1 @c Copyright (C) 1988-2019 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 * D Language and ABI:: Controlling D ABI changes.
56 * Named Address Spaces:: Adding support for named address spaces
57 * Misc:: Everything else.
60 @node Target Structure
61 @section The Global @code{targetm} Variable
63 @cindex target functions
65 @deftypevar {struct gcc_target} targetm
66 The target @file{.c} file must define the global @code{targetm} variable
67 which contains pointers to functions and data relating to the target
68 machine. The variable is declared in @file{target.h};
69 @file{target-def.h} defines the macro @code{TARGET_INITIALIZER} which is
70 used to initialize the variable, and macros for the default initializers
71 for elements of the structure. The @file{.c} file should override those
72 macros for which the default definition is inappropriate. For example:
75 #include "target-def.h"
77 /* @r{Initialize the GCC target structure.} */
79 #undef TARGET_COMP_TYPE_ATTRIBUTES
80 #define TARGET_COMP_TYPE_ATTRIBUTES @var{machine}_comp_type_attributes
82 struct gcc_target targetm = TARGET_INITIALIZER;
86 Where a macro should be defined in the @file{.c} file in this manner to
87 form part of the @code{targetm} structure, it is documented below as a
88 ``Target Hook'' with a prototype. Many macros will change in future
89 from being defined in the @file{.h} file to being part of the
90 @code{targetm} structure.
92 Similarly, there is a @code{targetcm} variable for hooks that are
93 specific to front ends for C-family languages, documented as ``C
94 Target Hook''. This is declared in @file{c-family/c-target.h}, the
95 initializer @code{TARGETCM_INITIALIZER} in
96 @file{c-family/c-target-def.h}. If targets initialize @code{targetcm}
97 themselves, they should set @code{target_has_targetcm=yes} in
98 @file{config.gcc}; otherwise a default definition is used.
100 Similarly, there is a @code{targetm_common} variable for hooks that
101 are shared between the compiler driver and the compilers proper,
102 documented as ``Common Target Hook''. This is declared in
103 @file{common/common-target.h}, the initializer
104 @code{TARGETM_COMMON_INITIALIZER} in
105 @file{common/common-target-def.h}. If targets initialize
106 @code{targetm_common} themselves, they should set
107 @code{target_has_targetm_common=yes} in @file{config.gcc}; otherwise a
108 default definition is used.
110 Similarly, there is a @code{targetdm} variable for hooks that are
111 specific to the D language front end, documented as ``D Target Hook''.
112 This is declared in @file{d/d-target.h}, the initializer
113 @code{TARGETDM_INITIALIZER} in @file{d/d-target-def.h}. If targets
114 initialize @code{targetdm} themselves, they should set
115 @code{target_has_targetdm=yes} in @file{config.gcc}; otherwise a default
119 @section Controlling the Compilation Driver, @file{gcc}
121 @cindex controlling the compilation driver
123 @c prevent bad page break with this line
124 You can control the compilation driver.
126 @defmac DRIVER_SELF_SPECS
127 A list of specs for the driver itself. It should be a suitable
128 initializer for an array of strings, with no surrounding braces.
130 The driver applies these specs to its own command line between loading
131 default @file{specs} files (but not command-line specified ones) and
132 choosing the multilib directory or running any subcommands. It
133 applies them in the order given, so each spec can depend on the
134 options added by earlier ones. It is also possible to remove options
135 using @samp{%<@var{option}} in the usual way.
137 This macro can be useful when a port has several interdependent target
138 options. It provides a way of standardizing the command line so
139 that the other specs are easier to write.
141 Do not define this macro if it does not need to do anything.
144 @defmac OPTION_DEFAULT_SPECS
145 A list of specs used to support configure-time default options (i.e.@:
146 @option{--with} options) in the driver. It should be a suitable initializer
147 for an array of structures, each containing two strings, without the
148 outermost pair of surrounding braces.
150 The first item in the pair is the name of the default. This must match
151 the code in @file{config.gcc} for the target. The second item is a spec
152 to apply if a default with this name was specified. The string
153 @samp{%(VALUE)} in the spec will be replaced by the value of the default
154 everywhere it occurs.
156 The driver will apply these specs to its own command line between loading
157 default @file{specs} files and processing @code{DRIVER_SELF_SPECS}, using
158 the same mechanism as @code{DRIVER_SELF_SPECS}.
160 Do not define this macro if it does not need to do anything.
164 A C string constant that tells the GCC driver program options to
165 pass to CPP@. It can also specify how to translate options you
166 give to GCC into options for GCC to pass to the CPP@.
168 Do not define this macro if it does not need to do anything.
171 @defmac CPLUSPLUS_CPP_SPEC
172 This macro is just like @code{CPP_SPEC}, but is used for C++, rather
173 than C@. If you do not define this macro, then the value of
174 @code{CPP_SPEC} (if any) will be used instead.
178 A C string constant that tells the GCC driver program options to
179 pass to @code{cc1}, @code{cc1plus}, @code{f771}, and the other language
181 It can also specify how to translate options you give to GCC into options
182 for GCC to pass to front ends.
184 Do not define this macro if it does not need to do anything.
188 A C string constant that tells the GCC driver program options to
189 pass to @code{cc1plus}. It can also specify how to translate options you
190 give to GCC into options for GCC to pass to the @code{cc1plus}.
192 Do not define this macro if it does not need to do anything.
193 Note that everything defined in CC1_SPEC is already passed to
194 @code{cc1plus} so there is no need to duplicate the contents of
195 CC1_SPEC in CC1PLUS_SPEC@.
199 A C string constant that tells the GCC driver program options to
200 pass to the assembler. It can also specify how to translate options
201 you give to GCC into options for GCC to pass to the assembler.
202 See the file @file{sun3.h} for an example of this.
204 Do not define this macro if it does not need to do anything.
207 @defmac ASM_FINAL_SPEC
208 A C string constant that tells the GCC driver program how to
209 run any programs which cleanup after the normal assembler.
210 Normally, this is not needed. See the file @file{mips.h} for
213 Do not define this macro if it does not need to do anything.
216 @defmac AS_NEEDS_DASH_FOR_PIPED_INPUT
217 Define this macro, with no value, if the driver should give the assembler
218 an argument consisting of a single dash, @option{-}, to instruct it to
219 read from its standard input (which will be a pipe connected to the
220 output of the compiler proper). This argument is given after any
221 @option{-o} option specifying the name of the output file.
223 If you do not define this macro, the assembler is assumed to read its
224 standard input if given no non-option arguments. If your assembler
225 cannot read standard input at all, use a @samp{%@{pipe:%e@}} construct;
226 see @file{mips.h} for instance.
230 A C string constant that tells the GCC driver program options to
231 pass to the linker. It can also specify how to translate options you
232 give to GCC into options for GCC to pass to the linker.
234 Do not define this macro if it does not need to do anything.
238 Another C string constant used much like @code{LINK_SPEC}. The difference
239 between the two is that @code{LIB_SPEC} is used at the end of the
240 command given to the linker.
242 If this macro is not defined, a default is provided that
243 loads the standard C library from the usual place. See @file{gcc.c}.
247 Another C string constant that tells the GCC driver program
248 how and when to place a reference to @file{libgcc.a} into the
249 linker command line. This constant is placed both before and after
250 the value of @code{LIB_SPEC}.
252 If this macro is not defined, the GCC driver provides a default that
253 passes the string @option{-lgcc} to the linker.
256 @defmac REAL_LIBGCC_SPEC
257 By default, if @code{ENABLE_SHARED_LIBGCC} is defined, the
258 @code{LIBGCC_SPEC} is not directly used by the driver program but is
259 instead modified to refer to different versions of @file{libgcc.a}
260 depending on the values of the command line flags @option{-static},
261 @option{-shared}, @option{-static-libgcc}, and @option{-shared-libgcc}. On
262 targets where these modifications are inappropriate, define
263 @code{REAL_LIBGCC_SPEC} instead. @code{REAL_LIBGCC_SPEC} tells the
264 driver how to place a reference to @file{libgcc} on the link command
265 line, but, unlike @code{LIBGCC_SPEC}, it is used unmodified.
268 @defmac USE_LD_AS_NEEDED
269 A macro that controls the modifications to @code{LIBGCC_SPEC}
270 mentioned in @code{REAL_LIBGCC_SPEC}. If nonzero, a spec will be
271 generated that uses @option{--as-needed} or equivalent options and the
272 shared @file{libgcc} in place of the
273 static exception handler library, when linking without any of
274 @code{-static}, @code{-static-libgcc}, or @code{-shared-libgcc}.
278 If defined, this C string constant is added to @code{LINK_SPEC}.
279 When @code{USE_LD_AS_NEEDED} is zero or undefined, it also affects
280 the modifications to @code{LIBGCC_SPEC} mentioned in
281 @code{REAL_LIBGCC_SPEC}.
284 @defmac STARTFILE_SPEC
285 Another C string constant used much like @code{LINK_SPEC}. The
286 difference between the two is that @code{STARTFILE_SPEC} is used at
287 the very beginning of the command given to the linker.
289 If this macro is not defined, a default is provided that loads the
290 standard C startup file from the usual place. See @file{gcc.c}.
294 Another C string constant used much like @code{LINK_SPEC}. The
295 difference between the two is that @code{ENDFILE_SPEC} is used at
296 the very end of the command given to the linker.
298 Do not define this macro if it does not need to do anything.
301 @defmac THREAD_MODEL_SPEC
302 GCC @code{-v} will print the thread model GCC was configured to use.
303 However, this doesn't work on platforms that are multilibbed on thread
304 models, such as AIX 4.3. On such platforms, define
305 @code{THREAD_MODEL_SPEC} such that it evaluates to a string without
306 blanks that names one of the recognized thread models. @code{%*}, the
307 default value of this macro, will expand to the value of
308 @code{thread_file} set in @file{config.gcc}.
311 @defmac SYSROOT_SUFFIX_SPEC
312 Define this macro to add a suffix to the target sysroot when GCC is
313 configured with a sysroot. This will cause GCC to search for usr/lib,
314 et al, within sysroot+suffix.
317 @defmac SYSROOT_HEADERS_SUFFIX_SPEC
318 Define this macro to add a headers_suffix to the target sysroot when
319 GCC is configured with a sysroot. This will cause GCC to pass the
320 updated sysroot+headers_suffix to CPP, causing it to search for
321 usr/include, et al, within sysroot+headers_suffix.
325 Define this macro to provide additional specifications to put in the
326 @file{specs} file that can be used in various specifications like
329 The definition should be an initializer for an array of structures,
330 containing a string constant, that defines the specification name, and a
331 string constant that provides the specification.
333 Do not define this macro if it does not need to do anything.
335 @code{EXTRA_SPECS} is useful when an architecture contains several
336 related targets, which have various @code{@dots{}_SPECS} which are similar
337 to each other, and the maintainer would like one central place to keep
340 For example, the PowerPC System V.4 targets use @code{EXTRA_SPECS} to
341 define either @code{_CALL_SYSV} when the System V calling sequence is
342 used or @code{_CALL_AIX} when the older AIX-based calling sequence is
345 The @file{config/rs6000/rs6000.h} target file defines:
348 #define EXTRA_SPECS \
349 @{ "cpp_sysv_default", CPP_SYSV_DEFAULT @},
351 #define CPP_SYS_DEFAULT ""
354 The @file{config/rs6000/sysv.h} target file defines:
358 "%@{posix: -D_POSIX_SOURCE @} \
359 %@{mcall-sysv: -D_CALL_SYSV @} \
360 %@{!mcall-sysv: %(cpp_sysv_default) @} \
361 %@{msoft-float: -D_SOFT_FLOAT@} %@{mcpu=403: -D_SOFT_FLOAT@}"
363 #undef CPP_SYSV_DEFAULT
364 #define CPP_SYSV_DEFAULT "-D_CALL_SYSV"
367 while the @file{config/rs6000/eabiaix.h} target file defines
368 @code{CPP_SYSV_DEFAULT} as:
371 #undef CPP_SYSV_DEFAULT
372 #define CPP_SYSV_DEFAULT "-D_CALL_AIX"
376 @defmac LINK_LIBGCC_SPECIAL_1
377 Define this macro if the driver program should find the library
378 @file{libgcc.a}. If you do not define this macro, the driver program will pass
379 the argument @option{-lgcc} to tell the linker to do the search.
382 @defmac LINK_GCC_C_SEQUENCE_SPEC
383 The sequence in which libgcc and libc are specified to the linker.
384 By default this is @code{%G %L %G}.
387 @defmac POST_LINK_SPEC
388 Define this macro to add additional steps to be executed after linker.
389 The default value of this macro is empty string.
392 @defmac LINK_COMMAND_SPEC
393 A C string constant giving the complete command line need to execute the
394 linker. When you do this, you will need to update your port each time a
395 change is made to the link command line within @file{gcc.c}. Therefore,
396 define this macro only if you need to completely redefine the command
397 line for invoking the linker and there is no other way to accomplish
398 the effect you need. Overriding this macro may be avoidable by overriding
399 @code{LINK_GCC_C_SEQUENCE_SPEC} instead.
402 @hook TARGET_ALWAYS_STRIP_DOTDOT
404 @defmac MULTILIB_DEFAULTS
405 Define this macro as a C expression for the initializer of an array of
406 string to tell the driver program which options are defaults for this
407 target and thus do not need to be handled specially when using
408 @code{MULTILIB_OPTIONS}.
410 Do not define this macro if @code{MULTILIB_OPTIONS} is not defined in
411 the target makefile fragment or if none of the options listed in
412 @code{MULTILIB_OPTIONS} are set by default.
413 @xref{Target Fragment}.
416 @defmac RELATIVE_PREFIX_NOT_LINKDIR
417 Define this macro to tell @command{gcc} that it should only translate
418 a @option{-B} prefix into a @option{-L} linker option if the prefix
419 indicates an absolute file name.
422 @defmac MD_EXEC_PREFIX
423 If defined, this macro is an additional prefix to try after
424 @code{STANDARD_EXEC_PREFIX}. @code{MD_EXEC_PREFIX} is not searched
425 when the compiler is built as a cross
426 compiler. If you define @code{MD_EXEC_PREFIX}, then be sure to add it
427 to the list of directories used to find the assembler in @file{configure.ac}.
430 @defmac STANDARD_STARTFILE_PREFIX
431 Define this macro as a C string constant if you wish to override the
432 standard choice of @code{libdir} as the default prefix to
433 try when searching for startup files such as @file{crt0.o}.
434 @code{STANDARD_STARTFILE_PREFIX} is not searched when the compiler
435 is built as a cross compiler.
438 @defmac STANDARD_STARTFILE_PREFIX_1
439 Define this macro as a C string constant if you wish to override the
440 standard choice of @code{/lib} as a prefix to try after the default prefix
441 when searching for startup files such as @file{crt0.o}.
442 @code{STANDARD_STARTFILE_PREFIX_1} is not searched when the compiler
443 is built as a cross compiler.
446 @defmac STANDARD_STARTFILE_PREFIX_2
447 Define this macro as a C string constant if you wish to override the
448 standard choice of @code{/lib} as yet another prefix to try after the
449 default prefix when searching for startup files such as @file{crt0.o}.
450 @code{STANDARD_STARTFILE_PREFIX_2} is not searched when the compiler
451 is built as a cross compiler.
454 @defmac MD_STARTFILE_PREFIX
455 If defined, this macro supplies an additional prefix to try after the
456 standard prefixes. @code{MD_EXEC_PREFIX} is not searched when the
457 compiler is built as a cross compiler.
460 @defmac MD_STARTFILE_PREFIX_1
461 If defined, this macro supplies yet another prefix to try after the
462 standard prefixes. It is not searched when the compiler is built as a
466 @defmac INIT_ENVIRONMENT
467 Define this macro as a C string constant if you wish to set environment
468 variables for programs called by the driver, such as the assembler and
469 loader. The driver passes the value of this macro to @code{putenv} to
470 initialize the necessary environment variables.
473 @defmac LOCAL_INCLUDE_DIR
474 Define this macro as a C string constant if you wish to override the
475 standard choice of @file{/usr/local/include} as the default prefix to
476 try when searching for local header files. @code{LOCAL_INCLUDE_DIR}
477 comes before @code{NATIVE_SYSTEM_HEADER_DIR} (set in
478 @file{config.gcc}, normally @file{/usr/include}) in the search order.
480 Cross compilers do not search either @file{/usr/local/include} or its
484 @defmac NATIVE_SYSTEM_HEADER_COMPONENT
485 The ``component'' corresponding to @code{NATIVE_SYSTEM_HEADER_DIR}.
486 See @code{INCLUDE_DEFAULTS}, below, for the description of components.
487 If you do not define this macro, no component is used.
490 @defmac INCLUDE_DEFAULTS
491 Define this macro if you wish to override the entire default search path
492 for include files. For a native compiler, the default search path
493 usually consists of @code{GCC_INCLUDE_DIR}, @code{LOCAL_INCLUDE_DIR},
494 @code{GPLUSPLUS_INCLUDE_DIR}, and
495 @code{NATIVE_SYSTEM_HEADER_DIR}. In addition, @code{GPLUSPLUS_INCLUDE_DIR}
496 and @code{GCC_INCLUDE_DIR} are defined automatically by @file{Makefile},
497 and specify private search areas for GCC@. The directory
498 @code{GPLUSPLUS_INCLUDE_DIR} is used only for C++ programs.
500 The definition should be an initializer for an array of structures.
501 Each array element should have four elements: the directory name (a
502 string constant), the component name (also a string constant), a flag
503 for C++-only directories,
504 and a flag showing that the includes in the directory don't need to be
505 wrapped in @code{extern @samp{C}} when compiling C++. Mark the end of
506 the array with a null element.
508 The component name denotes what GNU package the include file is part of,
509 if any, in all uppercase letters. For example, it might be @samp{GCC}
510 or @samp{BINUTILS}. If the package is part of a vendor-supplied
511 operating system, code the component name as @samp{0}.
513 For example, here is the definition used for VAX/VMS:
516 #define INCLUDE_DEFAULTS \
518 @{ "GNU_GXX_INCLUDE:", "G++", 1, 1@}, \
519 @{ "GNU_CC_INCLUDE:", "GCC", 0, 0@}, \
520 @{ "SYS$SYSROOT:[SYSLIB.]", 0, 0, 0@}, \
527 Here is the order of prefixes tried for exec files:
531 Any prefixes specified by the user with @option{-B}.
534 The environment variable @code{GCC_EXEC_PREFIX} or, if @code{GCC_EXEC_PREFIX}
535 is not set and the compiler has not been installed in the configure-time
536 @var{prefix}, the location in which the compiler has actually been installed.
539 The directories specified by the environment variable @code{COMPILER_PATH}.
542 The macro @code{STANDARD_EXEC_PREFIX}, if the compiler has been installed
543 in the configured-time @var{prefix}.
546 The location @file{/usr/libexec/gcc/}, but only if this is a native compiler.
549 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
552 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
556 Here is the order of prefixes tried for startfiles:
560 Any prefixes specified by the user with @option{-B}.
563 The environment variable @code{GCC_EXEC_PREFIX} or its automatically determined
564 value based on the installed toolchain location.
567 The directories specified by the environment variable @code{LIBRARY_PATH}
568 (or port-specific name; native only, cross compilers do not use this).
571 The macro @code{STANDARD_EXEC_PREFIX}, but only if the toolchain is installed
572 in the configured @var{prefix} or this is a native compiler.
575 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
578 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
582 The macro @code{MD_STARTFILE_PREFIX}, if defined, but only if this is a
583 native compiler, or we have a target system root.
586 The macro @code{MD_STARTFILE_PREFIX_1}, if defined, but only if this is a
587 native compiler, or we have a target system root.
590 The macro @code{STANDARD_STARTFILE_PREFIX}, with any sysroot modifications.
591 If this path is relative it will be prefixed by @code{GCC_EXEC_PREFIX} and
592 the machine suffix or @code{STANDARD_EXEC_PREFIX} and the machine suffix.
595 The macro @code{STANDARD_STARTFILE_PREFIX_1}, but only if this is a native
596 compiler, or we have a target system root. The default for this macro is
600 The macro @code{STANDARD_STARTFILE_PREFIX_2}, but only if this is a native
601 compiler, or we have a target system root. The default for this macro is
605 @node Run-time Target
606 @section Run-time Target Specification
607 @cindex run-time target specification
608 @cindex predefined macros
609 @cindex target specifications
611 @c prevent bad page break with this line
612 Here are run-time target specifications.
614 @defmac TARGET_CPU_CPP_BUILTINS ()
615 This function-like macro expands to a block of code that defines
616 built-in preprocessor macros and assertions for the target CPU, using
617 the functions @code{builtin_define}, @code{builtin_define_std} and
618 @code{builtin_assert}. When the front end
619 calls this macro it provides a trailing semicolon, and since it has
620 finished command line option processing your code can use those
623 @code{builtin_assert} takes a string in the form you pass to the
624 command-line option @option{-A}, such as @code{cpu=mips}, and creates
625 the assertion. @code{builtin_define} takes a string in the form
626 accepted by option @option{-D} and unconditionally defines the macro.
628 @code{builtin_define_std} takes a string representing the name of an
629 object-like macro. If it doesn't lie in the user's namespace,
630 @code{builtin_define_std} defines it unconditionally. Otherwise, it
631 defines a version with two leading underscores, and another version
632 with two leading and trailing underscores, and defines the original
633 only if an ISO standard was not requested on the command line. For
634 example, passing @code{unix} defines @code{__unix}, @code{__unix__}
635 and possibly @code{unix}; passing @code{_mips} defines @code{__mips},
636 @code{__mips__} and possibly @code{_mips}, and passing @code{_ABI64}
637 defines only @code{_ABI64}.
639 You can also test for the C dialect being compiled. The variable
640 @code{c_language} is set to one of @code{clk_c}, @code{clk_cplusplus}
641 or @code{clk_objective_c}. Note that if we are preprocessing
642 assembler, this variable will be @code{clk_c} but the function-like
643 macro @code{preprocessing_asm_p()} will return true, so you might want
644 to check for that first. If you need to check for strict ANSI, the
645 variable @code{flag_iso} can be used. The function-like macro
646 @code{preprocessing_trad_p()} can be used to check for traditional
650 @defmac TARGET_OS_CPP_BUILTINS ()
651 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
652 and is used for the target operating system instead.
655 @defmac TARGET_OBJFMT_CPP_BUILTINS ()
656 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
657 and is used for the target object format. @file{elfos.h} uses this
658 macro to define @code{__ELF__}, so you probably do not need to define
662 @deftypevar {extern int} target_flags
663 This variable is declared in @file{options.h}, which is included before
664 any target-specific headers.
667 @hook TARGET_DEFAULT_TARGET_FLAGS
668 This variable specifies the initial value of @code{target_flags}.
669 Its default setting is 0.
672 @cindex optional hardware or system features
673 @cindex features, optional, in system conventions
675 @hook TARGET_HANDLE_OPTION
676 This hook is called whenever the user specifies one of the
677 target-specific options described by the @file{.opt} definition files
678 (@pxref{Options}). It has the opportunity to do some option-specific
679 processing and should return true if the option is valid. The default
680 definition does nothing but return true.
682 @var{decoded} specifies the option and its arguments. @var{opts} and
683 @var{opts_set} are the @code{gcc_options} structures to be used for
684 storing option state, and @var{loc} is the location at which the
685 option was passed (@code{UNKNOWN_LOCATION} except for options passed
689 @hook TARGET_HANDLE_C_OPTION
690 This target hook is called whenever the user specifies one of the
691 target-specific C language family options described by the @file{.opt}
692 definition files(@pxref{Options}). It has the opportunity to do some
693 option-specific processing and should return true if the option is
694 valid. The arguments are like for @code{TARGET_HANDLE_OPTION}. The
695 default definition does nothing but return false.
697 In general, you should use @code{TARGET_HANDLE_OPTION} to handle
698 options. However, if processing an option requires routines that are
699 only available in the C (and related language) front ends, then you
700 should use @code{TARGET_HANDLE_C_OPTION} instead.
703 @hook TARGET_OBJC_CONSTRUCT_STRING_OBJECT
705 @hook TARGET_OBJC_DECLARE_UNRESOLVED_CLASS_REFERENCE
707 @hook TARGET_OBJC_DECLARE_CLASS_DEFINITION
709 @hook TARGET_STRING_OBJECT_REF_TYPE_P
711 @hook TARGET_CHECK_STRING_OBJECT_FORMAT_ARG
713 @hook TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE
715 @defmac C_COMMON_OVERRIDE_OPTIONS
716 This is similar to the @code{TARGET_OPTION_OVERRIDE} hook
717 but is only used in the C
718 language frontends (C, Objective-C, C++, Objective-C++) and so can be
719 used to alter option flag variables which only exist in those
723 @hook TARGET_OPTION_OPTIMIZATION_TABLE
724 Some machines may desire to change what optimizations are performed for
725 various optimization levels. This variable, if defined, describes
726 options to enable at particular sets of optimization levels. These
727 options are processed once
728 just after the optimization level is determined and before the remainder
729 of the command options have been parsed, so may be overridden by other
730 options passed explicitly.
732 This processing is run once at program startup and when the optimization
733 options are changed via @code{#pragma GCC optimize} or by using the
734 @code{optimize} attribute.
737 @hook TARGET_OPTION_INIT_STRUCT
739 @hook TARGET_OPTION_DEFAULT_PARAMS
741 @hook TARGET_OPTION_VALIDATE_PARAM
743 @defmac SWITCHABLE_TARGET
744 Some targets need to switch between substantially different subtargets
745 during compilation. For example, the MIPS target has one subtarget for
746 the traditional MIPS architecture and another for MIPS16. Source code
747 can switch between these two subarchitectures using the @code{mips16}
748 and @code{nomips16} attributes.
750 Such subtargets can differ in things like the set of available
751 registers, the set of available instructions, the costs of various
752 operations, and so on. GCC caches a lot of this type of information
753 in global variables, and recomputing them for each subtarget takes a
754 significant amount of time. The compiler therefore provides a facility
755 for maintaining several versions of the global variables and quickly
756 switching between them; see @file{target-globals.h} for details.
758 Define this macro to 1 if your target needs this facility. The default
762 @hook TARGET_FLOAT_EXCEPTIONS_ROUNDING_SUPPORTED_P
764 @node Per-Function Data
765 @section Defining data structures for per-function information.
766 @cindex per-function data
767 @cindex data structures
769 If the target needs to store information on a per-function basis, GCC
770 provides a macro and a couple of variables to allow this. Note, just
771 using statics to store the information is a bad idea, since GCC supports
772 nested functions, so you can be halfway through encoding one function
773 when another one comes along.
775 GCC defines a data structure called @code{struct function} which
776 contains all of the data specific to an individual function. This
777 structure contains a field called @code{machine} whose type is
778 @code{struct machine_function *}, which can be used by targets to point
779 to their own specific data.
781 If a target needs per-function specific data it should define the type
782 @code{struct machine_function} and also the macro @code{INIT_EXPANDERS}.
783 This macro should be used to initialize the function pointer
784 @code{init_machine_status}. This pointer is explained below.
786 One typical use of per-function, target specific data is to create an
787 RTX to hold the register containing the function's return address. This
788 RTX can then be used to implement the @code{__builtin_return_address}
789 function, for level 0.
791 Note---earlier implementations of GCC used a single data area to hold
792 all of the per-function information. Thus when processing of a nested
793 function began the old per-function data had to be pushed onto a
794 stack, and when the processing was finished, it had to be popped off the
795 stack. GCC used to provide function pointers called
796 @code{save_machine_status} and @code{restore_machine_status} to handle
797 the saving and restoring of the target specific information. Since the
798 single data area approach is no longer used, these pointers are no
801 @defmac INIT_EXPANDERS
802 Macro called to initialize any target specific information. This macro
803 is called once per function, before generation of any RTL has begun.
804 The intention of this macro is to allow the initialization of the
805 function pointer @code{init_machine_status}.
808 @deftypevar {void (*)(struct function *)} init_machine_status
809 If this function pointer is non-@code{NULL} it will be called once per
810 function, before function compilation starts, in order to allow the
811 target to perform any target specific initialization of the
812 @code{struct function} structure. It is intended that this would be
813 used to initialize the @code{machine} of that structure.
815 @code{struct machine_function} structures are expected to be freed by GC@.
816 Generally, any memory that they reference must be allocated by using
817 GC allocation, including the structure itself.
821 @section Storage Layout
822 @cindex storage layout
824 Note that the definitions of the macros in this table which are sizes or
825 alignments measured in bits do not need to be constant. They can be C
826 expressions that refer to static variables, such as the @code{target_flags}.
827 @xref{Run-time Target}.
829 @defmac BITS_BIG_ENDIAN
830 Define this macro to have the value 1 if the most significant bit in a
831 byte has the lowest number; otherwise define it to have the value zero.
832 This means that bit-field instructions count from the most significant
833 bit. If the machine has no bit-field instructions, then this must still
834 be defined, but it doesn't matter which value it is defined to. This
835 macro need not be a constant.
837 This macro does not affect the way structure fields are packed into
838 bytes or words; that is controlled by @code{BYTES_BIG_ENDIAN}.
841 @defmac BYTES_BIG_ENDIAN
842 Define this macro to have the value 1 if the most significant byte in a
843 word has the lowest number. This macro need not be a constant.
846 @defmac WORDS_BIG_ENDIAN
847 Define this macro to have the value 1 if, in a multiword object, the
848 most significant word has the lowest number. This applies to both
849 memory locations and registers; see @code{REG_WORDS_BIG_ENDIAN} if the
850 order of words in memory is not the same as the order in registers. This
851 macro need not be a constant.
854 @defmac REG_WORDS_BIG_ENDIAN
855 On some machines, the order of words in a multiword object differs between
856 registers in memory. In such a situation, define this macro to describe
857 the order of words in a register. The macro @code{WORDS_BIG_ENDIAN} controls
858 the order of words in memory.
861 @defmac FLOAT_WORDS_BIG_ENDIAN
862 Define this macro to have the value 1 if @code{DFmode}, @code{XFmode} or
863 @code{TFmode} floating point numbers are stored in memory with the word
864 containing the sign bit at the lowest address; otherwise define it to
865 have the value 0. This macro need not be a constant.
867 You need not define this macro if the ordering is the same as for
871 @defmac BITS_PER_WORD
872 Number of bits in a word. If you do not define this macro, the default
873 is @code{BITS_PER_UNIT * UNITS_PER_WORD}.
876 @defmac MAX_BITS_PER_WORD
877 Maximum number of bits in a word. If this is undefined, the default is
878 @code{BITS_PER_WORD}. Otherwise, it is the constant value that is the
879 largest value that @code{BITS_PER_WORD} can have at run-time.
882 @defmac UNITS_PER_WORD
883 Number of storage units in a word; normally the size of a general-purpose
884 register, a power of two from 1 or 8.
887 @defmac MIN_UNITS_PER_WORD
888 Minimum number of units in a word. If this is undefined, the default is
889 @code{UNITS_PER_WORD}. Otherwise, it is the constant value that is the
890 smallest value that @code{UNITS_PER_WORD} can have at run-time.
894 Width of a pointer, in bits. You must specify a value no wider than the
895 width of @code{Pmode}. If it is not equal to the width of @code{Pmode},
896 you must define @code{POINTERS_EXTEND_UNSIGNED}. If you do not specify
897 a value the default is @code{BITS_PER_WORD}.
900 @defmac POINTERS_EXTEND_UNSIGNED
901 A C expression that determines how pointers should be extended from
902 @code{ptr_mode} to either @code{Pmode} or @code{word_mode}. It is
903 greater than zero if pointers should be zero-extended, zero if they
904 should be sign-extended, and negative if some other sort of conversion
905 is needed. In the last case, the extension is done by the target's
906 @code{ptr_extend} instruction.
908 You need not define this macro if the @code{ptr_mode}, @code{Pmode}
909 and @code{word_mode} are all the same width.
912 @defmac PROMOTE_MODE (@var{m}, @var{unsignedp}, @var{type})
913 A macro to update @var{m} and @var{unsignedp} when an object whose type
914 is @var{type} and which has the specified mode and signedness is to be
915 stored in a register. This macro is only called when @var{type} is a
918 On most RISC machines, which only have operations that operate on a full
919 register, define this macro to set @var{m} to @code{word_mode} if
920 @var{m} is an integer mode narrower than @code{BITS_PER_WORD}. In most
921 cases, only integer modes should be widened because wider-precision
922 floating-point operations are usually more expensive than their narrower
925 For most machines, the macro definition does not change @var{unsignedp}.
926 However, some machines, have instructions that preferentially handle
927 either signed or unsigned quantities of certain modes. For example, on
928 the DEC Alpha, 32-bit loads from memory and 32-bit add instructions
929 sign-extend the result to 64 bits. On such machines, set
930 @var{unsignedp} according to which kind of extension is more efficient.
932 Do not define this macro if it would never modify @var{m}.
935 @hook TARGET_C_EXCESS_PRECISION
937 @hook TARGET_PROMOTE_FUNCTION_MODE
939 @defmac PARM_BOUNDARY
940 Normal alignment required for function parameters on the stack, in
941 bits. All stack parameters receive at least this much alignment
942 regardless of data type. On most machines, this is the same as the
946 @defmac STACK_BOUNDARY
947 Define this macro to the minimum alignment enforced by hardware for the
948 stack pointer on this machine. The definition is a C expression for the
949 desired alignment (measured in bits). This value is used as a default
950 if @code{PREFERRED_STACK_BOUNDARY} is not defined. On most machines,
951 this should be the same as @code{PARM_BOUNDARY}.
954 @defmac PREFERRED_STACK_BOUNDARY
955 Define this macro if you wish to preserve a certain alignment for the
956 stack pointer, greater than what the hardware enforces. The definition
957 is a C expression for the desired alignment (measured in bits). This
958 macro must evaluate to a value equal to or larger than
959 @code{STACK_BOUNDARY}.
962 @defmac INCOMING_STACK_BOUNDARY
963 Define this macro if the incoming stack boundary may be different
964 from @code{PREFERRED_STACK_BOUNDARY}. This macro must evaluate
965 to a value equal to or larger than @code{STACK_BOUNDARY}.
968 @defmac FUNCTION_BOUNDARY
969 Alignment required for a function entry point, in bits.
972 @defmac BIGGEST_ALIGNMENT
973 Biggest alignment that any data type can require on this machine, in
974 bits. Note that this is not the biggest alignment that is supported,
975 just the biggest alignment that, when violated, may cause a fault.
978 @hook TARGET_ABSOLUTE_BIGGEST_ALIGNMENT
980 @defmac MALLOC_ABI_ALIGNMENT
981 Alignment, in bits, a C conformant malloc implementation has to
982 provide. If not defined, the default value is @code{BITS_PER_WORD}.
985 @defmac ATTRIBUTE_ALIGNED_VALUE
986 Alignment used by the @code{__attribute__ ((aligned))} construct. If
987 not defined, the default value is @code{BIGGEST_ALIGNMENT}.
990 @defmac MINIMUM_ATOMIC_ALIGNMENT
991 If defined, the smallest alignment, in bits, that can be given to an
992 object that can be referenced in one operation, without disturbing any
993 nearby object. Normally, this is @code{BITS_PER_UNIT}, but may be larger
994 on machines that don't have byte or half-word store operations.
997 @defmac BIGGEST_FIELD_ALIGNMENT
998 Biggest alignment that any structure or union field can require on this
999 machine, in bits. If defined, this overrides @code{BIGGEST_ALIGNMENT} for
1000 structure and union fields only, unless the field alignment has been set
1001 by the @code{__attribute__ ((aligned (@var{n})))} construct.
1004 @defmac ADJUST_FIELD_ALIGN (@var{field}, @var{type}, @var{computed})
1005 An expression for the alignment of a structure field @var{field} of
1006 type @var{type} if the alignment computed in the usual way (including
1007 applying of @code{BIGGEST_ALIGNMENT} and @code{BIGGEST_FIELD_ALIGNMENT} to the
1008 alignment) is @var{computed}. It overrides alignment only if the
1009 field alignment has not been set by the
1010 @code{__attribute__ ((aligned (@var{n})))} construct. Note that @var{field}
1011 may be @code{NULL_TREE} in case we just query for the minimum alignment
1012 of a field of type @var{type} in structure context.
1015 @defmac MAX_STACK_ALIGNMENT
1016 Biggest stack alignment guaranteed by the backend. Use this macro
1017 to specify the maximum alignment of a variable on stack.
1019 If not defined, the default value is @code{STACK_BOUNDARY}.
1021 @c FIXME: The default should be @code{PREFERRED_STACK_BOUNDARY}.
1022 @c But the fix for PR 32893 indicates that we can only guarantee
1023 @c maximum stack alignment on stack up to @code{STACK_BOUNDARY}, not
1024 @c @code{PREFERRED_STACK_BOUNDARY}, if stack alignment isn't supported.
1027 @defmac MAX_OFILE_ALIGNMENT
1028 Biggest alignment supported by the object file format of this machine.
1029 Use this macro to limit the alignment which can be specified using the
1030 @code{__attribute__ ((aligned (@var{n})))} construct for functions and
1031 objects with static storage duration. The alignment of automatic
1032 objects may exceed the object file format maximum up to the maximum
1033 supported by GCC. If not defined, the default value is
1034 @code{BIGGEST_ALIGNMENT}.
1036 On systems that use ELF, the default (in @file{config/elfos.h}) is
1037 the largest supported 32-bit ELF section alignment representable on
1038 a 32-bit host e.g.@: @samp{(((uint64_t) 1 << 28) * 8)}.
1039 On 32-bit ELF the largest supported section alignment in bits is
1040 @samp{(0x80000000 * 8)}, but this is not representable on 32-bit hosts.
1043 @hook TARGET_STATIC_RTX_ALIGNMENT
1045 @defmac DATA_ALIGNMENT (@var{type}, @var{basic-align})
1046 If defined, a C expression to compute the alignment for a variable in
1047 the static store. @var{type} is the data type, and @var{basic-align} is
1048 the alignment that the object would ordinarily have. The value of this
1049 macro is used instead of that alignment to align the object.
1051 If this macro is not defined, then @var{basic-align} is used.
1054 One use of this macro is to increase alignment of medium-size data to
1055 make it all fit in fewer cache lines. Another is to cause character
1056 arrays to be word-aligned so that @code{strcpy} calls that copy
1057 constants to character arrays can be done inline.
1060 @defmac DATA_ABI_ALIGNMENT (@var{type}, @var{basic-align})
1061 Similar to @code{DATA_ALIGNMENT}, but for the cases where the ABI mandates
1062 some alignment increase, instead of optimization only purposes. E.g.@
1063 AMD x86-64 psABI says that variables with array type larger than 15 bytes
1064 must be aligned to 16 byte boundaries.
1066 If this macro is not defined, then @var{basic-align} is used.
1069 @hook TARGET_CONSTANT_ALIGNMENT
1071 @defmac LOCAL_ALIGNMENT (@var{type}, @var{basic-align})
1072 If defined, a C expression to compute the alignment for a variable in
1073 the local store. @var{type} is the data type, and @var{basic-align} is
1074 the alignment that the object would ordinarily have. The value of this
1075 macro is used instead of that alignment to align the object.
1077 If this macro is not defined, then @var{basic-align} is used.
1079 One use of this macro is to increase alignment of medium-size data to
1080 make it all fit in fewer cache lines.
1082 If the value of this macro has a type, it should be an unsigned type.
1085 @hook TARGET_VECTOR_ALIGNMENT
1087 @defmac STACK_SLOT_ALIGNMENT (@var{type}, @var{mode}, @var{basic-align})
1088 If defined, a C expression to compute the alignment for stack slot.
1089 @var{type} is the data type, @var{mode} is the widest mode available,
1090 and @var{basic-align} is the alignment that the slot would ordinarily
1091 have. The value of this macro is used instead of that alignment to
1094 If this macro is not defined, then @var{basic-align} is used when
1095 @var{type} is @code{NULL}. Otherwise, @code{LOCAL_ALIGNMENT} will
1098 This macro is to set alignment of stack slot to the maximum alignment
1099 of all possible modes which the slot may have.
1101 If the value of this macro has a type, it should be an unsigned type.
1104 @defmac LOCAL_DECL_ALIGNMENT (@var{decl})
1105 If defined, a C expression to compute the alignment for a local
1106 variable @var{decl}.
1108 If this macro is not defined, then
1109 @code{LOCAL_ALIGNMENT (TREE_TYPE (@var{decl}), DECL_ALIGN (@var{decl}))}
1112 One use of this macro is to increase alignment of medium-size data to
1113 make it all fit in fewer cache lines.
1115 If the value of this macro has a type, it should be an unsigned type.
1118 @defmac MINIMUM_ALIGNMENT (@var{exp}, @var{mode}, @var{align})
1119 If defined, a C expression to compute the minimum required alignment
1120 for dynamic stack realignment purposes for @var{exp} (a type or decl),
1121 @var{mode}, assuming normal alignment @var{align}.
1123 If this macro is not defined, then @var{align} will be used.
1126 @defmac EMPTY_FIELD_BOUNDARY
1127 Alignment in bits to be given to a structure bit-field that follows an
1128 empty field such as @code{int : 0;}.
1130 If @code{PCC_BITFIELD_TYPE_MATTERS} is true, it overrides this macro.
1133 @defmac STRUCTURE_SIZE_BOUNDARY
1134 Number of bits which any structure or union's size must be a multiple of.
1135 Each structure or union's size is rounded up to a multiple of this.
1137 If you do not define this macro, the default is the same as
1138 @code{BITS_PER_UNIT}.
1141 @defmac STRICT_ALIGNMENT
1142 Define this macro to be the value 1 if instructions will fail to work
1143 if given data not on the nominal alignment. If instructions will merely
1144 go slower in that case, define this macro as 0.
1147 @defmac PCC_BITFIELD_TYPE_MATTERS
1148 Define this if you wish to imitate the way many other C compilers handle
1149 alignment of bit-fields and the structures that contain them.
1151 The behavior is that the type written for a named bit-field (@code{int},
1152 @code{short}, or other integer type) imposes an alignment for the entire
1153 structure, as if the structure really did contain an ordinary field of
1154 that type. In addition, the bit-field is placed within the structure so
1155 that it would fit within such a field, not crossing a boundary for it.
1157 Thus, on most machines, a named bit-field whose type is written as
1158 @code{int} would not cross a four-byte boundary, and would force
1159 four-byte alignment for the whole structure. (The alignment used may
1160 not be four bytes; it is controlled by the other alignment parameters.)
1162 An unnamed bit-field will not affect the alignment of the containing
1165 If the macro is defined, its definition should be a C expression;
1166 a nonzero value for the expression enables this behavior.
1168 Note that if this macro is not defined, or its value is zero, some
1169 bit-fields may cross more than one alignment boundary. The compiler can
1170 support such references if there are @samp{insv}, @samp{extv}, and
1171 @samp{extzv} insns that can directly reference memory.
1173 The other known way of making bit-fields work is to define
1174 @code{STRUCTURE_SIZE_BOUNDARY} as large as @code{BIGGEST_ALIGNMENT}.
1175 Then every structure can be accessed with fullwords.
1177 Unless the machine has bit-field instructions or you define
1178 @code{STRUCTURE_SIZE_BOUNDARY} that way, you must define
1179 @code{PCC_BITFIELD_TYPE_MATTERS} to have a nonzero value.
1181 If your aim is to make GCC use the same conventions for laying out
1182 bit-fields as are used by another compiler, here is how to investigate
1183 what the other compiler does. Compile and run this program:
1202 printf ("Size of foo1 is %d\n",
1203 sizeof (struct foo1));
1204 printf ("Size of foo2 is %d\n",
1205 sizeof (struct foo2));
1210 If this prints 2 and 5, then the compiler's behavior is what you would
1211 get from @code{PCC_BITFIELD_TYPE_MATTERS}.
1214 @defmac BITFIELD_NBYTES_LIMITED
1215 Like @code{PCC_BITFIELD_TYPE_MATTERS} except that its effect is limited
1216 to aligning a bit-field within the structure.
1219 @hook TARGET_ALIGN_ANON_BITFIELD
1221 @hook TARGET_NARROW_VOLATILE_BITFIELD
1223 @hook TARGET_MEMBER_TYPE_FORCES_BLK
1225 @defmac ROUND_TYPE_ALIGN (@var{type}, @var{computed}, @var{specified})
1226 Define this macro as an expression for the alignment of a type (given
1227 by @var{type} as a tree node) if the alignment computed in the usual
1228 way is @var{computed} and the alignment explicitly specified was
1231 The default is to use @var{specified} if it is larger; otherwise, use
1232 the smaller of @var{computed} and @code{BIGGEST_ALIGNMENT}
1235 @defmac MAX_FIXED_MODE_SIZE
1236 An integer expression for the size in bits of the largest integer
1237 machine mode that should actually be used. All integer machine modes of
1238 this size or smaller can be used for structures and unions with the
1239 appropriate sizes. If this macro is undefined, @code{GET_MODE_BITSIZE
1240 (DImode)} is assumed.
1243 @defmac STACK_SAVEAREA_MODE (@var{save_level})
1244 If defined, an expression of type @code{machine_mode} that
1245 specifies the mode of the save area operand of a
1246 @code{save_stack_@var{level}} named pattern (@pxref{Standard Names}).
1247 @var{save_level} is one of @code{SAVE_BLOCK}, @code{SAVE_FUNCTION}, or
1248 @code{SAVE_NONLOCAL} and selects which of the three named patterns is
1249 having its mode specified.
1251 You need not define this macro if it always returns @code{Pmode}. You
1252 would most commonly define this macro if the
1253 @code{save_stack_@var{level}} patterns need to support both a 32- and a
1257 @defmac STACK_SIZE_MODE
1258 If defined, an expression of type @code{machine_mode} that
1259 specifies the mode of the size increment operand of an
1260 @code{allocate_stack} named pattern (@pxref{Standard Names}).
1262 You need not define this macro if it always returns @code{word_mode}.
1263 You would most commonly define this macro if the @code{allocate_stack}
1264 pattern needs to support both a 32- and a 64-bit mode.
1267 @hook TARGET_LIBGCC_CMP_RETURN_MODE
1269 @hook TARGET_LIBGCC_SHIFT_COUNT_MODE
1271 @hook TARGET_UNWIND_WORD_MODE
1273 @hook TARGET_MS_BITFIELD_LAYOUT_P
1275 @hook TARGET_DECIMAL_FLOAT_SUPPORTED_P
1277 @hook TARGET_FIXED_POINT_SUPPORTED_P
1279 @hook TARGET_EXPAND_TO_RTL_HOOK
1281 @hook TARGET_INSTANTIATE_DECLS
1283 @hook TARGET_MANGLE_TYPE
1286 @section Layout of Source Language Data Types
1288 These macros define the sizes and other characteristics of the standard
1289 basic data types used in programs being compiled. Unlike the macros in
1290 the previous section, these apply to specific features of C and related
1291 languages, rather than to fundamental aspects of storage layout.
1293 @defmac INT_TYPE_SIZE
1294 A C expression for the size in bits of the type @code{int} on the
1295 target machine. If you don't define this, the default is one word.
1298 @defmac SHORT_TYPE_SIZE
1299 A C expression for the size in bits of the type @code{short} on the
1300 target machine. If you don't define this, the default is half a word.
1301 (If this would be less than one storage unit, it is rounded up to one
1305 @defmac LONG_TYPE_SIZE
1306 A C expression for the size in bits of the type @code{long} on the
1307 target machine. If you don't define this, the default is one word.
1310 @defmac ADA_LONG_TYPE_SIZE
1311 On some machines, the size used for the Ada equivalent of the type
1312 @code{long} by a native Ada compiler differs from that used by C@. In
1313 that situation, define this macro to be a C expression to be used for
1314 the size of that type. If you don't define this, the default is the
1315 value of @code{LONG_TYPE_SIZE}.
1318 @defmac LONG_LONG_TYPE_SIZE
1319 A C expression for the size in bits of the type @code{long long} on the
1320 target machine. If you don't define this, the default is two
1321 words. If you want to support GNU Ada on your machine, the value of this
1322 macro must be at least 64.
1325 @defmac CHAR_TYPE_SIZE
1326 A C expression for the size in bits of the type @code{char} on the
1327 target machine. If you don't define this, the default is
1328 @code{BITS_PER_UNIT}.
1331 @defmac BOOL_TYPE_SIZE
1332 A C expression for the size in bits of the C++ type @code{bool} and
1333 C99 type @code{_Bool} on the target machine. If you don't define
1334 this, and you probably shouldn't, the default is @code{CHAR_TYPE_SIZE}.
1337 @defmac FLOAT_TYPE_SIZE
1338 A C expression for the size in bits of the type @code{float} on the
1339 target machine. If you don't define this, the default is one word.
1342 @defmac DOUBLE_TYPE_SIZE
1343 A C expression for the size in bits of the type @code{double} on the
1344 target machine. If you don't define this, the default is two
1348 @defmac LONG_DOUBLE_TYPE_SIZE
1349 A C expression for the size in bits of the type @code{long double} on
1350 the target machine. If you don't define this, the default is two
1354 @defmac SHORT_FRACT_TYPE_SIZE
1355 A C expression for the size in bits of the type @code{short _Fract} on
1356 the target machine. If you don't define this, the default is
1357 @code{BITS_PER_UNIT}.
1360 @defmac FRACT_TYPE_SIZE
1361 A C expression for the size in bits of the type @code{_Fract} on
1362 the target machine. If you don't define this, the default is
1363 @code{BITS_PER_UNIT * 2}.
1366 @defmac LONG_FRACT_TYPE_SIZE
1367 A C expression for the size in bits of the type @code{long _Fract} on
1368 the target machine. If you don't define this, the default is
1369 @code{BITS_PER_UNIT * 4}.
1372 @defmac LONG_LONG_FRACT_TYPE_SIZE
1373 A C expression for the size in bits of the type @code{long long _Fract} on
1374 the target machine. If you don't define this, the default is
1375 @code{BITS_PER_UNIT * 8}.
1378 @defmac SHORT_ACCUM_TYPE_SIZE
1379 A C expression for the size in bits of the type @code{short _Accum} on
1380 the target machine. If you don't define this, the default is
1381 @code{BITS_PER_UNIT * 2}.
1384 @defmac ACCUM_TYPE_SIZE
1385 A C expression for the size in bits of the type @code{_Accum} on
1386 the target machine. If you don't define this, the default is
1387 @code{BITS_PER_UNIT * 4}.
1390 @defmac LONG_ACCUM_TYPE_SIZE
1391 A C expression for the size in bits of the type @code{long _Accum} on
1392 the target machine. If you don't define this, the default is
1393 @code{BITS_PER_UNIT * 8}.
1396 @defmac LONG_LONG_ACCUM_TYPE_SIZE
1397 A C expression for the size in bits of the type @code{long long _Accum} on
1398 the target machine. If you don't define this, the default is
1399 @code{BITS_PER_UNIT * 16}.
1402 @defmac LIBGCC2_GNU_PREFIX
1403 This macro corresponds to the @code{TARGET_LIBFUNC_GNU_PREFIX} target
1404 hook and should be defined if that hook is overriden to be true. It
1405 causes function names in libgcc to be changed to use a @code{__gnu_}
1406 prefix for their name rather than the default @code{__}. A port which
1407 uses this macro should also arrange to use @file{t-gnu-prefix} in
1408 the libgcc @file{config.host}.
1411 @defmac WIDEST_HARDWARE_FP_SIZE
1412 A C expression for the size in bits of the widest floating-point format
1413 supported by the hardware. If you define this macro, you must specify a
1414 value less than or equal to the value of @code{LONG_DOUBLE_TYPE_SIZE}.
1415 If you do not define this macro, the value of @code{LONG_DOUBLE_TYPE_SIZE}
1419 @defmac DEFAULT_SIGNED_CHAR
1420 An expression whose value is 1 or 0, according to whether the type
1421 @code{char} should be signed or unsigned by default. The user can
1422 always override this default with the options @option{-fsigned-char}
1423 and @option{-funsigned-char}.
1426 @hook TARGET_DEFAULT_SHORT_ENUMS
1429 A C expression for a string describing the name of the data type to use
1430 for size values. The typedef name @code{size_t} is defined using the
1431 contents of the string.
1433 The string can contain more than one keyword. If so, separate them with
1434 spaces, and write first any length keyword, then @code{unsigned} if
1435 appropriate, and finally @code{int}. The string must exactly match one
1436 of the data type names defined in the function
1437 @code{c_common_nodes_and_builtins} in the file @file{c-family/c-common.c}.
1438 You may not omit @code{int} or change the order---that would cause the
1439 compiler to crash on startup.
1441 If you don't define this macro, the default is @code{"long unsigned
1446 GCC defines internal types (@code{sizetype}, @code{ssizetype},
1447 @code{bitsizetype} and @code{sbitsizetype}) for expressions
1448 dealing with size. This macro is a C expression for a string describing
1449 the name of the data type from which the precision of @code{sizetype}
1452 The string has the same restrictions as @code{SIZE_TYPE} string.
1454 If you don't define this macro, the default is @code{SIZE_TYPE}.
1457 @defmac PTRDIFF_TYPE
1458 A C expression for a string describing the name of the data type to use
1459 for the result of subtracting two pointers. The typedef name
1460 @code{ptrdiff_t} is defined using the contents of the string. See
1461 @code{SIZE_TYPE} above for more information.
1463 If you don't define this macro, the default is @code{"long int"}.
1467 A C expression for a string describing the name of the data type to use
1468 for wide characters. The typedef name @code{wchar_t} is defined using
1469 the contents of the string. See @code{SIZE_TYPE} above for more
1472 If you don't define this macro, the default is @code{"int"}.
1475 @defmac WCHAR_TYPE_SIZE
1476 A C expression for the size in bits of the data type for wide
1477 characters. This is used in @code{cpp}, which cannot make use of
1482 A C expression for a string describing the name of the data type to
1483 use for wide characters passed to @code{printf} and returned from
1484 @code{getwc}. The typedef name @code{wint_t} is defined using the
1485 contents of the string. See @code{SIZE_TYPE} above for more
1488 If you don't define this macro, the default is @code{"unsigned int"}.
1492 A C expression for a string describing the name of the data type that
1493 can represent any value of any standard or extended signed integer type.
1494 The typedef name @code{intmax_t} is defined using the contents of the
1495 string. See @code{SIZE_TYPE} above for more information.
1497 If you don't define this macro, the default is the first of
1498 @code{"int"}, @code{"long int"}, or @code{"long long int"} that has as
1499 much precision as @code{long long int}.
1502 @defmac UINTMAX_TYPE
1503 A C expression for a string describing the name of the data type that
1504 can represent any value of any standard or extended unsigned integer
1505 type. The typedef name @code{uintmax_t} is defined using the contents
1506 of the string. See @code{SIZE_TYPE} above for more information.
1508 If you don't define this macro, the default is the first of
1509 @code{"unsigned int"}, @code{"long unsigned int"}, or @code{"long long
1510 unsigned int"} that has as much precision as @code{long long unsigned
1514 @defmac SIG_ATOMIC_TYPE
1520 @defmacx UINT16_TYPE
1521 @defmacx UINT32_TYPE
1522 @defmacx UINT64_TYPE
1523 @defmacx INT_LEAST8_TYPE
1524 @defmacx INT_LEAST16_TYPE
1525 @defmacx INT_LEAST32_TYPE
1526 @defmacx INT_LEAST64_TYPE
1527 @defmacx UINT_LEAST8_TYPE
1528 @defmacx UINT_LEAST16_TYPE
1529 @defmacx UINT_LEAST32_TYPE
1530 @defmacx UINT_LEAST64_TYPE
1531 @defmacx INT_FAST8_TYPE
1532 @defmacx INT_FAST16_TYPE
1533 @defmacx INT_FAST32_TYPE
1534 @defmacx INT_FAST64_TYPE
1535 @defmacx UINT_FAST8_TYPE
1536 @defmacx UINT_FAST16_TYPE
1537 @defmacx UINT_FAST32_TYPE
1538 @defmacx UINT_FAST64_TYPE
1539 @defmacx INTPTR_TYPE
1540 @defmacx UINTPTR_TYPE
1541 C expressions for the standard types @code{sig_atomic_t},
1542 @code{int8_t}, @code{int16_t}, @code{int32_t}, @code{int64_t},
1543 @code{uint8_t}, @code{uint16_t}, @code{uint32_t}, @code{uint64_t},
1544 @code{int_least8_t}, @code{int_least16_t}, @code{int_least32_t},
1545 @code{int_least64_t}, @code{uint_least8_t}, @code{uint_least16_t},
1546 @code{uint_least32_t}, @code{uint_least64_t}, @code{int_fast8_t},
1547 @code{int_fast16_t}, @code{int_fast32_t}, @code{int_fast64_t},
1548 @code{uint_fast8_t}, @code{uint_fast16_t}, @code{uint_fast32_t},
1549 @code{uint_fast64_t}, @code{intptr_t}, and @code{uintptr_t}. See
1550 @code{SIZE_TYPE} above for more information.
1552 If any of these macros evaluates to a null pointer, the corresponding
1553 type is not supported; if GCC is configured to provide
1554 @code{<stdint.h>} in such a case, the header provided may not conform
1555 to C99, depending on the type in question. The defaults for all of
1556 these macros are null pointers.
1559 @defmac TARGET_PTRMEMFUNC_VBIT_LOCATION
1560 The C++ compiler represents a pointer-to-member-function with a struct
1567 ptrdiff_t vtable_index;
1574 The C++ compiler must use one bit to indicate whether the function that
1575 will be called through a pointer-to-member-function is virtual.
1576 Normally, we assume that the low-order bit of a function pointer must
1577 always be zero. Then, by ensuring that the vtable_index is odd, we can
1578 distinguish which variant of the union is in use. But, on some
1579 platforms function pointers can be odd, and so this doesn't work. In
1580 that case, we use the low-order bit of the @code{delta} field, and shift
1581 the remainder of the @code{delta} field to the left.
1583 GCC will automatically make the right selection about where to store
1584 this bit using the @code{FUNCTION_BOUNDARY} setting for your platform.
1585 However, some platforms such as ARM/Thumb have @code{FUNCTION_BOUNDARY}
1586 set such that functions always start at even addresses, but the lowest
1587 bit of pointers to functions indicate whether the function at that
1588 address is in ARM or Thumb mode. If this is the case of your
1589 architecture, you should define this macro to
1590 @code{ptrmemfunc_vbit_in_delta}.
1592 In general, you should not have to define this macro. On architectures
1593 in which function addresses are always even, according to
1594 @code{FUNCTION_BOUNDARY}, GCC will automatically define this macro to
1595 @code{ptrmemfunc_vbit_in_pfn}.
1598 @defmac TARGET_VTABLE_USES_DESCRIPTORS
1599 Normally, the C++ compiler uses function pointers in vtables. This
1600 macro allows the target to change to use ``function descriptors''
1601 instead. Function descriptors are found on targets for whom a
1602 function pointer is actually a small data structure. Normally the
1603 data structure consists of the actual code address plus a data
1604 pointer to which the function's data is relative.
1606 If vtables are used, the value of this macro should be the number
1607 of words that the function descriptor occupies.
1610 @defmac TARGET_VTABLE_ENTRY_ALIGN
1611 By default, the vtable entries are void pointers, the so the alignment
1612 is the same as pointer alignment. The value of this macro specifies
1613 the alignment of the vtable entry in bits. It should be defined only
1614 when special alignment is necessary. */
1617 @defmac TARGET_VTABLE_DATA_ENTRY_DISTANCE
1618 There are a few non-descriptor entries in the vtable at offsets below
1619 zero. If these entries must be padded (say, to preserve the alignment
1620 specified by @code{TARGET_VTABLE_ENTRY_ALIGN}), set this to the number
1621 of words in each data entry.
1625 @section Register Usage
1626 @cindex register usage
1628 This section explains how to describe what registers the target machine
1629 has, and how (in general) they can be used.
1631 The description of which registers a specific instruction can use is
1632 done with register classes; see @ref{Register Classes}. For information
1633 on using registers to access a stack frame, see @ref{Frame Registers}.
1634 For passing values in registers, see @ref{Register Arguments}.
1635 For returning values in registers, see @ref{Scalar Return}.
1638 * Register Basics:: Number and kinds of registers.
1639 * Allocation Order:: Order in which registers are allocated.
1640 * Values in Registers:: What kinds of values each reg can hold.
1641 * Leaf Functions:: Renumbering registers for leaf functions.
1642 * Stack Registers:: Handling a register stack such as 80387.
1645 @node Register Basics
1646 @subsection Basic Characteristics of Registers
1648 @c prevent bad page break with this line
1649 Registers have various characteristics.
1651 @defmac FIRST_PSEUDO_REGISTER
1652 Number of hardware registers known to the compiler. They receive
1653 numbers 0 through @code{FIRST_PSEUDO_REGISTER-1}; thus, the first
1654 pseudo register's number really is assigned the number
1655 @code{FIRST_PSEUDO_REGISTER}.
1658 @defmac FIXED_REGISTERS
1659 @cindex fixed register
1660 An initializer that says which registers are used for fixed purposes
1661 all throughout the compiled code and are therefore not available for
1662 general allocation. These would include the stack pointer, the frame
1663 pointer (except on machines where that can be used as a general
1664 register when no frame pointer is needed), the program counter on
1665 machines where that is considered one of the addressable registers,
1666 and any other numbered register with a standard use.
1668 This information is expressed as a sequence of numbers, separated by
1669 commas and surrounded by braces. The @var{n}th number is 1 if
1670 register @var{n} is fixed, 0 otherwise.
1672 The table initialized from this macro, and the table initialized by
1673 the following one, may be overridden at run time either automatically,
1674 by the actions of the macro @code{CONDITIONAL_REGISTER_USAGE}, or by
1675 the user with the command options @option{-ffixed-@var{reg}},
1676 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}.
1679 @defmac CALL_USED_REGISTERS
1680 @cindex call-used register
1681 @cindex call-clobbered register
1682 @cindex call-saved register
1683 Like @code{FIXED_REGISTERS} but has 1 for each register that is
1684 clobbered (in general) by function calls as well as for fixed
1685 registers. This macro therefore identifies the registers that are not
1686 available for general allocation of values that must live across
1689 If a register has 0 in @code{CALL_USED_REGISTERS}, the compiler
1690 automatically saves it on function entry and restores it on function
1691 exit, if the register is used within the function.
1693 Exactly one of @code{CALL_USED_REGISTERS} and @code{CALL_REALLY_USED_REGISTERS}
1694 must be defined. Modern ports should define @code{CALL_REALLY_USED_REGISTERS}.
1697 @defmac CALL_REALLY_USED_REGISTERS
1698 @cindex call-used register
1699 @cindex call-clobbered register
1700 @cindex call-saved register
1701 Like @code{CALL_USED_REGISTERS} except this macro doesn't require
1702 that the entire set of @code{FIXED_REGISTERS} be included.
1703 (@code{CALL_USED_REGISTERS} must be a superset of @code{FIXED_REGISTERS}).
1705 Exactly one of @code{CALL_USED_REGISTERS} and @code{CALL_REALLY_USED_REGISTERS}
1706 must be defined. Modern ports should define @code{CALL_REALLY_USED_REGISTERS}.
1709 @cindex call-used register
1710 @cindex call-clobbered register
1711 @cindex call-saved register
1712 @hook TARGET_HARD_REGNO_CALL_PART_CLOBBERED
1714 @hook TARGET_REMOVE_EXTRA_CALL_PRESERVED_REGS
1716 @hook TARGET_RETURN_CALL_WITH_MAX_CLOBBERS
1718 @hook TARGET_GET_MULTILIB_ABI_NAME
1721 @findex call_used_regs
1724 @findex reg_class_contents
1725 @hook TARGET_CONDITIONAL_REGISTER_USAGE
1727 @defmac INCOMING_REGNO (@var{out})
1728 Define this macro if the target machine has register windows. This C
1729 expression returns the register number as seen by the called function
1730 corresponding to the register number @var{out} as seen by the calling
1731 function. Return @var{out} if register number @var{out} is not an
1735 @defmac OUTGOING_REGNO (@var{in})
1736 Define this macro if the target machine has register windows. This C
1737 expression returns the register number as seen by the calling function
1738 corresponding to the register number @var{in} as seen by the called
1739 function. Return @var{in} if register number @var{in} is not an inbound
1743 @defmac LOCAL_REGNO (@var{regno})
1744 Define this macro if the target machine has register windows. This C
1745 expression returns true if the register is call-saved but is in the
1746 register window. Unlike most call-saved registers, such registers
1747 need not be explicitly restored on function exit or during non-local
1752 If the program counter has a register number, define this as that
1753 register number. Otherwise, do not define it.
1756 @node Allocation Order
1757 @subsection Order of Allocation of Registers
1758 @cindex order of register allocation
1759 @cindex register allocation order
1761 @c prevent bad page break with this line
1762 Registers are allocated in order.
1764 @defmac REG_ALLOC_ORDER
1765 If defined, an initializer for a vector of integers, containing the
1766 numbers of hard registers in the order in which GCC should prefer
1767 to use them (from most preferred to least).
1769 If this macro is not defined, registers are used lowest numbered first
1770 (all else being equal).
1772 One use of this macro is on machines where the highest numbered
1773 registers must always be saved and the save-multiple-registers
1774 instruction supports only sequences of consecutive registers. On such
1775 machines, define @code{REG_ALLOC_ORDER} to be an initializer that lists
1776 the highest numbered allocable register first.
1779 @defmac ADJUST_REG_ALLOC_ORDER
1780 A C statement (sans semicolon) to choose the order in which to allocate
1781 hard registers for pseudo-registers local to a basic block.
1783 Store the desired register order in the array @code{reg_alloc_order}.
1784 Element 0 should be the register to allocate first; element 1, the next
1785 register; and so on.
1787 The macro body should not assume anything about the contents of
1788 @code{reg_alloc_order} before execution of the macro.
1790 On most machines, it is not necessary to define this macro.
1793 @defmac HONOR_REG_ALLOC_ORDER
1794 Normally, IRA tries to estimate the costs for saving a register in the
1795 prologue and restoring it in the epilogue. This discourages it from
1796 using call-saved registers. If a machine wants to ensure that IRA
1797 allocates registers in the order given by REG_ALLOC_ORDER even if some
1798 call-saved registers appear earlier than call-used ones, then define this
1799 macro as a C expression to nonzero. Default is 0.
1802 @defmac IRA_HARD_REGNO_ADD_COST_MULTIPLIER (@var{regno})
1803 In some case register allocation order is not enough for the
1804 Integrated Register Allocator (@acronym{IRA}) to generate a good code.
1805 If this macro is defined, it should return a floating point value
1806 based on @var{regno}. The cost of using @var{regno} for a pseudo will
1807 be increased by approximately the pseudo's usage frequency times the
1808 value returned by this macro. Not defining this macro is equivalent
1809 to having it always return @code{0.0}.
1811 On most machines, it is not necessary to define this macro.
1814 @node Values in Registers
1815 @subsection How Values Fit in Registers
1817 This section discusses the macros that describe which kinds of values
1818 (specifically, which machine modes) each register can hold, and how many
1819 consecutive registers are needed for a given mode.
1821 @hook TARGET_HARD_REGNO_NREGS
1823 @defmac HARD_REGNO_NREGS_HAS_PADDING (@var{regno}, @var{mode})
1824 A C expression that is nonzero if a value of mode @var{mode}, stored
1825 in memory, ends with padding that causes it to take up more space than
1826 in registers starting at register number @var{regno} (as determined by
1827 multiplying GCC's notion of the size of the register when containing
1828 this mode by the number of registers returned by
1829 @code{TARGET_HARD_REGNO_NREGS}). By default this is zero.
1831 For example, if a floating-point value is stored in three 32-bit
1832 registers but takes up 128 bits in memory, then this would be
1835 This macros only needs to be defined if there are cases where
1836 @code{subreg_get_info}
1837 would otherwise wrongly determine that a @code{subreg} can be
1838 represented by an offset to the register number, when in fact such a
1839 @code{subreg} would contain some of the padding not stored in
1840 registers and so not be representable.
1843 @defmac HARD_REGNO_NREGS_WITH_PADDING (@var{regno}, @var{mode})
1844 For values of @var{regno} and @var{mode} for which
1845 @code{HARD_REGNO_NREGS_HAS_PADDING} returns nonzero, a C expression
1846 returning the greater number of registers required to hold the value
1847 including any padding. In the example above, the value would be four.
1850 @defmac REGMODE_NATURAL_SIZE (@var{mode})
1851 Define this macro if the natural size of registers that hold values
1852 of mode @var{mode} is not the word size. It is a C expression that
1853 should give the natural size in bytes for the specified mode. It is
1854 used by the register allocator to try to optimize its results. This
1855 happens for example on SPARC 64-bit where the natural size of
1856 floating-point registers is still 32-bit.
1859 @hook TARGET_HARD_REGNO_MODE_OK
1861 @defmac HARD_REGNO_RENAME_OK (@var{from}, @var{to})
1862 A C expression that is nonzero if it is OK to rename a hard register
1863 @var{from} to another hard register @var{to}.
1865 One common use of this macro is to prevent renaming of a register to
1866 another register that is not saved by a prologue in an interrupt
1869 The default is always nonzero.
1872 @hook TARGET_MODES_TIEABLE_P
1874 @hook TARGET_HARD_REGNO_SCRATCH_OK
1876 @defmac AVOID_CCMODE_COPIES
1877 Define this macro if the compiler should avoid copies to/from @code{CCmode}
1878 registers. You should only define this macro if support for copying to/from
1879 @code{CCmode} is incomplete.
1882 @node Leaf Functions
1883 @subsection Handling Leaf Functions
1885 @cindex leaf functions
1886 @cindex functions, leaf
1887 On some machines, a leaf function (i.e., one which makes no calls) can run
1888 more efficiently if it does not make its own register window. Often this
1889 means it is required to receive its arguments in the registers where they
1890 are passed by the caller, instead of the registers where they would
1893 The special treatment for leaf functions generally applies only when
1894 other conditions are met; for example, often they may use only those
1895 registers for its own variables and temporaries. We use the term ``leaf
1896 function'' to mean a function that is suitable for this special
1897 handling, so that functions with no calls are not necessarily ``leaf
1900 GCC assigns register numbers before it knows whether the function is
1901 suitable for leaf function treatment. So it needs to renumber the
1902 registers in order to output a leaf function. The following macros
1905 @defmac LEAF_REGISTERS
1906 Name of a char vector, indexed by hard register number, which
1907 contains 1 for a register that is allowable in a candidate for leaf
1910 If leaf function treatment involves renumbering the registers, then the
1911 registers marked here should be the ones before renumbering---those that
1912 GCC would ordinarily allocate. The registers which will actually be
1913 used in the assembler code, after renumbering, should not be marked with 1
1916 Define this macro only if the target machine offers a way to optimize
1917 the treatment of leaf functions.
1920 @defmac LEAF_REG_REMAP (@var{regno})
1921 A C expression whose value is the register number to which @var{regno}
1922 should be renumbered, when a function is treated as a leaf function.
1924 If @var{regno} is a register number which should not appear in a leaf
1925 function before renumbering, then the expression should yield @minus{}1, which
1926 will cause the compiler to abort.
1928 Define this macro only if the target machine offers a way to optimize the
1929 treatment of leaf functions, and registers need to be renumbered to do
1933 @findex current_function_is_leaf
1934 @findex current_function_uses_only_leaf_regs
1935 @code{TARGET_ASM_FUNCTION_PROLOGUE} and
1936 @code{TARGET_ASM_FUNCTION_EPILOGUE} must usually treat leaf functions
1937 specially. They can test the C variable @code{current_function_is_leaf}
1938 which is nonzero for leaf functions. @code{current_function_is_leaf} is
1939 set prior to local register allocation and is valid for the remaining
1940 compiler passes. They can also test the C variable
1941 @code{current_function_uses_only_leaf_regs} which is nonzero for leaf
1942 functions which only use leaf registers.
1943 @code{current_function_uses_only_leaf_regs} is valid after all passes
1944 that modify the instructions have been run and is only useful if
1945 @code{LEAF_REGISTERS} is defined.
1946 @c changed this to fix overfull. ALSO: why the "it" at the beginning
1947 @c of the next paragraph?! --mew 2feb93
1949 @node Stack Registers
1950 @subsection Registers That Form a Stack
1952 There are special features to handle computers where some of the
1953 ``registers'' form a stack. Stack registers are normally written by
1954 pushing onto the stack, and are numbered relative to the top of the
1957 Currently, GCC can only handle one group of stack-like registers, and
1958 they must be consecutively numbered. Furthermore, the existing
1959 support for stack-like registers is specific to the 80387 floating
1960 point coprocessor. If you have a new architecture that uses
1961 stack-like registers, you will need to do substantial work on
1962 @file{reg-stack.c} and write your machine description to cooperate
1963 with it, as well as defining these macros.
1966 Define this if the machine has any stack-like registers.
1969 @defmac STACK_REG_COVER_CLASS
1970 This is a cover class containing the stack registers. Define this if
1971 the machine has any stack-like registers.
1974 @defmac FIRST_STACK_REG
1975 The number of the first stack-like register. This one is the top
1979 @defmac LAST_STACK_REG
1980 The number of the last stack-like register. This one is the bottom of
1984 @node Register Classes
1985 @section Register Classes
1986 @cindex register class definitions
1987 @cindex class definitions, register
1989 On many machines, the numbered registers are not all equivalent.
1990 For example, certain registers may not be allowed for indexed addressing;
1991 certain registers may not be allowed in some instructions. These machine
1992 restrictions are described to the compiler using @dfn{register classes}.
1994 You define a number of register classes, giving each one a name and saying
1995 which of the registers belong to it. Then you can specify register classes
1996 that are allowed as operands to particular instruction patterns.
2000 In general, each register will belong to several classes. In fact, one
2001 class must be named @code{ALL_REGS} and contain all the registers. Another
2002 class must be named @code{NO_REGS} and contain no registers. Often the
2003 union of two classes will be another class; however, this is not required.
2005 @findex GENERAL_REGS
2006 One of the classes must be named @code{GENERAL_REGS}. There is nothing
2007 terribly special about the name, but the operand constraint letters
2008 @samp{r} and @samp{g} specify this class. If @code{GENERAL_REGS} is
2009 the same as @code{ALL_REGS}, just define it as a macro which expands
2012 Order the classes so that if class @var{x} is contained in class @var{y}
2013 then @var{x} has a lower class number than @var{y}.
2015 The way classes other than @code{GENERAL_REGS} are specified in operand
2016 constraints is through machine-dependent operand constraint letters.
2017 You can define such letters to correspond to various classes, then use
2018 them in operand constraints.
2020 You must define the narrowest register classes for allocatable
2021 registers, so that each class either has no subclasses, or that for
2022 some mode, the move cost between registers within the class is
2023 cheaper than moving a register in the class to or from memory
2026 You should define a class for the union of two classes whenever some
2027 instruction allows both classes. For example, if an instruction allows
2028 either a floating point (coprocessor) register or a general register for a
2029 certain operand, you should define a class @code{FLOAT_OR_GENERAL_REGS}
2030 which includes both of them. Otherwise you will get suboptimal code,
2031 or even internal compiler errors when reload cannot find a register in the
2032 class computed via @code{reg_class_subunion}.
2034 You must also specify certain redundant information about the register
2035 classes: for each class, which classes contain it and which ones are
2036 contained in it; for each pair of classes, the largest class contained
2039 When a value occupying several consecutive registers is expected in a
2040 certain class, all the registers used must belong to that class.
2041 Therefore, register classes cannot be used to enforce a requirement for
2042 a register pair to start with an even-numbered register. The way to
2043 specify this requirement is with @code{TARGET_HARD_REGNO_MODE_OK}.
2045 Register classes used for input-operands of bitwise-and or shift
2046 instructions have a special requirement: each such class must have, for
2047 each fixed-point machine mode, a subclass whose registers can transfer that
2048 mode to or from memory. For example, on some machines, the operations for
2049 single-byte values (@code{QImode}) are limited to certain registers. When
2050 this is so, each register class that is used in a bitwise-and or shift
2051 instruction must have a subclass consisting of registers from which
2052 single-byte values can be loaded or stored. This is so that
2053 @code{PREFERRED_RELOAD_CLASS} can always have a possible value to return.
2055 @deftp {Data type} {enum reg_class}
2056 An enumerated type that must be defined with all the register class names
2057 as enumerated values. @code{NO_REGS} must be first. @code{ALL_REGS}
2058 must be the last register class, followed by one more enumerated value,
2059 @code{LIM_REG_CLASSES}, which is not a register class but rather
2060 tells how many classes there are.
2062 Each register class has a number, which is the value of casting
2063 the class name to type @code{int}. The number serves as an index
2064 in many of the tables described below.
2067 @defmac N_REG_CLASSES
2068 The number of distinct register classes, defined as follows:
2071 #define N_REG_CLASSES (int) LIM_REG_CLASSES
2075 @defmac REG_CLASS_NAMES
2076 An initializer containing the names of the register classes as C string
2077 constants. These names are used in writing some of the debugging dumps.
2080 @defmac REG_CLASS_CONTENTS
2081 An initializer containing the contents of the register classes, as integers
2082 which are bit masks. The @var{n}th integer specifies the contents of class
2083 @var{n}. The way the integer @var{mask} is interpreted is that
2084 register @var{r} is in the class if @code{@var{mask} & (1 << @var{r})} is 1.
2086 When the machine has more than 32 registers, an integer does not suffice.
2087 Then the integers are replaced by sub-initializers, braced groupings containing
2088 several integers. Each sub-initializer must be suitable as an initializer
2089 for the type @code{HARD_REG_SET} which is defined in @file{hard-reg-set.h}.
2090 In this situation, the first integer in each sub-initializer corresponds to
2091 registers 0 through 31, the second integer to registers 32 through 63, and
2095 @defmac REGNO_REG_CLASS (@var{regno})
2096 A C expression whose value is a register class containing hard register
2097 @var{regno}. In general there is more than one such class; choose a class
2098 which is @dfn{minimal}, meaning that no smaller class also contains the
2102 @defmac BASE_REG_CLASS
2103 A macro whose definition is the name of the class to which a valid
2104 base register must belong. A base register is one used in an address
2105 which is the register value plus a displacement.
2108 @defmac MODE_BASE_REG_CLASS (@var{mode})
2109 This is a variation of the @code{BASE_REG_CLASS} macro which allows
2110 the selection of a base register in a mode dependent manner. If
2111 @var{mode} is VOIDmode then it should return the same value as
2112 @code{BASE_REG_CLASS}.
2115 @defmac MODE_BASE_REG_REG_CLASS (@var{mode})
2116 A C expression whose value is the register class to which a valid
2117 base register must belong in order to be used in a base plus index
2118 register address. You should define this macro if base plus index
2119 addresses have different requirements than other base register uses.
2122 @defmac MODE_CODE_BASE_REG_CLASS (@var{mode}, @var{address_space}, @var{outer_code}, @var{index_code})
2123 A C expression whose value is the register class to which a valid
2124 base register for a memory reference in mode @var{mode} to address
2125 space @var{address_space} must belong. @var{outer_code} and @var{index_code}
2126 define the context in which the base register occurs. @var{outer_code} is
2127 the code of the immediately enclosing expression (@code{MEM} for the top level
2128 of an address, @code{ADDRESS} for something that occurs in an
2129 @code{address_operand}). @var{index_code} is the code of the corresponding
2130 index expression if @var{outer_code} is @code{PLUS}; @code{SCRATCH} otherwise.
2133 @defmac INDEX_REG_CLASS
2134 A macro whose definition is the name of the class to which a valid
2135 index register must belong. An index register is one used in an
2136 address where its value is either multiplied by a scale factor or
2137 added to another register (as well as added to a displacement).
2140 @defmac REGNO_OK_FOR_BASE_P (@var{num})
2141 A C expression which is nonzero if register number @var{num} is
2142 suitable for use as a base register in operand addresses.
2145 @defmac REGNO_MODE_OK_FOR_BASE_P (@var{num}, @var{mode})
2146 A C expression that is just like @code{REGNO_OK_FOR_BASE_P}, except that
2147 that expression may examine the mode of the memory reference in
2148 @var{mode}. You should define this macro if the mode of the memory
2149 reference affects whether a register may be used as a base register. If
2150 you define this macro, the compiler will use it instead of
2151 @code{REGNO_OK_FOR_BASE_P}. The mode may be @code{VOIDmode} for
2152 addresses that appear outside a @code{MEM}, i.e., as an
2153 @code{address_operand}.
2156 @defmac REGNO_MODE_OK_FOR_REG_BASE_P (@var{num}, @var{mode})
2157 A C expression which is nonzero if register number @var{num} is suitable for
2158 use as a base register in base plus index operand addresses, accessing
2159 memory in mode @var{mode}. It may be either a suitable hard register or a
2160 pseudo register that has been allocated such a hard register. You should
2161 define this macro if base plus index addresses have different requirements
2162 than other base register uses.
2164 Use of this macro is deprecated; please use the more general
2165 @code{REGNO_MODE_CODE_OK_FOR_BASE_P}.
2168 @defmac REGNO_MODE_CODE_OK_FOR_BASE_P (@var{num}, @var{mode}, @var{address_space}, @var{outer_code}, @var{index_code})
2169 A C expression which is nonzero if register number @var{num} is
2170 suitable for use as a base register in operand addresses, accessing
2171 memory in mode @var{mode} in address space @var{address_space}.
2172 This is similar to @code{REGNO_MODE_OK_FOR_BASE_P}, except
2173 that that expression may examine the context in which the register
2174 appears in the memory reference. @var{outer_code} is the code of the
2175 immediately enclosing expression (@code{MEM} if at the top level of the
2176 address, @code{ADDRESS} for something that occurs in an
2177 @code{address_operand}). @var{index_code} is the code of the
2178 corresponding index expression if @var{outer_code} is @code{PLUS};
2179 @code{SCRATCH} otherwise. The mode may be @code{VOIDmode} for addresses
2180 that appear outside a @code{MEM}, i.e., as an @code{address_operand}.
2183 @defmac REGNO_OK_FOR_INDEX_P (@var{num})
2184 A C expression which is nonzero if register number @var{num} is
2185 suitable for use as an index register in operand addresses. It may be
2186 either a suitable hard register or a pseudo register that has been
2187 allocated such a hard register.
2189 The difference between an index register and a base register is that
2190 the index register may be scaled. If an address involves the sum of
2191 two registers, neither one of them scaled, then either one may be
2192 labeled the ``base'' and the other the ``index''; but whichever
2193 labeling is used must fit the machine's constraints of which registers
2194 may serve in each capacity. The compiler will try both labelings,
2195 looking for one that is valid, and will reload one or both registers
2196 only if neither labeling works.
2199 @hook TARGET_PREFERRED_RENAME_CLASS
2201 @hook TARGET_PREFERRED_RELOAD_CLASS
2203 @defmac PREFERRED_RELOAD_CLASS (@var{x}, @var{class})
2204 A C expression that places additional restrictions on the register class
2205 to use when it is necessary to copy value @var{x} into a register in class
2206 @var{class}. The value is a register class; perhaps @var{class}, or perhaps
2207 another, smaller class. On many machines, the following definition is
2211 #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS
2214 Sometimes returning a more restrictive class makes better code. For
2215 example, on the 68000, when @var{x} is an integer constant that is in range
2216 for a @samp{moveq} instruction, the value of this macro is always
2217 @code{DATA_REGS} as long as @var{class} includes the data registers.
2218 Requiring a data register guarantees that a @samp{moveq} will be used.
2220 One case where @code{PREFERRED_RELOAD_CLASS} must not return
2221 @var{class} is if @var{x} is a legitimate constant which cannot be
2222 loaded into some register class. By returning @code{NO_REGS} you can
2223 force @var{x} into a memory location. For example, rs6000 can load
2224 immediate values into general-purpose registers, but does not have an
2225 instruction for loading an immediate value into a floating-point
2226 register, so @code{PREFERRED_RELOAD_CLASS} returns @code{NO_REGS} when
2227 @var{x} is a floating-point constant. If the constant cannot be loaded
2228 into any kind of register, code generation will be better if
2229 @code{TARGET_LEGITIMATE_CONSTANT_P} makes the constant illegitimate instead
2230 of using @code{TARGET_PREFERRED_RELOAD_CLASS}.
2232 If an insn has pseudos in it after register allocation, reload will go
2233 through the alternatives and call repeatedly @code{PREFERRED_RELOAD_CLASS}
2234 to find the best one. Returning @code{NO_REGS}, in this case, makes
2235 reload add a @code{!} in front of the constraint: the x86 back-end uses
2236 this feature to discourage usage of 387 registers when math is done in
2237 the SSE registers (and vice versa).
2240 @hook TARGET_PREFERRED_OUTPUT_RELOAD_CLASS
2242 @defmac LIMIT_RELOAD_CLASS (@var{mode}, @var{class})
2243 A C expression that places additional restrictions on the register class
2244 to use when it is necessary to be able to hold a value of mode
2245 @var{mode} in a reload register for which class @var{class} would
2248 Unlike @code{PREFERRED_RELOAD_CLASS}, this macro should be used when
2249 there are certain modes that simply cannot go in certain reload classes.
2251 The value is a register class; perhaps @var{class}, or perhaps another,
2254 Don't define this macro unless the target machine has limitations which
2255 require the macro to do something nontrivial.
2258 @hook TARGET_SECONDARY_RELOAD
2260 @defmac SECONDARY_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2261 @defmacx SECONDARY_INPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2262 @defmacx SECONDARY_OUTPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2263 These macros are obsolete, new ports should use the target hook
2264 @code{TARGET_SECONDARY_RELOAD} instead.
2266 These are obsolete macros, replaced by the @code{TARGET_SECONDARY_RELOAD}
2267 target hook. Older ports still define these macros to indicate to the
2268 reload phase that it may
2269 need to allocate at least one register for a reload in addition to the
2270 register to contain the data. Specifically, if copying @var{x} to a
2271 register @var{class} in @var{mode} requires an intermediate register,
2272 you were supposed to define @code{SECONDARY_INPUT_RELOAD_CLASS} to return the
2273 largest register class all of whose registers can be used as
2274 intermediate registers or scratch registers.
2276 If copying a register @var{class} in @var{mode} to @var{x} requires an
2277 intermediate or scratch register, @code{SECONDARY_OUTPUT_RELOAD_CLASS}
2278 was supposed to be defined be defined to return the largest register
2279 class required. If the
2280 requirements for input and output reloads were the same, the macro
2281 @code{SECONDARY_RELOAD_CLASS} should have been used instead of defining both
2284 The values returned by these macros are often @code{GENERAL_REGS}.
2285 Return @code{NO_REGS} if no spare register is needed; i.e., if @var{x}
2286 can be directly copied to or from a register of @var{class} in
2287 @var{mode} without requiring a scratch register. Do not define this
2288 macro if it would always return @code{NO_REGS}.
2290 If a scratch register is required (either with or without an
2291 intermediate register), you were supposed to define patterns for
2292 @samp{reload_in@var{m}} or @samp{reload_out@var{m}}, as required
2293 (@pxref{Standard Names}. These patterns, which were normally
2294 implemented with a @code{define_expand}, should be similar to the
2295 @samp{mov@var{m}} patterns, except that operand 2 is the scratch
2298 These patterns need constraints for the reload register and scratch
2300 contain a single register class. If the original reload register (whose
2301 class is @var{class}) can meet the constraint given in the pattern, the
2302 value returned by these macros is used for the class of the scratch
2303 register. Otherwise, two additional reload registers are required.
2304 Their classes are obtained from the constraints in the insn pattern.
2306 @var{x} might be a pseudo-register or a @code{subreg} of a
2307 pseudo-register, which could either be in a hard register or in memory.
2308 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2309 in memory and the hard register number if it is in a register.
2311 These macros should not be used in the case where a particular class of
2312 registers can only be copied to memory and not to another class of
2313 registers. In that case, secondary reload registers are not needed and
2314 would not be helpful. Instead, a stack location must be used to perform
2315 the copy and the @code{mov@var{m}} pattern should use memory as an
2316 intermediate storage. This case often occurs between floating-point and
2320 @hook TARGET_SECONDARY_MEMORY_NEEDED
2322 @defmac SECONDARY_MEMORY_NEEDED_RTX (@var{mode})
2323 Normally when @code{TARGET_SECONDARY_MEMORY_NEEDED} is defined, the compiler
2324 allocates a stack slot for a memory location needed for register copies.
2325 If this macro is defined, the compiler instead uses the memory location
2326 defined by this macro.
2328 Do not define this macro if you do not define
2329 @code{TARGET_SECONDARY_MEMORY_NEEDED}.
2332 @hook TARGET_SECONDARY_MEMORY_NEEDED_MODE
2334 @hook TARGET_SELECT_EARLY_REMAT_MODES
2336 @hook TARGET_CLASS_LIKELY_SPILLED_P
2338 @hook TARGET_CLASS_MAX_NREGS
2340 @defmac CLASS_MAX_NREGS (@var{class}, @var{mode})
2341 A C expression for the maximum number of consecutive registers
2342 of class @var{class} needed to hold a value of mode @var{mode}.
2344 This is closely related to the macro @code{TARGET_HARD_REGNO_NREGS}. In fact,
2345 the value of the macro @code{CLASS_MAX_NREGS (@var{class}, @var{mode})}
2346 should be the maximum value of @code{TARGET_HARD_REGNO_NREGS (@var{regno},
2347 @var{mode})} for all @var{regno} values in the class @var{class}.
2349 This macro helps control the handling of multiple-word values
2353 @hook TARGET_CAN_CHANGE_MODE_CLASS
2355 @hook TARGET_IRA_CHANGE_PSEUDO_ALLOCNO_CLASS
2359 @hook TARGET_REGISTER_PRIORITY
2361 @hook TARGET_REGISTER_USAGE_LEVELING_P
2363 @hook TARGET_DIFFERENT_ADDR_DISPLACEMENT_P
2365 @hook TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P
2367 @hook TARGET_LEGITIMIZE_ADDRESS_DISPLACEMENT
2369 @hook TARGET_SPILL_CLASS
2371 @hook TARGET_ADDITIONAL_ALLOCNO_CLASS_P
2373 @hook TARGET_CSTORE_MODE
2375 @hook TARGET_COMPUTE_PRESSURE_CLASSES
2377 @node Stack and Calling
2378 @section Stack Layout and Calling Conventions
2379 @cindex calling conventions
2381 @c prevent bad page break with this line
2382 This describes the stack layout and calling conventions.
2386 * Exception Handling::
2391 * Register Arguments::
2393 * Aggregate Return::
2398 * Shrink-wrapping separate components::
2399 * Stack Smashing Protection::
2400 * Miscellaneous Register Hooks::
2404 @subsection Basic Stack Layout
2405 @cindex stack frame layout
2406 @cindex frame layout
2408 @c prevent bad page break with this line
2409 Here is the basic stack layout.
2411 @defmac STACK_GROWS_DOWNWARD
2412 Define this macro to be true if pushing a word onto the stack moves the stack
2413 pointer to a smaller address, and false otherwise.
2416 @defmac STACK_PUSH_CODE
2417 This macro defines the operation used when something is pushed
2418 on the stack. In RTL, a push operation will be
2419 @code{(set (mem (STACK_PUSH_CODE (reg sp))) @dots{})}
2421 The choices are @code{PRE_DEC}, @code{POST_DEC}, @code{PRE_INC},
2422 and @code{POST_INC}. Which of these is correct depends on
2423 the stack direction and on whether the stack pointer points
2424 to the last item on the stack or whether it points to the
2425 space for the next item on the stack.
2427 The default is @code{PRE_DEC} when @code{STACK_GROWS_DOWNWARD} is
2428 true, which is almost always right, and @code{PRE_INC} otherwise,
2429 which is often wrong.
2432 @defmac FRAME_GROWS_DOWNWARD
2433 Define this macro to nonzero value if the addresses of local variable slots
2434 are at negative offsets from the frame pointer.
2437 @defmac ARGS_GROW_DOWNWARD
2438 Define this macro if successive arguments to a function occupy decreasing
2439 addresses on the stack.
2442 @hook TARGET_STARTING_FRAME_OFFSET
2444 @defmac STACK_ALIGNMENT_NEEDED
2445 Define to zero to disable final alignment of the stack during reload.
2446 The nonzero default for this macro is suitable for most ports.
2448 On ports where @code{TARGET_STARTING_FRAME_OFFSET} is nonzero or where there
2449 is a register save block following the local block that doesn't require
2450 alignment to @code{STACK_BOUNDARY}, it may be beneficial to disable
2451 stack alignment and do it in the backend.
2454 @defmac STACK_POINTER_OFFSET
2455 Offset from the stack pointer register to the first location at which
2456 outgoing arguments are placed. If not specified, the default value of
2457 zero is used. This is the proper value for most machines.
2459 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
2460 the first location at which outgoing arguments are placed.
2463 @defmac FIRST_PARM_OFFSET (@var{fundecl})
2464 Offset from the argument pointer register to the first argument's
2465 address. On some machines it may depend on the data type of the
2468 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
2469 the first argument's address.
2472 @defmac STACK_DYNAMIC_OFFSET (@var{fundecl})
2473 Offset from the stack pointer register to an item dynamically allocated
2474 on the stack, e.g., by @code{alloca}.
2476 The default value for this macro is @code{STACK_POINTER_OFFSET} plus the
2477 length of the outgoing arguments. The default is correct for most
2478 machines. See @file{function.c} for details.
2481 @defmac INITIAL_FRAME_ADDRESS_RTX
2482 A C expression whose value is RTL representing the address of the initial
2483 stack frame. This address is passed to @code{RETURN_ADDR_RTX} and
2484 @code{DYNAMIC_CHAIN_ADDRESS}. If you don't define this macro, a reasonable
2485 default value will be used. Define this macro in order to make frame pointer
2486 elimination work in the presence of @code{__builtin_frame_address (count)} and
2487 @code{__builtin_return_address (count)} for @code{count} not equal to zero.
2490 @defmac DYNAMIC_CHAIN_ADDRESS (@var{frameaddr})
2491 A C expression whose value is RTL representing the address in a stack
2492 frame where the pointer to the caller's frame is stored. Assume that
2493 @var{frameaddr} is an RTL expression for the address of the stack frame
2496 If you don't define this macro, the default is to return the value
2497 of @var{frameaddr}---that is, the stack frame address is also the
2498 address of the stack word that points to the previous frame.
2501 @defmac SETUP_FRAME_ADDRESSES
2502 A C expression that produces the machine-specific code to
2503 setup the stack so that arbitrary frames can be accessed. For example,
2504 on the SPARC, we must flush all of the register windows to the stack
2505 before we can access arbitrary stack frames. You will seldom need to
2506 define this macro. The default is to do nothing.
2509 @hook TARGET_BUILTIN_SETJMP_FRAME_VALUE
2511 @defmac FRAME_ADDR_RTX (@var{frameaddr})
2512 A C expression whose value is RTL representing the value of the frame
2513 address for the current frame. @var{frameaddr} is the frame pointer
2514 of the current frame. This is used for __builtin_frame_address.
2515 You need only define this macro if the frame address is not the same
2516 as the frame pointer. Most machines do not need to define it.
2519 @defmac RETURN_ADDR_RTX (@var{count}, @var{frameaddr})
2520 A C expression whose value is RTL representing the value of the return
2521 address for the frame @var{count} steps up from the current frame, after
2522 the prologue. @var{frameaddr} is the frame pointer of the @var{count}
2523 frame, or the frame pointer of the @var{count} @minus{} 1 frame if
2524 @code{RETURN_ADDR_IN_PREVIOUS_FRAME} is nonzero.
2526 The value of the expression must always be the correct address when
2527 @var{count} is zero, but may be @code{NULL_RTX} if there is no way to
2528 determine the return address of other frames.
2531 @defmac RETURN_ADDR_IN_PREVIOUS_FRAME
2532 Define this macro to nonzero value if the return address of a particular
2533 stack frame is accessed from the frame pointer of the previous stack
2534 frame. The zero default for this macro is suitable for most ports.
2537 @defmac INCOMING_RETURN_ADDR_RTX
2538 A C expression whose value is RTL representing the location of the
2539 incoming return address at the beginning of any function, before the
2540 prologue. This RTL is either a @code{REG}, indicating that the return
2541 value is saved in @samp{REG}, or a @code{MEM} representing a location in
2544 You only need to define this macro if you want to support call frame
2545 debugging information like that provided by DWARF 2.
2547 If this RTL is a @code{REG}, you should also define
2548 @code{DWARF_FRAME_RETURN_COLUMN} to @code{DWARF_FRAME_REGNUM (REGNO)}.
2551 @defmac DWARF_ALT_FRAME_RETURN_COLUMN
2552 A C expression whose value is an integer giving a DWARF 2 column
2553 number that may be used as an alternative return column. The column
2554 must not correspond to any gcc hard register (that is, it must not
2555 be in the range of @code{DWARF_FRAME_REGNUM}).
2557 This macro can be useful if @code{DWARF_FRAME_RETURN_COLUMN} is set to a
2558 general register, but an alternative column needs to be used for signal
2559 frames. Some targets have also used different frame return columns
2563 @defmac DWARF_ZERO_REG
2564 A C expression whose value is an integer giving a DWARF 2 register
2565 number that is considered to always have the value zero. This should
2566 only be defined if the target has an architected zero register, and
2567 someone decided it was a good idea to use that register number to
2568 terminate the stack backtrace. New ports should avoid this.
2571 @hook TARGET_DWARF_HANDLE_FRAME_UNSPEC
2573 @hook TARGET_DWARF_POLY_INDETERMINATE_VALUE
2575 @defmac INCOMING_FRAME_SP_OFFSET
2576 A C expression whose value is an integer giving the offset, in bytes,
2577 from the value of the stack pointer register to the top of the stack
2578 frame at the beginning of any function, before the prologue. The top of
2579 the frame is defined to be the value of the stack pointer in the
2580 previous frame, just before the call instruction.
2582 You only need to define this macro if you want to support call frame
2583 debugging information like that provided by DWARF 2.
2586 @defmac DEFAULT_INCOMING_FRAME_SP_OFFSET
2587 Like @code{INCOMING_FRAME_SP_OFFSET}, but must be the same for all
2588 functions of the same ABI, and when using GAS @code{.cfi_*} directives
2589 must also agree with the default CFI GAS emits. Define this macro
2590 only if @code{INCOMING_FRAME_SP_OFFSET} can have different values
2591 between different functions of the same ABI or when
2592 @code{INCOMING_FRAME_SP_OFFSET} does not agree with GAS default CFI.
2595 @defmac ARG_POINTER_CFA_OFFSET (@var{fundecl})
2596 A C expression whose value is an integer giving the offset, in bytes,
2597 from the argument pointer to the canonical frame address (cfa). The
2598 final value should coincide with that calculated by
2599 @code{INCOMING_FRAME_SP_OFFSET}. Which is unfortunately not usable
2600 during virtual register instantiation.
2602 The default value for this macro is
2603 @code{FIRST_PARM_OFFSET (fundecl) + crtl->args.pretend_args_size},
2604 which is correct for most machines; in general, the arguments are found
2605 immediately before the stack frame. Note that this is not the case on
2606 some targets that save registers into the caller's frame, such as SPARC
2607 and rs6000, and so such targets need to define this macro.
2609 You only need to define this macro if the default is incorrect, and you
2610 want to support call frame debugging information like that provided by
2614 @defmac FRAME_POINTER_CFA_OFFSET (@var{fundecl})
2615 If defined, a C expression whose value is an integer giving the offset
2616 in bytes from the frame pointer to the canonical frame address (cfa).
2617 The final value should coincide with that calculated by
2618 @code{INCOMING_FRAME_SP_OFFSET}.
2620 Normally the CFA is calculated as an offset from the argument pointer,
2621 via @code{ARG_POINTER_CFA_OFFSET}, but if the argument pointer is
2622 variable due to the ABI, this may not be possible. If this macro is
2623 defined, it implies that the virtual register instantiation should be
2624 based on the frame pointer instead of the argument pointer. Only one
2625 of @code{FRAME_POINTER_CFA_OFFSET} and @code{ARG_POINTER_CFA_OFFSET}
2629 @defmac CFA_FRAME_BASE_OFFSET (@var{fundecl})
2630 If defined, a C expression whose value is an integer giving the offset
2631 in bytes from the canonical frame address (cfa) to the frame base used
2632 in DWARF 2 debug information. The default is zero. A different value
2633 may reduce the size of debug information on some ports.
2636 @node Exception Handling
2637 @subsection Exception Handling Support
2638 @cindex exception handling
2640 @defmac EH_RETURN_DATA_REGNO (@var{N})
2641 A C expression whose value is the @var{N}th register number used for
2642 data by exception handlers, or @code{INVALID_REGNUM} if fewer than
2643 @var{N} registers are usable.
2645 The exception handling library routines communicate with the exception
2646 handlers via a set of agreed upon registers. Ideally these registers
2647 should be call-clobbered; it is possible to use call-saved registers,
2648 but may negatively impact code size. The target must support at least
2649 2 data registers, but should define 4 if there are enough free registers.
2651 You must define this macro if you want to support call frame exception
2652 handling like that provided by DWARF 2.
2655 @defmac EH_RETURN_STACKADJ_RTX
2656 A C expression whose value is RTL representing a location in which
2657 to store a stack adjustment to be applied before function return.
2658 This is used to unwind the stack to an exception handler's call frame.
2659 It will be assigned zero on code paths that return normally.
2661 Typically this is a call-clobbered hard register that is otherwise
2662 untouched by the epilogue, but could also be a stack slot.
2664 Do not define this macro if the stack pointer is saved and restored
2665 by the regular prolog and epilog code in the call frame itself; in
2666 this case, the exception handling library routines will update the
2667 stack location to be restored in place. Otherwise, you must define
2668 this macro if you want to support call frame exception handling like
2669 that provided by DWARF 2.
2672 @defmac EH_RETURN_HANDLER_RTX
2673 A C expression whose value is RTL representing a location in which
2674 to store the address of an exception handler to which we should
2675 return. It will not be assigned on code paths that return normally.
2677 Typically this is the location in the call frame at which the normal
2678 return address is stored. For targets that return by popping an
2679 address off the stack, this might be a memory address just below
2680 the @emph{target} call frame rather than inside the current call
2681 frame. If defined, @code{EH_RETURN_STACKADJ_RTX} will have already
2682 been assigned, so it may be used to calculate the location of the
2685 Some targets have more complex requirements than storing to an
2686 address calculable during initial code generation. In that case
2687 the @code{eh_return} instruction pattern should be used instead.
2689 If you want to support call frame exception handling, you must
2690 define either this macro or the @code{eh_return} instruction pattern.
2693 @defmac RETURN_ADDR_OFFSET
2694 If defined, an integer-valued C expression for which rtl will be generated
2695 to add it to the exception handler address before it is searched in the
2696 exception handling tables, and to subtract it again from the address before
2697 using it to return to the exception handler.
2700 @defmac ASM_PREFERRED_EH_DATA_FORMAT (@var{code}, @var{global})
2701 This macro chooses the encoding of pointers embedded in the exception
2702 handling sections. If at all possible, this should be defined such
2703 that the exception handling section will not require dynamic relocations,
2704 and so may be read-only.
2706 @var{code} is 0 for data, 1 for code labels, 2 for function pointers.
2707 @var{global} is true if the symbol may be affected by dynamic relocations.
2708 The macro should return a combination of the @code{DW_EH_PE_*} defines
2709 as found in @file{dwarf2.h}.
2711 If this macro is not defined, pointers will not be encoded but
2712 represented directly.
2715 @defmac ASM_MAYBE_OUTPUT_ENCODED_ADDR_RTX (@var{file}, @var{encoding}, @var{size}, @var{addr}, @var{done})
2716 This macro allows the target to emit whatever special magic is required
2717 to represent the encoding chosen by @code{ASM_PREFERRED_EH_DATA_FORMAT}.
2718 Generic code takes care of pc-relative and indirect encodings; this must
2719 be defined if the target uses text-relative or data-relative encodings.
2721 This is a C statement that branches to @var{done} if the format was
2722 handled. @var{encoding} is the format chosen, @var{size} is the number
2723 of bytes that the format occupies, @var{addr} is the @code{SYMBOL_REF}
2727 @defmac MD_FALLBACK_FRAME_STATE_FOR (@var{context}, @var{fs})
2728 This macro allows the target to add CPU and operating system specific
2729 code to the call-frame unwinder for use when there is no unwind data
2730 available. The most common reason to implement this macro is to unwind
2731 through signal frames.
2733 This macro is called from @code{uw_frame_state_for} in
2734 @file{unwind-dw2.c}, @file{unwind-dw2-xtensa.c} and
2735 @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context};
2736 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{context->ra}
2737 for the address of the code being executed and @code{context->cfa} for
2738 the stack pointer value. If the frame can be decoded, the register
2739 save addresses should be updated in @var{fs} and the macro should
2740 evaluate to @code{_URC_NO_REASON}. If the frame cannot be decoded,
2741 the macro should evaluate to @code{_URC_END_OF_STACK}.
2743 For proper signal handling in Java this macro is accompanied by
2744 @code{MAKE_THROW_FRAME}, defined in @file{libjava/include/*-signal.h} headers.
2747 @defmac MD_HANDLE_UNWABI (@var{context}, @var{fs})
2748 This macro allows the target to add operating system specific code to the
2749 call-frame unwinder to handle the IA-64 @code{.unwabi} unwinding directive,
2750 usually used for signal or interrupt frames.
2752 This macro is called from @code{uw_update_context} in libgcc's
2753 @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context};
2754 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{fs->unwabi}
2755 for the abi and context in the @code{.unwabi} directive. If the
2756 @code{.unwabi} directive can be handled, the register save addresses should
2757 be updated in @var{fs}.
2760 @defmac TARGET_USES_WEAK_UNWIND_INFO
2761 A C expression that evaluates to true if the target requires unwind
2762 info to be given comdat linkage. Define it to be @code{1} if comdat
2763 linkage is necessary. The default is @code{0}.
2766 @node Stack Checking
2767 @subsection Specifying How Stack Checking is Done
2769 GCC will check that stack references are within the boundaries of the
2770 stack, if the option @option{-fstack-check} is specified, in one of
2775 If the value of the @code{STACK_CHECK_BUILTIN} macro is nonzero, GCC
2776 will assume that you have arranged for full stack checking to be done
2777 at appropriate places in the configuration files. GCC will not do
2778 other special processing.
2781 If @code{STACK_CHECK_BUILTIN} is zero and the value of the
2782 @code{STACK_CHECK_STATIC_BUILTIN} macro is nonzero, GCC will assume
2783 that you have arranged for static stack checking (checking of the
2784 static stack frame of functions) to be done at appropriate places
2785 in the configuration files. GCC will only emit code to do dynamic
2786 stack checking (checking on dynamic stack allocations) using the third
2790 If neither of the above are true, GCC will generate code to periodically
2791 ``probe'' the stack pointer using the values of the macros defined below.
2794 If neither STACK_CHECK_BUILTIN nor STACK_CHECK_STATIC_BUILTIN is defined,
2795 GCC will change its allocation strategy for large objects if the option
2796 @option{-fstack-check} is specified: they will always be allocated
2797 dynamically if their size exceeds @code{STACK_CHECK_MAX_VAR_SIZE} bytes.
2799 @defmac STACK_CHECK_BUILTIN
2800 A nonzero value if stack checking is done by the configuration files in a
2801 machine-dependent manner. You should define this macro if stack checking
2802 is required by the ABI of your machine or if you would like to do stack
2803 checking in some more efficient way than the generic approach. The default
2804 value of this macro is zero.
2807 @defmac STACK_CHECK_STATIC_BUILTIN
2808 A nonzero value if static stack checking is done by the configuration files
2809 in a machine-dependent manner. You should define this macro if you would
2810 like to do static stack checking in some more efficient way than the generic
2811 approach. The default value of this macro is zero.
2814 @defmac STACK_CHECK_PROBE_INTERVAL_EXP
2815 An integer specifying the interval at which GCC must generate stack probe
2816 instructions, defined as 2 raised to this integer. You will normally
2817 define this macro so that the interval be no larger than the size of
2818 the ``guard pages'' at the end of a stack area. The default value
2819 of 12 (4096-byte interval) is suitable for most systems.
2822 @defmac STACK_CHECK_MOVING_SP
2823 An integer which is nonzero if GCC should move the stack pointer page by page
2824 when doing probes. This can be necessary on systems where the stack pointer
2825 contains the bottom address of the memory area accessible to the executing
2826 thread at any point in time. In this situation an alternate signal stack
2827 is required in order to be able to recover from a stack overflow. The
2828 default value of this macro is zero.
2831 @defmac STACK_CHECK_PROTECT
2832 The number of bytes of stack needed to recover from a stack overflow, for
2833 languages where such a recovery is supported. The default value of 4KB/8KB
2834 with the @code{setjmp}/@code{longjmp}-based exception handling mechanism and
2835 8KB/12KB with other exception handling mechanisms should be adequate for most
2836 architectures and operating systems.
2839 The following macros are relevant only if neither STACK_CHECK_BUILTIN
2840 nor STACK_CHECK_STATIC_BUILTIN is defined; you can omit them altogether
2841 in the opposite case.
2843 @defmac STACK_CHECK_MAX_FRAME_SIZE
2844 The maximum size of a stack frame, in bytes. GCC will generate probe
2845 instructions in non-leaf functions to ensure at least this many bytes of
2846 stack are available. If a stack frame is larger than this size, stack
2847 checking will not be reliable and GCC will issue a warning. The
2848 default is chosen so that GCC only generates one instruction on most
2849 systems. You should normally not change the default value of this macro.
2852 @defmac STACK_CHECK_FIXED_FRAME_SIZE
2853 GCC uses this value to generate the above warning message. It
2854 represents the amount of fixed frame used by a function, not including
2855 space for any callee-saved registers, temporaries and user variables.
2856 You need only specify an upper bound for this amount and will normally
2857 use the default of four words.
2860 @defmac STACK_CHECK_MAX_VAR_SIZE
2861 The maximum size, in bytes, of an object that GCC will place in the
2862 fixed area of the stack frame when the user specifies
2863 @option{-fstack-check}.
2864 GCC computed the default from the values of the above macros and you will
2865 normally not need to override that default.
2868 @hook TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE
2871 @node Frame Registers
2872 @subsection Registers That Address the Stack Frame
2874 @c prevent bad page break with this line
2875 This discusses registers that address the stack frame.
2877 @defmac STACK_POINTER_REGNUM
2878 The register number of the stack pointer register, which must also be a
2879 fixed register according to @code{FIXED_REGISTERS}. On most machines,
2880 the hardware determines which register this is.
2883 @defmac FRAME_POINTER_REGNUM
2884 The register number of the frame pointer register, which is used to
2885 access automatic variables in the stack frame. On some machines, the
2886 hardware determines which register this is. On other machines, you can
2887 choose any register you wish for this purpose.
2890 @defmac HARD_FRAME_POINTER_REGNUM
2891 On some machines the offset between the frame pointer and starting
2892 offset of the automatic variables is not known until after register
2893 allocation has been done (for example, because the saved registers are
2894 between these two locations). On those machines, define
2895 @code{FRAME_POINTER_REGNUM} the number of a special, fixed register to
2896 be used internally until the offset is known, and define
2897 @code{HARD_FRAME_POINTER_REGNUM} to be the actual hard register number
2898 used for the frame pointer.
2900 You should define this macro only in the very rare circumstances when it
2901 is not possible to calculate the offset between the frame pointer and
2902 the automatic variables until after register allocation has been
2903 completed. When this macro is defined, you must also indicate in your
2904 definition of @code{ELIMINABLE_REGS} how to eliminate
2905 @code{FRAME_POINTER_REGNUM} into either @code{HARD_FRAME_POINTER_REGNUM}
2906 or @code{STACK_POINTER_REGNUM}.
2908 Do not define this macro if it would be the same as
2909 @code{FRAME_POINTER_REGNUM}.
2912 @defmac ARG_POINTER_REGNUM
2913 The register number of the arg pointer register, which is used to access
2914 the function's argument list. On some machines, this is the same as the
2915 frame pointer register. On some machines, the hardware determines which
2916 register this is. On other machines, you can choose any register you
2917 wish for this purpose. If this is not the same register as the frame
2918 pointer register, then you must mark it as a fixed register according to
2919 @code{FIXED_REGISTERS}, or arrange to be able to eliminate it
2920 (@pxref{Elimination}).
2923 @defmac HARD_FRAME_POINTER_IS_FRAME_POINTER
2924 Define this to a preprocessor constant that is nonzero if
2925 @code{hard_frame_pointer_rtx} and @code{frame_pointer_rtx} should be
2926 the same. The default definition is @samp{(HARD_FRAME_POINTER_REGNUM
2927 == FRAME_POINTER_REGNUM)}; you only need to define this macro if that
2928 definition is not suitable for use in preprocessor conditionals.
2931 @defmac HARD_FRAME_POINTER_IS_ARG_POINTER
2932 Define this to a preprocessor constant that is nonzero if
2933 @code{hard_frame_pointer_rtx} and @code{arg_pointer_rtx} should be the
2934 same. The default definition is @samp{(HARD_FRAME_POINTER_REGNUM ==
2935 ARG_POINTER_REGNUM)}; you only need to define this macro if that
2936 definition is not suitable for use in preprocessor conditionals.
2939 @defmac RETURN_ADDRESS_POINTER_REGNUM
2940 The register number of the return address pointer register, which is used to
2941 access the current function's return address from the stack. On some
2942 machines, the return address is not at a fixed offset from the frame
2943 pointer or stack pointer or argument pointer. This register can be defined
2944 to point to the return address on the stack, and then be converted by
2945 @code{ELIMINABLE_REGS} into either the frame pointer or stack pointer.
2947 Do not define this macro unless there is no other way to get the return
2948 address from the stack.
2951 @defmac STATIC_CHAIN_REGNUM
2952 @defmacx STATIC_CHAIN_INCOMING_REGNUM
2953 Register numbers used for passing a function's static chain pointer. If
2954 register windows are used, the register number as seen by the called
2955 function is @code{STATIC_CHAIN_INCOMING_REGNUM}, while the register
2956 number as seen by the calling function is @code{STATIC_CHAIN_REGNUM}. If
2957 these registers are the same, @code{STATIC_CHAIN_INCOMING_REGNUM} need
2960 The static chain register need not be a fixed register.
2962 If the static chain is passed in memory, these macros should not be
2963 defined; instead, the @code{TARGET_STATIC_CHAIN} hook should be used.
2966 @hook TARGET_STATIC_CHAIN
2968 @defmac DWARF_FRAME_REGISTERS
2969 This macro specifies the maximum number of hard registers that can be
2970 saved in a call frame. This is used to size data structures used in
2971 DWARF2 exception handling.
2973 Prior to GCC 3.0, this macro was needed in order to establish a stable
2974 exception handling ABI in the face of adding new hard registers for ISA
2975 extensions. In GCC 3.0 and later, the EH ABI is insulated from changes
2976 in the number of hard registers. Nevertheless, this macro can still be
2977 used to reduce the runtime memory requirements of the exception handling
2978 routines, which can be substantial if the ISA contains a lot of
2979 registers that are not call-saved.
2981 If this macro is not defined, it defaults to
2982 @code{FIRST_PSEUDO_REGISTER}.
2985 @defmac PRE_GCC3_DWARF_FRAME_REGISTERS
2987 This macro is similar to @code{DWARF_FRAME_REGISTERS}, but is provided
2988 for backward compatibility in pre GCC 3.0 compiled code.
2990 If this macro is not defined, it defaults to
2991 @code{DWARF_FRAME_REGISTERS}.
2994 @defmac DWARF_REG_TO_UNWIND_COLUMN (@var{regno})
2996 Define this macro if the target's representation for dwarf registers
2997 is different than the internal representation for unwind column.
2998 Given a dwarf register, this macro should return the internal unwind
2999 column number to use instead.
3002 @defmac DWARF_FRAME_REGNUM (@var{regno})
3004 Define this macro if the target's representation for dwarf registers
3005 used in .eh_frame or .debug_frame is different from that used in other
3006 debug info sections. Given a GCC hard register number, this macro
3007 should return the .eh_frame register number. The default is
3008 @code{DBX_REGISTER_NUMBER (@var{regno})}.
3012 @defmac DWARF2_FRAME_REG_OUT (@var{regno}, @var{for_eh})
3014 Define this macro to map register numbers held in the call frame info
3015 that GCC has collected using @code{DWARF_FRAME_REGNUM} to those that
3016 should be output in .debug_frame (@code{@var{for_eh}} is zero) and
3017 .eh_frame (@code{@var{for_eh}} is nonzero). The default is to
3018 return @code{@var{regno}}.
3022 @defmac REG_VALUE_IN_UNWIND_CONTEXT
3024 Define this macro if the target stores register values as
3025 @code{_Unwind_Word} type in unwind context. It should be defined if
3026 target register size is larger than the size of @code{void *}. The
3027 default is to store register values as @code{void *} type.
3031 @defmac ASSUME_EXTENDED_UNWIND_CONTEXT
3033 Define this macro to be 1 if the target always uses extended unwind
3034 context with version, args_size and by_value fields. If it is undefined,
3035 it will be defined to 1 when @code{REG_VALUE_IN_UNWIND_CONTEXT} is
3036 defined and 0 otherwise.
3040 @defmac DWARF_LAZY_REGISTER_VALUE (@var{regno}, @var{value})
3041 Define this macro if the target has pseudo DWARF registers whose
3042 values need to be computed lazily on demand by the unwinder (such as when
3043 referenced in a CFA expression). The macro returns true if @var{regno}
3044 is such a register and stores its value in @samp{*@var{value}} if so.
3048 @subsection Eliminating Frame Pointer and Arg Pointer
3050 @c prevent bad page break with this line
3051 This is about eliminating the frame pointer and arg pointer.
3053 @hook TARGET_FRAME_POINTER_REQUIRED
3055 @defmac ELIMINABLE_REGS
3056 This macro specifies a table of register pairs used to eliminate
3057 unneeded registers that point into the stack frame.
3059 The definition of this macro is a list of structure initializations, each
3060 of which specifies an original and replacement register.
3062 On some machines, the position of the argument pointer is not known until
3063 the compilation is completed. In such a case, a separate hard register
3064 must be used for the argument pointer. This register can be eliminated by
3065 replacing it with either the frame pointer or the argument pointer,
3066 depending on whether or not the frame pointer has been eliminated.
3068 In this case, you might specify:
3070 #define ELIMINABLE_REGS \
3071 @{@{ARG_POINTER_REGNUM, STACK_POINTER_REGNUM@}, \
3072 @{ARG_POINTER_REGNUM, FRAME_POINTER_REGNUM@}, \
3073 @{FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM@}@}
3076 Note that the elimination of the argument pointer with the stack pointer is
3077 specified first since that is the preferred elimination.
3080 @hook TARGET_CAN_ELIMINATE
3082 @defmac INITIAL_ELIMINATION_OFFSET (@var{from-reg}, @var{to-reg}, @var{offset-var})
3083 This macro returns the initial difference between the specified pair
3084 of registers. The value would be computed from information
3085 such as the result of @code{get_frame_size ()} and the tables of
3086 registers @code{df_regs_ever_live_p} and @code{call_used_regs}.
3089 @hook TARGET_COMPUTE_FRAME_LAYOUT
3091 @node Stack Arguments
3092 @subsection Passing Function Arguments on the Stack
3093 @cindex arguments on stack
3094 @cindex stack arguments
3096 The macros in this section control how arguments are passed
3097 on the stack. See the following section for other macros that
3098 control passing certain arguments in registers.
3100 @hook TARGET_PROMOTE_PROTOTYPES
3103 A C expression. If nonzero, push insns will be used to pass
3105 If the target machine does not have a push instruction, set it to zero.
3106 That directs GCC to use an alternate strategy: to
3107 allocate the entire argument block and then store the arguments into
3108 it. When @code{PUSH_ARGS} is nonzero, @code{PUSH_ROUNDING} must be defined too.
3111 @defmac PUSH_ARGS_REVERSED
3112 A C expression. If nonzero, function arguments will be evaluated from
3113 last to first, rather than from first to last. If this macro is not
3114 defined, it defaults to @code{PUSH_ARGS} on targets where the stack
3115 and args grow in opposite directions, and 0 otherwise.
3118 @defmac PUSH_ROUNDING (@var{npushed})
3119 A C expression that is the number of bytes actually pushed onto the
3120 stack when an instruction attempts to push @var{npushed} bytes.
3122 On some machines, the definition
3125 #define PUSH_ROUNDING(BYTES) (BYTES)
3129 will suffice. But on other machines, instructions that appear
3130 to push one byte actually push two bytes in an attempt to maintain
3131 alignment. Then the definition should be
3134 #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1)
3137 If the value of this macro has a type, it should be an unsigned type.
3140 @findex outgoing_args_size
3141 @findex crtl->outgoing_args_size
3142 @defmac ACCUMULATE_OUTGOING_ARGS
3143 A C expression. If nonzero, the maximum amount of space required for outgoing arguments
3144 will be computed and placed into
3145 @code{crtl->outgoing_args_size}. No space will be pushed
3146 onto the stack for each call; instead, the function prologue should
3147 increase the stack frame size by this amount.
3149 Setting both @code{PUSH_ARGS} and @code{ACCUMULATE_OUTGOING_ARGS}
3153 @defmac REG_PARM_STACK_SPACE (@var{fndecl})
3154 Define this macro if functions should assume that stack space has been
3155 allocated for arguments even when their values are passed in
3158 The value of this macro is the size, in bytes, of the area reserved for
3159 arguments passed in registers for the function represented by @var{fndecl},
3160 which can be zero if GCC is calling a library function.
3161 The argument @var{fndecl} can be the FUNCTION_DECL, or the type itself
3164 This space can be allocated by the caller, or be a part of the
3165 machine-dependent stack frame: @code{OUTGOING_REG_PARM_STACK_SPACE} says
3168 @c above is overfull. not sure what to do. --mew 5feb93 did
3169 @c something, not sure if it looks good. --mew 10feb93
3171 @defmac INCOMING_REG_PARM_STACK_SPACE (@var{fndecl})
3172 Like @code{REG_PARM_STACK_SPACE}, but for incoming register arguments.
3173 Define this macro if space guaranteed when compiling a function body
3174 is different to space required when making a call, a situation that
3175 can arise with K&R style function definitions.
3178 @defmac OUTGOING_REG_PARM_STACK_SPACE (@var{fntype})
3179 Define this to a nonzero value if it is the responsibility of the
3180 caller to allocate the area reserved for arguments passed in registers
3181 when calling a function of @var{fntype}. @var{fntype} may be NULL
3182 if the function called is a library function.
3184 If @code{ACCUMULATE_OUTGOING_ARGS} is defined, this macro controls
3185 whether the space for these arguments counts in the value of
3186 @code{crtl->outgoing_args_size}.
3189 @defmac STACK_PARMS_IN_REG_PARM_AREA
3190 Define this macro if @code{REG_PARM_STACK_SPACE} is defined, but the
3191 stack parameters don't skip the area specified by it.
3192 @c i changed this, makes more sens and it should have taken care of the
3193 @c overfull.. not as specific, tho. --mew 5feb93
3195 Normally, when a parameter is not passed in registers, it is placed on the
3196 stack beyond the @code{REG_PARM_STACK_SPACE} area. Defining this macro
3197 suppresses this behavior and causes the parameter to be passed on the
3198 stack in its natural location.
3201 @hook TARGET_RETURN_POPS_ARGS
3203 @defmac CALL_POPS_ARGS (@var{cum})
3204 A C expression that should indicate the number of bytes a call sequence
3205 pops off the stack. It is added to the value of @code{RETURN_POPS_ARGS}
3206 when compiling a function call.
3208 @var{cum} is the variable in which all arguments to the called function
3209 have been accumulated.
3211 On certain architectures, such as the SH5, a call trampoline is used
3212 that pops certain registers off the stack, depending on the arguments
3213 that have been passed to the function. Since this is a property of the
3214 call site, not of the called function, @code{RETURN_POPS_ARGS} is not
3218 @node Register Arguments
3219 @subsection Passing Arguments in Registers
3220 @cindex arguments in registers
3221 @cindex registers arguments
3223 This section describes the macros which let you control how various
3224 types of arguments are passed in registers or how they are arranged in
3227 @hook TARGET_FUNCTION_ARG
3229 @hook TARGET_MUST_PASS_IN_STACK
3231 @hook TARGET_FUNCTION_INCOMING_ARG
3233 @hook TARGET_USE_PSEUDO_PIC_REG
3235 @hook TARGET_INIT_PIC_REG
3237 @hook TARGET_ARG_PARTIAL_BYTES
3239 @hook TARGET_PASS_BY_REFERENCE
3241 @hook TARGET_CALLEE_COPIES
3243 @defmac CUMULATIVE_ARGS
3244 A C type for declaring a variable that is used as the first argument
3245 of @code{TARGET_FUNCTION_ARG} and other related values. For some
3246 target machines, the type @code{int} suffices and can hold the number
3247 of bytes of argument so far.
3249 There is no need to record in @code{CUMULATIVE_ARGS} anything about the
3250 arguments that have been passed on the stack. The compiler has other
3251 variables to keep track of that. For target machines on which all
3252 arguments are passed on the stack, there is no need to store anything in
3253 @code{CUMULATIVE_ARGS}; however, the data structure must exist and
3254 should not be empty, so use @code{int}.
3257 @defmac OVERRIDE_ABI_FORMAT (@var{fndecl})
3258 If defined, this macro is called before generating any code for a
3259 function, but after the @var{cfun} descriptor for the function has been
3260 created. The back end may use this macro to update @var{cfun} to
3261 reflect an ABI other than that which would normally be used by default.
3262 If the compiler is generating code for a compiler-generated function,
3263 @var{fndecl} may be @code{NULL}.
3266 @defmac INIT_CUMULATIVE_ARGS (@var{cum}, @var{fntype}, @var{libname}, @var{fndecl}, @var{n_named_args})
3267 A C statement (sans semicolon) for initializing the variable
3268 @var{cum} for the state at the beginning of the argument list. The
3269 variable has type @code{CUMULATIVE_ARGS}. The value of @var{fntype}
3270 is the tree node for the data type of the function which will receive
3271 the args, or 0 if the args are to a compiler support library function.
3272 For direct calls that are not libcalls, @var{fndecl} contain the
3273 declaration node of the function. @var{fndecl} is also set when
3274 @code{INIT_CUMULATIVE_ARGS} is used to find arguments for the function
3275 being compiled. @var{n_named_args} is set to the number of named
3276 arguments, including a structure return address if it is passed as a
3277 parameter, when making a call. When processing incoming arguments,
3278 @var{n_named_args} is set to @minus{}1.
3280 When processing a call to a compiler support library function,
3281 @var{libname} identifies which one. It is a @code{symbol_ref} rtx which
3282 contains the name of the function, as a string. @var{libname} is 0 when
3283 an ordinary C function call is being processed. Thus, each time this
3284 macro is called, either @var{libname} or @var{fntype} is nonzero, but
3285 never both of them at once.
3288 @defmac INIT_CUMULATIVE_LIBCALL_ARGS (@var{cum}, @var{mode}, @var{libname})
3289 Like @code{INIT_CUMULATIVE_ARGS} but only used for outgoing libcalls,
3290 it gets a @code{MODE} argument instead of @var{fntype}, that would be
3291 @code{NULL}. @var{indirect} would always be zero, too. If this macro
3292 is not defined, @code{INIT_CUMULATIVE_ARGS (cum, NULL_RTX, libname,
3293 0)} is used instead.
3296 @defmac INIT_CUMULATIVE_INCOMING_ARGS (@var{cum}, @var{fntype}, @var{libname})
3297 Like @code{INIT_CUMULATIVE_ARGS} but overrides it for the purposes of
3298 finding the arguments for the function being compiled. If this macro is
3299 undefined, @code{INIT_CUMULATIVE_ARGS} is used instead.
3301 The value passed for @var{libname} is always 0, since library routines
3302 with special calling conventions are never compiled with GCC@. The
3303 argument @var{libname} exists for symmetry with
3304 @code{INIT_CUMULATIVE_ARGS}.
3305 @c could use "this macro" in place of @code{INIT_CUMULATIVE_ARGS}, maybe.
3306 @c --mew 5feb93 i switched the order of the sentences. --mew 10feb93
3309 @hook TARGET_FUNCTION_ARG_ADVANCE
3311 @hook TARGET_FUNCTION_ARG_OFFSET
3313 @hook TARGET_FUNCTION_ARG_PADDING
3315 @defmac PAD_VARARGS_DOWN
3316 If defined, a C expression which determines whether the default
3317 implementation of va_arg will attempt to pad down before reading the
3318 next argument, if that argument is smaller than its aligned space as
3319 controlled by @code{PARM_BOUNDARY}. If this macro is not defined, all such
3320 arguments are padded down if @code{BYTES_BIG_ENDIAN} is true.
3323 @defmac BLOCK_REG_PADDING (@var{mode}, @var{type}, @var{first})
3324 Specify padding for the last element of a block move between registers and
3325 memory. @var{first} is nonzero if this is the only element. Defining this
3326 macro allows better control of register function parameters on big-endian
3327 machines, without using @code{PARALLEL} rtl. In particular,
3328 @code{MUST_PASS_IN_STACK} need not test padding and mode of types in
3329 registers, as there is no longer a "wrong" part of a register; For example,
3330 a three byte aggregate may be passed in the high part of a register if so
3334 @hook TARGET_FUNCTION_ARG_BOUNDARY
3336 @hook TARGET_FUNCTION_ARG_ROUND_BOUNDARY
3338 @defmac FUNCTION_ARG_REGNO_P (@var{regno})
3339 A C expression that is nonzero if @var{regno} is the number of a hard
3340 register in which function arguments are sometimes passed. This does
3341 @emph{not} include implicit arguments such as the static chain and
3342 the structure-value address. On many machines, no registers can be
3343 used for this purpose since all function arguments are pushed on the
3347 @hook TARGET_SPLIT_COMPLEX_ARG
3349 @hook TARGET_BUILD_BUILTIN_VA_LIST
3351 @hook TARGET_ENUM_VA_LIST_P
3353 @hook TARGET_FN_ABI_VA_LIST
3355 @hook TARGET_CANONICAL_VA_LIST_TYPE
3357 @hook TARGET_GIMPLIFY_VA_ARG_EXPR
3359 @hook TARGET_VALID_POINTER_MODE
3361 @hook TARGET_REF_MAY_ALIAS_ERRNO
3363 @hook TARGET_TRANSLATE_MODE_ATTRIBUTE
3365 @hook TARGET_SCALAR_MODE_SUPPORTED_P
3367 @hook TARGET_VECTOR_MODE_SUPPORTED_P
3369 @hook TARGET_ARRAY_MODE
3371 @hook TARGET_ARRAY_MODE_SUPPORTED_P
3373 @hook TARGET_LIBGCC_FLOATING_MODE_SUPPORTED_P
3375 @hook TARGET_FLOATN_MODE
3377 @hook TARGET_FLOATN_BUILTIN_P
3379 @hook TARGET_SMALL_REGISTER_CLASSES_FOR_MODE_P
3382 @subsection How Scalar Function Values Are Returned
3383 @cindex return values in registers
3384 @cindex values, returned by functions
3385 @cindex scalars, returned as values
3387 This section discusses the macros that control returning scalars as
3388 values---values that can fit in registers.
3390 @hook TARGET_FUNCTION_VALUE
3392 @defmac FUNCTION_VALUE (@var{valtype}, @var{func})
3393 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE} for
3394 a new target instead.
3397 @defmac LIBCALL_VALUE (@var{mode})
3398 A C expression to create an RTX representing the place where a library
3399 function returns a value of mode @var{mode}.
3401 Note that ``library function'' in this context means a compiler
3402 support routine, used to perform arithmetic, whose name is known
3403 specially by the compiler and was not mentioned in the C code being
3407 @hook TARGET_LIBCALL_VALUE
3409 @defmac FUNCTION_VALUE_REGNO_P (@var{regno})
3410 A C expression that is nonzero if @var{regno} is the number of a hard
3411 register in which the values of called function may come back.
3413 A register whose use for returning values is limited to serving as the
3414 second of a pair (for a value of type @code{double}, say) need not be
3415 recognized by this macro. So for most machines, this definition
3419 #define FUNCTION_VALUE_REGNO_P(N) ((N) == 0)
3422 If the machine has register windows, so that the caller and the called
3423 function use different registers for the return value, this macro
3424 should recognize only the caller's register numbers.
3426 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE_REGNO_P}
3427 for a new target instead.
3430 @hook TARGET_FUNCTION_VALUE_REGNO_P
3432 @defmac APPLY_RESULT_SIZE
3433 Define this macro if @samp{untyped_call} and @samp{untyped_return}
3434 need more space than is implied by @code{FUNCTION_VALUE_REGNO_P} for
3435 saving and restoring an arbitrary return value.
3438 @hook TARGET_OMIT_STRUCT_RETURN_REG
3440 @hook TARGET_RETURN_IN_MSB
3442 @node Aggregate Return
3443 @subsection How Large Values Are Returned
3444 @cindex aggregates as return values
3445 @cindex large return values
3446 @cindex returning aggregate values
3447 @cindex structure value address
3449 When a function value's mode is @code{BLKmode} (and in some other
3450 cases), the value is not returned according to
3451 @code{TARGET_FUNCTION_VALUE} (@pxref{Scalar Return}). Instead, the
3452 caller passes the address of a block of memory in which the value
3453 should be stored. This address is called the @dfn{structure value
3456 This section describes how to control returning structure values in
3459 @hook TARGET_RETURN_IN_MEMORY
3461 @defmac DEFAULT_PCC_STRUCT_RETURN
3462 Define this macro to be 1 if all structure and union return values must be
3463 in memory. Since this results in slower code, this should be defined
3464 only if needed for compatibility with other compilers or with an ABI@.
3465 If you define this macro to be 0, then the conventions used for structure
3466 and union return values are decided by the @code{TARGET_RETURN_IN_MEMORY}
3469 If not defined, this defaults to the value 1.
3472 @hook TARGET_STRUCT_VALUE_RTX
3474 @defmac PCC_STATIC_STRUCT_RETURN
3475 Define this macro if the usual system convention on the target machine
3476 for returning structures and unions is for the called function to return
3477 the address of a static variable containing the value.
3479 Do not define this if the usual system convention is for the caller to
3480 pass an address to the subroutine.
3482 This macro has effect in @option{-fpcc-struct-return} mode, but it does
3483 nothing when you use @option{-freg-struct-return} mode.
3486 @hook TARGET_GET_RAW_RESULT_MODE
3488 @hook TARGET_GET_RAW_ARG_MODE
3490 @hook TARGET_EMPTY_RECORD_P
3492 @hook TARGET_WARN_PARAMETER_PASSING_ABI
3495 @subsection Caller-Saves Register Allocation
3497 If you enable it, GCC can save registers around function calls. This
3498 makes it possible to use call-clobbered registers to hold variables that
3499 must live across calls.
3501 @defmac HARD_REGNO_CALLER_SAVE_MODE (@var{regno}, @var{nregs})
3502 A C expression specifying which mode is required for saving @var{nregs}
3503 of a pseudo-register in call-clobbered hard register @var{regno}. If
3504 @var{regno} is unsuitable for caller save, @code{VOIDmode} should be
3505 returned. For most machines this macro need not be defined since GCC
3506 will select the smallest suitable mode.
3509 @node Function Entry
3510 @subsection Function Entry and Exit
3511 @cindex function entry and exit
3515 This section describes the macros that output function entry
3516 (@dfn{prologue}) and exit (@dfn{epilogue}) code.
3518 @hook TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY
3520 @hook TARGET_ASM_FUNCTION_PROLOGUE
3522 @hook TARGET_ASM_FUNCTION_END_PROLOGUE
3524 @hook TARGET_ASM_FUNCTION_BEGIN_EPILOGUE
3526 @hook TARGET_ASM_FUNCTION_EPILOGUE
3530 @findex pretend_args_size
3531 @findex crtl->args.pretend_args_size
3532 A region of @code{crtl->args.pretend_args_size} bytes of
3533 uninitialized space just underneath the first argument arriving on the
3534 stack. (This may not be at the very start of the allocated stack region
3535 if the calling sequence has pushed anything else since pushing the stack
3536 arguments. But usually, on such machines, nothing else has been pushed
3537 yet, because the function prologue itself does all the pushing.) This
3538 region is used on machines where an argument may be passed partly in
3539 registers and partly in memory, and, in some cases to support the
3540 features in @code{<stdarg.h>}.
3543 An area of memory used to save certain registers used by the function.
3544 The size of this area, which may also include space for such things as
3545 the return address and pointers to previous stack frames, is
3546 machine-specific and usually depends on which registers have been used
3547 in the function. Machines with register windows often do not require
3551 A region of at least @var{size} bytes, possibly rounded up to an allocation
3552 boundary, to contain the local variables of the function. On some machines,
3553 this region and the save area may occur in the opposite order, with the
3554 save area closer to the top of the stack.
3557 @cindex @code{ACCUMULATE_OUTGOING_ARGS} and stack frames
3558 Optionally, when @code{ACCUMULATE_OUTGOING_ARGS} is defined, a region of
3559 @code{crtl->outgoing_args_size} bytes to be used for outgoing
3560 argument lists of the function. @xref{Stack Arguments}.
3563 @defmac EXIT_IGNORE_STACK
3564 Define this macro as a C expression that is nonzero if the return
3565 instruction or the function epilogue ignores the value of the stack
3566 pointer; in other words, if it is safe to delete an instruction to
3567 adjust the stack pointer before a return from the function. The
3570 Note that this macro's value is relevant only for functions for which
3571 frame pointers are maintained. It is never safe to delete a final
3572 stack adjustment in a function that has no frame pointer, and the
3573 compiler knows this regardless of @code{EXIT_IGNORE_STACK}.
3576 @defmac EPILOGUE_USES (@var{regno})
3577 Define this macro as a C expression that is nonzero for registers that are
3578 used by the epilogue or the @samp{return} pattern. The stack and frame
3579 pointer registers are already assumed to be used as needed.
3582 @defmac EH_USES (@var{regno})
3583 Define this macro as a C expression that is nonzero for registers that are
3584 used by the exception handling mechanism, and so should be considered live
3585 on entry to an exception edge.
3588 @hook TARGET_ASM_OUTPUT_MI_THUNK
3590 @hook TARGET_ASM_CAN_OUTPUT_MI_THUNK
3593 @subsection Generating Code for Profiling
3594 @cindex profiling, code generation
3596 These macros will help you generate code for profiling.
3598 @defmac FUNCTION_PROFILER (@var{file}, @var{labelno})
3599 A C statement or compound statement to output to @var{file} some
3600 assembler code to call the profiling subroutine @code{mcount}.
3603 The details of how @code{mcount} expects to be called are determined by
3604 your operating system environment, not by GCC@. To figure them out,
3605 compile a small program for profiling using the system's installed C
3606 compiler and look at the assembler code that results.
3608 Older implementations of @code{mcount} expect the address of a counter
3609 variable to be loaded into some register. The name of this variable is
3610 @samp{LP} followed by the number @var{labelno}, so you would generate
3611 the name using @samp{LP%d} in a @code{fprintf}.
3614 @defmac PROFILE_HOOK
3615 A C statement or compound statement to output to @var{file} some assembly
3616 code to call the profiling subroutine @code{mcount} even the target does
3617 not support profiling.
3620 @defmac NO_PROFILE_COUNTERS
3621 Define this macro to be an expression with a nonzero value if the
3622 @code{mcount} subroutine on your system does not need a counter variable
3623 allocated for each function. This is true for almost all modern
3624 implementations. If you define this macro, you must not use the
3625 @var{labelno} argument to @code{FUNCTION_PROFILER}.
3628 @defmac PROFILE_BEFORE_PROLOGUE
3629 Define this macro if the code for function profiling should come before
3630 the function prologue. Normally, the profiling code comes after.
3633 @hook TARGET_KEEP_LEAF_WHEN_PROFILED
3636 @subsection Permitting tail calls
3639 @hook TARGET_FUNCTION_OK_FOR_SIBCALL
3641 @hook TARGET_EXTRA_LIVE_ON_ENTRY
3643 @hook TARGET_SET_UP_BY_PROLOGUE
3645 @hook TARGET_WARN_FUNC_RETURN
3647 @node Shrink-wrapping separate components
3648 @subsection Shrink-wrapping separate components
3649 @cindex shrink-wrapping separate components
3651 The prologue may perform a variety of target dependent tasks such as
3652 saving callee-saved registers, saving the return address, aligning the
3653 stack, creating a stack frame, initializing the PIC register, setting
3654 up the static chain, etc.
3656 On some targets some of these tasks may be independent of others and
3657 thus may be shrink-wrapped separately. These independent tasks are
3658 referred to as components and are handled generically by the target
3659 independent parts of GCC.
3661 Using the following hooks those prologue or epilogue components can be
3662 shrink-wrapped separately, so that the initialization (and possibly
3663 teardown) those components do is not done as frequently on execution
3664 paths where this would unnecessary.
3666 What exactly those components are is up to the target code; the generic
3667 code treats them abstractly, as a bit in an @code{sbitmap}. These
3668 @code{sbitmap}s are allocated by the @code{shrink_wrap.get_separate_components}
3669 and @code{shrink_wrap.components_for_bb} hooks, and deallocated by the
3672 @hook TARGET_SHRINK_WRAP_GET_SEPARATE_COMPONENTS
3674 @hook TARGET_SHRINK_WRAP_COMPONENTS_FOR_BB
3676 @hook TARGET_SHRINK_WRAP_DISQUALIFY_COMPONENTS
3678 @hook TARGET_SHRINK_WRAP_EMIT_PROLOGUE_COMPONENTS
3680 @hook TARGET_SHRINK_WRAP_EMIT_EPILOGUE_COMPONENTS
3682 @hook TARGET_SHRINK_WRAP_SET_HANDLED_COMPONENTS
3684 @node Stack Smashing Protection
3685 @subsection Stack smashing protection
3686 @cindex stack smashing protection
3688 @hook TARGET_STACK_PROTECT_GUARD
3690 @hook TARGET_STACK_PROTECT_FAIL
3692 @hook TARGET_STACK_PROTECT_RUNTIME_ENABLED_P
3694 @hook TARGET_SUPPORTS_SPLIT_STACK
3696 @hook TARGET_GET_VALID_OPTION_VALUES
3698 @node Miscellaneous Register Hooks
3699 @subsection Miscellaneous register hooks
3700 @cindex miscellaneous register hooks
3702 @hook TARGET_CALL_FUSAGE_CONTAINS_NON_CALLEE_CLOBBERS
3705 @section Implementing the Varargs Macros
3706 @cindex varargs implementation
3708 GCC comes with an implementation of @code{<varargs.h>} and
3709 @code{<stdarg.h>} that work without change on machines that pass arguments
3710 on the stack. Other machines require their own implementations of
3711 varargs, and the two machine independent header files must have
3712 conditionals to include it.
3714 ISO @code{<stdarg.h>} differs from traditional @code{<varargs.h>} mainly in
3715 the calling convention for @code{va_start}. The traditional
3716 implementation takes just one argument, which is the variable in which
3717 to store the argument pointer. The ISO implementation of
3718 @code{va_start} takes an additional second argument. The user is
3719 supposed to write the last named argument of the function here.
3721 However, @code{va_start} should not use this argument. The way to find
3722 the end of the named arguments is with the built-in functions described
3725 @defmac __builtin_saveregs ()
3726 Use this built-in function to save the argument registers in memory so
3727 that the varargs mechanism can access them. Both ISO and traditional
3728 versions of @code{va_start} must use @code{__builtin_saveregs}, unless
3729 you use @code{TARGET_SETUP_INCOMING_VARARGS} (see below) instead.
3731 On some machines, @code{__builtin_saveregs} is open-coded under the
3732 control of the target hook @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. On
3733 other machines, it calls a routine written in assembler language,
3734 found in @file{libgcc2.c}.
3736 Code generated for the call to @code{__builtin_saveregs} appears at the
3737 beginning of the function, as opposed to where the call to
3738 @code{__builtin_saveregs} is written, regardless of what the code is.
3739 This is because the registers must be saved before the function starts
3740 to use them for its own purposes.
3741 @c i rewrote the first sentence above to fix an overfull hbox. --mew
3745 @defmac __builtin_next_arg (@var{lastarg})
3746 This builtin returns the address of the first anonymous stack
3747 argument, as type @code{void *}. If @code{ARGS_GROW_DOWNWARD}, it
3748 returns the address of the location above the first anonymous stack
3749 argument. Use it in @code{va_start} to initialize the pointer for
3750 fetching arguments from the stack. Also use it in @code{va_start} to
3751 verify that the second parameter @var{lastarg} is the last named argument
3752 of the current function.
3755 @defmac __builtin_classify_type (@var{object})
3756 Since each machine has its own conventions for which data types are
3757 passed in which kind of register, your implementation of @code{va_arg}
3758 has to embody these conventions. The easiest way to categorize the
3759 specified data type is to use @code{__builtin_classify_type} together
3760 with @code{sizeof} and @code{__alignof__}.
3762 @code{__builtin_classify_type} ignores the value of @var{object},
3763 considering only its data type. It returns an integer describing what
3764 kind of type that is---integer, floating, pointer, structure, and so on.
3766 The file @file{typeclass.h} defines an enumeration that you can use to
3767 interpret the values of @code{__builtin_classify_type}.
3770 These machine description macros help implement varargs:
3772 @hook TARGET_EXPAND_BUILTIN_SAVEREGS
3774 @hook TARGET_SETUP_INCOMING_VARARGS
3776 @hook TARGET_STRICT_ARGUMENT_NAMING
3778 @hook TARGET_CALL_ARGS
3780 @hook TARGET_END_CALL_ARGS
3782 @hook TARGET_PRETEND_OUTGOING_VARARGS_NAMED
3784 @hook TARGET_LOAD_BOUNDS_FOR_ARG
3786 @hook TARGET_STORE_BOUNDS_FOR_ARG
3788 @hook TARGET_LOAD_RETURNED_BOUNDS
3790 @hook TARGET_STORE_RETURNED_BOUNDS
3793 @section Support for Nested Functions
3794 @cindex support for nested functions
3795 @cindex trampolines for nested functions
3796 @cindex descriptors for nested functions
3797 @cindex nested functions, support for
3799 Taking the address of a nested function requires special compiler
3800 handling to ensure that the static chain register is loaded when
3801 the function is invoked via an indirect call.
3803 GCC has traditionally supported nested functions by creating an
3804 executable @dfn{trampoline} at run time when the address of a nested
3805 function is taken. This is a small piece of code which normally
3806 resides on the stack, in the stack frame of the containing function.
3807 The trampoline loads the static chain register and then jumps to the
3808 real address of the nested function.
3810 The use of trampolines requires an executable stack, which is a
3811 security risk. To avoid this problem, GCC also supports another
3812 strategy: using descriptors for nested functions. Under this model,
3813 taking the address of a nested function results in a pointer to a
3814 non-executable function descriptor object. Initializing the static chain
3815 from the descriptor is handled at indirect call sites.
3817 On some targets, including HPPA and IA-64, function descriptors may be
3818 mandated by the ABI or be otherwise handled in a target-specific way
3819 by the back end in its code generation strategy for indirect calls.
3820 GCC also provides its own generic descriptor implementation to support the
3821 @option{-fno-trampolines} option. In this case runtime detection of
3822 function descriptors at indirect call sites relies on descriptor
3823 pointers being tagged with a bit that is never set in bare function
3824 addresses. Since GCC's generic function descriptors are
3825 not ABI-compliant, this option is typically used only on a
3826 per-language basis (notably by Ada) or when it can otherwise be
3827 applied to the whole program.
3829 Define the following hook if your backend either implements ABI-specified
3830 descriptor support, or can use GCC's generic descriptor implementation
3831 for nested functions.
3833 @hook TARGET_CUSTOM_FUNCTION_DESCRIPTORS
3835 The following macros tell GCC how to generate code to allocate and
3836 initialize an executable trampoline. You can also use this interface
3837 if your back end needs to create ABI-specified non-executable descriptors; in
3838 this case the "trampoline" created is the descriptor containing data only.
3840 The instructions in an executable trampoline must do two things: load
3841 a constant address into the static chain register, and jump to the real
3842 address of the nested function. On CISC machines such as the m68k,
3843 this requires two instructions, a move immediate and a jump. Then the
3844 two addresses exist in the trampoline as word-long immediate operands.
3845 On RISC machines, it is often necessary to load each address into a
3846 register in two parts. Then pieces of each address form separate
3849 The code generated to initialize the trampoline must store the variable
3850 parts---the static chain value and the function address---into the
3851 immediate operands of the instructions. On a CISC machine, this is
3852 simply a matter of copying each address to a memory reference at the
3853 proper offset from the start of the trampoline. On a RISC machine, it
3854 may be necessary to take out pieces of the address and store them
3857 @hook TARGET_ASM_TRAMPOLINE_TEMPLATE
3859 @defmac TRAMPOLINE_SECTION
3860 Return the section into which the trampoline template is to be placed
3861 (@pxref{Sections}). The default value is @code{readonly_data_section}.
3864 @defmac TRAMPOLINE_SIZE
3865 A C expression for the size in bytes of the trampoline, as an integer.
3868 @defmac TRAMPOLINE_ALIGNMENT
3869 Alignment required for trampolines, in bits.
3871 If you don't define this macro, the value of @code{FUNCTION_ALIGNMENT}
3872 is used for aligning trampolines.
3875 @hook TARGET_TRAMPOLINE_INIT
3877 @hook TARGET_TRAMPOLINE_ADJUST_ADDRESS
3879 Implementing trampolines is difficult on many machines because they have
3880 separate instruction and data caches. Writing into a stack location
3881 fails to clear the memory in the instruction cache, so when the program
3882 jumps to that location, it executes the old contents.
3884 Here are two possible solutions. One is to clear the relevant parts of
3885 the instruction cache whenever a trampoline is set up. The other is to
3886 make all trampolines identical, by having them jump to a standard
3887 subroutine. The former technique makes trampoline execution faster; the
3888 latter makes initialization faster.
3890 To clear the instruction cache when a trampoline is initialized, define
3891 the following macro.
3893 @defmac CLEAR_INSN_CACHE (@var{beg}, @var{end})
3894 If defined, expands to a C expression clearing the @emph{instruction
3895 cache} in the specified interval. The definition of this macro would
3896 typically be a series of @code{asm} statements. Both @var{beg} and
3897 @var{end} are both pointer expressions.
3900 To use a standard subroutine, define the following macro. In addition,
3901 you must make sure that the instructions in a trampoline fill an entire
3902 cache line with identical instructions, or else ensure that the
3903 beginning of the trampoline code is always aligned at the same point in
3904 its cache line. Look in @file{m68k.h} as a guide.
3906 @defmac TRANSFER_FROM_TRAMPOLINE
3907 Define this macro if trampolines need a special subroutine to do their
3908 work. The macro should expand to a series of @code{asm} statements
3909 which will be compiled with GCC@. They go in a library function named
3910 @code{__transfer_from_trampoline}.
3912 If you need to avoid executing the ordinary prologue code of a compiled
3913 C function when you jump to the subroutine, you can do so by placing a
3914 special label of your own in the assembler code. Use one @code{asm}
3915 statement to generate an assembler label, and another to make the label
3916 global. Then trampolines can use that label to jump directly to your
3917 special assembler code.
3921 @section Implicit Calls to Library Routines
3922 @cindex library subroutine names
3923 @cindex @file{libgcc.a}
3925 @c prevent bad page break with this line
3926 Here is an explanation of implicit calls to library routines.
3928 @defmac DECLARE_LIBRARY_RENAMES
3929 This macro, if defined, should expand to a piece of C code that will get
3930 expanded when compiling functions for libgcc.a. It can be used to
3931 provide alternate names for GCC's internal library functions if there
3932 are ABI-mandated names that the compiler should provide.
3935 @findex set_optab_libfunc
3936 @findex init_one_libfunc
3937 @hook TARGET_INIT_LIBFUNCS
3939 @hook TARGET_LIBFUNC_GNU_PREFIX
3941 @defmac FLOAT_LIB_COMPARE_RETURNS_BOOL (@var{mode}, @var{comparison})
3942 This macro should return @code{true} if the library routine that
3943 implements the floating point comparison operator @var{comparison} in
3944 mode @var{mode} will return a boolean, and @var{false} if it will
3947 GCC's own floating point libraries return tristates from the
3948 comparison operators, so the default returns false always. Most ports
3949 don't need to define this macro.
3952 @defmac TARGET_LIB_INT_CMP_BIASED
3953 This macro should evaluate to @code{true} if the integer comparison
3954 functions (like @code{__cmpdi2}) return 0 to indicate that the first
3955 operand is smaller than the second, 1 to indicate that they are equal,
3956 and 2 to indicate that the first operand is greater than the second.
3957 If this macro evaluates to @code{false} the comparison functions return
3958 @minus{}1, 0, and 1 instead of 0, 1, and 2. If the target uses the routines
3959 in @file{libgcc.a}, you do not need to define this macro.
3962 @defmac TARGET_HAS_NO_HW_DIVIDE
3963 This macro should be defined if the target has no hardware divide
3964 instructions. If this macro is defined, GCC will use an algorithm which
3965 make use of simple logical and arithmetic operations for 64-bit
3966 division. If the macro is not defined, GCC will use an algorithm which
3967 make use of a 64-bit by 32-bit divide primitive.
3970 @cindex @code{EDOM}, implicit usage
3973 The value of @code{EDOM} on the target machine, as a C integer constant
3974 expression. If you don't define this macro, GCC does not attempt to
3975 deposit the value of @code{EDOM} into @code{errno} directly. Look in
3976 @file{/usr/include/errno.h} to find the value of @code{EDOM} on your
3979 If you do not define @code{TARGET_EDOM}, then compiled code reports
3980 domain errors by calling the library function and letting it report the
3981 error. If mathematical functions on your system use @code{matherr} when
3982 there is an error, then you should leave @code{TARGET_EDOM} undefined so
3983 that @code{matherr} is used normally.
3986 @cindex @code{errno}, implicit usage
3987 @defmac GEN_ERRNO_RTX
3988 Define this macro as a C expression to create an rtl expression that
3989 refers to the global ``variable'' @code{errno}. (On certain systems,
3990 @code{errno} may not actually be a variable.) If you don't define this
3991 macro, a reasonable default is used.
3994 @hook TARGET_LIBC_HAS_FUNCTION
3996 @hook TARGET_LIBC_HAS_FAST_FUNCTION
3998 @defmac NEXT_OBJC_RUNTIME
3999 Set this macro to 1 to use the "NeXT" Objective-C message sending conventions
4000 by default. This calling convention involves passing the object, the selector
4001 and the method arguments all at once to the method-lookup library function.
4002 This is the usual setting when targeting Darwin/Mac OS X systems, which have
4003 the NeXT runtime installed.
4005 If the macro is set to 0, the "GNU" Objective-C message sending convention
4006 will be used by default. This convention passes just the object and the
4007 selector to the method-lookup function, which returns a pointer to the method.
4009 In either case, it remains possible to select code-generation for the alternate
4010 scheme, by means of compiler command line switches.
4013 @node Addressing Modes
4014 @section Addressing Modes
4015 @cindex addressing modes
4017 @c prevent bad page break with this line
4018 This is about addressing modes.
4020 @defmac HAVE_PRE_INCREMENT
4021 @defmacx HAVE_PRE_DECREMENT
4022 @defmacx HAVE_POST_INCREMENT
4023 @defmacx HAVE_POST_DECREMENT
4024 A C expression that is nonzero if the machine supports pre-increment,
4025 pre-decrement, post-increment, or post-decrement addressing respectively.
4028 @defmac HAVE_PRE_MODIFY_DISP
4029 @defmacx HAVE_POST_MODIFY_DISP
4030 A C expression that is nonzero if the machine supports pre- or
4031 post-address side-effect generation involving constants other than
4032 the size of the memory operand.
4035 @defmac HAVE_PRE_MODIFY_REG
4036 @defmacx HAVE_POST_MODIFY_REG
4037 A C expression that is nonzero if the machine supports pre- or
4038 post-address side-effect generation involving a register displacement.
4041 @defmac CONSTANT_ADDRESS_P (@var{x})
4042 A C expression that is 1 if the RTX @var{x} is a constant which
4043 is a valid address. On most machines the default definition of
4044 @code{(CONSTANT_P (@var{x}) && GET_CODE (@var{x}) != CONST_DOUBLE)}
4045 is acceptable, but a few machines are more restrictive as to which
4046 constant addresses are supported.
4049 @defmac CONSTANT_P (@var{x})
4050 @code{CONSTANT_P}, which is defined by target-independent code,
4051 accepts integer-values expressions whose values are not explicitly
4052 known, such as @code{symbol_ref}, @code{label_ref}, and @code{high}
4053 expressions and @code{const} arithmetic expressions, in addition to
4054 @code{const_int} and @code{const_double} expressions.
4057 @defmac MAX_REGS_PER_ADDRESS
4058 A number, the maximum number of registers that can appear in a valid
4059 memory address. Note that it is up to you to specify a value equal to
4060 the maximum number that @code{TARGET_LEGITIMATE_ADDRESS_P} would ever
4064 @hook TARGET_LEGITIMATE_ADDRESS_P
4066 @defmac TARGET_MEM_CONSTRAINT
4067 A single character to be used instead of the default @code{'m'}
4068 character for general memory addresses. This defines the constraint
4069 letter which matches the memory addresses accepted by
4070 @code{TARGET_LEGITIMATE_ADDRESS_P}. Define this macro if you want to
4071 support new address formats in your back end without changing the
4072 semantics of the @code{'m'} constraint. This is necessary in order to
4073 preserve functionality of inline assembly constructs using the
4074 @code{'m'} constraint.
4077 @defmac FIND_BASE_TERM (@var{x})
4078 A C expression to determine the base term of address @var{x},
4079 or to provide a simplified version of @var{x} from which @file{alias.c}
4080 can easily find the base term. This macro is used in only two places:
4081 @code{find_base_value} and @code{find_base_term} in @file{alias.c}.
4083 It is always safe for this macro to not be defined. It exists so
4084 that alias analysis can understand machine-dependent addresses.
4086 The typical use of this macro is to handle addresses containing
4087 a label_ref or symbol_ref within an UNSPEC@.
4090 @hook TARGET_LEGITIMIZE_ADDRESS
4092 @defmac LEGITIMIZE_RELOAD_ADDRESS (@var{x}, @var{mode}, @var{opnum}, @var{type}, @var{ind_levels}, @var{win})
4093 A C compound statement that attempts to replace @var{x}, which is an address
4094 that needs reloading, with a valid memory address for an operand of mode
4095 @var{mode}. @var{win} will be a C statement label elsewhere in the code.
4096 It is not necessary to define this macro, but it might be useful for
4097 performance reasons.
4099 For example, on the i386, it is sometimes possible to use a single
4100 reload register instead of two by reloading a sum of two pseudo
4101 registers into a register. On the other hand, for number of RISC
4102 processors offsets are limited so that often an intermediate address
4103 needs to be generated in order to address a stack slot. By defining
4104 @code{LEGITIMIZE_RELOAD_ADDRESS} appropriately, the intermediate addresses
4105 generated for adjacent some stack slots can be made identical, and thus
4108 @emph{Note}: This macro should be used with caution. It is necessary
4109 to know something of how reload works in order to effectively use this,
4110 and it is quite easy to produce macros that build in too much knowledge
4111 of reload internals.
4113 @emph{Note}: This macro must be able to reload an address created by a
4114 previous invocation of this macro. If it fails to handle such addresses
4115 then the compiler may generate incorrect code or abort.
4118 The macro definition should use @code{push_reload} to indicate parts that
4119 need reloading; @var{opnum}, @var{type} and @var{ind_levels} are usually
4120 suitable to be passed unaltered to @code{push_reload}.
4122 The code generated by this macro must not alter the substructure of
4123 @var{x}. If it transforms @var{x} into a more legitimate form, it
4124 should assign @var{x} (which will always be a C variable) a new value.
4125 This also applies to parts that you change indirectly by calling
4128 @findex strict_memory_address_p
4129 The macro definition may use @code{strict_memory_address_p} to test if
4130 the address has become legitimate.
4133 If you want to change only a part of @var{x}, one standard way of doing
4134 this is to use @code{copy_rtx}. Note, however, that it unshares only a
4135 single level of rtl. Thus, if the part to be changed is not at the
4136 top level, you'll need to replace first the top level.
4137 It is not necessary for this macro to come up with a legitimate
4138 address; but often a machine-dependent strategy can generate better code.
4141 @hook TARGET_MODE_DEPENDENT_ADDRESS_P
4143 @hook TARGET_LEGITIMATE_CONSTANT_P
4145 @hook TARGET_DELEGITIMIZE_ADDRESS
4147 @hook TARGET_CONST_NOT_OK_FOR_DEBUG_P
4149 @hook TARGET_CANNOT_FORCE_CONST_MEM
4151 @hook TARGET_USE_BLOCKS_FOR_CONSTANT_P
4153 @hook TARGET_USE_BLOCKS_FOR_DECL_P
4155 @hook TARGET_BUILTIN_RECIPROCAL
4157 @hook TARGET_VECTORIZE_BUILTIN_MASK_FOR_LOAD
4159 @hook TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST
4161 @hook TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT
4163 @hook TARGET_VECTORIZE_VECTOR_ALIGNMENT_REACHABLE
4165 @hook TARGET_VECTORIZE_VEC_PERM_CONST
4167 @hook TARGET_VECTORIZE_BUILTIN_CONVERSION
4169 @hook TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION
4171 @hook TARGET_VECTORIZE_BUILTIN_MD_VECTORIZED_FUNCTION
4173 @hook TARGET_VECTORIZE_SUPPORT_VECTOR_MISALIGNMENT
4175 @hook TARGET_VECTORIZE_PREFERRED_SIMD_MODE
4177 @hook TARGET_VECTORIZE_SPLIT_REDUCTION
4179 @hook TARGET_VECTORIZE_AUTOVECTORIZE_VECTOR_SIZES
4181 @hook TARGET_VECTORIZE_GET_MASK_MODE
4183 @hook TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE
4185 @hook TARGET_VECTORIZE_INIT_COST
4187 @hook TARGET_VECTORIZE_ADD_STMT_COST
4189 @hook TARGET_VECTORIZE_FINISH_COST
4191 @hook TARGET_VECTORIZE_DESTROY_COST_DATA
4193 @hook TARGET_VECTORIZE_BUILTIN_GATHER
4195 @hook TARGET_VECTORIZE_BUILTIN_SCATTER
4197 @hook TARGET_SIMD_CLONE_COMPUTE_VECSIZE_AND_SIMDLEN
4199 @hook TARGET_SIMD_CLONE_ADJUST
4201 @hook TARGET_SIMD_CLONE_USABLE
4203 @hook TARGET_SIMT_VF
4205 @hook TARGET_GOACC_VALIDATE_DIMS
4207 @hook TARGET_GOACC_DIM_LIMIT
4209 @hook TARGET_GOACC_FORK_JOIN
4211 @hook TARGET_GOACC_REDUCTION
4213 @hook TARGET_PREFERRED_ELSE_VALUE
4215 @node Anchored Addresses
4216 @section Anchored Addresses
4217 @cindex anchored addresses
4218 @cindex @option{-fsection-anchors}
4220 GCC usually addresses every static object as a separate entity.
4221 For example, if we have:
4225 int foo (void) @{ return a + b + c; @}
4228 the code for @code{foo} will usually calculate three separate symbolic
4229 addresses: those of @code{a}, @code{b} and @code{c}. On some targets,
4230 it would be better to calculate just one symbolic address and access
4231 the three variables relative to it. The equivalent pseudocode would
4237 register int *xr = &x;
4238 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
4242 (which isn't valid C). We refer to shared addresses like @code{x} as
4243 ``section anchors''. Their use is controlled by @option{-fsection-anchors}.
4245 The hooks below describe the target properties that GCC needs to know
4246 in order to make effective use of section anchors. It won't use
4247 section anchors at all unless either @code{TARGET_MIN_ANCHOR_OFFSET}
4248 or @code{TARGET_MAX_ANCHOR_OFFSET} is set to a nonzero value.
4250 @hook TARGET_MIN_ANCHOR_OFFSET
4252 @hook TARGET_MAX_ANCHOR_OFFSET
4254 @hook TARGET_ASM_OUTPUT_ANCHOR
4256 @hook TARGET_USE_ANCHORS_FOR_SYMBOL_P
4258 @node Condition Code
4259 @section Condition Code Status
4260 @cindex condition code status
4262 The macros in this section can be split in two families, according to the
4263 two ways of representing condition codes in GCC.
4265 The first representation is the so called @code{(cc0)} representation
4266 (@pxref{Jump Patterns}), where all instructions can have an implicit
4267 clobber of the condition codes. The second is the condition code
4268 register representation, which provides better schedulability for
4269 architectures that do have a condition code register, but on which
4270 most instructions do not affect it. The latter category includes
4273 The implicit clobbering poses a strong restriction on the placement of
4274 the definition and use of the condition code. In the past the definition
4275 and use were always adjacent. However, recent changes to support trapping
4276 arithmatic may result in the definition and user being in different blocks.
4277 Thus, there may be a @code{NOTE_INSN_BASIC_BLOCK} between them. Additionally,
4278 the definition may be the source of exception handling edges.
4280 These restrictions can prevent important
4281 optimizations on some machines. For example, on the IBM RS/6000, there
4282 is a delay for taken branches unless the condition code register is set
4283 three instructions earlier than the conditional branch. The instruction
4284 scheduler cannot perform this optimization if it is not permitted to
4285 separate the definition and use of the condition code register.
4287 For this reason, it is possible and suggested to use a register to
4288 represent the condition code for new ports. If there is a specific
4289 condition code register in the machine, use a hard register. If the
4290 condition code or comparison result can be placed in any general register,
4291 or if there are multiple condition registers, use a pseudo register.
4292 Registers used to store the condition code value will usually have a mode
4293 that is in class @code{MODE_CC}.
4295 Alternatively, you can use @code{BImode} if the comparison operator is
4296 specified already in the compare instruction. In this case, you are not
4297 interested in most macros in this section.
4300 * CC0 Condition Codes:: Old style representation of condition codes.
4301 * MODE_CC Condition Codes:: Modern representation of condition codes.
4304 @node CC0 Condition Codes
4305 @subsection Representation of condition codes using @code{(cc0)}
4309 The file @file{conditions.h} defines a variable @code{cc_status} to
4310 describe how the condition code was computed (in case the interpretation of
4311 the condition code depends on the instruction that it was set by). This
4312 variable contains the RTL expressions on which the condition code is
4313 currently based, and several standard flags.
4315 Sometimes additional machine-specific flags must be defined in the machine
4316 description header file. It can also add additional machine-specific
4317 information by defining @code{CC_STATUS_MDEP}.
4319 @defmac CC_STATUS_MDEP
4320 C code for a data type which is used for declaring the @code{mdep}
4321 component of @code{cc_status}. It defaults to @code{int}.
4323 This macro is not used on machines that do not use @code{cc0}.
4326 @defmac CC_STATUS_MDEP_INIT
4327 A C expression to initialize the @code{mdep} field to ``empty''.
4328 The default definition does nothing, since most machines don't use
4329 the field anyway. If you want to use the field, you should probably
4330 define this macro to initialize it.
4332 This macro is not used on machines that do not use @code{cc0}.
4335 @defmac NOTICE_UPDATE_CC (@var{exp}, @var{insn})
4336 A C compound statement to set the components of @code{cc_status}
4337 appropriately for an insn @var{insn} whose body is @var{exp}. It is
4338 this macro's responsibility to recognize insns that set the condition
4339 code as a byproduct of other activity as well as those that explicitly
4342 This macro is not used on machines that do not use @code{cc0}.
4344 If there are insns that do not set the condition code but do alter
4345 other machine registers, this macro must check to see whether they
4346 invalidate the expressions that the condition code is recorded as
4347 reflecting. For example, on the 68000, insns that store in address
4348 registers do not set the condition code, which means that usually
4349 @code{NOTICE_UPDATE_CC} can leave @code{cc_status} unaltered for such
4350 insns. But suppose that the previous insn set the condition code
4351 based on location @samp{a4@@(102)} and the current insn stores a new
4352 value in @samp{a4}. Although the condition code is not changed by
4353 this, it will no longer be true that it reflects the contents of
4354 @samp{a4@@(102)}. Therefore, @code{NOTICE_UPDATE_CC} must alter
4355 @code{cc_status} in this case to say that nothing is known about the
4356 condition code value.
4358 The definition of @code{NOTICE_UPDATE_CC} must be prepared to deal
4359 with the results of peephole optimization: insns whose patterns are
4360 @code{parallel} RTXs containing various @code{reg}, @code{mem} or
4361 constants which are just the operands. The RTL structure of these
4362 insns is not sufficient to indicate what the insns actually do. What
4363 @code{NOTICE_UPDATE_CC} should do when it sees one is just to run
4364 @code{CC_STATUS_INIT}.
4366 A possible definition of @code{NOTICE_UPDATE_CC} is to call a function
4367 that looks at an attribute (@pxref{Insn Attributes}) named, for example,
4368 @samp{cc}. This avoids having detailed information about patterns in
4369 two places, the @file{md} file and in @code{NOTICE_UPDATE_CC}.
4372 @node MODE_CC Condition Codes
4373 @subsection Representation of condition codes using registers
4377 @defmac SELECT_CC_MODE (@var{op}, @var{x}, @var{y})
4378 On many machines, the condition code may be produced by other instructions
4379 than compares, for example the branch can use directly the condition
4380 code set by a subtract instruction. However, on some machines
4381 when the condition code is set this way some bits (such as the overflow
4382 bit) are not set in the same way as a test instruction, so that a different
4383 branch instruction must be used for some conditional branches. When
4384 this happens, use the machine mode of the condition code register to
4385 record different formats of the condition code register. Modes can
4386 also be used to record which compare instruction (e.g.@: a signed or an
4387 unsigned comparison) produced the condition codes.
4389 If other modes than @code{CCmode} are required, add them to
4390 @file{@var{machine}-modes.def} and define @code{SELECT_CC_MODE} to choose
4391 a mode given an operand of a compare. This is needed because the modes
4392 have to be chosen not only during RTL generation but also, for example,
4393 by instruction combination. The result of @code{SELECT_CC_MODE} should
4394 be consistent with the mode used in the patterns; for example to support
4395 the case of the add on the SPARC discussed above, we have the pattern
4401 (plus:SI (match_operand:SI 0 "register_operand" "%r")
4402 (match_operand:SI 1 "arith_operand" "rI"))
4409 together with a @code{SELECT_CC_MODE} that returns @code{CCNZmode}
4410 for comparisons whose argument is a @code{plus}:
4413 #define SELECT_CC_MODE(OP,X,Y) \
4414 (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \
4415 ? ((OP == LT || OP == LE || OP == GT || OP == GE) \
4416 ? CCFPEmode : CCFPmode) \
4417 : ((GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \
4418 || GET_CODE (X) == NEG || GET_CODE (x) == ASHIFT) \
4419 ? CCNZmode : CCmode))
4422 Another reason to use modes is to retain information on which operands
4423 were used by the comparison; see @code{REVERSIBLE_CC_MODE} later in
4426 You should define this macro if and only if you define extra CC modes
4427 in @file{@var{machine}-modes.def}.
4430 @hook TARGET_CANONICALIZE_COMPARISON
4432 @defmac REVERSIBLE_CC_MODE (@var{mode})
4433 A C expression whose value is one if it is always safe to reverse a
4434 comparison whose mode is @var{mode}. If @code{SELECT_CC_MODE}
4435 can ever return @var{mode} for a floating-point inequality comparison,
4436 then @code{REVERSIBLE_CC_MODE (@var{mode})} must be zero.
4438 You need not define this macro if it would always returns zero or if the
4439 floating-point format is anything other than @code{IEEE_FLOAT_FORMAT}.
4440 For example, here is the definition used on the SPARC, where floating-point
4441 inequality comparisons are given either @code{CCFPEmode} or @code{CCFPmode}:
4444 #define REVERSIBLE_CC_MODE(MODE) \
4445 ((MODE) != CCFPEmode && (MODE) != CCFPmode)
4449 @defmac REVERSE_CONDITION (@var{code}, @var{mode})
4450 A C expression whose value is reversed condition code of the @var{code} for
4451 comparison done in CC_MODE @var{mode}. The macro is used only in case
4452 @code{REVERSIBLE_CC_MODE (@var{mode})} is nonzero. Define this macro in case
4453 machine has some non-standard way how to reverse certain conditionals. For
4454 instance in case all floating point conditions are non-trapping, compiler may
4455 freely convert unordered compares to ordered ones. Then definition may look
4459 #define REVERSE_CONDITION(CODE, MODE) \
4460 ((MODE) != CCFPmode ? reverse_condition (CODE) \
4461 : reverse_condition_maybe_unordered (CODE))
4465 @hook TARGET_FIXED_CONDITION_CODE_REGS
4467 @hook TARGET_CC_MODES_COMPATIBLE
4469 @hook TARGET_FLAGS_REGNUM
4472 @section Describing Relative Costs of Operations
4473 @cindex costs of instructions
4474 @cindex relative costs
4475 @cindex speed of instructions
4477 These macros let you describe the relative speed of various operations
4478 on the target machine.
4480 @defmac REGISTER_MOVE_COST (@var{mode}, @var{from}, @var{to})
4481 A C expression for the cost of moving data of mode @var{mode} from a
4482 register in class @var{from} to one in class @var{to}. The classes are
4483 expressed using the enumeration values such as @code{GENERAL_REGS}. A
4484 value of 2 is the default; other values are interpreted relative to
4487 It is not required that the cost always equal 2 when @var{from} is the
4488 same as @var{to}; on some machines it is expensive to move between
4489 registers if they are not general registers.
4491 If reload sees an insn consisting of a single @code{set} between two
4492 hard registers, and if @code{REGISTER_MOVE_COST} applied to their
4493 classes returns a value of 2, reload does not check to ensure that the
4494 constraints of the insn are met. Setting a cost of other than 2 will
4495 allow reload to verify that the constraints are met. You should do this
4496 if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
4498 These macros are obsolete, new ports should use the target hook
4499 @code{TARGET_REGISTER_MOVE_COST} instead.
4502 @hook TARGET_REGISTER_MOVE_COST
4504 @defmac MEMORY_MOVE_COST (@var{mode}, @var{class}, @var{in})
4505 A C expression for the cost of moving data of mode @var{mode} between a
4506 register of class @var{class} and memory; @var{in} is zero if the value
4507 is to be written to memory, nonzero if it is to be read in. This cost
4508 is relative to those in @code{REGISTER_MOVE_COST}. If moving between
4509 registers and memory is more expensive than between two registers, you
4510 should define this macro to express the relative cost.
4512 If you do not define this macro, GCC uses a default cost of 4 plus
4513 the cost of copying via a secondary reload register, if one is
4514 needed. If your machine requires a secondary reload register to copy
4515 between memory and a register of @var{class} but the reload mechanism is
4516 more complex than copying via an intermediate, define this macro to
4517 reflect the actual cost of the move.
4519 GCC defines the function @code{memory_move_secondary_cost} if
4520 secondary reloads are needed. It computes the costs due to copying via
4521 a secondary register. If your machine copies from memory using a
4522 secondary register in the conventional way but the default base value of
4523 4 is not correct for your machine, define this macro to add some other
4524 value to the result of that function. The arguments to that function
4525 are the same as to this macro.
4527 These macros are obsolete, new ports should use the target hook
4528 @code{TARGET_MEMORY_MOVE_COST} instead.
4531 @hook TARGET_MEMORY_MOVE_COST
4533 @defmac BRANCH_COST (@var{speed_p}, @var{predictable_p})
4534 A C expression for the cost of a branch instruction. A value of 1 is
4535 the default; other values are interpreted relative to that. Parameter
4536 @var{speed_p} is true when the branch in question should be optimized
4537 for speed. When it is false, @code{BRANCH_COST} should return a value
4538 optimal for code size rather than performance. @var{predictable_p} is
4539 true for well-predicted branches. On many architectures the
4540 @code{BRANCH_COST} can be reduced then.
4543 Here are additional macros which do not specify precise relative costs,
4544 but only that certain actions are more expensive than GCC would
4547 @defmac SLOW_BYTE_ACCESS
4548 Define this macro as a C expression which is nonzero if accessing less
4549 than a word of memory (i.e.@: a @code{char} or a @code{short}) is no
4550 faster than accessing a word of memory, i.e., if such access
4551 require more than one instruction or if there is no difference in cost
4552 between byte and (aligned) word loads.
4554 When this macro is not defined, the compiler will access a field by
4555 finding the smallest containing object; when it is defined, a fullword
4556 load will be used if alignment permits. Unless bytes accesses are
4557 faster than word accesses, using word accesses is preferable since it
4558 may eliminate subsequent memory access if subsequent accesses occur to
4559 other fields in the same word of the structure, but to different bytes.
4562 @hook TARGET_SLOW_UNALIGNED_ACCESS
4564 @defmac MOVE_RATIO (@var{speed})
4565 The threshold of number of scalar memory-to-memory move insns, @emph{below}
4566 which a sequence of insns should be generated instead of a
4567 string move insn or a library call. Increasing the value will always
4568 make code faster, but eventually incurs high cost in increased code size.
4570 Note that on machines where the corresponding move insn is a
4571 @code{define_expand} that emits a sequence of insns, this macro counts
4572 the number of such sequences.
4574 The parameter @var{speed} is true if the code is currently being
4575 optimized for speed rather than size.
4577 If you don't define this, a reasonable default is used.
4580 @hook TARGET_USE_BY_PIECES_INFRASTRUCTURE_P
4582 @hook TARGET_COMPARE_BY_PIECES_BRANCH_RATIO
4584 @defmac MOVE_MAX_PIECES
4585 A C expression used by @code{move_by_pieces} to determine the largest unit
4586 a load or store used to copy memory is. Defaults to @code{MOVE_MAX}.
4589 @defmac STORE_MAX_PIECES
4590 A C expression used by @code{store_by_pieces} to determine the largest unit
4591 a store used to memory is. Defaults to @code{MOVE_MAX_PIECES}, or two times
4592 the size of @code{HOST_WIDE_INT}, whichever is smaller.
4595 @defmac COMPARE_MAX_PIECES
4596 A C expression used by @code{compare_by_pieces} to determine the largest unit
4597 a load or store used to compare memory is. Defaults to
4598 @code{MOVE_MAX_PIECES}.
4601 @defmac CLEAR_RATIO (@var{speed})
4602 The threshold of number of scalar move insns, @emph{below} which a sequence
4603 of insns should be generated to clear memory instead of a string clear insn
4604 or a library call. Increasing the value will always make code faster, but
4605 eventually incurs high cost in increased code size.
4607 The parameter @var{speed} is true if the code is currently being
4608 optimized for speed rather than size.
4610 If you don't define this, a reasonable default is used.
4613 @defmac SET_RATIO (@var{speed})
4614 The threshold of number of scalar move insns, @emph{below} which a sequence
4615 of insns should be generated to set memory to a constant value, instead of
4616 a block set insn or a library call.
4617 Increasing the value will always make code faster, but
4618 eventually incurs high cost in increased code size.
4620 The parameter @var{speed} is true if the code is currently being
4621 optimized for speed rather than size.
4623 If you don't define this, it defaults to the value of @code{MOVE_RATIO}.
4626 @defmac USE_LOAD_POST_INCREMENT (@var{mode})
4627 A C expression used to determine whether a load postincrement is a good
4628 thing to use for a given mode. Defaults to the value of
4629 @code{HAVE_POST_INCREMENT}.
4632 @defmac USE_LOAD_POST_DECREMENT (@var{mode})
4633 A C expression used to determine whether a load postdecrement is a good
4634 thing to use for a given mode. Defaults to the value of
4635 @code{HAVE_POST_DECREMENT}.
4638 @defmac USE_LOAD_PRE_INCREMENT (@var{mode})
4639 A C expression used to determine whether a load preincrement is a good
4640 thing to use for a given mode. Defaults to the value of
4641 @code{HAVE_PRE_INCREMENT}.
4644 @defmac USE_LOAD_PRE_DECREMENT (@var{mode})
4645 A C expression used to determine whether a load predecrement is a good
4646 thing to use for a given mode. Defaults to the value of
4647 @code{HAVE_PRE_DECREMENT}.
4650 @defmac USE_STORE_POST_INCREMENT (@var{mode})
4651 A C expression used to determine whether a store postincrement is a good
4652 thing to use for a given mode. Defaults to the value of
4653 @code{HAVE_POST_INCREMENT}.
4656 @defmac USE_STORE_POST_DECREMENT (@var{mode})
4657 A C expression used to determine whether a store postdecrement is a good
4658 thing to use for a given mode. Defaults to the value of
4659 @code{HAVE_POST_DECREMENT}.
4662 @defmac USE_STORE_PRE_INCREMENT (@var{mode})
4663 This macro is used to determine whether a store preincrement is a good
4664 thing to use for a given mode. Defaults to the value of
4665 @code{HAVE_PRE_INCREMENT}.
4668 @defmac USE_STORE_PRE_DECREMENT (@var{mode})
4669 This macro is used to determine whether a store predecrement is a good
4670 thing to use for a given mode. Defaults to the value of
4671 @code{HAVE_PRE_DECREMENT}.
4674 @defmac NO_FUNCTION_CSE
4675 Define this macro to be true if it is as good or better to call a constant
4676 function address than to call an address kept in a register.
4679 @defmac LOGICAL_OP_NON_SHORT_CIRCUIT
4680 Define this macro if a non-short-circuit operation produced by
4681 @samp{fold_range_test ()} is optimal. This macro defaults to true if
4682 @code{BRANCH_COST} is greater than or equal to the value 2.
4685 @hook TARGET_OPTAB_SUPPORTED_P
4687 @hook TARGET_RTX_COSTS
4689 @hook TARGET_ADDRESS_COST
4691 @hook TARGET_INSN_COST
4693 @hook TARGET_MAX_NOCE_IFCVT_SEQ_COST
4695 @hook TARGET_NOCE_CONVERSION_PROFITABLE_P
4697 @hook TARGET_NO_SPECULATION_IN_DELAY_SLOTS_P
4699 @hook TARGET_ESTIMATED_POLY_VALUE
4702 @section Adjusting the Instruction Scheduler
4704 The instruction scheduler may need a fair amount of machine-specific
4705 adjustment in order to produce good code. GCC provides several target
4706 hooks for this purpose. It is usually enough to define just a few of
4707 them: try the first ones in this list first.
4709 @hook TARGET_SCHED_ISSUE_RATE
4711 @hook TARGET_SCHED_VARIABLE_ISSUE
4713 @hook TARGET_SCHED_ADJUST_COST
4715 @hook TARGET_SCHED_ADJUST_PRIORITY
4717 @hook TARGET_SCHED_REORDER
4719 @hook TARGET_SCHED_REORDER2
4721 @hook TARGET_SCHED_MACRO_FUSION_P
4723 @hook TARGET_SCHED_MACRO_FUSION_PAIR_P
4725 @hook TARGET_SCHED_DEPENDENCIES_EVALUATION_HOOK
4727 @hook TARGET_SCHED_INIT
4729 @hook TARGET_SCHED_FINISH
4731 @hook TARGET_SCHED_INIT_GLOBAL
4733 @hook TARGET_SCHED_FINISH_GLOBAL
4735 @hook TARGET_SCHED_DFA_PRE_CYCLE_INSN
4737 @hook TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN
4739 @hook TARGET_SCHED_DFA_POST_CYCLE_INSN
4741 @hook TARGET_SCHED_INIT_DFA_POST_CYCLE_INSN
4743 @hook TARGET_SCHED_DFA_PRE_ADVANCE_CYCLE
4745 @hook TARGET_SCHED_DFA_POST_ADVANCE_CYCLE
4747 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD
4749 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD
4751 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_BEGIN
4753 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_ISSUE
4755 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_BACKTRACK
4757 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_END
4759 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_INIT
4761 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_FINI
4763 @hook TARGET_SCHED_DFA_NEW_CYCLE
4765 @hook TARGET_SCHED_IS_COSTLY_DEPENDENCE
4767 @hook TARGET_SCHED_H_I_D_EXTENDED
4769 @hook TARGET_SCHED_ALLOC_SCHED_CONTEXT
4771 @hook TARGET_SCHED_INIT_SCHED_CONTEXT
4773 @hook TARGET_SCHED_SET_SCHED_CONTEXT
4775 @hook TARGET_SCHED_CLEAR_SCHED_CONTEXT
4777 @hook TARGET_SCHED_FREE_SCHED_CONTEXT
4779 @hook TARGET_SCHED_SPECULATE_INSN
4781 @hook TARGET_SCHED_NEEDS_BLOCK_P
4783 @hook TARGET_SCHED_GEN_SPEC_CHECK
4785 @hook TARGET_SCHED_SET_SCHED_FLAGS
4787 @hook TARGET_SCHED_CAN_SPECULATE_INSN
4789 @hook TARGET_SCHED_SMS_RES_MII
4791 @hook TARGET_SCHED_DISPATCH
4793 @hook TARGET_SCHED_DISPATCH_DO
4795 @hook TARGET_SCHED_EXPOSED_PIPELINE
4797 @hook TARGET_SCHED_REASSOCIATION_WIDTH
4799 @hook TARGET_SCHED_FUSION_PRIORITY
4801 @hook TARGET_EXPAND_DIVMOD_LIBFUNC
4804 @section Dividing the Output into Sections (Texts, Data, @dots{})
4805 @c the above section title is WAY too long. maybe cut the part between
4806 @c the (...)? --mew 10feb93
4808 An object file is divided into sections containing different types of
4809 data. In the most common case, there are three sections: the @dfn{text
4810 section}, which holds instructions and read-only data; the @dfn{data
4811 section}, which holds initialized writable data; and the @dfn{bss
4812 section}, which holds uninitialized data. Some systems have other kinds
4815 @file{varasm.c} provides several well-known sections, such as
4816 @code{text_section}, @code{data_section} and @code{bss_section}.
4817 The normal way of controlling a @code{@var{foo}_section} variable
4818 is to define the associated @code{@var{FOO}_SECTION_ASM_OP} macro,
4819 as described below. The macros are only read once, when @file{varasm.c}
4820 initializes itself, so their values must be run-time constants.
4821 They may however depend on command-line flags.
4823 @emph{Note:} Some run-time files, such @file{crtstuff.c}, also make
4824 use of the @code{@var{FOO}_SECTION_ASM_OP} macros, and expect them
4825 to be string literals.
4827 Some assemblers require a different string to be written every time a
4828 section is selected. If your assembler falls into this category, you
4829 should define the @code{TARGET_ASM_INIT_SECTIONS} hook and use
4830 @code{get_unnamed_section} to set up the sections.
4832 You must always create a @code{text_section}, either by defining
4833 @code{TEXT_SECTION_ASM_OP} or by initializing @code{text_section}
4834 in @code{TARGET_ASM_INIT_SECTIONS}. The same is true of
4835 @code{data_section} and @code{DATA_SECTION_ASM_OP}. If you do not
4836 create a distinct @code{readonly_data_section}, the default is to
4837 reuse @code{text_section}.
4839 All the other @file{varasm.c} sections are optional, and are null
4840 if the target does not provide them.
4842 @defmac TEXT_SECTION_ASM_OP
4843 A C expression whose value is a string, including spacing, containing the
4844 assembler operation that should precede instructions and read-only data.
4845 Normally @code{"\t.text"} is right.
4848 @defmac HOT_TEXT_SECTION_NAME
4849 If defined, a C string constant for the name of the section containing most
4850 frequently executed functions of the program. If not defined, GCC will provide
4851 a default definition if the target supports named sections.
4854 @defmac UNLIKELY_EXECUTED_TEXT_SECTION_NAME
4855 If defined, a C string constant for the name of the section containing unlikely
4856 executed functions in the program.
4859 @defmac DATA_SECTION_ASM_OP
4860 A C expression whose value is a string, including spacing, containing the
4861 assembler operation to identify the following data as writable initialized
4862 data. Normally @code{"\t.data"} is right.
4865 @defmac SDATA_SECTION_ASM_OP
4866 If defined, a C expression whose value is a string, including spacing,
4867 containing the assembler operation to identify the following data as
4868 initialized, writable small data.
4871 @defmac READONLY_DATA_SECTION_ASM_OP
4872 A C expression whose value is a string, including spacing, containing the
4873 assembler operation to identify the following data as read-only initialized
4877 @defmac BSS_SECTION_ASM_OP
4878 If defined, a C expression whose value is a string, including spacing,
4879 containing the assembler operation to identify the following data as
4880 uninitialized global data. If not defined, and
4881 @code{ASM_OUTPUT_ALIGNED_BSS} not defined,
4882 uninitialized global data will be output in the data section if
4883 @option{-fno-common} is passed, otherwise @code{ASM_OUTPUT_COMMON} will be
4887 @defmac SBSS_SECTION_ASM_OP
4888 If defined, a C expression whose value is a string, including spacing,
4889 containing the assembler operation to identify the following data as
4890 uninitialized, writable small data.
4893 @defmac TLS_COMMON_ASM_OP
4894 If defined, a C expression whose value is a string containing the
4895 assembler operation to identify the following data as thread-local
4896 common data. The default is @code{".tls_common"}.
4899 @defmac TLS_SECTION_ASM_FLAG
4900 If defined, a C expression whose value is a character constant
4901 containing the flag used to mark a section as a TLS section. The
4902 default is @code{'T'}.
4905 @defmac INIT_SECTION_ASM_OP
4906 If defined, a C expression whose value is a string, including spacing,
4907 containing the assembler operation to identify the following data as
4908 initialization code. If not defined, GCC will assume such a section does
4909 not exist. This section has no corresponding @code{init_section}
4910 variable; it is used entirely in runtime code.
4913 @defmac FINI_SECTION_ASM_OP
4914 If defined, a C expression whose value is a string, including spacing,
4915 containing the assembler operation to identify the following data as
4916 finalization code. If not defined, GCC will assume such a section does
4917 not exist. This section has no corresponding @code{fini_section}
4918 variable; it is used entirely in runtime code.
4921 @defmac INIT_ARRAY_SECTION_ASM_OP
4922 If defined, a C expression whose value is a string, including spacing,
4923 containing the assembler operation to identify the following data as
4924 part of the @code{.init_array} (or equivalent) section. If not
4925 defined, GCC will assume such a section does not exist. Do not define
4926 both this macro and @code{INIT_SECTION_ASM_OP}.
4929 @defmac FINI_ARRAY_SECTION_ASM_OP
4930 If defined, a C expression whose value is a string, including spacing,
4931 containing the assembler operation to identify the following data as
4932 part of the @code{.fini_array} (or equivalent) section. If not
4933 defined, GCC will assume such a section does not exist. Do not define
4934 both this macro and @code{FINI_SECTION_ASM_OP}.
4937 @defmac MACH_DEP_SECTION_ASM_FLAG
4938 If defined, a C expression whose value is a character constant
4939 containing the flag used to mark a machine-dependent section. This
4940 corresponds to the @code{SECTION_MACH_DEP} section flag.
4943 @defmac CRT_CALL_STATIC_FUNCTION (@var{section_op}, @var{function})
4944 If defined, an ASM statement that switches to a different section
4945 via @var{section_op}, calls @var{function}, and switches back to
4946 the text section. This is used in @file{crtstuff.c} if
4947 @code{INIT_SECTION_ASM_OP} or @code{FINI_SECTION_ASM_OP} to calls
4948 to initialization and finalization functions from the init and fini
4949 sections. By default, this macro uses a simple function call. Some
4950 ports need hand-crafted assembly code to avoid dependencies on
4951 registers initialized in the function prologue or to ensure that
4952 constant pools don't end up too far way in the text section.
4955 @defmac TARGET_LIBGCC_SDATA_SECTION
4956 If defined, a string which names the section into which small
4957 variables defined in crtstuff and libgcc should go. This is useful
4958 when the target has options for optimizing access to small data, and
4959 you want the crtstuff and libgcc routines to be conservative in what
4960 they expect of your application yet liberal in what your application
4961 expects. For example, for targets with a @code{.sdata} section (like
4962 MIPS), you could compile crtstuff with @code{-G 0} so that it doesn't
4963 require small data support from your application, but use this macro
4964 to put small data into @code{.sdata} so that your application can
4965 access these variables whether it uses small data or not.
4968 @defmac FORCE_CODE_SECTION_ALIGN
4969 If defined, an ASM statement that aligns a code section to some
4970 arbitrary boundary. This is used to force all fragments of the
4971 @code{.init} and @code{.fini} sections to have to same alignment
4972 and thus prevent the linker from having to add any padding.
4975 @defmac JUMP_TABLES_IN_TEXT_SECTION
4976 Define this macro to be an expression with a nonzero value if jump
4977 tables (for @code{tablejump} insns) should be output in the text
4978 section, along with the assembler instructions. Otherwise, the
4979 readonly data section is used.
4981 This macro is irrelevant if there is no separate readonly data section.
4984 @hook TARGET_ASM_INIT_SECTIONS
4986 @hook TARGET_ASM_RELOC_RW_MASK
4988 @hook TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC
4990 @hook TARGET_ASM_SELECT_SECTION
4992 @defmac USE_SELECT_SECTION_FOR_FUNCTIONS
4993 Define this macro if you wish TARGET_ASM_SELECT_SECTION to be called
4994 for @code{FUNCTION_DECL}s as well as for variables and constants.
4996 In the case of a @code{FUNCTION_DECL}, @var{reloc} will be zero if the
4997 function has been determined to be likely to be called, and nonzero if
4998 it is unlikely to be called.
5001 @hook TARGET_ASM_UNIQUE_SECTION
5003 @hook TARGET_ASM_FUNCTION_RODATA_SECTION
5005 @hook TARGET_ASM_MERGEABLE_RODATA_PREFIX
5007 @hook TARGET_ASM_TM_CLONE_TABLE_SECTION
5009 @hook TARGET_ASM_SELECT_RTX_SECTION
5011 @hook TARGET_MANGLE_DECL_ASSEMBLER_NAME
5013 @hook TARGET_ENCODE_SECTION_INFO
5015 @hook TARGET_STRIP_NAME_ENCODING
5017 @hook TARGET_IN_SMALL_DATA_P
5019 @hook TARGET_HAVE_SRODATA_SECTION
5021 @hook TARGET_PROFILE_BEFORE_PROLOGUE
5023 @hook TARGET_BINDS_LOCAL_P
5025 @hook TARGET_HAVE_TLS
5029 @section Position Independent Code
5030 @cindex position independent code
5033 This section describes macros that help implement generation of position
5034 independent code. Simply defining these macros is not enough to
5035 generate valid PIC; you must also add support to the hook
5036 @code{TARGET_LEGITIMATE_ADDRESS_P} and to the macro
5037 @code{PRINT_OPERAND_ADDRESS}, as well as @code{LEGITIMIZE_ADDRESS}. You
5038 must modify the definition of @samp{movsi} to do something appropriate
5039 when the source operand contains a symbolic address. You may also
5040 need to alter the handling of switch statements so that they use
5042 @c i rearranged the order of the macros above to try to force one of
5043 @c them to the next line, to eliminate an overfull hbox. --mew 10feb93
5045 @defmac PIC_OFFSET_TABLE_REGNUM
5046 The register number of the register used to address a table of static
5047 data addresses in memory. In some cases this register is defined by a
5048 processor's ``application binary interface'' (ABI)@. When this macro
5049 is defined, RTL is generated for this register once, as with the stack
5050 pointer and frame pointer registers. If this macro is not defined, it
5051 is up to the machine-dependent files to allocate such a register (if
5052 necessary). Note that this register must be fixed when in use (e.g.@:
5053 when @code{flag_pic} is true).
5056 @defmac PIC_OFFSET_TABLE_REG_CALL_CLOBBERED
5057 A C expression that is nonzero if the register defined by
5058 @code{PIC_OFFSET_TABLE_REGNUM} is clobbered by calls. If not defined,
5059 the default is zero. Do not define
5060 this macro if @code{PIC_OFFSET_TABLE_REGNUM} is not defined.
5063 @defmac LEGITIMATE_PIC_OPERAND_P (@var{x})
5064 A C expression that is nonzero if @var{x} is a legitimate immediate
5065 operand on the target machine when generating position independent code.
5066 You can assume that @var{x} satisfies @code{CONSTANT_P}, so you need not
5067 check this. You can also assume @var{flag_pic} is true, so you need not
5068 check it either. You need not define this macro if all constants
5069 (including @code{SYMBOL_REF}) can be immediate operands when generating
5070 position independent code.
5073 @node Assembler Format
5074 @section Defining the Output Assembler Language
5076 This section describes macros whose principal purpose is to describe how
5077 to write instructions in assembler language---rather than what the
5081 * File Framework:: Structural information for the assembler file.
5082 * Data Output:: Output of constants (numbers, strings, addresses).
5083 * Uninitialized Data:: Output of uninitialized variables.
5084 * Label Output:: Output and generation of labels.
5085 * Initialization:: General principles of initialization
5086 and termination routines.
5087 * Macros for Initialization::
5088 Specific macros that control the handling of
5089 initialization and termination routines.
5090 * Instruction Output:: Output of actual instructions.
5091 * Dispatch Tables:: Output of jump tables.
5092 * Exception Region Output:: Output of exception region code.
5093 * Alignment Output:: Pseudo ops for alignment and skipping data.
5096 @node File Framework
5097 @subsection The Overall Framework of an Assembler File
5098 @cindex assembler format
5099 @cindex output of assembler code
5101 @c prevent bad page break with this line
5102 This describes the overall framework of an assembly file.
5104 @findex default_file_start
5105 @hook TARGET_ASM_FILE_START
5107 @hook TARGET_ASM_FILE_START_APP_OFF
5109 @hook TARGET_ASM_FILE_START_FILE_DIRECTIVE
5111 @hook TARGET_ASM_FILE_END
5113 @deftypefun void file_end_indicate_exec_stack ()
5114 Some systems use a common convention, the @samp{.note.GNU-stack}
5115 special section, to indicate whether or not an object file relies on
5116 the stack being executable. If your system uses this convention, you
5117 should define @code{TARGET_ASM_FILE_END} to this function. If you
5118 need to do other things in that hook, have your hook function call
5122 @hook TARGET_ASM_LTO_START
5124 @hook TARGET_ASM_LTO_END
5126 @hook TARGET_ASM_CODE_END
5128 @defmac ASM_COMMENT_START
5129 A C string constant describing how to begin a comment in the target
5130 assembler language. The compiler assumes that the comment will end at
5131 the end of the line.
5135 A C string constant for text to be output before each @code{asm}
5136 statement or group of consecutive ones. Normally this is
5137 @code{"#APP"}, which is a comment that has no effect on most
5138 assemblers but tells the GNU assembler that it must check the lines
5139 that follow for all valid assembler constructs.
5143 A C string constant for text to be output after each @code{asm}
5144 statement or group of consecutive ones. Normally this is
5145 @code{"#NO_APP"}, which tells the GNU assembler to resume making the
5146 time-saving assumptions that are valid for ordinary compiler output.
5149 @defmac ASM_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name})
5150 A C statement to output COFF information or DWARF debugging information
5151 which indicates that filename @var{name} is the current source file to
5152 the stdio stream @var{stream}.
5154 This macro need not be defined if the standard form of output
5155 for the file format in use is appropriate.
5158 @hook TARGET_ASM_OUTPUT_SOURCE_FILENAME
5160 @hook TARGET_ASM_OUTPUT_IDENT
5162 @defmac OUTPUT_QUOTED_STRING (@var{stream}, @var{string})
5163 A C statement to output the string @var{string} to the stdio stream
5164 @var{stream}. If you do not call the function @code{output_quoted_string}
5165 in your config files, GCC will only call it to output filenames to
5166 the assembler source. So you can use it to canonicalize the format
5167 of the filename using this macro.
5170 @hook TARGET_ASM_NAMED_SECTION
5172 @hook TARGET_ASM_ELF_FLAGS_NUMERIC
5174 @hook TARGET_ASM_FUNCTION_SECTION
5176 @hook TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS
5178 @hook TARGET_HAVE_NAMED_SECTIONS
5179 This flag is true if the target supports @code{TARGET_ASM_NAMED_SECTION}.
5180 It must not be modified by command-line option processing.
5183 @anchor{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}
5184 @hook TARGET_HAVE_SWITCHABLE_BSS_SECTIONS
5186 @hook TARGET_SECTION_TYPE_FLAGS
5188 @hook TARGET_ASM_RECORD_GCC_SWITCHES
5190 @hook TARGET_ASM_RECORD_GCC_SWITCHES_SECTION
5194 @subsection Output of Data
5197 @hook TARGET_ASM_BYTE_OP
5199 @hook TARGET_ASM_INTEGER
5201 @hook TARGET_ASM_DECL_END
5203 @hook TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA
5205 @defmac ASM_OUTPUT_ASCII (@var{stream}, @var{ptr}, @var{len})
5206 A C statement to output to the stdio stream @var{stream} an assembler
5207 instruction to assemble a string constant containing the @var{len}
5208 bytes at @var{ptr}. @var{ptr} will be a C expression of type
5209 @code{char *} and @var{len} a C expression of type @code{int}.
5211 If the assembler has a @code{.ascii} pseudo-op as found in the
5212 Berkeley Unix assembler, do not define the macro
5213 @code{ASM_OUTPUT_ASCII}.
5216 @defmac ASM_OUTPUT_FDESC (@var{stream}, @var{decl}, @var{n})
5217 A C statement to output word @var{n} of a function descriptor for
5218 @var{decl}. This must be defined if @code{TARGET_VTABLE_USES_DESCRIPTORS}
5219 is defined, and is otherwise unused.
5222 @defmac CONSTANT_POOL_BEFORE_FUNCTION
5223 You may define this macro as a C expression. You should define the
5224 expression to have a nonzero value if GCC should output the constant
5225 pool for a function before the code for the function, or a zero value if
5226 GCC should output the constant pool after the function. If you do
5227 not define this macro, the usual case, GCC will output the constant
5228 pool before the function.
5231 @defmac ASM_OUTPUT_POOL_PROLOGUE (@var{file}, @var{funname}, @var{fundecl}, @var{size})
5232 A C statement to output assembler commands to define the start of the
5233 constant pool for a function. @var{funname} is a string giving
5234 the name of the function. Should the return type of the function
5235 be required, it can be obtained via @var{fundecl}. @var{size}
5236 is the size, in bytes, of the constant pool that will be written
5237 immediately after this call.
5239 If no constant-pool prefix is required, the usual case, this macro need
5243 @defmac ASM_OUTPUT_SPECIAL_POOL_ENTRY (@var{file}, @var{x}, @var{mode}, @var{align}, @var{labelno}, @var{jumpto})
5244 A C statement (with or without semicolon) to output a constant in the
5245 constant pool, if it needs special treatment. (This macro need not do
5246 anything for RTL expressions that can be output normally.)
5248 The argument @var{file} is the standard I/O stream to output the
5249 assembler code on. @var{x} is the RTL expression for the constant to
5250 output, and @var{mode} is the machine mode (in case @var{x} is a
5251 @samp{const_int}). @var{align} is the required alignment for the value
5252 @var{x}; you should output an assembler directive to force this much
5255 The argument @var{labelno} is a number to use in an internal label for
5256 the address of this pool entry. The definition of this macro is
5257 responsible for outputting the label definition at the proper place.
5258 Here is how to do this:
5261 @code{(*targetm.asm_out.internal_label)} (@var{file}, "LC", @var{labelno});
5264 When you output a pool entry specially, you should end with a
5265 @code{goto} to the label @var{jumpto}. This will prevent the same pool
5266 entry from being output a second time in the usual manner.
5268 You need not define this macro if it would do nothing.
5271 @defmac ASM_OUTPUT_POOL_EPILOGUE (@var{file} @var{funname} @var{fundecl} @var{size})
5272 A C statement to output assembler commands to at the end of the constant
5273 pool for a function. @var{funname} is a string giving the name of the
5274 function. Should the return type of the function be required, you can
5275 obtain it via @var{fundecl}. @var{size} is the size, in bytes, of the
5276 constant pool that GCC wrote immediately before this call.
5278 If no constant-pool epilogue is required, the usual case, you need not
5282 @defmac IS_ASM_LOGICAL_LINE_SEPARATOR (@var{C}, @var{STR})
5283 Define this macro as a C expression which is nonzero if @var{C} is
5284 used as a logical line separator by the assembler. @var{STR} points
5285 to the position in the string where @var{C} was found; this can be used if
5286 a line separator uses multiple characters.
5288 If you do not define this macro, the default is that only
5289 the character @samp{;} is treated as a logical line separator.
5292 @hook TARGET_ASM_OPEN_PAREN
5294 These macros are provided by @file{real.h} for writing the definitions
5295 of @code{ASM_OUTPUT_DOUBLE} and the like:
5297 @defmac REAL_VALUE_TO_TARGET_SINGLE (@var{x}, @var{l})
5298 @defmacx REAL_VALUE_TO_TARGET_DOUBLE (@var{x}, @var{l})
5299 @defmacx REAL_VALUE_TO_TARGET_LONG_DOUBLE (@var{x}, @var{l})
5300 @defmacx REAL_VALUE_TO_TARGET_DECIMAL32 (@var{x}, @var{l})
5301 @defmacx REAL_VALUE_TO_TARGET_DECIMAL64 (@var{x}, @var{l})
5302 @defmacx REAL_VALUE_TO_TARGET_DECIMAL128 (@var{x}, @var{l})
5303 These translate @var{x}, of type @code{REAL_VALUE_TYPE}, to the
5304 target's floating point representation, and store its bit pattern in
5305 the variable @var{l}. For @code{REAL_VALUE_TO_TARGET_SINGLE} and
5306 @code{REAL_VALUE_TO_TARGET_DECIMAL32}, this variable should be a
5307 simple @code{long int}. For the others, it should be an array of
5308 @code{long int}. The number of elements in this array is determined
5309 by the size of the desired target floating point data type: 32 bits of
5310 it go in each @code{long int} array element. Each array element holds
5311 32 bits of the result, even if @code{long int} is wider than 32 bits
5312 on the host machine.
5314 The array element values are designed so that you can print them out
5315 using @code{fprintf} in the order they should appear in the target
5319 @node Uninitialized Data
5320 @subsection Output of Uninitialized Variables
5322 Each of the macros in this section is used to do the whole job of
5323 outputting a single uninitialized variable.
5325 @defmac ASM_OUTPUT_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
5326 A C statement (sans semicolon) to output to the stdio stream
5327 @var{stream} the assembler definition of a common-label named
5328 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
5329 is the size rounded up to whatever alignment the caller wants. It is
5330 possible that @var{size} may be zero, for instance if a struct with no
5331 other member than a zero-length array is defined. In this case, the
5332 backend must output a symbol definition that allocates at least one
5333 byte, both so that the address of the resulting object does not compare
5334 equal to any other, and because some object formats cannot even express
5335 the concept of a zero-sized common symbol, as that is how they represent
5336 an ordinary undefined external.
5338 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
5339 output the name itself; before and after that, output the additional
5340 assembler syntax for defining the name, and a newline.
5342 This macro controls how the assembler definitions of uninitialized
5343 common global variables are output.
5346 @defmac ASM_OUTPUT_ALIGNED_COMMON (@var{stream}, @var{name}, @var{size}, @var{alignment})
5347 Like @code{ASM_OUTPUT_COMMON} except takes the required alignment as a
5348 separate, explicit argument. If you define this macro, it is used in
5349 place of @code{ASM_OUTPUT_COMMON}, and gives you more flexibility in
5350 handling the required alignment of the variable. The alignment is specified
5351 as the number of bits.
5354 @defmac ASM_OUTPUT_ALIGNED_DECL_COMMON (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
5355 Like @code{ASM_OUTPUT_ALIGNED_COMMON} except that @var{decl} of the
5356 variable to be output, if there is one, or @code{NULL_TREE} if there
5357 is no corresponding variable. If you define this macro, GCC will use it
5358 in place of both @code{ASM_OUTPUT_COMMON} and
5359 @code{ASM_OUTPUT_ALIGNED_COMMON}. Define this macro when you need to see
5360 the variable's decl in order to chose what to output.
5363 @defmac ASM_OUTPUT_ALIGNED_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
5364 A C statement (sans semicolon) to output to the stdio stream
5365 @var{stream} the assembler definition of uninitialized global @var{decl} named
5366 @var{name} whose size is @var{size} bytes. The variable @var{alignment}
5367 is the alignment specified as the number of bits.
5369 Try to use function @code{asm_output_aligned_bss} defined in file
5370 @file{varasm.c} when defining this macro. If unable, use the expression
5371 @code{assemble_name (@var{stream}, @var{name})} to output the name itself;
5372 before and after that, output the additional assembler syntax for defining
5373 the name, and a newline.
5375 There are two ways of handling global BSS@. One is to define this macro.
5376 The other is to have @code{TARGET_ASM_SELECT_SECTION} return a
5377 switchable BSS section (@pxref{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}).
5378 You do not need to do both.
5380 Some languages do not have @code{common} data, and require a
5381 non-common form of global BSS in order to handle uninitialized globals
5382 efficiently. C++ is one example of this. However, if the target does
5383 not support global BSS, the front end may choose to make globals
5384 common in order to save space in the object file.
5387 @defmac ASM_OUTPUT_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded})
5388 A C statement (sans semicolon) to output to the stdio stream
5389 @var{stream} the assembler definition of a local-common-label named
5390 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
5391 is the size rounded up to whatever alignment the caller wants.
5393 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
5394 output the name itself; before and after that, output the additional
5395 assembler syntax for defining the name, and a newline.
5397 This macro controls how the assembler definitions of uninitialized
5398 static variables are output.
5401 @defmac ASM_OUTPUT_ALIGNED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{alignment})
5402 Like @code{ASM_OUTPUT_LOCAL} except takes the required alignment as a
5403 separate, explicit argument. If you define this macro, it is used in
5404 place of @code{ASM_OUTPUT_LOCAL}, and gives you more flexibility in
5405 handling the required alignment of the variable. The alignment is specified
5406 as the number of bits.
5409 @defmac ASM_OUTPUT_ALIGNED_DECL_LOCAL (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
5410 Like @code{ASM_OUTPUT_ALIGNED_DECL} except that @var{decl} of the
5411 variable to be output, if there is one, or @code{NULL_TREE} if there
5412 is no corresponding variable. If you define this macro, GCC will use it
5413 in place of both @code{ASM_OUTPUT_DECL} and
5414 @code{ASM_OUTPUT_ALIGNED_DECL}. Define this macro when you need to see
5415 the variable's decl in order to chose what to output.
5419 @subsection Output and Generation of Labels
5421 @c prevent bad page break with this line
5422 This is about outputting labels.
5424 @findex assemble_name
5425 @defmac ASM_OUTPUT_LABEL (@var{stream}, @var{name})
5426 A C statement (sans semicolon) to output to the stdio stream
5427 @var{stream} the assembler definition of a label named @var{name}.
5428 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
5429 output the name itself; before and after that, output the additional
5430 assembler syntax for defining the name, and a newline. A default
5431 definition of this macro is provided which is correct for most systems.
5434 @defmac ASM_OUTPUT_FUNCTION_LABEL (@var{stream}, @var{name}, @var{decl})
5435 A C statement (sans semicolon) to output to the stdio stream
5436 @var{stream} the assembler definition of a label named @var{name} of
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.
5443 If this macro is not defined, then the function name is defined in the
5444 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
5447 @findex assemble_name_raw
5448 @defmac ASM_OUTPUT_INTERNAL_LABEL (@var{stream}, @var{name})
5449 Identical to @code{ASM_OUTPUT_LABEL}, except that @var{name} is known
5450 to refer to a compiler-generated label. The default definition uses
5451 @code{assemble_name_raw}, which is like @code{assemble_name} except
5452 that it is more efficient.
5456 A C string containing the appropriate assembler directive to specify the
5457 size of a symbol, without any arguments. On systems that use ELF, the
5458 default (in @file{config/elfos.h}) is @samp{"\t.size\t"}; on other
5459 systems, the default is not to define this macro.
5461 Define this macro only if it is correct to use the default definitions
5462 of @code{ASM_OUTPUT_SIZE_DIRECTIVE} and @code{ASM_OUTPUT_MEASURED_SIZE}
5463 for your system. If you need your own custom definitions of those
5464 macros, or if you do not need explicit symbol sizes at all, do not
5468 @defmac ASM_OUTPUT_SIZE_DIRECTIVE (@var{stream}, @var{name}, @var{size})
5469 A C statement (sans semicolon) to output to the stdio stream
5470 @var{stream} a directive telling the assembler that the size of the
5471 symbol @var{name} is @var{size}. @var{size} is a @code{HOST_WIDE_INT}.
5472 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
5476 @defmac ASM_OUTPUT_MEASURED_SIZE (@var{stream}, @var{name})
5477 A C statement (sans semicolon) to output to the stdio stream
5478 @var{stream} a directive telling the assembler to calculate the size of
5479 the symbol @var{name} by subtracting its address from the current
5482 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
5483 provided. The default assumes that the assembler recognizes a special
5484 @samp{.} symbol as referring to the current address, and can calculate
5485 the difference between this and another symbol. If your assembler does
5486 not recognize @samp{.} or cannot do calculations with it, you will need
5487 to redefine @code{ASM_OUTPUT_MEASURED_SIZE} to use some other technique.
5490 @defmac NO_DOLLAR_IN_LABEL
5491 Define this macro if the assembler does not accept the character
5492 @samp{$} in label names. By default constructors and destructors in
5493 G++ have @samp{$} in the identifiers. If this macro is defined,
5494 @samp{.} is used instead.
5497 @defmac NO_DOT_IN_LABEL
5498 Define this macro if the assembler does not accept the character
5499 @samp{.} in label names. By default constructors and destructors in G++
5500 have names that use @samp{.}. If this macro is defined, these names
5501 are rewritten to avoid @samp{.}.
5505 A C string containing the appropriate assembler directive to specify the
5506 type of a symbol, without any arguments. On systems that use ELF, the
5507 default (in @file{config/elfos.h}) is @samp{"\t.type\t"}; on other
5508 systems, the default is not to define this macro.
5510 Define this macro only if it is correct to use the default definition of
5511 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
5512 custom definition of this macro, or if you do not need explicit symbol
5513 types at all, do not define this macro.
5516 @defmac TYPE_OPERAND_FMT
5517 A C string which specifies (using @code{printf} syntax) the format of
5518 the second operand to @code{TYPE_ASM_OP}. On systems that use ELF, the
5519 default (in @file{config/elfos.h}) is @samp{"@@%s"}; on other systems,
5520 the default is not to define this macro.
5522 Define this macro only if it is correct to use the default definition of
5523 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
5524 custom definition of this macro, or if you do not need explicit symbol
5525 types at all, do not define this macro.
5528 @defmac ASM_OUTPUT_TYPE_DIRECTIVE (@var{stream}, @var{type})
5529 A C statement (sans semicolon) to output to the stdio stream
5530 @var{stream} a directive telling the assembler that the type of the
5531 symbol @var{name} is @var{type}. @var{type} is a C string; currently,
5532 that string is always either @samp{"function"} or @samp{"object"}, but
5533 you should not count on this.
5535 If you define @code{TYPE_ASM_OP} and @code{TYPE_OPERAND_FMT}, a default
5536 definition of this macro is provided.
5539 @defmac ASM_DECLARE_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl})
5540 A C statement (sans semicolon) to output to the stdio stream
5541 @var{stream} any text necessary for declaring the name @var{name} of a
5542 function which is being defined. This macro is responsible for
5543 outputting the label definition (perhaps using
5544 @code{ASM_OUTPUT_FUNCTION_LABEL}). The argument @var{decl} is the
5545 @code{FUNCTION_DECL} tree node representing the function.
5547 If this macro is not defined, then the function name is defined in the
5548 usual manner as a label (by means of @code{ASM_OUTPUT_FUNCTION_LABEL}).
5550 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition
5554 @defmac ASM_DECLARE_FUNCTION_SIZE (@var{stream}, @var{name}, @var{decl})
5555 A C statement (sans semicolon) to output to the stdio stream
5556 @var{stream} any text necessary for declaring the size of a function
5557 which is being defined. The argument @var{name} is the name of the
5558 function. The argument @var{decl} is the @code{FUNCTION_DECL} tree node
5559 representing the function.
5561 If this macro is not defined, then the function size is not defined.
5563 You may wish to use @code{ASM_OUTPUT_MEASURED_SIZE} in the definition
5567 @defmac ASM_DECLARE_COLD_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl})
5568 A C statement (sans semicolon) to output to the stdio stream
5569 @var{stream} any text necessary for declaring the name @var{name} of a
5570 cold function partition which is being defined. This macro is responsible
5571 for outputting the label definition (perhaps using
5572 @code{ASM_OUTPUT_FUNCTION_LABEL}). The argument @var{decl} is the
5573 @code{FUNCTION_DECL} tree node representing the function.
5575 If this macro is not defined, then the cold partition name is defined in the
5576 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
5578 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition
5582 @defmac ASM_DECLARE_COLD_FUNCTION_SIZE (@var{stream}, @var{name}, @var{decl})
5583 A C statement (sans semicolon) to output to the stdio stream
5584 @var{stream} any text necessary for declaring the size of a cold function
5585 partition which is being defined. The argument @var{name} is the name of the
5586 cold partition of the function. The argument @var{decl} is the
5587 @code{FUNCTION_DECL} tree node representing the function.
5589 If this macro is not defined, then the partition size is not defined.
5591 You may wish to use @code{ASM_OUTPUT_MEASURED_SIZE} in the definition
5595 @defmac ASM_DECLARE_OBJECT_NAME (@var{stream}, @var{name}, @var{decl})
5596 A C statement (sans semicolon) to output to the stdio stream
5597 @var{stream} any text necessary for declaring the name @var{name} of an
5598 initialized variable which is being defined. This macro must output the
5599 label definition (perhaps using @code{ASM_OUTPUT_LABEL}). The argument
5600 @var{decl} is the @code{VAR_DECL} tree node representing the variable.
5602 If this macro is not defined, then the variable name is defined in the
5603 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
5605 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} and/or
5606 @code{ASM_OUTPUT_SIZE_DIRECTIVE} in the definition of this macro.
5609 @hook TARGET_ASM_DECLARE_CONSTANT_NAME
5611 @defmac ASM_DECLARE_REGISTER_GLOBAL (@var{stream}, @var{decl}, @var{regno}, @var{name})
5612 A C statement (sans semicolon) to output to the stdio stream
5613 @var{stream} any text necessary for claiming a register @var{regno}
5614 for a global variable @var{decl} with name @var{name}.
5616 If you don't define this macro, that is equivalent to defining it to do
5620 @defmac ASM_FINISH_DECLARE_OBJECT (@var{stream}, @var{decl}, @var{toplevel}, @var{atend})
5621 A C statement (sans semicolon) to finish up declaring a variable name
5622 once the compiler has processed its initializer fully and thus has had a
5623 chance to determine the size of an array when controlled by an
5624 initializer. This is used on systems where it's necessary to declare
5625 something about the size of the object.
5627 If you don't define this macro, that is equivalent to defining it to do
5630 You may wish to use @code{ASM_OUTPUT_SIZE_DIRECTIVE} and/or
5631 @code{ASM_OUTPUT_MEASURED_SIZE} in the definition of this macro.
5634 @hook TARGET_ASM_GLOBALIZE_LABEL
5636 @hook TARGET_ASM_GLOBALIZE_DECL_NAME
5638 @hook TARGET_ASM_ASSEMBLE_UNDEFINED_DECL
5640 @defmac ASM_WEAKEN_LABEL (@var{stream}, @var{name})
5641 A C statement (sans semicolon) to output to the stdio stream
5642 @var{stream} some commands that will make the label @var{name} weak;
5643 that is, available for reference from other files but only used if
5644 no other definition is available. Use the expression
5645 @code{assemble_name (@var{stream}, @var{name})} to output the name
5646 itself; before and after that, output the additional assembler syntax
5647 for making that name weak, and a newline.
5649 If you don't define this macro or @code{ASM_WEAKEN_DECL}, GCC will not
5650 support weak symbols and you should not define the @code{SUPPORTS_WEAK}
5654 @defmac ASM_WEAKEN_DECL (@var{stream}, @var{decl}, @var{name}, @var{value})
5655 Combines (and replaces) the function of @code{ASM_WEAKEN_LABEL} and
5656 @code{ASM_OUTPUT_WEAK_ALIAS}, allowing access to the associated function
5657 or variable decl. If @var{value} is not @code{NULL}, this C statement
5658 should output to the stdio stream @var{stream} assembler code which
5659 defines (equates) the weak symbol @var{name} to have the value
5660 @var{value}. If @var{value} is @code{NULL}, it should output commands
5661 to make @var{name} weak.
5664 @defmac ASM_OUTPUT_WEAKREF (@var{stream}, @var{decl}, @var{name}, @var{value})
5665 Outputs a directive that enables @var{name} to be used to refer to
5666 symbol @var{value} with weak-symbol semantics. @code{decl} is the
5667 declaration of @code{name}.
5670 @defmac SUPPORTS_WEAK
5671 A preprocessor constant expression which evaluates to true if the target
5672 supports weak symbols.
5674 If you don't define this macro, @file{defaults.h} provides a default
5675 definition. If either @code{ASM_WEAKEN_LABEL} or @code{ASM_WEAKEN_DECL}
5676 is defined, the default definition is @samp{1}; otherwise, it is @samp{0}.
5679 @defmac TARGET_SUPPORTS_WEAK
5680 A C expression which evaluates to true if the target supports weak symbols.
5682 If you don't define this macro, @file{defaults.h} provides a default
5683 definition. The default definition is @samp{(SUPPORTS_WEAK)}. Define
5684 this macro if you want to control weak symbol support with a compiler
5685 flag such as @option{-melf}.
5688 @defmac MAKE_DECL_ONE_ONLY (@var{decl})
5689 A C statement (sans semicolon) to mark @var{decl} to be emitted as a
5690 public symbol such that extra copies in multiple translation units will
5691 be discarded by the linker. Define this macro if your object file
5692 format provides support for this concept, such as the @samp{COMDAT}
5693 section flags in the Microsoft Windows PE/COFF format, and this support
5694 requires changes to @var{decl}, such as putting it in a separate section.
5697 @defmac SUPPORTS_ONE_ONLY
5698 A C expression which evaluates to true if the target supports one-only
5701 If you don't define this macro, @file{varasm.c} provides a default
5702 definition. If @code{MAKE_DECL_ONE_ONLY} is defined, the default
5703 definition is @samp{1}; otherwise, it is @samp{0}. Define this macro if
5704 you want to control one-only symbol support with a compiler flag, or if
5705 setting the @code{DECL_ONE_ONLY} flag is enough to mark a declaration to
5706 be emitted as one-only.
5709 @hook TARGET_ASM_ASSEMBLE_VISIBILITY
5711 @defmac TARGET_WEAK_NOT_IN_ARCHIVE_TOC
5712 A C expression that evaluates to true if the target's linker expects
5713 that weak symbols do not appear in a static archive's table of contents.
5714 The default is @code{0}.
5716 Leaving weak symbols out of an archive's table of contents means that,
5717 if a symbol will only have a definition in one translation unit and
5718 will have undefined references from other translation units, that
5719 symbol should not be weak. Defining this macro to be nonzero will
5720 thus have the effect that certain symbols that would normally be weak
5721 (explicit template instantiations, and vtables for polymorphic classes
5722 with noninline key methods) will instead be nonweak.
5724 The C++ ABI requires this macro to be zero. Define this macro for
5725 targets where full C++ ABI compliance is impossible and where linker
5726 restrictions require weak symbols to be left out of a static archive's
5730 @defmac ASM_OUTPUT_EXTERNAL (@var{stream}, @var{decl}, @var{name})
5731 A C statement (sans semicolon) to output to the stdio stream
5732 @var{stream} any text necessary for declaring the name of an external
5733 symbol named @var{name} which is referenced in this compilation but
5734 not defined. The value of @var{decl} is the tree node for the
5737 This macro need not be defined if it does not need to output anything.
5738 The GNU assembler and most Unix assemblers don't require anything.
5741 @hook TARGET_ASM_EXTERNAL_LIBCALL
5743 @hook TARGET_ASM_MARK_DECL_PRESERVED
5745 @defmac ASM_OUTPUT_LABELREF (@var{stream}, @var{name})
5746 A C statement (sans semicolon) to output to the stdio stream
5747 @var{stream} a reference in assembler syntax to a label named
5748 @var{name}. This should add @samp{_} to the front of the name, if that
5749 is customary on your operating system, as it is in most Berkeley Unix
5750 systems. This macro is used in @code{assemble_name}.
5753 @hook TARGET_MANGLE_ASSEMBLER_NAME
5755 @defmac ASM_OUTPUT_SYMBOL_REF (@var{stream}, @var{sym})
5756 A C statement (sans semicolon) to output a reference to
5757 @code{SYMBOL_REF} @var{sym}. If not defined, @code{assemble_name}
5758 will be used to output the name of the symbol. This macro may be used
5759 to modify the way a symbol is referenced depending on information
5760 encoded by @code{TARGET_ENCODE_SECTION_INFO}.
5763 @defmac ASM_OUTPUT_LABEL_REF (@var{stream}, @var{buf})
5764 A C statement (sans semicolon) to output a reference to @var{buf}, the
5765 result of @code{ASM_GENERATE_INTERNAL_LABEL}. If not defined,
5766 @code{assemble_name} will be used to output the name of the symbol.
5767 This macro is not used by @code{output_asm_label}, or the @code{%l}
5768 specifier that calls it; the intention is that this macro should be set
5769 when it is necessary to output a label differently when its address is
5773 @hook TARGET_ASM_INTERNAL_LABEL
5775 @defmac ASM_OUTPUT_DEBUG_LABEL (@var{stream}, @var{prefix}, @var{num})
5776 A C statement to output to the stdio stream @var{stream} a debug info
5777 label whose name is made from the string @var{prefix} and the number
5778 @var{num}. This is useful for VLIW targets, where debug info labels
5779 may need to be treated differently than branch target labels. On some
5780 systems, branch target labels must be at the beginning of instruction
5781 bundles, but debug info labels can occur in the middle of instruction
5784 If this macro is not defined, then @code{(*targetm.asm_out.internal_label)} will be
5788 @defmac ASM_GENERATE_INTERNAL_LABEL (@var{string}, @var{prefix}, @var{num})
5789 A C statement to store into the string @var{string} a label whose name
5790 is made from the string @var{prefix} and the number @var{num}.
5792 This string, when output subsequently by @code{assemble_name}, should
5793 produce the output that @code{(*targetm.asm_out.internal_label)} would produce
5794 with the same @var{prefix} and @var{num}.
5796 If the string begins with @samp{*}, then @code{assemble_name} will
5797 output the rest of the string unchanged. It is often convenient for
5798 @code{ASM_GENERATE_INTERNAL_LABEL} to use @samp{*} in this way. If the
5799 string doesn't start with @samp{*}, then @code{ASM_OUTPUT_LABELREF} gets
5800 to output the string, and may change it. (Of course,
5801 @code{ASM_OUTPUT_LABELREF} is also part of your machine description, so
5802 you should know what it does on your machine.)
5805 @defmac ASM_FORMAT_PRIVATE_NAME (@var{outvar}, @var{name}, @var{number})
5806 A C expression to assign to @var{outvar} (which is a variable of type
5807 @code{char *}) a newly allocated string made from the string
5808 @var{name} and the number @var{number}, with some suitable punctuation
5809 added. Use @code{alloca} to get space for the string.
5811 The string will be used as an argument to @code{ASM_OUTPUT_LABELREF} to
5812 produce an assembler label for an internal static variable whose name is
5813 @var{name}. Therefore, the string must be such as to result in valid
5814 assembler code. The argument @var{number} is different each time this
5815 macro is executed; it prevents conflicts between similarly-named
5816 internal static variables in different scopes.
5818 Ideally this string should not be a valid C identifier, to prevent any
5819 conflict with the user's own symbols. Most assemblers allow periods
5820 or percent signs in assembler symbols; putting at least one of these
5821 between the name and the number will suffice.
5823 If this macro is not defined, a default definition will be provided
5824 which is correct for most systems.
5827 @defmac ASM_OUTPUT_DEF (@var{stream}, @var{name}, @var{value})
5828 A C statement to output to the stdio stream @var{stream} assembler code
5829 which defines (equates) the symbol @var{name} to have the value @var{value}.
5832 If @code{SET_ASM_OP} is defined, a default definition is provided which is
5833 correct for most systems.
5836 @defmac ASM_OUTPUT_DEF_FROM_DECLS (@var{stream}, @var{decl_of_name}, @var{decl_of_value})
5837 A C statement to output to the stdio stream @var{stream} assembler code
5838 which defines (equates) the symbol whose tree node is @var{decl_of_name}
5839 to have the value of the tree node @var{decl_of_value}. This macro will
5840 be used in preference to @samp{ASM_OUTPUT_DEF} if it is defined and if
5841 the tree nodes are available.
5844 If @code{SET_ASM_OP} is defined, a default definition is provided which is
5845 correct for most systems.
5848 @defmac TARGET_DEFERRED_OUTPUT_DEFS (@var{decl_of_name}, @var{decl_of_value})
5849 A C statement that evaluates to true if the assembler code which defines
5850 (equates) the symbol whose tree node is @var{decl_of_name} to have the value
5851 of the tree node @var{decl_of_value} should be emitted near the end of the
5852 current compilation unit. The default is to not defer output of defines.
5853 This macro affects defines output by @samp{ASM_OUTPUT_DEF} and
5854 @samp{ASM_OUTPUT_DEF_FROM_DECLS}.
5857 @defmac ASM_OUTPUT_WEAK_ALIAS (@var{stream}, @var{name}, @var{value})
5858 A C statement to output to the stdio stream @var{stream} assembler code
5859 which defines (equates) the weak symbol @var{name} to have the value
5860 @var{value}. If @var{value} is @code{NULL}, it defines @var{name} as
5861 an undefined weak symbol.
5863 Define this macro if the target only supports weak aliases; define
5864 @code{ASM_OUTPUT_DEF} instead if possible.
5867 @defmac OBJC_GEN_METHOD_LABEL (@var{buf}, @var{is_inst}, @var{class_name}, @var{cat_name}, @var{sel_name})
5868 Define this macro to override the default assembler names used for
5869 Objective-C methods.
5871 The default name is a unique method number followed by the name of the
5872 class (e.g.@: @samp{_1_Foo}). For methods in categories, the name of
5873 the category is also included in the assembler name (e.g.@:
5876 These names are safe on most systems, but make debugging difficult since
5877 the method's selector is not present in the name. Therefore, particular
5878 systems define other ways of computing names.
5880 @var{buf} is an expression of type @code{char *} which gives you a
5881 buffer in which to store the name; its length is as long as
5882 @var{class_name}, @var{cat_name} and @var{sel_name} put together, plus
5883 50 characters extra.
5885 The argument @var{is_inst} specifies whether the method is an instance
5886 method or a class method; @var{class_name} is the name of the class;
5887 @var{cat_name} is the name of the category (or @code{NULL} if the method is not
5888 in a category); and @var{sel_name} is the name of the selector.
5890 On systems where the assembler can handle quoted names, you can use this
5891 macro to provide more human-readable names.
5894 @node Initialization
5895 @subsection How Initialization Functions Are Handled
5896 @cindex initialization routines
5897 @cindex termination routines
5898 @cindex constructors, output of
5899 @cindex destructors, output of
5901 The compiled code for certain languages includes @dfn{constructors}
5902 (also called @dfn{initialization routines})---functions to initialize
5903 data in the program when the program is started. These functions need
5904 to be called before the program is ``started''---that is to say, before
5905 @code{main} is called.
5907 Compiling some languages generates @dfn{destructors} (also called
5908 @dfn{termination routines}) that should be called when the program
5911 To make the initialization and termination functions work, the compiler
5912 must output something in the assembler code to cause those functions to
5913 be called at the appropriate time. When you port the compiler to a new
5914 system, you need to specify how to do this.
5916 There are two major ways that GCC currently supports the execution of
5917 initialization and termination functions. Each way has two variants.
5918 Much of the structure is common to all four variations.
5920 @findex __CTOR_LIST__
5921 @findex __DTOR_LIST__
5922 The linker must build two lists of these functions---a list of
5923 initialization functions, called @code{__CTOR_LIST__}, and a list of
5924 termination functions, called @code{__DTOR_LIST__}.
5926 Each list always begins with an ignored function pointer (which may hold
5927 0, @minus{}1, or a count of the function pointers after it, depending on
5928 the environment). This is followed by a series of zero or more function
5929 pointers to constructors (or destructors), followed by a function
5930 pointer containing zero.
5932 Depending on the operating system and its executable file format, either
5933 @file{crtstuff.c} or @file{libgcc2.c} traverses these lists at startup
5934 time and exit time. Constructors are called in reverse order of the
5935 list; destructors in forward order.
5937 The best way to handle static constructors works only for object file
5938 formats which provide arbitrarily-named sections. A section is set
5939 aside for a list of constructors, and another for a list of destructors.
5940 Traditionally these are called @samp{.ctors} and @samp{.dtors}. Each
5941 object file that defines an initialization function also puts a word in
5942 the constructor section to point to that function. The linker
5943 accumulates all these words into one contiguous @samp{.ctors} section.
5944 Termination functions are handled similarly.
5946 This method will be chosen as the default by @file{target-def.h} if
5947 @code{TARGET_ASM_NAMED_SECTION} is defined. A target that does not
5948 support arbitrary sections, but does support special designated
5949 constructor and destructor sections may define @code{CTORS_SECTION_ASM_OP}
5950 and @code{DTORS_SECTION_ASM_OP} to achieve the same effect.
5952 When arbitrary sections are available, there are two variants, depending
5953 upon how the code in @file{crtstuff.c} is called. On systems that
5954 support a @dfn{.init} section which is executed at program startup,
5955 parts of @file{crtstuff.c} are compiled into that section. The
5956 program is linked by the @command{gcc} driver like this:
5959 ld -o @var{output_file} crti.o crtbegin.o @dots{} -lgcc crtend.o crtn.o
5962 The prologue of a function (@code{__init}) appears in the @code{.init}
5963 section of @file{crti.o}; the epilogue appears in @file{crtn.o}. Likewise
5964 for the function @code{__fini} in the @dfn{.fini} section. Normally these
5965 files are provided by the operating system or by the GNU C library, but
5966 are provided by GCC for a few targets.
5968 The objects @file{crtbegin.o} and @file{crtend.o} are (for most targets)
5969 compiled from @file{crtstuff.c}. They contain, among other things, code
5970 fragments within the @code{.init} and @code{.fini} sections that branch
5971 to routines in the @code{.text} section. The linker will pull all parts
5972 of a section together, which results in a complete @code{__init} function
5973 that invokes the routines we need at startup.
5975 To use this variant, you must define the @code{INIT_SECTION_ASM_OP}
5978 If no init section is available, when GCC compiles any function called
5979 @code{main} (or more accurately, any function designated as a program
5980 entry point by the language front end calling @code{expand_main_function}),
5981 it inserts a procedure call to @code{__main} as the first executable code
5982 after the function prologue. The @code{__main} function is defined
5983 in @file{libgcc2.c} and runs the global constructors.
5985 In file formats that don't support arbitrary sections, there are again
5986 two variants. In the simplest variant, the GNU linker (GNU @code{ld})
5987 and an `a.out' format must be used. In this case,
5988 @code{TARGET_ASM_CONSTRUCTOR} is defined to produce a @code{.stabs}
5989 entry of type @samp{N_SETT}, referencing the name @code{__CTOR_LIST__},
5990 and with the address of the void function containing the initialization
5991 code as its value. The GNU linker recognizes this as a request to add
5992 the value to a @dfn{set}; the values are accumulated, and are eventually
5993 placed in the executable as a vector in the format described above, with
5994 a leading (ignored) count and a trailing zero element.
5995 @code{TARGET_ASM_DESTRUCTOR} is handled similarly. Since no init
5996 section is available, the absence of @code{INIT_SECTION_ASM_OP} causes
5997 the compilation of @code{main} to call @code{__main} as above, starting
5998 the initialization process.
6000 The last variant uses neither arbitrary sections nor the GNU linker.
6001 This is preferable when you want to do dynamic linking and when using
6002 file formats which the GNU linker does not support, such as `ECOFF'@. In
6003 this case, @code{TARGET_HAVE_CTORS_DTORS} is false, initialization and
6004 termination functions are recognized simply by their names. This requires
6005 an extra program in the linkage step, called @command{collect2}. This program
6006 pretends to be the linker, for use with GCC; it does its job by running
6007 the ordinary linker, but also arranges to include the vectors of
6008 initialization and termination functions. These functions are called
6009 via @code{__main} as described above. In order to use this method,
6010 @code{use_collect2} must be defined in the target in @file{config.gcc}.
6013 The following section describes the specific macros that control and
6014 customize the handling of initialization and termination functions.
6017 @node Macros for Initialization
6018 @subsection Macros Controlling Initialization Routines
6020 Here are the macros that control how the compiler handles initialization
6021 and termination functions:
6023 @defmac INIT_SECTION_ASM_OP
6024 If defined, a C string constant, including spacing, for the assembler
6025 operation to identify the following data as initialization code. If not
6026 defined, GCC will assume such a section does not exist. When you are
6027 using special sections for initialization and termination functions, this
6028 macro also controls how @file{crtstuff.c} and @file{libgcc2.c} arrange to
6029 run the initialization functions.
6032 @defmac HAS_INIT_SECTION
6033 If defined, @code{main} will not call @code{__main} as described above.
6034 This macro should be defined for systems that control start-up code
6035 on a symbol-by-symbol basis, such as OSF/1, and should not
6036 be defined explicitly for systems that support @code{INIT_SECTION_ASM_OP}.
6039 @defmac LD_INIT_SWITCH
6040 If defined, a C string constant for a switch that tells the linker that
6041 the following symbol is an initialization routine.
6044 @defmac LD_FINI_SWITCH
6045 If defined, a C string constant for a switch that tells the linker that
6046 the following symbol is a finalization routine.
6049 @defmac COLLECT_SHARED_INIT_FUNC (@var{stream}, @var{func})
6050 If defined, a C statement that will write a function that can be
6051 automatically called when a shared library is loaded. The function
6052 should call @var{func}, which takes no arguments. If not defined, and
6053 the object format requires an explicit initialization function, then a
6054 function called @code{_GLOBAL__DI} will be generated.
6056 This function and the following one are used by collect2 when linking a
6057 shared library that needs constructors or destructors, or has DWARF2
6058 exception tables embedded in the code.
6061 @defmac COLLECT_SHARED_FINI_FUNC (@var{stream}, @var{func})
6062 If defined, a C statement that will write a function that can be
6063 automatically called when a shared library is unloaded. The function
6064 should call @var{func}, which takes no arguments. If not defined, and
6065 the object format requires an explicit finalization function, then a
6066 function called @code{_GLOBAL__DD} will be generated.
6069 @defmac INVOKE__main
6070 If defined, @code{main} will call @code{__main} despite the presence of
6071 @code{INIT_SECTION_ASM_OP}. This macro should be defined for systems
6072 where the init section is not actually run automatically, but is still
6073 useful for collecting the lists of constructors and destructors.
6076 @defmac SUPPORTS_INIT_PRIORITY
6077 If nonzero, the C++ @code{init_priority} attribute is supported and the
6078 compiler should emit instructions to control the order of initialization
6079 of objects. If zero, the compiler will issue an error message upon
6080 encountering an @code{init_priority} attribute.
6083 @hook TARGET_HAVE_CTORS_DTORS
6085 @hook TARGET_ASM_CONSTRUCTOR
6087 @hook TARGET_ASM_DESTRUCTOR
6089 If @code{TARGET_HAVE_CTORS_DTORS} is true, the initialization routine
6090 generated for the generated object file will have static linkage.
6092 If your system uses @command{collect2} as the means of processing
6093 constructors, then that program normally uses @command{nm} to scan
6094 an object file for constructor functions to be called.
6096 On certain kinds of systems, you can define this macro to make
6097 @command{collect2} work faster (and, in some cases, make it work at all):
6099 @defmac OBJECT_FORMAT_COFF
6100 Define this macro if the system uses COFF (Common Object File Format)
6101 object files, so that @command{collect2} can assume this format and scan
6102 object files directly for dynamic constructor/destructor functions.
6104 This macro is effective only in a native compiler; @command{collect2} as
6105 part of a cross compiler always uses @command{nm} for the target machine.
6108 @defmac REAL_NM_FILE_NAME
6109 Define this macro as a C string constant containing the file name to use
6110 to execute @command{nm}. The default is to search the path normally for
6115 @command{collect2} calls @command{nm} to scan object files for static
6116 constructors and destructors and LTO info. By default, @option{-n} is
6117 passed. Define @code{NM_FLAGS} to a C string constant if other options
6118 are needed to get the same output format as GNU @command{nm -n}
6122 If your system supports shared libraries and has a program to list the
6123 dynamic dependencies of a given library or executable, you can define
6124 these macros to enable support for running initialization and
6125 termination functions in shared libraries:
6128 Define this macro to a C string constant containing the name of the program
6129 which lists dynamic dependencies, like @command{ldd} under SunOS 4.
6132 @defmac PARSE_LDD_OUTPUT (@var{ptr})
6133 Define this macro to be C code that extracts filenames from the output
6134 of the program denoted by @code{LDD_SUFFIX}. @var{ptr} is a variable
6135 of type @code{char *} that points to the beginning of a line of output
6136 from @code{LDD_SUFFIX}. If the line lists a dynamic dependency, the
6137 code must advance @var{ptr} to the beginning of the filename on that
6138 line. Otherwise, it must set @var{ptr} to @code{NULL}.
6141 @defmac SHLIB_SUFFIX
6142 Define this macro to a C string constant containing the default shared
6143 library extension of the target (e.g., @samp{".so"}). @command{collect2}
6144 strips version information after this suffix when generating global
6145 constructor and destructor names. This define is only needed on targets
6146 that use @command{collect2} to process constructors and destructors.
6149 @node Instruction Output
6150 @subsection Output of Assembler Instructions
6152 @c prevent bad page break with this line
6153 This describes assembler instruction output.
6155 @defmac REGISTER_NAMES
6156 A C initializer containing the assembler's names for the machine
6157 registers, each one as a C string constant. This is what translates
6158 register numbers in the compiler into assembler language.
6161 @defmac ADDITIONAL_REGISTER_NAMES
6162 If defined, a C initializer for an array of structures containing a name
6163 and a register number. This macro defines additional names for hard
6164 registers, thus allowing the @code{asm} option in declarations to refer
6165 to registers using alternate names.
6168 @defmac OVERLAPPING_REGISTER_NAMES
6169 If defined, a C initializer for an array of structures containing a
6170 name, a register number and a count of the number of consecutive
6171 machine registers the name overlaps. This macro defines additional
6172 names for hard registers, thus allowing the @code{asm} option in
6173 declarations to refer to registers using alternate names. Unlike
6174 @code{ADDITIONAL_REGISTER_NAMES}, this macro should be used when the
6175 register name implies multiple underlying registers.
6177 This macro should be used when it is important that a clobber in an
6178 @code{asm} statement clobbers all the underlying values implied by the
6179 register name. For example, on ARM, clobbering the double-precision
6180 VFP register ``d0'' implies clobbering both single-precision registers
6184 @defmac ASM_OUTPUT_OPCODE (@var{stream}, @var{ptr})
6185 Define this macro if you are using an unusual assembler that
6186 requires different names for the machine instructions.
6188 The definition is a C statement or statements which output an
6189 assembler instruction opcode to the stdio stream @var{stream}. The
6190 macro-operand @var{ptr} is a variable of type @code{char *} which
6191 points to the opcode name in its ``internal'' form---the form that is
6192 written in the machine description. The definition should output the
6193 opcode name to @var{stream}, performing any translation you desire, and
6194 increment the variable @var{ptr} to point at the end of the opcode
6195 so that it will not be output twice.
6197 In fact, your macro definition may process less than the entire opcode
6198 name, or more than the opcode name; but if you want to process text
6199 that includes @samp{%}-sequences to substitute operands, you must take
6200 care of the substitution yourself. Just be sure to increment
6201 @var{ptr} over whatever text should not be output normally.
6203 @findex recog_data.operand
6204 If you need to look at the operand values, they can be found as the
6205 elements of @code{recog_data.operand}.
6207 If the macro definition does nothing, the instruction is output
6211 @defmac FINAL_PRESCAN_INSN (@var{insn}, @var{opvec}, @var{noperands})
6212 If defined, a C statement to be executed just prior to the output of
6213 assembler code for @var{insn}, to modify the extracted operands so
6214 they will be output differently.
6216 Here the argument @var{opvec} is the vector containing the operands
6217 extracted from @var{insn}, and @var{noperands} is the number of
6218 elements of the vector which contain meaningful data for this insn.
6219 The contents of this vector are what will be used to convert the insn
6220 template into assembler code, so you can change the assembler output
6221 by changing the contents of the vector.
6223 This macro is useful when various assembler syntaxes share a single
6224 file of instruction patterns; by defining this macro differently, you
6225 can cause a large class of instructions to be output differently (such
6226 as with rearranged operands). Naturally, variations in assembler
6227 syntax affecting individual insn patterns ought to be handled by
6228 writing conditional output routines in those patterns.
6230 If this macro is not defined, it is equivalent to a null statement.
6233 @hook TARGET_ASM_FINAL_POSTSCAN_INSN
6235 @defmac PRINT_OPERAND (@var{stream}, @var{x}, @var{code})
6236 A C compound statement to output to stdio stream @var{stream} the
6237 assembler syntax for an instruction operand @var{x}. @var{x} is an
6240 @var{code} is a value that can be used to specify one of several ways
6241 of printing the operand. It is used when identical operands must be
6242 printed differently depending on the context. @var{code} comes from
6243 the @samp{%} specification that was used to request printing of the
6244 operand. If the specification was just @samp{%@var{digit}} then
6245 @var{code} is 0; if the specification was @samp{%@var{ltr}
6246 @var{digit}} then @var{code} is the ASCII code for @var{ltr}.
6249 If @var{x} is a register, this macro should print the register's name.
6250 The names can be found in an array @code{reg_names} whose type is
6251 @code{char *[]}. @code{reg_names} is initialized from
6252 @code{REGISTER_NAMES}.
6254 When the machine description has a specification @samp{%@var{punct}}
6255 (a @samp{%} followed by a punctuation character), this macro is called
6256 with a null pointer for @var{x} and the punctuation character for
6260 @defmac PRINT_OPERAND_PUNCT_VALID_P (@var{code})
6261 A C expression which evaluates to true if @var{code} is a valid
6262 punctuation character for use in the @code{PRINT_OPERAND} macro. If
6263 @code{PRINT_OPERAND_PUNCT_VALID_P} is not defined, it means that no
6264 punctuation characters (except for the standard one, @samp{%}) are used
6268 @defmac PRINT_OPERAND_ADDRESS (@var{stream}, @var{x})
6269 A C compound statement to output to stdio stream @var{stream} the
6270 assembler syntax for an instruction operand that is a memory reference
6271 whose address is @var{x}. @var{x} is an RTL expression.
6273 @cindex @code{TARGET_ENCODE_SECTION_INFO} usage
6274 On some machines, the syntax for a symbolic address depends on the
6275 section that the address refers to. On these machines, define the hook
6276 @code{TARGET_ENCODE_SECTION_INFO} to store the information into the
6277 @code{symbol_ref}, and then check for it here. @xref{Assembler
6281 @findex dbr_sequence_length
6282 @defmac DBR_OUTPUT_SEQEND (@var{file})
6283 A C statement, to be executed after all slot-filler instructions have
6284 been output. If necessary, call @code{dbr_sequence_length} to
6285 determine the number of slots filled in a sequence (zero if not
6286 currently outputting a sequence), to decide how many no-ops to output,
6289 Don't define this macro if it has nothing to do, but it is helpful in
6290 reading assembly output if the extent of the delay sequence is made
6291 explicit (e.g.@: with white space).
6294 @findex final_sequence
6295 Note that output routines for instructions with delay slots must be
6296 prepared to deal with not being output as part of a sequence
6297 (i.e.@: when the scheduling pass is not run, or when no slot fillers could be
6298 found.) The variable @code{final_sequence} is null when not
6299 processing a sequence, otherwise it contains the @code{sequence} rtx
6303 @defmac REGISTER_PREFIX
6304 @defmacx LOCAL_LABEL_PREFIX
6305 @defmacx USER_LABEL_PREFIX
6306 @defmacx IMMEDIATE_PREFIX
6307 If defined, C string expressions to be used for the @samp{%R}, @samp{%L},
6308 @samp{%U}, and @samp{%I} options of @code{asm_fprintf} (see
6309 @file{final.c}). These are useful when a single @file{md} file must
6310 support multiple assembler formats. In that case, the various @file{tm.h}
6311 files can define these macros differently.
6314 @defmac ASM_FPRINTF_EXTENSIONS (@var{file}, @var{argptr}, @var{format})
6315 If defined this macro should expand to a series of @code{case}
6316 statements which will be parsed inside the @code{switch} statement of
6317 the @code{asm_fprintf} function. This allows targets to define extra
6318 printf formats which may useful when generating their assembler
6319 statements. Note that uppercase letters are reserved for future
6320 generic extensions to asm_fprintf, and so are not available to target
6321 specific code. The output file is given by the parameter @var{file}.
6322 The varargs input pointer is @var{argptr} and the rest of the format
6323 string, starting the character after the one that is being switched
6324 upon, is pointed to by @var{format}.
6327 @defmac ASSEMBLER_DIALECT
6328 If your target supports multiple dialects of assembler language (such as
6329 different opcodes), define this macro as a C expression that gives the
6330 numeric index of the assembler language dialect to use, with zero as the
6333 If this macro is defined, you may use constructs of the form
6335 @samp{@{option0|option1|option2@dots{}@}}
6338 in the output templates of patterns (@pxref{Output Template}) or in the
6339 first argument of @code{asm_fprintf}. This construct outputs
6340 @samp{option0}, @samp{option1}, @samp{option2}, etc., if the value of
6341 @code{ASSEMBLER_DIALECT} is zero, one, two, etc. Any special characters
6342 within these strings retain their usual meaning. If there are fewer
6343 alternatives within the braces than the value of
6344 @code{ASSEMBLER_DIALECT}, the construct outputs nothing. If it's needed
6345 to print curly braces or @samp{|} character in assembler output directly,
6346 @samp{%@{}, @samp{%@}} and @samp{%|} can be used.
6348 If you do not define this macro, the characters @samp{@{}, @samp{|} and
6349 @samp{@}} do not have any special meaning when used in templates or
6350 operands to @code{asm_fprintf}.
6352 Define the macros @code{REGISTER_PREFIX}, @code{LOCAL_LABEL_PREFIX},
6353 @code{USER_LABEL_PREFIX} and @code{IMMEDIATE_PREFIX} if you can express
6354 the variations in assembler language syntax with that mechanism. Define
6355 @code{ASSEMBLER_DIALECT} and use the @samp{@{option0|option1@}} syntax
6356 if the syntax variant are larger and involve such things as different
6357 opcodes or operand order.
6360 @defmac ASM_OUTPUT_REG_PUSH (@var{stream}, @var{regno})
6361 A C expression to output to @var{stream} some assembler code
6362 which will push hard register number @var{regno} onto the stack.
6363 The code need not be optimal, since this macro is used only when
6367 @defmac ASM_OUTPUT_REG_POP (@var{stream}, @var{regno})
6368 A C expression to output to @var{stream} some assembler code
6369 which will pop hard register number @var{regno} off of the stack.
6370 The code need not be optimal, since this macro is used only when
6374 @node Dispatch Tables
6375 @subsection Output of Dispatch Tables
6377 @c prevent bad page break with this line
6378 This concerns dispatch tables.
6380 @cindex dispatch table
6381 @defmac ASM_OUTPUT_ADDR_DIFF_ELT (@var{stream}, @var{body}, @var{value}, @var{rel})
6382 A C statement to output to the stdio stream @var{stream} an assembler
6383 pseudo-instruction to generate a difference between two labels.
6384 @var{value} and @var{rel} are the numbers of two internal labels. The
6385 definitions of these labels are output using
6386 @code{(*targetm.asm_out.internal_label)}, and they must be printed in the same
6387 way here. For example,
6390 fprintf (@var{stream}, "\t.word L%d-L%d\n",
6391 @var{value}, @var{rel})
6394 You must provide this macro on machines where the addresses in a
6395 dispatch table are relative to the table's own address. If defined, GCC
6396 will also use this macro on all machines when producing PIC@.
6397 @var{body} is the body of the @code{ADDR_DIFF_VEC}; it is provided so that the
6398 mode and flags can be read.
6401 @defmac ASM_OUTPUT_ADDR_VEC_ELT (@var{stream}, @var{value})
6402 This macro should be provided on machines where the addresses
6403 in a dispatch table are absolute.
6405 The definition should be a C statement to output to the stdio stream
6406 @var{stream} an assembler pseudo-instruction to generate a reference to
6407 a label. @var{value} is the number of an internal label whose
6408 definition is output using @code{(*targetm.asm_out.internal_label)}.
6412 fprintf (@var{stream}, "\t.word L%d\n", @var{value})
6416 @defmac ASM_OUTPUT_CASE_LABEL (@var{stream}, @var{prefix}, @var{num}, @var{table})
6417 Define this if the label before a jump-table needs to be output
6418 specially. The first three arguments are the same as for
6419 @code{(*targetm.asm_out.internal_label)}; the fourth argument is the
6420 jump-table which follows (a @code{jump_table_data} containing an
6421 @code{addr_vec} or @code{addr_diff_vec}).
6423 This feature is used on system V to output a @code{swbeg} statement
6426 If this macro is not defined, these labels are output with
6427 @code{(*targetm.asm_out.internal_label)}.
6430 @defmac ASM_OUTPUT_CASE_END (@var{stream}, @var{num}, @var{table})
6431 Define this if something special must be output at the end of a
6432 jump-table. The definition should be a C statement to be executed
6433 after the assembler code for the table is written. It should write
6434 the appropriate code to stdio stream @var{stream}. The argument
6435 @var{table} is the jump-table insn, and @var{num} is the label-number
6436 of the preceding label.
6438 If this macro is not defined, nothing special is output at the end of
6442 @hook TARGET_ASM_POST_CFI_STARTPROC
6444 @hook TARGET_ASM_EMIT_UNWIND_LABEL
6446 @hook TARGET_ASM_EMIT_EXCEPT_TABLE_LABEL
6448 @hook TARGET_ASM_EMIT_EXCEPT_PERSONALITY
6450 @hook TARGET_ASM_UNWIND_EMIT
6452 @hook TARGET_ASM_UNWIND_EMIT_BEFORE_INSN
6454 @node Exception Region Output
6455 @subsection Assembler Commands for Exception Regions
6457 @c prevent bad page break with this line
6459 This describes commands marking the start and the end of an exception
6462 @defmac EH_FRAME_SECTION_NAME
6463 If defined, a C string constant for the name of the section containing
6464 exception handling frame unwind information. If not defined, GCC will
6465 provide a default definition if the target supports named sections.
6466 @file{crtstuff.c} uses this macro to switch to the appropriate section.
6468 You should define this symbol if your target supports DWARF 2 frame
6469 unwind information and the default definition does not work.
6472 @defmac EH_FRAME_THROUGH_COLLECT2
6473 If defined, DWARF 2 frame unwind information will identified by
6474 specially named labels. The collect2 process will locate these
6475 labels and generate code to register the frames.
6477 This might be necessary, for instance, if the system linker will not
6478 place the eh_frames in-between the sentinals from @file{crtstuff.c},
6479 or if the system linker does garbage collection and sections cannot
6480 be marked as not to be collected.
6483 @defmac EH_TABLES_CAN_BE_READ_ONLY
6484 Define this macro to 1 if your target is such that no frame unwind
6485 information encoding used with non-PIC code will ever require a
6486 runtime relocation, but the linker may not support merging read-only
6487 and read-write sections into a single read-write section.
6490 @defmac MASK_RETURN_ADDR
6491 An rtx used to mask the return address found via @code{RETURN_ADDR_RTX}, so
6492 that it does not contain any extraneous set bits in it.
6495 @defmac DWARF2_UNWIND_INFO
6496 Define this macro to 0 if your target supports DWARF 2 frame unwind
6497 information, but it does not yet work with exception handling.
6498 Otherwise, if your target supports this information (if it defines
6499 @code{INCOMING_RETURN_ADDR_RTX} and @code{OBJECT_FORMAT_ELF}),
6500 GCC will provide a default definition of 1.
6503 @hook TARGET_EXCEPT_UNWIND_INFO
6504 This hook defines the mechanism that will be used for exception handling
6505 by the target. If the target has ABI specified unwind tables, the hook
6506 should return @code{UI_TARGET}. If the target is to use the
6507 @code{setjmp}/@code{longjmp}-based exception handling scheme, the hook
6508 should return @code{UI_SJLJ}. If the target supports DWARF 2 frame unwind
6509 information, the hook should return @code{UI_DWARF2}.
6511 A target may, if exceptions are disabled, choose to return @code{UI_NONE}.
6512 This may end up simplifying other parts of target-specific code. The
6513 default implementation of this hook never returns @code{UI_NONE}.
6515 Note that the value returned by this hook should be constant. It should
6516 not depend on anything except the command-line switches described by
6517 @var{opts}. In particular, the
6518 setting @code{UI_SJLJ} must be fixed at compiler start-up as C pre-processor
6519 macros and builtin functions related to exception handling are set up
6520 depending on this setting.
6522 The default implementation of the hook first honors the
6523 @option{--enable-sjlj-exceptions} configure option, then
6524 @code{DWARF2_UNWIND_INFO}, and finally defaults to @code{UI_SJLJ}. If
6525 @code{DWARF2_UNWIND_INFO} depends on command-line options, the target
6526 must define this hook so that @var{opts} is used correctly.
6529 @hook TARGET_UNWIND_TABLES_DEFAULT
6530 This variable should be set to @code{true} if the target ABI requires unwinding
6531 tables even when exceptions are not used. It must not be modified by
6532 command-line option processing.
6535 @defmac DONT_USE_BUILTIN_SETJMP
6536 Define this macro to 1 if the @code{setjmp}/@code{longjmp}-based scheme
6537 should use the @code{setjmp}/@code{longjmp} functions from the C library
6538 instead of the @code{__builtin_setjmp}/@code{__builtin_longjmp} machinery.
6541 @defmac JMP_BUF_SIZE
6542 This macro has no effect unless @code{DONT_USE_BUILTIN_SETJMP} is also
6543 defined. Define this macro if the default size of @code{jmp_buf} buffer
6544 for the @code{setjmp}/@code{longjmp}-based exception handling mechanism
6545 is not large enough, or if it is much too large.
6546 The default size is @code{FIRST_PSEUDO_REGISTER * sizeof(void *)}.
6549 @defmac DWARF_CIE_DATA_ALIGNMENT
6550 This macro need only be defined if the target might save registers in the
6551 function prologue at an offset to the stack pointer that is not aligned to
6552 @code{UNITS_PER_WORD}. The definition should be the negative minimum
6553 alignment if @code{STACK_GROWS_DOWNWARD} is true, and the positive
6554 minimum alignment otherwise. @xref{DWARF}. Only applicable if
6555 the target supports DWARF 2 frame unwind information.
6558 @hook TARGET_TERMINATE_DW2_EH_FRAME_INFO
6560 @hook TARGET_DWARF_REGISTER_SPAN
6562 @hook TARGET_DWARF_FRAME_REG_MODE
6564 @hook TARGET_INIT_DWARF_REG_SIZES_EXTRA
6566 @hook TARGET_ASM_TTYPE
6568 @hook TARGET_ARM_EABI_UNWINDER
6570 @node Alignment Output
6571 @subsection Assembler Commands for Alignment
6573 @c prevent bad page break with this line
6574 This describes commands for alignment.
6576 @defmac JUMP_ALIGN (@var{label})
6577 The alignment (log base 2) to put in front of @var{label}, which is
6578 a common destination of jumps and has no fallthru incoming edge.
6580 This macro need not be defined if you don't want any special alignment
6581 to be done at such a time. Most machine descriptions do not currently
6584 Unless it's necessary to inspect the @var{label} parameter, it is better
6585 to set the variable @var{align_jumps} in the target's
6586 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
6587 selection in @var{align_jumps} in a @code{JUMP_ALIGN} implementation.
6590 @defmac LABEL_ALIGN_AFTER_BARRIER (@var{label})
6591 The alignment (log base 2) to put in front of @var{label}, which follows
6594 This macro need not be defined if you don't want any special alignment
6595 to be done at such a time. Most machine descriptions do not currently
6599 @defmac LOOP_ALIGN (@var{label})
6600 The alignment (log base 2) to put in front of @var{label} that heads
6601 a frequently executed basic block (usually the header of a loop).
6603 This macro need not be defined if you don't want any special alignment
6604 to be done at such a time. Most machine descriptions do not currently
6607 Unless it's necessary to inspect the @var{label} parameter, it is better
6608 to set the variable @code{align_loops} in the target's
6609 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
6610 selection in @code{align_loops} in a @code{LOOP_ALIGN} implementation.
6613 @defmac LABEL_ALIGN (@var{label})
6614 The alignment (log base 2) to put in front of @var{label}.
6615 If @code{LABEL_ALIGN_AFTER_BARRIER} / @code{LOOP_ALIGN} specify a different alignment,
6616 the maximum of the specified values is used.
6618 Unless it's necessary to inspect the @var{label} parameter, it is better
6619 to set the variable @code{align_labels} in the target's
6620 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
6621 selection in @code{align_labels} in a @code{LABEL_ALIGN} implementation.
6624 @defmac ASM_OUTPUT_SKIP (@var{stream}, @var{nbytes})
6625 A C statement to output to the stdio stream @var{stream} an assembler
6626 instruction to advance the location counter by @var{nbytes} bytes.
6627 Those bytes should be zero when loaded. @var{nbytes} will be a C
6628 expression of type @code{unsigned HOST_WIDE_INT}.
6631 @defmac ASM_NO_SKIP_IN_TEXT
6632 Define this macro if @code{ASM_OUTPUT_SKIP} should not be used in the
6633 text section because it fails to put zeros in the bytes that are skipped.
6634 This is true on many Unix systems, where the pseudo--op to skip bytes
6635 produces no-op instructions rather than zeros when used in the text
6639 @defmac ASM_OUTPUT_ALIGN (@var{stream}, @var{power})
6640 A C statement to output to the stdio stream @var{stream} an assembler
6641 command to advance the location counter to a multiple of 2 to the
6642 @var{power} bytes. @var{power} will be a C expression of type @code{int}.
6645 @defmac ASM_OUTPUT_ALIGN_WITH_NOP (@var{stream}, @var{power})
6646 Like @code{ASM_OUTPUT_ALIGN}, except that the ``nop'' instruction is used
6647 for padding, if necessary.
6650 @defmac ASM_OUTPUT_MAX_SKIP_ALIGN (@var{stream}, @var{power}, @var{max_skip})
6651 A C statement to output to the stdio stream @var{stream} an assembler
6652 command to advance the location counter to a multiple of 2 to the
6653 @var{power} bytes, but only if @var{max_skip} or fewer bytes are needed to
6654 satisfy the alignment request. @var{power} and @var{max_skip} will be
6655 a C expression of type @code{int}.
6659 @node Debugging Info
6660 @section Controlling Debugging Information Format
6662 @c prevent bad page break with this line
6663 This describes how to specify debugging information.
6666 * All Debuggers:: Macros that affect all debugging formats uniformly.
6667 * DBX Options:: Macros enabling specific options in DBX format.
6668 * DBX Hooks:: Hook macros for varying DBX format.
6669 * File Names and DBX:: Macros controlling output of file names in DBX format.
6670 * DWARF:: Macros for DWARF format.
6671 * VMS Debug:: Macros for VMS debug format.
6675 @subsection Macros Affecting All Debugging Formats
6677 @c prevent bad page break with this line
6678 These macros affect all debugging formats.
6680 @defmac DBX_REGISTER_NUMBER (@var{regno})
6681 A C expression that returns the DBX register number for the compiler
6682 register number @var{regno}. In the default macro provided, the value
6683 of this expression will be @var{regno} itself. But sometimes there are
6684 some registers that the compiler knows about and DBX does not, or vice
6685 versa. In such cases, some register may need to have one number in the
6686 compiler and another for DBX@.
6688 If two registers have consecutive numbers inside GCC, and they can be
6689 used as a pair to hold a multiword value, then they @emph{must} have
6690 consecutive numbers after renumbering with @code{DBX_REGISTER_NUMBER}.
6691 Otherwise, debuggers will be unable to access such a pair, because they
6692 expect register pairs to be consecutive in their own numbering scheme.
6694 If you find yourself defining @code{DBX_REGISTER_NUMBER} in way that
6695 does not preserve register pairs, then what you must do instead is
6696 redefine the actual register numbering scheme.
6699 @defmac DEBUGGER_AUTO_OFFSET (@var{x})
6700 A C expression that returns the integer offset value for an automatic
6701 variable having address @var{x} (an RTL expression). The default
6702 computation assumes that @var{x} is based on the frame-pointer and
6703 gives the offset from the frame-pointer. This is required for targets
6704 that produce debugging output for DBX and allow the frame-pointer to be
6705 eliminated when the @option{-g} option is used.
6708 @defmac DEBUGGER_ARG_OFFSET (@var{offset}, @var{x})
6709 A C expression that returns the integer offset value for an argument
6710 having address @var{x} (an RTL expression). The nominal offset is
6714 @defmac PREFERRED_DEBUGGING_TYPE
6715 A C expression that returns the type of debugging output GCC should
6716 produce when the user specifies just @option{-g}. Define
6717 this if you have arranged for GCC to support more than one format of
6718 debugging output. Currently, the allowable values are @code{DBX_DEBUG},
6719 @code{DWARF2_DEBUG}, @code{XCOFF_DEBUG}, @code{VMS_DEBUG},
6720 and @code{VMS_AND_DWARF2_DEBUG}.
6722 When the user specifies @option{-ggdb}, GCC normally also uses the
6723 value of this macro to select the debugging output format, but with two
6724 exceptions. If @code{DWARF2_DEBUGGING_INFO} is defined, GCC uses the
6725 value @code{DWARF2_DEBUG}. Otherwise, if @code{DBX_DEBUGGING_INFO} is
6726 defined, GCC uses @code{DBX_DEBUG}.
6728 The value of this macro only affects the default debugging output; the
6729 user can always get a specific type of output by using @option{-gstabs},
6730 @option{-gdwarf-2}, @option{-gxcoff}, or @option{-gvms}.
6734 @subsection Specific Options for DBX Output
6736 @c prevent bad page break with this line
6737 These are specific options for DBX output.
6739 @defmac DBX_DEBUGGING_INFO
6740 Define this macro if GCC should produce debugging output for DBX
6741 in response to the @option{-g} option.
6744 @defmac XCOFF_DEBUGGING_INFO
6745 Define this macro if GCC should produce XCOFF format debugging output
6746 in response to the @option{-g} option. This is a variant of DBX format.
6749 @defmac DEFAULT_GDB_EXTENSIONS
6750 Define this macro to control whether GCC should by default generate
6751 GDB's extended version of DBX debugging information (assuming DBX-format
6752 debugging information is enabled at all). If you don't define the
6753 macro, the default is 1: always generate the extended information
6754 if there is any occasion to.
6757 @defmac DEBUG_SYMS_TEXT
6758 Define this macro if all @code{.stabs} commands should be output while
6759 in the text section.
6762 @defmac ASM_STABS_OP
6763 A C string constant, including spacing, naming the assembler pseudo op to
6764 use instead of @code{"\t.stabs\t"} to define an ordinary debugging symbol.
6765 If you don't define this macro, @code{"\t.stabs\t"} is used. This macro
6766 applies only to DBX debugging information format.
6769 @defmac ASM_STABD_OP
6770 A C string constant, including spacing, naming the assembler pseudo op to
6771 use instead of @code{"\t.stabd\t"} to define a debugging symbol whose
6772 value is the current location. If you don't define this macro,
6773 @code{"\t.stabd\t"} is used. This macro applies only to DBX debugging
6777 @defmac ASM_STABN_OP
6778 A C string constant, including spacing, naming the assembler pseudo op to
6779 use instead of @code{"\t.stabn\t"} to define a debugging symbol with no
6780 name. If you don't define this macro, @code{"\t.stabn\t"} is used. This
6781 macro applies only to DBX debugging information format.
6784 @defmac DBX_NO_XREFS
6785 Define this macro if DBX on your system does not support the construct
6786 @samp{xs@var{tagname}}. On some systems, this construct is used to
6787 describe a forward reference to a structure named @var{tagname}.
6788 On other systems, this construct is not supported at all.
6791 @defmac DBX_CONTIN_LENGTH
6792 A symbol name in DBX-format debugging information is normally
6793 continued (split into two separate @code{.stabs} directives) when it
6794 exceeds a certain length (by default, 80 characters). On some
6795 operating systems, DBX requires this splitting; on others, splitting
6796 must not be done. You can inhibit splitting by defining this macro
6797 with the value zero. You can override the default splitting-length by
6798 defining this macro as an expression for the length you desire.
6801 @defmac DBX_CONTIN_CHAR
6802 Normally continuation is indicated by adding a @samp{\} character to
6803 the end of a @code{.stabs} string when a continuation follows. To use
6804 a different character instead, define this macro as a character
6805 constant for the character you want to use. Do not define this macro
6806 if backslash is correct for your system.
6809 @defmac DBX_STATIC_STAB_DATA_SECTION
6810 Define this macro if it is necessary to go to the data section before
6811 outputting the @samp{.stabs} pseudo-op for a non-global static
6815 @defmac DBX_TYPE_DECL_STABS_CODE
6816 The value to use in the ``code'' field of the @code{.stabs} directive
6817 for a typedef. The default is @code{N_LSYM}.
6820 @defmac DBX_STATIC_CONST_VAR_CODE
6821 The value to use in the ``code'' field of the @code{.stabs} directive
6822 for a static variable located in the text section. DBX format does not
6823 provide any ``right'' way to do this. The default is @code{N_FUN}.
6826 @defmac DBX_REGPARM_STABS_CODE
6827 The value to use in the ``code'' field of the @code{.stabs} directive
6828 for a parameter passed in registers. DBX format does not provide any
6829 ``right'' way to do this. The default is @code{N_RSYM}.
6832 @defmac DBX_REGPARM_STABS_LETTER
6833 The letter to use in DBX symbol data to identify a symbol as a parameter
6834 passed in registers. DBX format does not customarily provide any way to
6835 do this. The default is @code{'P'}.
6838 @defmac DBX_FUNCTION_FIRST
6839 Define this macro if the DBX information for a function and its
6840 arguments should precede the assembler code for the function. Normally,
6841 in DBX format, the debugging information entirely follows the assembler
6845 @defmac DBX_BLOCKS_FUNCTION_RELATIVE
6846 Define this macro, with value 1, if the value of a symbol describing
6847 the scope of a block (@code{N_LBRAC} or @code{N_RBRAC}) should be
6848 relative to the start of the enclosing function. Normally, GCC uses
6849 an absolute address.
6852 @defmac DBX_LINES_FUNCTION_RELATIVE
6853 Define this macro, with value 1, if the value of a symbol indicating
6854 the current line number (@code{N_SLINE}) should be relative to the
6855 start of the enclosing function. Normally, GCC uses an absolute address.
6858 @defmac DBX_USE_BINCL
6859 Define this macro if GCC should generate @code{N_BINCL} and
6860 @code{N_EINCL} stabs for included header files, as on Sun systems. This
6861 macro also directs GCC to output a type number as a pair of a file
6862 number and a type number within the file. Normally, GCC does not
6863 generate @code{N_BINCL} or @code{N_EINCL} stabs, and it outputs a single
6864 number for a type number.
6868 @subsection Open-Ended Hooks for DBX Format
6870 @c prevent bad page break with this line
6871 These are hooks for DBX format.
6873 @defmac DBX_OUTPUT_SOURCE_LINE (@var{stream}, @var{line}, @var{counter})
6874 A C statement to output DBX debugging information before code for line
6875 number @var{line} of the current source file to the stdio stream
6876 @var{stream}. @var{counter} is the number of time the macro was
6877 invoked, including the current invocation; it is intended to generate
6878 unique labels in the assembly output.
6880 This macro should not be defined if the default output is correct, or
6881 if it can be made correct by defining @code{DBX_LINES_FUNCTION_RELATIVE}.
6884 @defmac NO_DBX_FUNCTION_END
6885 Some stabs encapsulation formats (in particular ECOFF), cannot handle the
6886 @code{.stabs "",N_FUN,,0,0,Lscope-function-1} gdb dbx extension construct.
6887 On those machines, define this macro to turn this feature off without
6888 disturbing the rest of the gdb extensions.
6891 @defmac NO_DBX_BNSYM_ENSYM
6892 Some assemblers cannot handle the @code{.stabd BNSYM/ENSYM,0,0} gdb dbx
6893 extension construct. On those machines, define this macro to turn this
6894 feature off without disturbing the rest of the gdb extensions.
6897 @node File Names and DBX
6898 @subsection File Names in DBX Format
6900 @c prevent bad page break with this line
6901 This describes file names in DBX format.
6903 @defmac DBX_OUTPUT_MAIN_SOURCE_FILENAME (@var{stream}, @var{name})
6904 A C statement to output DBX debugging information to the stdio stream
6905 @var{stream}, which indicates that file @var{name} is the main source
6906 file---the file specified as the input file for compilation.
6907 This macro is called only once, at the beginning of compilation.
6909 This macro need not be defined if the standard form of output
6910 for DBX debugging information is appropriate.
6912 It may be necessary to refer to a label equal to the beginning of the
6913 text section. You can use @samp{assemble_name (stream, ltext_label_name)}
6914 to do so. If you do this, you must also set the variable
6915 @var{used_ltext_label_name} to @code{true}.
6918 @defmac NO_DBX_MAIN_SOURCE_DIRECTORY
6919 Define this macro, with value 1, if GCC should not emit an indication
6920 of the current directory for compilation and current source language at
6921 the beginning of the file.
6924 @defmac NO_DBX_GCC_MARKER
6925 Define this macro, with value 1, if GCC should not emit an indication
6926 that this object file was compiled by GCC@. The default is to emit
6927 an @code{N_OPT} stab at the beginning of every source file, with
6928 @samp{gcc2_compiled.} for the string and value 0.
6931 @defmac DBX_OUTPUT_MAIN_SOURCE_FILE_END (@var{stream}, @var{name})
6932 A C statement to output DBX debugging information at the end of
6933 compilation of the main source file @var{name}. Output should be
6934 written to the stdio stream @var{stream}.
6936 If you don't define this macro, nothing special is output at the end
6937 of compilation, which is correct for most machines.
6940 @defmac DBX_OUTPUT_NULL_N_SO_AT_MAIN_SOURCE_FILE_END
6941 Define this macro @emph{instead of} defining
6942 @code{DBX_OUTPUT_MAIN_SOURCE_FILE_END}, if what needs to be output at
6943 the end of compilation is an @code{N_SO} stab with an empty string,
6944 whose value is the highest absolute text address in the file.
6949 @subsection Macros for DWARF Output
6951 @c prevent bad page break with this line
6952 Here are macros for DWARF output.
6954 @defmac DWARF2_DEBUGGING_INFO
6955 Define this macro if GCC should produce dwarf version 2 format
6956 debugging output in response to the @option{-g} option.
6958 @hook TARGET_DWARF_CALLING_CONVENTION
6960 To support optional call frame debugging information, you must also
6961 define @code{INCOMING_RETURN_ADDR_RTX} and either set
6962 @code{RTX_FRAME_RELATED_P} on the prologue insns if you use RTL for the
6963 prologue, or call @code{dwarf2out_def_cfa} and @code{dwarf2out_reg_save}
6964 as appropriate from @code{TARGET_ASM_FUNCTION_PROLOGUE} if you don't.
6967 @defmac DWARF2_FRAME_INFO
6968 Define this macro to a nonzero value if GCC should always output
6969 Dwarf 2 frame information. If @code{TARGET_EXCEPT_UNWIND_INFO}
6970 (@pxref{Exception Region Output}) returns @code{UI_DWARF2}, and
6971 exceptions are enabled, GCC will output this information not matter
6972 how you define @code{DWARF2_FRAME_INFO}.
6975 @hook TARGET_DEBUG_UNWIND_INFO
6977 @defmac DWARF2_ASM_LINE_DEBUG_INFO
6978 Define this macro to be a nonzero value if the assembler can generate Dwarf 2
6979 line debug info sections. This will result in much more compact line number
6980 tables, and hence is desirable if it works.
6983 @defmac DWARF2_ASM_VIEW_DEBUG_INFO
6984 Define this macro to be a nonzero value if the assembler supports view
6985 assignment and verification in @code{.loc}. If it does not, but the
6986 user enables location views, the compiler may have to fallback to
6987 internal line number tables.
6990 @hook TARGET_RESET_LOCATION_VIEW
6992 @hook TARGET_WANT_DEBUG_PUB_SECTIONS
6994 @hook TARGET_DELAY_SCHED2
6996 @hook TARGET_DELAY_VARTRACK
6998 @hook TARGET_NO_REGISTER_ALLOCATION
7000 @defmac ASM_OUTPUT_DWARF_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
7001 A C statement to issue assembly directives that create a difference
7002 @var{lab1} minus @var{lab2}, using an integer of the given @var{size}.
7005 @defmac ASM_OUTPUT_DWARF_VMS_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
7006 A C statement to issue assembly directives that create a difference
7007 between the two given labels in system defined units, e.g.@: instruction
7008 slots on IA64 VMS, using an integer of the given size.
7011 @defmac ASM_OUTPUT_DWARF_OFFSET (@var{stream}, @var{size}, @var{label}, @var{offset}, @var{section})
7012 A C statement to issue assembly directives that create a
7013 section-relative reference to the given @var{label} plus @var{offset}, using
7014 an integer of the given @var{size}. The label is known to be defined in the
7015 given @var{section}.
7018 @defmac ASM_OUTPUT_DWARF_PCREL (@var{stream}, @var{size}, @var{label})
7019 A C statement to issue assembly directives that create a self-relative
7020 reference to the given @var{label}, using an integer of the given @var{size}.
7023 @defmac ASM_OUTPUT_DWARF_DATAREL (@var{stream}, @var{size}, @var{label})
7024 A C statement to issue assembly directives that create a reference to the
7025 given @var{label} relative to the dbase, using an integer of the given @var{size}.
7028 @defmac ASM_OUTPUT_DWARF_TABLE_REF (@var{label})
7029 A C statement to issue assembly directives that create a reference to
7030 the DWARF table identifier @var{label} from the current section. This
7031 is used on some systems to avoid garbage collecting a DWARF table which
7032 is referenced by a function.
7035 @hook TARGET_ASM_OUTPUT_DWARF_DTPREL
7039 @subsection Macros for VMS Debug Format
7041 @c prevent bad page break with this line
7042 Here are macros for VMS debug format.
7044 @defmac VMS_DEBUGGING_INFO
7045 Define this macro if GCC should produce debugging output for VMS
7046 in response to the @option{-g} option. The default behavior for VMS
7047 is to generate minimal debug info for a traceback in the absence of
7048 @option{-g} unless explicitly overridden with @option{-g0}. This
7049 behavior is controlled by @code{TARGET_OPTION_OPTIMIZATION} and
7050 @code{TARGET_OPTION_OVERRIDE}.
7053 @node Floating Point
7054 @section Cross Compilation and Floating Point
7055 @cindex cross compilation and floating point
7056 @cindex floating point and cross compilation
7058 While all modern machines use twos-complement representation for integers,
7059 there are a variety of representations for floating point numbers. This
7060 means that in a cross-compiler the representation of floating point numbers
7061 in the compiled program may be different from that used in the machine
7062 doing the compilation.
7064 Because different representation systems may offer different amounts of
7065 range and precision, all floating point constants must be represented in
7066 the target machine's format. Therefore, the cross compiler cannot
7067 safely use the host machine's floating point arithmetic; it must emulate
7068 the target's arithmetic. To ensure consistency, GCC always uses
7069 emulation to work with floating point values, even when the host and
7070 target floating point formats are identical.
7072 The following macros are provided by @file{real.h} for the compiler to
7073 use. All parts of the compiler which generate or optimize
7074 floating-point calculations must use these macros. They may evaluate
7075 their operands more than once, so operands must not have side effects.
7077 @defmac REAL_VALUE_TYPE
7078 The C data type to be used to hold a floating point value in the target
7079 machine's format. Typically this is a @code{struct} containing an
7080 array of @code{HOST_WIDE_INT}, but all code should treat it as an opaque
7084 @deftypefn Macro HOST_WIDE_INT REAL_VALUE_FIX (REAL_VALUE_TYPE @var{x})
7085 Truncates @var{x} to a signed integer, rounding toward zero.
7088 @deftypefn Macro {unsigned HOST_WIDE_INT} REAL_VALUE_UNSIGNED_FIX (REAL_VALUE_TYPE @var{x})
7089 Truncates @var{x} to an unsigned integer, rounding toward zero. If
7090 @var{x} is negative, returns zero.
7093 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ATOF (const char *@var{string}, machine_mode @var{mode})
7094 Converts @var{string} into a floating point number in the target machine's
7095 representation for mode @var{mode}. This routine can handle both
7096 decimal and hexadecimal floating point constants, using the syntax
7097 defined by the C language for both.
7100 @deftypefn Macro int REAL_VALUE_NEGATIVE (REAL_VALUE_TYPE @var{x})
7101 Returns 1 if @var{x} is negative (including negative zero), 0 otherwise.
7104 @deftypefn Macro int REAL_VALUE_ISINF (REAL_VALUE_TYPE @var{x})
7105 Determines whether @var{x} represents infinity (positive or negative).
7108 @deftypefn Macro int REAL_VALUE_ISNAN (REAL_VALUE_TYPE @var{x})
7109 Determines whether @var{x} represents a ``NaN'' (not-a-number).
7112 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_NEGATE (REAL_VALUE_TYPE @var{x})
7113 Returns the negative of the floating point value @var{x}.
7116 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ABS (REAL_VALUE_TYPE @var{x})
7117 Returns the absolute value of @var{x}.
7120 @node Mode Switching
7121 @section Mode Switching Instructions
7122 @cindex mode switching
7123 The following macros control mode switching optimizations:
7125 @defmac OPTIMIZE_MODE_SWITCHING (@var{entity})
7126 Define this macro if the port needs extra instructions inserted for mode
7127 switching in an optimizing compilation.
7129 For an example, the SH4 can perform both single and double precision
7130 floating point operations, but to perform a single precision operation,
7131 the FPSCR PR bit has to be cleared, while for a double precision
7132 operation, this bit has to be set. Changing the PR bit requires a general
7133 purpose register as a scratch register, hence these FPSCR sets have to
7134 be inserted before reload, i.e.@: you cannot put this into instruction emitting
7135 or @code{TARGET_MACHINE_DEPENDENT_REORG}.
7137 You can have multiple entities that are mode-switched, and select at run time
7138 which entities actually need it. @code{OPTIMIZE_MODE_SWITCHING} should
7139 return nonzero for any @var{entity} that needs mode-switching.
7140 If you define this macro, you also have to define
7141 @code{NUM_MODES_FOR_MODE_SWITCHING}, @code{TARGET_MODE_NEEDED},
7142 @code{TARGET_MODE_PRIORITY} and @code{TARGET_MODE_EMIT}.
7143 @code{TARGET_MODE_AFTER}, @code{TARGET_MODE_ENTRY}, and @code{TARGET_MODE_EXIT}
7147 @defmac NUM_MODES_FOR_MODE_SWITCHING
7148 If you define @code{OPTIMIZE_MODE_SWITCHING}, you have to define this as
7149 initializer for an array of integers. Each initializer element
7150 N refers to an entity that needs mode switching, and specifies the number
7151 of different modes that might need to be set for this entity.
7152 The position of the initializer in the initializer---starting counting at
7153 zero---determines the integer that is used to refer to the mode-switched
7155 In macros that take mode arguments / yield a mode result, modes are
7156 represented as numbers 0 @dots{} N @minus{} 1. N is used to specify that no mode
7157 switch is needed / supplied.
7160 @hook TARGET_MODE_EMIT
7162 @hook TARGET_MODE_NEEDED
7164 @hook TARGET_MODE_AFTER
7166 @hook TARGET_MODE_ENTRY
7168 @hook TARGET_MODE_EXIT
7170 @hook TARGET_MODE_PRIORITY
7172 @node Target Attributes
7173 @section Defining target-specific uses of @code{__attribute__}
7174 @cindex target attributes
7175 @cindex machine attributes
7176 @cindex attributes, target-specific
7178 Target-specific attributes may be defined for functions, data and types.
7179 These are described using the following target hooks; they also need to
7180 be documented in @file{extend.texi}.
7182 @hook TARGET_ATTRIBUTE_TABLE
7184 @hook TARGET_ATTRIBUTE_TAKES_IDENTIFIER_P
7186 @hook TARGET_COMP_TYPE_ATTRIBUTES
7188 @hook TARGET_SET_DEFAULT_TYPE_ATTRIBUTES
7190 @hook TARGET_MERGE_TYPE_ATTRIBUTES
7192 @hook TARGET_MERGE_DECL_ATTRIBUTES
7194 @hook TARGET_VALID_DLLIMPORT_ATTRIBUTE_P
7196 @defmac TARGET_DECLSPEC
7197 Define this macro to a nonzero value if you want to treat
7198 @code{__declspec(X)} as equivalent to @code{__attribute((X))}. By
7199 default, this behavior is enabled only for targets that define
7200 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES}. The current implementation
7201 of @code{__declspec} is via a built-in macro, but you should not rely
7202 on this implementation detail.
7205 @hook TARGET_INSERT_ATTRIBUTES
7207 @hook TARGET_HANDLE_GENERIC_ATTRIBUTE
7209 @hook TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P
7211 @hook TARGET_OPTION_VALID_ATTRIBUTE_P
7213 @hook TARGET_OPTION_SAVE
7215 @hook TARGET_OPTION_RESTORE
7217 @hook TARGET_OPTION_POST_STREAM_IN
7219 @hook TARGET_OPTION_PRINT
7221 @hook TARGET_OPTION_PRAGMA_PARSE
7223 @hook TARGET_OPTION_OVERRIDE
7225 @hook TARGET_OPTION_FUNCTION_VERSIONS
7227 @hook TARGET_CAN_INLINE_P
7229 @hook TARGET_RELAYOUT_FUNCTION
7232 @section Emulating TLS
7233 @cindex Emulated TLS
7235 For targets whose psABI does not provide Thread Local Storage via
7236 specific relocations and instruction sequences, an emulation layer is
7237 used. A set of target hooks allows this emulation layer to be
7238 configured for the requirements of a particular target. For instance
7239 the psABI may in fact specify TLS support in terms of an emulation
7242 The emulation layer works by creating a control object for every TLS
7243 object. To access the TLS object, a lookup function is provided
7244 which, when given the address of the control object, will return the
7245 address of the current thread's instance of the TLS object.
7247 @hook TARGET_EMUTLS_GET_ADDRESS
7249 @hook TARGET_EMUTLS_REGISTER_COMMON
7251 @hook TARGET_EMUTLS_VAR_SECTION
7253 @hook TARGET_EMUTLS_TMPL_SECTION
7255 @hook TARGET_EMUTLS_VAR_PREFIX
7257 @hook TARGET_EMUTLS_TMPL_PREFIX
7259 @hook TARGET_EMUTLS_VAR_FIELDS
7261 @hook TARGET_EMUTLS_VAR_INIT
7263 @hook TARGET_EMUTLS_VAR_ALIGN_FIXED
7265 @hook TARGET_EMUTLS_DEBUG_FORM_TLS_ADDRESS
7267 @node MIPS Coprocessors
7268 @section Defining coprocessor specifics for MIPS targets.
7269 @cindex MIPS coprocessor-definition macros
7271 The MIPS specification allows MIPS implementations to have as many as 4
7272 coprocessors, each with as many as 32 private registers. GCC supports
7273 accessing these registers and transferring values between the registers
7274 and memory using asm-ized variables. For example:
7277 register unsigned int cp0count asm ("c0r1");
7283 (``c0r1'' is the default name of register 1 in coprocessor 0; alternate
7284 names may be added as described below, or the default names may be
7285 overridden entirely in @code{SUBTARGET_CONDITIONAL_REGISTER_USAGE}.)
7287 Coprocessor registers are assumed to be epilogue-used; sets to them will
7288 be preserved even if it does not appear that the register is used again
7289 later in the function.
7291 Another note: according to the MIPS spec, coprocessor 1 (if present) is
7292 the FPU@. One accesses COP1 registers through standard mips
7293 floating-point support; they are not included in this mechanism.
7296 @section Parameters for Precompiled Header Validity Checking
7297 @cindex parameters, precompiled headers
7299 @hook TARGET_GET_PCH_VALIDITY
7301 @hook TARGET_PCH_VALID_P
7303 @hook TARGET_CHECK_PCH_TARGET_FLAGS
7305 @hook TARGET_PREPARE_PCH_SAVE
7308 @section C++ ABI parameters
7309 @cindex parameters, c++ abi
7311 @hook TARGET_CXX_GUARD_TYPE
7313 @hook TARGET_CXX_GUARD_MASK_BIT
7315 @hook TARGET_CXX_GET_COOKIE_SIZE
7317 @hook TARGET_CXX_COOKIE_HAS_SIZE
7319 @hook TARGET_CXX_IMPORT_EXPORT_CLASS
7321 @hook TARGET_CXX_CDTOR_RETURNS_THIS
7323 @hook TARGET_CXX_KEY_METHOD_MAY_BE_INLINE
7325 @hook TARGET_CXX_DETERMINE_CLASS_DATA_VISIBILITY
7327 @hook TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT
7329 @hook TARGET_CXX_LIBRARY_RTTI_COMDAT
7331 @hook TARGET_CXX_USE_AEABI_ATEXIT
7333 @hook TARGET_CXX_USE_ATEXIT_FOR_CXA_ATEXIT
7335 @hook TARGET_CXX_ADJUST_CLASS_AT_DEFINITION
7337 @hook TARGET_CXX_DECL_MANGLING_CONTEXT
7339 @node D Language and ABI
7340 @section D ABI parameters
7341 @cindex parameters, d abi
7343 @hook TARGET_D_CPU_VERSIONS
7345 @hook TARGET_D_OS_VERSIONS
7347 @hook TARGET_D_CRITSEC_SIZE
7349 @node Named Address Spaces
7350 @section Adding support for named address spaces
7351 @cindex named address spaces
7353 The draft technical report of the ISO/IEC JTC1 S22 WG14 N1275
7354 standards committee, @cite{Programming Languages - C - Extensions to
7355 support embedded processors}, specifies a syntax for embedded
7356 processors to specify alternate address spaces. You can configure a
7357 GCC port to support section 5.1 of the draft report to add support for
7358 address spaces other than the default address space. These address
7359 spaces are new keywords that are similar to the @code{volatile} and
7360 @code{const} type attributes.
7362 Pointers to named address spaces can have a different size than
7363 pointers to the generic address space.
7365 For example, the SPU port uses the @code{__ea} address space to refer
7366 to memory in the host processor, rather than memory local to the SPU
7367 processor. Access to memory in the @code{__ea} address space involves
7368 issuing DMA operations to move data between the host processor and the
7369 local processor memory address space. Pointers in the @code{__ea}
7370 address space are either 32 bits or 64 bits based on the
7371 @option{-mea32} or @option{-mea64} switches (native SPU pointers are
7374 Internally, address spaces are represented as a small integer in the
7375 range 0 to 15 with address space 0 being reserved for the generic
7378 To register a named address space qualifier keyword with the C front end,
7379 the target may call the @code{c_register_addr_space} routine. For example,
7380 the SPU port uses the following to declare @code{__ea} as the keyword for
7381 named address space #1:
7383 #define ADDR_SPACE_EA 1
7384 c_register_addr_space ("__ea", ADDR_SPACE_EA);
7387 @hook TARGET_ADDR_SPACE_POINTER_MODE
7389 @hook TARGET_ADDR_SPACE_ADDRESS_MODE
7391 @hook TARGET_ADDR_SPACE_VALID_POINTER_MODE
7393 @hook TARGET_ADDR_SPACE_LEGITIMATE_ADDRESS_P
7395 @hook TARGET_ADDR_SPACE_LEGITIMIZE_ADDRESS
7397 @hook TARGET_ADDR_SPACE_SUBSET_P
7399 @hook TARGET_ADDR_SPACE_ZERO_ADDRESS_VALID
7401 @hook TARGET_ADDR_SPACE_CONVERT
7403 @hook TARGET_ADDR_SPACE_DEBUG
7405 @hook TARGET_ADDR_SPACE_DIAGNOSE_USAGE
7408 @section Miscellaneous Parameters
7409 @cindex parameters, miscellaneous
7411 @c prevent bad page break with this line
7412 Here are several miscellaneous parameters.
7414 @defmac HAS_LONG_COND_BRANCH
7415 Define this boolean macro to indicate whether or not your architecture
7416 has conditional branches that can span all of memory. It is used in
7417 conjunction with an optimization that partitions hot and cold basic
7418 blocks into separate sections of the executable. If this macro is
7419 set to false, gcc will convert any conditional branches that attempt
7420 to cross between sections into unconditional branches or indirect jumps.
7423 @defmac HAS_LONG_UNCOND_BRANCH
7424 Define this boolean macro to indicate whether or not your architecture
7425 has unconditional branches that can span all of memory. It is used in
7426 conjunction with an optimization that partitions hot and cold basic
7427 blocks into separate sections of the executable. If this macro is
7428 set to false, gcc will convert any unconditional branches that attempt
7429 to cross between sections into indirect jumps.
7432 @defmac CASE_VECTOR_MODE
7433 An alias for a machine mode name. This is the machine mode that
7434 elements of a jump-table should have.
7437 @defmac CASE_VECTOR_SHORTEN_MODE (@var{min_offset}, @var{max_offset}, @var{body})
7438 Optional: return the preferred mode for an @code{addr_diff_vec}
7439 when the minimum and maximum offset are known. If you define this,
7440 it enables extra code in branch shortening to deal with @code{addr_diff_vec}.
7441 To make this work, you also have to define @code{INSN_ALIGN} and
7442 make the alignment for @code{addr_diff_vec} explicit.
7443 The @var{body} argument is provided so that the offset_unsigned and scale
7444 flags can be updated.
7447 @defmac CASE_VECTOR_PC_RELATIVE
7448 Define this macro to be a C expression to indicate when jump-tables
7449 should contain relative addresses. You need not define this macro if
7450 jump-tables never contain relative addresses, or jump-tables should
7451 contain relative addresses only when @option{-fPIC} or @option{-fPIC}
7455 @hook TARGET_CASE_VALUES_THRESHOLD
7457 @defmac WORD_REGISTER_OPERATIONS
7458 Define this macro to 1 if operations between registers with integral mode
7459 smaller than a word are always performed on the entire register. To be
7460 more explicit, if you start with a pair of @code{word_mode} registers with
7461 known values and you do a subword, for example @code{QImode}, addition on
7462 the low part of the registers, then the compiler may consider that the
7463 result has a known value in @code{word_mode} too if the macro is defined
7464 to 1. Most RISC machines have this property and most CISC machines do not.
7467 @hook TARGET_MIN_ARITHMETIC_PRECISION
7469 @defmac LOAD_EXTEND_OP (@var{mem_mode})
7470 Define this macro to be a C expression indicating when insns that read
7471 memory in @var{mem_mode}, an integral mode narrower than a word, set the
7472 bits outside of @var{mem_mode} to be either the sign-extension or the
7473 zero-extension of the data read. Return @code{SIGN_EXTEND} for values
7474 of @var{mem_mode} for which the
7475 insn sign-extends, @code{ZERO_EXTEND} for which it zero-extends, and
7476 @code{UNKNOWN} for other modes.
7478 This macro is not called with @var{mem_mode} non-integral or with a width
7479 greater than or equal to @code{BITS_PER_WORD}, so you may return any
7480 value in this case. Do not define this macro if it would always return
7481 @code{UNKNOWN}. On machines where this macro is defined, you will normally
7482 define it as the constant @code{SIGN_EXTEND} or @code{ZERO_EXTEND}.
7484 You may return a non-@code{UNKNOWN} value even if for some hard registers
7485 the sign extension is not performed, if for the @code{REGNO_REG_CLASS}
7486 of these hard registers @code{TARGET_CAN_CHANGE_MODE_CLASS} returns false
7487 when the @var{from} mode is @var{mem_mode} and the @var{to} mode is any
7488 integral mode larger than this but not larger than @code{word_mode}.
7490 You must return @code{UNKNOWN} if for some hard registers that allow this
7491 mode, @code{TARGET_CAN_CHANGE_MODE_CLASS} says that they cannot change to
7492 @code{word_mode}, but that they can change to another integral mode that
7493 is larger then @var{mem_mode} but still smaller than @code{word_mode}.
7496 @defmac SHORT_IMMEDIATES_SIGN_EXTEND
7497 Define this macro to 1 if loading short immediate values into registers sign
7501 @hook TARGET_MIN_DIVISIONS_FOR_RECIP_MUL
7504 The maximum number of bytes that a single instruction can move quickly
7505 between memory and registers or between two memory locations.
7508 @defmac MAX_MOVE_MAX
7509 The maximum number of bytes that a single instruction can move quickly
7510 between memory and registers or between two memory locations. If this
7511 is undefined, the default is @code{MOVE_MAX}. Otherwise, it is the
7512 constant value that is the largest value that @code{MOVE_MAX} can have
7516 @defmac SHIFT_COUNT_TRUNCATED
7517 A C expression that is nonzero if on this machine the number of bits
7518 actually used for the count of a shift operation is equal to the number
7519 of bits needed to represent the size of the object being shifted. When
7520 this macro is nonzero, the compiler will assume that it is safe to omit
7521 a sign-extend, zero-extend, and certain bitwise `and' instructions that
7522 truncates the count of a shift operation. On machines that have
7523 instructions that act on bit-fields at variable positions, which may
7524 include `bit test' instructions, a nonzero @code{SHIFT_COUNT_TRUNCATED}
7525 also enables deletion of truncations of the values that serve as
7526 arguments to bit-field instructions.
7528 If both types of instructions truncate the count (for shifts) and
7529 position (for bit-field operations), or if no variable-position bit-field
7530 instructions exist, you should define this macro.
7532 However, on some machines, such as the 80386 and the 680x0, truncation
7533 only applies to shift operations and not the (real or pretended)
7534 bit-field operations. Define @code{SHIFT_COUNT_TRUNCATED} to be zero on
7535 such machines. Instead, add patterns to the @file{md} file that include
7536 the implied truncation of the shift instructions.
7538 You need not define this macro if it would always have the value of zero.
7541 @anchor{TARGET_SHIFT_TRUNCATION_MASK}
7542 @hook TARGET_SHIFT_TRUNCATION_MASK
7544 @hook TARGET_TRULY_NOOP_TRUNCATION
7546 @hook TARGET_MODE_REP_EXTENDED
7548 @hook TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P
7550 @defmac STORE_FLAG_VALUE
7551 A C expression describing the value returned by a comparison operator
7552 with an integral mode and stored by a store-flag instruction
7553 (@samp{cstore@var{mode}4}) when the condition is true. This description must
7554 apply to @emph{all} the @samp{cstore@var{mode}4} patterns and all the
7555 comparison operators whose results have a @code{MODE_INT} mode.
7557 A value of 1 or @minus{}1 means that the instruction implementing the
7558 comparison operator returns exactly 1 or @minus{}1 when the comparison is true
7559 and 0 when the comparison is false. Otherwise, the value indicates
7560 which bits of the result are guaranteed to be 1 when the comparison is
7561 true. This value is interpreted in the mode of the comparison
7562 operation, which is given by the mode of the first operand in the
7563 @samp{cstore@var{mode}4} pattern. Either the low bit or the sign bit of
7564 @code{STORE_FLAG_VALUE} be on. Presently, only those bits are used by
7567 If @code{STORE_FLAG_VALUE} is neither 1 or @minus{}1, the compiler will
7568 generate code that depends only on the specified bits. It can also
7569 replace comparison operators with equivalent operations if they cause
7570 the required bits to be set, even if the remaining bits are undefined.
7571 For example, on a machine whose comparison operators return an
7572 @code{SImode} value and where @code{STORE_FLAG_VALUE} is defined as
7573 @samp{0x80000000}, saying that just the sign bit is relevant, the
7577 (ne:SI (and:SI @var{x} (const_int @var{power-of-2})) (const_int 0))
7584 (ashift:SI @var{x} (const_int @var{n}))
7588 where @var{n} is the appropriate shift count to move the bit being
7589 tested into the sign bit.
7591 There is no way to describe a machine that always sets the low-order bit
7592 for a true value, but does not guarantee the value of any other bits,
7593 but we do not know of any machine that has such an instruction. If you
7594 are trying to port GCC to such a machine, include an instruction to
7595 perform a logical-and of the result with 1 in the pattern for the
7596 comparison operators and let us know at @email{gcc@@gcc.gnu.org}.
7598 Often, a machine will have multiple instructions that obtain a value
7599 from a comparison (or the condition codes). Here are rules to guide the
7600 choice of value for @code{STORE_FLAG_VALUE}, and hence the instructions
7605 Use the shortest sequence that yields a valid definition for
7606 @code{STORE_FLAG_VALUE}. It is more efficient for the compiler to
7607 ``normalize'' the value (convert it to, e.g., 1 or 0) than for the
7608 comparison operators to do so because there may be opportunities to
7609 combine the normalization with other operations.
7612 For equal-length sequences, use a value of 1 or @minus{}1, with @minus{}1 being
7613 slightly preferred on machines with expensive jumps and 1 preferred on
7617 As a second choice, choose a value of @samp{0x80000001} if instructions
7618 exist that set both the sign and low-order bits but do not define the
7622 Otherwise, use a value of @samp{0x80000000}.
7625 Many machines can produce both the value chosen for
7626 @code{STORE_FLAG_VALUE} and its negation in the same number of
7627 instructions. On those machines, you should also define a pattern for
7628 those cases, e.g., one matching
7631 (set @var{A} (neg:@var{m} (ne:@var{m} @var{B} @var{C})))
7634 Some machines can also perform @code{and} or @code{plus} operations on
7635 condition code values with less instructions than the corresponding
7636 @samp{cstore@var{mode}4} insn followed by @code{and} or @code{plus}. On those
7637 machines, define the appropriate patterns. Use the names @code{incscc}
7638 and @code{decscc}, respectively, for the patterns which perform
7639 @code{plus} or @code{minus} operations on condition code values. See
7640 @file{rs6000.md} for some examples. The GNU Superoptimizer can be used to
7641 find such instruction sequences on other machines.
7643 If this macro is not defined, the default value, 1, is used. You need
7644 not define @code{STORE_FLAG_VALUE} if the machine has no store-flag
7645 instructions, or if the value generated by these instructions is 1.
7648 @defmac FLOAT_STORE_FLAG_VALUE (@var{mode})
7649 A C expression that gives a nonzero @code{REAL_VALUE_TYPE} value that is
7650 returned when comparison operators with floating-point results are true.
7651 Define this macro on machines that have comparison operations that return
7652 floating-point values. If there are no such operations, do not define
7656 @defmac VECTOR_STORE_FLAG_VALUE (@var{mode})
7657 A C expression that gives a rtx representing the nonzero true element
7658 for vector comparisons. The returned rtx should be valid for the inner
7659 mode of @var{mode} which is guaranteed to be a vector mode. Define
7660 this macro on machines that have vector comparison operations that
7661 return a vector result. If there are no such operations, do not define
7662 this macro. Typically, this macro is defined as @code{const1_rtx} or
7663 @code{constm1_rtx}. This macro may return @code{NULL_RTX} to prevent
7664 the compiler optimizing such vector comparison operations for the
7668 @defmac CLZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
7669 @defmacx CTZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
7670 A C expression that indicates whether the architecture defines a value
7671 for @code{clz} or @code{ctz} with a zero operand.
7672 A result of @code{0} indicates the value is undefined.
7673 If the value is defined for only the RTL expression, the macro should
7674 evaluate to @code{1}; if the value applies also to the corresponding optab
7675 entry (which is normally the case if it expands directly into
7676 the corresponding RTL), then the macro should evaluate to @code{2}.
7677 In the cases where the value is defined, @var{value} should be set to
7680 If this macro is not defined, the value of @code{clz} or
7681 @code{ctz} at zero is assumed to be undefined.
7683 This macro must be defined if the target's expansion for @code{ffs}
7684 relies on a particular value to get correct results. Otherwise it
7685 is not necessary, though it may be used to optimize some corner cases, and
7686 to provide a default expansion for the @code{ffs} optab.
7688 Note that regardless of this macro the ``definedness'' of @code{clz}
7689 and @code{ctz} at zero do @emph{not} extend to the builtin functions
7690 visible to the user. Thus one may be free to adjust the value at will
7691 to match the target expansion of these operations without fear of
7696 An alias for the machine mode for pointers. On most machines, define
7697 this to be the integer mode corresponding to the width of a hardware
7698 pointer; @code{SImode} on 32-bit machine or @code{DImode} on 64-bit machines.
7699 On some machines you must define this to be one of the partial integer
7700 modes, such as @code{PSImode}.
7702 The width of @code{Pmode} must be at least as large as the value of
7703 @code{POINTER_SIZE}. If it is not equal, you must define the macro
7704 @code{POINTERS_EXTEND_UNSIGNED} to specify how pointers are extended
7708 @defmac FUNCTION_MODE
7709 An alias for the machine mode used for memory references to functions
7710 being called, in @code{call} RTL expressions. On most CISC machines,
7711 where an instruction can begin at any byte address, this should be
7712 @code{QImode}. On most RISC machines, where all instructions have fixed
7713 size and alignment, this should be a mode with the same size and alignment
7714 as the machine instruction words - typically @code{SImode} or @code{HImode}.
7717 @defmac STDC_0_IN_SYSTEM_HEADERS
7718 In normal operation, the preprocessor expands @code{__STDC__} to the
7719 constant 1, to signify that GCC conforms to ISO Standard C@. On some
7720 hosts, like Solaris, the system compiler uses a different convention,
7721 where @code{__STDC__} is normally 0, but is 1 if the user specifies
7722 strict conformance to the C Standard.
7724 Defining @code{STDC_0_IN_SYSTEM_HEADERS} makes GNU CPP follows the host
7725 convention when processing system header files, but when processing user
7726 files @code{__STDC__} will always expand to 1.
7729 @hook TARGET_C_PREINCLUDE
7731 @hook TARGET_CXX_IMPLICIT_EXTERN_C
7733 @defmac SYSTEM_IMPLICIT_EXTERN_C
7734 Define this macro if the system header files do not support C++@.
7735 This macro handles system header files by pretending that system
7736 header files are enclosed in @samp{extern "C" @{@dots{}@}}.
7741 @defmac REGISTER_TARGET_PRAGMAS ()
7742 Define this macro if you want to implement any target-specific pragmas.
7743 If defined, it is a C expression which makes a series of calls to
7744 @code{c_register_pragma} or @code{c_register_pragma_with_expansion}
7745 for each pragma. The macro may also do any
7746 setup required for the pragmas.
7748 The primary reason to define this macro is to provide compatibility with
7749 other compilers for the same target. In general, we discourage
7750 definition of target-specific pragmas for GCC@.
7752 If the pragma can be implemented by attributes then you should consider
7753 defining the target hook @samp{TARGET_INSERT_ATTRIBUTES} as well.
7755 Preprocessor macros that appear on pragma lines are not expanded. All
7756 @samp{#pragma} directives that do not match any registered pragma are
7757 silently ignored, unless the user specifies @option{-Wunknown-pragmas}.
7760 @deftypefun void c_register_pragma (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
7761 @deftypefunx void c_register_pragma_with_expansion (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
7763 Each call to @code{c_register_pragma} or
7764 @code{c_register_pragma_with_expansion} establishes one pragma. The
7765 @var{callback} routine will be called when the preprocessor encounters a
7769 #pragma [@var{space}] @var{name} @dots{}
7772 @var{space} is the case-sensitive namespace of the pragma, or
7773 @code{NULL} to put the pragma in the global namespace. The callback
7774 routine receives @var{pfile} as its first argument, which can be passed
7775 on to cpplib's functions if necessary. You can lex tokens after the
7776 @var{name} by calling @code{pragma_lex}. Tokens that are not read by the
7777 callback will be silently ignored. The end of the line is indicated by
7778 a token of type @code{CPP_EOF}. Macro expansion occurs on the
7779 arguments of pragmas registered with
7780 @code{c_register_pragma_with_expansion} but not on the arguments of
7781 pragmas registered with @code{c_register_pragma}.
7783 Note that the use of @code{pragma_lex} is specific to the C and C++
7784 compilers. It will not work in the Java or Fortran compilers, or any
7785 other language compilers for that matter. Thus if @code{pragma_lex} is going
7786 to be called from target-specific code, it must only be done so when
7787 building the C and C++ compilers. This can be done by defining the
7788 variables @code{c_target_objs} and @code{cxx_target_objs} in the
7789 target entry in the @file{config.gcc} file. These variables should name
7790 the target-specific, language-specific object file which contains the
7791 code that uses @code{pragma_lex}. Note it will also be necessary to add a
7792 rule to the makefile fragment pointed to by @code{tmake_file} that shows
7793 how to build this object file.
7796 @defmac HANDLE_PRAGMA_PACK_WITH_EXPANSION
7797 Define this macro if macros should be expanded in the
7798 arguments of @samp{#pragma pack}.
7801 @defmac TARGET_DEFAULT_PACK_STRUCT
7802 If your target requires a structure packing default other than 0 (meaning
7803 the machine default), define this macro to the necessary value (in bytes).
7804 This must be a value that would also be valid to use with
7805 @samp{#pragma pack()} (that is, a small power of two).
7808 @defmac DOLLARS_IN_IDENTIFIERS
7809 Define this macro to control use of the character @samp{$} in
7810 identifier names for the C family of languages. 0 means @samp{$} is
7811 not allowed by default; 1 means it is allowed. 1 is the default;
7812 there is no need to define this macro in that case.
7815 @defmac INSN_SETS_ARE_DELAYED (@var{insn})
7816 Define this macro as a C expression that is nonzero if it is safe for the
7817 delay slot scheduler to place instructions in the delay slot of @var{insn},
7818 even if they appear to use a resource set or clobbered in @var{insn}.
7819 @var{insn} is always a @code{jump_insn} or an @code{insn}; GCC knows that
7820 every @code{call_insn} has this behavior. On machines where some @code{insn}
7821 or @code{jump_insn} is really a function call and hence has this behavior,
7822 you should define this macro.
7824 You need not define this macro if it would always return zero.
7827 @defmac INSN_REFERENCES_ARE_DELAYED (@var{insn})
7828 Define this macro as a C expression that is nonzero if it is safe for the
7829 delay slot scheduler to place instructions in the delay slot of @var{insn},
7830 even if they appear to set or clobber a resource referenced in @var{insn}.
7831 @var{insn} is always a @code{jump_insn} or an @code{insn}. On machines where
7832 some @code{insn} or @code{jump_insn} is really a function call and its operands
7833 are registers whose use is actually in the subroutine it calls, you should
7834 define this macro. Doing so allows the delay slot scheduler to move
7835 instructions which copy arguments into the argument registers into the delay
7838 You need not define this macro if it would always return zero.
7841 @defmac MULTIPLE_SYMBOL_SPACES
7842 Define this macro as a C expression that is nonzero if, in some cases,
7843 global symbols from one translation unit may not be bound to undefined
7844 symbols in another translation unit without user intervention. For
7845 instance, under Microsoft Windows symbols must be explicitly imported
7846 from shared libraries (DLLs).
7848 You need not define this macro if it would always evaluate to zero.
7851 @hook TARGET_MD_ASM_ADJUST
7853 @defmac MATH_LIBRARY
7854 Define this macro as a C string constant for the linker argument to link
7855 in the system math library, minus the initial @samp{"-l"}, or
7856 @samp{""} if the target does not have a
7857 separate math library.
7859 You need only define this macro if the default of @samp{"m"} is wrong.
7862 @defmac LIBRARY_PATH_ENV
7863 Define this macro as a C string constant for the environment variable that
7864 specifies where the linker should look for libraries.
7866 You need only define this macro if the default of @samp{"LIBRARY_PATH"}
7870 @defmac TARGET_POSIX_IO
7871 Define this macro if the target supports the following POSIX@ file
7872 functions, access, mkdir and file locking with fcntl / F_SETLKW@.
7873 Defining @code{TARGET_POSIX_IO} will enable the test coverage code
7874 to use file locking when exiting a program, which avoids race conditions
7875 if the program has forked. It will also create directories at run-time
7876 for cross-profiling.
7879 @defmac MAX_CONDITIONAL_EXECUTE
7881 A C expression for the maximum number of instructions to execute via
7882 conditional execution instructions instead of a branch. A value of
7883 @code{BRANCH_COST}+1 is the default if the machine does not use cc0, and
7884 1 if it does use cc0.
7887 @defmac IFCVT_MODIFY_TESTS (@var{ce_info}, @var{true_expr}, @var{false_expr})
7888 Used if the target needs to perform machine-dependent modifications on the
7889 conditionals used for turning basic blocks into conditionally executed code.
7890 @var{ce_info} points to a data structure, @code{struct ce_if_block}, which
7891 contains information about the currently processed blocks. @var{true_expr}
7892 and @var{false_expr} are the tests that are used for converting the
7893 then-block and the else-block, respectively. Set either @var{true_expr} or
7894 @var{false_expr} to a null pointer if the tests cannot be converted.
7897 @defmac IFCVT_MODIFY_MULTIPLE_TESTS (@var{ce_info}, @var{bb}, @var{true_expr}, @var{false_expr})
7898 Like @code{IFCVT_MODIFY_TESTS}, but used when converting more complicated
7899 if-statements into conditions combined by @code{and} and @code{or} operations.
7900 @var{bb} contains the basic block that contains the test that is currently
7901 being processed and about to be turned into a condition.
7904 @defmac IFCVT_MODIFY_INSN (@var{ce_info}, @var{pattern}, @var{insn})
7905 A C expression to modify the @var{PATTERN} of an @var{INSN} that is to
7906 be converted to conditional execution format. @var{ce_info} points to
7907 a data structure, @code{struct ce_if_block}, which contains information
7908 about the currently processed blocks.
7911 @defmac IFCVT_MODIFY_FINAL (@var{ce_info})
7912 A C expression to perform any final machine dependent modifications in
7913 converting code to conditional execution. The involved basic blocks
7914 can be found in the @code{struct ce_if_block} structure that is pointed
7915 to by @var{ce_info}.
7918 @defmac IFCVT_MODIFY_CANCEL (@var{ce_info})
7919 A C expression to cancel any machine dependent modifications in
7920 converting code to conditional execution. The involved basic blocks
7921 can be found in the @code{struct ce_if_block} structure that is pointed
7922 to by @var{ce_info}.
7925 @defmac IFCVT_MACHDEP_INIT (@var{ce_info})
7926 A C expression to initialize any machine specific data for if-conversion
7927 of the if-block in the @code{struct ce_if_block} structure that is pointed
7928 to by @var{ce_info}.
7931 @hook TARGET_MACHINE_DEPENDENT_REORG
7933 @hook TARGET_INIT_BUILTINS
7935 @hook TARGET_BUILTIN_DECL
7937 @hook TARGET_EXPAND_BUILTIN
7939 @hook TARGET_RESOLVE_OVERLOADED_BUILTIN
7941 @hook TARGET_CHECK_BUILTIN_CALL
7943 @hook TARGET_FOLD_BUILTIN
7945 @hook TARGET_GIMPLE_FOLD_BUILTIN
7947 @hook TARGET_COMPARE_VERSION_PRIORITY
7949 @hook TARGET_GET_FUNCTION_VERSIONS_DISPATCHER
7951 @hook TARGET_GENERATE_VERSION_DISPATCHER_BODY
7953 @hook TARGET_PREDICT_DOLOOP_P
7955 @hook TARGET_HAVE_COUNT_REG_DECR_P
7957 @hook TARGET_DOLOOP_COST_FOR_GENERIC
7959 @hook TARGET_DOLOOP_COST_FOR_ADDRESS
7961 @hook TARGET_CAN_USE_DOLOOP_P
7963 @hook TARGET_INVALID_WITHIN_DOLOOP
7965 @hook TARGET_LEGITIMATE_COMBINED_INSN
7967 @hook TARGET_CAN_FOLLOW_JUMP
7969 @hook TARGET_COMMUTATIVE_P
7971 @hook TARGET_ALLOCATE_INITIAL_VALUE
7973 @hook TARGET_UNSPEC_MAY_TRAP_P
7975 @hook TARGET_SET_CURRENT_FUNCTION
7977 @defmac TARGET_OBJECT_SUFFIX
7978 Define this macro to be a C string representing the suffix for object
7979 files on your target machine. If you do not define this macro, GCC will
7980 use @samp{.o} as the suffix for object files.
7983 @defmac TARGET_EXECUTABLE_SUFFIX
7984 Define this macro to be a C string representing the suffix to be
7985 automatically added to executable files on your target machine. If you
7986 do not define this macro, GCC will use the null string as the suffix for
7990 @defmac COLLECT_EXPORT_LIST
7991 If defined, @code{collect2} will scan the individual object files
7992 specified on its command line and create an export list for the linker.
7993 Define this macro for systems like AIX, where the linker discards
7994 object files that are not referenced from @code{main} and uses export
7998 @defmac MODIFY_JNI_METHOD_CALL (@var{mdecl})
7999 Define this macro to a C expression representing a variant of the
8000 method call @var{mdecl}, if Java Native Interface (JNI) methods
8001 must be invoked differently from other methods on your target.
8002 For example, on 32-bit Microsoft Windows, JNI methods must be invoked using
8003 the @code{stdcall} calling convention and this macro is then
8004 defined as this expression:
8007 build_type_attribute_variant (@var{mdecl},
8009 (get_identifier ("stdcall"),
8014 @hook TARGET_CANNOT_MODIFY_JUMPS_P
8016 @hook TARGET_HAVE_CONDITIONAL_EXECUTION
8018 @hook TARGET_GEN_CCMP_FIRST
8020 @hook TARGET_GEN_CCMP_NEXT
8022 @hook TARGET_LOOP_UNROLL_ADJUST
8024 @defmac POWI_MAX_MULTS
8025 If defined, this macro is interpreted as a signed integer C expression
8026 that specifies the maximum number of floating point multiplications
8027 that should be emitted when expanding exponentiation by an integer
8028 constant inline. When this value is defined, exponentiation requiring
8029 more than this number of multiplications is implemented by calling the
8030 system library's @code{pow}, @code{powf} or @code{powl} routines.
8031 The default value places no upper bound on the multiplication count.
8034 @deftypefn Macro void TARGET_EXTRA_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
8035 This target hook should register any extra include files for the
8036 target. The parameter @var{stdinc} indicates if normal include files
8037 are present. The parameter @var{sysroot} is the system root directory.
8038 The parameter @var{iprefix} is the prefix for the gcc directory.
8041 @deftypefn Macro void TARGET_EXTRA_PRE_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
8042 This target hook should register any extra include files for the
8043 target before any standard headers. The parameter @var{stdinc}
8044 indicates if normal include files are present. The parameter
8045 @var{sysroot} is the system root directory. The parameter
8046 @var{iprefix} is the prefix for the gcc directory.
8049 @deftypefn Macro void TARGET_OPTF (char *@var{path})
8050 This target hook should register special include paths for the target.
8051 The parameter @var{path} is the include to register. On Darwin
8052 systems, this is used for Framework includes, which have semantics
8053 that are different from @option{-I}.
8056 @defmac bool TARGET_USE_LOCAL_THUNK_ALIAS_P (tree @var{fndecl})
8057 This target macro returns @code{true} if it is safe to use a local alias
8058 for a virtual function @var{fndecl} when constructing thunks,
8059 @code{false} otherwise. By default, the macro returns @code{true} for all
8060 functions, if a target supports aliases (i.e.@: defines
8061 @code{ASM_OUTPUT_DEF}), @code{false} otherwise,
8064 @defmac TARGET_FORMAT_TYPES
8065 If defined, this macro is the name of a global variable containing
8066 target-specific format checking information for the @option{-Wformat}
8067 option. The default is to have no target-specific format checks.
8070 @defmac TARGET_N_FORMAT_TYPES
8071 If defined, this macro is the number of entries in
8072 @code{TARGET_FORMAT_TYPES}.
8075 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES
8076 If defined, this macro is the name of a global variable containing
8077 target-specific format overrides for the @option{-Wformat} option. The
8078 default is to have no target-specific format overrides. If defined,
8079 @code{TARGET_FORMAT_TYPES} must be defined, too.
8082 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES_COUNT
8083 If defined, this macro specifies the number of entries in
8084 @code{TARGET_OVERRIDES_FORMAT_ATTRIBUTES}.
8087 @defmac TARGET_OVERRIDES_FORMAT_INIT
8088 If defined, this macro specifies the optional initialization
8089 routine for target specific customizations of the system printf
8090 and scanf formatter settings.
8093 @hook TARGET_INVALID_ARG_FOR_UNPROTOTYPED_FN
8095 @hook TARGET_INVALID_CONVERSION
8097 @hook TARGET_INVALID_UNARY_OP
8099 @hook TARGET_INVALID_BINARY_OP
8101 @hook TARGET_PROMOTED_TYPE
8103 @hook TARGET_CONVERT_TO_TYPE
8106 This macro determines the size of the objective C jump buffer for the
8107 NeXT runtime. By default, OBJC_JBLEN is defined to an innocuous value.
8110 @defmac LIBGCC2_UNWIND_ATTRIBUTE
8111 Define this macro if any target-specific attributes need to be attached
8112 to the functions in @file{libgcc} that provide low-level support for
8113 call stack unwinding. It is used in declarations in @file{unwind-generic.h}
8114 and the associated definitions of those functions.
8117 @hook TARGET_UPDATE_STACK_BOUNDARY
8119 @hook TARGET_GET_DRAP_RTX
8121 @hook TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS
8123 @hook TARGET_CONST_ANCHOR
8125 @hook TARGET_ASAN_SHADOW_OFFSET
8127 @hook TARGET_MEMMODEL_CHECK
8129 @hook TARGET_ATOMIC_TEST_AND_SET_TRUEVAL
8131 @hook TARGET_HAS_IFUNC_P
8133 @hook TARGET_ATOMIC_ALIGN_FOR_MODE
8135 @hook TARGET_ATOMIC_ASSIGN_EXPAND_FENV
8137 @hook TARGET_RECORD_OFFLOAD_SYMBOL
8139 @hook TARGET_OFFLOAD_OPTIONS
8141 @defmac TARGET_SUPPORTS_WIDE_INT
8143 On older ports, large integers are stored in @code{CONST_DOUBLE} rtl
8144 objects. Newer ports define @code{TARGET_SUPPORTS_WIDE_INT} to be nonzero
8145 to indicate that large integers are stored in
8146 @code{CONST_WIDE_INT} rtl objects. The @code{CONST_WIDE_INT} allows
8147 very large integer constants to be represented. @code{CONST_DOUBLE}
8148 is limited to twice the size of the host's @code{HOST_WIDE_INT}
8151 Converting a port mostly requires looking for the places where
8152 @code{CONST_DOUBLE}s are used with @code{VOIDmode} and replacing that
8153 code with code that accesses @code{CONST_WIDE_INT}s. @samp{"grep -i
8154 const_double"} at the port level gets you to 95% of the changes that
8155 need to be made. There are a few places that require a deeper look.
8159 There is no equivalent to @code{hval} and @code{lval} for
8160 @code{CONST_WIDE_INT}s. This would be difficult to express in the md
8161 language since there are a variable number of elements.
8163 Most ports only check that @code{hval} is either 0 or -1 to see if the
8164 value is small. As mentioned above, this will no longer be necessary
8165 since small constants are always @code{CONST_INT}. Of course there
8166 are still a few exceptions, the alpha's constraint used by the zap
8167 instruction certainly requires careful examination by C code.
8168 However, all the current code does is pass the hval and lval to C
8169 code, so evolving the c code to look at the @code{CONST_WIDE_INT} is
8170 not really a large change.
8173 Because there is no standard template that ports use to materialize
8174 constants, there is likely to be some futzing that is unique to each
8178 The rtx costs may have to be adjusted to properly account for larger
8179 constants that are represented as @code{CONST_WIDE_INT}.
8182 All and all it does not take long to convert ports that the
8183 maintainer is familiar with.
8187 @hook TARGET_HAVE_SPECULATION_SAFE_VALUE
8189 @hook TARGET_SPECULATION_SAFE_VALUE
8191 @hook TARGET_RUN_TARGET_SELFTESTS