MOVE_COST (@var{mode}, @var{from}, @var{to})
4516 A C expression for the cost of moving data of mode @var{mode} from a
4517 register in class @var{from} to one in class @var{to}. The classes are
4518 expressed using the enumeration values such as @code{GENERAL_REGS}. A
4519 value of 2 is the default; other values are interpreted relative to
4522 It is not required that the cost always equal 2 when @var{from} is the
4523 same as @var{to}; on some machines it is expensive to move between
4524 registers if they are not general registers.
4526 If reload sees an insn consisting of a single @code{set} between two
4527 hard registers, and if @code{REGISTER_MOVE_COST} applied to their
4528 classes returns a value of 2, reload does not check to ensure that the
4529 constraints of the insn are met. Setting a cost of other than 2 will
4530 allow reload to verify that the constraints are met. You should do this
4531 if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
4533 These macros are obsolete, new ports should use the target hook
4534 @code{TARGET_REGISTER_MOVE_COST} instead.
4537 @hook TARGET_REGISTER_MOVE_COST
4539 @defmac MEMORY_MOVE_COST (@var{mode}, @var{class}, @var{in})
4540 A C expression for the cost of moving data of mode @var{mode} between a
4541 register of class @var{class} and memory; @var{in} is zero if the value
4542 is to be written to memory, nonzero if it is to be read in. This cost
4543 is relative to those in @code{REGISTER_MOVE_COST}. If moving between
4544 registers and memory is more expensive than between two registers, you
4545 should define this macro to express the relative cost.
4547 If you do not define this macro, GCC uses a default cost of 4 plus
4548 the cost of copying via a secondary reload register, if one is
4549 needed. If your machine requires a secondary reload register to copy
4550 between memory and a register of @var{class} but the reload mechanism is
4551 more complex than copying via an intermediate, define this macro to
4552 reflect the actual cost of the move.
4554 GCC defines the function @code{memory_move_secondary_cost} if
4555 secondary reloads are needed. It computes the costs due to copying via
4556 a secondary register. If your machine copies from memory using a
4557 secondary register in the conventional way but the default base value of
4558 4 is not correct for your machine, define this macro to add some other
4559 value to the result of that function. The arguments to that function
4560 are the same as to this macro.
4562 These macros are obsolete, new ports should use the target hook
4563 @code{TARGET_MEMORY_MOVE_COST} instead.
4566 @hook TARGET_MEMORY_MOVE_COST
4568 @defmac BRANCH_COST (@var{speed_p}, @var{predictable_p})
4569 A C expression for the cost of a branch instruction. A value of 1 is
4570 the default; other values are interpreted relative to that. Parameter
4571 @var{speed_p} is true when the branch in question should be optimized
4572 for speed. When it is false, @code{BRANCH_COST} should return a value
4573 optimal for code size rather than performance. @var{predictable_p} is
4574 true for well-predicted branches. On many architectures the
4575 @code{BRANCH_COST} can be reduced then.
4578 Here are additional macros which do not specify precise relative costs,
4579 but only that certain actions are more expensive than GCC would
4582 @defmac SLOW_BYTE_ACCESS
4583 Define this macro as a C expression which is nonzero if accessing less
4584 than a word of memory (i.e.@: a @code{char} or a @code{short}) is no
4585 faster than accessing a word of memory, i.e., if such access
4586 require more than one instruction or if there is no difference in cost
4587 between byte and (aligned) word loads.
4589 When this macro is not defined, the compiler will access a field by
4590 finding the smallest containing object; when it is defined, a fullword
4591 load will be used if alignment permits. Unless bytes accesses are
4592 faster than word accesses, using word accesses is preferable since it
4593 may eliminate subsequent memory access if subsequent accesses occur to
4594 other fields in the same word of the structure, but to different bytes.
4597 @defmac SLOW_UNALIGNED_ACCESS (@var{mode}, @var{alignment})
4598 Define this macro to be the value 1 if memory accesses described by the
4599 @var{mode} and @var{alignment} parameters have a cost many times greater
4600 than aligned accesses, for example if they are emulated in a trap
4603 When this macro is nonzero, the compiler will act as if
4604 @code{STRICT_ALIGNMENT} were nonzero when generating code for block
4605 moves. This can cause significantly more instructions to be produced.
4606 Therefore, do not set this macro nonzero if unaligned accesses only add a
4607 cycle or two to the time for a memory access.
4609 If the value of this macro is always zero, it need not be defined. If
4610 this macro is defined, it should produce a nonzero value when
4611 @code{STRICT_ALIGNMENT} is nonzero.
4614 @defmac MOVE_RATIO (@var{speed})
4615 The threshold of number of scalar memory-to-memory move insns, @emph{below}
4616 which a sequence of insns should be generated instead of a
4617 string move insn or a library call. Increasing the value will always
4618 make code faster, but eventually incurs high cost in increased code size.
4620 Note that on machines where the corresponding move insn is a
4621 @code{define_expand} that emits a sequence of insns, this macro counts
4622 the number of such sequences.
4624 The parameter @var{speed} is true if the code is currently being
4625 optimized for speed rather than size.
4627 If you don't define this, a reasonable default is used.
4630 @hook TARGET_USE_BY_PIECES_INFRASTRUCTURE_P
4632 @defmac MOVE_MAX_PIECES
4633 A C expression used by @code{move_by_pieces} to determine the largest unit
4634 a load or store used to copy memory is. Defaults to @code{MOVE_MAX}.
4637 @defmac CLEAR_RATIO (@var{speed})
4638 The threshold of number of scalar move insns, @emph{below} which a sequence
4639 of insns should be generated to clear memory instead of a string clear insn
4640 or a library call. Increasing the value will always make code faster, but
4641 eventually incurs high cost in increased code size.
4643 The parameter @var{speed} is true if the code is currently being
4644 optimized for speed rather than size.
4646 If you don't define this, a reasonable default is used.
4649 @defmac SET_RATIO (@var{speed})
4650 The threshold of number of scalar move insns, @emph{below} which a sequence
4651 of insns should be generated to set memory to a constant value, instead of
4652 a block set insn or a library call.
4653 Increasing the value will always make code faster, but
4654 eventually incurs high cost in increased code size.
4656 The parameter @var{speed} is true if the code is currently being
4657 optimized for speed rather than size.
4659 If you don't define this, it defaults to the value of @code{MOVE_RATIO}.
4662 @defmac USE_LOAD_POST_INCREMENT (@var{mode})
4663 A C expression used to determine whether a load postincrement is a good
4664 thing to use for a given mode. Defaults to the value of
4665 @code{HAVE_POST_INCREMENT}.
4668 @defmac USE_LOAD_POST_DECREMENT (@var{mode})
4669 A C expression used to determine whether a load postdecrement is a good
4670 thing to use for a given mode. Defaults to the value of
4671 @code{HAVE_POST_DECREMENT}.
4674 @defmac USE_LOAD_PRE_INCREMENT (@var{mode})
4675 A C expression used to determine whether a load preincrement is a good
4676 thing to use for a given mode. Defaults to the value of
4677 @code{HAVE_PRE_INCREMENT}.
4680 @defmac USE_LOAD_PRE_DECREMENT (@var{mode})
4681 A C expression used to determine whether a load predecrement is a good
4682 thing to use for a given mode. Defaults to the value of
4683 @code{HAVE_PRE_DECREMENT}.
4686 @defmac USE_STORE_POST_INCREMENT (@var{mode})
4687 A C expression used to determine whether a store postincrement is a good
4688 thing to use for a given mode. Defaults to the value of
4689 @code{HAVE_POST_INCREMENT}.
4692 @defmac USE_STORE_POST_DECREMENT (@var{mode})
4693 A C expression used to determine whether a store postdecrement is a good
4694 thing to use for a given mode. Defaults to the value of
4695 @code{HAVE_POST_DECREMENT}.
4698 @defmac USE_STORE_PRE_INCREMENT (@var{mode})
4699 This macro is used to determine whether a store preincrement is a good
4700 thing to use for a given mode. Defaults to the value of
4701 @code{HAVE_PRE_INCREMENT}.
4704 @defmac USE_STORE_PRE_DECREMENT (@var{mode})
4705 This macro is used to determine whether a store predecrement is a good
4706 thing to use for a given mode. Defaults to the value of
4707 @code{HAVE_PRE_DECREMENT}.
4710 @defmac NO_FUNCTION_CSE
4711 Define this macro if it is as good or better to call a constant
4712 function address than to call an address kept in a register.
4715 @defmac LOGICAL_OP_NON_SHORT_CIRCUIT
4716 Define this macro if a non-short-circuit operation produced by
4717 @samp{fold_range_test ()} is optimal. This macro defaults to true if
4718 @code{BRANCH_COST} is greater than or equal to the value 2.
4721 @hook TARGET_RTX_COSTS
4723 @hook TARGET_ADDRESS_COST
4726 @section Adjusting the Instruction Scheduler
4728 The instruction scheduler may need a fair amount of machine-specific
4729 adjustment in order to produce good code. GCC provides several target
4730 hooks for this purpose. It is usually enough to define just a few of
4731 them: try the first ones in this list first.
4733 @hook TARGET_SCHED_ISSUE_RATE
4735 @hook TARGET_SCHED_VARIABLE_ISSUE
4737 @hook TARGET_SCHED_ADJUST_COST
4739 @hook TARGET_SCHED_ADJUST_PRIORITY
4741 @hook TARGET_SCHED_REORDER
4743 @hook TARGET_SCHED_REORDER2
4745 @hook TARGET_SCHED_MACRO_FUSION_P
4747 @hook TARGET_SCHED_MACRO_FUSION_PAIR_P
4749 @hook TARGET_SCHED_DEPENDENCIES_EVALUATION_HOOK
4751 @hook TARGET_SCHED_INIT
4753 @hook TARGET_SCHED_FINISH
4755 @hook TARGET_SCHED_INIT_GLOBAL
4757 @hook TARGET_SCHED_FINISH_GLOBAL
4759 @hook TARGET_SCHED_DFA_PRE_CYCLE_INSN
4761 @hook TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN
4763 @hook TARGET_SCHED_DFA_POST_CYCLE_INSN
4765 @hook TARGET_SCHED_INIT_DFA_POST_CYCLE_INSN
4767 @hook TARGET_SCHED_DFA_PRE_ADVANCE_CYCLE
4769 @hook TARGET_SCHED_DFA_POST_ADVANCE_CYCLE
4771 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD
4773 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD
4775 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_BEGIN
4777 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_ISSUE
4779 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_BACKTRACK
4781 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_END
4783 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_INIT
4785 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_FINI
4787 @hook TARGET_SCHED_DFA_NEW_CYCLE
4789 @hook TARGET_SCHED_IS_COSTLY_DEPENDENCE
4791 @hook TARGET_SCHED_H_I_D_EXTENDED
4793 @hook TARGET_SCHED_ALLOC_SCHED_CONTEXT
4795 @hook TARGET_SCHED_INIT_SCHED_CONTEXT
4797 @hook TARGET_SCHED_SET_SCHED_CONTEXT
4799 @hook TARGET_SCHED_CLEAR_SCHED_CONTEXT
4801 @hook TARGET_SCHED_FREE_SCHED_CONTEXT
4803 @hook TARGET_SCHED_SPECULATE_INSN
4805 @hook TARGET_SCHED_NEEDS_BLOCK_P
4807 @hook TARGET_SCHED_GEN_SPEC_CHECK
4809 @hook TARGET_SCHED_SET_SCHED_FLAGS
4811 @hook TARGET_SCHED_SMS_RES_MII
4813 @hook TARGET_SCHED_DISPATCH
4815 @hook TARGET_SCHED_DISPATCH_DO
4817 @hook TARGET_SCHED_EXPOSED_PIPELINE
4819 @hook TARGET_SCHED_REASSOCIATION_WIDTH
4821 @hook TARGET_SCHED_FUSION_PRIORITY
4824 @section Dividing the Output into Sections (Texts, Data, @dots{})
4825 @c the above section title is WAY too long. maybe cut the part between
4826 @c the (...)? --mew 10feb93
4828 An object file is divided into sections containing different types of
4829 data. In the most common case, there are three sections: the @dfn{text
4830 section}, which holds instructions and read-only data; the @dfn{data
4831 section}, which holds initialized writable data; and the @dfn{bss
4832 section}, which holds uninitialized data. Some systems have other kinds
4835 @file{varasm.c} provides several well-known sections, such as
4836 @code{text_section}, @code{data_section} and @code{bss_section}.
4837 The normal way of controlling a @code{@var{foo}_section} variable
4838 is to define the associated @code{@var{FOO}_SECTION_ASM_OP} macro,
4839 as described below. The macros are only read once, when @file{varasm.c}
4840 initializes itself, so their values must be run-time constants.
4841 They may however depend on command-line flags.
4843 @emph{Note:} Some run-time files, such @file{crtstuff.c}, also make
4844 use of the @code{@var{FOO}_SECTION_ASM_OP} macros, and expect them
4845 to be string literals.
4847 Some assemblers require a different string to be written every time a
4848 section is selected. If your assembler falls into this category, you
4849 should define the @code{TARGET_ASM_INIT_SECTIONS} hook and use
4850 @code{get_unnamed_section} to set up the sections.
4852 You must always create a @code{text_section}, either by defining
4853 @code{TEXT_SECTION_ASM_OP} or by initializing @code{text_section}
4854 in @code{TARGET_ASM_INIT_SECTIONS}. The same is true of
4855 @code{data_section} and @code{DATA_SECTION_ASM_OP}. If you do not
4856 create a distinct @code{readonly_data_section}, the default is to
4857 reuse @code{text_section}.
4859 All the other @file{varasm.c} sections are optional, and are null
4860 if the target does not provide them.
4862 @defmac TEXT_SECTION_ASM_OP
4863 A C expression whose value is a string, including spacing, containing the
4864 assembler operation that should precede instructions and read-only data.
4865 Normally @code{"\t.text"} is right.
4868 @defmac HOT_TEXT_SECTION_NAME
4869 If defined, a C string constant for the name of the section containing most
4870 frequently executed functions of the program. If not defined, GCC will provide
4871 a default definition if the target supports named sections.
4874 @defmac UNLIKELY_EXECUTED_TEXT_SECTION_NAME
4875 If defined, a C string constant for the name of the section containing unlikely
4876 executed functions in the program.
4879 @defmac DATA_SECTION_ASM_OP
4880 A C expression whose value is a string, including spacing, containing the
4881 assembler operation to identify the following data as writable initialized
4882 data. Normally @code{"\t.data"} is right.
4885 @defmac SDATA_SECTION_ASM_OP
4886 If defined, a C expression whose value is a string, including spacing,
4887 containing the assembler operation to identify the following data as
4888 initialized, writable small data.
4891 @defmac READONLY_DATA_SECTION_ASM_OP
4892 A C expression whose value is a string, including spacing, containing the
4893 assembler operation to identify the following data as read-only initialized
4897 @defmac BSS_SECTION_ASM_OP
4898 If defined, a C expression whose value is a string, including spacing,
4899 containing the assembler operation to identify the following data as
4900 uninitialized global data. If not defined, and
4901 @code{ASM_OUTPUT_ALIGNED_BSS} not defined,
4902 uninitialized global data will be output in the data section if
4903 @option{-fno-common} is passed, otherwise @code{ASM_OUTPUT_COMMON} will be
4907 @defmac SBSS_SECTION_ASM_OP
4908 If defined, a C expression whose value is a string, including spacing,
4909 containing the assembler operation to identify the following data as
4910 uninitialized, writable small data.
4913 @defmac TLS_COMMON_ASM_OP
4914 If defined, a C expression whose value is a string containing the
4915 assembler operation to identify the following data as thread-local
4916 common data. The default is @code{".tls_common"}.
4919 @defmac TLS_SECTION_ASM_FLAG
4920 If defined, a C expression whose value is a character constant
4921 containing the flag used to mark a section as a TLS section. The
4922 default is @code{'T'}.
4925 @defmac INIT_SECTION_ASM_OP
4926 If defined, a C expression whose value is a string, including spacing,
4927 containing the assembler operation to identify the following data as
4928 initialization code. If not defined, GCC will assume such a section does
4929 not exist. This section has no corresponding @code{init_section}
4930 variable; it is used entirely in runtime code.
4933 @defmac FINI_SECTION_ASM_OP
4934 If defined, a C expression whose value is a string, including spacing,
4935 containing the assembler operation to identify the following data as
4936 finalization code. If not defined, GCC will assume such a section does
4937 not exist. This section has no corresponding @code{fini_section}
4938 variable; it is used entirely in runtime code.
4941 @defmac INIT_ARRAY_SECTION_ASM_OP
4942 If defined, a C expression whose value is a string, including spacing,
4943 containing the assembler operation to identify the following data as
4944 part of the @code{.init_array} (or equivalent) section. If not
4945 defined, GCC will assume such a section does not exist. Do not define
4946 both this macro and @code{INIT_SECTION_ASM_OP}.
4949 @defmac FINI_ARRAY_SECTION_ASM_OP
4950 If defined, a C expression whose value is a string, including spacing,
4951 containing the assembler operation to identify the following data as
4952 part of the @code{.fini_array} (or equivalent) section. If not
4953 defined, GCC will assume such a section does not exist. Do not define
4954 both this macro and @code{FINI_SECTION_ASM_OP}.
4957 @defmac CRT_CALL_STATIC_FUNCTION (@var{section_op}, @var{function})
4958 If defined, an ASM statement that switches to a different section
4959 via @var{section_op}, calls @var{function}, and switches back to
4960 the text section. This is used in @file{crtstuff.c} if
4961 @code{INIT_SECTION_ASM_OP} or @code{FINI_SECTION_ASM_OP} to calls
4962 to initialization and finalization functions from the init and fini
4963 sections. By default, this macro uses a simple function call. Some
4964 ports need hand-crafted assembly code to avoid dependencies on
4965 registers initialized in the function prologue or to ensure that
4966 constant pools don't end up too far way in the text section.
4969 @defmac TARGET_LIBGCC_SDATA_SECTION
4970 If defined, a string which names the section into which small
4971 variables defined in crtstuff and libgcc should go. This is useful
4972 when the target has options for optimizing access to small data, and
4973 you want the crtstuff and libgcc routines to be conservative in what
4974 they expect of your application yet liberal in what your application
4975 expects. For example, for targets with a @code{.sdata} section (like
4976 MIPS), you could compile crtstuff with @code{-G 0} so that it doesn't
4977 require small data support from your application, but use this macro
4978 to put small data into @code{.sdata} so that your application can
4979 access these variables whether it uses small data or not.
4982 @defmac FORCE_CODE_SECTION_ALIGN
4983 If defined, an ASM statement that aligns a code section to some
4984 arbitrary boundary. This is used to force all fragments of the
4985 @code{.init} and @code{.fini} sections to have to same alignment
4986 and thus prevent the linker from having to add any padding.
4989 @defmac JUMP_TABLES_IN_TEXT_SECTION
4990 Define this macro to be an expression with a nonzero value if jump
4991 tables (for @code{tablejump} insns) should be output in the text
4992 section, along with the assembler instructions. Otherwise, the
4993 readonly data section is used.
4995 This macro is irrelevant if there is no separate readonly data section.
4998 @hook TARGET_ASM_INIT_SECTIONS
5000 @hook TARGET_ASM_RELOC_RW_MASK
5002 @hook TARGET_ASM_SELECT_SECTION
5004 @defmac USE_SELECT_SECTION_FOR_FUNCTIONS
5005 Define this macro if you wish TARGET_ASM_SELECT_SECTION to be called
5006 for @code{FUNCTION_DECL}s as well as for variables and constants.
5008 In the case of a @code{FUNCTION_DECL}, @var{reloc} will be zero if the
5009 function has been determined to be likely to be called, and nonzero if
5010 it is unlikely to be called.
5013 @hook TARGET_ASM_UNIQUE_SECTION
5015 @hook TARGET_ASM_FUNCTION_RODATA_SECTION
5017 @hook TARGET_ASM_MERGEABLE_RODATA_PREFIX
5019 @hook TARGET_ASM_TM_CLONE_TABLE_SECTION
5021 @hook TARGET_ASM_SELECT_RTX_SECTION
5023 @hook TARGET_MANGLE_DECL_ASSEMBLER_NAME
5025 @hook TARGET_ENCODE_SECTION_INFO
5027 @hook TARGET_STRIP_NAME_ENCODING
5029 @hook TARGET_IN_SMALL_DATA_P
5031 @hook TARGET_HAVE_SRODATA_SECTION
5033 @hook TARGET_PROFILE_BEFORE_PROLOGUE
5035 @hook TARGET_BINDS_LOCAL_P
5037 @hook TARGET_HAVE_TLS
5041 @section Position Independent Code
5042 @cindex position independent code
5045 This section describes macros that help implement generation of position
5046 independent code. Simply defining these macros is not enough to
5047 generate valid PIC; you must also add support to the hook
5048 @code{TARGET_LEGITIMATE_ADDRESS_P} and to the macro
5049 @code{PRINT_OPERAND_ADDRESS}, as well as @code{LEGITIMIZE_ADDRESS}. You
5050 must modify the definition of @samp{movsi} to do something appropriate
5051 when the source operand contains a symbolic address. You may also
5052 need to alter the handling of switch statements so that they use
5054 @c i rearranged the order of the macros above to try to force one of
5055 @c them to the next line, to eliminate an overfull hbox. --mew 10feb93
5057 @defmac PIC_OFFSET_TABLE_REGNUM
5058 The register number of the register used to address a table of static
5059 data addresses in memory. In some cases this register is defined by a
5060 processor's ``application binary interface'' (ABI)@. When this macro
5061 is defined, RTL is generated for this register once, as with the stack
5062 pointer and frame pointer registers. If this macro is not defined, it
5063 is up to the machine-dependent files to allocate such a register (if
5064 necessary). Note that this register must be fixed when in use (e.g.@:
5065 when @code{flag_pic} is true).
5068 @defmac PIC_OFFSET_TABLE_REG_CALL_CLOBBERED
5069 A C expression that is nonzero if the register defined by
5070 @code{PIC_OFFSET_TABLE_REGNUM} is clobbered by calls. If not defined,
5071 the default is zero. Do not define
5072 this macro if @code{PIC_OFFSET_TABLE_REGNUM} is not defined.
5075 @defmac LEGITIMATE_PIC_OPERAND_P (@var{x})
5076 A C expression that is nonzero if @var{x} is a legitimate immediate
5077 operand on the target machine when generating position independent code.
5078 You can assume that @var{x} satisfies @code{CONSTANT_P}, so you need not
5079 check this. You can also assume @var{flag_pic} is true, so you need not
5080 check it either. You need not define this macro if all constants
5081 (including @code{SYMBOL_REF}) can be immediate operands when generating
5082 position independent code.
5085 @node Assembler Format
5086 @section Defining the Output Assembler Language
5088 This section describes macros whose principal purpose is to describe how
5089 to write instructions in assembler language---rather than what the
5093 * File Framework:: Structural information for the assembler file.
5094 * Data Output:: Output of constants (numbers, strings, addresses).
5095 * Uninitialized Data:: Output of uninitialized variables.
5096 * Label Output:: Output and generation of labels.
5097 * Initialization:: General principles of initialization
5098 and termination routines.
5099 * Macros for Initialization::
5100 Specific macros that control the handling of
5101 initialization and termination routines.
5102 * Instruction Output:: Output of actual instructions.
5103 * Dispatch Tables:: Output of jump tables.
5104 * Exception Region Output:: Output of exception region code.
5105 * Alignment Output:: Pseudo ops for alignment and skipping data.
5108 @node File Framework
5109 @subsection The Overall Framework of an Assembler File
5110 @cindex assembler format
5111 @cindex output of assembler code
5113 @c prevent bad page break with this line
5114 This describes the overall framework of an assembly file.
5116 @findex default_file_start
5117 @hook TARGET_ASM_FILE_START
5119 @hook TARGET_ASM_FILE_START_APP_OFF
5121 @hook TARGET_ASM_FILE_START_FILE_DIRECTIVE
5123 @hook TARGET_ASM_FILE_END
5125 @deftypefun void file_end_indicate_exec_stack ()
5126 Some systems use a common convention, the @samp{.note.GNU-stack}
5127 special section, to indicate whether or not an object file relies on
5128 the stack being executable. If your system uses this convention, you
5129 should define @code{TARGET_ASM_FILE_END} to this function. If you
5130 need to do other things in that hook, have your hook function call
5134 @hook TARGET_ASM_LTO_START
5136 @hook TARGET_ASM_LTO_END
5138 @hook TARGET_ASM_CODE_END
5140 @defmac ASM_COMMENT_START
5141 A C string constant describing how to begin a comment in the target
5142 assembler language. The compiler assumes that the comment will end at
5143 the end of the line.
5147 A C string constant for text to be output before each @code{asm}
5148 statement or group of consecutive ones. Normally this is
5149 @code{"#APP"}, which is a comment that has no effect on most
5150 assemblers but tells the GNU assembler that it must check the lines
5151 that follow for all valid assembler constructs.
5155 A C string constant for text to be output after each @code{asm}
5156 statement or group of consecutive ones. Normally this is
5157 @code{"#NO_APP"}, which tells the GNU assembler to resume making the
5158 time-saving assumptions that are valid for ordinary compiler output.
5161 @defmac ASM_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name})
5162 A C statement to output COFF information or DWARF debugging information
5163 which indicates that filename @var{name} is the current source file to
5164 the stdio stream @var{stream}.
5166 This macro need not be defined if the standard form of output
5167 for the file format in use is appropriate.
5170 @hook TARGET_ASM_OUTPUT_SOURCE_FILENAME
5172 @hook TARGET_ASM_OUTPUT_IDENT
5174 @defmac OUTPUT_QUOTED_STRING (@var{stream}, @var{string})
5175 A C statement to output the string @var{string} to the stdio stream
5176 @var{stream}. If you do not call the function @code{output_quoted_string}
5177 in your config files, GCC will only call it to output filenames to
5178 the assembler source. So you can use it to canonicalize the format
5179 of the filename using this macro.
5182 @hook TARGET_ASM_NAMED_SECTION
5184 @hook TARGET_ASM_FUNCTION_SECTION
5186 @hook TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS
5188 @hook TARGET_HAVE_NAMED_SECTIONS
5189 This flag is true if the target supports @code{TARGET_ASM_NAMED_SECTION}.
5190 It must not be modified by command-line option processing.
5193 @anchor{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}
5194 @hook TARGET_HAVE_SWITCHABLE_BSS_SECTIONS
5196 @hook TARGET_SECTION_TYPE_FLAGS
5198 @hook TARGET_ASM_RECORD_GCC_SWITCHES
5200 @hook TARGET_ASM_RECORD_GCC_SWITCHES_SECTION
5204 @subsection Output of Data
5207 @hook TARGET_ASM_BYTE_OP
5209 @hook TARGET_ASM_INTEGER
5211 @hook TARGET_ASM_DECL_END
5213 @hook TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA
5215 @defmac ASM_OUTPUT_ASCII (@var{stream}, @var{ptr}, @var{len})
5216 A C statement to output to the stdio stream @var{stream} an assembler
5217 instruction to assemble a string constant containing the @var{len}
5218 bytes at @var{ptr}. @var{ptr} will be a C expression of type
5219 @code{char *} and @var{len} a C expression of type @code{int}.
5221 If the assembler has a @code{.ascii} pseudo-op as found in the
5222 Berkeley Unix assembler, do not define the macro
5223 @code{ASM_OUTPUT_ASCII}.
5226 @defmac ASM_OUTPUT_FDESC (@var{stream}, @var{decl}, @var{n})
5227 A C statement to output word @var{n} of a function descriptor for
5228 @var{decl}. This must be defined if @code{TARGET_VTABLE_USES_DESCRIPTORS}
5229 is defined, and is otherwise unused.
5232 @defmac CONSTANT_POOL_BEFORE_FUNCTION
5233 You may define this macro as a C expression. You should define the
5234 expression to have a nonzero value if GCC should output the constant
5235 pool for a function before the code for the function, or a zero value if
5236 GCC should output the constant pool after the function. If you do
5237 not define this macro, the usual case, GCC will output the constant
5238 pool before the function.
5241 @defmac ASM_OUTPUT_POOL_PROLOGUE (@var{file}, @var{funname}, @var{fundecl}, @var{size})
5242 A C statement to output assembler commands to define the start of the
5243 constant pool for a function. @var{funname} is a string giving
5244 the name of the function. Should the return type of the function
5245 be required, it can be obtained via @var{fundecl}. @var{size}
5246 is the size, in bytes, of the constant pool that will be written
5247 immediately after this call.
5249 If no constant-pool prefix is required, the usual case, this macro need
5253 @defmac ASM_OUTPUT_SPECIAL_POOL_ENTRY (@var{file}, @var{x}, @var{mode}, @var{align}, @var{labelno}, @var{jumpto})
5254 A C statement (with or without semicolon) to output a constant in the
5255 constant pool, if it needs special treatment. (This macro need not do
5256 anything for RTL expressions that can be output normally.)
5258 The argument @var{file} is the standard I/O stream to output the
5259 assembler code on. @var{x} is the RTL expression for the constant to
5260 output, and @var{mode} is the machine mode (in case @var{x} is a
5261 @samp{const_int}). @var{align} is the required alignment for the value
5262 @var{x}; you should output an assembler directive to force this much
5265 The argument @var{labelno} is a number to use in an internal label for
5266 the address of this pool entry. The definition of this macro is
5267 responsible for outputting the label definition at the proper place.
5268 Here is how to do this:
5271 @code{(*targetm.asm_out.internal_label)} (@var{file}, "LC", @var{labelno});
5274 When you output a pool entry specially, you should end with a
5275 @code{goto} to the label @var{jumpto}. This will prevent the same pool
5276 entry from being output a second time in the usual manner.
5278 You need not define this macro if it would do nothing.
5281 @defmac ASM_OUTPUT_POOL_EPILOGUE (@var{file} @var{funname} @var{fundecl} @var{size})
5282 A C statement to output assembler commands to at the end of the constant
5283 pool for a function. @var{funname} is a string giving the name of the
5284 function. Should the return type of the function be required, you can
5285 obtain it via @var{fundecl}. @var{size} is the size, in bytes, of the
5286 constant pool that GCC wrote immediately before this call.
5288 If no constant-pool epilogue is required, the usual case, you need not
5292 @defmac IS_ASM_LOGICAL_LINE_SEPARATOR (@var{C}, @var{STR})
5293 Define this macro as a C expression which is nonzero if @var{C} is
5294 used as a logical line separator by the assembler. @var{STR} points
5295 to the position in the string where @var{C} was found; this can be used if
5296 a line separator uses multiple characters.
5298 If you do not define this macro, the default is that only
5299 the character @samp{;} is treated as a logical line separator.
5302 @hook TARGET_ASM_OPEN_PAREN
5304 These macros are provided by @file{real.h} for writing the definitions
5305 of @code{ASM_OUTPUT_DOUBLE} and the like:
5307 @defmac REAL_VALUE_TO_TARGET_SINGLE (@var{x}, @var{l})
5308 @defmacx REAL_VALUE_TO_TARGET_DOUBLE (@var{x}, @var{l})
5309 @defmacx REAL_VALUE_TO_TARGET_LONG_DOUBLE (@var{x}, @var{l})
5310 @defmacx REAL_VALUE_TO_TARGET_DECIMAL32 (@var{x}, @var{l})
5311 @defmacx REAL_VALUE_TO_TARGET_DECIMAL64 (@var{x}, @var{l})
5312 @defmacx REAL_VALUE_TO_TARGET_DECIMAL128 (@var{x}, @var{l})
5313 These translate @var{x}, of type @code{REAL_VALUE_TYPE}, to the
5314 target's floating point representation, and store its bit pattern in
5315 the variable @var{l}. For @code{REAL_VALUE_TO_TARGET_SINGLE} and
5316 @code{REAL_VALUE_TO_TARGET_DECIMAL32}, this variable should be a
5317 simple @code{long int}. For the others, it should be an array of
5318 @code{long int}. The number of elements in this array is determined
5319 by the size of the desired target floating point data type: 32 bits of
5320 it go in each @code{long int} array element. Each array element holds
5321 32 bits of the result, even if @code{long int} is wider than 32 bits
5322 on the host machine.
5324 The array element values are designed so that you can print them out
5325 using @code{fprintf} in the order they should appear in the target
5329 @node Uninitialized Data
5330 @subsection Output of Uninitialized Variables
5332 Each of the macros in this section is used to do the whole job of
5333 outputting a single uninitialized variable.
5335 @defmac ASM_OUTPUT_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
5336 A C statement (sans semicolon) to output to the stdio stream
5337 @var{stream} the assembler definition of a common-label named
5338 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
5339 is the size rounded up to whatever alignment the caller wants. It is
5340 possible that @var{size} may be zero, for instance if a struct with no
5341 other member than a zero-length array is defined. In this case, the
5342 backend must output a symbol definition that allocates at least one
5343 byte, both so that the address of the resulting object does not compare
5344 equal to any other, and because some object formats cannot even express
5345 the concept of a zero-sized common symbol, as that is how they represent
5346 an ordinary undefined external.
5348 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
5349 output the name itself; before and after that, output the additional
5350 assembler syntax for defining the name, and a newline.
5352 This macro controls how the assembler definitions of uninitialized
5353 common global variables are output.
5356 @defmac ASM_OUTPUT_ALIGNED_COMMON (@var{stream}, @var{name}, @var{size}, @var{alignment})
5357 Like @code{ASM_OUTPUT_COMMON} except takes the required alignment as a
5358 separate, explicit argument. If you define this macro, it is used in
5359 place of @code{ASM_OUTPUT_COMMON}, and gives you more flexibility in
5360 handling the required alignment of the variable. The alignment is specified
5361 as the number of bits.
5364 @defmac ASM_OUTPUT_ALIGNED_DECL_COMMON (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
5365 Like @code{ASM_OUTPUT_ALIGNED_COMMON} except that @var{decl} of the
5366 variable to be output, if there is one, or @code{NULL_TREE} if there
5367 is no corresponding variable. If you define this macro, GCC will use it
5368 in place of both @code{ASM_OUTPUT_COMMON} and
5369 @code{ASM_OUTPUT_ALIGNED_COMMON}. Define this macro when you need to see
5370 the variable's decl in order to chose what to output.
5373 @defmac ASM_OUTPUT_ALIGNED_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
5374 A C statement (sans semicolon) to output to the stdio stream
5375 @var{stream} the assembler definition of uninitialized global @var{decl} named
5376 @var{name} whose size is @var{size} bytes. The variable @var{alignment}
5377 is the alignment specified as the number of bits.
5379 Try to use function @code{asm_output_aligned_bss} defined in file
5380 @file{varasm.c} when defining this macro. If unable, use the expression
5381 @code{assemble_name (@var{stream}, @var{name})} to output the name itself;
5382 before and after that, output the additional assembler syntax for defining
5383 the name, and a newline.
5385 There are two ways of handling global BSS@. One is to define this macro.
5386 The other is to have @code{TARGET_ASM_SELECT_SECTION} return a
5387 switchable BSS section (@pxref{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}).
5388 You do not need to do both.
5390 Some languages do not have @code{common} data, and require a
5391 non-common form of global BSS in order to handle uninitialized globals
5392 efficiently. C++ is one example of this. However, if the target does
5393 not support global BSS, the front end may choose to make globals
5394 common in order to save space in the object file.
5397 @defmac ASM_OUTPUT_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded})
5398 A C statement (sans semicolon) to output to the stdio stream
5399 @var{stream} the assembler definition of a local-common-label named
5400 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
5401 is the size rounded up to whatever alignment the caller wants.
5403 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
5404 output the name itself; before and after that, output the additional
5405 assembler syntax for defining the name, and a newline.
5407 This macro controls how the assembler definitions of uninitialized
5408 static variables are output.
5411 @defmac ASM_OUTPUT_ALIGNED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{alignment})
5412 Like @code{ASM_OUTPUT_LOCAL} except takes the required alignment as a
5413 separate, explicit argument. If you define this macro, it is used in
5414 place of @code{ASM_OUTPUT_LOCAL}, and gives you more flexibility in
5415 handling the required alignment of the variable. The alignment is specified
5416 as the number of bits.
5419 @defmac ASM_OUTPUT_ALIGNED_DECL_LOCAL (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
5420 Like @code{ASM_OUTPUT_ALIGNED_DECL} except that @var{decl} of the
5421 variable to be output, if there is one, or @code{NULL_TREE} if there
5422 is no corresponding variable. If you define this macro, GCC will use it
5423 in place of both @code{ASM_OUTPUT_DECL} and
5424 @code{ASM_OUTPUT_ALIGNED_DECL}. Define this macro when you need to see
5425 the variable's decl in order to chose what to output.
5429 @subsection Output and Generation of Labels
5431 @c prevent bad page break with this line
5432 This is about outputting labels.
5434 @findex assemble_name
5435 @defmac ASM_OUTPUT_LABEL (@var{stream}, @var{name})
5436 A C statement (sans semicolon) to output to the stdio stream
5437 @var{stream} the assembler definition of a label named @var{name}.
5438 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
5439 output the name itself; before and after that, output the additional
5440 assembler syntax for defining the name, and a newline. A default
5441 definition of this macro is provided which is correct for most systems.
5444 @defmac ASM_OUTPUT_FUNCTION_LABEL (@var{stream}, @var{name}, @var{decl})
5445 A C statement (sans semicolon) to output to the stdio stream
5446 @var{stream} the assembler definition of a label named @var{name} of
5448 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
5449 output the name itself; before and after that, output the additional
5450 assembler syntax for defining the name, and a newline. A default
5451 definition of this macro is provided which is correct for most systems.
5453 If this macro is not defined, then the function name is defined in the
5454 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
5457 @findex assemble_name_raw
5458 @defmac ASM_OUTPUT_INTERNAL_LABEL (@var{stream}, @var{name})
5459 Identical to @code{ASM_OUTPUT_LABEL}, except that @var{name} is known
5460 to refer to a compiler-generated label. The default definition uses
5461 @code{assemble_name_raw}, which is like @code{assemble_name} except
5462 that it is more efficient.
5466 A C string containing the appropriate assembler directive to specify the
5467 size of a symbol, without any arguments. On systems that use ELF, the
5468 default (in @file{config/elfos.h}) is @samp{"\t.size\t"}; on other
5469 systems, the default is not to define this macro.
5471 Define this macro only if it is correct to use the default definitions
5472 of @code{ASM_OUTPUT_SIZE_DIRECTIVE} and @code{ASM_OUTPUT_MEASURED_SIZE}
5473 for your system. If you need your own custom definitions of those
5474 macros, or if you do not need explicit symbol sizes at all, do not
5478 @defmac ASM_OUTPUT_SIZE_DIRECTIVE (@var{stream}, @var{name}, @var{size})
5479 A C statement (sans semicolon) to output to the stdio stream
5480 @var{stream} a directive telling the assembler that the size of the
5481 symbol @var{name} is @var{size}. @var{size} is a @code{HOST_WIDE_INT}.
5482 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
5486 @defmac ASM_OUTPUT_MEASURED_SIZE (@var{stream}, @var{name})
5487 A C statement (sans semicolon) to output to the stdio stream
5488 @var{stream} a directive telling the assembler to calculate the size of
5489 the symbol @var{name} by subtracting its address from the current
5492 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
5493 provided. The default assumes that the assembler recognizes a special
5494 @samp{.} symbol as referring to the current address, and can calculate
5495 the difference between this and another symbol. If your assembler does
5496 not recognize @samp{.} or cannot do calculations with it, you will need
5497 to redefine @code{ASM_OUTPUT_MEASURED_SIZE} to use some other technique.
5500 @defmac NO_DOLLAR_IN_LABEL
5501 Define this macro if the assembler does not accept the character
5502 @samp{$} in label names. By default constructors and destructors in
5503 G++ have @samp{$} in the identifiers. If this macro is defined,
5504 @samp{.} is used instead.
5507 @defmac NO_DOT_IN_LABEL
5508 Define this macro if the assembler does not accept the character
5509 @samp{.} in label names. By default constructors and destructors in G++
5510 have names that use @samp{.}. If this macro is defined, these names
5511 are rewritten to avoid @samp{.}.
5515 A C string containing the appropriate assembler directive to specify the
5516 type of a symbol, without any arguments. On systems that use ELF, the
5517 default (in @file{config/elfos.h}) is @samp{"\t.type\t"}; on other
5518 systems, the default is not to define this macro.
5520 Define this macro only if it is correct to use the default definition of
5521 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
5522 custom definition of this macro, or if you do not need explicit symbol
5523 types at all, do not define this macro.
5526 @defmac TYPE_OPERAND_FMT
5527 A C string which specifies (using @code{printf} syntax) the format of
5528 the second operand to @code{TYPE_ASM_OP}. On systems that use ELF, the
5529 default (in @file{config/elfos.h}) is @samp{"@@%s"}; on other systems,
5530 the default is not to define this macro.
5532 Define this macro only if it is correct to use the default definition of
5533 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
5534 custom definition of this macro, or if you do not need explicit symbol
5535 types at all, do not define this macro.
5538 @defmac ASM_OUTPUT_TYPE_DIRECTIVE (@var{stream}, @var{type})
5539 A C statement (sans semicolon) to output to the stdio stream
5540 @var{stream} a directive telling the assembler that the type of the
5541 symbol @var{name} is @var{type}. @var{type} is a C string; currently,
5542 that string is always either @samp{"function"} or @samp{"object"}, but
5543 you should not count on this.
5545 If you define @code{TYPE_ASM_OP} and @code{TYPE_OPERAND_FMT}, a default
5546 definition of this macro is provided.
5549 @defmac ASM_DECLARE_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl})
5550 A C statement (sans semicolon) to output to the stdio stream
5551 @var{stream} any text necessary for declaring the name @var{name} of a
5552 function which is being defined. This macro is responsible for
5553 outputting the label definition (perhaps using
5554 @code{ASM_OUTPUT_FUNCTION_LABEL}). The argument @var{decl} is the
5555 @code{FUNCTION_DECL} tree node representing the function.
5557 If this macro is not defined, then the function name is defined in the
5558 usual manner as a label (by means of @code{ASM_OUTPUT_FUNCTION_LABEL}).
5560 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition
5564 @defmac ASM_DECLARE_FUNCTION_SIZE (@var{stream}, @var{name}, @var{decl})
5565 A C statement (sans semicolon) to output to the stdio stream
5566 @var{stream} any text necessary for declaring the size of a function
5567 which is being defined. The argument @var{name} is the name of the
5568 function. The argument @var{decl} is the @code{FUNCTION_DECL} tree node
5569 representing the function.
5571 If this macro is not defined, then the function size is not defined.
5573 You may wish to use @code{ASM_OUTPUT_MEASURED_SIZE} in the definition
5577 @defmac ASM_DECLARE_OBJECT_NAME (@var{stream}, @var{name}, @var{decl})
5578 A C statement (sans semicolon) to output to the stdio stream
5579 @var{stream} any text necessary for declaring the name @var{name} of an
5580 initialized variable which is being defined. This macro must output the
5581 label definition (perhaps using @code{ASM_OUTPUT_LABEL}). The argument
5582 @var{decl} is the @code{VAR_DECL} tree node representing the variable.
5584 If this macro is not defined, then the variable name is defined in the
5585 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
5587 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} and/or
5588 @code{ASM_OUTPUT_SIZE_DIRECTIVE} in the definition of this macro.
5591 @hook TARGET_ASM_DECLARE_CONSTANT_NAME
5593 @defmac ASM_DECLARE_REGISTER_GLOBAL (@var{stream}, @var{decl}, @var{regno}, @var{name})
5594 A C statement (sans semicolon) to output to the stdio stream
5595 @var{stream} any text necessary for claiming a register @var{regno}
5596 for a global variable @var{decl} with name @var{name}.
5598 If you don't define this macro, that is equivalent to defining it to do
5602 @defmac ASM_FINISH_DECLARE_OBJECT (@var{stream}, @var{decl}, @var{toplevel}, @var{atend})
5603 A C statement (sans semicolon) to finish up declaring a variable name
5604 once the compiler has processed its initializer fully and thus has had a
5605 chance to determine the size of an array when controlled by an
5606 initializer. This is used on systems where it's necessary to declare
5607 something about the size of the object.
5609 If you don't define this macro, that is equivalent to defining it to do
5612 You may wish to use @code{ASM_OUTPUT_SIZE_DIRECTIVE} and/or
5613 @code{ASM_OUTPUT_MEASURED_SIZE} in the definition of this macro.
5616 @hook TARGET_ASM_GLOBALIZE_LABEL
5618 @hook TARGET_ASM_GLOBALIZE_DECL_NAME
5620 @hook TARGET_ASM_ASSEMBLE_UNDEFINED_DECL
5622 @defmac ASM_WEAKEN_LABEL (@var{stream}, @var{name})
5623 A C statement (sans semicolon) to output to the stdio stream
5624 @var{stream} some commands that will make the label @var{name} weak;
5625 that is, available for reference from other files but only used if
5626 no other definition is available. Use the expression
5627 @code{assemble_name (@var{stream}, @var{name})} to output the name
5628 itself; before and after that, output the additional assembler syntax
5629 for making that name weak, and a newline.
5631 If you don't define this macro or @code{ASM_WEAKEN_DECL}, GCC will not
5632 support weak symbols and you should not define the @code{SUPPORTS_WEAK}
5636 @defmac ASM_WEAKEN_DECL (@var{stream}, @var{decl}, @var{name}, @var{value})
5637 Combines (and replaces) the function of @code{ASM_WEAKEN_LABEL} and
5638 @code{ASM_OUTPUT_WEAK_ALIAS}, allowing access to the associated function
5639 or variable decl. If @var{value} is not @code{NULL}, this C statement
5640 should output to the stdio stream @var{stream} assembler code which
5641 defines (equates) the weak symbol @var{name} to have the value
5642 @var{value}. If @var{value} is @code{NULL}, it should output commands
5643 to make @var{name} weak.
5646 @defmac ASM_OUTPUT_WEAKREF (@var{stream}, @var{decl}, @var{name}, @var{value})
5647 Outputs a directive that enables @var{name} to be used to refer to
5648 symbol @var{value} with weak-symbol semantics. @code{decl} is the
5649 declaration of @code{name}.
5652 @defmac SUPPORTS_WEAK
5653 A preprocessor constant expression which evaluates to true if the target
5654 supports weak symbols.
5656 If you don't define this macro, @file{defaults.h} provides a default
5657 definition. If either @code{ASM_WEAKEN_LABEL} or @code{ASM_WEAKEN_DECL}
5658 is defined, the default definition is @samp{1}; otherwise, it is @samp{0}.
5661 @defmac TARGET_SUPPORTS_WEAK
5662 A C expression which evaluates to true if the target supports weak symbols.
5664 If you don't define this macro, @file{defaults.h} provides a default
5665 definition. The default definition is @samp{(SUPPORTS_WEAK)}. Define
5666 this macro if you want to control weak symbol support with a compiler
5667 flag such as @option{-melf}.
5670 @defmac MAKE_DECL_ONE_ONLY (@var{decl})
5671 A C statement (sans semicolon) to mark @var{decl} to be emitted as a
5672 public symbol such that extra copies in multiple translation units will
5673 be discarded by the linker. Define this macro if your object file
5674 format provides support for this concept, such as the @samp{COMDAT}
5675 section flags in the Microsoft Windows PE/COFF format, and this support
5676 requires changes to @var{decl}, such as putting it in a separate section.
5679 @defmac SUPPORTS_ONE_ONLY
5680 A C expression which evaluates to true if the target supports one-only
5683 If you don't define this macro, @file{varasm.c} provides a default
5684 definition. If @code{MAKE_DECL_ONE_ONLY} is defined, the default
5685 definition is @samp{1}; otherwise, it is @samp{0}. Define this macro if
5686 you want to control one-only symbol support with a compiler flag, or if
5687 setting the @code{DECL_ONE_ONLY} flag is enough to mark a declaration to
5688 be emitted as one-only.
5691 @hook TARGET_ASM_ASSEMBLE_VISIBILITY
5693 @defmac TARGET_WEAK_NOT_IN_ARCHIVE_TOC
5694 A C expression that evaluates to true if the target's linker expects
5695 that weak symbols do not appear in a static archive's table of contents.
5696 The default is @code{0}.
5698 Leaving weak symbols out of an archive's table of contents means that,
5699 if a symbol will only have a definition in one translation unit and
5700 will have undefined references from other translation units, that
5701 symbol should not be weak. Defining this macro to be nonzero will
5702 thus have the effect that certain symbols that would normally be weak
5703 (explicit template instantiations, and vtables for polymorphic classes
5704 with noninline key methods) will instead be nonweak.
5706 The C++ ABI requires this macro to be zero. Define this macro for
5707 targets where full C++ ABI compliance is impossible and where linker
5708 restrictions require weak symbols to be left out of a static archive's
5712 @defmac ASM_OUTPUT_EXTERNAL (@var{stream}, @var{decl}, @var{name})
5713 A C statement (sans semicolon) to output to the stdio stream
5714 @var{stream} any text necessary for declaring the name of an external
5715 symbol named @var{name} which is referenced in this compilation but
5716 not defined. The value of @var{decl} is the tree node for the
5719 This macro need not be defined if it does not need to output anything.
5720 The GNU assembler and most Unix assemblers don't require anything.
5723 @hook TARGET_ASM_EXTERNAL_LIBCALL
5725 @hook TARGET_ASM_MARK_DECL_PRESERVED
5727 @defmac ASM_OUTPUT_LABELREF (@var{stream}, @var{name})
5728 A C statement (sans semicolon) to output to the stdio stream
5729 @var{stream} a reference in assembler syntax to a label named
5730 @var{name}. This should add @samp{_} to the front of the name, if that
5731 is customary on your operating system, as it is in most Berkeley Unix
5732 systems. This macro is used in @code{assemble_name}.
5735 @hook TARGET_MANGLE_ASSEMBLER_NAME
5737 @defmac ASM_OUTPUT_SYMBOL_REF (@var{stream}, @var{sym})
5738 A C statement (sans semicolon) to output a reference to
5739 @code{SYMBOL_REF} @var{sym}. If not defined, @code{assemble_name}
5740 will be used to output the name of the symbol. This macro may be used
5741 to modify the way a symbol is referenced depending on information
5742 encoded by @code{TARGET_ENCODE_SECTION_INFO}.
5745 @defmac ASM_OUTPUT_LABEL_REF (@var{stream}, @var{buf})
5746 A C statement (sans semicolon) to output a reference to @var{buf}, the
5747 result of @code{ASM_GENERATE_INTERNAL_LABEL}. If not defined,
5748 @code{assemble_name} will be used to output the name of the symbol.
5749 This macro is not used by @code{output_asm_label}, or the @code{%l}
5750 specifier that calls it; the intention is that this macro should be set
5751 when it is necessary to output a label differently when its address is
5755 @hook TARGET_ASM_INTERNAL_LABEL
5757 @defmac ASM_OUTPUT_DEBUG_LABEL (@var{stream}, @var{prefix}, @var{num})
5758 A C statement to output to the stdio stream @var{stream} a debug info
5759 label whose name is made from the string @var{prefix} and the number
5760 @var{num}. This is useful for VLIW targets, where debug info labels
5761 may need to be treated differently than branch target labels. On some
5762 systems, branch target labels must be at the beginning of instruction
5763 bundles, but debug info labels can occur in the middle of instruction
5766 If this macro is not defined, then @code{(*targetm.asm_out.internal_label)} will be
5770 @defmac ASM_GENERATE_INTERNAL_LABEL (@var{string}, @var{prefix}, @var{num})
5771 A C statement to store into the string @var{string} a label whose name
5772 is made from the string @var{prefix} and the number @var{num}.
5774 This string, when output subsequently by @code{assemble_name}, should
5775 produce the output that @code{(*targetm.asm_out.internal_label)} would produce
5776 with the same @var{prefix} and @var{num}.
5778 If the string begins with @samp{*}, then @code{assemble_name} will
5779 output the rest of the string unchanged. It is often convenient for
5780 @code{ASM_GENERATE_INTERNAL_LABEL} to use @samp{*} in this way. If the
5781 string doesn't start with @samp{*}, then @code{ASM_OUTPUT_LABELREF} gets
5782 to output the string, and may change it. (Of course,
5783 @code{ASM_OUTPUT_LABELREF} is also part of your machine description, so
5784 you should know what it does on your machine.)
5787 @defmac ASM_FORMAT_PRIVATE_NAME (@var{outvar}, @var{name}, @var{number})
5788 A C expression to assign to @var{outvar} (which is a variable of type
5789 @code{char *}) a newly allocated string made from the string
5790 @var{name} and the number @var{number}, with some suitable punctuation
5791 added. Use @code{alloca} to get space for the string.
5793 The string will be used as an argument to @code{ASM_OUTPUT_LABELREF} to
5794 produce an assembler label for an internal static variable whose name is
5795 @var{name}. Therefore, the string must be such as to result in valid
5796 assembler code. The argument @var{number} is different each time this
5797 macro is executed; it prevents conflicts between similarly-named
5798 internal static variables in different scopes.
5800 Ideally this string should not be a valid C identifier, to prevent any
5801 conflict with the user's own symbols. Most assemblers allow periods
5802 or percent signs in assembler symbols; putting at least one of these
5803 between the name and the number will suffice.
5805 If this macro is not defined, a default definition will be provided
5806 which is correct for most systems.
5809 @defmac ASM_OUTPUT_DEF (@var{stream}, @var{name}, @var{value})
5810 A C statement to output to the stdio stream @var{stream} assembler code
5811 which defines (equates) the symbol @var{name} to have the value @var{value}.
5814 If @code{SET_ASM_OP} is defined, a default definition is provided which is
5815 correct for most systems.
5818 @defmac ASM_OUTPUT_DEF_FROM_DECLS (@var{stream}, @var{decl_of_name}, @var{decl_of_value})
5819 A C statement to output to the stdio stream @var{stream} assembler code
5820 which defines (equates) the symbol whose tree node is @var{decl_of_name}
5821 to have the value of the tree node @var{decl_of_value}. This macro will
5822 be used in preference to @samp{ASM_OUTPUT_DEF} if it is defined and if
5823 the tree nodes are available.
5826 If @code{SET_ASM_OP} is defined, a default definition is provided which is
5827 correct for most systems.
5830 @defmac TARGET_DEFERRED_OUTPUT_DEFS (@var{decl_of_name}, @var{decl_of_value})
5831 A C statement that evaluates to true if the assembler code which defines
5832 (equates) the symbol whose tree node is @var{decl_of_name} to have the value
5833 of the tree node @var{decl_of_value} should be emitted near the end of the
5834 current compilation unit. The default is to not defer output of defines.
5835 This macro affects defines output by @samp{ASM_OUTPUT_DEF} and
5836 @samp{ASM_OUTPUT_DEF_FROM_DECLS}.
5839 @defmac ASM_OUTPUT_WEAK_ALIAS (@var{stream}, @var{name}, @var{value})
5840 A C statement to output to the stdio stream @var{stream} assembler code
5841 which defines (equates) the weak symbol @var{name} to have the value
5842 @var{value}. If @var{value} is @code{NULL}, it defines @var{name} as
5843 an undefined weak symbol.
5845 Define this macro if the target only supports weak aliases; define
5846 @code{ASM_OUTPUT_DEF} instead if possible.
5849 @defmac OBJC_GEN_METHOD_LABEL (@var{buf}, @var{is_inst}, @var{class_name}, @var{cat_name}, @var{sel_name})
5850 Define this macro to override the default assembler names used for
5851 Objective-C methods.
5853 The default name is a unique method number followed by the name of the
5854 class (e.g.@: @samp{_1_Foo}). For methods in categories, the name of
5855 the category is also included in the assembler name (e.g.@:
5858 These names are safe on most systems, but make debugging difficult since
5859 the method's selector is not present in the name. Therefore, particular
5860 systems define other ways of computing names.
5862 @var{buf} is an expression of type @code{char *} which gives you a
5863 buffer in which to store the name; its length is as long as
5864 @var{class_name}, @var{cat_name} and @var{sel_name} put together, plus
5865 50 characters extra.
5867 The argument @var{is_inst} specifies whether the method is an instance
5868 method or a class method; @var{class_name} is the name of the class;
5869 @var{cat_name} is the name of the category (or @code{NULL} if the method is not
5870 in a category); and @var{sel_name} is the name of the selector.
5872 On systems where the assembler can handle quoted names, you can use this
5873 macro to provide more human-readable names.
5876 @node Initialization
5877 @subsection How Initialization Functions Are Handled
5878 @cindex initialization routines
5879 @cindex termination routines
5880 @cindex constructors, output of
5881 @cindex destructors, output of
5883 The compiled code for certain languages includes @dfn{constructors}
5884 (also called @dfn{initialization routines})---functions to initialize
5885 data in the program when the program is started. These functions need
5886 to be called before the program is ``started''---that is to say, before
5887 @code{main} is called.
5889 Compiling some languages generates @dfn{destructors} (also called
5890 @dfn{termination routines}) that should be called when the program
5893 To make the initialization and termination functions work, the compiler
5894 must output something in the assembler code to cause those functions to
5895 be called at the appropriate time. When you port the compiler to a new
5896 system, you need to specify how to do this.
5898 There are two major ways that GCC currently supports the execution of
5899 initialization and termination functions. Each way has two variants.
5900 Much of the structure is common to all four variations.
5902 @findex __CTOR_LIST__
5903 @findex __DTOR_LIST__
5904 The linker must build two lists of these functions---a list of
5905 initialization functions, called @code{__CTOR_LIST__}, and a list of
5906 termination functions, called @code{__DTOR_LIST__}.
5908 Each list always begins with an ignored function pointer (which may hold
5909 0, @minus{}1, or a count of the function pointers after it, depending on
5910 the environment). This is followed by a series of zero or more function
5911 pointers to constructors (or destructors), followed by a function
5912 pointer containing zero.
5914 Depending on the operating system and its executable file format, either
5915 @file{crtstuff.c} or @file{libgcc2.c} traverses these lists at startup
5916 time and exit time. Constructors are called in reverse order of the
5917 list; destructors in forward order.
5919 The best way to handle static constructors works only for object file
5920 formats which provide arbitrarily-named sections. A section is set
5921 aside for a list of constructors, and another for a list of destructors.
5922 Traditionally these are called @samp{.ctors} and @samp{.dtors}. Each
5923 object file that defines an initialization function also puts a word in
5924 the constructor section to point to that function. The linker
5925 accumulates all these words into one contiguous @samp{.ctors} section.
5926 Termination functions are handled similarly.
5928 This method will be chosen as the default by @file{target-def.h} if
5929 @code{TARGET_ASM_NAMED_SECTION} is defined. A target that does not
5930 support arbitrary sections, but does support special designated
5931 constructor and destructor sections may define @code{CTORS_SECTION_ASM_OP}
5932 and @code{DTORS_SECTION_ASM_OP} to achieve the same effect.
5934 When arbitrary sections are available, there are two variants, depending
5935 upon how the code in @file{crtstuff.c} is called. On systems that
5936 support a @dfn{.init} section which is executed at program startup,
5937 parts of @file{crtstuff.c} are compiled into that section. The
5938 program is linked by the @command{gcc} driver like this:
5941 ld -o @var{output_file} crti.o crtbegin.o @dots{} -lgcc crtend.o crtn.o
5944 The prologue of a function (@code{__init}) appears in the @code{.init}
5945 section of @file{crti.o}; the epilogue appears in @file{crtn.o}. Likewise
5946 for the function @code{__fini} in the @dfn{.fini} section. Normally these
5947 files are provided by the operating system or by the GNU C library, but
5948 are provided by GCC for a few targets.
5950 The objects @file{crtbegin.o} and @file{crtend.o} are (for most targets)
5951 compiled from @file{crtstuff.c}. They contain, among other things, code
5952 fragments within the @code{.init} and @code{.fini} sections that branch
5953 to routines in the @code{.text} section. The linker will pull all parts
5954 of a section together, which results in a complete @code{__init} function
5955 that invokes the routines we need at startup.
5957 To use this variant, you must define the @code{INIT_SECTION_ASM_OP}
5960 If no init section is available, when GCC compiles any function called
5961 @code{main} (or more accurately, any function designated as a program
5962 entry point by the language front end calling @code{expand_main_function}),
5963 it inserts a procedure call to @code{__main} as the first executable code
5964 after the function prologue. The @code{__main} function is defined
5965 in @file{libgcc2.c} and runs the global constructors.
5967 In file formats that don't support arbitrary sections, there are again
5968 two variants. In the simplest variant, the GNU linker (GNU @code{ld})
5969 and an `a.out' format must be used. In this case,
5970 @code{TARGET_ASM_CONSTRUCTOR} is defined to produce a @code{.stabs}
5971 entry of type @samp{N_SETT}, referencing the name @code{__CTOR_LIST__},
5972 and with the address of the void function containing the initialization
5973 code as its value. The GNU linker recognizes this as a request to add
5974 the value to a @dfn{set}; the values are accumulated, and are eventually
5975 placed in the executable as a vector in the format described above, with
5976 a leading (ignored) count and a trailing zero element.
5977 @code{TARGET_ASM_DESTRUCTOR} is handled similarly. Since no init
5978 section is available, the absence of @code{INIT_SECTION_ASM_OP} causes
5979 the compilation of @code{main} to call @code{__main} as above, starting
5980 the initialization process.
5982 The last variant uses neither arbitrary sections nor the GNU linker.
5983 This is preferable when you want to do dynamic linking and when using
5984 file formats which the GNU linker does not support, such as `ECOFF'@. In
5985 this case, @code{TARGET_HAVE_CTORS_DTORS} is false, initialization and
5986 termination functions are recognized simply by their names. This requires
5987 an extra program in the linkage step, called @command{collect2}. This program
5988 pretends to be the linker, for use with GCC; it does its job by running
5989 the ordinary linker, but also arranges to include the vectors of
5990 initialization and termination functions. These functions are called
5991 via @code{__main} as described above. In order to use this method,
5992 @code{use_collect2} must be defined in the target in @file{config.gcc}.
5995 The following section describes the specific macros that control and
5996 customize the handling of initialization and termination functions.
5999 @node Macros for Initialization
6000 @subsection Macros Controlling Initialization Routines
6002 Here are the macros that control how the compiler handles initialization
6003 and termination functions:
6005 @defmac INIT_SECTION_ASM_OP
6006 If defined, a C string constant, including spacing, for the assembler
6007 operation to identify the following data as initialization code. If not
6008 defined, GCC will assume such a section does not exist. When you are
6009 using special sections for initialization and termination functions, this
6010 macro also controls how @file{crtstuff.c} and @file{libgcc2.c} arrange to
6011 run the initialization functions.
6014 @defmac HAS_INIT_SECTION
6015 If defined, @code{main} will not call @code{__main} as described above.
6016 This macro should be defined for systems that control start-up code
6017 on a symbol-by-symbol basis, such as OSF/1, and should not
6018 be defined explicitly for systems that support @code{INIT_SECTION_ASM_OP}.
6021 @defmac LD_INIT_SWITCH
6022 If defined, a C string constant for a switch that tells the linker that
6023 the following symbol is an initialization routine.
6026 @defmac LD_FINI_SWITCH
6027 If defined, a C string constant for a switch that tells the linker that
6028 the following symbol is a finalization routine.
6031 @defmac COLLECT_SHARED_INIT_FUNC (@var{stream}, @var{func})
6032 If defined, a C statement that will write a function that can be
6033 automatically called when a shared library is loaded. The function
6034 should call @var{func}, which takes no arguments. If not defined, and
6035 the object format requires an explicit initialization function, then a
6036 function called @code{_GLOBAL__DI} will be generated.
6038 This function and the following one are used by collect2 when linking a
6039 shared library that needs constructors or destructors, or has DWARF2
6040 exception tables embedded in the code.
6043 @defmac COLLECT_SHARED_FINI_FUNC (@var{stream}, @var{func})
6044 If defined, a C statement that will write a function that can be
6045 automatically called when a shared library is unloaded. The function
6046 should call @var{func}, which takes no arguments. If not defined, and
6047 the object format requires an explicit finalization function, then a
6048 function called @code{_GLOBAL__DD} will be generated.
6051 @defmac INVOKE__main
6052 If defined, @code{main} will call @code{__main} despite the presence of
6053 @code{INIT_SECTION_ASM_OP}. This macro should be defined for systems
6054 where the init section is not actually run automatically, but is still
6055 useful for collecting the lists of constructors and destructors.
6058 @defmac SUPPORTS_INIT_PRIORITY
6059 If nonzero, the C++ @code{init_priority} attribute is supported and the
6060 compiler should emit instructions to control the order of initialization
6061 of objects. If zero, the compiler will issue an error message upon
6062 encountering an @code{init_priority} attribute.
6065 @hook TARGET_HAVE_CTORS_DTORS
6067 @hook TARGET_ASM_CONSTRUCTOR
6069 @hook TARGET_ASM_DESTRUCTOR
6071 If @code{TARGET_HAVE_CTORS_DTORS} is true, the initialization routine
6072 generated for the generated object file will have static linkage.
6074 If your system uses @command{collect2} as the means of processing
6075 constructors, then that program normally uses @command{nm} to scan
6076 an object file for constructor functions to be called.
6078 On certain kinds of systems, you can define this macro to make
6079 @command{collect2} work faster (and, in some cases, make it work at all):
6081 @defmac OBJECT_FORMAT_COFF
6082 Define this macro if the system uses COFF (Common Object File Format)
6083 object files, so that @command{collect2} can assume this format and scan
6084 object files directly for dynamic constructor/destructor functions.
6086 This macro is effective only in a native compiler; @command{collect2} as
6087 part of a cross compiler always uses @command{nm} for the target machine.
6090 @defmac REAL_NM_FILE_NAME
6091 Define this macro as a C string constant containing the file name to use
6092 to execute @command{nm}. The default is to search the path normally for
6097 @command{collect2} calls @command{nm} to scan object files for static
6098 constructors and destructors and LTO info. By default, @option{-n} is
6099 passed. Define @code{NM_FLAGS} to a C string constant if other options
6100 are needed to get the same output format as GNU @command{nm -n}
6104 If your system supports shared libraries and has a program to list the
6105 dynamic dependencies of a given library or executable, you can define
6106 these macros to enable support for running initialization and
6107 termination functions in shared libraries:
6110 Define this macro to a C string constant containing the name of the program
6111 which lists dynamic dependencies, like @command{ldd} under SunOS 4.
6114 @defmac PARSE_LDD_OUTPUT (@var{ptr})
6115 Define this macro to be C code that extracts filenames from the output
6116 of the program denoted by @code{LDD_SUFFIX}. @var{ptr} is a variable
6117 of type @code{char *} that points to the beginning of a line of output
6118 from @code{LDD_SUFFIX}. If the line lists a dynamic dependency, the
6119 code must advance @var{ptr} to the beginning of the filename on that
6120 line. Otherwise, it must set @var{ptr} to @code{NULL}.
6123 @defmac SHLIB_SUFFIX
6124 Define this macro to a C string constant containing the default shared
6125 library extension of the target (e.g., @samp{".so"}). @command{collect2}
6126 strips version information after this suffix when generating global
6127 constructor and destructor names. This define is only needed on targets
6128 that use @command{collect2} to process constructors and destructors.
6131 @node Instruction Output
6132 @subsection Output of Assembler Instructions
6134 @c prevent bad page break with this line
6135 This describes assembler instruction output.
6137 @defmac REGISTER_NAMES
6138 A C initializer containing the assembler's names for the machine
6139 registers, each one as a C string constant. This is what translates
6140 register numbers in the compiler into assembler language.
6143 @defmac ADDITIONAL_REGISTER_NAMES
6144 If defined, a C initializer for an array of structures containing a name
6145 and a register number. This macro defines additional names for hard
6146 registers, thus allowing the @code{asm} option in declarations to refer
6147 to registers using alternate names.
6150 @defmac OVERLAPPING_REGISTER_NAMES
6151 If defined, a C initializer for an array of structures containing a
6152 name, a register number and a count of the number of consecutive
6153 machine registers the name overlaps. This macro defines additional
6154 names for hard registers, thus allowing the @code{asm} option in
6155 declarations to refer to registers using alternate names. Unlike
6156 @code{ADDITIONAL_REGISTER_NAMES}, this macro should be used when the
6157 register name implies multiple underlying registers.
6159 This macro should be used when it is important that a clobber in an
6160 @code{asm} statement clobbers all the underlying values implied by the
6161 register name. For example, on ARM, clobbering the double-precision
6162 VFP register ``d0'' implies clobbering both single-precision registers
6166 @defmac ASM_OUTPUT_OPCODE (@var{stream}, @var{ptr})
6167 Define this macro if you are using an unusual assembler that
6168 requires different names for the machine instructions.
6170 The definition is a C statement or statements which output an
6171 assembler instruction opcode to the stdio stream @var{stream}. The
6172 macro-operand @var{ptr} is a variable of type @code{char *} which
6173 points to the opcode name in its ``internal'' form---the form that is
6174 written in the machine description. The definition should output the
6175 opcode name to @var{stream}, performing any translation you desire, and
6176 increment the variable @var{ptr} to point at the end of the opcode
6177 so that it will not be output twice.
6179 In fact, your macro definition may process less than the entire opcode
6180 name, or more than the opcode name; but if you want to process text
6181 that includes @samp{%}-sequences to substitute operands, you must take
6182 care of the substitution yourself. Just be sure to increment
6183 @var{ptr} over whatever text should not be output normally.
6185 @findex recog_data.operand
6186 If you need to look at the operand values, they can be found as the
6187 elements of @code{recog_data.operand}.
6189 If the macro definition does nothing, the instruction is output
6193 @defmac FINAL_PRESCAN_INSN (@var{insn}, @var{opvec}, @var{noperands})
6194 If defined, a C statement to be executed just prior to the output of
6195 assembler code for @var{insn}, to modify the extracted operands so
6196 they will be output differently.
6198 Here the argument @var{opvec} is the vector containing the operands
6199 extracted from @var{insn}, and @var{noperands} is the number of
6200 elements of the vector which contain meaningful data for this insn.
6201 The contents of this vector are what will be used to convert the insn
6202 template into assembler code, so you can change the assembler output
6203 by changing the contents of the vector.
6205 This macro is useful when various assembler syntaxes share a single
6206 file of instruction patterns; by defining this macro differently, you
6207 can cause a large class of instructions to be output differently (such
6208 as with rearranged operands). Naturally, variations in assembler
6209 syntax affecting individual insn patterns ought to be handled by
6210 writing conditional output routines in those patterns.
6212 If this macro is not defined, it is equivalent to a null statement.
6215 @hook TARGET_ASM_FINAL_POSTSCAN_INSN
6217 @defmac PRINT_OPERAND (@var{stream}, @var{x}, @var{code})
6218 A C compound statement to output to stdio stream @var{stream} the
6219 assembler syntax for an instruction operand @var{x}. @var{x} is an
6222 @var{code} is a value that can be used to specify one of several ways
6223 of printing the operand. It is used when identical operands must be
6224 printed differently depending on the context. @var{code} comes from
6225 the @samp{%} specification that was used to request printing of the
6226 operand. If the specification was just @samp{%@var{digit}} then
6227 @var{code} is 0; if the specification was @samp{%@var{ltr}
6228 @var{digit}} then @var{code} is the ASCII code for @var{ltr}.
6231 If @var{x} is a register, this macro should print the register's name.
6232 The names can be found in an array @code{reg_names} whose type is
6233 @code{char *[]}. @code{reg_names} is initialized from
6234 @code{REGISTER_NAMES}.
6236 When the machine description has a specification @samp{%@var{punct}}
6237 (a @samp{%} followed by a punctuation character), this macro is called
6238 with a null pointer for @var{x} and the punctuation character for
6242 @defmac PRINT_OPERAND_PUNCT_VALID_P (@var{code})
6243 A C expression which evaluates to true if @var{code} is a valid
6244 punctuation character for use in the @code{PRINT_OPERAND} macro. If
6245 @code{PRINT_OPERAND_PUNCT_VALID_P} is not defined, it means that no
6246 punctuation characters (except for the standard one, @samp{%}) are used
6250 @defmac PRINT_OPERAND_ADDRESS (@var{stream}, @var{x})
6251 A C compound statement to output to stdio stream @var{stream} the
6252 assembler syntax for an instruction operand that is a memory reference
6253 whose address is @var{x}. @var{x} is an RTL expression.
6255 @cindex @code{TARGET_ENCODE_SECTION_INFO} usage
6256 On some machines, the syntax for a symbolic address depends on the
6257 section that the address refers to. On these machines, define the hook
6258 @code{TARGET_ENCODE_SECTION_INFO} to store the information into the
6259 @code{symbol_ref}, and then check for it here. @xref{Assembler
6263 @findex dbr_sequence_length
6264 @defmac DBR_OUTPUT_SEQEND (@var{file})
6265 A C statement, to be executed after all slot-filler instructions have
6266 been output. If necessary, call @code{dbr_sequence_length} to
6267 determine the number of slots filled in a sequence (zero if not
6268 currently outputting a sequence), to decide how many no-ops to output,
6271 Don't define this macro if it has nothing to do, but it is helpful in
6272 reading assembly output if the extent of the delay sequence is made
6273 explicit (e.g.@: with white space).
6276 @findex final_sequence
6277 Note that output routines for instructions with delay slots must be
6278 prepared to deal with not being output as part of a sequence
6279 (i.e.@: when the scheduling pass is not run, or when no slot fillers could be
6280 found.) The variable @code{final_sequence} is null when not
6281 processing a sequence, otherwise it contains the @code{sequence} rtx
6285 @defmac REGISTER_PREFIX
6286 @defmacx LOCAL_LABEL_PREFIX
6287 @defmacx USER_LABEL_PREFIX
6288 @defmacx IMMEDIATE_PREFIX
6289 If defined, C string expressions to be used for the @samp{%R}, @samp{%L},
6290 @samp{%U}, and @samp{%I} options of @code{asm_fprintf} (see
6291 @file{final.c}). These are useful when a single @file{md} file must
6292 support multiple assembler formats. In that case, the various @file{tm.h}
6293 files can define these macros differently.
6296 @defmac ASM_FPRINTF_EXTENSIONS (@var{file}, @var{argptr}, @var{format})
6297 If defined this macro should expand to a series of @code{case}
6298 statements which will be parsed inside the @code{switch} statement of
6299 the @code{asm_fprintf} function. This allows targets to define extra
6300 printf formats which may useful when generating their assembler
6301 statements. Note that uppercase letters are reserved for future
6302 generic extensions to asm_fprintf, and so are not available to target
6303 specific code. The output file is given by the parameter @var{file}.
6304 The varargs input pointer is @var{argptr} and the rest of the format
6305 string, starting the character after the one that is being switched
6306 upon, is pointed to by @var{format}.
6309 @defmac ASSEMBLER_DIALECT
6310 If your target supports multiple dialects of assembler language (such as
6311 different opcodes), define this macro as a C expression that gives the
6312 numeric index of the assembler language dialect to use, with zero as the
6315 If this macro is defined, you may use constructs of the form
6317 @samp{@{option0|option1|option2@dots{}@}}
6320 in the output templates of patterns (@pxref{Output Template}) or in the
6321 first argument of @code{asm_fprintf}. This construct outputs
6322 @samp{option0}, @samp{option1}, @samp{option2}, etc., if the value of
6323 @code{ASSEMBLER_DIALECT} is zero, one, two, etc. Any special characters
6324 within these strings retain their usual meaning. If there are fewer
6325 alternatives within the braces than the value of
6326 @code{ASSEMBLER_DIALECT}, the construct outputs nothing. If it's needed
6327 to print curly braces or @samp{|} character in assembler output directly,
6328 @samp{%@{}, @samp{%@}} and @samp{%|} can be used.
6330 If you do not define this macro, the characters @samp{@{}, @samp{|} and
6331 @samp{@}} do not have any special meaning when used in templates or
6332 operands to @code{asm_fprintf}.
6334 Define the macros @code{REGISTER_PREFIX}, @code{LOCAL_LABEL_PREFIX},
6335 @code{USER_LABEL_PREFIX} and @code{IMMEDIATE_PREFIX} if you can express
6336 the variations in assembler language syntax with that mechanism. Define
6337 @code{ASSEMBLER_DIALECT} and use the @samp{@{option0|option1@}} syntax
6338 if the syntax variant are larger and involve such things as different
6339 opcodes or operand order.
6342 @defmac ASM_OUTPUT_REG_PUSH (@var{stream}, @var{regno})
6343 A C expression to output to @var{stream} some assembler code
6344 which will push hard register number @var{regno} onto the stack.
6345 The code need not be optimal, since this macro is used only when
6349 @defmac ASM_OUTPUT_REG_POP (@var{stream}, @var{regno})
6350 A C expression to output to @var{stream} some assembler code
6351 which will pop hard register number @var{regno} off of the stack.
6352 The code need not be optimal, since this macro is used only when
6356 @node Dispatch Tables
6357 @subsection Output of Dispatch Tables
6359 @c prevent bad page break with this line
6360 This concerns dispatch tables.
6362 @cindex dispatch table
6363 @defmac ASM_OUTPUT_ADDR_DIFF_ELT (@var{stream}, @var{body}, @var{value}, @var{rel})
6364 A C statement to output to the stdio stream @var{stream} an assembler
6365 pseudo-instruction to generate a difference between two labels.
6366 @var{value} and @var{rel} are the numbers of two internal labels. The
6367 definitions of these labels are output using
6368 @code{(*targetm.asm_out.internal_label)}, and they must be printed in the same
6369 way here. For example,
6372 fprintf (@var{stream}, "\t.word L%d-L%d\n",
6373 @var{value}, @var{rel})
6376 You must provide this macro on machines where the addresses in a
6377 dispatch table are relative to the table's own address. If defined, GCC
6378 will also use this macro on all machines when producing PIC@.
6379 @var{body} is the body of the @code{ADDR_DIFF_VEC}; it is provided so that the
6380 mode and flags can be read.
6383 @defmac ASM_OUTPUT_ADDR_VEC_ELT (@var{stream}, @var{value})
6384 This macro should be provided on machines where the addresses
6385 in a dispatch table are absolute.
6387 The definition should be a C statement to output to the stdio stream
6388 @var{stream} an assembler pseudo-instruction to generate a reference to
6389 a label. @var{value} is the number of an internal label whose
6390 definition is output using @code{(*targetm.asm_out.internal_label)}.
6394 fprintf (@var{stream}, "\t.word L%d\n", @var{value})
6398 @defmac ASM_OUTPUT_CASE_LABEL (@var{stream}, @var{prefix}, @var{num}, @var{table})
6399 Define this if the label before a jump-table needs to be output
6400 specially. The first three arguments are the same as for
6401 @code{(*targetm.asm_out.internal_label)}; the fourth argument is the
6402 jump-table which follows (a @code{jump_table_data} containing an
6403 @code{addr_vec} or @code{addr_diff_vec}).
6405 This feature is used on system V to output a @code{swbeg} statement
6408 If this macro is not defined, these labels are output with
6409 @code{(*targetm.asm_out.internal_label)}.
6412 @defmac ASM_OUTPUT_CASE_END (@var{stream}, @var{num}, @var{table})
6413 Define this if something special must be output at the end of a
6414 jump-table. The definition should be a C statement to be executed
6415 after the assembler code for the table is written. It should write
6416 the appropriate code to stdio stream @var{stream}. The argument
6417 @var{table} is the jump-table insn, and @var{num} is the label-number
6418 of the preceding label.
6420 If this macro is not defined, nothing special is output at the end of
6424 @hook TARGET_ASM_EMIT_UNWIND_LABEL
6426 @hook TARGET_ASM_EMIT_EXCEPT_TABLE_LABEL
6428 @hook TARGET_ASM_EMIT_EXCEPT_PERSONALITY
6430 @hook TARGET_ASM_UNWIND_EMIT
6432 @hook TARGET_ASM_UNWIND_EMIT_BEFORE_INSN
6434 @node Exception Region Output
6435 @subsection Assembler Commands for Exception Regions
6437 @c prevent bad page break with this line
6439 This describes commands marking the start and the end of an exception
6442 @defmac EH_FRAME_SECTION_NAME
6443 If defined, a C string constant for the name of the section containing
6444 exception handling frame unwind information. If not defined, GCC will
6445 provide a default definition if the target supports named sections.
6446 @file{crtstuff.c} uses this macro to switch to the appropriate section.
6448 You should define this symbol if your target supports DWARF 2 frame
6449 unwind information and the default definition does not work.
6452 @defmac EH_FRAME_IN_DATA_SECTION
6453 If defined, DWARF 2 frame unwind information will be placed in the
6454 data section even though the target supports named sections. This
6455 might be necessary, for instance, if the system linker does garbage
6456 collection and sections cannot be marked as not to be collected.
6458 Do not define this macro unless @code{TARGET_ASM_NAMED_SECTION} is
6462 @defmac EH_TABLES_CAN_BE_READ_ONLY
6463 Define this macro to 1 if your target is such that no frame unwind
6464 information encoding used with non-PIC code will ever require a
6465 runtime relocation, but the linker may not support merging read-only
6466 and read-write sections into a single read-write section.
6469 @defmac MASK_RETURN_ADDR
6470 An rtx used to mask the return address found via @code{RETURN_ADDR_RTX}, so
6471 that it does not contain any extraneous set bits in it.
6474 @defmac DWARF2_UNWIND_INFO
6475 Define this macro to 0 if your target supports DWARF 2 frame unwind
6476 information, but it does not yet work with exception handling.
6477 Otherwise, if your target supports this information (if it defines
6478 @code{INCOMING_RETURN_ADDR_RTX} and @code{OBJECT_FORMAT_ELF}),
6479 GCC will provide a default definition of 1.
6482 @hook TARGET_EXCEPT_UNWIND_INFO
6483 This hook defines the mechanism that will be used for exception handling
6484 by the target. If the target has ABI specified unwind tables, the hook
6485 should return @code{UI_TARGET}. If the target is to use the
6486 @code{setjmp}/@code{longjmp}-based exception handling scheme, the hook
6487 should return @code{UI_SJLJ}. If the target supports DWARF 2 frame unwind
6488 information, the hook should return @code{UI_DWARF2}.
6490 A target may, if exceptions are disabled, choose to return @code{UI_NONE}.
6491 This may end up simplifying other parts of target-specific code. The
6492 default implementation of this hook never returns @code{UI_NONE}.
6494 Note that the value returned by this hook should be constant. It should
6495 not depend on anything except the command-line switches described by
6496 @var{opts}. In particular, the
6497 setting @code{UI_SJLJ} must be fixed at compiler start-up as C pre-processor
6498 macros and builtin functions related to exception handling are set up
6499 depending on this setting.
6501 The default implementation of the hook first honors the
6502 @option{--enable-sjlj-exceptions} configure option, then
6503 @code{DWARF2_UNWIND_INFO}, and finally defaults to @code{UI_SJLJ}. If
6504 @code{DWARF2_UNWIND_INFO} depends on command-line options, the target
6505 must define this hook so that @var{opts} is used correctly.
6508 @hook TARGET_UNWIND_TABLES_DEFAULT
6509 This variable should be set to @code{true} if the target ABI requires unwinding
6510 tables even when exceptions are not used. It must not be modified by
6511 command-line option processing.
6514 @defmac DONT_USE_BUILTIN_SETJMP
6515 Define this macro to 1 if the @code{setjmp}/@code{longjmp}-based scheme
6516 should use the @code{setjmp}/@code{longjmp} functions from the C library
6517 instead of the @code{__builtin_setjmp}/@code{__builtin_longjmp} machinery.
6520 @defmac JMP_BUF_SIZE
6521 This macro has no effect unless @code{DONT_USE_BUILTIN_SETJMP} is also
6522 defined. Define this macro if the default size of @code{jmp_buf} buffer
6523 for the @code{setjmp}/@code{longjmp}-based exception handling mechanism
6524 is not large enough, or if it is much too large.
6525 The default size is @code{FIRST_PSEUDO_REGISTER * sizeof(void *)}.
6528 @defmac DWARF_CIE_DATA_ALIGNMENT
6529 This macro need only be defined if the target might save registers in the
6530 function prologue at an offset to the stack pointer that is not aligned to
6531 @code{UNITS_PER_WORD}. The definition should be the negative minimum
6532 alignment if @code{STACK_GROWS_DOWNWARD} is defined, and the positive
6533 minimum alignment otherwise. @xref{SDB and DWARF}. Only applicable if
6534 the target supports DWARF 2 frame unwind information.
6537 @hook TARGET_TERMINATE_DW2_EH_FRAME_INFO
6539 @hook TARGET_DWARF_REGISTER_SPAN
6541 @hook TARGET_DWARF_FRAME_REG_MODE
6543 @hook TARGET_INIT_DWARF_REG_SIZES_EXTRA
6545 @hook TARGET_ASM_TTYPE
6547 @hook TARGET_ARM_EABI_UNWINDER
6549 @node Alignment Output
6550 @subsection Assembler Commands for Alignment
6552 @c prevent bad page break with this line
6553 This describes commands for alignment.
6555 @defmac JUMP_ALIGN (@var{label})
6556 The alignment (log base 2) to put in front of @var{label}, which is
6557 a common destination of jumps and has no fallthru incoming edge.
6559 This macro need not be defined if you don't want any special alignment
6560 to be done at such a time. Most machine descriptions do not currently
6563 Unless it's necessary to inspect the @var{label} parameter, it is better
6564 to set the variable @var{align_jumps} in the target's
6565 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
6566 selection in @var{align_jumps} in a @code{JUMP_ALIGN} implementation.
6569 @hook TARGET_ASM_JUMP_ALIGN_MAX_SKIP
6571 @defmac LABEL_ALIGN_AFTER_BARRIER (@var{label})
6572 The alignment (log base 2) to put in front of @var{label}, which follows
6575 This macro need not be defined if you don't want any special alignment
6576 to be done at such a time. Most machine descriptions do not currently
6580 @hook TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP
6582 @defmac LOOP_ALIGN (@var{label})
6583 The alignment (log base 2) to put in front of @var{label} that heads
6584 a frequently executed basic block (usually the header of a loop).
6586 This macro need not be defined if you don't want any special alignment
6587 to be done at such a time. Most machine descriptions do not currently
6590 Unless it's necessary to inspect the @var{label} parameter, it is better
6591 to set the variable @code{align_loops} in the target's
6592 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
6593 selection in @code{align_loops} in a @code{LOOP_ALIGN} implementation.
6596 @hook TARGET_ASM_LOOP_ALIGN_MAX_SKIP
6598 @defmac LABEL_ALIGN (@var{label})
6599 The alignment (log base 2) to put in front of @var{label}.
6600 If @code{LABEL_ALIGN_AFTER_BARRIER} / @code{LOOP_ALIGN} specify a different alignment,
6601 the maximum of the specified values is used.
6603 Unless it's necessary to inspect the @var{label} parameter, it is better
6604 to set the variable @code{align_labels} in the target's
6605 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
6606 selection in @code{align_labels} in a @code{LABEL_ALIGN} implementation.
6609 @hook TARGET_ASM_LABEL_ALIGN_MAX_SKIP
6611 @defmac ASM_OUTPUT_SKIP (@var{stream}, @var{nbytes})
6612 A C statement to output to the stdio stream @var{stream} an assembler
6613 instruction to advance the location counter by @var{nbytes} bytes.
6614 Those bytes should be zero when loaded. @var{nbytes} will be a C
6615 expression of type @code{unsigned HOST_WIDE_INT}.
6618 @defmac ASM_NO_SKIP_IN_TEXT
6619 Define this macro if @code{ASM_OUTPUT_SKIP} should not be used in the
6620 text section because it fails to put zeros in the bytes that are skipped.
6621 This is true on many Unix systems, where the pseudo--op to skip bytes
6622 produces no-op instructions rather than zeros when used in the text
6626 @defmac ASM_OUTPUT_ALIGN (@var{stream}, @var{power})
6627 A C statement to output to the stdio stream @var{stream} an assembler
6628 command to advance the location counter to a multiple of 2 to the
6629 @var{power} bytes. @var{power} will be a C expression of type @code{int}.
6632 @defmac ASM_OUTPUT_ALIGN_WITH_NOP (@var{stream}, @var{power})
6633 Like @code{ASM_OUTPUT_ALIGN}, except that the ``nop'' instruction is used
6634 for padding, if necessary.
6637 @defmac ASM_OUTPUT_MAX_SKIP_ALIGN (@var{stream}, @var{power}, @var{max_skip})
6638 A C statement to output to the stdio stream @var{stream} an assembler
6639 command to advance the location counter to a multiple of 2 to the
6640 @var{power} bytes, but only if @var{max_skip} or fewer bytes are needed to
6641 satisfy the alignment request. @var{power} and @var{max_skip} will be
6642 a C expression of type @code{int}.
6646 @node Debugging Info
6647 @section Controlling Debugging Information Format
6649 @c prevent bad page break with this line
6650 This describes how to specify debugging information.
6653 * All Debuggers:: Macros that affect all debugging formats uniformly.
6654 * DBX Options:: Macros enabling specific options in DBX format.
6655 * DBX Hooks:: Hook macros for varying DBX format.
6656 * File Names and DBX:: Macros controlling output of file names in DBX format.
6657 * SDB and DWARF:: Macros for SDB (COFF) and DWARF formats.
6658 * VMS Debug:: Macros for VMS debug format.
6662 @subsection Macros Affecting All Debugging Formats
6664 @c prevent bad page break with this line
6665 These macros affect all debugging formats.
6667 @defmac DBX_REGISTER_NUMBER (@var{regno})
6668 A C expression that returns the DBX register number for the compiler
6669 register number @var{regno}. In the default macro provided, the value
6670 of this expression will be @var{regno} itself. But sometimes there are
6671 some registers that the compiler knows about and DBX does not, or vice
6672 versa. In such cases, some register may need to have one number in the
6673 compiler and another for DBX@.
6675 If two registers have consecutive numbers inside GCC, and they can be
6676 used as a pair to hold a multiword value, then they @emph{must} have
6677 consecutive numbers after renumbering with @code{DBX_REGISTER_NUMBER}.
6678 Otherwise, debuggers will be unable to access such a pair, because they
6679 expect register pairs to be consecutive in their own numbering scheme.
6681 If you find yourself defining @code{DBX_REGISTER_NUMBER} in way that
6682 does not preserve register pairs, then what you must do instead is
6683 redefine the actual register numbering scheme.
6686 @defmac DEBUGGER_AUTO_OFFSET (@var{x})
6687 A C expression that returns the integer offset value for an automatic
6688 variable having address @var{x} (an RTL expression). The default
6689 computation assumes that @var{x} is based on the frame-pointer and
6690 gives the offset from the frame-pointer. This is required for targets
6691 that produce debugging output for DBX or COFF-style debugging output
6692 for SDB and allow the frame-pointer to be eliminated when the
6693 @option{-g} options is used.
6696 @defmac DEBUGGER_ARG_OFFSET (@var{offset}, @var{x})
6697 A C expression that returns the integer offset value for an argument
6698 having address @var{x} (an RTL expression). The nominal offset is
6702 @defmac PREFERRED_DEBUGGING_TYPE
6703 A C expression that returns the type of debugging output GCC should
6704 produce when the user specifies just @option{-g}. Define
6705 this if you have arranged for GCC to support more than one format of
6706 debugging output. Currently, the allowable values are @code{DBX_DEBUG},
6707 @code{SDB_DEBUG}, @code{DWARF_DEBUG}, @code{DWARF2_DEBUG},
6708 @code{XCOFF_DEBUG}, @code{VMS_DEBUG}, and @code{VMS_AND_DWARF2_DEBUG}.
6710 When the user specifies @option{-ggdb}, GCC normally also uses the
6711 value of this macro to select the debugging output format, but with two
6712 exceptions. If @code{DWARF2_DEBUGGING_INFO} is defined, GCC uses the
6713 value @code{DWARF2_DEBUG}. Otherwise, if @code{DBX_DEBUGGING_INFO} is
6714 defined, GCC uses @code{DBX_DEBUG}.
6716 The value of this macro only affects the default debugging output; the
6717 user can always get a specific type of output by using @option{-gstabs},
6718 @option{-gcoff}, @option{-gdwarf-2}, @option{-gxcoff}, or @option{-gvms}.
6722 @subsection Specific Options for DBX Output
6724 @c prevent bad page break with this line
6725 These are specific options for DBX output.
6727 @defmac DBX_DEBUGGING_INFO
6728 Define this macro if GCC should produce debugging output for DBX
6729 in response to the @option{-g} option.
6732 @defmac XCOFF_DEBUGGING_INFO
6733 Define this macro if GCC should produce XCOFF format debugging output
6734 in response to the @option{-g} option. This is a variant of DBX format.
6737 @defmac DEFAULT_GDB_EXTENSIONS
6738 Define this macro to control whether GCC should by default generate
6739 GDB's extended version of DBX debugging information (assuming DBX-format
6740 debugging information is enabled at all). If you don't define the
6741 macro, the default is 1: always generate the extended information
6742 if there is any occasion to.
6745 @defmac DEBUG_SYMS_TEXT
6746 Define this macro if all @code{.stabs} commands should be output while
6747 in the text section.
6750 @defmac ASM_STABS_OP
6751 A C string constant, including spacing, naming the assembler pseudo op to
6752 use instead of @code{"\t.stabs\t"} to define an ordinary debugging symbol.
6753 If you don't define this macro, @code{"\t.stabs\t"} is used. This macro
6754 applies only to DBX debugging information format.
6757 @defmac ASM_STABD_OP
6758 A C string constant, including spacing, naming the assembler pseudo op to
6759 use instead of @code{"\t.stabd\t"} to define a debugging symbol whose
6760 value is the current location. If you don't define this macro,
6761 @code{"\t.stabd\t"} is used. This macro applies only to DBX debugging
6765 @defmac ASM_STABN_OP
6766 A C string constant, including spacing, naming the assembler pseudo op to
6767 use instead of @code{"\t.stabn\t"} to define a debugging symbol with no
6768 name. If you don't define this macro, @code{"\t.stabn\t"} is used. This
6769 macro applies only to DBX debugging information format.
6772 @defmac DBX_NO_XREFS
6773 Define this macro if DBX on your system does not support the construct
6774 @samp{xs@var{tagname}}. On some systems, this construct is used to
6775 describe a forward reference to a structure named @var{tagname}.
6776 On other systems, this construct is not supported at all.
6779 @defmac DBX_CONTIN_LENGTH
6780 A symbol name in DBX-format debugging information is normally
6781 continued (split into two separate @code{.stabs} directives) when it
6782 exceeds a certain length (by default, 80 characters). On some
6783 operating systems, DBX requires this splitting; on others, splitting
6784 must not be done. You can inhibit splitting by defining this macro
6785 with the value zero. You can override the default splitting-length by
6786 defining this macro as an expression for the length you desire.
6789 @defmac DBX_CONTIN_CHAR
6790 Normally continuation is indicated by adding a @samp{\} character to
6791 the end of a @code{.stabs} string when a continuation follows. To use
6792 a different character instead, define this macro as a character
6793 constant for the character you want to use. Do not define this macro
6794 if backslash is correct for your system.
6797 @defmac DBX_STATIC_STAB_DATA_SECTION
6798 Define this macro if it is necessary to go to the data section before
6799 outputting the @samp{.stabs} pseudo-op for a non-global static
6803 @defmac DBX_TYPE_DECL_STABS_CODE
6804 The value to use in the ``code'' field of the @code{.stabs} directive
6805 for a typedef. The default is @code{N_LSYM}.
6808 @defmac DBX_STATIC_CONST_VAR_CODE
6809 The value to use in the ``code'' field of the @code{.stabs} directive
6810 for a static variable located in the text section. DBX format does not
6811 provide any ``right'' way to do this. The default is @code{N_FUN}.
6814 @defmac DBX_REGPARM_STABS_CODE
6815 The value to use in the ``code'' field of the @code{.stabs} directive
6816 for a parameter passed in registers. DBX format does not provide any
6817 ``right'' way to do this. The default is @code{N_RSYM}.
6820 @defmac DBX_REGPARM_STABS_LETTER
6821 The letter to use in DBX symbol data to identify a symbol as a parameter
6822 passed in registers. DBX format does not customarily provide any way to
6823 do this. The default is @code{'P'}.
6826 @defmac DBX_FUNCTION_FIRST
6827 Define this macro if the DBX information for a function and its
6828 arguments should precede the assembler code for the function. Normally,
6829 in DBX format, the debugging information entirely follows the assembler
6833 @defmac DBX_BLOCKS_FUNCTION_RELATIVE
6834 Define this macro, with value 1, if the value of a symbol describing
6835 the scope of a block (@code{N_LBRAC} or @code{N_RBRAC}) should be
6836 relative to the start of the enclosing function. Normally, GCC uses
6837 an absolute address.
6840 @defmac DBX_LINES_FUNCTION_RELATIVE
6841 Define this macro, with value 1, if the value of a symbol indicating
6842 the current line number (@code{N_SLINE}) should be relative to the
6843 start of the enclosing function. Normally, GCC uses an absolute address.
6846 @defmac DBX_USE_BINCL
6847 Define this macro if GCC should generate @code{N_BINCL} and
6848 @code{N_EINCL} stabs for included header files, as on Sun systems. This
6849 macro also directs GCC to output a type number as a pair of a file
6850 number and a type number within the file. Normally, GCC does not
6851 generate @code{N_BINCL} or @code{N_EINCL} stabs, and it outputs a single
6852 number for a type number.
6856 @subsection Open-Ended Hooks for DBX Format
6858 @c prevent bad page break with this line
6859 These are hooks for DBX format.
6861 @defmac DBX_OUTPUT_SOURCE_LINE (@var{stream}, @var{line}, @var{counter})
6862 A C statement to output DBX debugging information before code for line
6863 number @var{line} of the current source file to the stdio stream
6864 @var{stream}. @var{counter} is the number of time the macro was
6865 invoked, including the current invocation; it is intended to generate
6866 unique labels in the assembly output.
6868 This macro should not be defined if the default output is correct, or
6869 if it can be made correct by defining @code{DBX_LINES_FUNCTION_RELATIVE}.
6872 @defmac NO_DBX_FUNCTION_END
6873 Some stabs encapsulation formats (in particular ECOFF), cannot handle the
6874 @code{.stabs "",N_FUN,,0,0,Lscope-function-1} gdb dbx extension construct.
6875 On those machines, define this macro to turn this feature off without
6876 disturbing the rest of the gdb extensions.
6879 @defmac NO_DBX_BNSYM_ENSYM
6880 Some assemblers cannot handle the @code{.stabd BNSYM/ENSYM,0,0} gdb dbx
6881 extension construct. On those machines, define this macro to turn this
6882 feature off without disturbing the rest of the gdb extensions.
6885 @node File Names and DBX
6886 @subsection File Names in DBX Format
6888 @c prevent bad page break with this line
6889 This describes file names in DBX format.
6891 @defmac DBX_OUTPUT_MAIN_SOURCE_FILENAME (@var{stream}, @var{name})
6892 A C statement to output DBX debugging information to the stdio stream
6893 @var{stream}, which indicates that file @var{name} is the main source
6894 file---the file specified as the input file for compilation.
6895 This macro is called only once, at the beginning of compilation.
6897 This macro need not be defined if the standard form of output
6898 for DBX debugging information is appropriate.
6900 It may be necessary to refer to a label equal to the beginning of the
6901 text section. You can use @samp{assemble_name (stream, ltext_label_name)}
6902 to do so. If you do this, you must also set the variable
6903 @var{used_ltext_label_name} to @code{true}.
6906 @defmac NO_DBX_MAIN_SOURCE_DIRECTORY
6907 Define this macro, with value 1, if GCC should not emit an indication
6908 of the current directory for compilation and current source language at
6909 the beginning of the file.
6912 @defmac NO_DBX_GCC_MARKER
6913 Define this macro, with value 1, if GCC should not emit an indication
6914 that this object file was compiled by GCC@. The default is to emit
6915 an @code{N_OPT} stab at the beginning of every source file, with
6916 @samp{gcc2_compiled.} for the string and value 0.
6919 @defmac DBX_OUTPUT_MAIN_SOURCE_FILE_END (@var{stream}, @var{name})
6920 A C statement to output DBX debugging information at the end of
6921 compilation of the main source file @var{name}. Output should be
6922 written to the stdio stream @var{stream}.
6924 If you don't define this macro, nothing special is output at the end
6925 of compilation, which is correct for most machines.
6928 @defmac DBX_OUTPUT_NULL_N_SO_AT_MAIN_SOURCE_FILE_END
6929 Define this macro @emph{instead of} defining
6930 @code{DBX_OUTPUT_MAIN_SOURCE_FILE_END}, if what needs to be output at
6931 the end of compilation is an @code{N_SO} stab with an empty string,
6932 whose value is the highest absolute text address in the file.
6937 @subsection Macros for SDB and DWARF Output
6939 @c prevent bad page break with this line
6940 Here are macros for SDB and DWARF output.
6942 @defmac SDB_DEBUGGING_INFO
6943 Define this macro if GCC should produce COFF-style debugging output
6944 for SDB in response to the @option{-g} option.
6947 @defmac DWARF2_DEBUGGING_INFO
6948 Define this macro if GCC should produce dwarf version 2 format
6949 debugging output in response to the @option{-g} option.
6951 @hook TARGET_DWARF_CALLING_CONVENTION
6953 To support optional call frame debugging information, you must also
6954 define @code{INCOMING_RETURN_ADDR_RTX} and either set
6955 @code{RTX_FRAME_RELATED_P} on the prologue insns if you use RTL for the
6956 prologue, or call @code{dwarf2out_def_cfa} and @code{dwarf2out_reg_save}
6957 as appropriate from @code{TARGET_ASM_FUNCTION_PROLOGUE} if you don't.
6960 @defmac DWARF2_FRAME_INFO
6961 Define this macro to a nonzero value if GCC should always output
6962 Dwarf 2 frame information. If @code{TARGET_EXCEPT_UNWIND_INFO}
6963 (@pxref{Exception Region Output}) returns @code{UI_DWARF2}, and
6964 exceptions are enabled, GCC will output this information not matter
6965 how you define @code{DWARF2_FRAME_INFO}.
6968 @hook TARGET_DEBUG_UNWIND_INFO
6970 @defmac DWARF2_ASM_LINE_DEBUG_INFO
6971 Define this macro to be a nonzero value if the assembler can generate Dwarf 2
6972 line debug info sections. This will result in much more compact line number
6973 tables, and hence is desirable if it works.
6976 @hook TARGET_WANT_DEBUG_PUB_SECTIONS
6978 @hook TARGET_FORCE_AT_COMP_DIR
6980 @hook TARGET_DELAY_SCHED2
6982 @hook TARGET_DELAY_VARTRACK
6984 @hook TARGET_NO_REGISTER_ALLOCATION
6986 @defmac ASM_OUTPUT_DWARF_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
6987 A C statement to issue assembly directives that create a difference
6988 @var{lab1} minus @var{lab2}, using an integer of the given @var{size}.
6991 @defmac ASM_OUTPUT_DWARF_VMS_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
6992 A C statement to issue assembly directives that create a difference
6993 between the two given labels in system defined units, e.g. instruction
6994 slots on IA64 VMS, using an integer of the given size.
6997 @defmac ASM_OUTPUT_DWARF_OFFSET (@var{stream}, @var{size}, @var{label}, @var{section})
6998 A C statement to issue assembly directives that create a
6999 section-relative reference to the given @var{label}, using an integer of the
7000 given @var{size}. The label is known to be defined in the given @var{section}.
7003 @defmac ASM_OUTPUT_DWARF_PCREL (@var{stream}, @var{size}, @var{label})
7004 A C statement to issue assembly directives that create a self-relative
7005 reference to the given @var{label}, using an integer of the given @var{size}.
7008 @defmac ASM_OUTPUT_DWARF_TABLE_REF (@var{label})
7009 A C statement to issue assembly directives that create a reference to
7010 the DWARF table identifier @var{label} from the current section. This
7011 is used on some systems to avoid garbage collecting a DWARF table which
7012 is referenced by a function.
7015 @hook TARGET_ASM_OUTPUT_DWARF_DTPREL
7017 @defmac PUT_SDB_@dots{}
7018 Define these macros to override the assembler syntax for the special
7019 SDB assembler directives. See @file{sdbout.c} for a list of these
7020 macros and their arguments. If the standard syntax is used, you need
7021 not define them yourself.
7025 Some assemblers do not support a semicolon as a delimiter, even between
7026 SDB assembler directives. In that case, define this macro to be the
7027 delimiter to use (usually @samp{\n}). It is not necessary to define
7028 a new set of @code{PUT_SDB_@var{op}} macros if this is the only change
7032 @defmac SDB_ALLOW_UNKNOWN_REFERENCES
7033 Define this macro to allow references to unknown structure,
7034 union, or enumeration tags to be emitted. Standard COFF does not
7035 allow handling of unknown references, MIPS ECOFF has support for
7039 @defmac SDB_ALLOW_FORWARD_REFERENCES
7040 Define this macro to allow references to structure, union, or
7041 enumeration tags that have not yet been seen to be handled. Some
7042 assemblers choke if forward tags are used, while some require it.
7045 @defmac SDB_OUTPUT_SOURCE_LINE (@var{stream}, @var{line})
7046 A C statement to output SDB debugging information before code for line
7047 number @var{line} of the current source file to the stdio stream
7048 @var{stream}. The default is to emit an @code{.ln} directive.
7053 @subsection Macros for VMS Debug Format
7055 @c prevent bad page break with this line
7056 Here are macros for VMS debug format.
7058 @defmac VMS_DEBUGGING_INFO
7059 Define this macro if GCC should produce debugging output for VMS
7060 in response to the @option{-g} option. The default behavior for VMS
7061 is to generate minimal debug info for a traceback in the absence of
7062 @option{-g} unless explicitly overridden with @option{-g0}. This
7063 behavior is controlled by @code{TARGET_OPTION_OPTIMIZATION} and
7064 @code{TARGET_OPTION_OVERRIDE}.
7067 @node Floating Point
7068 @section Cross Compilation and Floating Point
7069 @cindex cross compilation and floating point
7070 @cindex floating point and cross compilation
7072 While all modern machines use twos-complement representation for integers,
7073 there are a variety of representations for floating point numbers. This
7074 means that in a cross-compiler the representation of floating point numbers
7075 in the compiled program may be different from that used in the machine
7076 doing the compilation.
7078 Because different representation systems may offer different amounts of
7079 range and precision, all floating point constants must be represented in
7080 the target machine's format. Therefore, the cross compiler cannot
7081 safely use the host machine's floating point arithmetic; it must emulate
7082 the target's arithmetic. To ensure consistency, GCC always uses
7083 emulation to work with floating point values, even when the host and
7084 target floating point formats are identical.
7086 The following macros are provided by @file{real.h} for the compiler to
7087 use. All parts of the compiler which generate or optimize
7088 floating-point calculations must use these macros. They may evaluate
7089 their operands more than once, so operands must not have side effects.
7091 @defmac REAL_VALUE_TYPE
7092 The C data type to be used to hold a floating point value in the target
7093 machine's format. Typically this is a @code{struct} containing an
7094 array of @code{HOST_WIDE_INT}, but all code should treat it as an opaque
7098 @deftypefn Macro int REAL_VALUES_EQUAL (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
7099 Compares for equality the two values, @var{x} and @var{y}. If the target
7100 floating point format supports negative zeroes and/or NaNs,
7101 @samp{REAL_VALUES_EQUAL (-0.0, 0.0)} is true, and
7102 @samp{REAL_VALUES_EQUAL (NaN, NaN)} is false.
7105 @deftypefn Macro int REAL_VALUES_LESS (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
7106 Tests whether @var{x} is less than @var{y}.
7109 @deftypefn Macro HOST_WIDE_INT REAL_VALUE_FIX (REAL_VALUE_TYPE @var{x})
7110 Truncates @var{x} to a signed integer, rounding toward zero.
7113 @deftypefn Macro {unsigned HOST_WIDE_INT} REAL_VALUE_UNSIGNED_FIX (REAL_VALUE_TYPE @var{x})
7114 Truncates @var{x} to an unsigned integer, rounding toward zero. If
7115 @var{x} is negative, returns zero.
7118 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ATOF (const char *@var{string}, machine_mode @var{mode})
7119 Converts @var{string} into a floating point number in the target machine's
7120 representation for mode @var{mode}. This routine can handle both
7121 decimal and hexadecimal floating point constants, using the syntax
7122 defined by the C language for both.
7125 @deftypefn Macro int REAL_VALUE_NEGATIVE (REAL_VALUE_TYPE @var{x})
7126 Returns 1 if @var{x} is negative (including negative zero), 0 otherwise.
7129 @deftypefn Macro int REAL_VALUE_ISINF (REAL_VALUE_TYPE @var{x})
7130 Determines whether @var{x} represents infinity (positive or negative).
7133 @deftypefn Macro int REAL_VALUE_ISNAN (REAL_VALUE_TYPE @var{x})
7134 Determines whether @var{x} represents a ``NaN'' (not-a-number).
7137 @deftypefn Macro void REAL_ARITHMETIC (REAL_VALUE_TYPE @var{output}, enum tree_code @var{code}, REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
7138 Calculates an arithmetic operation on the two floating point values
7139 @var{x} and @var{y}, storing the result in @var{output} (which must be a
7142 The operation to be performed is specified by @var{code}. Only the
7143 following codes are supported: @code{PLUS_EXPR}, @code{MINUS_EXPR},
7144 @code{MULT_EXPR}, @code{RDIV_EXPR}, @code{MAX_EXPR}, @code{MIN_EXPR}.
7146 If @code{REAL_ARITHMETIC} is asked to evaluate division by zero and the
7147 target's floating point format cannot represent infinity, it will call
7148 @code{abort}. Callers should check for this situation first, using
7149 @code{MODE_HAS_INFINITIES}. @xref{Storage Layout}.
7152 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_NEGATE (REAL_VALUE_TYPE @var{x})
7153 Returns the negative of the floating point value @var{x}.
7156 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ABS (REAL_VALUE_TYPE @var{x})
7157 Returns the absolute value of @var{x}.
7160 @node Mode Switching
7161 @section Mode Switching Instructions
7162 @cindex mode switching
7163 The following macros control mode switching optimizations:
7165 @defmac OPTIMIZE_MODE_SWITCHING (@var{entity})
7166 Define this macro if the port needs extra instructions inserted for mode
7167 switching in an optimizing compilation.
7169 For an example, the SH4 can perform both single and double precision
7170 floating point operations, but to perform a single precision operation,
7171 the FPSCR PR bit has to be cleared, while for a double precision
7172 operation, this bit has to be set. Changing the PR bit requires a general
7173 purpose register as a scratch register, hence these FPSCR sets have to
7174 be inserted before reload, i.e.@: you can't put this into instruction emitting
7175 or @code{TARGET_MACHINE_DEPENDENT_REORG}.
7177 You can have multiple entities that are mode-switched, and select at run time
7178 which entities actually need it. @code{OPTIMIZE_MODE_SWITCHING} should
7179 return nonzero for any @var{entity} that needs mode-switching.
7180 If you define this macro, you also have to define
7181 @code{NUM_MODES_FOR_MODE_SWITCHING}, @code{TARGET_MODE_NEEDED},
7182 @code{TARGET_MODE_PRIORITY} and @code{TARGET_MODE_EMIT}.
7183 @code{TARGET_MODE_AFTER}, @code{TARGET_MODE_ENTRY}, and @code{TARGET_MODE_EXIT}
7187 @defmac NUM_MODES_FOR_MODE_SWITCHING
7188 If you define @code{OPTIMIZE_MODE_SWITCHING}, you have to define this as
7189 initializer for an array of integers. Each initializer element
7190 N refers to an entity that needs mode switching, and specifies the number
7191 of different modes that might need to be set for this entity.
7192 The position of the initializer in the initializer---starting counting at
7193 zero---determines the integer that is used to refer to the mode-switched
7195 In macros that take mode arguments / yield a mode result, modes are
7196 represented as numbers 0 @dots{} N @minus{} 1. N is used to specify that no mode
7197 switch is needed / supplied.
7200 @hook TARGET_MODE_EMIT
7202 @hook TARGET_MODE_NEEDED
7204 @hook TARGET_MODE_AFTER
7206 @hook TARGET_MODE_ENTRY
7208 @hook TARGET_MODE_EXIT
7210 @hook TARGET_MODE_PRIORITY
7212 @node Target Attributes
7213 @section Defining target-specific uses of @code{__attribute__}
7214 @cindex target attributes
7215 @cindex machine attributes
7216 @cindex attributes, target-specific
7218 Target-specific attributes may be defined for functions, data and types.
7219 These are described using the following target hooks; they also need to
7220 be documented in @file{extend.texi}.
7222 @hook TARGET_ATTRIBUTE_TABLE
7224 @hook TARGET_ATTRIBUTE_TAKES_IDENTIFIER_P
7226 @hook TARGET_COMP_TYPE_ATTRIBUTES
7228 @hook TARGET_SET_DEFAULT_TYPE_ATTRIBUTES
7230 @hook TARGET_MERGE_TYPE_ATTRIBUTES
7232 @hook TARGET_MERGE_DECL_ATTRIBUTES
7234 @hook TARGET_VALID_DLLIMPORT_ATTRIBUTE_P
7236 @defmac TARGET_DECLSPEC
7237 Define this macro to a nonzero value if you want to treat
7238 @code{__declspec(X)} as equivalent to @code{__attribute((X))}. By
7239 default, this behavior is enabled only for targets that define
7240 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES}. The current implementation
7241 of @code{__declspec} is via a built-in macro, but you should not rely
7242 on this implementation detail.
7245 @hook TARGET_INSERT_ATTRIBUTES
7247 @hook TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P
7249 @hook TARGET_OPTION_VALID_ATTRIBUTE_P
7251 @hook TARGET_OPTION_SAVE
7253 @hook TARGET_OPTION_RESTORE
7255 @hook TARGET_OPTION_POST_STREAM_IN
7257 @hook TARGET_OPTION_PRINT
7259 @hook TARGET_OPTION_PRAGMA_PARSE
7261 @hook TARGET_OPTION_OVERRIDE
7263 @hook TARGET_OPTION_FUNCTION_VERSIONS
7265 @hook TARGET_CAN_INLINE_P
7268 @section Emulating TLS
7269 @cindex Emulated TLS
7271 For targets whose psABI does not provide Thread Local Storage via
7272 specific relocations and instruction sequences, an emulation layer is
7273 used. A set of target hooks allows this emulation layer to be
7274 configured for the requirements of a particular target. For instance
7275 the psABI may in fact specify TLS support in terms of an emulation
7278 The emulation layer works by creating a control object for every TLS
7279 object. To access the TLS object, a lookup function is provided
7280 which, when given the address of the control object, will return the
7281 address of the current thread's instance of the TLS object.
7283 @hook TARGET_EMUTLS_GET_ADDRESS
7285 @hook TARGET_EMUTLS_REGISTER_COMMON
7287 @hook TARGET_EMUTLS_VAR_SECTION
7289 @hook TARGET_EMUTLS_TMPL_SECTION
7291 @hook TARGET_EMUTLS_VAR_PREFIX
7293 @hook TARGET_EMUTLS_TMPL_PREFIX
7295 @hook TARGET_EMUTLS_VAR_FIELDS
7297 @hook TARGET_EMUTLS_VAR_INIT
7299 @hook TARGET_EMUTLS_VAR_ALIGN_FIXED
7301 @hook TARGET_EMUTLS_DEBUG_FORM_TLS_ADDRESS
7303 @node MIPS Coprocessors
7304 @section Defining coprocessor specifics for MIPS targets.
7305 @cindex MIPS coprocessor-definition macros
7307 The MIPS specification allows MIPS implementations to have as many as 4
7308 coprocessors, each with as many as 32 private registers. GCC supports
7309 accessing these registers and transferring values between the registers
7310 and memory using asm-ized variables. For example:
7313 register unsigned int cp0count asm ("c0r1");
7319 (``c0r1'' is the default name of register 1 in coprocessor 0; alternate
7320 names may be added as described below, or the default names may be
7321 overridden entirely in @code{SUBTARGET_CONDITIONAL_REGISTER_USAGE}.)
7323 Coprocessor registers are assumed to be epilogue-used; sets to them will
7324 be preserved even if it does not appear that the register is used again
7325 later in the function.
7327 Another note: according to the MIPS spec, coprocessor 1 (if present) is
7328 the FPU@. One accesses COP1 registers through standard mips
7329 floating-point support; they are not included in this mechanism.
7332 @section Parameters for Precompiled Header Validity Checking
7333 @cindex parameters, precompiled headers
7335 @hook TARGET_GET_PCH_VALIDITY
7337 @hook TARGET_PCH_VALID_P
7339 @hook TARGET_CHECK_PCH_TARGET_FLAGS
7341 @hook TARGET_PREPARE_PCH_SAVE
7344 @section C++ ABI parameters
7345 @cindex parameters, c++ abi
7347 @hook TARGET_CXX_GUARD_TYPE
7349 @hook TARGET_CXX_GUARD_MASK_BIT
7351 @hook TARGET_CXX_GET_COOKIE_SIZE
7353 @hook TARGET_CXX_COOKIE_HAS_SIZE
7355 @hook TARGET_CXX_IMPORT_EXPORT_CLASS
7357 @hook TARGET_CXX_CDTOR_RETURNS_THIS
7359 @hook TARGET_CXX_KEY_METHOD_MAY_BE_INLINE
7361 @hook TARGET_CXX_DETERMINE_CLASS_DATA_VISIBILITY
7363 @hook TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT
7365 @hook TARGET_CXX_LIBRARY_RTTI_COMDAT
7367 @hook TARGET_CXX_USE_AEABI_ATEXIT
7369 @hook TARGET_CXX_USE_ATEXIT_FOR_CXA_ATEXIT
7371 @hook TARGET_CXX_ADJUST_CLASS_AT_DEFINITION
7373 @hook TARGET_CXX_DECL_MANGLING_CONTEXT
7375 @node Named Address Spaces
7376 @section Adding support for named address spaces
7377 @cindex named address spaces
7379 The draft technical report of the ISO/IEC JTC1 S22 WG14 N1275
7380 standards committee, @cite{Programming Languages - C - Extensions to
7381 support embedded processors}, specifies a syntax for embedded
7382 processors to specify alternate address spaces. You can configure a
7383 GCC port to support section 5.1 of the draft report to add support for
7384 address spaces other than the default address space. These address
7385 spaces are new keywords that are similar to the @code{volatile} and
7386 @code{const} type attributes.
7388 Pointers to named address spaces can have a different size than
7389 pointers to the generic address space.
7391 For example, the SPU port uses the @code{__ea} address space to refer
7392 to memory in the host processor, rather than memory local to the SPU
7393 processor. Access to memory in the @code{__ea} address space involves
7394 issuing DMA operations to move data between the host processor and the
7395 local processor memory address space. Pointers in the @code{__ea}
7396 address space are either 32 bits or 64 bits based on the
7397 @option{-mea32} or @option{-mea64} switches (native SPU pointers are
7400 Internally, address spaces are represented as a small integer in the
7401 range 0 to 15 with address space 0 being reserved for the generic
7404 To register a named address space qualifier keyword with the C front end,
7405 the target may call the @code{c_register_addr_space} routine. For example,
7406 the SPU port uses the following to declare @code{__ea} as the keyword for
7407 named address space #1:
7409 #define ADDR_SPACE_EA 1
7410 c_register_addr_space ("__ea", ADDR_SPACE_EA);
7413 @hook TARGET_ADDR_SPACE_POINTER_MODE
7415 @hook TARGET_ADDR_SPACE_ADDRESS_MODE
7417 @hook TARGET_ADDR_SPACE_VALID_POINTER_MODE
7419 @hook TARGET_ADDR_SPACE_LEGITIMATE_ADDRESS_P
7421 @hook TARGET_ADDR_SPACE_LEGITIMIZE_ADDRESS
7423 @hook TARGET_ADDR_SPACE_SUBSET_P
7425 @hook TARGET_ADDR_SPACE_CONVERT
7428 @section Miscellaneous Parameters
7429 @cindex parameters, miscellaneous
7431 @c prevent bad page break with this line
7432 Here are several miscellaneous parameters.
7434 @defmac HAS_LONG_COND_BRANCH
7435 Define this boolean macro to indicate whether or not your architecture
7436 has conditional branches that can span all of memory. It is used in
7437 conjunction with an optimization that partitions hot and cold basic
7438 blocks into separate sections of the executable. If this macro is
7439 set to false, gcc will convert any conditional branches that attempt
7440 to cross between sections into unconditional branches or indirect jumps.
7443 @defmac HAS_LONG_UNCOND_BRANCH
7444 Define this boolean macro to indicate whether or not your architecture
7445 has unconditional branches that can span all of memory. It is used in
7446 conjunction with an optimization that partitions hot and cold basic
7447 blocks into separate sections of the executable. If this macro is
7448 set to false, gcc will convert any unconditional branches that attempt
7449 to cross between sections into indirect jumps.
7452 @defmac CASE_VECTOR_MODE
7453 An alias for a machine mode name. This is the machine mode that
7454 elements of a jump-table should have.
7457 @defmac CASE_VECTOR_SHORTEN_MODE (@var{min_offset}, @var{max_offset}, @var{body})
7458 Optional: return the preferred mode for an @code{addr_diff_vec}
7459 when the minimum and maximum offset are known. If you define this,
7460 it enables extra code in branch shortening to deal with @code{addr_diff_vec}.
7461 To make this work, you also have to define @code{INSN_ALIGN} and
7462 make the alignment for @code{addr_diff_vec} explicit.
7463 The @var{body} argument is provided so that the offset_unsigned and scale
7464 flags can be updated.
7467 @defmac CASE_VECTOR_PC_RELATIVE
7468 Define this macro to be a C expression to indicate when jump-tables
7469 should contain relative addresses. You need not define this macro if
7470 jump-tables never contain relative addresses, or jump-tables should
7471 contain relative addresses only when @option{-fPIC} or @option{-fPIC}
7475 @hook TARGET_CASE_VALUES_THRESHOLD
7477 @defmac WORD_REGISTER_OPERATIONS
7478 Define this macro if operations between registers with integral mode
7479 smaller than a word are always performed on the entire register.
7480 Most RISC machines have this property and most CISC machines do not.
7483 @defmac LOAD_EXTEND_OP (@var{mem_mode})
7484 Define this macro to be a C expression indicating when insns that read
7485 memory in @var{mem_mode}, an integral mode narrower than a word, set the
7486 bits outside of @var{mem_mode} to be either the sign-extension or the
7487 zero-extension of the data read. Return @code{SIGN_EXTEND} for values
7488 of @var{mem_mode} for which the
7489 insn sign-extends, @code{ZERO_EXTEND} for which it zero-extends, and
7490 @code{UNKNOWN} for other modes.
7492 This macro is not called with @var{mem_mode} non-integral or with a width
7493 greater than or equal to @code{BITS_PER_WORD}, so you may return any
7494 value in this case. Do not define this macro if it would always return
7495 @code{UNKNOWN}. On machines where this macro is defined, you will normally
7496 define it as the constant @code{SIGN_EXTEND} or @code{ZERO_EXTEND}.
7498 You may return a non-@code{UNKNOWN} value even if for some hard registers
7499 the sign extension is not performed, if for the @code{REGNO_REG_CLASS}
7500 of these hard registers @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero
7501 when the @var{from} mode is @var{mem_mode} and the @var{to} mode is any
7502 integral mode larger than this but not larger than @code{word_mode}.
7504 You must return @code{UNKNOWN} if for some hard registers that allow this
7505 mode, @code{CANNOT_CHANGE_MODE_CLASS} says that they cannot change to
7506 @code{word_mode}, but that they can change to another integral mode that
7507 is larger then @var{mem_mode} but still smaller than @code{word_mode}.
7510 @defmac SHORT_IMMEDIATES_SIGN_EXTEND
7511 Define this macro if loading short immediate values into registers sign
7515 @hook TARGET_MIN_DIVISIONS_FOR_RECIP_MUL
7518 The maximum number of bytes that a single instruction can move quickly
7519 between memory and registers or between two memory locations.
7522 @defmac MAX_MOVE_MAX
7523 The maximum number of bytes that a single instruction can move quickly
7524 between memory and registers or between two memory locations. If this
7525 is undefined, the default is @code{MOVE_MAX}. Otherwise, it is the
7526 constant value that is the largest value that @code{MOVE_MAX} can have
7530 @defmac SHIFT_COUNT_TRUNCATED
7531 A C expression that is nonzero if on this machine the number of bits
7532 actually used for the count of a shift operation is equal to the number
7533 of bits needed to represent the size of the object being shifted. When
7534 this macro is nonzero, the compiler will assume that it is safe to omit
7535 a sign-extend, zero-extend, and certain bitwise `and' instructions that
7536 truncates the count of a shift operation. On machines that have
7537 instructions that act on bit-fields at variable positions, which may
7538 include `bit test' instructions, a nonzero @code{SHIFT_COUNT_TRUNCATED}
7539 also enables deletion of truncations of the values that serve as
7540 arguments to bit-field instructions.
7542 If both types of instructions truncate the count (for shifts) and
7543 position (for bit-field operations), or if no variable-position bit-field
7544 instructions exist, you should define this macro.
7546 However, on some machines, such as the 80386 and the 680x0, truncation
7547 only applies to shift operations and not the (real or pretended)
7548 bit-field operations. Define @code{SHIFT_COUNT_TRUNCATED} to be zero on
7549 such machines. Instead, add patterns to the @file{md} file that include
7550 the implied truncation of the shift instructions.
7552 You need not define this macro if it would always have the value of zero.
7555 @anchor{TARGET_SHIFT_TRUNCATION_MASK}
7556 @hook TARGET_SHIFT_TRUNCATION_MASK
7558 @defmac TRULY_NOOP_TRUNCATION (@var{outprec}, @var{inprec})
7559 A C expression which is nonzero if on this machine it is safe to
7560 ``convert'' an integer of @var{inprec} bits to one of @var{outprec}
7561 bits (where @var{outprec} is smaller than @var{inprec}) by merely
7562 operating on it as if it had only @var{outprec} bits.
7564 On many machines, this expression can be 1.
7566 @c rearranged this, removed the phrase "it is reported that". this was
7567 @c to fix an overfull hbox. --mew 10feb93
7568 When @code{TRULY_NOOP_TRUNCATION} returns 1 for a pair of sizes for
7569 modes for which @code{MODES_TIEABLE_P} is 0, suboptimal code can result.
7570 If this is the case, making @code{TRULY_NOOP_TRUNCATION} return 0 in
7571 such cases may improve things.
7574 @hook TARGET_MODE_REP_EXTENDED
7576 @defmac STORE_FLAG_VALUE
7577 A C expression describing the value returned by a comparison operator
7578 with an integral mode and stored by a store-flag instruction
7579 (@samp{cstore@var{mode}4}) when the condition is true. This description must
7580 apply to @emph{all} the @samp{cstore@var{mode}4} patterns and all the
7581 comparison operators whose results have a @code{MODE_INT} mode.
7583 A value of 1 or @minus{}1 means that the instruction implementing the
7584 comparison operator returns exactly 1 or @minus{}1 when the comparison is true
7585 and 0 when the comparison is false. Otherwise, the value indicates
7586 which bits of the result are guaranteed to be 1 when the comparison is
7587 true. This value is interpreted in the mode of the comparison
7588 operation, which is given by the mode of the first operand in the
7589 @samp{cstore@var{mode}4} pattern. Either the low bit or the sign bit of
7590 @code{STORE_FLAG_VALUE} be on. Presently, only those bits are used by
7593 If @code{STORE_FLAG_VALUE} is neither 1 or @minus{}1, the compiler will
7594 generate code that depends only on the specified bits. It can also
7595 replace comparison operators with equivalent operations if they cause
7596 the required bits to be set, even if the remaining bits are undefined.
7597 For example, on a machine whose comparison operators return an
7598 @code{SImode} value and where @code{STORE_FLAG_VALUE} is defined as
7599 @samp{0x80000000}, saying that just the sign bit is relevant, the
7603 (ne:SI (and:SI @var{x} (const_int @var{power-of-2})) (const_int 0))
7610 (ashift:SI @var{x} (const_int @var{n}))
7614 where @var{n} is the appropriate shift count to move the bit being
7615 tested into the sign bit.
7617 There is no way to describe a machine that always sets the low-order bit
7618 for a true value, but does not guarantee the value of any other bits,
7619 but we do not know of any machine that has such an instruction. If you
7620 are trying to port GCC to such a machine, include an instruction to
7621 perform a logical-and of the result with 1 in the pattern for the
7622 comparison operators and let us know at @email{gcc@@gcc.gnu.org}.
7624 Often, a machine will have multiple instructions that obtain a value
7625 from a comparison (or the condition codes). Here are rules to guide the
7626 choice of value for @code{STORE_FLAG_VALUE}, and hence the instructions
7631 Use the shortest sequence that yields a valid definition for
7632 @code{STORE_FLAG_VALUE}. It is more efficient for the compiler to
7633 ``normalize'' the value (convert it to, e.g., 1 or 0) than for the
7634 comparison operators to do so because there may be opportunities to
7635 combine the normalization with other operations.
7638 For equal-length sequences, use a value of 1 or @minus{}1, with @minus{}1 being
7639 slightly preferred on machines with expensive jumps and 1 preferred on
7643 As a second choice, choose a value of @samp{0x80000001} if instructions
7644 exist that set both the sign and low-order bits but do not define the
7648 Otherwise, use a value of @samp{0x80000000}.
7651 Many machines can produce both the value chosen for
7652 @code{STORE_FLAG_VALUE} and its negation in the same number of
7653 instructions. On those machines, you should also define a pattern for
7654 those cases, e.g., one matching
7657 (set @var{A} (neg:@var{m} (ne:@var{m} @var{B} @var{C})))
7660 Some machines can also perform @code{and} or @code{plus} operations on
7661 condition code values with less instructions than the corresponding
7662 @samp{cstore@var{mode}4} insn followed by @code{and} or @code{plus}. On those
7663 machines, define the appropriate patterns. Use the names @code{incscc}
7664 and @code{decscc}, respectively, for the patterns which perform
7665 @code{plus} or @code{minus} operations on condition code values. See
7666 @file{rs6000.md} for some examples. The GNU Superoptimizer can be used to
7667 find such instruction sequences on other machines.
7669 If this macro is not defined, the default value, 1, is used. You need
7670 not define @code{STORE_FLAG_VALUE} if the machine has no store-flag
7671 instructions, or if the value generated by these instructions is 1.
7674 @defmac FLOAT_STORE_FLAG_VALUE (@var{mode})
7675 A C expression that gives a nonzero @code{REAL_VALUE_TYPE} value that is
7676 returned when comparison operators with floating-point results are true.
7677 Define this macro on machines that have comparison operations that return
7678 floating-point values. If there are no such operations, do not define
7682 @defmac VECTOR_STORE_FLAG_VALUE (@var{mode})
7683 A C expression that gives a rtx representing the nonzero true element
7684 for vector comparisons. The returned rtx should be valid for the inner
7685 mode of @var{mode} which is guaranteed to be a vector mode. Define
7686 this macro on machines that have vector comparison operations that
7687 return a vector result. If there are no such operations, do not define
7688 this macro. Typically, this macro is defined as @code{const1_rtx} or
7689 @code{constm1_rtx}. This macro may return @code{NULL_RTX} to prevent
7690 the compiler optimizing such vector comparison operations for the
7694 @defmac CLZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
7695 @defmacx CTZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
7696 A C expression that indicates whether the architecture defines a value
7697 for @code{clz} or @code{ctz} with a zero operand.
7698 A result of @code{0} indicates the value is undefined.
7699 If the value is defined for only the RTL expression, the macro should
7700 evaluate to @code{1}; if the value applies also to the corresponding optab
7701 entry (which is normally the case if it expands directly into
7702 the corresponding RTL), then the macro should evaluate to @code{2}.
7703 In the cases where the value is defined, @var{value} should be set to
7706 If this macro is not defined, the value of @code{clz} or
7707 @code{ctz} at zero is assumed to be undefined.
7709 This macro must be defined if the target's expansion for @code{ffs}
7710 relies on a particular value to get correct results. Otherwise it
7711 is not necessary, though it may be used to optimize some corner cases, and
7712 to provide a default expansion for the @code{ffs} optab.
7714 Note that regardless of this macro the ``definedness'' of @code{clz}
7715 and @code{ctz} at zero do @emph{not} extend to the builtin functions
7716 visible to the user. Thus one may be free to adjust the value at will
7717 to match the target expansion of these operations without fear of
7722 An alias for the machine mode for pointers. On most machines, define
7723 this to be the integer mode corresponding to the width of a hardware
7724 pointer; @code{SImode} on 32-bit machine or @code{DImode} on 64-bit machines.
7725 On some machines you must define this to be one of the partial integer
7726 modes, such as @code{PSImode}.
7728 The width of @code{Pmode} must be at least as large as the value of
7729 @code{POINTER_SIZE}. If it is not equal, you must define the macro
7730 @code{POINTERS_EXTEND_UNSIGNED} to specify how pointers are extended
7734 @defmac FUNCTION_MODE
7735 An alias for the machine mode used for memory references to functions
7736 being called, in @code{call} RTL expressions. On most CISC machines,
7737 where an instruction can begin at any byte address, this should be
7738 @code{QImode}. On most RISC machines, where all instructions have fixed
7739 size and alignment, this should be a mode with the same size and alignment
7740 as the machine instruction words - typically @code{SImode} or @code{HImode}.
7743 @defmac STDC_0_IN_SYSTEM_HEADERS
7744 In normal operation, the preprocessor expands @code{__STDC__} to the
7745 constant 1, to signify that GCC conforms to ISO Standard C@. On some
7746 hosts, like Solaris, the system compiler uses a different convention,
7747 where @code{__STDC__} is normally 0, but is 1 if the user specifies
7748 strict conformance to the C Standard.
7750 Defining @code{STDC_0_IN_SYSTEM_HEADERS} makes GNU CPP follows the host
7751 convention when processing system header files, but when processing user
7752 files @code{__STDC__} will always expand to 1.
7755 @hook TARGET_C_PREINCLUDE
7757 @hook TARGET_CXX_IMPLICIT_EXTERN_C
7759 @defmac NO_IMPLICIT_EXTERN_C
7760 Define this macro if the system header files support C++ as well as C@.
7761 This macro inhibits the usual method of using system header files in
7762 C++, which is to pretend that the file's contents are enclosed in
7763 @samp{extern "C" @{@dots{}@}}.
7768 @defmac REGISTER_TARGET_PRAGMAS ()
7769 Define this macro if you want to implement any target-specific pragmas.
7770 If defined, it is a C expression which makes a series of calls to
7771 @code{c_register_pragma} or @code{c_register_pragma_with_expansion}
7772 for each pragma. The macro may also do any
7773 setup required for the pragmas.
7775 The primary reason to define this macro is to provide compatibility with
7776 other compilers for the same target. In general, we discourage
7777 definition of target-specific pragmas for GCC@.
7779 If the pragma can be implemented by attributes then you should consider
7780 defining the target hook @samp{TARGET_INSERT_ATTRIBUTES} as well.
7782 Preprocessor macros that appear on pragma lines are not expanded. All
7783 @samp{#pragma} directives that do not match any registered pragma are
7784 silently ignored, unless the user specifies @option{-Wunknown-pragmas}.
7787 @deftypefun void c_register_pragma (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
7788 @deftypefunx void c_register_pragma_with_expansion (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
7790 Each call to @code{c_register_pragma} or
7791 @code{c_register_pragma_with_expansion} establishes one pragma. The
7792 @var{callback} routine will be called when the preprocessor encounters a
7796 #pragma [@var{space}] @var{name} @dots{}
7799 @var{space} is the case-sensitive namespace of the pragma, or
7800 @code{NULL} to put the pragma in the global namespace. The callback
7801 routine receives @var{pfile} as its first argument, which can be passed
7802 on to cpplib's functions if necessary. You can lex tokens after the
7803 @var{name} by calling @code{pragma_lex}. Tokens that are not read by the
7804 callback will be silently ignored. The end of the line is indicated by
7805 a token of type @code{CPP_EOF}. Macro expansion occurs on the
7806 arguments of pragmas registered with
7807 @code{c_register_pragma_with_expansion} but not on the arguments of
7808 pragmas registered with @code{c_register_pragma}.
7810 Note that the use of @code{pragma_lex} is specific to the C and C++
7811 compilers. It will not work in the Java or Fortran compilers, or any
7812 other language compilers for that matter. Thus if @code{pragma_lex} is going
7813 to be called from target-specific code, it must only be done so when
7814 building the C and C++ compilers. This can be done by defining the
7815 variables @code{c_target_objs} and @code{cxx_target_objs} in the
7816 target entry in the @file{config.gcc} file. These variables should name
7817 the target-specific, language-specific object file which contains the
7818 code that uses @code{pragma_lex}. Note it will also be necessary to add a
7819 rule to the makefile fragment pointed to by @code{tmake_file} that shows
7820 how to build this object file.
7823 @defmac HANDLE_PRAGMA_PACK_WITH_EXPANSION
7824 Define this macro if macros should be expanded in the
7825 arguments of @samp{#pragma pack}.
7828 @defmac TARGET_DEFAULT_PACK_STRUCT
7829 If your target requires a structure packing default other than 0 (meaning
7830 the machine default), define this macro to the necessary value (in bytes).
7831 This must be a value that would also be valid to use with
7832 @samp{#pragma pack()} (that is, a small power of two).
7835 @defmac DOLLARS_IN_IDENTIFIERS
7836 Define this macro to control use of the character @samp{$} in
7837 identifier names for the C family of languages. 0 means @samp{$} is
7838 not allowed by default; 1 means it is allowed. 1 is the default;
7839 there is no need to define this macro in that case.
7842 @defmac INSN_SETS_ARE_DELAYED (@var{insn})
7843 Define this macro as a C expression that is nonzero if it is safe for the
7844 delay slot scheduler to place instructions in the delay slot of @var{insn},
7845 even if they appear to use a resource set or clobbered in @var{insn}.
7846 @var{insn} is always a @code{jump_insn} or an @code{insn}; GCC knows that
7847 every @code{call_insn} has this behavior. On machines where some @code{insn}
7848 or @code{jump_insn} is really a function call and hence has this behavior,
7849 you should define this macro.
7851 You need not define this macro if it would always return zero.
7854 @defmac INSN_REFERENCES_ARE_DELAYED (@var{insn})
7855 Define this macro as a C expression that is nonzero if it is safe for the
7856 delay slot scheduler to place instructions in the delay slot of @var{insn},
7857 even if they appear to set or clobber a resource referenced in @var{insn}.
7858 @var{insn} is always a @code{jump_insn} or an @code{insn}. On machines where
7859 some @code{insn} or @code{jump_insn} is really a function call and its operands
7860 are registers whose use is actually in the subroutine it calls, you should
7861 define this macro. Doing so allows the delay slot scheduler to move
7862 instructions which copy arguments into the argument registers into the delay
7865 You need not define this macro if it would always return zero.
7868 @defmac MULTIPLE_SYMBOL_SPACES
7869 Define this macro as a C expression that is nonzero if, in some cases,
7870 global symbols from one translation unit may not be bound to undefined
7871 symbols in another translation unit without user intervention. For
7872 instance, under Microsoft Windows symbols must be explicitly imported
7873 from shared libraries (DLLs).
7875 You need not define this macro if it would always evaluate to zero.
7878 @hook TARGET_MD_ASM_CLOBBERS
7880 @defmac MATH_LIBRARY
7881 Define this macro as a C string constant for the linker argument to link
7882 in the system math library, minus the initial @samp{"-l"}, or
7883 @samp{""} if the target does not have a
7884 separate math library.
7886 You need only define this macro if the default of @samp{"m"} is wrong.
7889 @defmac LIBRARY_PATH_ENV
7890 Define this macro as a C string constant for the environment variable that
7891 specifies where the linker should look for libraries.
7893 You need only define this macro if the default of @samp{"LIBRARY_PATH"}
7897 @defmac TARGET_POSIX_IO
7898 Define this macro if the target supports the following POSIX@ file
7899 functions, access, mkdir and file locking with fcntl / F_SETLKW@.
7900 Defining @code{TARGET_POSIX_IO} will enable the test coverage code
7901 to use file locking when exiting a program, which avoids race conditions
7902 if the program has forked. It will also create directories at run-time
7903 for cross-profiling.
7906 @defmac MAX_CONDITIONAL_EXECUTE
7908 A C expression for the maximum number of instructions to execute via
7909 conditional execution instructions instead of a branch. A value of
7910 @code{BRANCH_COST}+1 is the default if the machine does not use cc0, and
7911 1 if it does use cc0.
7914 @defmac IFCVT_MODIFY_TESTS (@var{ce_info}, @var{true_expr}, @var{false_expr})
7915 Used if the target needs to perform machine-dependent modifications on the
7916 conditionals used for turning basic blocks into conditionally executed code.
7917 @var{ce_info} points to a data structure, @code{struct ce_if_block}, which
7918 contains information about the currently processed blocks. @var{true_expr}
7919 and @var{false_expr} are the tests that are used for converting the
7920 then-block and the else-block, respectively. Set either @var{true_expr} or
7921 @var{false_expr} to a null pointer if the tests cannot be converted.
7924 @defmac IFCVT_MODIFY_MULTIPLE_TESTS (@var{ce_info}, @var{bb}, @var{true_expr}, @var{false_expr})
7925 Like @code{IFCVT_MODIFY_TESTS}, but used when converting more complicated
7926 if-statements into conditions combined by @code{and} and @code{or} operations.
7927 @var{bb} contains the basic block that contains the test that is currently
7928 being processed and about to be turned into a condition.
7931 @defmac IFCVT_MODIFY_INSN (@var{ce_info}, @var{pattern}, @var{insn})
7932 A C expression to modify the @var{PATTERN} of an @var{INSN} that is to
7933 be converted to conditional execution format. @var{ce_info} points to
7934 a data structure, @code{struct ce_if_block}, which contains information
7935 about the currently processed blocks.
7938 @defmac IFCVT_MODIFY_FINAL (@var{ce_info})
7939 A C expression to perform any final machine dependent modifications in
7940 converting code to conditional execution. The involved basic blocks
7941 can be found in the @code{struct ce_if_block} structure that is pointed
7942 to by @var{ce_info}.
7945 @defmac IFCVT_MODIFY_CANCEL (@var{ce_info})
7946 A C expression to cancel any machine dependent modifications in
7947 converting code to conditional execution. The involved basic blocks
7948 can be found in the @code{struct ce_if_block} structure that is pointed
7949 to by @var{ce_info}.
7952 @defmac IFCVT_MACHDEP_INIT (@var{ce_info})
7953 A C expression to initialize any machine specific data for if-conversion
7954 of the if-block in the @code{struct ce_if_block} structure that is pointed
7955 to by @var{ce_info}.
7958 @hook TARGET_MACHINE_DEPENDENT_REORG
7960 @hook TARGET_INIT_BUILTINS
7962 @hook TARGET_BUILTIN_DECL
7964 @hook TARGET_EXPAND_BUILTIN
7966 @hook TARGET_BUILTIN_CHKP_FUNCTION
7967 @hook TARGET_CHKP_BOUND_TYPE
7968 @hook TARGET_CHKP_BOUND_MODE
7969 @hook TARGET_CHKP_MAKE_BOUNDS_CONSTANT
7970 @hook TARGET_CHKP_INITIALIZE_BOUNDS
7972 @hook TARGET_RESOLVE_OVERLOADED_BUILTIN
7974 @hook TARGET_FOLD_BUILTIN
7976 @hook TARGET_GIMPLE_FOLD_BUILTIN
7978 @hook TARGET_COMPARE_VERSION_PRIORITY
7980 @hook TARGET_GET_FUNCTION_VERSIONS_DISPATCHER
7982 @hook TARGET_GENERATE_VERSION_DISPATCHER_BODY
7984 @hook TARGET_CAN_USE_DOLOOP_P
7986 @hook TARGET_INVALID_WITHIN_DOLOOP
7988 @hook TARGET_LEGITIMATE_COMBINED_INSN
7990 @hook TARGET_CAN_FOLLOW_JUMP
7992 @hook TARGET_COMMUTATIVE_P
7994 @hook TARGET_ALLOCATE_INITIAL_VALUE
7996 @hook TARGET_UNSPEC_MAY_TRAP_P
7998 @hook TARGET_SET_CURRENT_FUNCTION
8000 @defmac TARGET_OBJECT_SUFFIX
8001 Define this macro to be a C string representing the suffix for object
8002 files on your target machine. If you do not define this macro, GCC will
8003 use @samp{.o} as the suffix for object files.
8006 @defmac TARGET_EXECUTABLE_SUFFIX
8007 Define this macro to be a C string representing the suffix to be
8008 automatically added to executable files on your target machine. If you
8009 do not define this macro, GCC will use the null string as the suffix for
8013 @defmac COLLECT_EXPORT_LIST
8014 If defined, @code{collect2} will scan the individual object files
8015 specified on its command line and create an export list for the linker.
8016 Define this macro for systems like AIX, where the linker discards
8017 object files that are not referenced from @code{main} and uses export
8021 @defmac MODIFY_JNI_METHOD_CALL (@var{mdecl})
8022 Define this macro to a C expression representing a variant of the
8023 method call @var{mdecl}, if Java Native Interface (JNI) methods
8024 must be invoked differently from other methods on your target.
8025 For example, on 32-bit Microsoft Windows, JNI methods must be invoked using
8026 the @code{stdcall} calling convention and this macro is then
8027 defined as this expression:
8030 build_type_attribute_variant (@var{mdecl},
8032 (get_identifier ("stdcall"),
8037 @hook TARGET_CANNOT_MODIFY_JUMPS_P
8039 @hook TARGET_BRANCH_TARGET_REGISTER_CLASS
8041 @hook TARGET_BRANCH_TARGET_REGISTER_CALLEE_SAVED
8043 @hook TARGET_HAVE_CONDITIONAL_EXECUTION
8045 @hook TARGET_GEN_CCMP_FIRST
8047 @hook TARGET_GEN_CCMP_NEXT
8049 @hook TARGET_LOOP_UNROLL_ADJUST
8051 @defmac POWI_MAX_MULTS
8052 If defined, this macro is interpreted as a signed integer C expression
8053 that specifies the maximum number of floating point multiplications
8054 that should be emitted when expanding exponentiation by an integer
8055 constant inline. When this value is defined, exponentiation requiring
8056 more than this number of multiplications is implemented by calling the
8057 system library's @code{pow}, @code{powf} or @code{powl} routines.
8058 The default value places no upper bound on the multiplication count.
8061 @deftypefn Macro void TARGET_EXTRA_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
8062 This target hook should register any extra include files for the
8063 target. The parameter @var{stdinc} indicates if normal include files
8064 are present. The parameter @var{sysroot} is the system root directory.
8065 The parameter @var{iprefix} is the prefix for the gcc directory.
8068 @deftypefn Macro void TARGET_EXTRA_PRE_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
8069 This target hook should register any extra include files for the
8070 target before any standard headers. The parameter @var{stdinc}
8071 indicates if normal include files are present. The parameter
8072 @var{sysroot} is the system root directory. The parameter
8073 @var{iprefix} is the prefix for the gcc directory.
8076 @deftypefn Macro void TARGET_OPTF (char *@var{path})
8077 This target hook should register special include paths for the target.
8078 The parameter @var{path} is the include to register. On Darwin
8079 systems, this is used for Framework includes, which have semantics
8080 that are different from @option{-I}.
8083 @defmac bool TARGET_USE_LOCAL_THUNK_ALIAS_P (tree @var{fndecl})
8084 This target macro returns @code{true} if it is safe to use a local alias
8085 for a virtual function @var{fndecl} when constructing thunks,
8086 @code{false} otherwise. By default, the macro returns @code{true} for all
8087 functions, if a target supports aliases (i.e.@: defines
8088 @code{ASM_OUTPUT_DEF}), @code{false} otherwise,
8091 @defmac TARGET_FORMAT_TYPES
8092 If defined, this macro is the name of a global variable containing
8093 target-specific format checking information for the @option{-Wformat}
8094 option. The default is to have no target-specific format checks.
8097 @defmac TARGET_N_FORMAT_TYPES
8098 If defined, this macro is the number of entries in
8099 @code{TARGET_FORMAT_TYPES}.
8102 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES
8103 If defined, this macro is the name of a global variable containing
8104 target-specific format overrides for the @option{-Wformat} option. The
8105 default is to have no target-specific format overrides. If defined,
8106 @code{TARGET_FORMAT_TYPES} must be defined, too.
8109 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES_COUNT
8110 If defined, this macro specifies the number of entries in
8111 @code{TARGET_OVERRIDES_FORMAT_ATTRIBUTES}.
8114 @defmac TARGET_OVERRIDES_FORMAT_INIT
8115 If defined, this macro specifies the optional initialization
8116 routine for target specific customizations of the system printf
8117 and scanf formatter settings.
8120 @hook TARGET_RELAXED_ORDERING
8122 @hook TARGET_INVALID_ARG_FOR_UNPROTOTYPED_FN
8124 @hook TARGET_INVALID_CONVERSION
8126 @hook TARGET_INVALID_UNARY_OP
8128 @hook TARGET_INVALID_BINARY_OP
8130 @hook TARGET_INVALID_PARAMETER_TYPE
8132 @hook TARGET_INVALID_RETURN_TYPE
8134 @hook TARGET_PROMOTED_TYPE
8136 @hook TARGET_CONVERT_TO_TYPE
8138 @defmac TARGET_USE_JCR_SECTION
8139 This macro determines whether to use the JCR section to register Java
8140 classes. By default, TARGET_USE_JCR_SECTION is defined to 1 if both
8141 SUPPORTS_WEAK and TARGET_HAVE_NAMED_SECTIONS are true, else 0.
8145 This macro determines the size of the objective C jump buffer for the
8146 NeXT runtime. By default, OBJC_JBLEN is defined to an innocuous value.
8149 @defmac LIBGCC2_UNWIND_ATTRIBUTE
8150 Define this macro if any target-specific attributes need to be attached
8151 to the functions in @file{libgcc} that provide low-level support for
8152 call stack unwinding. It is used in declarations in @file{unwind-generic.h}
8153 and the associated definitions of those functions.
8156 @hook TARGET_UPDATE_STACK_BOUNDARY
8158 @hook TARGET_GET_DRAP_RTX
8160 @hook TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS
8162 @hook TARGET_CONST_ANCHOR
8164 @hook TARGET_ASAN_SHADOW_OFFSET
8166 @hook TARGET_MEMMODEL_CHECK
8168 @hook TARGET_ATOMIC_TEST_AND_SET_TRUEVAL
8170 @hook TARGET_HAS_IFUNC_P
8172 @hook TARGET_ATOMIC_ALIGN_FOR_MODE
8174 @hook TARGET_ATOMIC_ASSIGN_EXPAND_FENV
8176 @hook TARGET_RECORD_OFFLOAD_SYMBOL
8178 @hook TARGET_OFFLOAD_OPTIONS
8180 @defmac TARGET_SUPPORTS_WIDE_INT
8182 On older ports, large integers are stored in @code{CONST_DOUBLE} rtl
8183 objects. Newer ports define @code{TARGET_SUPPORTS_WIDE_INT} to be nonzero
8184 to indicate that large integers are stored in
8185 @code{CONST_WIDE_INT} rtl objects. The @code{CONST_WIDE_INT} allows
8186 very large integer constants to be represented. @code{CONST_DOUBLE}
8187 is limited to twice the size of the host's @code{HOST_WIDE_INT}
8190 Converting a port mostly requires looking for the places where
8191 @code{CONST_DOUBLE}s are used with @code{VOIDmode} and replacing that
8192 code with code that accesses @code{CONST_WIDE_INT}s. @samp{"grep -i
8193 const_double"} at the port level gets you to 95% of the changes that
8194 need to be made. There are a few places that require a deeper look.
8198 There is no equivalent to @code{hval} and @code{lval} for
8199 @code{CONST_WIDE_INT}s. This would be difficult to express in the md
8200 language since there are a variable number of elements.
8202 Most ports only check that @code{hval} is either 0 or -1 to see if the
8203 value is small. As mentioned above, this will no longer be necessary
8204 since small constants are always @code{CONST_INT}. Of course there
8205 are still a few exceptions, the alpha's constraint used by the zap
8206 instruction certainly requires careful examination by C code.
8207 However, all the current code does is pass the hval and lval to C
8208 code, so evolving the c code to look at the @code{CONST_WIDE_INT} is
8209 not really a large change.
8212 Because there is no standard template that ports use to materialize
8213 constants, there is likely to be some futzing that is unique to each
8217 The rtx costs may have to be adjusted to properly account for larger
8218 constants that are represented as @code{CONST_WIDE_INT}.
8221 All and all it does not take long to convert ports that the
8222 maintainer is familiar with.