1 @c Copyright (C) 1988,1989,1992,1993,1994,1995,1996,1997,1998,1999,2000,2001,
2 @c 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012
3 @c Free Software Foundation, Inc.
4 @c This is part of the GCC manual.
5 @c For copying conditions, see the file gcc.texi.
8 @chapter Target Description Macros and Functions
9 @cindex machine description macros
10 @cindex target description macros
11 @cindex macros, target description
12 @cindex @file{tm.h} macros
14 In addition to the file @file{@var{machine}.md}, a machine description
15 includes a C header file conventionally given the name
16 @file{@var{machine}.h} and a C source file named @file{@var{machine}.c}.
17 The header file defines numerous macros that convey the information
18 about the target machine that does not fit into the scheme of the
19 @file{.md} file. The file @file{tm.h} should be a link to
20 @file{@var{machine}.h}. The header file @file{config.h} includes
21 @file{tm.h} and most compiler source files include @file{config.h}. The
22 source file defines a variable @code{targetm}, which is a structure
23 containing pointers to functions and data relating to the target
24 machine. @file{@var{machine}.c} should also contain their definitions,
25 if they are not defined elsewhere in GCC, and other functions called
26 through the macros defined in the @file{.h} file.
29 * Target Structure:: The @code{targetm} variable.
30 * Driver:: Controlling how the driver runs the compilation passes.
31 * Run-time Target:: Defining @samp{-m} options like @option{-m68000} and @option{-m68020}.
32 * Per-Function Data:: Defining data structures for per-function information.
33 * Storage Layout:: Defining sizes and alignments of data.
34 * Type Layout:: Defining sizes and properties of basic user data types.
35 * Registers:: Naming and describing the hardware registers.
36 * Register Classes:: Defining the classes of hardware registers.
37 * Old Constraints:: The old way to define machine-specific constraints.
38 * Stack and Calling:: Defining which way the stack grows and by how much.
39 * Varargs:: Defining the varargs macros.
40 * Trampolines:: Code set up at run time to enter a nested function.
41 * Library Calls:: Controlling how library routines are implicitly called.
42 * Addressing Modes:: Defining addressing modes valid for memory operands.
43 * Anchored Addresses:: Defining how @option{-fsection-anchors} should work.
44 * Condition Code:: Defining how insns update the condition code.
45 * Costs:: Defining relative costs of different operations.
46 * Scheduling:: Adjusting the behavior of the instruction scheduler.
47 * Sections:: Dividing storage into text, data, and other sections.
48 * PIC:: Macros for position independent code.
49 * Assembler Format:: Defining how to write insns and pseudo-ops to output.
50 * Debugging Info:: Defining the format of debugging output.
51 * Floating Point:: Handling floating point for cross-compilers.
52 * Mode Switching:: Insertion of mode-switching instructions.
53 * Target Attributes:: Defining target-specific uses of @code{__attribute__}.
54 * Emulated TLS:: Emulated TLS support.
55 * MIPS Coprocessors:: MIPS coprocessor support and how to customize it.
56 * PCH Target:: Validity checking for precompiled headers.
57 * C++ ABI:: Controlling C++ ABI changes.
58 * Named Address Spaces:: Adding support for named address spaces
59 * Misc:: Everything else.
62 @node Target Structure
63 @section The Global @code{targetm} Variable
65 @cindex target functions
67 @deftypevar {struct gcc_target} targetm
68 The target @file{.c} file must define the global @code{targetm} variable
69 which contains pointers to functions and data relating to the target
70 machine. The variable is declared in @file{target.h};
71 @file{target-def.h} defines the macro @code{TARGET_INITIALIZER} which is
72 used to initialize the variable, and macros for the default initializers
73 for elements of the structure. The @file{.c} file should override those
74 macros for which the default definition is inappropriate. For example:
77 #include "target-def.h"
79 /* @r{Initialize the GCC target structure.} */
81 #undef TARGET_COMP_TYPE_ATTRIBUTES
82 #define TARGET_COMP_TYPE_ATTRIBUTES @var{machine}_comp_type_attributes
84 struct gcc_target targetm = TARGET_INITIALIZER;
88 Where a macro should be defined in the @file{.c} file in this manner to
89 form part of the @code{targetm} structure, it is documented below as a
90 ``Target Hook'' with a prototype. Many macros will change in future
91 from being defined in the @file{.h} file to being part of the
92 @code{targetm} structure.
94 Similarly, there is a @code{targetcm} variable for hooks that are
95 specific to front ends for C-family languages, documented as ``C
96 Target Hook''. This is declared in @file{c-family/c-target.h}, the
97 initializer @code{TARGETCM_INITIALIZER} in
98 @file{c-family/c-target-def.h}. If targets initialize @code{targetcm}
99 themselves, they should set @code{target_has_targetcm=yes} in
100 @file{config.gcc}; otherwise a default definition is used.
102 Similarly, there is a @code{targetm_common} variable for hooks that
103 are shared between the compiler driver and the compilers proper,
104 documented as ``Common Target Hook''. This is declared in
105 @file{common/common-target.h}, the initializer
106 @code{TARGETM_COMMON_INITIALIZER} in
107 @file{common/common-target-def.h}. If targets initialize
108 @code{targetm_common} themselves, they should set
109 @code{target_has_targetm_common=yes} in @file{config.gcc}; otherwise a
110 default definition is used.
113 @section Controlling the Compilation Driver, @file{gcc}
115 @cindex controlling the compilation driver
117 @c prevent bad page break with this line
118 You can control the compilation driver.
120 @defmac DRIVER_SELF_SPECS
121 A list of specs for the driver itself. It should be a suitable
122 initializer for an array of strings, with no surrounding braces.
124 The driver applies these specs to its own command line between loading
125 default @file{specs} files (but not command-line specified ones) and
126 choosing the multilib directory or running any subcommands. It
127 applies them in the order given, so each spec can depend on the
128 options added by earlier ones. It is also possible to remove options
129 using @samp{%<@var{option}} in the usual way.
131 This macro can be useful when a port has several interdependent target
132 options. It provides a way of standardizing the command line so
133 that the other specs are easier to write.
135 Do not define this macro if it does not need to do anything.
138 @defmac OPTION_DEFAULT_SPECS
139 A list of specs used to support configure-time default options (i.e.@:
140 @option{--with} options) in the driver. It should be a suitable initializer
141 for an array of structures, each containing two strings, without the
142 outermost pair of surrounding braces.
144 The first item in the pair is the name of the default. This must match
145 the code in @file{config.gcc} for the target. The second item is a spec
146 to apply if a default with this name was specified. The string
147 @samp{%(VALUE)} in the spec will be replaced by the value of the default
148 everywhere it occurs.
150 The driver will apply these specs to its own command line between loading
151 default @file{specs} files and processing @code{DRIVER_SELF_SPECS}, using
152 the same mechanism as @code{DRIVER_SELF_SPECS}.
154 Do not define this macro if it does not need to do anything.
158 A C string constant that tells the GCC driver program options to
159 pass to CPP@. It can also specify how to translate options you
160 give to GCC into options for GCC to pass to the CPP@.
162 Do not define this macro if it does not need to do anything.
165 @defmac CPLUSPLUS_CPP_SPEC
166 This macro is just like @code{CPP_SPEC}, but is used for C++, rather
167 than C@. If you do not define this macro, then the value of
168 @code{CPP_SPEC} (if any) will be used instead.
172 A C string constant that tells the GCC driver program options to
173 pass to @code{cc1}, @code{cc1plus}, @code{f771}, and the other language
175 It can also specify how to translate options you give to GCC into options
176 for GCC to pass to front ends.
178 Do not define this macro if it does not need to do anything.
182 A C string constant that tells the GCC driver program options to
183 pass to @code{cc1plus}. It can also specify how to translate options you
184 give to GCC into options for GCC to pass to the @code{cc1plus}.
186 Do not define this macro if it does not need to do anything.
187 Note that everything defined in CC1_SPEC is already passed to
188 @code{cc1plus} so there is no need to duplicate the contents of
189 CC1_SPEC in CC1PLUS_SPEC@.
193 A C string constant that tells the GCC driver program options to
194 pass to the assembler. It can also specify how to translate options
195 you give to GCC into options for GCC to pass to the assembler.
196 See the file @file{sun3.h} for an example of this.
198 Do not define this macro if it does not need to do anything.
201 @defmac ASM_FINAL_SPEC
202 A C string constant that tells the GCC driver program how to
203 run any programs which cleanup after the normal assembler.
204 Normally, this is not needed. See the file @file{mips.h} for
207 Do not define this macro if it does not need to do anything.
210 @defmac AS_NEEDS_DASH_FOR_PIPED_INPUT
211 Define this macro, with no value, if the driver should give the assembler
212 an argument consisting of a single dash, @option{-}, to instruct it to
213 read from its standard input (which will be a pipe connected to the
214 output of the compiler proper). This argument is given after any
215 @option{-o} option specifying the name of the output file.
217 If you do not define this macro, the assembler is assumed to read its
218 standard input if given no non-option arguments. If your assembler
219 cannot read standard input at all, use a @samp{%@{pipe:%e@}} construct;
220 see @file{mips.h} for instance.
224 A C string constant that tells the GCC driver program options to
225 pass to the linker. It can also specify how to translate options you
226 give to GCC into options for GCC to pass to the linker.
228 Do not define this macro if it does not need to do anything.
232 Another C string constant used much like @code{LINK_SPEC}. The difference
233 between the two is that @code{LIB_SPEC} is used at the end of the
234 command given to the linker.
236 If this macro is not defined, a default is provided that
237 loads the standard C library from the usual place. See @file{gcc.c}.
241 Another C string constant that tells the GCC driver program
242 how and when to place a reference to @file{libgcc.a} into the
243 linker command line. This constant is placed both before and after
244 the value of @code{LIB_SPEC}.
246 If this macro is not defined, the GCC driver provides a default that
247 passes the string @option{-lgcc} to the linker.
250 @defmac REAL_LIBGCC_SPEC
251 By default, if @code{ENABLE_SHARED_LIBGCC} is defined, the
252 @code{LIBGCC_SPEC} is not directly used by the driver program but is
253 instead modified to refer to different versions of @file{libgcc.a}
254 depending on the values of the command line flags @option{-static},
255 @option{-shared}, @option{-static-libgcc}, and @option{-shared-libgcc}. On
256 targets where these modifications are inappropriate, define
257 @code{REAL_LIBGCC_SPEC} instead. @code{REAL_LIBGCC_SPEC} tells the
258 driver how to place a reference to @file{libgcc} on the link command
259 line, but, unlike @code{LIBGCC_SPEC}, it is used unmodified.
262 @defmac USE_LD_AS_NEEDED
263 A macro that controls the modifications to @code{LIBGCC_SPEC}
264 mentioned in @code{REAL_LIBGCC_SPEC}. If nonzero, a spec will be
265 generated that uses --as-needed and the shared libgcc in place of the
266 static exception handler library, when linking without any of
267 @code{-static}, @code{-static-libgcc}, or @code{-shared-libgcc}.
271 If defined, this C string constant is added to @code{LINK_SPEC}.
272 When @code{USE_LD_AS_NEEDED} is zero or undefined, it also affects
273 the modifications to @code{LIBGCC_SPEC} mentioned in
274 @code{REAL_LIBGCC_SPEC}.
277 @defmac STARTFILE_SPEC
278 Another C string constant used much like @code{LINK_SPEC}. The
279 difference between the two is that @code{STARTFILE_SPEC} is used at
280 the very beginning of the command given to the linker.
282 If this macro is not defined, a default is provided that loads the
283 standard C startup file from the usual place. See @file{gcc.c}.
287 Another C string constant used much like @code{LINK_SPEC}. The
288 difference between the two is that @code{ENDFILE_SPEC} is used at
289 the very end of the command given to the linker.
291 Do not define this macro if it does not need to do anything.
294 @defmac THREAD_MODEL_SPEC
295 GCC @code{-v} will print the thread model GCC was configured to use.
296 However, this doesn't work on platforms that are multilibbed on thread
297 models, such as AIX 4.3. On such platforms, define
298 @code{THREAD_MODEL_SPEC} such that it evaluates to a string without
299 blanks that names one of the recognized thread models. @code{%*}, the
300 default value of this macro, will expand to the value of
301 @code{thread_file} set in @file{config.gcc}.
304 @defmac SYSROOT_SUFFIX_SPEC
305 Define this macro to add a suffix to the target sysroot when GCC is
306 configured with a sysroot. This will cause GCC to search for usr/lib,
307 et al, within sysroot+suffix.
310 @defmac SYSROOT_HEADERS_SUFFIX_SPEC
311 Define this macro to add a headers_suffix to the target sysroot when
312 GCC is configured with a sysroot. This will cause GCC to pass the
313 updated sysroot+headers_suffix to CPP, causing it to search for
314 usr/include, et al, within sysroot+headers_suffix.
318 Define this macro to provide additional specifications to put in the
319 @file{specs} file that can be used in various specifications like
322 The definition should be an initializer for an array of structures,
323 containing a string constant, that defines the specification name, and a
324 string constant that provides the specification.
326 Do not define this macro if it does not need to do anything.
328 @code{EXTRA_SPECS} is useful when an architecture contains several
329 related targets, which have various @code{@dots{}_SPECS} which are similar
330 to each other, and the maintainer would like one central place to keep
333 For example, the PowerPC System V.4 targets use @code{EXTRA_SPECS} to
334 define either @code{_CALL_SYSV} when the System V calling sequence is
335 used or @code{_CALL_AIX} when the older AIX-based calling sequence is
338 The @file{config/rs6000/rs6000.h} target file defines:
341 #define EXTRA_SPECS \
342 @{ "cpp_sysv_default", CPP_SYSV_DEFAULT @},
344 #define CPP_SYS_DEFAULT ""
347 The @file{config/rs6000/sysv.h} target file defines:
351 "%@{posix: -D_POSIX_SOURCE @} \
352 %@{mcall-sysv: -D_CALL_SYSV @} \
353 %@{!mcall-sysv: %(cpp_sysv_default) @} \
354 %@{msoft-float: -D_SOFT_FLOAT@} %@{mcpu=403: -D_SOFT_FLOAT@}"
356 #undef CPP_SYSV_DEFAULT
357 #define CPP_SYSV_DEFAULT "-D_CALL_SYSV"
360 while the @file{config/rs6000/eabiaix.h} target file defines
361 @code{CPP_SYSV_DEFAULT} as:
364 #undef CPP_SYSV_DEFAULT
365 #define CPP_SYSV_DEFAULT "-D_CALL_AIX"
369 @defmac LINK_LIBGCC_SPECIAL_1
370 Define this macro if the driver program should find the library
371 @file{libgcc.a}. If you do not define this macro, the driver program will pass
372 the argument @option{-lgcc} to tell the linker to do the search.
375 @defmac LINK_GCC_C_SEQUENCE_SPEC
376 The sequence in which libgcc and libc are specified to the linker.
377 By default this is @code{%G %L %G}.
380 @defmac LINK_COMMAND_SPEC
381 A C string constant giving the complete command line need to execute the
382 linker. When you do this, you will need to update your port each time a
383 change is made to the link command line within @file{gcc.c}. Therefore,
384 define this macro only if you need to completely redefine the command
385 line for invoking the linker and there is no other way to accomplish
386 the effect you need. Overriding this macro may be avoidable by overriding
387 @code{LINK_GCC_C_SEQUENCE_SPEC} instead.
390 @hook TARGET_ALWAYS_STRIP_DOTDOT
392 @defmac MULTILIB_DEFAULTS
393 Define this macro as a C expression for the initializer of an array of
394 string to tell the driver program which options are defaults for this
395 target and thus do not need to be handled specially when using
396 @code{MULTILIB_OPTIONS}.
398 Do not define this macro if @code{MULTILIB_OPTIONS} is not defined in
399 the target makefile fragment or if none of the options listed in
400 @code{MULTILIB_OPTIONS} are set by default.
401 @xref{Target Fragment}.
404 @defmac RELATIVE_PREFIX_NOT_LINKDIR
405 Define this macro to tell @command{gcc} that it should only translate
406 a @option{-B} prefix into a @option{-L} linker option if the prefix
407 indicates an absolute file name.
410 @defmac MD_EXEC_PREFIX
411 If defined, this macro is an additional prefix to try after
412 @code{STANDARD_EXEC_PREFIX}. @code{MD_EXEC_PREFIX} is not searched
413 when the compiler is built as a cross
414 compiler. If you define @code{MD_EXEC_PREFIX}, then be sure to add it
415 to the list of directories used to find the assembler in @file{configure.in}.
418 @defmac STANDARD_STARTFILE_PREFIX
419 Define this macro as a C string constant if you wish to override the
420 standard choice of @code{libdir} as the default prefix to
421 try when searching for startup files such as @file{crt0.o}.
422 @code{STANDARD_STARTFILE_PREFIX} is not searched when the compiler
423 is built as a cross compiler.
426 @defmac STANDARD_STARTFILE_PREFIX_1
427 Define this macro as a C string constant if you wish to override the
428 standard choice of @code{/lib} as a prefix to try after the default prefix
429 when searching for startup files such as @file{crt0.o}.
430 @code{STANDARD_STARTFILE_PREFIX_1} is not searched when the compiler
431 is built as a cross compiler.
434 @defmac STANDARD_STARTFILE_PREFIX_2
435 Define this macro as a C string constant if you wish to override the
436 standard choice of @code{/lib} as yet another prefix to try after the
437 default prefix when searching for startup files such as @file{crt0.o}.
438 @code{STANDARD_STARTFILE_PREFIX_2} is not searched when the compiler
439 is built as a cross compiler.
442 @defmac MD_STARTFILE_PREFIX
443 If defined, this macro supplies an additional prefix to try after the
444 standard prefixes. @code{MD_EXEC_PREFIX} is not searched when the
445 compiler is built as a cross compiler.
448 @defmac MD_STARTFILE_PREFIX_1
449 If defined, this macro supplies yet another prefix to try after the
450 standard prefixes. It is not searched when the compiler is built as a
454 @defmac INIT_ENVIRONMENT
455 Define this macro as a C string constant if you wish to set environment
456 variables for programs called by the driver, such as the assembler and
457 loader. The driver passes the value of this macro to @code{putenv} to
458 initialize the necessary environment variables.
461 @defmac LOCAL_INCLUDE_DIR
462 Define this macro as a C string constant if you wish to override the
463 standard choice of @file{/usr/local/include} as the default prefix to
464 try when searching for local header files. @code{LOCAL_INCLUDE_DIR}
465 comes before @code{NATIVE_SYSTEM_HEADER_DIR} (set in
466 @file{config.gcc}, normally @file{/usr/include}) in the search order.
468 Cross compilers do not search either @file{/usr/local/include} or its
472 @defmac NATIVE_SYSTEM_HEADER_COMPONENT
473 The ``component'' corresponding to @code{NATIVE_SYSTEM_HEADER_DIR}.
474 See @code{INCLUDE_DEFAULTS}, below, for the description of components.
475 If you do not define this macro, no component is used.
478 @defmac INCLUDE_DEFAULTS
479 Define this macro if you wish to override the entire default search path
480 for include files. For a native compiler, the default search path
481 usually consists of @code{GCC_INCLUDE_DIR}, @code{LOCAL_INCLUDE_DIR},
482 @code{GPLUSPLUS_INCLUDE_DIR}, and
483 @code{NATIVE_SYSTEM_HEADER_DIR}. In addition, @code{GPLUSPLUS_INCLUDE_DIR}
484 and @code{GCC_INCLUDE_DIR} are defined automatically by @file{Makefile},
485 and specify private search areas for GCC@. The directory
486 @code{GPLUSPLUS_INCLUDE_DIR} is used only for C++ programs.
488 The definition should be an initializer for an array of structures.
489 Each array element should have four elements: the directory name (a
490 string constant), the component name (also a string constant), a flag
491 for C++-only directories,
492 and a flag showing that the includes in the directory don't need to be
493 wrapped in @code{extern @samp{C}} when compiling C++. Mark the end of
494 the array with a null element.
496 The component name denotes what GNU package the include file is part of,
497 if any, in all uppercase letters. For example, it might be @samp{GCC}
498 or @samp{BINUTILS}. If the package is part of a vendor-supplied
499 operating system, code the component name as @samp{0}.
501 For example, here is the definition used for VAX/VMS:
504 #define INCLUDE_DEFAULTS \
506 @{ "GNU_GXX_INCLUDE:", "G++", 1, 1@}, \
507 @{ "GNU_CC_INCLUDE:", "GCC", 0, 0@}, \
508 @{ "SYS$SYSROOT:[SYSLIB.]", 0, 0, 0@}, \
515 Here is the order of prefixes tried for exec files:
519 Any prefixes specified by the user with @option{-B}.
522 The environment variable @code{GCC_EXEC_PREFIX} or, if @code{GCC_EXEC_PREFIX}
523 is not set and the compiler has not been installed in the configure-time
524 @var{prefix}, the location in which the compiler has actually been installed.
527 The directories specified by the environment variable @code{COMPILER_PATH}.
530 The macro @code{STANDARD_EXEC_PREFIX}, if the compiler has been installed
531 in the configured-time @var{prefix}.
534 The location @file{/usr/libexec/gcc/}, but only if this is a native compiler.
537 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
540 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
544 Here is the order of prefixes tried for startfiles:
548 Any prefixes specified by the user with @option{-B}.
551 The environment variable @code{GCC_EXEC_PREFIX} or its automatically determined
552 value based on the installed toolchain location.
555 The directories specified by the environment variable @code{LIBRARY_PATH}
556 (or port-specific name; native only, cross compilers do not use this).
559 The macro @code{STANDARD_EXEC_PREFIX}, but only if the toolchain is installed
560 in the configured @var{prefix} or this is a native compiler.
563 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
566 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
570 The macro @code{MD_STARTFILE_PREFIX}, if defined, but only if this is a
571 native compiler, or we have a target system root.
574 The macro @code{MD_STARTFILE_PREFIX_1}, if defined, but only if this is a
575 native compiler, or we have a target system root.
578 The macro @code{STANDARD_STARTFILE_PREFIX}, with any sysroot modifications.
579 If this path is relative it will be prefixed by @code{GCC_EXEC_PREFIX} and
580 the machine suffix or @code{STANDARD_EXEC_PREFIX} and the machine suffix.
583 The macro @code{STANDARD_STARTFILE_PREFIX_1}, but only if this is a native
584 compiler, or we have a target system root. The default for this macro is
588 The macro @code{STANDARD_STARTFILE_PREFIX_2}, but only if this is a native
589 compiler, or we have a target system root. The default for this macro is
593 @node Run-time Target
594 @section Run-time Target Specification
595 @cindex run-time target specification
596 @cindex predefined macros
597 @cindex target specifications
599 @c prevent bad page break with this line
600 Here are run-time target specifications.
602 @defmac TARGET_CPU_CPP_BUILTINS ()
603 This function-like macro expands to a block of code that defines
604 built-in preprocessor macros and assertions for the target CPU, using
605 the functions @code{builtin_define}, @code{builtin_define_std} and
606 @code{builtin_assert}. When the front end
607 calls this macro it provides a trailing semicolon, and since it has
608 finished command line option processing your code can use those
611 @code{builtin_assert} takes a string in the form you pass to the
612 command-line option @option{-A}, such as @code{cpu=mips}, and creates
613 the assertion. @code{builtin_define} takes a string in the form
614 accepted by option @option{-D} and unconditionally defines the macro.
616 @code{builtin_define_std} takes a string representing the name of an
617 object-like macro. If it doesn't lie in the user's namespace,
618 @code{builtin_define_std} defines it unconditionally. Otherwise, it
619 defines a version with two leading underscores, and another version
620 with two leading and trailing underscores, and defines the original
621 only if an ISO standard was not requested on the command line. For
622 example, passing @code{unix} defines @code{__unix}, @code{__unix__}
623 and possibly @code{unix}; passing @code{_mips} defines @code{__mips},
624 @code{__mips__} and possibly @code{_mips}, and passing @code{_ABI64}
625 defines only @code{_ABI64}.
627 You can also test for the C dialect being compiled. The variable
628 @code{c_language} is set to one of @code{clk_c}, @code{clk_cplusplus}
629 or @code{clk_objective_c}. Note that if we are preprocessing
630 assembler, this variable will be @code{clk_c} but the function-like
631 macro @code{preprocessing_asm_p()} will return true, so you might want
632 to check for that first. If you need to check for strict ANSI, the
633 variable @code{flag_iso} can be used. The function-like macro
634 @code{preprocessing_trad_p()} can be used to check for traditional
638 @defmac TARGET_OS_CPP_BUILTINS ()
639 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
640 and is used for the target operating system instead.
643 @defmac TARGET_OBJFMT_CPP_BUILTINS ()
644 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
645 and is used for the target object format. @file{elfos.h} uses this
646 macro to define @code{__ELF__}, so you probably do not need to define
650 @deftypevar {extern int} target_flags
651 This variable is declared in @file{options.h}, which is included before
652 any target-specific headers.
655 @hook TARGET_DEFAULT_TARGET_FLAGS
656 This variable specifies the initial value of @code{target_flags}.
657 Its default setting is 0.
660 @cindex optional hardware or system features
661 @cindex features, optional, in system conventions
663 @hook TARGET_HANDLE_OPTION
664 This hook is called whenever the user specifies one of the
665 target-specific options described by the @file{.opt} definition files
666 (@pxref{Options}). It has the opportunity to do some option-specific
667 processing and should return true if the option is valid. The default
668 definition does nothing but return true.
670 @var{decoded} specifies the option and its arguments. @var{opts} and
671 @var{opts_set} are the @code{gcc_options} structures to be used for
672 storing option state, and @var{loc} is the location at which the
673 option was passed (@code{UNKNOWN_LOCATION} except for options passed
677 @hook TARGET_HANDLE_C_OPTION
678 This target hook is called whenever the user specifies one of the
679 target-specific C language family options described by the @file{.opt}
680 definition files(@pxref{Options}). It has the opportunity to do some
681 option-specific processing and should return true if the option is
682 valid. The arguments are like for @code{TARGET_HANDLE_OPTION}. The
683 default definition does nothing but return false.
685 In general, you should use @code{TARGET_HANDLE_OPTION} to handle
686 options. However, if processing an option requires routines that are
687 only available in the C (and related language) front ends, then you
688 should use @code{TARGET_HANDLE_C_OPTION} instead.
691 @hook TARGET_OBJC_CONSTRUCT_STRING_OBJECT
693 @hook TARGET_OBJC_DECLARE_UNRESOLVED_CLASS_REFERENCE
695 @hook TARGET_OBJC_DECLARE_CLASS_DEFINITION
697 @hook TARGET_STRING_OBJECT_REF_TYPE_P
699 @hook TARGET_CHECK_STRING_OBJECT_FORMAT_ARG
701 @hook TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE
702 This target function is similar to the hook @code{TARGET_OPTION_OVERRIDE}
703 but is called when the optimize level is changed via an attribute or
704 pragma or when it is reset at the end of the code affected by the
705 attribute or pragma. It is not called at the beginning of compilation
706 when @code{TARGET_OPTION_OVERRIDE} is called so if you want to perform these
707 actions then, you should have @code{TARGET_OPTION_OVERRIDE} call
708 @code{TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE}.
711 @defmac C_COMMON_OVERRIDE_OPTIONS
712 This is similar to the @code{TARGET_OPTION_OVERRIDE} hook
713 but is only used in the C
714 language frontends (C, Objective-C, C++, Objective-C++) and so can be
715 used to alter option flag variables which only exist in those
719 @hook TARGET_OPTION_OPTIMIZATION_TABLE
720 Some machines may desire to change what optimizations are performed for
721 various optimization levels. This variable, if defined, describes
722 options to enable at particular sets of optimization levels. These
723 options are processed once
724 just after the optimization level is determined and before the remainder
725 of the command options have been parsed, so may be overridden by other
726 options passed explicitly.
728 This processing is run once at program startup and when the optimization
729 options are changed via @code{#pragma GCC optimize} or by using the
730 @code{optimize} attribute.
733 @hook TARGET_OPTION_INIT_STRUCT
735 @hook TARGET_OPTION_DEFAULT_PARAMS
737 @defmac SWITCHABLE_TARGET
738 Some targets need to switch between substantially different subtargets
739 during compilation. For example, the MIPS target has one subtarget for
740 the traditional MIPS architecture and another for MIPS16. Source code
741 can switch between these two subarchitectures using the @code{mips16}
742 and @code{nomips16} attributes.
744 Such subtargets can differ in things like the set of available
745 registers, the set of available instructions, the costs of various
746 operations, and so on. GCC caches a lot of this type of information
747 in global variables, and recomputing them for each subtarget takes a
748 significant amount of time. The compiler therefore provides a facility
749 for maintaining several versions of the global variables and quickly
750 switching between them; see @file{target-globals.h} for details.
752 Define this macro to 1 if your target needs this facility. The default
756 @node Per-Function Data
757 @section Defining data structures for per-function information.
758 @cindex per-function data
759 @cindex data structures
761 If the target needs to store information on a per-function basis, GCC
762 provides a macro and a couple of variables to allow this. Note, just
763 using statics to store the information is a bad idea, since GCC supports
764 nested functions, so you can be halfway through encoding one function
765 when another one comes along.
767 GCC defines a data structure called @code{struct function} which
768 contains all of the data specific to an individual function. This
769 structure contains a field called @code{machine} whose type is
770 @code{struct machine_function *}, which can be used by targets to point
771 to their own specific data.
773 If a target needs per-function specific data it should define the type
774 @code{struct machine_function} and also the macro @code{INIT_EXPANDERS}.
775 This macro should be used to initialize the function pointer
776 @code{init_machine_status}. This pointer is explained below.
778 One typical use of per-function, target specific data is to create an
779 RTX to hold the register containing the function's return address. This
780 RTX can then be used to implement the @code{__builtin_return_address}
781 function, for level 0.
783 Note---earlier implementations of GCC used a single data area to hold
784 all of the per-function information. Thus when processing of a nested
785 function began the old per-function data had to be pushed onto a
786 stack, and when the processing was finished, it had to be popped off the
787 stack. GCC used to provide function pointers called
788 @code{save_machine_status} and @code{restore_machine_status} to handle
789 the saving and restoring of the target specific information. Since the
790 single data area approach is no longer used, these pointers are no
793 @defmac INIT_EXPANDERS
794 Macro called to initialize any target specific information. This macro
795 is called once per function, before generation of any RTL has begun.
796 The intention of this macro is to allow the initialization of the
797 function pointer @code{init_machine_status}.
800 @deftypevar {void (*)(struct function *)} init_machine_status
801 If this function pointer is non-@code{NULL} it will be called once per
802 function, before function compilation starts, in order to allow the
803 target to perform any target specific initialization of the
804 @code{struct function} structure. It is intended that this would be
805 used to initialize the @code{machine} of that structure.
807 @code{struct machine_function} structures are expected to be freed by GC@.
808 Generally, any memory that they reference must be allocated by using
809 GC allocation, including the structure itself.
813 @section Storage Layout
814 @cindex storage layout
816 Note that the definitions of the macros in this table which are sizes or
817 alignments measured in bits do not need to be constant. They can be C
818 expressions that refer to static variables, such as the @code{target_flags}.
819 @xref{Run-time Target}.
821 @defmac BITS_BIG_ENDIAN
822 Define this macro to have the value 1 if the most significant bit in a
823 byte has the lowest number; otherwise define it to have the value zero.
824 This means that bit-field instructions count from the most significant
825 bit. If the machine has no bit-field instructions, then this must still
826 be defined, but it doesn't matter which value it is defined to. This
827 macro need not be a constant.
829 This macro does not affect the way structure fields are packed into
830 bytes or words; that is controlled by @code{BYTES_BIG_ENDIAN}.
833 @defmac BYTES_BIG_ENDIAN
834 Define this macro to have the value 1 if the most significant byte in a
835 word has the lowest number. This macro need not be a constant.
838 @defmac WORDS_BIG_ENDIAN
839 Define this macro to have the value 1 if, in a multiword object, the
840 most significant word has the lowest number. This applies to both
841 memory locations and registers; see @code{REG_WORDS_BIG_ENDIAN} if the
842 order of words in memory is not the same as the order in registers. This
843 macro need not be a constant.
846 @defmac REG_WORDS_BIG_ENDIAN
847 On some machines, the order of words in a multiword object differs between
848 registers in memory. In such a situation, define this macro to describe
849 the order of words in a register. The macro @code{WORDS_BIG_ENDIAN} controls
850 the order of words in memory.
853 @defmac FLOAT_WORDS_BIG_ENDIAN
854 Define this macro to have the value 1 if @code{DFmode}, @code{XFmode} or
855 @code{TFmode} floating point numbers are stored in memory with the word
856 containing the sign bit at the lowest address; otherwise define it to
857 have the value 0. This macro need not be a constant.
859 You need not define this macro if the ordering is the same as for
863 @defmac BITS_PER_UNIT
864 Define this macro to be the number of bits in an addressable storage
865 unit (byte). If you do not define this macro the default is 8.
868 @defmac BITS_PER_WORD
869 Number of bits in a word. If you do not define this macro, the default
870 is @code{BITS_PER_UNIT * UNITS_PER_WORD}.
873 @defmac MAX_BITS_PER_WORD
874 Maximum number of bits in a word. If this is undefined, the default is
875 @code{BITS_PER_WORD}. Otherwise, it is the constant value that is the
876 largest value that @code{BITS_PER_WORD} can have at run-time.
879 @defmac UNITS_PER_WORD
880 Number of storage units in a word; normally the size of a general-purpose
881 register, a power of two from 1 or 8.
884 @defmac MIN_UNITS_PER_WORD
885 Minimum number of units in a word. If this is undefined, the default is
886 @code{UNITS_PER_WORD}. Otherwise, it is the constant value that is the
887 smallest value that @code{UNITS_PER_WORD} can have at run-time.
891 Width of a pointer, in bits. You must specify a value no wider than the
892 width of @code{Pmode}. If it is not equal to the width of @code{Pmode},
893 you must define @code{POINTERS_EXTEND_UNSIGNED}. If you do not specify
894 a value the default is @code{BITS_PER_WORD}.
897 @defmac POINTERS_EXTEND_UNSIGNED
898 A C expression that determines how pointers should be extended from
899 @code{ptr_mode} to either @code{Pmode} or @code{word_mode}. It is
900 greater than zero if pointers should be zero-extended, zero if they
901 should be sign-extended, and negative if some other sort of conversion
902 is needed. In the last case, the extension is done by the target's
903 @code{ptr_extend} instruction.
905 You need not define this macro if the @code{ptr_mode}, @code{Pmode}
906 and @code{word_mode} are all the same width.
909 @defmac PROMOTE_MODE (@var{m}, @var{unsignedp}, @var{type})
910 A macro to update @var{m} and @var{unsignedp} when an object whose type
911 is @var{type} and which has the specified mode and signedness is to be
912 stored in a register. This macro is only called when @var{type} is a
915 On most RISC machines, which only have operations that operate on a full
916 register, define this macro to set @var{m} to @code{word_mode} if
917 @var{m} is an integer mode narrower than @code{BITS_PER_WORD}. In most
918 cases, only integer modes should be widened because wider-precision
919 floating-point operations are usually more expensive than their narrower
922 For most machines, the macro definition does not change @var{unsignedp}.
923 However, some machines, have instructions that preferentially handle
924 either signed or unsigned quantities of certain modes. For example, on
925 the DEC Alpha, 32-bit loads from memory and 32-bit add instructions
926 sign-extend the result to 64 bits. On such machines, set
927 @var{unsignedp} according to which kind of extension is more efficient.
929 Do not define this macro if it would never modify @var{m}.
932 @hook TARGET_PROMOTE_FUNCTION_MODE
933 Like @code{PROMOTE_MODE}, but it is applied to outgoing function arguments or
934 function return values. The target hook should return the new mode
935 and possibly change @code{*@var{punsignedp}} if the promotion should
936 change signedness. This function is called only for scalar @emph{or
939 @var{for_return} allows to distinguish the promotion of arguments and
940 return values. If it is @code{1}, a return value is being promoted and
941 @code{TARGET_FUNCTION_VALUE} must perform the same promotions done here.
942 If it is @code{2}, the returned mode should be that of the register in
943 which an incoming parameter is copied, or the outgoing result is computed;
944 then the hook should return the same mode as @code{promote_mode}, though
945 the signedness may be different.
947 @var{type} can be NULL when promoting function arguments of libcalls.
949 The default is to not promote arguments and return values. You can
950 also define the hook to @code{default_promote_function_mode_always_promote}
951 if you would like to apply the same rules given by @code{PROMOTE_MODE}.
954 @defmac PARM_BOUNDARY
955 Normal alignment required for function parameters on the stack, in
956 bits. All stack parameters receive at least this much alignment
957 regardless of data type. On most machines, this is the same as the
961 @defmac STACK_BOUNDARY
962 Define this macro to the minimum alignment enforced by hardware for the
963 stack pointer on this machine. The definition is a C expression for the
964 desired alignment (measured in bits). This value is used as a default
965 if @code{PREFERRED_STACK_BOUNDARY} is not defined. On most machines,
966 this should be the same as @code{PARM_BOUNDARY}.
969 @defmac PREFERRED_STACK_BOUNDARY
970 Define this macro if you wish to preserve a certain alignment for the
971 stack pointer, greater than what the hardware enforces. The definition
972 is a C expression for the desired alignment (measured in bits). This
973 macro must evaluate to a value equal to or larger than
974 @code{STACK_BOUNDARY}.
977 @defmac INCOMING_STACK_BOUNDARY
978 Define this macro if the incoming stack boundary may be different
979 from @code{PREFERRED_STACK_BOUNDARY}. This macro must evaluate
980 to a value equal to or larger than @code{STACK_BOUNDARY}.
983 @defmac FUNCTION_BOUNDARY
984 Alignment required for a function entry point, in bits.
987 @defmac BIGGEST_ALIGNMENT
988 Biggest alignment that any data type can require on this machine, in
989 bits. Note that this is not the biggest alignment that is supported,
990 just the biggest alignment that, when violated, may cause a fault.
993 @defmac MALLOC_ABI_ALIGNMENT
994 Alignment, in bits, a C conformant malloc implementation has to
995 provide. If not defined, the default value is @code{BITS_PER_WORD}.
998 @defmac ATTRIBUTE_ALIGNED_VALUE
999 Alignment used by the @code{__attribute__ ((aligned))} construct. If
1000 not defined, the default value is @code{BIGGEST_ALIGNMENT}.
1003 @defmac MINIMUM_ATOMIC_ALIGNMENT
1004 If defined, the smallest alignment, in bits, that can be given to an
1005 object that can be referenced in one operation, without disturbing any
1006 nearby object. Normally, this is @code{BITS_PER_UNIT}, but may be larger
1007 on machines that don't have byte or half-word store operations.
1010 @defmac BIGGEST_FIELD_ALIGNMENT
1011 Biggest alignment that any structure or union field can require on this
1012 machine, in bits. If defined, this overrides @code{BIGGEST_ALIGNMENT} for
1013 structure and union fields only, unless the field alignment has been set
1014 by the @code{__attribute__ ((aligned (@var{n})))} construct.
1017 @defmac ADJUST_FIELD_ALIGN (@var{field}, @var{computed})
1018 An expression for the alignment of a structure field @var{field} if the
1019 alignment computed in the usual way (including applying of
1020 @code{BIGGEST_ALIGNMENT} and @code{BIGGEST_FIELD_ALIGNMENT} to the
1021 alignment) is @var{computed}. It overrides alignment only if the
1022 field alignment has not been set by the
1023 @code{__attribute__ ((aligned (@var{n})))} construct.
1026 @defmac MAX_STACK_ALIGNMENT
1027 Biggest stack alignment guaranteed by the backend. Use this macro
1028 to specify the maximum alignment of a variable on stack.
1030 If not defined, the default value is @code{STACK_BOUNDARY}.
1032 @c FIXME: The default should be @code{PREFERRED_STACK_BOUNDARY}.
1033 @c But the fix for PR 32893 indicates that we can only guarantee
1034 @c maximum stack alignment on stack up to @code{STACK_BOUNDARY}, not
1035 @c @code{PREFERRED_STACK_BOUNDARY}, if stack alignment isn't supported.
1038 @defmac MAX_OFILE_ALIGNMENT
1039 Biggest alignment supported by the object file format of this machine.
1040 Use this macro to limit the alignment which can be specified using the
1041 @code{__attribute__ ((aligned (@var{n})))} construct. If not defined,
1042 the default value is @code{BIGGEST_ALIGNMENT}.
1044 On systems that use ELF, the default (in @file{config/elfos.h}) is
1045 the largest supported 32-bit ELF section alignment representable on
1046 a 32-bit host e.g. @samp{(((unsigned HOST_WIDEST_INT) 1 << 28) * 8)}.
1047 On 32-bit ELF the largest supported section alignment in bits is
1048 @samp{(0x80000000 * 8)}, but this is not representable on 32-bit hosts.
1051 @defmac DATA_ALIGNMENT (@var{type}, @var{basic-align})
1052 If defined, a C expression to compute the alignment for a variable in
1053 the static store. @var{type} is the data type, and @var{basic-align} is
1054 the alignment that the object would ordinarily have. The value of this
1055 macro is used instead of that alignment to align the object.
1057 If this macro is not defined, then @var{basic-align} is used.
1060 One use of this macro is to increase alignment of medium-size data to
1061 make it all fit in fewer cache lines. Another is to cause character
1062 arrays to be word-aligned so that @code{strcpy} calls that copy
1063 constants to character arrays can be done inline.
1066 @defmac CONSTANT_ALIGNMENT (@var{constant}, @var{basic-align})
1067 If defined, a C expression to compute the alignment given to a constant
1068 that is being placed in memory. @var{constant} is the constant and
1069 @var{basic-align} is the alignment that the object would ordinarily
1070 have. The value of this macro is used instead of that alignment to
1073 If this macro is not defined, then @var{basic-align} is used.
1075 The typical use of this macro is to increase alignment for string
1076 constants to be word aligned so that @code{strcpy} calls that copy
1077 constants can be done inline.
1080 @defmac LOCAL_ALIGNMENT (@var{type}, @var{basic-align})
1081 If defined, a C expression to compute the alignment for a variable in
1082 the local store. @var{type} is the data type, and @var{basic-align} is
1083 the alignment that the object would ordinarily have. The value of this
1084 macro is used instead of that alignment to align the object.
1086 If this macro is not defined, then @var{basic-align} is used.
1088 One use of this macro is to increase alignment of medium-size data to
1089 make it all fit in fewer cache lines.
1091 If the value of this macro has a type, it should be an unsigned type.
1094 @hook TARGET_VECTOR_ALIGNMENT
1096 @defmac STACK_SLOT_ALIGNMENT (@var{type}, @var{mode}, @var{basic-align})
1097 If defined, a C expression to compute the alignment for stack slot.
1098 @var{type} is the data type, @var{mode} is the widest mode available,
1099 and @var{basic-align} is the alignment that the slot would ordinarily
1100 have. The value of this macro is used instead of that alignment to
1103 If this macro is not defined, then @var{basic-align} is used when
1104 @var{type} is @code{NULL}. Otherwise, @code{LOCAL_ALIGNMENT} will
1107 This macro is to set alignment of stack slot to the maximum alignment
1108 of all possible modes which the slot may have.
1110 If the value of this macro has a type, it should be an unsigned type.
1113 @defmac LOCAL_DECL_ALIGNMENT (@var{decl})
1114 If defined, a C expression to compute the alignment for a local
1115 variable @var{decl}.
1117 If this macro is not defined, then
1118 @code{LOCAL_ALIGNMENT (TREE_TYPE (@var{decl}), DECL_ALIGN (@var{decl}))}
1121 One use of this macro is to increase alignment of medium-size data to
1122 make it all fit in fewer cache lines.
1124 If the value of this macro has a type, it should be an unsigned type.
1127 @defmac MINIMUM_ALIGNMENT (@var{exp}, @var{mode}, @var{align})
1128 If defined, a C expression to compute the minimum required alignment
1129 for dynamic stack realignment purposes for @var{exp} (a type or decl),
1130 @var{mode}, assuming normal alignment @var{align}.
1132 If this macro is not defined, then @var{align} will be used.
1135 @defmac EMPTY_FIELD_BOUNDARY
1136 Alignment in bits to be given to a structure bit-field that follows an
1137 empty field such as @code{int : 0;}.
1139 If @code{PCC_BITFIELD_TYPE_MATTERS} is true, it overrides this macro.
1142 @defmac STRUCTURE_SIZE_BOUNDARY
1143 Number of bits which any structure or union's size must be a multiple of.
1144 Each structure or union's size is rounded up to a multiple of this.
1146 If you do not define this macro, the default is the same as
1147 @code{BITS_PER_UNIT}.
1150 @defmac STRICT_ALIGNMENT
1151 Define this macro to be the value 1 if instructions will fail to work
1152 if given data not on the nominal alignment. If instructions will merely
1153 go slower in that case, define this macro as 0.
1156 @defmac PCC_BITFIELD_TYPE_MATTERS
1157 Define this if you wish to imitate the way many other C compilers handle
1158 alignment of bit-fields and the structures that contain them.
1160 The behavior is that the type written for a named bit-field (@code{int},
1161 @code{short}, or other integer type) imposes an alignment for the entire
1162 structure, as if the structure really did contain an ordinary field of
1163 that type. In addition, the bit-field is placed within the structure so
1164 that it would fit within such a field, not crossing a boundary for it.
1166 Thus, on most machines, a named bit-field whose type is written as
1167 @code{int} would not cross a four-byte boundary, and would force
1168 four-byte alignment for the whole structure. (The alignment used may
1169 not be four bytes; it is controlled by the other alignment parameters.)
1171 An unnamed bit-field will not affect the alignment of the containing
1174 If the macro is defined, its definition should be a C expression;
1175 a nonzero value for the expression enables this behavior.
1177 Note that if this macro is not defined, or its value is zero, some
1178 bit-fields may cross more than one alignment boundary. The compiler can
1179 support such references if there are @samp{insv}, @samp{extv}, and
1180 @samp{extzv} insns that can directly reference memory.
1182 The other known way of making bit-fields work is to define
1183 @code{STRUCTURE_SIZE_BOUNDARY} as large as @code{BIGGEST_ALIGNMENT}.
1184 Then every structure can be accessed with fullwords.
1186 Unless the machine has bit-field instructions or you define
1187 @code{STRUCTURE_SIZE_BOUNDARY} that way, you must define
1188 @code{PCC_BITFIELD_TYPE_MATTERS} to have a nonzero value.
1190 If your aim is to make GCC use the same conventions for laying out
1191 bit-fields as are used by another compiler, here is how to investigate
1192 what the other compiler does. Compile and run this program:
1211 printf ("Size of foo1 is %d\n",
1212 sizeof (struct foo1));
1213 printf ("Size of foo2 is %d\n",
1214 sizeof (struct foo2));
1219 If this prints 2 and 5, then the compiler's behavior is what you would
1220 get from @code{PCC_BITFIELD_TYPE_MATTERS}.
1223 @defmac BITFIELD_NBYTES_LIMITED
1224 Like @code{PCC_BITFIELD_TYPE_MATTERS} except that its effect is limited
1225 to aligning a bit-field within the structure.
1228 @hook TARGET_ALIGN_ANON_BITFIELD
1229 When @code{PCC_BITFIELD_TYPE_MATTERS} is true this hook will determine
1230 whether unnamed bitfields affect the alignment of the containing
1231 structure. The hook should return true if the structure should inherit
1232 the alignment requirements of an unnamed bitfield's type.
1235 @hook TARGET_NARROW_VOLATILE_BITFIELD
1236 This target hook should return @code{true} if accesses to volatile bitfields
1237 should use the narrowest mode possible. It should return @code{false} if
1238 these accesses should use the bitfield container type.
1240 The default is @code{!TARGET_STRICT_ALIGN}.
1243 @hook TARGET_MEMBER_TYPE_FORCES_BLK
1244 Return true if a structure, union or array containing @var{field} should
1245 be accessed using @code{BLKMODE}.
1247 If @var{field} is the only field in the structure, @var{mode} is its
1248 mode, otherwise @var{mode} is VOIDmode. @var{mode} is provided in the
1249 case where structures of one field would require the structure's mode to
1250 retain the field's mode.
1252 Normally, this is not needed.
1255 @defmac ROUND_TYPE_ALIGN (@var{type}, @var{computed}, @var{specified})
1256 Define this macro as an expression for the alignment of a type (given
1257 by @var{type} as a tree node) if the alignment computed in the usual
1258 way is @var{computed} and the alignment explicitly specified was
1261 The default is to use @var{specified} if it is larger; otherwise, use
1262 the smaller of @var{computed} and @code{BIGGEST_ALIGNMENT}
1265 @defmac MAX_FIXED_MODE_SIZE
1266 An integer expression for the size in bits of the largest integer
1267 machine mode that should actually be used. All integer machine modes of
1268 this size or smaller can be used for structures and unions with the
1269 appropriate sizes. If this macro is undefined, @code{GET_MODE_BITSIZE
1270 (DImode)} is assumed.
1273 @defmac STACK_SAVEAREA_MODE (@var{save_level})
1274 If defined, an expression of type @code{enum machine_mode} that
1275 specifies the mode of the save area operand of a
1276 @code{save_stack_@var{level}} named pattern (@pxref{Standard Names}).
1277 @var{save_level} is one of @code{SAVE_BLOCK}, @code{SAVE_FUNCTION}, or
1278 @code{SAVE_NONLOCAL} and selects which of the three named patterns is
1279 having its mode specified.
1281 You need not define this macro if it always returns @code{Pmode}. You
1282 would most commonly define this macro if the
1283 @code{save_stack_@var{level}} patterns need to support both a 32- and a
1287 @defmac STACK_SIZE_MODE
1288 If defined, an expression of type @code{enum machine_mode} that
1289 specifies the mode of the size increment operand of an
1290 @code{allocate_stack} named pattern (@pxref{Standard Names}).
1292 You need not define this macro if it always returns @code{word_mode}.
1293 You would most commonly define this macro if the @code{allocate_stack}
1294 pattern needs to support both a 32- and a 64-bit mode.
1297 @hook TARGET_LIBGCC_CMP_RETURN_MODE
1298 This target hook should return the mode to be used for the return value
1299 of compare instructions expanded to libgcc calls. If not defined
1300 @code{word_mode} is returned which is the right choice for a majority of
1304 @hook TARGET_LIBGCC_SHIFT_COUNT_MODE
1305 This target hook should return the mode to be used for the shift count operand
1306 of shift instructions expanded to libgcc calls. If not defined
1307 @code{word_mode} is returned which is the right choice for a majority of
1311 @hook TARGET_UNWIND_WORD_MODE
1312 Return machine mode to be used for @code{_Unwind_Word} type.
1313 The default is to use @code{word_mode}.
1316 @defmac ROUND_TOWARDS_ZERO
1317 If defined, this macro should be true if the prevailing rounding
1318 mode is towards zero.
1320 Defining this macro only affects the way @file{libgcc.a} emulates
1321 floating-point arithmetic.
1323 Not defining this macro is equivalent to returning zero.
1326 @defmac LARGEST_EXPONENT_IS_NORMAL (@var{size})
1327 This macro should return true if floats with @var{size}
1328 bits do not have a NaN or infinity representation, but use the largest
1329 exponent for normal numbers instead.
1331 Defining this macro only affects the way @file{libgcc.a} emulates
1332 floating-point arithmetic.
1334 The default definition of this macro returns false for all sizes.
1337 @hook TARGET_MS_BITFIELD_LAYOUT_P
1338 This target hook returns @code{true} if bit-fields in the given
1339 @var{record_type} are to be laid out following the rules of Microsoft
1340 Visual C/C++, namely: (i) a bit-field won't share the same storage
1341 unit with the previous bit-field if their underlying types have
1342 different sizes, and the bit-field will be aligned to the highest
1343 alignment of the underlying types of itself and of the previous
1344 bit-field; (ii) a zero-sized bit-field will affect the alignment of
1345 the whole enclosing structure, even if it is unnamed; except that
1346 (iii) a zero-sized bit-field will be disregarded unless it follows
1347 another bit-field of nonzero size. If this hook returns @code{true},
1348 other macros that control bit-field layout are ignored.
1350 When a bit-field is inserted into a packed record, the whole size
1351 of the underlying type is used by one or more same-size adjacent
1352 bit-fields (that is, if its long:3, 32 bits is used in the record,
1353 and any additional adjacent long bit-fields are packed into the same
1354 chunk of 32 bits. However, if the size changes, a new field of that
1355 size is allocated). In an unpacked record, this is the same as using
1356 alignment, but not equivalent when packing.
1358 If both MS bit-fields and @samp{__attribute__((packed))} are used,
1359 the latter will take precedence. If @samp{__attribute__((packed))} is
1360 used on a single field when MS bit-fields are in use, it will take
1361 precedence for that field, but the alignment of the rest of the structure
1362 may affect its placement.
1365 @hook TARGET_DECIMAL_FLOAT_SUPPORTED_P
1366 Returns true if the target supports decimal floating point.
1369 @hook TARGET_FIXED_POINT_SUPPORTED_P
1370 Returns true if the target supports fixed-point arithmetic.
1373 @hook TARGET_EXPAND_TO_RTL_HOOK
1374 This hook is called just before expansion into rtl, allowing the target
1375 to perform additional initializations or analysis before the expansion.
1376 For example, the rs6000 port uses it to allocate a scratch stack slot
1377 for use in copying SDmode values between memory and floating point
1378 registers whenever the function being expanded has any SDmode
1382 @hook TARGET_INSTANTIATE_DECLS
1383 This hook allows the backend to perform additional instantiations on rtl
1384 that are not actually in any insns yet, but will be later.
1387 @hook TARGET_MANGLE_TYPE
1388 If your target defines any fundamental types, or any types your target
1389 uses should be mangled differently from the default, define this hook
1390 to return the appropriate encoding for these types as part of a C++
1391 mangled name. The @var{type} argument is the tree structure representing
1392 the type to be mangled. The hook may be applied to trees which are
1393 not target-specific fundamental types; it should return @code{NULL}
1394 for all such types, as well as arguments it does not recognize. If the
1395 return value is not @code{NULL}, it must point to a statically-allocated
1398 Target-specific fundamental types might be new fundamental types or
1399 qualified versions of ordinary fundamental types. Encode new
1400 fundamental types as @samp{@w{u @var{n} @var{name}}}, where @var{name}
1401 is the name used for the type in source code, and @var{n} is the
1402 length of @var{name} in decimal. Encode qualified versions of
1403 ordinary types as @samp{@w{U @var{n} @var{name} @var{code}}}, where
1404 @var{name} is the name used for the type qualifier in source code,
1405 @var{n} is the length of @var{name} as above, and @var{code} is the
1406 code used to represent the unqualified version of this type. (See
1407 @code{write_builtin_type} in @file{cp/mangle.c} for the list of
1408 codes.) In both cases the spaces are for clarity; do not include any
1409 spaces in your string.
1411 This hook is applied to types prior to typedef resolution. If the mangled
1412 name for a particular type depends only on that type's main variant, you
1413 can perform typedef resolution yourself using @code{TYPE_MAIN_VARIANT}
1416 The default version of this hook always returns @code{NULL}, which is
1417 appropriate for a target that does not define any new fundamental
1422 @section Layout of Source Language Data Types
1424 These macros define the sizes and other characteristics of the standard
1425 basic data types used in programs being compiled. Unlike the macros in
1426 the previous section, these apply to specific features of C and related
1427 languages, rather than to fundamental aspects of storage layout.
1429 @defmac INT_TYPE_SIZE
1430 A C expression for the size in bits of the type @code{int} on the
1431 target machine. If you don't define this, the default is one word.
1434 @defmac SHORT_TYPE_SIZE
1435 A C expression for the size in bits of the type @code{short} on the
1436 target machine. If you don't define this, the default is half a word.
1437 (If this would be less than one storage unit, it is rounded up to one
1441 @defmac LONG_TYPE_SIZE
1442 A C expression for the size in bits of the type @code{long} on the
1443 target machine. If you don't define this, the default is one word.
1446 @defmac ADA_LONG_TYPE_SIZE
1447 On some machines, the size used for the Ada equivalent of the type
1448 @code{long} by a native Ada compiler differs from that used by C@. In
1449 that situation, define this macro to be a C expression to be used for
1450 the size of that type. If you don't define this, the default is the
1451 value of @code{LONG_TYPE_SIZE}.
1454 @defmac LONG_LONG_TYPE_SIZE
1455 A C expression for the size in bits of the type @code{long long} on the
1456 target machine. If you don't define this, the default is two
1457 words. If you want to support GNU Ada on your machine, the value of this
1458 macro must be at least 64.
1461 @defmac CHAR_TYPE_SIZE
1462 A C expression for the size in bits of the type @code{char} on the
1463 target machine. If you don't define this, the default is
1464 @code{BITS_PER_UNIT}.
1467 @defmac BOOL_TYPE_SIZE
1468 A C expression for the size in bits of the C++ type @code{bool} and
1469 C99 type @code{_Bool} on the target machine. If you don't define
1470 this, and you probably shouldn't, the default is @code{CHAR_TYPE_SIZE}.
1473 @defmac FLOAT_TYPE_SIZE
1474 A C expression for the size in bits of the type @code{float} on the
1475 target machine. If you don't define this, the default is one word.
1478 @defmac DOUBLE_TYPE_SIZE
1479 A C expression for the size in bits of the type @code{double} on the
1480 target machine. If you don't define this, the default is two
1484 @defmac LONG_DOUBLE_TYPE_SIZE
1485 A C expression for the size in bits of the type @code{long double} on
1486 the target machine. If you don't define this, the default is two
1490 @defmac SHORT_FRACT_TYPE_SIZE
1491 A C expression for the size in bits of the type @code{short _Fract} on
1492 the target machine. If you don't define this, the default is
1493 @code{BITS_PER_UNIT}.
1496 @defmac FRACT_TYPE_SIZE
1497 A C expression for the size in bits of the type @code{_Fract} on
1498 the target machine. If you don't define this, the default is
1499 @code{BITS_PER_UNIT * 2}.
1502 @defmac LONG_FRACT_TYPE_SIZE
1503 A C expression for the size in bits of the type @code{long _Fract} on
1504 the target machine. If you don't define this, the default is
1505 @code{BITS_PER_UNIT * 4}.
1508 @defmac LONG_LONG_FRACT_TYPE_SIZE
1509 A C expression for the size in bits of the type @code{long long _Fract} on
1510 the target machine. If you don't define this, the default is
1511 @code{BITS_PER_UNIT * 8}.
1514 @defmac SHORT_ACCUM_TYPE_SIZE
1515 A C expression for the size in bits of the type @code{short _Accum} on
1516 the target machine. If you don't define this, the default is
1517 @code{BITS_PER_UNIT * 2}.
1520 @defmac ACCUM_TYPE_SIZE
1521 A C expression for the size in bits of the type @code{_Accum} on
1522 the target machine. If you don't define this, the default is
1523 @code{BITS_PER_UNIT * 4}.
1526 @defmac LONG_ACCUM_TYPE_SIZE
1527 A C expression for the size in bits of the type @code{long _Accum} on
1528 the target machine. If you don't define this, the default is
1529 @code{BITS_PER_UNIT * 8}.
1532 @defmac LONG_LONG_ACCUM_TYPE_SIZE
1533 A C expression for the size in bits of the type @code{long long _Accum} on
1534 the target machine. If you don't define this, the default is
1535 @code{BITS_PER_UNIT * 16}.
1538 @defmac LIBGCC2_LONG_DOUBLE_TYPE_SIZE
1539 Define this macro if @code{LONG_DOUBLE_TYPE_SIZE} is not constant or
1540 if you want routines in @file{libgcc2.a} for a size other than
1541 @code{LONG_DOUBLE_TYPE_SIZE}. If you don't define this, the
1542 default is @code{LONG_DOUBLE_TYPE_SIZE}.
1545 @defmac LIBGCC2_HAS_DF_MODE
1546 Define this macro if neither @code{DOUBLE_TYPE_SIZE} nor
1547 @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is
1548 @code{DFmode} but you want @code{DFmode} routines in @file{libgcc2.a}
1549 anyway. If you don't define this and either @code{DOUBLE_TYPE_SIZE}
1550 or @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is 64 then the default is 1,
1554 @defmac LIBGCC2_HAS_XF_MODE
1555 Define this macro if @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is not
1556 @code{XFmode} but you want @code{XFmode} routines in @file{libgcc2.a}
1557 anyway. If you don't define this and @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE}
1558 is 80 then the default is 1, otherwise it is 0.
1561 @defmac LIBGCC2_HAS_TF_MODE
1562 Define this macro if @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is not
1563 @code{TFmode} but you want @code{TFmode} routines in @file{libgcc2.a}
1564 anyway. If you don't define this and @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE}
1565 is 128 then the default is 1, otherwise it is 0.
1568 @defmac LIBGCC2_GNU_PREFIX
1569 This macro corresponds to the @code{TARGET_LIBFUNC_GNU_PREFIX} target
1570 hook and should be defined if that hook is overriden to be true. It
1571 causes function names in libgcc to be changed to use a @code{__gnu_}
1572 prefix for their name rather than the default @code{__}. A port which
1573 uses this macro should also arrange to use @file{t-gnu-prefix} in
1574 the libgcc @file{config.host}.
1581 Define these macros to be the size in bits of the mantissa of
1582 @code{SFmode}, @code{DFmode}, @code{XFmode} and @code{TFmode} values,
1583 if the defaults in @file{libgcc2.h} are inappropriate. By default,
1584 @code{FLT_MANT_DIG} is used for @code{SF_SIZE}, @code{LDBL_MANT_DIG}
1585 for @code{XF_SIZE} and @code{TF_SIZE}, and @code{DBL_MANT_DIG} or
1586 @code{LDBL_MANT_DIG} for @code{DF_SIZE} according to whether
1587 @code{DOUBLE_TYPE_SIZE} or
1588 @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is 64.
1591 @defmac TARGET_FLT_EVAL_METHOD
1592 A C expression for the value for @code{FLT_EVAL_METHOD} in @file{float.h},
1593 assuming, if applicable, that the floating-point control word is in its
1594 default state. If you do not define this macro the value of
1595 @code{FLT_EVAL_METHOD} will be zero.
1598 @defmac WIDEST_HARDWARE_FP_SIZE
1599 A C expression for the size in bits of the widest floating-point format
1600 supported by the hardware. If you define this macro, you must specify a
1601 value less than or equal to the value of @code{LONG_DOUBLE_TYPE_SIZE}.
1602 If you do not define this macro, the value of @code{LONG_DOUBLE_TYPE_SIZE}
1606 @defmac DEFAULT_SIGNED_CHAR
1607 An expression whose value is 1 or 0, according to whether the type
1608 @code{char} should be signed or unsigned by default. The user can
1609 always override this default with the options @option{-fsigned-char}
1610 and @option{-funsigned-char}.
1613 @hook TARGET_DEFAULT_SHORT_ENUMS
1614 This target hook should return true if the compiler should give an
1615 @code{enum} type only as many bytes as it takes to represent the range
1616 of possible values of that type. It should return false if all
1617 @code{enum} types should be allocated like @code{int}.
1619 The default is to return false.
1623 A C expression for a string describing the name of the data type to use
1624 for size values. The typedef name @code{size_t} is defined using the
1625 contents of the string.
1627 The string can contain more than one keyword. If so, separate them with
1628 spaces, and write first any length keyword, then @code{unsigned} if
1629 appropriate, and finally @code{int}. The string must exactly match one
1630 of the data type names defined in the function
1631 @code{c_common_nodes_and_builtins} in the file @file{c-family/c-common.c}.
1632 You may not omit @code{int} or change the order---that would cause the
1633 compiler to crash on startup.
1635 If you don't define this macro, the default is @code{"long unsigned
1640 GCC defines internal types (@code{sizetype}, @code{ssizetype},
1641 @code{bitsizetype} and @code{sbitsizetype}) for expressions
1642 dealing with size. This macro is a C expression for a string describing
1643 the name of the data type from which the precision of @code{sizetype}
1646 The string has the same restrictions as @code{SIZE_TYPE} string.
1648 If you don't define this macro, the default is @code{SIZE_TYPE}.
1651 @defmac PTRDIFF_TYPE
1652 A C expression for a string describing the name of the data type to use
1653 for the result of subtracting two pointers. The typedef name
1654 @code{ptrdiff_t} is defined using the contents of the string. See
1655 @code{SIZE_TYPE} above for more information.
1657 If you don't define this macro, the default is @code{"long int"}.
1661 A C expression for a string describing the name of the data type to use
1662 for wide characters. The typedef name @code{wchar_t} is defined using
1663 the contents of the string. See @code{SIZE_TYPE} above for more
1666 If you don't define this macro, the default is @code{"int"}.
1669 @defmac WCHAR_TYPE_SIZE
1670 A C expression for the size in bits of the data type for wide
1671 characters. This is used in @code{cpp}, which cannot make use of
1676 A C expression for a string describing the name of the data type to
1677 use for wide characters passed to @code{printf} and returned from
1678 @code{getwc}. The typedef name @code{wint_t} is defined using the
1679 contents of the string. See @code{SIZE_TYPE} above for more
1682 If you don't define this macro, the default is @code{"unsigned int"}.
1686 A C expression for a string describing the name of the data type that
1687 can represent any value of any standard or extended signed integer type.
1688 The typedef name @code{intmax_t} is defined using the contents of the
1689 string. See @code{SIZE_TYPE} above for more information.
1691 If you don't define this macro, the default is the first of
1692 @code{"int"}, @code{"long int"}, or @code{"long long int"} that has as
1693 much precision as @code{long long int}.
1696 @defmac UINTMAX_TYPE
1697 A C expression for a string describing the name of the data type that
1698 can represent any value of any standard or extended unsigned integer
1699 type. The typedef name @code{uintmax_t} is defined using the contents
1700 of the string. See @code{SIZE_TYPE} above for more information.
1702 If you don't define this macro, the default is the first of
1703 @code{"unsigned int"}, @code{"long unsigned int"}, or @code{"long long
1704 unsigned int"} that has as much precision as @code{long long unsigned
1708 @defmac SIG_ATOMIC_TYPE
1714 @defmacx UINT16_TYPE
1715 @defmacx UINT32_TYPE
1716 @defmacx UINT64_TYPE
1717 @defmacx INT_LEAST8_TYPE
1718 @defmacx INT_LEAST16_TYPE
1719 @defmacx INT_LEAST32_TYPE
1720 @defmacx INT_LEAST64_TYPE
1721 @defmacx UINT_LEAST8_TYPE
1722 @defmacx UINT_LEAST16_TYPE
1723 @defmacx UINT_LEAST32_TYPE
1724 @defmacx UINT_LEAST64_TYPE
1725 @defmacx INT_FAST8_TYPE
1726 @defmacx INT_FAST16_TYPE
1727 @defmacx INT_FAST32_TYPE
1728 @defmacx INT_FAST64_TYPE
1729 @defmacx UINT_FAST8_TYPE
1730 @defmacx UINT_FAST16_TYPE
1731 @defmacx UINT_FAST32_TYPE
1732 @defmacx UINT_FAST64_TYPE
1733 @defmacx INTPTR_TYPE
1734 @defmacx UINTPTR_TYPE
1735 C expressions for the standard types @code{sig_atomic_t},
1736 @code{int8_t}, @code{int16_t}, @code{int32_t}, @code{int64_t},
1737 @code{uint8_t}, @code{uint16_t}, @code{uint32_t}, @code{uint64_t},
1738 @code{int_least8_t}, @code{int_least16_t}, @code{int_least32_t},
1739 @code{int_least64_t}, @code{uint_least8_t}, @code{uint_least16_t},
1740 @code{uint_least32_t}, @code{uint_least64_t}, @code{int_fast8_t},
1741 @code{int_fast16_t}, @code{int_fast32_t}, @code{int_fast64_t},
1742 @code{uint_fast8_t}, @code{uint_fast16_t}, @code{uint_fast32_t},
1743 @code{uint_fast64_t}, @code{intptr_t}, and @code{uintptr_t}. See
1744 @code{SIZE_TYPE} above for more information.
1746 If any of these macros evaluates to a null pointer, the corresponding
1747 type is not supported; if GCC is configured to provide
1748 @code{<stdint.h>} in such a case, the header provided may not conform
1749 to C99, depending on the type in question. The defaults for all of
1750 these macros are null pointers.
1753 @defmac TARGET_PTRMEMFUNC_VBIT_LOCATION
1754 The C++ compiler represents a pointer-to-member-function with a struct
1761 ptrdiff_t vtable_index;
1768 The C++ compiler must use one bit to indicate whether the function that
1769 will be called through a pointer-to-member-function is virtual.
1770 Normally, we assume that the low-order bit of a function pointer must
1771 always be zero. Then, by ensuring that the vtable_index is odd, we can
1772 distinguish which variant of the union is in use. But, on some
1773 platforms function pointers can be odd, and so this doesn't work. In
1774 that case, we use the low-order bit of the @code{delta} field, and shift
1775 the remainder of the @code{delta} field to the left.
1777 GCC will automatically make the right selection about where to store
1778 this bit using the @code{FUNCTION_BOUNDARY} setting for your platform.
1779 However, some platforms such as ARM/Thumb have @code{FUNCTION_BOUNDARY}
1780 set such that functions always start at even addresses, but the lowest
1781 bit of pointers to functions indicate whether the function at that
1782 address is in ARM or Thumb mode. If this is the case of your
1783 architecture, you should define this macro to
1784 @code{ptrmemfunc_vbit_in_delta}.
1786 In general, you should not have to define this macro. On architectures
1787 in which function addresses are always even, according to
1788 @code{FUNCTION_BOUNDARY}, GCC will automatically define this macro to
1789 @code{ptrmemfunc_vbit_in_pfn}.
1792 @defmac TARGET_VTABLE_USES_DESCRIPTORS
1793 Normally, the C++ compiler uses function pointers in vtables. This
1794 macro allows the target to change to use ``function descriptors''
1795 instead. Function descriptors are found on targets for whom a
1796 function pointer is actually a small data structure. Normally the
1797 data structure consists of the actual code address plus a data
1798 pointer to which the function's data is relative.
1800 If vtables are used, the value of this macro should be the number
1801 of words that the function descriptor occupies.
1804 @defmac TARGET_VTABLE_ENTRY_ALIGN
1805 By default, the vtable entries are void pointers, the so the alignment
1806 is the same as pointer alignment. The value of this macro specifies
1807 the alignment of the vtable entry in bits. It should be defined only
1808 when special alignment is necessary. */
1811 @defmac TARGET_VTABLE_DATA_ENTRY_DISTANCE
1812 There are a few non-descriptor entries in the vtable at offsets below
1813 zero. If these entries must be padded (say, to preserve the alignment
1814 specified by @code{TARGET_VTABLE_ENTRY_ALIGN}), set this to the number
1815 of words in each data entry.
1819 @section Register Usage
1820 @cindex register usage
1822 This section explains how to describe what registers the target machine
1823 has, and how (in general) they can be used.
1825 The description of which registers a specific instruction can use is
1826 done with register classes; see @ref{Register Classes}. For information
1827 on using registers to access a stack frame, see @ref{Frame Registers}.
1828 For passing values in registers, see @ref{Register Arguments}.
1829 For returning values in registers, see @ref{Scalar Return}.
1832 * Register Basics:: Number and kinds of registers.
1833 * Allocation Order:: Order in which registers are allocated.
1834 * Values in Registers:: What kinds of values each reg can hold.
1835 * Leaf Functions:: Renumbering registers for leaf functions.
1836 * Stack Registers:: Handling a register stack such as 80387.
1839 @node Register Basics
1840 @subsection Basic Characteristics of Registers
1842 @c prevent bad page break with this line
1843 Registers have various characteristics.
1845 @defmac FIRST_PSEUDO_REGISTER
1846 Number of hardware registers known to the compiler. They receive
1847 numbers 0 through @code{FIRST_PSEUDO_REGISTER-1}; thus, the first
1848 pseudo register's number really is assigned the number
1849 @code{FIRST_PSEUDO_REGISTER}.
1852 @defmac FIXED_REGISTERS
1853 @cindex fixed register
1854 An initializer that says which registers are used for fixed purposes
1855 all throughout the compiled code and are therefore not available for
1856 general allocation. These would include the stack pointer, the frame
1857 pointer (except on machines where that can be used as a general
1858 register when no frame pointer is needed), the program counter on
1859 machines where that is considered one of the addressable registers,
1860 and any other numbered register with a standard use.
1862 This information is expressed as a sequence of numbers, separated by
1863 commas and surrounded by braces. The @var{n}th number is 1 if
1864 register @var{n} is fixed, 0 otherwise.
1866 The table initialized from this macro, and the table initialized by
1867 the following one, may be overridden at run time either automatically,
1868 by the actions of the macro @code{CONDITIONAL_REGISTER_USAGE}, or by
1869 the user with the command options @option{-ffixed-@var{reg}},
1870 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}.
1873 @defmac CALL_USED_REGISTERS
1874 @cindex call-used register
1875 @cindex call-clobbered register
1876 @cindex call-saved register
1877 Like @code{FIXED_REGISTERS} but has 1 for each register that is
1878 clobbered (in general) by function calls as well as for fixed
1879 registers. This macro therefore identifies the registers that are not
1880 available for general allocation of values that must live across
1883 If a register has 0 in @code{CALL_USED_REGISTERS}, the compiler
1884 automatically saves it on function entry and restores it on function
1885 exit, if the register is used within the function.
1888 @defmac CALL_REALLY_USED_REGISTERS
1889 @cindex call-used register
1890 @cindex call-clobbered register
1891 @cindex call-saved register
1892 Like @code{CALL_USED_REGISTERS} except this macro doesn't require
1893 that the entire set of @code{FIXED_REGISTERS} be included.
1894 (@code{CALL_USED_REGISTERS} must be a superset of @code{FIXED_REGISTERS}).
1895 This macro is optional. If not specified, it defaults to the value
1896 of @code{CALL_USED_REGISTERS}.
1899 @defmac HARD_REGNO_CALL_PART_CLOBBERED (@var{regno}, @var{mode})
1900 @cindex call-used register
1901 @cindex call-clobbered register
1902 @cindex call-saved register
1903 A C expression that is nonzero if it is not permissible to store a
1904 value of mode @var{mode} in hard register number @var{regno} across a
1905 call without some part of it being clobbered. For most machines this
1906 macro need not be defined. It is only required for machines that do not
1907 preserve the entire contents of a register across a call.
1911 @findex call_used_regs
1914 @findex reg_class_contents
1915 @hook TARGET_CONDITIONAL_REGISTER_USAGE
1916 This hook may conditionally modify five variables
1917 @code{fixed_regs}, @code{call_used_regs}, @code{global_regs},
1918 @code{reg_names}, and @code{reg_class_contents}, to take into account
1919 any dependence of these register sets on target flags. The first three
1920 of these are of type @code{char []} (interpreted as Boolean vectors).
1921 @code{global_regs} is a @code{const char *[]}, and
1922 @code{reg_class_contents} is a @code{HARD_REG_SET}. Before the macro is
1923 called, @code{fixed_regs}, @code{call_used_regs},
1924 @code{reg_class_contents}, and @code{reg_names} have been initialized
1925 from @code{FIXED_REGISTERS}, @code{CALL_USED_REGISTERS},
1926 @code{REG_CLASS_CONTENTS}, and @code{REGISTER_NAMES}, respectively.
1927 @code{global_regs} has been cleared, and any @option{-ffixed-@var{reg}},
1928 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}
1929 command options have been applied.
1931 @cindex disabling certain registers
1932 @cindex controlling register usage
1933 If the usage of an entire class of registers depends on the target
1934 flags, you may indicate this to GCC by using this macro to modify
1935 @code{fixed_regs} and @code{call_used_regs} to 1 for each of the
1936 registers in the classes which should not be used by GCC@. Also define
1937 the macro @code{REG_CLASS_FROM_LETTER} / @code{REG_CLASS_FROM_CONSTRAINT}
1938 to return @code{NO_REGS} if it
1939 is called with a letter for a class that shouldn't be used.
1941 (However, if this class is not included in @code{GENERAL_REGS} and all
1942 of the insn patterns whose constraints permit this class are
1943 controlled by target switches, then GCC will automatically avoid using
1944 these registers when the target switches are opposed to them.)
1947 @defmac INCOMING_REGNO (@var{out})
1948 Define this macro if the target machine has register windows. This C
1949 expression returns the register number as seen by the called function
1950 corresponding to the register number @var{out} as seen by the calling
1951 function. Return @var{out} if register number @var{out} is not an
1955 @defmac OUTGOING_REGNO (@var{in})
1956 Define this macro if the target machine has register windows. This C
1957 expression returns the register number as seen by the calling function
1958 corresponding to the register number @var{in} as seen by the called
1959 function. Return @var{in} if register number @var{in} is not an inbound
1963 @defmac LOCAL_REGNO (@var{regno})
1964 Define this macro if the target machine has register windows. This C
1965 expression returns true if the register is call-saved but is in the
1966 register window. Unlike most call-saved registers, such registers
1967 need not be explicitly restored on function exit or during non-local
1972 If the program counter has a register number, define this as that
1973 register number. Otherwise, do not define it.
1976 @node Allocation Order
1977 @subsection Order of Allocation of Registers
1978 @cindex order of register allocation
1979 @cindex register allocation order
1981 @c prevent bad page break with this line
1982 Registers are allocated in order.
1984 @defmac REG_ALLOC_ORDER
1985 If defined, an initializer for a vector of integers, containing the
1986 numbers of hard registers in the order in which GCC should prefer
1987 to use them (from most preferred to least).
1989 If this macro is not defined, registers are used lowest numbered first
1990 (all else being equal).
1992 One use of this macro is on machines where the highest numbered
1993 registers must always be saved and the save-multiple-registers
1994 instruction supports only sequences of consecutive registers. On such
1995 machines, define @code{REG_ALLOC_ORDER} to be an initializer that lists
1996 the highest numbered allocable register first.
1999 @defmac ADJUST_REG_ALLOC_ORDER
2000 A C statement (sans semicolon) to choose the order in which to allocate
2001 hard registers for pseudo-registers local to a basic block.
2003 Store the desired register order in the array @code{reg_alloc_order}.
2004 Element 0 should be the register to allocate first; element 1, the next
2005 register; and so on.
2007 The macro body should not assume anything about the contents of
2008 @code{reg_alloc_order} before execution of the macro.
2010 On most machines, it is not necessary to define this macro.
2013 @defmac HONOR_REG_ALLOC_ORDER
2014 Normally, IRA tries to estimate the costs for saving a register in the
2015 prologue and restoring it in the epilogue. This discourages it from
2016 using call-saved registers. If a machine wants to ensure that IRA
2017 allocates registers in the order given by REG_ALLOC_ORDER even if some
2018 call-saved registers appear earlier than call-used ones, this macro
2022 @defmac IRA_HARD_REGNO_ADD_COST_MULTIPLIER (@var{regno})
2023 In some case register allocation order is not enough for the
2024 Integrated Register Allocator (@acronym{IRA}) to generate a good code.
2025 If this macro is defined, it should return a floating point value
2026 based on @var{regno}. The cost of using @var{regno} for a pseudo will
2027 be increased by approximately the pseudo's usage frequency times the
2028 value returned by this macro. Not defining this macro is equivalent
2029 to having it always return @code{0.0}.
2031 On most machines, it is not necessary to define this macro.
2034 @node Values in Registers
2035 @subsection How Values Fit in Registers
2037 This section discusses the macros that describe which kinds of values
2038 (specifically, which machine modes) each register can hold, and how many
2039 consecutive registers are needed for a given mode.
2041 @defmac HARD_REGNO_NREGS (@var{regno}, @var{mode})
2042 A C expression for the number of consecutive hard registers, starting
2043 at register number @var{regno}, required to hold a value of mode
2044 @var{mode}. This macro must never return zero, even if a register
2045 cannot hold the requested mode - indicate that with HARD_REGNO_MODE_OK
2046 and/or CANNOT_CHANGE_MODE_CLASS instead.
2048 On a machine where all registers are exactly one word, a suitable
2049 definition of this macro is
2052 #define HARD_REGNO_NREGS(REGNO, MODE) \
2053 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) \
2058 @defmac HARD_REGNO_NREGS_HAS_PADDING (@var{regno}, @var{mode})
2059 A C expression that is nonzero if a value of mode @var{mode}, stored
2060 in memory, ends with padding that causes it to take up more space than
2061 in registers starting at register number @var{regno} (as determined by
2062 multiplying GCC's notion of the size of the register when containing
2063 this mode by the number of registers returned by
2064 @code{HARD_REGNO_NREGS}). By default this is zero.
2066 For example, if a floating-point value is stored in three 32-bit
2067 registers but takes up 128 bits in memory, then this would be
2070 This macros only needs to be defined if there are cases where
2071 @code{subreg_get_info}
2072 would otherwise wrongly determine that a @code{subreg} can be
2073 represented by an offset to the register number, when in fact such a
2074 @code{subreg} would contain some of the padding not stored in
2075 registers and so not be representable.
2078 @defmac HARD_REGNO_NREGS_WITH_PADDING (@var{regno}, @var{mode})
2079 For values of @var{regno} and @var{mode} for which
2080 @code{HARD_REGNO_NREGS_HAS_PADDING} returns nonzero, a C expression
2081 returning the greater number of registers required to hold the value
2082 including any padding. In the example above, the value would be four.
2085 @defmac REGMODE_NATURAL_SIZE (@var{mode})
2086 Define this macro if the natural size of registers that hold values
2087 of mode @var{mode} is not the word size. It is a C expression that
2088 should give the natural size in bytes for the specified mode. It is
2089 used by the register allocator to try to optimize its results. This
2090 happens for example on SPARC 64-bit where the natural size of
2091 floating-point registers is still 32-bit.
2094 @defmac HARD_REGNO_MODE_OK (@var{regno}, @var{mode})
2095 A C expression that is nonzero if it is permissible to store a value
2096 of mode @var{mode} in hard register number @var{regno} (or in several
2097 registers starting with that one). For a machine where all registers
2098 are equivalent, a suitable definition is
2101 #define HARD_REGNO_MODE_OK(REGNO, MODE) 1
2104 You need not include code to check for the numbers of fixed registers,
2105 because the allocation mechanism considers them to be always occupied.
2107 @cindex register pairs
2108 On some machines, double-precision values must be kept in even/odd
2109 register pairs. You can implement that by defining this macro to reject
2110 odd register numbers for such modes.
2112 The minimum requirement for a mode to be OK in a register is that the
2113 @samp{mov@var{mode}} instruction pattern support moves between the
2114 register and other hard register in the same class and that moving a
2115 value into the register and back out not alter it.
2117 Since the same instruction used to move @code{word_mode} will work for
2118 all narrower integer modes, it is not necessary on any machine for
2119 @code{HARD_REGNO_MODE_OK} to distinguish between these modes, provided
2120 you define patterns @samp{movhi}, etc., to take advantage of this. This
2121 is useful because of the interaction between @code{HARD_REGNO_MODE_OK}
2122 and @code{MODES_TIEABLE_P}; it is very desirable for all integer modes
2125 Many machines have special registers for floating point arithmetic.
2126 Often people assume that floating point machine modes are allowed only
2127 in floating point registers. This is not true. Any registers that
2128 can hold integers can safely @emph{hold} a floating point machine
2129 mode, whether or not floating arithmetic can be done on it in those
2130 registers. Integer move instructions can be used to move the values.
2132 On some machines, though, the converse is true: fixed-point machine
2133 modes may not go in floating registers. This is true if the floating
2134 registers normalize any value stored in them, because storing a
2135 non-floating value there would garble it. In this case,
2136 @code{HARD_REGNO_MODE_OK} should reject fixed-point machine modes in
2137 floating registers. But if the floating registers do not automatically
2138 normalize, if you can store any bit pattern in one and retrieve it
2139 unchanged without a trap, then any machine mode may go in a floating
2140 register, so you can define this macro to say so.
2142 The primary significance of special floating registers is rather that
2143 they are the registers acceptable in floating point arithmetic
2144 instructions. However, this is of no concern to
2145 @code{HARD_REGNO_MODE_OK}. You handle it by writing the proper
2146 constraints for those instructions.
2148 On some machines, the floating registers are especially slow to access,
2149 so that it is better to store a value in a stack frame than in such a
2150 register if floating point arithmetic is not being done. As long as the
2151 floating registers are not in class @code{GENERAL_REGS}, they will not
2152 be used unless some pattern's constraint asks for one.
2155 @defmac HARD_REGNO_RENAME_OK (@var{from}, @var{to})
2156 A C expression that is nonzero if it is OK to rename a hard register
2157 @var{from} to another hard register @var{to}.
2159 One common use of this macro is to prevent renaming of a register to
2160 another register that is not saved by a prologue in an interrupt
2163 The default is always nonzero.
2166 @defmac MODES_TIEABLE_P (@var{mode1}, @var{mode2})
2167 A C expression that is nonzero if a value of mode
2168 @var{mode1} is accessible in mode @var{mode2} without copying.
2170 If @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode1})} and
2171 @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode2})} are always the same for
2172 any @var{r}, then @code{MODES_TIEABLE_P (@var{mode1}, @var{mode2})}
2173 should be nonzero. If they differ for any @var{r}, you should define
2174 this macro to return zero unless some other mechanism ensures the
2175 accessibility of the value in a narrower mode.
2177 You should define this macro to return nonzero in as many cases as
2178 possible since doing so will allow GCC to perform better register
2182 @hook TARGET_HARD_REGNO_SCRATCH_OK
2183 This target hook should return @code{true} if it is OK to use a hard register
2184 @var{regno} as scratch reg in peephole2.
2186 One common use of this macro is to prevent using of a register that
2187 is not saved by a prologue in an interrupt handler.
2189 The default version of this hook always returns @code{true}.
2192 @defmac AVOID_CCMODE_COPIES
2193 Define this macro if the compiler should avoid copies to/from @code{CCmode}
2194 registers. You should only define this macro if support for copying to/from
2195 @code{CCmode} is incomplete.
2198 @node Leaf Functions
2199 @subsection Handling Leaf Functions
2201 @cindex leaf functions
2202 @cindex functions, leaf
2203 On some machines, a leaf function (i.e., one which makes no calls) can run
2204 more efficiently if it does not make its own register window. Often this
2205 means it is required to receive its arguments in the registers where they
2206 are passed by the caller, instead of the registers where they would
2209 The special treatment for leaf functions generally applies only when
2210 other conditions are met; for example, often they may use only those
2211 registers for its own variables and temporaries. We use the term ``leaf
2212 function'' to mean a function that is suitable for this special
2213 handling, so that functions with no calls are not necessarily ``leaf
2216 GCC assigns register numbers before it knows whether the function is
2217 suitable for leaf function treatment. So it needs to renumber the
2218 registers in order to output a leaf function. The following macros
2221 @defmac LEAF_REGISTERS
2222 Name of a char vector, indexed by hard register number, which
2223 contains 1 for a register that is allowable in a candidate for leaf
2226 If leaf function treatment involves renumbering the registers, then the
2227 registers marked here should be the ones before renumbering---those that
2228 GCC would ordinarily allocate. The registers which will actually be
2229 used in the assembler code, after renumbering, should not be marked with 1
2232 Define this macro only if the target machine offers a way to optimize
2233 the treatment of leaf functions.
2236 @defmac LEAF_REG_REMAP (@var{regno})
2237 A C expression whose value is the register number to which @var{regno}
2238 should be renumbered, when a function is treated as a leaf function.
2240 If @var{regno} is a register number which should not appear in a leaf
2241 function before renumbering, then the expression should yield @minus{}1, which
2242 will cause the compiler to abort.
2244 Define this macro only if the target machine offers a way to optimize the
2245 treatment of leaf functions, and registers need to be renumbered to do
2249 @findex current_function_is_leaf
2250 @findex current_function_uses_only_leaf_regs
2251 @code{TARGET_ASM_FUNCTION_PROLOGUE} and
2252 @code{TARGET_ASM_FUNCTION_EPILOGUE} must usually treat leaf functions
2253 specially. They can test the C variable @code{current_function_is_leaf}
2254 which is nonzero for leaf functions. @code{current_function_is_leaf} is
2255 set prior to local register allocation and is valid for the remaining
2256 compiler passes. They can also test the C variable
2257 @code{current_function_uses_only_leaf_regs} which is nonzero for leaf
2258 functions which only use leaf registers.
2259 @code{current_function_uses_only_leaf_regs} is valid after all passes
2260 that modify the instructions have been run and is only useful if
2261 @code{LEAF_REGISTERS} is defined.
2262 @c changed this to fix overfull. ALSO: why the "it" at the beginning
2263 @c of the next paragraph?! --mew 2feb93
2265 @node Stack Registers
2266 @subsection Registers That Form a Stack
2268 There are special features to handle computers where some of the
2269 ``registers'' form a stack. Stack registers are normally written by
2270 pushing onto the stack, and are numbered relative to the top of the
2273 Currently, GCC can only handle one group of stack-like registers, and
2274 they must be consecutively numbered. Furthermore, the existing
2275 support for stack-like registers is specific to the 80387 floating
2276 point coprocessor. If you have a new architecture that uses
2277 stack-like registers, you will need to do substantial work on
2278 @file{reg-stack.c} and write your machine description to cooperate
2279 with it, as well as defining these macros.
2282 Define this if the machine has any stack-like registers.
2285 @defmac STACK_REG_COVER_CLASS
2286 This is a cover class containing the stack registers. Define this if
2287 the machine has any stack-like registers.
2290 @defmac FIRST_STACK_REG
2291 The number of the first stack-like register. This one is the top
2295 @defmac LAST_STACK_REG
2296 The number of the last stack-like register. This one is the bottom of
2300 @node Register Classes
2301 @section Register Classes
2302 @cindex register class definitions
2303 @cindex class definitions, register
2305 On many machines, the numbered registers are not all equivalent.
2306 For example, certain registers may not be allowed for indexed addressing;
2307 certain registers may not be allowed in some instructions. These machine
2308 restrictions are described to the compiler using @dfn{register classes}.
2310 You define a number of register classes, giving each one a name and saying
2311 which of the registers belong to it. Then you can specify register classes
2312 that are allowed as operands to particular instruction patterns.
2316 In general, each register will belong to several classes. In fact, one
2317 class must be named @code{ALL_REGS} and contain all the registers. Another
2318 class must be named @code{NO_REGS} and contain no registers. Often the
2319 union of two classes will be another class; however, this is not required.
2321 @findex GENERAL_REGS
2322 One of the classes must be named @code{GENERAL_REGS}. There is nothing
2323 terribly special about the name, but the operand constraint letters
2324 @samp{r} and @samp{g} specify this class. If @code{GENERAL_REGS} is
2325 the same as @code{ALL_REGS}, just define it as a macro which expands
2328 Order the classes so that if class @var{x} is contained in class @var{y}
2329 then @var{x} has a lower class number than @var{y}.
2331 The way classes other than @code{GENERAL_REGS} are specified in operand
2332 constraints is through machine-dependent operand constraint letters.
2333 You can define such letters to correspond to various classes, then use
2334 them in operand constraints.
2336 You must define the narrowest register classes for allocatable
2337 registers, so that each class either has no subclasses, or that for
2338 some mode, the move cost between registers within the class is
2339 cheaper than moving a register in the class to or from memory
2342 You should define a class for the union of two classes whenever some
2343 instruction allows both classes. For example, if an instruction allows
2344 either a floating point (coprocessor) register or a general register for a
2345 certain operand, you should define a class @code{FLOAT_OR_GENERAL_REGS}
2346 which includes both of them. Otherwise you will get suboptimal code,
2347 or even internal compiler errors when reload cannot find a register in the
2348 class computed via @code{reg_class_subunion}.
2350 You must also specify certain redundant information about the register
2351 classes: for each class, which classes contain it and which ones are
2352 contained in it; for each pair of classes, the largest class contained
2355 When a value occupying several consecutive registers is expected in a
2356 certain class, all the registers used must belong to that class.
2357 Therefore, register classes cannot be used to enforce a requirement for
2358 a register pair to start with an even-numbered register. The way to
2359 specify this requirement is with @code{HARD_REGNO_MODE_OK}.
2361 Register classes used for input-operands of bitwise-and or shift
2362 instructions have a special requirement: each such class must have, for
2363 each fixed-point machine mode, a subclass whose registers can transfer that
2364 mode to or from memory. For example, on some machines, the operations for
2365 single-byte values (@code{QImode}) are limited to certain registers. When
2366 this is so, each register class that is used in a bitwise-and or shift
2367 instruction must have a subclass consisting of registers from which
2368 single-byte values can be loaded or stored. This is so that
2369 @code{PREFERRED_RELOAD_CLASS} can always have a possible value to return.
2371 @deftp {Data type} {enum reg_class}
2372 An enumerated type that must be defined with all the register class names
2373 as enumerated values. @code{NO_REGS} must be first. @code{ALL_REGS}
2374 must be the last register class, followed by one more enumerated value,
2375 @code{LIM_REG_CLASSES}, which is not a register class but rather
2376 tells how many classes there are.
2378 Each register class has a number, which is the value of casting
2379 the class name to type @code{int}. The number serves as an index
2380 in many of the tables described below.
2383 @defmac N_REG_CLASSES
2384 The number of distinct register classes, defined as follows:
2387 #define N_REG_CLASSES (int) LIM_REG_CLASSES
2391 @defmac REG_CLASS_NAMES
2392 An initializer containing the names of the register classes as C string
2393 constants. These names are used in writing some of the debugging dumps.
2396 @defmac REG_CLASS_CONTENTS
2397 An initializer containing the contents of the register classes, as integers
2398 which are bit masks. The @var{n}th integer specifies the contents of class
2399 @var{n}. The way the integer @var{mask} is interpreted is that
2400 register @var{r} is in the class if @code{@var{mask} & (1 << @var{r})} is 1.
2402 When the machine has more than 32 registers, an integer does not suffice.
2403 Then the integers are replaced by sub-initializers, braced groupings containing
2404 several integers. Each sub-initializer must be suitable as an initializer
2405 for the type @code{HARD_REG_SET} which is defined in @file{hard-reg-set.h}.
2406 In this situation, the first integer in each sub-initializer corresponds to
2407 registers 0 through 31, the second integer to registers 32 through 63, and
2411 @defmac REGNO_REG_CLASS (@var{regno})
2412 A C expression whose value is a register class containing hard register
2413 @var{regno}. In general there is more than one such class; choose a class
2414 which is @dfn{minimal}, meaning that no smaller class also contains the
2418 @defmac BASE_REG_CLASS
2419 A macro whose definition is the name of the class to which a valid
2420 base register must belong. A base register is one used in an address
2421 which is the register value plus a displacement.
2424 @defmac MODE_BASE_REG_CLASS (@var{mode})
2425 This is a variation of the @code{BASE_REG_CLASS} macro which allows
2426 the selection of a base register in a mode dependent manner. If
2427 @var{mode} is VOIDmode then it should return the same value as
2428 @code{BASE_REG_CLASS}.
2431 @defmac MODE_BASE_REG_REG_CLASS (@var{mode})
2432 A C expression whose value is the register class to which a valid
2433 base register must belong in order to be used in a base plus index
2434 register address. You should define this macro if base plus index
2435 addresses have different requirements than other base register uses.
2438 @defmac MODE_CODE_BASE_REG_CLASS (@var{mode}, @var{address_space}, @var{outer_code}, @var{index_code})
2439 A C expression whose value is the register class to which a valid
2440 base register for a memory reference in mode @var{mode} to address
2441 space @var{address_space} must belong. @var{outer_code} and @var{index_code}
2442 define the context in which the base register occurs. @var{outer_code} is
2443 the code of the immediately enclosing expression (@code{MEM} for the top level
2444 of an address, @code{ADDRESS} for something that occurs in an
2445 @code{address_operand}). @var{index_code} is the code of the corresponding
2446 index expression if @var{outer_code} is @code{PLUS}; @code{SCRATCH} otherwise.
2449 @defmac INDEX_REG_CLASS
2450 A macro whose definition is the name of the class to which a valid
2451 index register must belong. An index register is one used in an
2452 address where its value is either multiplied by a scale factor or
2453 added to another register (as well as added to a displacement).
2456 @defmac REGNO_OK_FOR_BASE_P (@var{num})
2457 A C expression which is nonzero if register number @var{num} is
2458 suitable for use as a base register in operand addresses.
2461 @defmac REGNO_MODE_OK_FOR_BASE_P (@var{num}, @var{mode})
2462 A C expression that is just like @code{REGNO_OK_FOR_BASE_P}, except that
2463 that expression may examine the mode of the memory reference in
2464 @var{mode}. You should define this macro if the mode of the memory
2465 reference affects whether a register may be used as a base register. If
2466 you define this macro, the compiler will use it instead of
2467 @code{REGNO_OK_FOR_BASE_P}. The mode may be @code{VOIDmode} for
2468 addresses that appear outside a @code{MEM}, i.e., as an
2469 @code{address_operand}.
2472 @defmac REGNO_MODE_OK_FOR_REG_BASE_P (@var{num}, @var{mode})
2473 A C expression which is nonzero if register number @var{num} is suitable for
2474 use as a base register in base plus index operand addresses, accessing
2475 memory in mode @var{mode}. It may be either a suitable hard register or a
2476 pseudo register that has been allocated such a hard register. You should
2477 define this macro if base plus index addresses have different requirements
2478 than other base register uses.
2480 Use of this macro is deprecated; please use the more general
2481 @code{REGNO_MODE_CODE_OK_FOR_BASE_P}.
2484 @defmac REGNO_MODE_CODE_OK_FOR_BASE_P (@var{num}, @var{mode}, @var{address_space}, @var{outer_code}, @var{index_code})
2485 A C expression which is nonzero if register number @var{num} is
2486 suitable for use as a base register in operand addresses, accessing
2487 memory in mode @var{mode} in address space @var{address_space}.
2488 This is similar to @code{REGNO_MODE_OK_FOR_BASE_P}, except
2489 that that expression may examine the context in which the register
2490 appears in the memory reference. @var{outer_code} is the code of the
2491 immediately enclosing expression (@code{MEM} if at the top level of the
2492 address, @code{ADDRESS} for something that occurs in an
2493 @code{address_operand}). @var{index_code} is the code of the
2494 corresponding index expression if @var{outer_code} is @code{PLUS};
2495 @code{SCRATCH} otherwise. The mode may be @code{VOIDmode} for addresses
2496 that appear outside a @code{MEM}, i.e., as an @code{address_operand}.
2499 @defmac REGNO_OK_FOR_INDEX_P (@var{num})
2500 A C expression which is nonzero if register number @var{num} is
2501 suitable for use as an index register in operand addresses. It may be
2502 either a suitable hard register or a pseudo register that has been
2503 allocated such a hard register.
2505 The difference between an index register and a base register is that
2506 the index register may be scaled. If an address involves the sum of
2507 two registers, neither one of them scaled, then either one may be
2508 labeled the ``base'' and the other the ``index''; but whichever
2509 labeling is used must fit the machine's constraints of which registers
2510 may serve in each capacity. The compiler will try both labelings,
2511 looking for one that is valid, and will reload one or both registers
2512 only if neither labeling works.
2515 @hook TARGET_PREFERRED_RENAME_CLASS
2517 @hook TARGET_PREFERRED_RELOAD_CLASS
2518 A target hook that places additional restrictions on the register class
2519 to use when it is necessary to copy value @var{x} into a register in class
2520 @var{rclass}. The value is a register class; perhaps @var{rclass}, or perhaps
2521 another, smaller class.
2523 The default version of this hook always returns value of @code{rclass} argument.
2525 Sometimes returning a more restrictive class makes better code. For
2526 example, on the 68000, when @var{x} is an integer constant that is in range
2527 for a @samp{moveq} instruction, the value of this macro is always
2528 @code{DATA_REGS} as long as @var{rclass} includes the data registers.
2529 Requiring a data register guarantees that a @samp{moveq} will be used.
2531 One case where @code{TARGET_PREFERRED_RELOAD_CLASS} must not return
2532 @var{rclass} is if @var{x} is a legitimate constant which cannot be
2533 loaded into some register class. By returning @code{NO_REGS} you can
2534 force @var{x} into a memory location. For example, rs6000 can load
2535 immediate values into general-purpose registers, but does not have an
2536 instruction for loading an immediate value into a floating-point
2537 register, so @code{TARGET_PREFERRED_RELOAD_CLASS} returns @code{NO_REGS} when
2538 @var{x} is a floating-point constant. If the constant can't be loaded
2539 into any kind of register, code generation will be better if
2540 @code{TARGET_LEGITIMATE_CONSTANT_P} makes the constant illegitimate instead
2541 of using @code{TARGET_PREFERRED_RELOAD_CLASS}.
2543 If an insn has pseudos in it after register allocation, reload will go
2544 through the alternatives and call repeatedly @code{TARGET_PREFERRED_RELOAD_CLASS}
2545 to find the best one. Returning @code{NO_REGS}, in this case, makes
2546 reload add a @code{!} in front of the constraint: the x86 back-end uses
2547 this feature to discourage usage of 387 registers when math is done in
2548 the SSE registers (and vice versa).
2551 @defmac PREFERRED_RELOAD_CLASS (@var{x}, @var{class})
2552 A C expression that places additional restrictions on the register class
2553 to use when it is necessary to copy value @var{x} into a register in class
2554 @var{class}. The value is a register class; perhaps @var{class}, or perhaps
2555 another, smaller class. On many machines, the following definition is
2559 #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS
2562 Sometimes returning a more restrictive class makes better code. For
2563 example, on the 68000, when @var{x} is an integer constant that is in range
2564 for a @samp{moveq} instruction, the value of this macro is always
2565 @code{DATA_REGS} as long as @var{class} includes the data registers.
2566 Requiring a data register guarantees that a @samp{moveq} will be used.
2568 One case where @code{PREFERRED_RELOAD_CLASS} must not return
2569 @var{class} is if @var{x} is a legitimate constant which cannot be
2570 loaded into some register class. By returning @code{NO_REGS} you can
2571 force @var{x} into a memory location. For example, rs6000 can load
2572 immediate values into general-purpose registers, but does not have an
2573 instruction for loading an immediate value into a floating-point
2574 register, so @code{PREFERRED_RELOAD_CLASS} returns @code{NO_REGS} when
2575 @var{x} is a floating-point constant. If the constant can't be loaded
2576 into any kind of register, code generation will be better if
2577 @code{TARGET_LEGITIMATE_CONSTANT_P} makes the constant illegitimate instead
2578 of using @code{TARGET_PREFERRED_RELOAD_CLASS}.
2580 If an insn has pseudos in it after register allocation, reload will go
2581 through the alternatives and call repeatedly @code{PREFERRED_RELOAD_CLASS}
2582 to find the best one. Returning @code{NO_REGS}, in this case, makes
2583 reload add a @code{!} in front of the constraint: the x86 back-end uses
2584 this feature to discourage usage of 387 registers when math is done in
2585 the SSE registers (and vice versa).
2588 @hook TARGET_PREFERRED_OUTPUT_RELOAD_CLASS
2589 Like @code{TARGET_PREFERRED_RELOAD_CLASS}, but for output reloads instead of
2592 The default version of this hook always returns value of @code{rclass}
2595 You can also use @code{TARGET_PREFERRED_OUTPUT_RELOAD_CLASS} to discourage
2596 reload from using some alternatives, like @code{TARGET_PREFERRED_RELOAD_CLASS}.
2599 @defmac LIMIT_RELOAD_CLASS (@var{mode}, @var{class})
2600 A C expression that places additional restrictions on the register class
2601 to use when it is necessary to be able to hold a value of mode
2602 @var{mode} in a reload register for which class @var{class} would
2605 Unlike @code{PREFERRED_RELOAD_CLASS}, this macro should be used when
2606 there are certain modes that simply can't go in certain reload classes.
2608 The value is a register class; perhaps @var{class}, or perhaps another,
2611 Don't define this macro unless the target machine has limitations which
2612 require the macro to do something nontrivial.
2615 @hook TARGET_SECONDARY_RELOAD
2616 Many machines have some registers that cannot be copied directly to or
2617 from memory or even from other types of registers. An example is the
2618 @samp{MQ} register, which on most machines, can only be copied to or
2619 from general registers, but not memory. Below, we shall be using the
2620 term 'intermediate register' when a move operation cannot be performed
2621 directly, but has to be done by copying the source into the intermediate
2622 register first, and then copying the intermediate register to the
2623 destination. An intermediate register always has the same mode as
2624 source and destination. Since it holds the actual value being copied,
2625 reload might apply optimizations to re-use an intermediate register
2626 and eliding the copy from the source when it can determine that the
2627 intermediate register still holds the required value.
2629 Another kind of secondary reload is required on some machines which
2630 allow copying all registers to and from memory, but require a scratch
2631 register for stores to some memory locations (e.g., those with symbolic
2632 address on the RT, and those with certain symbolic address on the SPARC
2633 when compiling PIC)@. Scratch registers need not have the same mode
2634 as the value being copied, and usually hold a different value than
2635 that being copied. Special patterns in the md file are needed to
2636 describe how the copy is performed with the help of the scratch register;
2637 these patterns also describe the number, register class(es) and mode(s)
2638 of the scratch register(s).
2640 In some cases, both an intermediate and a scratch register are required.
2642 For input reloads, this target hook is called with nonzero @var{in_p},
2643 and @var{x} is an rtx that needs to be copied to a register of class
2644 @var{reload_class} in @var{reload_mode}. For output reloads, this target
2645 hook is called with zero @var{in_p}, and a register of class @var{reload_class}
2646 needs to be copied to rtx @var{x} in @var{reload_mode}.
2648 If copying a register of @var{reload_class} from/to @var{x} requires
2649 an intermediate register, the hook @code{secondary_reload} should
2650 return the register class required for this intermediate register.
2651 If no intermediate register is required, it should return NO_REGS.
2652 If more than one intermediate register is required, describe the one
2653 that is closest in the copy chain to the reload register.
2655 If scratch registers are needed, you also have to describe how to
2656 perform the copy from/to the reload register to/from this
2657 closest intermediate register. Or if no intermediate register is
2658 required, but still a scratch register is needed, describe the
2659 copy from/to the reload register to/from the reload operand @var{x}.
2661 You do this by setting @code{sri->icode} to the instruction code of a pattern
2662 in the md file which performs the move. Operands 0 and 1 are the output
2663 and input of this copy, respectively. Operands from operand 2 onward are
2664 for scratch operands. These scratch operands must have a mode, and a
2665 single-register-class
2666 @c [later: or memory]
2669 When an intermediate register is used, the @code{secondary_reload}
2670 hook will be called again to determine how to copy the intermediate
2671 register to/from the reload operand @var{x}, so your hook must also
2672 have code to handle the register class of the intermediate operand.
2674 @c [For later: maybe we'll allow multi-alternative reload patterns -
2675 @c the port maintainer could name a mov<mode> pattern that has clobbers -
2676 @c and match the constraints of input and output to determine the required
2677 @c alternative. A restriction would be that constraints used to match
2678 @c against reloads registers would have to be written as register class
2679 @c constraints, or we need a new target macro / hook that tells us if an
2680 @c arbitrary constraint can match an unknown register of a given class.
2681 @c Such a macro / hook would also be useful in other places.]
2684 @var{x} might be a pseudo-register or a @code{subreg} of a
2685 pseudo-register, which could either be in a hard register or in memory.
2686 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2687 in memory and the hard register number if it is in a register.
2689 Scratch operands in memory (constraint @code{"=m"} / @code{"=&m"}) are
2690 currently not supported. For the time being, you will have to continue
2691 to use @code{SECONDARY_MEMORY_NEEDED} for that purpose.
2693 @code{copy_cost} also uses this target hook to find out how values are
2694 copied. If you want it to include some extra cost for the need to allocate
2695 (a) scratch register(s), set @code{sri->extra_cost} to the additional cost.
2696 Or if two dependent moves are supposed to have a lower cost than the sum
2697 of the individual moves due to expected fortuitous scheduling and/or special
2698 forwarding logic, you can set @code{sri->extra_cost} to a negative amount.
2701 @defmac SECONDARY_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2702 @defmacx SECONDARY_INPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2703 @defmacx SECONDARY_OUTPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2704 These macros are obsolete, new ports should use the target hook
2705 @code{TARGET_SECONDARY_RELOAD} instead.
2707 These are obsolete macros, replaced by the @code{TARGET_SECONDARY_RELOAD}
2708 target hook. Older ports still define these macros to indicate to the
2709 reload phase that it may
2710 need to allocate at least one register for a reload in addition to the
2711 register to contain the data. Specifically, if copying @var{x} to a
2712 register @var{class} in @var{mode} requires an intermediate register,
2713 you were supposed to define @code{SECONDARY_INPUT_RELOAD_CLASS} to return the
2714 largest register class all of whose registers can be used as
2715 intermediate registers or scratch registers.
2717 If copying a register @var{class} in @var{mode} to @var{x} requires an
2718 intermediate or scratch register, @code{SECONDARY_OUTPUT_RELOAD_CLASS}
2719 was supposed to be defined be defined to return the largest register
2720 class required. If the
2721 requirements for input and output reloads were the same, the macro
2722 @code{SECONDARY_RELOAD_CLASS} should have been used instead of defining both
2725 The values returned by these macros are often @code{GENERAL_REGS}.
2726 Return @code{NO_REGS} if no spare register is needed; i.e., if @var{x}
2727 can be directly copied to or from a register of @var{class} in
2728 @var{mode} without requiring a scratch register. Do not define this
2729 macro if it would always return @code{NO_REGS}.
2731 If a scratch register is required (either with or without an
2732 intermediate register), you were supposed to define patterns for
2733 @samp{reload_in@var{m}} or @samp{reload_out@var{m}}, as required
2734 (@pxref{Standard Names}. These patterns, which were normally
2735 implemented with a @code{define_expand}, should be similar to the
2736 @samp{mov@var{m}} patterns, except that operand 2 is the scratch
2739 These patterns need constraints for the reload register and scratch
2741 contain a single register class. If the original reload register (whose
2742 class is @var{class}) can meet the constraint given in the pattern, the
2743 value returned by these macros is used for the class of the scratch
2744 register. Otherwise, two additional reload registers are required.
2745 Their classes are obtained from the constraints in the insn pattern.
2747 @var{x} might be a pseudo-register or a @code{subreg} of a
2748 pseudo-register, which could either be in a hard register or in memory.
2749 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2750 in memory and the hard register number if it is in a register.
2752 These macros should not be used in the case where a particular class of
2753 registers can only be copied to memory and not to another class of
2754 registers. In that case, secondary reload registers are not needed and
2755 would not be helpful. Instead, a stack location must be used to perform
2756 the copy and the @code{mov@var{m}} pattern should use memory as an
2757 intermediate storage. This case often occurs between floating-point and
2761 @defmac SECONDARY_MEMORY_NEEDED (@var{class1}, @var{class2}, @var{m})
2762 Certain machines have the property that some registers cannot be copied
2763 to some other registers without using memory. Define this macro on
2764 those machines to be a C expression that is nonzero if objects of mode
2765 @var{m} in registers of @var{class1} can only be copied to registers of
2766 class @var{class2} by storing a register of @var{class1} into memory
2767 and loading that memory location into a register of @var{class2}.
2769 Do not define this macro if its value would always be zero.
2772 @defmac SECONDARY_MEMORY_NEEDED_RTX (@var{mode})
2773 Normally when @code{SECONDARY_MEMORY_NEEDED} is defined, the compiler
2774 allocates a stack slot for a memory location needed for register copies.
2775 If this macro is defined, the compiler instead uses the memory location
2776 defined by this macro.
2778 Do not define this macro if you do not define
2779 @code{SECONDARY_MEMORY_NEEDED}.
2782 @defmac SECONDARY_MEMORY_NEEDED_MODE (@var{mode})
2783 When the compiler needs a secondary memory location to copy between two
2784 registers of mode @var{mode}, it normally allocates sufficient memory to
2785 hold a quantity of @code{BITS_PER_WORD} bits and performs the store and
2786 load operations in a mode that many bits wide and whose class is the
2787 same as that of @var{mode}.
2789 This is right thing to do on most machines because it ensures that all
2790 bits of the register are copied and prevents accesses to the registers
2791 in a narrower mode, which some machines prohibit for floating-point
2794 However, this default behavior is not correct on some machines, such as
2795 the DEC Alpha, that store short integers in floating-point registers
2796 differently than in integer registers. On those machines, the default
2797 widening will not work correctly and you must define this macro to
2798 suppress that widening in some cases. See the file @file{alpha.h} for
2801 Do not define this macro if you do not define
2802 @code{SECONDARY_MEMORY_NEEDED} or if widening @var{mode} to a mode that
2803 is @code{BITS_PER_WORD} bits wide is correct for your machine.
2806 @hook TARGET_CLASS_LIKELY_SPILLED_P
2807 A target hook which returns @code{true} if pseudos that have been assigned
2808 to registers of class @var{rclass} would likely be spilled because
2809 registers of @var{rclass} are needed for spill registers.
2811 The default version of this target hook returns @code{true} if @var{rclass}
2812 has exactly one register and @code{false} otherwise. On most machines, this
2813 default should be used. Only use this target hook to some other expression
2814 if pseudos allocated by @file{local-alloc.c} end up in memory because their
2815 hard registers were needed for spill registers. If this target hook returns
2816 @code{false} for those classes, those pseudos will only be allocated by
2817 @file{global.c}, which knows how to reallocate the pseudo to another
2818 register. If there would not be another register available for reallocation,
2819 you should not change the implementation of this target hook since
2820 the only effect of such implementation would be to slow down register
2824 @hook TARGET_CLASS_MAX_NREGS
2825 A target hook returns the maximum number of consecutive registers
2826 of class @var{rclass} needed to hold a value of mode @var{mode}.
2828 This is closely related to the macro @code{HARD_REGNO_NREGS}. In fact,
2829 the value returned by @code{TARGET_CLASS_MAX_NREGS (@var{rclass},
2830 @var{mode})} target hook should be the maximum value of
2831 @code{HARD_REGNO_NREGS (@var{regno}, @var{mode})} for all @var{regno}
2832 values in the class @var{rclass}.
2834 This target hook helps control the handling of multiple-word values
2837 The default version of this target hook returns the size of @var{mode}
2841 @defmac CLASS_MAX_NREGS (@var{class}, @var{mode})
2842 A C expression for the maximum number of consecutive registers
2843 of class @var{class} needed to hold a value of mode @var{mode}.
2845 This is closely related to the macro @code{HARD_REGNO_NREGS}. In fact,
2846 the value of the macro @code{CLASS_MAX_NREGS (@var{class}, @var{mode})}
2847 should be the maximum value of @code{HARD_REGNO_NREGS (@var{regno},
2848 @var{mode})} for all @var{regno} values in the class @var{class}.
2850 This macro helps control the handling of multiple-word values
2854 @defmac CANNOT_CHANGE_MODE_CLASS (@var{from}, @var{to}, @var{class})
2855 If defined, a C expression that returns nonzero for a @var{class} for which
2856 a change from mode @var{from} to mode @var{to} is invalid.
2858 For the example, loading 32-bit integer or floating-point objects into
2859 floating-point registers on the Alpha extends them to 64 bits.
2860 Therefore loading a 64-bit object and then storing it as a 32-bit object
2861 does not store the low-order 32 bits, as would be the case for a normal
2862 register. Therefore, @file{alpha.h} defines @code{CANNOT_CHANGE_MODE_CLASS}
2866 #define CANNOT_CHANGE_MODE_CLASS(FROM, TO, CLASS) \
2867 (GET_MODE_SIZE (FROM) != GET_MODE_SIZE (TO) \
2868 ? reg_classes_intersect_p (FLOAT_REGS, (CLASS)) : 0)
2872 @node Old Constraints
2873 @section Obsolete Macros for Defining Constraints
2874 @cindex defining constraints, obsolete method
2875 @cindex constraints, defining, obsolete method
2877 Machine-specific constraints can be defined with these macros instead
2878 of the machine description constructs described in @ref{Define
2879 Constraints}. This mechanism is obsolete. New ports should not use
2880 it; old ports should convert to the new mechanism.
2882 @defmac CONSTRAINT_LEN (@var{char}, @var{str})
2883 For the constraint at the start of @var{str}, which starts with the letter
2884 @var{c}, return the length. This allows you to have register class /
2885 constant / extra constraints that are longer than a single letter;
2886 you don't need to define this macro if you can do with single-letter
2887 constraints only. The definition of this macro should use
2888 DEFAULT_CONSTRAINT_LEN for all the characters that you don't want
2889 to handle specially.
2890 There are some sanity checks in genoutput.c that check the constraint lengths
2891 for the md file, so you can also use this macro to help you while you are
2892 transitioning from a byzantine single-letter-constraint scheme: when you
2893 return a negative length for a constraint you want to re-use, genoutput
2894 will complain about every instance where it is used in the md file.
2897 @defmac REG_CLASS_FROM_LETTER (@var{char})
2898 A C expression which defines the machine-dependent operand constraint
2899 letters for register classes. If @var{char} is such a letter, the
2900 value should be the register class corresponding to it. Otherwise,
2901 the value should be @code{NO_REGS}. The register letter @samp{r},
2902 corresponding to class @code{GENERAL_REGS}, will not be passed
2903 to this macro; you do not need to handle it.
2906 @defmac REG_CLASS_FROM_CONSTRAINT (@var{char}, @var{str})
2907 Like @code{REG_CLASS_FROM_LETTER}, but you also get the constraint string
2908 passed in @var{str}, so that you can use suffixes to distinguish between
2912 @defmac CONST_OK_FOR_LETTER_P (@var{value}, @var{c})
2913 A C expression that defines the machine-dependent operand constraint
2914 letters (@samp{I}, @samp{J}, @samp{K}, @dots{} @samp{P}) that specify
2915 particular ranges of integer values. If @var{c} is one of those
2916 letters, the expression should check that @var{value}, an integer, is in
2917 the appropriate range and return 1 if so, 0 otherwise. If @var{c} is
2918 not one of those letters, the value should be 0 regardless of
2922 @defmac CONST_OK_FOR_CONSTRAINT_P (@var{value}, @var{c}, @var{str})
2923 Like @code{CONST_OK_FOR_LETTER_P}, but you also get the constraint
2924 string passed in @var{str}, so that you can use suffixes to distinguish
2925 between different variants.
2928 @defmac CONST_DOUBLE_OK_FOR_LETTER_P (@var{value}, @var{c})
2929 A C expression that defines the machine-dependent operand constraint
2930 letters that specify particular ranges of @code{const_double} values
2931 (@samp{G} or @samp{H}).
2933 If @var{c} is one of those letters, the expression should check that
2934 @var{value}, an RTX of code @code{const_double}, is in the appropriate
2935 range and return 1 if so, 0 otherwise. If @var{c} is not one of those
2936 letters, the value should be 0 regardless of @var{value}.
2938 @code{const_double} is used for all floating-point constants and for
2939 @code{DImode} fixed-point constants. A given letter can accept either
2940 or both kinds of values. It can use @code{GET_MODE} to distinguish
2941 between these kinds.
2944 @defmac CONST_DOUBLE_OK_FOR_CONSTRAINT_P (@var{value}, @var{c}, @var{str})
2945 Like @code{CONST_DOUBLE_OK_FOR_LETTER_P}, but you also get the constraint
2946 string passed in @var{str}, so that you can use suffixes to distinguish
2947 between different variants.
2950 @defmac EXTRA_CONSTRAINT (@var{value}, @var{c})
2951 A C expression that defines the optional machine-dependent constraint
2952 letters that can be used to segregate specific types of operands, usually
2953 memory references, for the target machine. Any letter that is not
2954 elsewhere defined and not matched by @code{REG_CLASS_FROM_LETTER} /
2955 @code{REG_CLASS_FROM_CONSTRAINT}
2956 may be used. Normally this macro will not be defined.
2958 If it is required for a particular target machine, it should return 1
2959 if @var{value} corresponds to the operand type represented by the
2960 constraint letter @var{c}. If @var{c} is not defined as an extra
2961 constraint, the value returned should be 0 regardless of @var{value}.
2963 For example, on the ROMP, load instructions cannot have their output
2964 in r0 if the memory reference contains a symbolic address. Constraint
2965 letter @samp{Q} is defined as representing a memory address that does
2966 @emph{not} contain a symbolic address. An alternative is specified with
2967 a @samp{Q} constraint on the input and @samp{r} on the output. The next
2968 alternative specifies @samp{m} on the input and a register class that
2969 does not include r0 on the output.
2972 @defmac EXTRA_CONSTRAINT_STR (@var{value}, @var{c}, @var{str})
2973 Like @code{EXTRA_CONSTRAINT}, but you also get the constraint string passed
2974 in @var{str}, so that you can use suffixes to distinguish between different
2978 @defmac EXTRA_MEMORY_CONSTRAINT (@var{c}, @var{str})
2979 A C expression that defines the optional machine-dependent constraint
2980 letters, amongst those accepted by @code{EXTRA_CONSTRAINT}, that should
2981 be treated like memory constraints by the reload pass.
2983 It should return 1 if the operand type represented by the constraint
2984 at the start of @var{str}, the first letter of which is the letter @var{c},
2985 comprises a subset of all memory references including
2986 all those whose address is simply a base register. This allows the reload
2987 pass to reload an operand, if it does not directly correspond to the operand
2988 type of @var{c}, by copying its address into a base register.
2990 For example, on the S/390, some instructions do not accept arbitrary
2991 memory references, but only those that do not make use of an index
2992 register. The constraint letter @samp{Q} is defined via
2993 @code{EXTRA_CONSTRAINT} as representing a memory address of this type.
2994 If the letter @samp{Q} is marked as @code{EXTRA_MEMORY_CONSTRAINT},
2995 a @samp{Q} constraint can handle any memory operand, because the
2996 reload pass knows it can be reloaded by copying the memory address
2997 into a base register if required. This is analogous to the way
2998 an @samp{o} constraint can handle any memory operand.
3001 @defmac EXTRA_ADDRESS_CONSTRAINT (@var{c}, @var{str})
3002 A C expression that defines the optional machine-dependent constraint
3003 letters, amongst those accepted by @code{EXTRA_CONSTRAINT} /
3004 @code{EXTRA_CONSTRAINT_STR}, that should
3005 be treated like address constraints by the reload pass.
3007 It should return 1 if the operand type represented by the constraint
3008 at the start of @var{str}, which starts with the letter @var{c}, comprises
3009 a subset of all memory addresses including
3010 all those that consist of just a base register. This allows the reload
3011 pass to reload an operand, if it does not directly correspond to the operand
3012 type of @var{str}, by copying it into a base register.
3014 Any constraint marked as @code{EXTRA_ADDRESS_CONSTRAINT} can only
3015 be used with the @code{address_operand} predicate. It is treated
3016 analogously to the @samp{p} constraint.
3019 @node Stack and Calling
3020 @section Stack Layout and Calling Conventions
3021 @cindex calling conventions
3023 @c prevent bad page break with this line
3024 This describes the stack layout and calling conventions.
3028 * Exception Handling::
3033 * Register Arguments::
3035 * Aggregate Return::
3040 * Stack Smashing Protection::
3044 @subsection Basic Stack Layout
3045 @cindex stack frame layout
3046 @cindex frame layout
3048 @c prevent bad page break with this line
3049 Here is the basic stack layout.
3051 @defmac STACK_GROWS_DOWNWARD
3052 Define this macro if pushing a word onto the stack moves the stack
3053 pointer to a smaller address.
3055 When we say, ``define this macro if @dots{}'', it means that the
3056 compiler checks this macro only with @code{#ifdef} so the precise
3057 definition used does not matter.
3060 @defmac STACK_PUSH_CODE
3061 This macro defines the operation used when something is pushed
3062 on the stack. In RTL, a push operation will be
3063 @code{(set (mem (STACK_PUSH_CODE (reg sp))) @dots{})}
3065 The choices are @code{PRE_DEC}, @code{POST_DEC}, @code{PRE_INC},
3066 and @code{POST_INC}. Which of these is correct depends on
3067 the stack direction and on whether the stack pointer points
3068 to the last item on the stack or whether it points to the
3069 space for the next item on the stack.
3071 The default is @code{PRE_DEC} when @code{STACK_GROWS_DOWNWARD} is
3072 defined, which is almost always right, and @code{PRE_INC} otherwise,
3073 which is often wrong.
3076 @defmac FRAME_GROWS_DOWNWARD
3077 Define this macro to nonzero value if the addresses of local variable slots
3078 are at negative offsets from the frame pointer.
3081 @defmac ARGS_GROW_DOWNWARD
3082 Define this macro if successive arguments to a function occupy decreasing
3083 addresses on the stack.
3086 @defmac STARTING_FRAME_OFFSET
3087 Offset from the frame pointer to the first local variable slot to be allocated.
3089 If @code{FRAME_GROWS_DOWNWARD}, find the next slot's offset by
3090 subtracting the first slot's length from @code{STARTING_FRAME_OFFSET}.
3091 Otherwise, it is found by adding the length of the first slot to the
3092 value @code{STARTING_FRAME_OFFSET}.
3093 @c i'm not sure if the above is still correct.. had to change it to get
3094 @c rid of an overfull. --mew 2feb93
3097 @defmac STACK_ALIGNMENT_NEEDED
3098 Define to zero to disable final alignment of the stack during reload.
3099 The nonzero default for this macro is suitable for most ports.
3101 On ports where @code{STARTING_FRAME_OFFSET} is nonzero or where there
3102 is a register save block following the local block that doesn't require
3103 alignment to @code{STACK_BOUNDARY}, it may be beneficial to disable
3104 stack alignment and do it in the backend.
3107 @defmac STACK_POINTER_OFFSET
3108 Offset from the stack pointer register to the first location at which
3109 outgoing arguments are placed. If not specified, the default value of
3110 zero is used. This is the proper value for most machines.
3112 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
3113 the first location at which outgoing arguments are placed.
3116 @defmac FIRST_PARM_OFFSET (@var{fundecl})
3117 Offset from the argument pointer register to the first argument's
3118 address. On some machines it may depend on the data type of the
3121 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
3122 the first argument's address.
3125 @defmac STACK_DYNAMIC_OFFSET (@var{fundecl})
3126 Offset from the stack pointer register to an item dynamically allocated
3127 on the stack, e.g., by @code{alloca}.
3129 The default value for this macro is @code{STACK_POINTER_OFFSET} plus the
3130 length of the outgoing arguments. The default is correct for most
3131 machines. See @file{function.c} for details.
3134 @defmac INITIAL_FRAME_ADDRESS_RTX
3135 A C expression whose value is RTL representing the address of the initial
3136 stack frame. This address is passed to @code{RETURN_ADDR_RTX} and
3137 @code{DYNAMIC_CHAIN_ADDRESS}. If you don't define this macro, a reasonable
3138 default value will be used. Define this macro in order to make frame pointer
3139 elimination work in the presence of @code{__builtin_frame_address (count)} and
3140 @code{__builtin_return_address (count)} for @code{count} not equal to zero.
3143 @defmac DYNAMIC_CHAIN_ADDRESS (@var{frameaddr})
3144 A C expression whose value is RTL representing the address in a stack
3145 frame where the pointer to the caller's frame is stored. Assume that
3146 @var{frameaddr} is an RTL expression for the address of the stack frame
3149 If you don't define this macro, the default is to return the value
3150 of @var{frameaddr}---that is, the stack frame address is also the
3151 address of the stack word that points to the previous frame.
3154 @defmac SETUP_FRAME_ADDRESSES
3155 If defined, a C expression that produces the machine-specific code to
3156 setup the stack so that arbitrary frames can be accessed. For example,
3157 on the SPARC, we must flush all of the register windows to the stack
3158 before we can access arbitrary stack frames. You will seldom need to
3162 @hook TARGET_BUILTIN_SETJMP_FRAME_VALUE
3163 This target hook should return an rtx that is used to store
3164 the address of the current frame into the built in @code{setjmp} buffer.
3165 The default value, @code{virtual_stack_vars_rtx}, is correct for most
3166 machines. One reason you may need to define this target hook is if
3167 @code{hard_frame_pointer_rtx} is the appropriate value on your machine.
3170 @defmac FRAME_ADDR_RTX (@var{frameaddr})
3171 A C expression whose value is RTL representing the value of the frame
3172 address for the current frame. @var{frameaddr} is the frame pointer
3173 of the current frame. This is used for __builtin_frame_address.
3174 You need only define this macro if the frame address is not the same
3175 as the frame pointer. Most machines do not need to define it.
3178 @defmac RETURN_ADDR_RTX (@var{count}, @var{frameaddr})
3179 A C expression whose value is RTL representing the value of the return
3180 address for the frame @var{count} steps up from the current frame, after
3181 the prologue. @var{frameaddr} is the frame pointer of the @var{count}
3182 frame, or the frame pointer of the @var{count} @minus{} 1 frame if
3183 @code{RETURN_ADDR_IN_PREVIOUS_FRAME} is defined.
3185 The value of the expression must always be the correct address when
3186 @var{count} is zero, but may be @code{NULL_RTX} if there is no way to
3187 determine the return address of other frames.
3190 @defmac RETURN_ADDR_IN_PREVIOUS_FRAME
3191 Define this if the return address of a particular stack frame is accessed
3192 from the frame pointer of the previous stack frame.
3195 @defmac INCOMING_RETURN_ADDR_RTX
3196 A C expression whose value is RTL representing the location of the
3197 incoming return address at the beginning of any function, before the
3198 prologue. This RTL is either a @code{REG}, indicating that the return
3199 value is saved in @samp{REG}, or a @code{MEM} representing a location in
3202 You only need to define this macro if you want to support call frame
3203 debugging information like that provided by DWARF 2.
3205 If this RTL is a @code{REG}, you should also define
3206 @code{DWARF_FRAME_RETURN_COLUMN} to @code{DWARF_FRAME_REGNUM (REGNO)}.
3209 @defmac DWARF_ALT_FRAME_RETURN_COLUMN
3210 A C expression whose value is an integer giving a DWARF 2 column
3211 number that may be used as an alternative return column. The column
3212 must not correspond to any gcc hard register (that is, it must not
3213 be in the range of @code{DWARF_FRAME_REGNUM}).
3215 This macro can be useful if @code{DWARF_FRAME_RETURN_COLUMN} is set to a
3216 general register, but an alternative column needs to be used for signal
3217 frames. Some targets have also used different frame return columns
3221 @defmac DWARF_ZERO_REG
3222 A C expression whose value is an integer giving a DWARF 2 register
3223 number that is considered to always have the value zero. This should
3224 only be defined if the target has an architected zero register, and
3225 someone decided it was a good idea to use that register number to
3226 terminate the stack backtrace. New ports should avoid this.
3229 @hook TARGET_DWARF_HANDLE_FRAME_UNSPEC
3230 This target hook allows the backend to emit frame-related insns that
3231 contain UNSPECs or UNSPEC_VOLATILEs. The DWARF 2 call frame debugging
3232 info engine will invoke it on insns of the form
3234 (set (reg) (unspec [@dots{}] UNSPEC_INDEX))
3238 (set (reg) (unspec_volatile [@dots{}] UNSPECV_INDEX)).
3240 to let the backend emit the call frame instructions. @var{label} is
3241 the CFI label attached to the insn, @var{pattern} is the pattern of
3242 the insn and @var{index} is @code{UNSPEC_INDEX} or @code{UNSPECV_INDEX}.
3245 @defmac INCOMING_FRAME_SP_OFFSET
3246 A C expression whose value is an integer giving the offset, in bytes,
3247 from the value of the stack pointer register to the top of the stack
3248 frame at the beginning of any function, before the prologue. The top of
3249 the frame is defined to be the value of the stack pointer in the
3250 previous frame, just before the call instruction.
3252 You only need to define this macro if you want to support call frame
3253 debugging information like that provided by DWARF 2.
3256 @defmac ARG_POINTER_CFA_OFFSET (@var{fundecl})
3257 A C expression whose value is an integer giving the offset, in bytes,
3258 from the argument pointer to the canonical frame address (cfa). The
3259 final value should coincide with that calculated by
3260 @code{INCOMING_FRAME_SP_OFFSET}. Which is unfortunately not usable
3261 during virtual register instantiation.
3263 The default value for this macro is
3264 @code{FIRST_PARM_OFFSET (fundecl) + crtl->args.pretend_args_size},
3265 which is correct for most machines; in general, the arguments are found
3266 immediately before the stack frame. Note that this is not the case on
3267 some targets that save registers into the caller's frame, such as SPARC
3268 and rs6000, and so such targets need to define this macro.
3270 You only need to define this macro if the default is incorrect, and you
3271 want to support call frame debugging information like that provided by
3275 @defmac FRAME_POINTER_CFA_OFFSET (@var{fundecl})
3276 If defined, a C expression whose value is an integer giving the offset
3277 in bytes from the frame pointer to the canonical frame address (cfa).
3278 The final value should coincide with that calculated by
3279 @code{INCOMING_FRAME_SP_OFFSET}.
3281 Normally the CFA is calculated as an offset from the argument pointer,
3282 via @code{ARG_POINTER_CFA_OFFSET}, but if the argument pointer is
3283 variable due to the ABI, this may not be possible. If this macro is
3284 defined, it implies that the virtual register instantiation should be
3285 based on the frame pointer instead of the argument pointer. Only one
3286 of @code{FRAME_POINTER_CFA_OFFSET} and @code{ARG_POINTER_CFA_OFFSET}
3290 @defmac CFA_FRAME_BASE_OFFSET (@var{fundecl})
3291 If defined, a C expression whose value is an integer giving the offset
3292 in bytes from the canonical frame address (cfa) to the frame base used
3293 in DWARF 2 debug information. The default is zero. A different value
3294 may reduce the size of debug information on some ports.
3297 @node Exception Handling
3298 @subsection Exception Handling Support
3299 @cindex exception handling
3301 @defmac EH_RETURN_DATA_REGNO (@var{N})
3302 A C expression whose value is the @var{N}th register number used for
3303 data by exception handlers, or @code{INVALID_REGNUM} if fewer than
3304 @var{N} registers are usable.
3306 The exception handling library routines communicate with the exception
3307 handlers via a set of agreed upon registers. Ideally these registers
3308 should be call-clobbered; it is possible to use call-saved registers,
3309 but may negatively impact code size. The target must support at least
3310 2 data registers, but should define 4 if there are enough free registers.
3312 You must define this macro if you want to support call frame exception
3313 handling like that provided by DWARF 2.
3316 @defmac EH_RETURN_STACKADJ_RTX
3317 A C expression whose value is RTL representing a location in which
3318 to store a stack adjustment to be applied before function return.
3319 This is used to unwind the stack to an exception handler's call frame.
3320 It will be assigned zero on code paths that return normally.
3322 Typically this is a call-clobbered hard register that is otherwise
3323 untouched by the epilogue, but could also be a stack slot.
3325 Do not define this macro if the stack pointer is saved and restored
3326 by the regular prolog and epilog code in the call frame itself; in
3327 this case, the exception handling library routines will update the
3328 stack location to be restored in place. Otherwise, you must define
3329 this macro if you want to support call frame exception handling like
3330 that provided by DWARF 2.
3333 @defmac EH_RETURN_HANDLER_RTX
3334 A C expression whose value is RTL representing a location in which
3335 to store the address of an exception handler to which we should
3336 return. It will not be assigned on code paths that return normally.
3338 Typically this is the location in the call frame at which the normal
3339 return address is stored. For targets that return by popping an
3340 address off the stack, this might be a memory address just below
3341 the @emph{target} call frame rather than inside the current call
3342 frame. If defined, @code{EH_RETURN_STACKADJ_RTX} will have already
3343 been assigned, so it may be used to calculate the location of the
3346 Some targets have more complex requirements than storing to an
3347 address calculable during initial code generation. In that case
3348 the @code{eh_return} instruction pattern should be used instead.
3350 If you want to support call frame exception handling, you must
3351 define either this macro or the @code{eh_return} instruction pattern.
3354 @defmac RETURN_ADDR_OFFSET
3355 If defined, an integer-valued C expression for which rtl will be generated
3356 to add it to the exception handler address before it is searched in the
3357 exception handling tables, and to subtract it again from the address before
3358 using it to return to the exception handler.
3361 @defmac ASM_PREFERRED_EH_DATA_FORMAT (@var{code}, @var{global})
3362 This macro chooses the encoding of pointers embedded in the exception
3363 handling sections. If at all possible, this should be defined such
3364 that the exception handling section will not require dynamic relocations,
3365 and so may be read-only.
3367 @var{code} is 0 for data, 1 for code labels, 2 for function pointers.
3368 @var{global} is true if the symbol may be affected by dynamic relocations.
3369 The macro should return a combination of the @code{DW_EH_PE_*} defines
3370 as found in @file{dwarf2.h}.
3372 If this macro is not defined, pointers will not be encoded but
3373 represented directly.
3376 @defmac ASM_MAYBE_OUTPUT_ENCODED_ADDR_RTX (@var{file}, @var{encoding}, @var{size}, @var{addr}, @var{done})
3377 This macro allows the target to emit whatever special magic is required
3378 to represent the encoding chosen by @code{ASM_PREFERRED_EH_DATA_FORMAT}.
3379 Generic code takes care of pc-relative and indirect encodings; this must
3380 be defined if the target uses text-relative or data-relative encodings.
3382 This is a C statement that branches to @var{done} if the format was
3383 handled. @var{encoding} is the format chosen, @var{size} is the number
3384 of bytes that the format occupies, @var{addr} is the @code{SYMBOL_REF}
3388 @defmac MD_FALLBACK_FRAME_STATE_FOR (@var{context}, @var{fs})
3389 This macro allows the target to add CPU and operating system specific
3390 code to the call-frame unwinder for use when there is no unwind data
3391 available. The most common reason to implement this macro is to unwind
3392 through signal frames.
3394 This macro is called from @code{uw_frame_state_for} in
3395 @file{unwind-dw2.c}, @file{unwind-dw2-xtensa.c} and
3396 @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context};
3397 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{context->ra}
3398 for the address of the code being executed and @code{context->cfa} for
3399 the stack pointer value. If the frame can be decoded, the register
3400 save addresses should be updated in @var{fs} and the macro should
3401 evaluate to @code{_URC_NO_REASON}. If the frame cannot be decoded,
3402 the macro should evaluate to @code{_URC_END_OF_STACK}.
3404 For proper signal handling in Java this macro is accompanied by
3405 @code{MAKE_THROW_FRAME}, defined in @file{libjava/include/*-signal.h} headers.
3408 @defmac MD_HANDLE_UNWABI (@var{context}, @var{fs})
3409 This macro allows the target to add operating system specific code to the
3410 call-frame unwinder to handle the IA-64 @code{.unwabi} unwinding directive,
3411 usually used for signal or interrupt frames.
3413 This macro is called from @code{uw_update_context} in libgcc's
3414 @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context};
3415 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{fs->unwabi}
3416 for the abi and context in the @code{.unwabi} directive. If the
3417 @code{.unwabi} directive can be handled, the register save addresses should
3418 be updated in @var{fs}.
3421 @defmac TARGET_USES_WEAK_UNWIND_INFO
3422 A C expression that evaluates to true if the target requires unwind
3423 info to be given comdat linkage. Define it to be @code{1} if comdat
3424 linkage is necessary. The default is @code{0}.
3427 @node Stack Checking
3428 @subsection Specifying How Stack Checking is Done
3430 GCC will check that stack references are within the boundaries of the
3431 stack, if the option @option{-fstack-check} is specified, in one of
3436 If the value of the @code{STACK_CHECK_BUILTIN} macro is nonzero, GCC
3437 will assume that you have arranged for full stack checking to be done
3438 at appropriate places in the configuration files. GCC will not do
3439 other special processing.
3442 If @code{STACK_CHECK_BUILTIN} is zero and the value of the
3443 @code{STACK_CHECK_STATIC_BUILTIN} macro is nonzero, GCC will assume
3444 that you have arranged for static stack checking (checking of the
3445 static stack frame of functions) to be done at appropriate places
3446 in the configuration files. GCC will only emit code to do dynamic
3447 stack checking (checking on dynamic stack allocations) using the third
3451 If neither of the above are true, GCC will generate code to periodically
3452 ``probe'' the stack pointer using the values of the macros defined below.
3455 If neither STACK_CHECK_BUILTIN nor STACK_CHECK_STATIC_BUILTIN is defined,
3456 GCC will change its allocation strategy for large objects if the option
3457 @option{-fstack-check} is specified: they will always be allocated
3458 dynamically if their size exceeds @code{STACK_CHECK_MAX_VAR_SIZE} bytes.
3460 @defmac STACK_CHECK_BUILTIN
3461 A nonzero value if stack checking is done by the configuration files in a
3462 machine-dependent manner. You should define this macro if stack checking
3463 is required by the ABI of your machine or if you would like to do stack
3464 checking in some more efficient way than the generic approach. The default
3465 value of this macro is zero.
3468 @defmac STACK_CHECK_STATIC_BUILTIN
3469 A nonzero value if static stack checking is done by the configuration files
3470 in a machine-dependent manner. You should define this macro if you would
3471 like to do static stack checking in some more efficient way than the generic
3472 approach. The default value of this macro is zero.
3475 @defmac STACK_CHECK_PROBE_INTERVAL_EXP
3476 An integer specifying the interval at which GCC must generate stack probe
3477 instructions, defined as 2 raised to this integer. You will normally
3478 define this macro so that the interval be no larger than the size of
3479 the ``guard pages'' at the end of a stack area. The default value
3480 of 12 (4096-byte interval) is suitable for most systems.
3483 @defmac STACK_CHECK_MOVING_SP
3484 An integer which is nonzero if GCC should move the stack pointer page by page
3485 when doing probes. This can be necessary on systems where the stack pointer
3486 contains the bottom address of the memory area accessible to the executing
3487 thread at any point in time. In this situation an alternate signal stack
3488 is required in order to be able to recover from a stack overflow. The
3489 default value of this macro is zero.
3492 @defmac STACK_CHECK_PROTECT
3493 The number of bytes of stack needed to recover from a stack overflow, for
3494 languages where such a recovery is supported. The default value of 75 words
3495 with the @code{setjmp}/@code{longjmp}-based exception handling mechanism and
3496 8192 bytes with other exception handling mechanisms should be adequate for
3500 The following macros are relevant only if neither STACK_CHECK_BUILTIN
3501 nor STACK_CHECK_STATIC_BUILTIN is defined; you can omit them altogether
3502 in the opposite case.
3504 @defmac STACK_CHECK_MAX_FRAME_SIZE
3505 The maximum size of a stack frame, in bytes. GCC will generate probe
3506 instructions in non-leaf functions to ensure at least this many bytes of
3507 stack are available. If a stack frame is larger than this size, stack
3508 checking will not be reliable and GCC will issue a warning. The
3509 default is chosen so that GCC only generates one instruction on most
3510 systems. You should normally not change the default value of this macro.
3513 @defmac STACK_CHECK_FIXED_FRAME_SIZE
3514 GCC uses this value to generate the above warning message. It
3515 represents the amount of fixed frame used by a function, not including
3516 space for any callee-saved registers, temporaries and user variables.
3517 You need only specify an upper bound for this amount and will normally
3518 use the default of four words.
3521 @defmac STACK_CHECK_MAX_VAR_SIZE
3522 The maximum size, in bytes, of an object that GCC will place in the
3523 fixed area of the stack frame when the user specifies
3524 @option{-fstack-check}.
3525 GCC computed the default from the values of the above macros and you will
3526 normally not need to override that default.
3530 @node Frame Registers
3531 @subsection Registers That Address the Stack Frame
3533 @c prevent bad page break with this line
3534 This discusses registers that address the stack frame.
3536 @defmac STACK_POINTER_REGNUM
3537 The register number of the stack pointer register, which must also be a
3538 fixed register according to @code{FIXED_REGISTERS}. On most machines,
3539 the hardware determines which register this is.
3542 @defmac FRAME_POINTER_REGNUM
3543 The register number of the frame pointer register, which is used to
3544 access automatic variables in the stack frame. On some machines, the
3545 hardware determines which register this is. On other machines, you can
3546 choose any register you wish for this purpose.
3549 @defmac HARD_FRAME_POINTER_REGNUM
3550 On some machines the offset between the frame pointer and starting
3551 offset of the automatic variables is not known until after register
3552 allocation has been done (for example, because the saved registers are
3553 between these two locations). On those machines, define
3554 @code{FRAME_POINTER_REGNUM} the number of a special, fixed register to
3555 be used internally until the offset is known, and define
3556 @code{HARD_FRAME_POINTER_REGNUM} to be the actual hard register number
3557 used for the frame pointer.
3559 You should define this macro only in the very rare circumstances when it
3560 is not possible to calculate the offset between the frame pointer and
3561 the automatic variables until after register allocation has been
3562 completed. When this macro is defined, you must also indicate in your
3563 definition of @code{ELIMINABLE_REGS} how to eliminate
3564 @code{FRAME_POINTER_REGNUM} into either @code{HARD_FRAME_POINTER_REGNUM}
3565 or @code{STACK_POINTER_REGNUM}.
3567 Do not define this macro if it would be the same as
3568 @code{FRAME_POINTER_REGNUM}.
3571 @defmac ARG_POINTER_REGNUM
3572 The register number of the arg pointer register, which is used to access
3573 the function's argument list. On some machines, this is the same as the
3574 frame pointer register. On some machines, the hardware determines which
3575 register this is. On other machines, you can choose any register you
3576 wish for this purpose. If this is not the same register as the frame
3577 pointer register, then you must mark it as a fixed register according to
3578 @code{FIXED_REGISTERS}, or arrange to be able to eliminate it
3579 (@pxref{Elimination}).
3582 @defmac HARD_FRAME_POINTER_IS_FRAME_POINTER
3583 Define this to a preprocessor constant that is nonzero if
3584 @code{hard_frame_pointer_rtx} and @code{frame_pointer_rtx} should be
3585 the same. The default definition is @samp{(HARD_FRAME_POINTER_REGNUM
3586 == FRAME_POINTER_REGNUM)}; you only need to define this macro if that
3587 definition is not suitable for use in preprocessor conditionals.
3590 @defmac HARD_FRAME_POINTER_IS_ARG_POINTER
3591 Define this to a preprocessor constant that is nonzero if
3592 @code{hard_frame_pointer_rtx} and @code{arg_pointer_rtx} should be the
3593 same. The default definition is @samp{(HARD_FRAME_POINTER_REGNUM ==
3594 ARG_POINTER_REGNUM)}; you only need to define this macro if that
3595 definition is not suitable for use in preprocessor conditionals.
3598 @defmac RETURN_ADDRESS_POINTER_REGNUM
3599 The register number of the return address pointer register, which is used to
3600 access the current function's return address from the stack. On some
3601 machines, the return address is not at a fixed offset from the frame
3602 pointer or stack pointer or argument pointer. This register can be defined
3603 to point to the return address on the stack, and then be converted by
3604 @code{ELIMINABLE_REGS} into either the frame pointer or stack pointer.
3606 Do not define this macro unless there is no other way to get the return
3607 address from the stack.
3610 @defmac STATIC_CHAIN_REGNUM
3611 @defmacx STATIC_CHAIN_INCOMING_REGNUM
3612 Register numbers used for passing a function's static chain pointer. If
3613 register windows are used, the register number as seen by the called
3614 function is @code{STATIC_CHAIN_INCOMING_REGNUM}, while the register
3615 number as seen by the calling function is @code{STATIC_CHAIN_REGNUM}. If
3616 these registers are the same, @code{STATIC_CHAIN_INCOMING_REGNUM} need
3619 The static chain register need not be a fixed register.
3621 If the static chain is passed in memory, these macros should not be
3622 defined; instead, the @code{TARGET_STATIC_CHAIN} hook should be used.
3625 @hook TARGET_STATIC_CHAIN
3626 This hook replaces the use of @code{STATIC_CHAIN_REGNUM} et al for
3627 targets that may use different static chain locations for different
3628 nested functions. This may be required if the target has function
3629 attributes that affect the calling conventions of the function and
3630 those calling conventions use different static chain locations.
3632 The default version of this hook uses @code{STATIC_CHAIN_REGNUM} et al.
3634 If the static chain is passed in memory, this hook should be used to
3635 provide rtx giving @code{mem} expressions that denote where they are stored.
3636 Often the @code{mem} expression as seen by the caller will be at an offset
3637 from the stack pointer and the @code{mem} expression as seen by the callee
3638 will be at an offset from the frame pointer.
3639 @findex stack_pointer_rtx
3640 @findex frame_pointer_rtx
3641 @findex arg_pointer_rtx
3642 The variables @code{stack_pointer_rtx}, @code{frame_pointer_rtx}, and
3643 @code{arg_pointer_rtx} will have been initialized and should be used
3644 to refer to those items.
3647 @defmac DWARF_FRAME_REGISTERS
3648 This macro specifies the maximum number of hard registers that can be
3649 saved in a call frame. This is used to size data structures used in
3650 DWARF2 exception handling.
3652 Prior to GCC 3.0, this macro was needed in order to establish a stable
3653 exception handling ABI in the face of adding new hard registers for ISA
3654 extensions. In GCC 3.0 and later, the EH ABI is insulated from changes
3655 in the number of hard registers. Nevertheless, this macro can still be
3656 used to reduce the runtime memory requirements of the exception handling
3657 routines, which can be substantial if the ISA contains a lot of
3658 registers that are not call-saved.
3660 If this macro is not defined, it defaults to
3661 @code{FIRST_PSEUDO_REGISTER}.
3664 @defmac PRE_GCC3_DWARF_FRAME_REGISTERS
3666 This macro is similar to @code{DWARF_FRAME_REGISTERS}, but is provided
3667 for backward compatibility in pre GCC 3.0 compiled code.
3669 If this macro is not defined, it defaults to
3670 @code{DWARF_FRAME_REGISTERS}.
3673 @defmac DWARF_REG_TO_UNWIND_COLUMN (@var{regno})
3675 Define this macro if the target's representation for dwarf registers
3676 is different than the internal representation for unwind column.
3677 Given a dwarf register, this macro should return the internal unwind
3678 column number to use instead.
3680 See the PowerPC's SPE target for an example.
3683 @defmac DWARF_FRAME_REGNUM (@var{regno})
3685 Define this macro if the target's representation for dwarf registers
3686 used in .eh_frame or .debug_frame is different from that used in other
3687 debug info sections. Given a GCC hard register number, this macro
3688 should return the .eh_frame register number. The default is
3689 @code{DBX_REGISTER_NUMBER (@var{regno})}.
3693 @defmac DWARF2_FRAME_REG_OUT (@var{regno}, @var{for_eh})
3695 Define this macro to map register numbers held in the call frame info
3696 that GCC has collected using @code{DWARF_FRAME_REGNUM} to those that
3697 should be output in .debug_frame (@code{@var{for_eh}} is zero) and
3698 .eh_frame (@code{@var{for_eh}} is nonzero). The default is to
3699 return @code{@var{regno}}.
3703 @defmac REG_VALUE_IN_UNWIND_CONTEXT
3705 Define this macro if the target stores register values as
3706 @code{_Unwind_Word} type in unwind context. It should be defined if
3707 target register size is larger than the size of @code{void *}. The
3708 default is to store register values as @code{void *} type.
3712 @defmac ASSUME_EXTENDED_UNWIND_CONTEXT
3714 Define this macro to be 1 if the target always uses extended unwind
3715 context with version, args_size and by_value fields. If it is undefined,
3716 it will be defined to 1 when @code{REG_VALUE_IN_UNWIND_CONTEXT} is
3717 defined and 0 otherwise.
3722 @subsection Eliminating Frame Pointer and Arg Pointer
3724 @c prevent bad page break with this line
3725 This is about eliminating the frame pointer and arg pointer.
3727 @hook TARGET_FRAME_POINTER_REQUIRED
3728 This target hook should return @code{true} if a function must have and use
3729 a frame pointer. This target hook is called in the reload pass. If its return
3730 value is @code{true} the function will have a frame pointer.
3732 This target hook can in principle examine the current function and decide
3733 according to the facts, but on most machines the constant @code{false} or the
3734 constant @code{true} suffices. Use @code{false} when the machine allows code
3735 to be generated with no frame pointer, and doing so saves some time or space.
3736 Use @code{true} when there is no possible advantage to avoiding a frame
3739 In certain cases, the compiler does not know how to produce valid code
3740 without a frame pointer. The compiler recognizes those cases and
3741 automatically gives the function a frame pointer regardless of what
3742 @code{TARGET_FRAME_POINTER_REQUIRED} returns. You don't need to worry about
3745 In a function that does not require a frame pointer, the frame pointer
3746 register can be allocated for ordinary usage, unless you mark it as a
3747 fixed register. See @code{FIXED_REGISTERS} for more information.
3749 Default return value is @code{false}.
3752 @findex get_frame_size
3753 @defmac INITIAL_FRAME_POINTER_OFFSET (@var{depth-var})
3754 A C statement to store in the variable @var{depth-var} the difference
3755 between the frame pointer and the stack pointer values immediately after
3756 the function prologue. The value would be computed from information
3757 such as the result of @code{get_frame_size ()} and the tables of
3758 registers @code{regs_ever_live} and @code{call_used_regs}.
3760 If @code{ELIMINABLE_REGS} is defined, this macro will be not be used and
3761 need not be defined. Otherwise, it must be defined even if
3762 @code{TARGET_FRAME_POINTER_REQUIRED} always returns true; in that
3763 case, you may set @var{depth-var} to anything.
3766 @defmac ELIMINABLE_REGS
3767 If defined, this macro specifies a table of register pairs used to
3768 eliminate unneeded registers that point into the stack frame. If it is not
3769 defined, the only elimination attempted by the compiler is to replace
3770 references to the frame pointer with references to the stack pointer.
3772 The definition of this macro is a list of structure initializations, each
3773 of which specifies an original and replacement register.
3775 On some machines, the position of the argument pointer is not known until
3776 the compilation is completed. In such a case, a separate hard register
3777 must be used for the argument pointer. This register can be eliminated by
3778 replacing it with either the frame pointer or the argument pointer,
3779 depending on whether or not the frame pointer has been eliminated.
3781 In this case, you might specify:
3783 #define ELIMINABLE_REGS \
3784 @{@{ARG_POINTER_REGNUM, STACK_POINTER_REGNUM@}, \
3785 @{ARG_POINTER_REGNUM, FRAME_POINTER_REGNUM@}, \
3786 @{FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM@}@}
3789 Note that the elimination of the argument pointer with the stack pointer is
3790 specified first since that is the preferred elimination.
3793 @hook TARGET_CAN_ELIMINATE
3794 This target hook should returns @code{true} if the compiler is allowed to
3795 try to replace register number @var{from_reg} with register number
3796 @var{to_reg}. This target hook need only be defined if @code{ELIMINABLE_REGS}
3797 is defined, and will usually be @code{true}, since most of the cases
3798 preventing register elimination are things that the compiler already
3801 Default return value is @code{true}.
3804 @defmac INITIAL_ELIMINATION_OFFSET (@var{from-reg}, @var{to-reg}, @var{offset-var})
3805 This macro is similar to @code{INITIAL_FRAME_POINTER_OFFSET}. It
3806 specifies the initial difference between the specified pair of
3807 registers. This macro must be defined if @code{ELIMINABLE_REGS} is
3811 @node Stack Arguments
3812 @subsection Passing Function Arguments on the Stack
3813 @cindex arguments on stack
3814 @cindex stack arguments
3816 The macros in this section control how arguments are passed
3817 on the stack. See the following section for other macros that
3818 control passing certain arguments in registers.
3820 @hook TARGET_PROMOTE_PROTOTYPES
3821 This target hook returns @code{true} if an argument declared in a
3822 prototype as an integral type smaller than @code{int} should actually be
3823 passed as an @code{int}. In addition to avoiding errors in certain
3824 cases of mismatch, it also makes for better code on certain machines.
3825 The default is to not promote prototypes.
3829 A C expression. If nonzero, push insns will be used to pass
3831 If the target machine does not have a push instruction, set it to zero.
3832 That directs GCC to use an alternate strategy: to
3833 allocate the entire argument block and then store the arguments into
3834 it. When @code{PUSH_ARGS} is nonzero, @code{PUSH_ROUNDING} must be defined too.
3837 @defmac PUSH_ARGS_REVERSED
3838 A C expression. If nonzero, function arguments will be evaluated from
3839 last to first, rather than from first to last. If this macro is not
3840 defined, it defaults to @code{PUSH_ARGS} on targets where the stack
3841 and args grow in opposite directions, and 0 otherwise.
3844 @defmac PUSH_ROUNDING (@var{npushed})
3845 A C expression that is the number of bytes actually pushed onto the
3846 stack when an instruction attempts to push @var{npushed} bytes.
3848 On some machines, the definition
3851 #define PUSH_ROUNDING(BYTES) (BYTES)
3855 will suffice. But on other machines, instructions that appear
3856 to push one byte actually push two bytes in an attempt to maintain
3857 alignment. Then the definition should be
3860 #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1)
3863 If the value of this macro has a type, it should be an unsigned type.
3866 @findex outgoing_args_size
3867 @findex crtl->outgoing_args_size
3868 @defmac ACCUMULATE_OUTGOING_ARGS
3869 A C expression. If nonzero, the maximum amount of space required for outgoing arguments
3870 will be computed and placed into
3871 @code{crtl->outgoing_args_size}. No space will be pushed
3872 onto the stack for each call; instead, the function prologue should
3873 increase the stack frame size by this amount.
3875 Setting both @code{PUSH_ARGS} and @code{ACCUMULATE_OUTGOING_ARGS}
3879 @defmac REG_PARM_STACK_SPACE (@var{fndecl})
3880 Define this macro if functions should assume that stack space has been
3881 allocated for arguments even when their values are passed in
3884 The value of this macro is the size, in bytes, of the area reserved for
3885 arguments passed in registers for the function represented by @var{fndecl},
3886 which can be zero if GCC is calling a library function.
3887 The argument @var{fndecl} can be the FUNCTION_DECL, or the type itself
3890 This space can be allocated by the caller, or be a part of the
3891 machine-dependent stack frame: @code{OUTGOING_REG_PARM_STACK_SPACE} says
3894 @c above is overfull. not sure what to do. --mew 5feb93 did
3895 @c something, not sure if it looks good. --mew 10feb93
3897 @defmac OUTGOING_REG_PARM_STACK_SPACE (@var{fntype})
3898 Define this to a nonzero value if it is the responsibility of the
3899 caller to allocate the area reserved for arguments passed in registers
3900 when calling a function of @var{fntype}. @var{fntype} may be NULL
3901 if the function called is a library function.
3903 If @code{ACCUMULATE_OUTGOING_ARGS} is defined, this macro controls
3904 whether the space for these arguments counts in the value of
3905 @code{crtl->outgoing_args_size}.
3908 @defmac STACK_PARMS_IN_REG_PARM_AREA
3909 Define this macro if @code{REG_PARM_STACK_SPACE} is defined, but the
3910 stack parameters don't skip the area specified by it.
3911 @c i changed this, makes more sens and it should have taken care of the
3912 @c overfull.. not as specific, tho. --mew 5feb93
3914 Normally, when a parameter is not passed in registers, it is placed on the
3915 stack beyond the @code{REG_PARM_STACK_SPACE} area. Defining this macro
3916 suppresses this behavior and causes the parameter to be passed on the
3917 stack in its natural location.
3920 @hook TARGET_RETURN_POPS_ARGS
3921 This target hook returns the number of bytes of its own arguments that
3922 a function pops on returning, or 0 if the function pops no arguments
3923 and the caller must therefore pop them all after the function returns.
3925 @var{fundecl} is a C variable whose value is a tree node that describes
3926 the function in question. Normally it is a node of type
3927 @code{FUNCTION_DECL} that describes the declaration of the function.
3928 From this you can obtain the @code{DECL_ATTRIBUTES} of the function.
3930 @var{funtype} is a C variable whose value is a tree node that
3931 describes the function in question. Normally it is a node of type
3932 @code{FUNCTION_TYPE} that describes the data type of the function.
3933 From this it is possible to obtain the data types of the value and
3934 arguments (if known).
3936 When a call to a library function is being considered, @var{fundecl}
3937 will contain an identifier node for the library function. Thus, if
3938 you need to distinguish among various library functions, you can do so
3939 by their names. Note that ``library function'' in this context means
3940 a function used to perform arithmetic, whose name is known specially
3941 in the compiler and was not mentioned in the C code being compiled.
3943 @var{size} is the number of bytes of arguments passed on the
3944 stack. If a variable number of bytes is passed, it is zero, and
3945 argument popping will always be the responsibility of the calling function.
3947 On the VAX, all functions always pop their arguments, so the definition
3948 of this macro is @var{size}. On the 68000, using the standard
3949 calling convention, no functions pop their arguments, so the value of
3950 the macro is always 0 in this case. But an alternative calling
3951 convention is available in which functions that take a fixed number of
3952 arguments pop them but other functions (such as @code{printf}) pop
3953 nothing (the caller pops all). When this convention is in use,
3954 @var{funtype} is examined to determine whether a function takes a fixed
3955 number of arguments.
3958 @defmac CALL_POPS_ARGS (@var{cum})
3959 A C expression that should indicate the number of bytes a call sequence
3960 pops off the stack. It is added to the value of @code{RETURN_POPS_ARGS}
3961 when compiling a function call.
3963 @var{cum} is the variable in which all arguments to the called function
3964 have been accumulated.
3966 On certain architectures, such as the SH5, a call trampoline is used
3967 that pops certain registers off the stack, depending on the arguments
3968 that have been passed to the function. Since this is a property of the
3969 call site, not of the called function, @code{RETURN_POPS_ARGS} is not
3973 @node Register Arguments
3974 @subsection Passing Arguments in Registers
3975 @cindex arguments in registers
3976 @cindex registers arguments
3978 This section describes the macros which let you control how various
3979 types of arguments are passed in registers or how they are arranged in
3982 @hook TARGET_FUNCTION_ARG
3983 Return an RTX indicating whether a function argument is passed in a
3984 register and if so, which register.
3986 The arguments are @var{ca}, which summarizes all the previous
3987 arguments; @var{mode}, the machine mode of the argument; @var{type},
3988 the data type of the argument as a tree node or 0 if that is not known
3989 (which happens for C support library functions); and @var{named},
3990 which is @code{true} for an ordinary argument and @code{false} for
3991 nameless arguments that correspond to @samp{@dots{}} in the called
3992 function's prototype. @var{type} can be an incomplete type if a
3993 syntax error has previously occurred.
3995 The return value is usually either a @code{reg} RTX for the hard
3996 register in which to pass the argument, or zero to pass the argument
3999 The value of the expression can also be a @code{parallel} RTX@. This is
4000 used when an argument is passed in multiple locations. The mode of the
4001 @code{parallel} should be the mode of the entire argument. The
4002 @code{parallel} holds any number of @code{expr_list} pairs; each one
4003 describes where part of the argument is passed. In each
4004 @code{expr_list} the first operand must be a @code{reg} RTX for the hard
4005 register in which to pass this part of the argument, and the mode of the
4006 register RTX indicates how large this part of the argument is. The
4007 second operand of the @code{expr_list} is a @code{const_int} which gives
4008 the offset in bytes into the entire argument of where this part starts.
4009 As a special exception the first @code{expr_list} in the @code{parallel}
4010 RTX may have a first operand of zero. This indicates that the entire
4011 argument is also stored on the stack.
4013 The last time this hook is called, it is called with @code{MODE ==
4014 VOIDmode}, and its result is passed to the @code{call} or @code{call_value}
4015 pattern as operands 2 and 3 respectively.
4017 @cindex @file{stdarg.h} and register arguments
4018 The usual way to make the ISO library @file{stdarg.h} work on a
4019 machine where some arguments are usually passed in registers, is to
4020 cause nameless arguments to be passed on the stack instead. This is
4021 done by making @code{TARGET_FUNCTION_ARG} return 0 whenever
4022 @var{named} is @code{false}.
4024 @cindex @code{TARGET_MUST_PASS_IN_STACK}, and @code{TARGET_FUNCTION_ARG}
4025 @cindex @code{REG_PARM_STACK_SPACE}, and @code{TARGET_FUNCTION_ARG}
4026 You may use the hook @code{targetm.calls.must_pass_in_stack}
4027 in the definition of this macro to determine if this argument is of a
4028 type that must be passed in the stack. If @code{REG_PARM_STACK_SPACE}
4029 is not defined and @code{TARGET_FUNCTION_ARG} returns nonzero for such an
4030 argument, the compiler will abort. If @code{REG_PARM_STACK_SPACE} is
4031 defined, the argument will be computed in the stack and then loaded into
4035 @hook TARGET_MUST_PASS_IN_STACK
4036 This target hook should return @code{true} if we should not pass @var{type}
4037 solely in registers. The file @file{expr.h} defines a
4038 definition that is usually appropriate, refer to @file{expr.h} for additional
4042 @hook TARGET_FUNCTION_INCOMING_ARG
4043 Define this hook if the target machine has ``register windows'', so
4044 that the register in which a function sees an arguments is not
4045 necessarily the same as the one in which the caller passed the
4048 For such machines, @code{TARGET_FUNCTION_ARG} computes the register in
4049 which the caller passes the value, and
4050 @code{TARGET_FUNCTION_INCOMING_ARG} should be defined in a similar
4051 fashion to tell the function being called where the arguments will
4054 If @code{TARGET_FUNCTION_INCOMING_ARG} is not defined,
4055 @code{TARGET_FUNCTION_ARG} serves both purposes.
4058 @hook TARGET_ARG_PARTIAL_BYTES
4059 This target hook returns the number of bytes at the beginning of an
4060 argument that must be put in registers. The value must be zero for
4061 arguments that are passed entirely in registers or that are entirely
4062 pushed on the stack.
4064 On some machines, certain arguments must be passed partially in
4065 registers and partially in memory. On these machines, typically the
4066 first few words of arguments are passed in registers, and the rest
4067 on the stack. If a multi-word argument (a @code{double} or a
4068 structure) crosses that boundary, its first few words must be passed
4069 in registers and the rest must be pushed. This macro tells the
4070 compiler when this occurs, and how many bytes should go in registers.
4072 @code{TARGET_FUNCTION_ARG} for these arguments should return the first
4073 register to be used by the caller for this argument; likewise
4074 @code{TARGET_FUNCTION_INCOMING_ARG}, for the called function.
4077 @hook TARGET_PASS_BY_REFERENCE
4078 This target hook should return @code{true} if an argument at the
4079 position indicated by @var{cum} should be passed by reference. This
4080 predicate is queried after target independent reasons for being
4081 passed by reference, such as @code{TREE_ADDRESSABLE (type)}.
4083 If the hook returns true, a copy of that argument is made in memory and a
4084 pointer to the argument is passed instead of the argument itself.
4085 The pointer is passed in whatever way is appropriate for passing a pointer
4089 @hook TARGET_CALLEE_COPIES
4090 The function argument described by the parameters to this hook is
4091 known to be passed by reference. The hook should return true if the
4092 function argument should be copied by the callee instead of copied
4095 For any argument for which the hook returns true, if it can be
4096 determined that the argument is not modified, then a copy need
4099 The default version of this hook always returns false.
4102 @defmac CUMULATIVE_ARGS
4103 A C type for declaring a variable that is used as the first argument
4104 of @code{TARGET_FUNCTION_ARG} and other related values. For some
4105 target machines, the type @code{int} suffices and can hold the number
4106 of bytes of argument so far.
4108 There is no need to record in @code{CUMULATIVE_ARGS} anything about the
4109 arguments that have been passed on the stack. The compiler has other
4110 variables to keep track of that. For target machines on which all
4111 arguments are passed on the stack, there is no need to store anything in
4112 @code{CUMULATIVE_ARGS}; however, the data structure must exist and
4113 should not be empty, so use @code{int}.
4116 @defmac OVERRIDE_ABI_FORMAT (@var{fndecl})
4117 If defined, this macro is called before generating any code for a
4118 function, but after the @var{cfun} descriptor for the function has been
4119 created. The back end may use this macro to update @var{cfun} to
4120 reflect an ABI other than that which would normally be used by default.
4121 If the compiler is generating code for a compiler-generated function,
4122 @var{fndecl} may be @code{NULL}.
4125 @defmac INIT_CUMULATIVE_ARGS (@var{cum}, @var{fntype}, @var{libname}, @var{fndecl}, @var{n_named_args})
4126 A C statement (sans semicolon) for initializing the variable
4127 @var{cum} for the state at the beginning of the argument list. The
4128 variable has type @code{CUMULATIVE_ARGS}. The value of @var{fntype}
4129 is the tree node for the data type of the function which will receive
4130 the args, or 0 if the args are to a compiler support library function.
4131 For direct calls that are not libcalls, @var{fndecl} contain the
4132 declaration node of the function. @var{fndecl} is also set when
4133 @code{INIT_CUMULATIVE_ARGS} is used to find arguments for the function
4134 being compiled. @var{n_named_args} is set to the number of named
4135 arguments, including a structure return address if it is passed as a
4136 parameter, when making a call. When processing incoming arguments,
4137 @var{n_named_args} is set to @minus{}1.
4139 When processing a call to a compiler support library function,
4140 @var{libname} identifies which one. It is a @code{symbol_ref} rtx which
4141 contains the name of the function, as a string. @var{libname} is 0 when
4142 an ordinary C function call is being processed. Thus, each time this
4143 macro is called, either @var{libname} or @var{fntype} is nonzero, but
4144 never both of them at once.
4147 @defmac INIT_CUMULATIVE_LIBCALL_ARGS (@var{cum}, @var{mode}, @var{libname})
4148 Like @code{INIT_CUMULATIVE_ARGS} but only used for outgoing libcalls,
4149 it gets a @code{MODE} argument instead of @var{fntype}, that would be
4150 @code{NULL}. @var{indirect} would always be zero, too. If this macro
4151 is not defined, @code{INIT_CUMULATIVE_ARGS (cum, NULL_RTX, libname,
4152 0)} is used instead.
4155 @defmac INIT_CUMULATIVE_INCOMING_ARGS (@var{cum}, @var{fntype}, @var{libname})
4156 Like @code{INIT_CUMULATIVE_ARGS} but overrides it for the purposes of
4157 finding the arguments for the function being compiled. If this macro is
4158 undefined, @code{INIT_CUMULATIVE_ARGS} is used instead.
4160 The value passed for @var{libname} is always 0, since library routines
4161 with special calling conventions are never compiled with GCC@. The
4162 argument @var{libname} exists for symmetry with
4163 @code{INIT_CUMULATIVE_ARGS}.
4164 @c could use "this macro" in place of @code{INIT_CUMULATIVE_ARGS}, maybe.
4165 @c --mew 5feb93 i switched the order of the sentences. --mew 10feb93
4168 @hook TARGET_FUNCTION_ARG_ADVANCE
4169 This hook updates the summarizer variable pointed to by @var{ca} to
4170 advance past an argument in the argument list. The values @var{mode},
4171 @var{type} and @var{named} describe that argument. Once this is done,
4172 the variable @var{cum} is suitable for analyzing the @emph{following}
4173 argument with @code{TARGET_FUNCTION_ARG}, etc.
4175 This hook need not do anything if the argument in question was passed
4176 on the stack. The compiler knows how to track the amount of stack space
4177 used for arguments without any special help.
4180 @defmac FUNCTION_ARG_OFFSET (@var{mode}, @var{type})
4181 If defined, a C expression that is the number of bytes to add to the
4182 offset of the argument passed in memory. This is needed for the SPU,
4183 which passes @code{char} and @code{short} arguments in the preferred
4184 slot that is in the middle of the quad word instead of starting at the
4188 @defmac FUNCTION_ARG_PADDING (@var{mode}, @var{type})
4189 If defined, a C expression which determines whether, and in which direction,
4190 to pad out an argument with extra space. The value should be of type
4191 @code{enum direction}: either @code{upward} to pad above the argument,
4192 @code{downward} to pad below, or @code{none} to inhibit padding.
4194 The @emph{amount} of padding is not controlled by this macro, but by the
4195 target hook @code{TARGET_FUNCTION_ARG_ROUND_BOUNDARY}. It is
4196 always just enough to reach the next multiple of that boundary.
4198 This macro has a default definition which is right for most systems.
4199 For little-endian machines, the default is to pad upward. For
4200 big-endian machines, the default is to pad downward for an argument of
4201 constant size shorter than an @code{int}, and upward otherwise.
4204 @defmac PAD_VARARGS_DOWN
4205 If defined, a C expression which determines whether the default
4206 implementation of va_arg will attempt to pad down before reading the
4207 next argument, if that argument is smaller than its aligned space as
4208 controlled by @code{PARM_BOUNDARY}. If this macro is not defined, all such
4209 arguments are padded down if @code{BYTES_BIG_ENDIAN} is true.
4212 @defmac BLOCK_REG_PADDING (@var{mode}, @var{type}, @var{first})
4213 Specify padding for the last element of a block move between registers and
4214 memory. @var{first} is nonzero if this is the only element. Defining this
4215 macro allows better control of register function parameters on big-endian
4216 machines, without using @code{PARALLEL} rtl. In particular,
4217 @code{MUST_PASS_IN_STACK} need not test padding and mode of types in
4218 registers, as there is no longer a "wrong" part of a register; For example,
4219 a three byte aggregate may be passed in the high part of a register if so
4223 @hook TARGET_FUNCTION_ARG_BOUNDARY
4224 This hook returns the alignment boundary, in bits, of an argument
4225 with the specified mode and type. The default hook returns
4226 @code{PARM_BOUNDARY} for all arguments.
4229 @hook TARGET_FUNCTION_ARG_ROUND_BOUNDARY
4231 @defmac FUNCTION_ARG_REGNO_P (@var{regno})
4232 A C expression that is nonzero if @var{regno} is the number of a hard
4233 register in which function arguments are sometimes passed. This does
4234 @emph{not} include implicit arguments such as the static chain and
4235 the structure-value address. On many machines, no registers can be
4236 used for this purpose since all function arguments are pushed on the
4240 @hook TARGET_SPLIT_COMPLEX_ARG
4241 This hook should return true if parameter of type @var{type} are passed
4242 as two scalar parameters. By default, GCC will attempt to pack complex
4243 arguments into the target's word size. Some ABIs require complex arguments
4244 to be split and treated as their individual components. For example, on
4245 AIX64, complex floats should be passed in a pair of floating point
4246 registers, even though a complex float would fit in one 64-bit floating
4249 The default value of this hook is @code{NULL}, which is treated as always
4253 @hook TARGET_BUILD_BUILTIN_VA_LIST
4254 This hook returns a type node for @code{va_list} for the target.
4255 The default version of the hook returns @code{void*}.
4258 @hook TARGET_ENUM_VA_LIST_P
4259 This target hook is used in function @code{c_common_nodes_and_builtins}
4260 to iterate through the target specific builtin types for va_list. The
4261 variable @var{idx} is used as iterator. @var{pname} has to be a pointer
4262 to a @code{const char *} and @var{ptree} a pointer to a @code{tree} typed
4264 The arguments @var{pname} and @var{ptree} are used to store the result of
4265 this macro and are set to the name of the va_list builtin type and its
4267 If the return value of this macro is zero, then there is no more element.
4268 Otherwise the @var{IDX} should be increased for the next call of this
4269 macro to iterate through all types.
4272 @hook TARGET_FN_ABI_VA_LIST
4273 This hook returns the va_list type of the calling convention specified by
4275 The default version of this hook returns @code{va_list_type_node}.
4278 @hook TARGET_CANONICAL_VA_LIST_TYPE
4279 This hook returns the va_list type of the calling convention specified by the
4280 type of @var{type}. If @var{type} is not a valid va_list type, it returns
4284 @hook TARGET_GIMPLIFY_VA_ARG_EXPR
4285 This hook performs target-specific gimplification of
4286 @code{VA_ARG_EXPR}. The first two parameters correspond to the
4287 arguments to @code{va_arg}; the latter two are as in
4288 @code{gimplify.c:gimplify_expr}.
4291 @hook TARGET_VALID_POINTER_MODE
4292 Define this to return nonzero if the port can handle pointers
4293 with machine mode @var{mode}. The default version of this
4294 hook returns true for both @code{ptr_mode} and @code{Pmode}.
4297 @hook TARGET_REF_MAY_ALIAS_ERRNO
4299 @hook TARGET_SCALAR_MODE_SUPPORTED_P
4300 Define this to return nonzero if the port is prepared to handle
4301 insns involving scalar mode @var{mode}. For a scalar mode to be
4302 considered supported, all the basic arithmetic and comparisons
4305 The default version of this hook returns true for any mode
4306 required to handle the basic C types (as defined by the port).
4307 Included here are the double-word arithmetic supported by the
4308 code in @file{optabs.c}.
4311 @hook TARGET_VECTOR_MODE_SUPPORTED_P
4312 Define this to return nonzero if the port is prepared to handle
4313 insns involving vector mode @var{mode}. At the very least, it
4314 must have move patterns for this mode.
4317 @hook TARGET_ARRAY_MODE_SUPPORTED_P
4319 @hook TARGET_SMALL_REGISTER_CLASSES_FOR_MODE_P
4320 Define this to return nonzero for machine modes for which the port has
4321 small register classes. If this target hook returns nonzero for a given
4322 @var{mode}, the compiler will try to minimize the lifetime of registers
4323 in @var{mode}. The hook may be called with @code{VOIDmode} as argument.
4324 In this case, the hook is expected to return nonzero if it returns nonzero
4327 On some machines, it is risky to let hard registers live across arbitrary
4328 insns. Typically, these machines have instructions that require values
4329 to be in specific registers (like an accumulator), and reload will fail
4330 if the required hard register is used for another purpose across such an
4333 Passes before reload do not know which hard registers will be used
4334 in an instruction, but the machine modes of the registers set or used in
4335 the instruction are already known. And for some machines, register
4336 classes are small for, say, integer registers but not for floating point
4337 registers. For example, the AMD x86-64 architecture requires specific
4338 registers for the legacy x86 integer instructions, but there are many
4339 SSE registers for floating point operations. On such targets, a good
4340 strategy may be to return nonzero from this hook for @code{INTEGRAL_MODE_P}
4341 machine modes but zero for the SSE register classes.
4343 The default version of this hook returns false for any mode. It is always
4344 safe to redefine this hook to return with a nonzero value. But if you
4345 unnecessarily define it, you will reduce the amount of optimizations
4346 that can be performed in some cases. If you do not define this hook
4347 to return a nonzero value when it is required, the compiler will run out
4348 of spill registers and print a fatal error message.
4351 @hook TARGET_FLAGS_REGNUM
4354 @subsection How Scalar Function Values Are Returned
4355 @cindex return values in registers
4356 @cindex values, returned by functions
4357 @cindex scalars, returned as values
4359 This section discusses the macros that control returning scalars as
4360 values---values that can fit in registers.
4362 @hook TARGET_FUNCTION_VALUE
4364 Define this to return an RTX representing the place where a function
4365 returns or receives a value of data type @var{ret_type}, a tree node
4366 representing a data type. @var{fn_decl_or_type} is a tree node
4367 representing @code{FUNCTION_DECL} or @code{FUNCTION_TYPE} of a
4368 function being called. If @var{outgoing} is false, the hook should
4369 compute the register in which the caller will see the return value.
4370 Otherwise, the hook should return an RTX representing the place where
4371 a function returns a value.
4373 On many machines, only @code{TYPE_MODE (@var{ret_type})} is relevant.
4374 (Actually, on most machines, scalar values are returned in the same
4375 place regardless of mode.) The value of the expression is usually a
4376 @code{reg} RTX for the hard register where the return value is stored.
4377 The value can also be a @code{parallel} RTX, if the return value is in
4378 multiple places. See @code{TARGET_FUNCTION_ARG} for an explanation of the
4379 @code{parallel} form. Note that the callee will populate every
4380 location specified in the @code{parallel}, but if the first element of
4381 the @code{parallel} contains the whole return value, callers will use
4382 that element as the canonical location and ignore the others. The m68k
4383 port uses this type of @code{parallel} to return pointers in both
4384 @samp{%a0} (the canonical location) and @samp{%d0}.
4386 If @code{TARGET_PROMOTE_FUNCTION_RETURN} returns true, you must apply
4387 the same promotion rules specified in @code{PROMOTE_MODE} if
4388 @var{valtype} is a scalar type.
4390 If the precise function being called is known, @var{func} is a tree
4391 node (@code{FUNCTION_DECL}) for it; otherwise, @var{func} is a null
4392 pointer. This makes it possible to use a different value-returning
4393 convention for specific functions when all their calls are
4396 Some target machines have ``register windows'' so that the register in
4397 which a function returns its value is not the same as the one in which
4398 the caller sees the value. For such machines, you should return
4399 different RTX depending on @var{outgoing}.
4401 @code{TARGET_FUNCTION_VALUE} is not used for return values with
4402 aggregate data types, because these are returned in another way. See
4403 @code{TARGET_STRUCT_VALUE_RTX} and related macros, below.
4406 @defmac FUNCTION_VALUE (@var{valtype}, @var{func})
4407 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE} for
4408 a new target instead.
4411 @defmac LIBCALL_VALUE (@var{mode})
4412 A C expression to create an RTX representing the place where a library
4413 function returns a value of mode @var{mode}.
4415 Note that ``library function'' in this context means a compiler
4416 support routine, used to perform arithmetic, whose name is known
4417 specially by the compiler and was not mentioned in the C code being
4421 @hook TARGET_LIBCALL_VALUE
4422 Define this hook if the back-end needs to know the name of the libcall
4423 function in order to determine where the result should be returned.
4425 The mode of the result is given by @var{mode} and the name of the called
4426 library function is given by @var{fun}. The hook should return an RTX
4427 representing the place where the library function result will be returned.
4429 If this hook is not defined, then LIBCALL_VALUE will be used.
4432 @defmac FUNCTION_VALUE_REGNO_P (@var{regno})
4433 A C expression that is nonzero if @var{regno} is the number of a hard
4434 register in which the values of called function may come back.
4436 A register whose use for returning values is limited to serving as the
4437 second of a pair (for a value of type @code{double}, say) need not be
4438 recognized by this macro. So for most machines, this definition
4442 #define FUNCTION_VALUE_REGNO_P(N) ((N) == 0)
4445 If the machine has register windows, so that the caller and the called
4446 function use different registers for the return value, this macro
4447 should recognize only the caller's register numbers.
4449 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE_REGNO_P}
4450 for a new target instead.
4453 @hook TARGET_FUNCTION_VALUE_REGNO_P
4454 A target hook that return @code{true} if @var{regno} is the number of a hard
4455 register in which the values of called function may come back.
4457 A register whose use for returning values is limited to serving as the
4458 second of a pair (for a value of type @code{double}, say) need not be
4459 recognized by this target hook.
4461 If the machine has register windows, so that the caller and the called
4462 function use different registers for the return value, this target hook
4463 should recognize only the caller's register numbers.
4465 If this hook is not defined, then FUNCTION_VALUE_REGNO_P will be used.
4468 @defmac APPLY_RESULT_SIZE
4469 Define this macro if @samp{untyped_call} and @samp{untyped_return}
4470 need more space than is implied by @code{FUNCTION_VALUE_REGNO_P} for
4471 saving and restoring an arbitrary return value.
4474 @hook TARGET_RETURN_IN_MSB
4475 This hook should return true if values of type @var{type} are returned
4476 at the most significant end of a register (in other words, if they are
4477 padded at the least significant end). You can assume that @var{type}
4478 is returned in a register; the caller is required to check this.
4480 Note that the register provided by @code{TARGET_FUNCTION_VALUE} must
4481 be able to hold the complete return value. For example, if a 1-, 2-
4482 or 3-byte structure is returned at the most significant end of a
4483 4-byte register, @code{TARGET_FUNCTION_VALUE} should provide an
4487 @node Aggregate Return
4488 @subsection How Large Values Are Returned
4489 @cindex aggregates as return values
4490 @cindex large return values
4491 @cindex returning aggregate values
4492 @cindex structure value address
4494 When a function value's mode is @code{BLKmode} (and in some other
4495 cases), the value is not returned according to
4496 @code{TARGET_FUNCTION_VALUE} (@pxref{Scalar Return}). Instead, the
4497 caller passes the address of a block of memory in which the value
4498 should be stored. This address is called the @dfn{structure value
4501 This section describes how to control returning structure values in
4504 @hook TARGET_RETURN_IN_MEMORY
4505 This target hook should return a nonzero value to say to return the
4506 function value in memory, just as large structures are always returned.
4507 Here @var{type} will be the data type of the value, and @var{fntype}
4508 will be the type of the function doing the returning, or @code{NULL} for
4511 Note that values of mode @code{BLKmode} must be explicitly handled
4512 by this function. Also, the option @option{-fpcc-struct-return}
4513 takes effect regardless of this macro. On most systems, it is
4514 possible to leave the hook undefined; this causes a default
4515 definition to be used, whose value is the constant 1 for @code{BLKmode}
4516 values, and 0 otherwise.
4518 Do not use this hook to indicate that structures and unions should always
4519 be returned in memory. You should instead use @code{DEFAULT_PCC_STRUCT_RETURN}
4523 @defmac DEFAULT_PCC_STRUCT_RETURN
4524 Define this macro to be 1 if all structure and union return values must be
4525 in memory. Since this results in slower code, this should be defined
4526 only if needed for compatibility with other compilers or with an ABI@.
4527 If you define this macro to be 0, then the conventions used for structure
4528 and union return values are decided by the @code{TARGET_RETURN_IN_MEMORY}
4531 If not defined, this defaults to the value 1.
4534 @hook TARGET_STRUCT_VALUE_RTX
4535 This target hook should return the location of the structure value
4536 address (normally a @code{mem} or @code{reg}), or 0 if the address is
4537 passed as an ``invisible'' first argument. Note that @var{fndecl} may
4538 be @code{NULL}, for libcalls. You do not need to define this target
4539 hook if the address is always passed as an ``invisible'' first
4542 On some architectures the place where the structure value address
4543 is found by the called function is not the same place that the
4544 caller put it. This can be due to register windows, or it could
4545 be because the function prologue moves it to a different place.
4546 @var{incoming} is @code{1} or @code{2} when the location is needed in
4547 the context of the called function, and @code{0} in the context of
4550 If @var{incoming} is nonzero and the address is to be found on the
4551 stack, return a @code{mem} which refers to the frame pointer. If
4552 @var{incoming} is @code{2}, the result is being used to fetch the
4553 structure value address at the beginning of a function. If you need
4554 to emit adjusting code, you should do it at this point.
4557 @defmac PCC_STATIC_STRUCT_RETURN
4558 Define this macro if the usual system convention on the target machine
4559 for returning structures and unions is for the called function to return
4560 the address of a static variable containing the value.
4562 Do not define this if the usual system convention is for the caller to
4563 pass an address to the subroutine.
4565 This macro has effect in @option{-fpcc-struct-return} mode, but it does
4566 nothing when you use @option{-freg-struct-return} mode.
4569 @hook TARGET_GET_RAW_RESULT_MODE
4571 @hook TARGET_GET_RAW_ARG_MODE
4574 @subsection Caller-Saves Register Allocation
4576 If you enable it, GCC can save registers around function calls. This
4577 makes it possible to use call-clobbered registers to hold variables that
4578 must live across calls.
4580 @defmac CALLER_SAVE_PROFITABLE (@var{refs}, @var{calls})
4581 A C expression to determine whether it is worthwhile to consider placing
4582 a pseudo-register in a call-clobbered hard register and saving and
4583 restoring it around each function call. The expression should be 1 when
4584 this is worth doing, and 0 otherwise.
4586 If you don't define this macro, a default is used which is good on most
4587 machines: @code{4 * @var{calls} < @var{refs}}.
4590 @defmac HARD_REGNO_CALLER_SAVE_MODE (@var{regno}, @var{nregs})
4591 A C expression specifying which mode is required for saving @var{nregs}
4592 of a pseudo-register in call-clobbered hard register @var{regno}. If
4593 @var{regno} is unsuitable for caller save, @code{VOIDmode} should be
4594 returned. For most machines this macro need not be defined since GCC
4595 will select the smallest suitable mode.
4598 @node Function Entry
4599 @subsection Function Entry and Exit
4600 @cindex function entry and exit
4604 This section describes the macros that output function entry
4605 (@dfn{prologue}) and exit (@dfn{epilogue}) code.
4607 @hook TARGET_ASM_FUNCTION_PROLOGUE
4608 If defined, a function that outputs the assembler code for entry to a
4609 function. The prologue is responsible for setting up the stack frame,
4610 initializing the frame pointer register, saving registers that must be
4611 saved, and allocating @var{size} additional bytes of storage for the
4612 local variables. @var{size} is an integer. @var{file} is a stdio
4613 stream to which the assembler code should be output.
4615 The label for the beginning of the function need not be output by this
4616 macro. That has already been done when the macro is run.
4618 @findex regs_ever_live
4619 To determine which registers to save, the macro can refer to the array
4620 @code{regs_ever_live}: element @var{r} is nonzero if hard register
4621 @var{r} is used anywhere within the function. This implies the function
4622 prologue should save register @var{r}, provided it is not one of the
4623 call-used registers. (@code{TARGET_ASM_FUNCTION_EPILOGUE} must likewise use
4624 @code{regs_ever_live}.)
4626 On machines that have ``register windows'', the function entry code does
4627 not save on the stack the registers that are in the windows, even if
4628 they are supposed to be preserved by function calls; instead it takes
4629 appropriate steps to ``push'' the register stack, if any non-call-used
4630 registers are used in the function.
4632 @findex frame_pointer_needed
4633 On machines where functions may or may not have frame-pointers, the
4634 function entry code must vary accordingly; it must set up the frame
4635 pointer if one is wanted, and not otherwise. To determine whether a
4636 frame pointer is in wanted, the macro can refer to the variable
4637 @code{frame_pointer_needed}. The variable's value will be 1 at run
4638 time in a function that needs a frame pointer. @xref{Elimination}.
4640 The function entry code is responsible for allocating any stack space
4641 required for the function. This stack space consists of the regions
4642 listed below. In most cases, these regions are allocated in the
4643 order listed, with the last listed region closest to the top of the
4644 stack (the lowest address if @code{STACK_GROWS_DOWNWARD} is defined, and
4645 the highest address if it is not defined). You can use a different order
4646 for a machine if doing so is more convenient or required for
4647 compatibility reasons. Except in cases where required by standard
4648 or by a debugger, there is no reason why the stack layout used by GCC
4649 need agree with that used by other compilers for a machine.
4652 @hook TARGET_ASM_FUNCTION_END_PROLOGUE
4653 If defined, a function that outputs assembler code at the end of a
4654 prologue. This should be used when the function prologue is being
4655 emitted as RTL, and you have some extra assembler that needs to be
4656 emitted. @xref{prologue instruction pattern}.
4659 @hook TARGET_ASM_FUNCTION_BEGIN_EPILOGUE
4660 If defined, a function that outputs assembler code at the start of an
4661 epilogue. This should be used when the function epilogue is being
4662 emitted as RTL, and you have some extra assembler that needs to be
4663 emitted. @xref{epilogue instruction pattern}.
4666 @hook TARGET_ASM_FUNCTION_EPILOGUE
4667 If defined, a function that outputs the assembler code for exit from a
4668 function. The epilogue is responsible for restoring the saved
4669 registers and stack pointer to their values when the function was
4670 called, and returning control to the caller. This macro takes the
4671 same arguments as the macro @code{TARGET_ASM_FUNCTION_PROLOGUE}, and the
4672 registers to restore are determined from @code{regs_ever_live} and
4673 @code{CALL_USED_REGISTERS} in the same way.
4675 On some machines, there is a single instruction that does all the work
4676 of returning from the function. On these machines, give that
4677 instruction the name @samp{return} and do not define the macro
4678 @code{TARGET_ASM_FUNCTION_EPILOGUE} at all.
4680 Do not define a pattern named @samp{return} if you want the
4681 @code{TARGET_ASM_FUNCTION_EPILOGUE} to be used. If you want the target
4682 switches to control whether return instructions or epilogues are used,
4683 define a @samp{return} pattern with a validity condition that tests the
4684 target switches appropriately. If the @samp{return} pattern's validity
4685 condition is false, epilogues will be used.
4687 On machines where functions may or may not have frame-pointers, the
4688 function exit code must vary accordingly. Sometimes the code for these
4689 two cases is completely different. To determine whether a frame pointer
4690 is wanted, the macro can refer to the variable
4691 @code{frame_pointer_needed}. The variable's value will be 1 when compiling
4692 a function that needs a frame pointer.
4694 Normally, @code{TARGET_ASM_FUNCTION_PROLOGUE} and
4695 @code{TARGET_ASM_FUNCTION_EPILOGUE} must treat leaf functions specially.
4696 The C variable @code{current_function_is_leaf} is nonzero for such a
4697 function. @xref{Leaf Functions}.
4699 On some machines, some functions pop their arguments on exit while
4700 others leave that for the caller to do. For example, the 68020 when
4701 given @option{-mrtd} pops arguments in functions that take a fixed
4702 number of arguments.
4705 @findex crtl->args.pops_args
4706 Your definition of the macro @code{RETURN_POPS_ARGS} decides which
4707 functions pop their own arguments. @code{TARGET_ASM_FUNCTION_EPILOGUE}
4708 needs to know what was decided. The number of bytes of the current
4709 function's arguments that this function should pop is available in
4710 @code{crtl->args.pops_args}. @xref{Scalar Return}.
4715 @findex pretend_args_size
4716 @findex crtl->args.pretend_args_size
4717 A region of @code{crtl->args.pretend_args_size} bytes of
4718 uninitialized space just underneath the first argument arriving on the
4719 stack. (This may not be at the very start of the allocated stack region
4720 if the calling sequence has pushed anything else since pushing the stack
4721 arguments. But usually, on such machines, nothing else has been pushed
4722 yet, because the function prologue itself does all the pushing.) This
4723 region is used on machines where an argument may be passed partly in
4724 registers and partly in memory, and, in some cases to support the
4725 features in @code{<stdarg.h>}.
4728 An area of memory used to save certain registers used by the function.
4729 The size of this area, which may also include space for such things as
4730 the return address and pointers to previous stack frames, is
4731 machine-specific and usually depends on which registers have been used
4732 in the function. Machines with register windows often do not require
4736 A region of at least @var{size} bytes, possibly rounded up to an allocation
4737 boundary, to contain the local variables of the function. On some machines,
4738 this region and the save area may occur in the opposite order, with the
4739 save area closer to the top of the stack.
4742 @cindex @code{ACCUMULATE_OUTGOING_ARGS} and stack frames
4743 Optionally, when @code{ACCUMULATE_OUTGOING_ARGS} is defined, a region of
4744 @code{crtl->outgoing_args_size} bytes to be used for outgoing
4745 argument lists of the function. @xref{Stack Arguments}.
4748 @defmac EXIT_IGNORE_STACK
4749 Define this macro as a C expression that is nonzero if the return
4750 instruction or the function epilogue ignores the value of the stack
4751 pointer; in other words, if it is safe to delete an instruction to
4752 adjust the stack pointer before a return from the function. The
4755 Note that this macro's value is relevant only for functions for which
4756 frame pointers are maintained. It is never safe to delete a final
4757 stack adjustment in a function that has no frame pointer, and the
4758 compiler knows this regardless of @code{EXIT_IGNORE_STACK}.
4761 @defmac EPILOGUE_USES (@var{regno})
4762 Define this macro as a C expression that is nonzero for registers that are
4763 used by the epilogue or the @samp{return} pattern. The stack and frame
4764 pointer registers are already assumed to be used as needed.
4767 @defmac EH_USES (@var{regno})
4768 Define this macro as a C expression that is nonzero for registers that are
4769 used by the exception handling mechanism, and so should be considered live
4770 on entry to an exception edge.
4773 @defmac DELAY_SLOTS_FOR_EPILOGUE
4774 Define this macro if the function epilogue contains delay slots to which
4775 instructions from the rest of the function can be ``moved''. The
4776 definition should be a C expression whose value is an integer
4777 representing the number of delay slots there.
4780 @defmac ELIGIBLE_FOR_EPILOGUE_DELAY (@var{insn}, @var{n})
4781 A C expression that returns 1 if @var{insn} can be placed in delay
4782 slot number @var{n} of the epilogue.
4784 The argument @var{n} is an integer which identifies the delay slot now
4785 being considered (since different slots may have different rules of
4786 eligibility). It is never negative and is always less than the number
4787 of epilogue delay slots (what @code{DELAY_SLOTS_FOR_EPILOGUE} returns).
4788 If you reject a particular insn for a given delay slot, in principle, it
4789 may be reconsidered for a subsequent delay slot. Also, other insns may
4790 (at least in principle) be considered for the so far unfilled delay
4793 @findex epilogue_delay_list
4794 @findex crtl->epilogue_delay_list
4795 @findex final_scan_insn
4796 The insns accepted to fill the epilogue delay slots are put in an RTL
4797 list made with @code{insn_list} objects, stored in
4798 @code{crtl->epilogue_delay_list}. The insn for the first
4799 delay slot comes first in the list. Your definition of the macro
4800 @code{TARGET_ASM_FUNCTION_EPILOGUE} should fill the delay slots by
4801 outputting the insns in this list, usually by calling
4802 @code{final_scan_insn}.
4804 You need not define this macro if you did not define
4805 @code{DELAY_SLOTS_FOR_EPILOGUE}.
4808 @hook TARGET_ASM_OUTPUT_MI_THUNK
4809 A function that outputs the assembler code for a thunk
4810 function, used to implement C++ virtual function calls with multiple
4811 inheritance. The thunk acts as a wrapper around a virtual function,
4812 adjusting the implicit object parameter before handing control off to
4815 First, emit code to add the integer @var{delta} to the location that
4816 contains the incoming first argument. Assume that this argument
4817 contains a pointer, and is the one used to pass the @code{this} pointer
4818 in C++. This is the incoming argument @emph{before} the function prologue,
4819 e.g.@: @samp{%o0} on a sparc. The addition must preserve the values of
4820 all other incoming arguments.
4822 Then, if @var{vcall_offset} is nonzero, an additional adjustment should be
4823 made after adding @code{delta}. In particular, if @var{p} is the
4824 adjusted pointer, the following adjustment should be made:
4827 p += (*((ptrdiff_t **)p))[vcall_offset/sizeof(ptrdiff_t)]
4830 After the additions, emit code to jump to @var{function}, which is a
4831 @code{FUNCTION_DECL}. This is a direct pure jump, not a call, and does
4832 not touch the return address. Hence returning from @var{FUNCTION} will
4833 return to whoever called the current @samp{thunk}.
4835 The effect must be as if @var{function} had been called directly with
4836 the adjusted first argument. This macro is responsible for emitting all
4837 of the code for a thunk function; @code{TARGET_ASM_FUNCTION_PROLOGUE}
4838 and @code{TARGET_ASM_FUNCTION_EPILOGUE} are not invoked.
4840 The @var{thunk_fndecl} is redundant. (@var{delta} and @var{function}
4841 have already been extracted from it.) It might possibly be useful on
4842 some targets, but probably not.
4844 If you do not define this macro, the target-independent code in the C++
4845 front end will generate a less efficient heavyweight thunk that calls
4846 @var{function} instead of jumping to it. The generic approach does
4847 not support varargs.
4850 @hook TARGET_ASM_CAN_OUTPUT_MI_THUNK
4851 A function that returns true if TARGET_ASM_OUTPUT_MI_THUNK would be able
4852 to output the assembler code for the thunk function specified by the
4853 arguments it is passed, and false otherwise. In the latter case, the
4854 generic approach will be used by the C++ front end, with the limitations
4859 @subsection Generating Code for Profiling
4860 @cindex profiling, code generation
4862 These macros will help you generate code for profiling.
4864 @defmac FUNCTION_PROFILER (@var{file}, @var{labelno})
4865 A C statement or compound statement to output to @var{file} some
4866 assembler code to call the profiling subroutine @code{mcount}.
4869 The details of how @code{mcount} expects to be called are determined by
4870 your operating system environment, not by GCC@. To figure them out,
4871 compile a small program for profiling using the system's installed C
4872 compiler and look at the assembler code that results.
4874 Older implementations of @code{mcount} expect the address of a counter
4875 variable to be loaded into some register. The name of this variable is
4876 @samp{LP} followed by the number @var{labelno}, so you would generate
4877 the name using @samp{LP%d} in a @code{fprintf}.
4880 @defmac PROFILE_HOOK
4881 A C statement or compound statement to output to @var{file} some assembly
4882 code to call the profiling subroutine @code{mcount} even the target does
4883 not support profiling.
4886 @defmac NO_PROFILE_COUNTERS
4887 Define this macro to be an expression with a nonzero value if the
4888 @code{mcount} subroutine on your system does not need a counter variable
4889 allocated for each function. This is true for almost all modern
4890 implementations. If you define this macro, you must not use the
4891 @var{labelno} argument to @code{FUNCTION_PROFILER}.
4894 @defmac PROFILE_BEFORE_PROLOGUE
4895 Define this macro if the code for function profiling should come before
4896 the function prologue. Normally, the profiling code comes after.
4900 @subsection Permitting tail calls
4903 @hook TARGET_FUNCTION_OK_FOR_SIBCALL
4904 True if it is ok to do sibling call optimization for the specified
4905 call expression @var{exp}. @var{decl} will be the called function,
4906 or @code{NULL} if this is an indirect call.
4908 It is not uncommon for limitations of calling conventions to prevent
4909 tail calls to functions outside the current unit of translation, or
4910 during PIC compilation. The hook is used to enforce these restrictions,
4911 as the @code{sibcall} md pattern can not fail, or fall over to a
4912 ``normal'' call. The criteria for successful sibling call optimization
4913 may vary greatly between different architectures.
4916 @hook TARGET_EXTRA_LIVE_ON_ENTRY
4917 Add any hard registers to @var{regs} that are live on entry to the
4918 function. This hook only needs to be defined to provide registers that
4919 cannot be found by examination of FUNCTION_ARG_REGNO_P, the callee saved
4920 registers, STATIC_CHAIN_INCOMING_REGNUM, STATIC_CHAIN_REGNUM,
4921 TARGET_STRUCT_VALUE_RTX, FRAME_POINTER_REGNUM, EH_USES,
4922 FRAME_POINTER_REGNUM, ARG_POINTER_REGNUM, and the PIC_OFFSET_TABLE_REGNUM.
4925 @hook TARGET_SET_UP_BY_PROLOGUE
4927 @hook TARGET_WARN_FUNC_RETURN
4929 @node Stack Smashing Protection
4930 @subsection Stack smashing protection
4931 @cindex stack smashing protection
4933 @hook TARGET_STACK_PROTECT_GUARD
4934 This hook returns a @code{DECL} node for the external variable to use
4935 for the stack protection guard. This variable is initialized by the
4936 runtime to some random value and is used to initialize the guard value
4937 that is placed at the top of the local stack frame. The type of this
4938 variable must be @code{ptr_type_node}.
4940 The default version of this hook creates a variable called
4941 @samp{__stack_chk_guard}, which is normally defined in @file{libgcc2.c}.
4944 @hook TARGET_STACK_PROTECT_FAIL
4945 This hook returns a @code{CALL_EXPR} that alerts the runtime that the
4946 stack protect guard variable has been modified. This expression should
4947 involve a call to a @code{noreturn} function.
4949 The default version of this hook invokes a function called
4950 @samp{__stack_chk_fail}, taking no arguments. This function is
4951 normally defined in @file{libgcc2.c}.
4954 @hook TARGET_SUPPORTS_SPLIT_STACK
4957 @section Implementing the Varargs Macros
4958 @cindex varargs implementation
4960 GCC comes with an implementation of @code{<varargs.h>} and
4961 @code{<stdarg.h>} that work without change on machines that pass arguments
4962 on the stack. Other machines require their own implementations of
4963 varargs, and the two machine independent header files must have
4964 conditionals to include it.
4966 ISO @code{<stdarg.h>} differs from traditional @code{<varargs.h>} mainly in
4967 the calling convention for @code{va_start}. The traditional
4968 implementation takes just one argument, which is the variable in which
4969 to store the argument pointer. The ISO implementation of
4970 @code{va_start} takes an additional second argument. The user is
4971 supposed to write the last named argument of the function here.
4973 However, @code{va_start} should not use this argument. The way to find
4974 the end of the named arguments is with the built-in functions described
4977 @defmac __builtin_saveregs ()
4978 Use this built-in function to save the argument registers in memory so
4979 that the varargs mechanism can access them. Both ISO and traditional
4980 versions of @code{va_start} must use @code{__builtin_saveregs}, unless
4981 you use @code{TARGET_SETUP_INCOMING_VARARGS} (see below) instead.
4983 On some machines, @code{__builtin_saveregs} is open-coded under the
4984 control of the target hook @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. On
4985 other machines, it calls a routine written in assembler language,
4986 found in @file{libgcc2.c}.
4988 Code generated for the call to @code{__builtin_saveregs} appears at the
4989 beginning of the function, as opposed to where the call to
4990 @code{__builtin_saveregs} is written, regardless of what the code is.
4991 This is because the registers must be saved before the function starts
4992 to use them for its own purposes.
4993 @c i rewrote the first sentence above to fix an overfull hbox. --mew
4997 @defmac __builtin_next_arg (@var{lastarg})
4998 This builtin returns the address of the first anonymous stack
4999 argument, as type @code{void *}. If @code{ARGS_GROW_DOWNWARD}, it
5000 returns the address of the location above the first anonymous stack
5001 argument. Use it in @code{va_start} to initialize the pointer for
5002 fetching arguments from the stack. Also use it in @code{va_start} to
5003 verify that the second parameter @var{lastarg} is the last named argument
5004 of the current function.
5007 @defmac __builtin_classify_type (@var{object})
5008 Since each machine has its own conventions for which data types are
5009 passed in which kind of register, your implementation of @code{va_arg}
5010 has to embody these conventions. The easiest way to categorize the
5011 specified data type is to use @code{__builtin_classify_type} together
5012 with @code{sizeof} and @code{__alignof__}.
5014 @code{__builtin_classify_type} ignores the value of @var{object},
5015 considering only its data type. It returns an integer describing what
5016 kind of type that is---integer, floating, pointer, structure, and so on.
5018 The file @file{typeclass.h} defines an enumeration that you can use to
5019 interpret the values of @code{__builtin_classify_type}.
5022 These machine description macros help implement varargs:
5024 @hook TARGET_EXPAND_BUILTIN_SAVEREGS
5025 If defined, this hook produces the machine-specific code for a call to
5026 @code{__builtin_saveregs}. This code will be moved to the very
5027 beginning of the function, before any parameter access are made. The
5028 return value of this function should be an RTX that contains the value
5029 to use as the return of @code{__builtin_saveregs}.
5032 @hook TARGET_SETUP_INCOMING_VARARGS
5033 This target hook offers an alternative to using
5034 @code{__builtin_saveregs} and defining the hook
5035 @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. Use it to store the anonymous
5036 register arguments into the stack so that all the arguments appear to
5037 have been passed consecutively on the stack. Once this is done, you can
5038 use the standard implementation of varargs that works for machines that
5039 pass all their arguments on the stack.
5041 The argument @var{args_so_far} points to the @code{CUMULATIVE_ARGS} data
5042 structure, containing the values that are obtained after processing the
5043 named arguments. The arguments @var{mode} and @var{type} describe the
5044 last named argument---its machine mode and its data type as a tree node.
5046 The target hook should do two things: first, push onto the stack all the
5047 argument registers @emph{not} used for the named arguments, and second,
5048 store the size of the data thus pushed into the @code{int}-valued
5049 variable pointed to by @var{pretend_args_size}. The value that you
5050 store here will serve as additional offset for setting up the stack
5053 Because you must generate code to push the anonymous arguments at
5054 compile time without knowing their data types,
5055 @code{TARGET_SETUP_INCOMING_VARARGS} is only useful on machines that
5056 have just a single category of argument register and use it uniformly
5059 If the argument @var{second_time} is nonzero, it means that the
5060 arguments of the function are being analyzed for the second time. This
5061 happens for an inline function, which is not actually compiled until the
5062 end of the source file. The hook @code{TARGET_SETUP_INCOMING_VARARGS} should
5063 not generate any instructions in this case.
5066 @hook TARGET_STRICT_ARGUMENT_NAMING
5067 Define this hook to return @code{true} if the location where a function
5068 argument is passed depends on whether or not it is a named argument.
5070 This hook controls how the @var{named} argument to @code{TARGET_FUNCTION_ARG}
5071 is set for varargs and stdarg functions. If this hook returns
5072 @code{true}, the @var{named} argument is always true for named
5073 arguments, and false for unnamed arguments. If it returns @code{false},
5074 but @code{TARGET_PRETEND_OUTGOING_VARARGS_NAMED} returns @code{true},
5075 then all arguments are treated as named. Otherwise, all named arguments
5076 except the last are treated as named.
5078 You need not define this hook if it always returns @code{false}.
5081 @hook TARGET_PRETEND_OUTGOING_VARARGS_NAMED
5082 If you need to conditionally change ABIs so that one works with
5083 @code{TARGET_SETUP_INCOMING_VARARGS}, but the other works like neither
5084 @code{TARGET_SETUP_INCOMING_VARARGS} nor @code{TARGET_STRICT_ARGUMENT_NAMING} was
5085 defined, then define this hook to return @code{true} if
5086 @code{TARGET_SETUP_INCOMING_VARARGS} is used, @code{false} otherwise.
5087 Otherwise, you should not define this hook.
5091 @section Trampolines for Nested Functions
5092 @cindex trampolines for nested functions
5093 @cindex nested functions, trampolines for
5095 A @dfn{trampoline} is a small piece of code that is created at run time
5096 when the address of a nested function is taken. It normally resides on
5097 the stack, in the stack frame of the containing function. These macros
5098 tell GCC how to generate code to allocate and initialize a
5101 The instructions in the trampoline must do two things: load a constant
5102 address into the static chain register, and jump to the real address of
5103 the nested function. On CISC machines such as the m68k, this requires
5104 two instructions, a move immediate and a jump. Then the two addresses
5105 exist in the trampoline as word-long immediate operands. On RISC
5106 machines, it is often necessary to load each address into a register in
5107 two parts. Then pieces of each address form separate immediate
5110 The code generated to initialize the trampoline must store the variable
5111 parts---the static chain value and the function address---into the
5112 immediate operands of the instructions. On a CISC machine, this is
5113 simply a matter of copying each address to a memory reference at the
5114 proper offset from the start of the trampoline. On a RISC machine, it
5115 may be necessary to take out pieces of the address and store them
5118 @hook TARGET_ASM_TRAMPOLINE_TEMPLATE
5119 This hook is called by @code{assemble_trampoline_template} to output,
5120 on the stream @var{f}, assembler code for a block of data that contains
5121 the constant parts of a trampoline. This code should not include a
5122 label---the label is taken care of automatically.
5124 If you do not define this hook, it means no template is needed
5125 for the target. Do not define this hook on systems where the block move
5126 code to copy the trampoline into place would be larger than the code
5127 to generate it on the spot.
5130 @defmac TRAMPOLINE_SECTION
5131 Return the section into which the trampoline template is to be placed
5132 (@pxref{Sections}). The default value is @code{readonly_data_section}.
5135 @defmac TRAMPOLINE_SIZE
5136 A C expression for the size in bytes of the trampoline, as an integer.
5139 @defmac TRAMPOLINE_ALIGNMENT
5140 Alignment required for trampolines, in bits.
5142 If you don't define this macro, the value of @code{FUNCTION_ALIGNMENT}
5143 is used for aligning trampolines.
5146 @hook TARGET_TRAMPOLINE_INIT
5147 This hook is called to initialize a trampoline.
5148 @var{m_tramp} is an RTX for the memory block for the trampoline; @var{fndecl}
5149 is the @code{FUNCTION_DECL} for the nested function; @var{static_chain} is an
5150 RTX for the static chain value that should be passed to the function
5153 If the target defines @code{TARGET_ASM_TRAMPOLINE_TEMPLATE}, then the
5154 first thing this hook should do is emit a block move into @var{m_tramp}
5155 from the memory block returned by @code{assemble_trampoline_template}.
5156 Note that the block move need only cover the constant parts of the
5157 trampoline. If the target isolates the variable parts of the trampoline
5158 to the end, not all @code{TRAMPOLINE_SIZE} bytes need be copied.
5160 If the target requires any other actions, such as flushing caches or
5161 enabling stack execution, these actions should be performed after
5162 initializing the trampoline proper.
5165 @hook TARGET_TRAMPOLINE_ADJUST_ADDRESS
5166 This hook should perform any machine-specific adjustment in
5167 the address of the trampoline. Its argument contains the address of the
5168 memory block that was passed to @code{TARGET_TRAMPOLINE_INIT}. In case
5169 the address to be used for a function call should be different from the
5170 address at which the template was stored, the different address should
5171 be returned; otherwise @var{addr} should be returned unchanged.
5172 If this hook is not defined, @var{addr} will be used for function calls.
5175 Implementing trampolines is difficult on many machines because they have
5176 separate instruction and data caches. Writing into a stack location
5177 fails to clear the memory in the instruction cache, so when the program
5178 jumps to that location, it executes the old contents.
5180 Here are two possible solutions. One is to clear the relevant parts of
5181 the instruction cache whenever a trampoline is set up. The other is to
5182 make all trampolines identical, by having them jump to a standard
5183 subroutine. The former technique makes trampoline execution faster; the
5184 latter makes initialization faster.
5186 To clear the instruction cache when a trampoline is initialized, define
5187 the following macro.
5189 @defmac CLEAR_INSN_CACHE (@var{beg}, @var{end})
5190 If defined, expands to a C expression clearing the @emph{instruction
5191 cache} in the specified interval. The definition of this macro would
5192 typically be a series of @code{asm} statements. Both @var{beg} and
5193 @var{end} are both pointer expressions.
5196 To use a standard subroutine, define the following macro. In addition,
5197 you must make sure that the instructions in a trampoline fill an entire
5198 cache line with identical instructions, or else ensure that the
5199 beginning of the trampoline code is always aligned at the same point in
5200 its cache line. Look in @file{m68k.h} as a guide.
5202 @defmac TRANSFER_FROM_TRAMPOLINE
5203 Define this macro if trampolines need a special subroutine to do their
5204 work. The macro should expand to a series of @code{asm} statements
5205 which will be compiled with GCC@. They go in a library function named
5206 @code{__transfer_from_trampoline}.
5208 If you need to avoid executing the ordinary prologue code of a compiled
5209 C function when you jump to the subroutine, you can do so by placing a
5210 special label of your own in the assembler code. Use one @code{asm}
5211 statement to generate an assembler label, and another to make the label
5212 global. Then trampolines can use that label to jump directly to your
5213 special assembler code.
5217 @section Implicit Calls to Library Routines
5218 @cindex library subroutine names
5219 @cindex @file{libgcc.a}
5221 @c prevent bad page break with this line
5222 Here is an explanation of implicit calls to library routines.
5224 @defmac DECLARE_LIBRARY_RENAMES
5225 This macro, if defined, should expand to a piece of C code that will get
5226 expanded when compiling functions for libgcc.a. It can be used to
5227 provide alternate names for GCC's internal library functions if there
5228 are ABI-mandated names that the compiler should provide.
5231 @findex set_optab_libfunc
5232 @findex init_one_libfunc
5233 @hook TARGET_INIT_LIBFUNCS
5234 This hook should declare additional library routines or rename
5235 existing ones, using the functions @code{set_optab_libfunc} and
5236 @code{init_one_libfunc} defined in @file{optabs.c}.
5237 @code{init_optabs} calls this macro after initializing all the normal
5240 The default is to do nothing. Most ports don't need to define this hook.
5243 @hook TARGET_LIBFUNC_GNU_PREFIX
5245 @defmac FLOAT_LIB_COMPARE_RETURNS_BOOL (@var{mode}, @var{comparison})
5246 This macro should return @code{true} if the library routine that
5247 implements the floating point comparison operator @var{comparison} in
5248 mode @var{mode} will return a boolean, and @var{false} if it will
5251 GCC's own floating point libraries return tristates from the
5252 comparison operators, so the default returns false always. Most ports
5253 don't need to define this macro.
5256 @defmac TARGET_LIB_INT_CMP_BIASED
5257 This macro should evaluate to @code{true} if the integer comparison
5258 functions (like @code{__cmpdi2}) return 0 to indicate that the first
5259 operand is smaller than the second, 1 to indicate that they are equal,
5260 and 2 to indicate that the first operand is greater than the second.
5261 If this macro evaluates to @code{false} the comparison functions return
5262 @minus{}1, 0, and 1 instead of 0, 1, and 2. If the target uses the routines
5263 in @file{libgcc.a}, you do not need to define this macro.
5266 @cindex @code{EDOM}, implicit usage
5269 The value of @code{EDOM} on the target machine, as a C integer constant
5270 expression. If you don't define this macro, GCC does not attempt to
5271 deposit the value of @code{EDOM} into @code{errno} directly. Look in
5272 @file{/usr/include/errno.h} to find the value of @code{EDOM} on your
5275 If you do not define @code{TARGET_EDOM}, then compiled code reports
5276 domain errors by calling the library function and letting it report the
5277 error. If mathematical functions on your system use @code{matherr} when
5278 there is an error, then you should leave @code{TARGET_EDOM} undefined so
5279 that @code{matherr} is used normally.
5282 @cindex @code{errno}, implicit usage
5283 @defmac GEN_ERRNO_RTX
5284 Define this macro as a C expression to create an rtl expression that
5285 refers to the global ``variable'' @code{errno}. (On certain systems,
5286 @code{errno} may not actually be a variable.) If you don't define this
5287 macro, a reasonable default is used.
5290 @cindex C99 math functions, implicit usage
5291 @defmac TARGET_C99_FUNCTIONS
5292 When this macro is nonzero, GCC will implicitly optimize @code{sin} calls into
5293 @code{sinf} and similarly for other functions defined by C99 standard. The
5294 default is zero because a number of existing systems lack support for these
5295 functions in their runtime so this macro needs to be redefined to one on
5296 systems that do support the C99 runtime.
5299 @cindex sincos math function, implicit usage
5300 @defmac TARGET_HAS_SINCOS
5301 When this macro is nonzero, GCC will implicitly optimize calls to @code{sin}
5302 and @code{cos} with the same argument to a call to @code{sincos}. The
5303 default is zero. The target has to provide the following functions:
5305 void sincos(double x, double *sin, double *cos);
5306 void sincosf(float x, float *sin, float *cos);
5307 void sincosl(long double x, long double *sin, long double *cos);
5311 @defmac NEXT_OBJC_RUNTIME
5312 Set this macro to 1 to use the "NeXT" Objective-C message sending conventions
5313 by default. This calling convention involves passing the object, the selector
5314 and the method arguments all at once to the method-lookup library function.
5315 This is the usual setting when targeting Darwin/Mac OS X systems, which have
5316 the NeXT runtime installed.
5318 If the macro is set to 0, the "GNU" Objective-C message sending convention
5319 will be used by default. This convention passes just the object and the
5320 selector to the method-lookup function, which returns a pointer to the method.
5322 In either case, it remains possible to select code-generation for the alternate
5323 scheme, by means of compiler command line switches.
5326 @node Addressing Modes
5327 @section Addressing Modes
5328 @cindex addressing modes
5330 @c prevent bad page break with this line
5331 This is about addressing modes.
5333 @defmac HAVE_PRE_INCREMENT
5334 @defmacx HAVE_PRE_DECREMENT
5335 @defmacx HAVE_POST_INCREMENT
5336 @defmacx HAVE_POST_DECREMENT
5337 A C expression that is nonzero if the machine supports pre-increment,
5338 pre-decrement, post-increment, or post-decrement addressing respectively.
5341 @defmac HAVE_PRE_MODIFY_DISP
5342 @defmacx HAVE_POST_MODIFY_DISP
5343 A C expression that is nonzero if the machine supports pre- or
5344 post-address side-effect generation involving constants other than
5345 the size of the memory operand.
5348 @defmac HAVE_PRE_MODIFY_REG
5349 @defmacx HAVE_POST_MODIFY_REG
5350 A C expression that is nonzero if the machine supports pre- or
5351 post-address side-effect generation involving a register displacement.
5354 @defmac CONSTANT_ADDRESS_P (@var{x})
5355 A C expression that is 1 if the RTX @var{x} is a constant which
5356 is a valid address. On most machines the default definition of
5357 @code{(CONSTANT_P (@var{x}) && GET_CODE (@var{x}) != CONST_DOUBLE)}
5358 is acceptable, but a few machines are more restrictive as to which
5359 constant addresses are supported.
5362 @defmac CONSTANT_P (@var{x})
5363 @code{CONSTANT_P}, which is defined by target-independent code,
5364 accepts integer-values expressions whose values are not explicitly
5365 known, such as @code{symbol_ref}, @code{label_ref}, and @code{high}
5366 expressions and @code{const} arithmetic expressions, in addition to
5367 @code{const_int} and @code{const_double} expressions.
5370 @defmac MAX_REGS_PER_ADDRESS
5371 A number, the maximum number of registers that can appear in a valid
5372 memory address. Note that it is up to you to specify a value equal to
5373 the maximum number that @code{TARGET_LEGITIMATE_ADDRESS_P} would ever
5377 @hook TARGET_LEGITIMATE_ADDRESS_P
5378 A function that returns whether @var{x} (an RTX) is a legitimate memory
5379 address on the target machine for a memory operand of mode @var{mode}.
5381 Legitimate addresses are defined in two variants: a strict variant and a
5382 non-strict one. The @var{strict} parameter chooses which variant is
5383 desired by the caller.
5385 The strict variant is used in the reload pass. It must be defined so
5386 that any pseudo-register that has not been allocated a hard register is
5387 considered a memory reference. This is because in contexts where some
5388 kind of register is required, a pseudo-register with no hard register
5389 must be rejected. For non-hard registers, the strict variant should look
5390 up the @code{reg_renumber} array; it should then proceed using the hard
5391 register number in the array, or treat the pseudo as a memory reference
5392 if the array holds @code{-1}.
5394 The non-strict variant is used in other passes. It must be defined to
5395 accept all pseudo-registers in every context where some kind of
5396 register is required.
5398 Normally, constant addresses which are the sum of a @code{symbol_ref}
5399 and an integer are stored inside a @code{const} RTX to mark them as
5400 constant. Therefore, there is no need to recognize such sums
5401 specifically as legitimate addresses. Normally you would simply
5402 recognize any @code{const} as legitimate.
5404 Usually @code{PRINT_OPERAND_ADDRESS} is not prepared to handle constant
5405 sums that are not marked with @code{const}. It assumes that a naked
5406 @code{plus} indicates indexing. If so, then you @emph{must} reject such
5407 naked constant sums as illegitimate addresses, so that none of them will
5408 be given to @code{PRINT_OPERAND_ADDRESS}.
5410 @cindex @code{TARGET_ENCODE_SECTION_INFO} and address validation
5411 On some machines, whether a symbolic address is legitimate depends on
5412 the section that the address refers to. On these machines, define the
5413 target hook @code{TARGET_ENCODE_SECTION_INFO} to store the information
5414 into the @code{symbol_ref}, and then check for it here. When you see a
5415 @code{const}, you will have to look inside it to find the
5416 @code{symbol_ref} in order to determine the section. @xref{Assembler
5419 @cindex @code{GO_IF_LEGITIMATE_ADDRESS}
5420 Some ports are still using a deprecated legacy substitute for
5421 this hook, the @code{GO_IF_LEGITIMATE_ADDRESS} macro. This macro
5425 #define GO_IF_LEGITIMATE_ADDRESS (@var{mode}, @var{x}, @var{label})
5429 and should @code{goto @var{label}} if the address @var{x} is a valid
5430 address on the target machine for a memory operand of mode @var{mode}.
5432 @findex REG_OK_STRICT
5433 Compiler source files that want to use the strict variant of this
5434 macro define the macro @code{REG_OK_STRICT}. You should use an
5435 @code{#ifdef REG_OK_STRICT} conditional to define the strict variant in
5436 that case and the non-strict variant otherwise.
5438 Using the hook is usually simpler because it limits the number of
5439 files that are recompiled when changes are made.
5442 @defmac TARGET_MEM_CONSTRAINT
5443 A single character to be used instead of the default @code{'m'}
5444 character for general memory addresses. This defines the constraint
5445 letter which matches the memory addresses accepted by
5446 @code{TARGET_LEGITIMATE_ADDRESS_P}. Define this macro if you want to
5447 support new address formats in your back end without changing the
5448 semantics of the @code{'m'} constraint. This is necessary in order to
5449 preserve functionality of inline assembly constructs using the
5450 @code{'m'} constraint.
5453 @defmac FIND_BASE_TERM (@var{x})
5454 A C expression to determine the base term of address @var{x},
5455 or to provide a simplified version of @var{x} from which @file{alias.c}
5456 can easily find the base term. This macro is used in only two places:
5457 @code{find_base_value} and @code{find_base_term} in @file{alias.c}.
5459 It is always safe for this macro to not be defined. It exists so
5460 that alias analysis can understand machine-dependent addresses.
5462 The typical use of this macro is to handle addresses containing
5463 a label_ref or symbol_ref within an UNSPEC@.
5466 @hook TARGET_LEGITIMIZE_ADDRESS
5467 This hook is given an invalid memory address @var{x} for an
5468 operand of mode @var{mode} and should try to return a valid memory
5471 @findex break_out_memory_refs
5472 @var{x} will always be the result of a call to @code{break_out_memory_refs},
5473 and @var{oldx} will be the operand that was given to that function to produce
5476 The code of the hook should not alter the substructure of
5477 @var{x}. If it transforms @var{x} into a more legitimate form, it
5478 should return the new @var{x}.
5480 It is not necessary for this hook to come up with a legitimate address,
5481 with the exception of native TLS addresses (@pxref{Emulated TLS}).
5482 The compiler has standard ways of doing so in all cases. In fact, if
5483 the target supports only emulated TLS, it
5484 is safe to omit this hook or make it return @var{x} if it cannot find
5485 a valid way to legitimize the address. But often a machine-dependent
5486 strategy can generate better code.
5489 @defmac LEGITIMIZE_RELOAD_ADDRESS (@var{x}, @var{mode}, @var{opnum}, @var{type}, @var{ind_levels}, @var{win})
5490 A C compound statement that attempts to replace @var{x}, which is an address
5491 that needs reloading, with a valid memory address for an operand of mode
5492 @var{mode}. @var{win} will be a C statement label elsewhere in the code.
5493 It is not necessary to define this macro, but it might be useful for
5494 performance reasons.
5496 For example, on the i386, it is sometimes possible to use a single
5497 reload register instead of two by reloading a sum of two pseudo
5498 registers into a register. On the other hand, for number of RISC
5499 processors offsets are limited so that often an intermediate address
5500 needs to be generated in order to address a stack slot. By defining
5501 @code{LEGITIMIZE_RELOAD_ADDRESS} appropriately, the intermediate addresses
5502 generated for adjacent some stack slots can be made identical, and thus
5505 @emph{Note}: This macro should be used with caution. It is necessary
5506 to know something of how reload works in order to effectively use this,
5507 and it is quite easy to produce macros that build in too much knowledge
5508 of reload internals.
5510 @emph{Note}: This macro must be able to reload an address created by a
5511 previous invocation of this macro. If it fails to handle such addresses
5512 then the compiler may generate incorrect code or abort.
5515 The macro definition should use @code{push_reload} to indicate parts that
5516 need reloading; @var{opnum}, @var{type} and @var{ind_levels} are usually
5517 suitable to be passed unaltered to @code{push_reload}.
5519 The code generated by this macro must not alter the substructure of
5520 @var{x}. If it transforms @var{x} into a more legitimate form, it
5521 should assign @var{x} (which will always be a C variable) a new value.
5522 This also applies to parts that you change indirectly by calling
5525 @findex strict_memory_address_p
5526 The macro definition may use @code{strict_memory_address_p} to test if
5527 the address has become legitimate.
5530 If you want to change only a part of @var{x}, one standard way of doing
5531 this is to use @code{copy_rtx}. Note, however, that it unshares only a
5532 single level of rtl. Thus, if the part to be changed is not at the
5533 top level, you'll need to replace first the top level.
5534 It is not necessary for this macro to come up with a legitimate
5535 address; but often a machine-dependent strategy can generate better code.
5538 @hook TARGET_MODE_DEPENDENT_ADDRESS_P
5539 This hook returns @code{true} if memory address @var{addr} in address
5540 space @var{addrspace} can have
5541 different meanings depending on the machine mode of the memory
5542 reference it is used for or if the address is valid for some modes
5545 Autoincrement and autodecrement addresses typically have mode-dependent
5546 effects because the amount of the increment or decrement is the size
5547 of the operand being addressed. Some machines have other mode-dependent
5548 addresses. Many RISC machines have no mode-dependent addresses.
5550 You may assume that @var{addr} is a valid address for the machine.
5552 The default version of this hook returns @code{false}.
5555 @hook TARGET_LEGITIMATE_CONSTANT_P
5556 This hook returns true if @var{x} is a legitimate constant for a
5557 @var{mode}-mode immediate operand on the target machine. You can assume that
5558 @var{x} satisfies @code{CONSTANT_P}, so you need not check this.
5560 The default definition returns true.
5563 @hook TARGET_DELEGITIMIZE_ADDRESS
5564 This hook is used to undo the possibly obfuscating effects of the
5565 @code{LEGITIMIZE_ADDRESS} and @code{LEGITIMIZE_RELOAD_ADDRESS} target
5566 macros. Some backend implementations of these macros wrap symbol
5567 references inside an @code{UNSPEC} rtx to represent PIC or similar
5568 addressing modes. This target hook allows GCC's optimizers to understand
5569 the semantics of these opaque @code{UNSPEC}s by converting them back
5570 into their original form.
5573 @hook TARGET_CONST_NOT_OK_FOR_DEBUG_P
5574 This hook should return true if @var{x} should not be emitted into
5578 @hook TARGET_CANNOT_FORCE_CONST_MEM
5579 This hook should return true if @var{x} is of a form that cannot (or
5580 should not) be spilled to the constant pool. @var{mode} is the mode
5583 The default version of this hook returns false.
5585 The primary reason to define this hook is to prevent reload from
5586 deciding that a non-legitimate constant would be better reloaded
5587 from the constant pool instead of spilling and reloading a register
5588 holding the constant. This restriction is often true of addresses
5589 of TLS symbols for various targets.
5592 @hook TARGET_USE_BLOCKS_FOR_CONSTANT_P
5593 This hook should return true if pool entries for constant @var{x} can
5594 be placed in an @code{object_block} structure. @var{mode} is the mode
5597 The default version returns false for all constants.
5600 @hook TARGET_BUILTIN_RECIPROCAL
5601 This hook should return the DECL of a function that implements reciprocal of
5602 the builtin function with builtin function code @var{fn}, or
5603 @code{NULL_TREE} if such a function is not available. @var{md_fn} is true
5604 when @var{fn} is a code of a machine-dependent builtin function. When
5605 @var{sqrt} is true, additional optimizations that apply only to the reciprocal
5606 of a square root function are performed, and only reciprocals of @code{sqrt}
5610 @hook TARGET_VECTORIZE_BUILTIN_MASK_FOR_LOAD
5611 This hook should return the DECL of a function @var{f} that given an
5612 address @var{addr} as an argument returns a mask @var{m} that can be
5613 used to extract from two vectors the relevant data that resides in
5614 @var{addr} in case @var{addr} is not properly aligned.
5616 The autovectorizer, when vectorizing a load operation from an address
5617 @var{addr} that may be unaligned, will generate two vector loads from
5618 the two aligned addresses around @var{addr}. It then generates a
5619 @code{REALIGN_LOAD} operation to extract the relevant data from the
5620 two loaded vectors. The first two arguments to @code{REALIGN_LOAD},
5621 @var{v1} and @var{v2}, are the two vectors, each of size @var{VS}, and
5622 the third argument, @var{OFF}, defines how the data will be extracted
5623 from these two vectors: if @var{OFF} is 0, then the returned vector is
5624 @var{v2}; otherwise, the returned vector is composed from the last
5625 @var{VS}-@var{OFF} elements of @var{v1} concatenated to the first
5626 @var{OFF} elements of @var{v2}.
5628 If this hook is defined, the autovectorizer will generate a call
5629 to @var{f} (using the DECL tree that this hook returns) and will
5630 use the return value of @var{f} as the argument @var{OFF} to
5631 @code{REALIGN_LOAD}. Therefore, the mask @var{m} returned by @var{f}
5632 should comply with the semantics expected by @code{REALIGN_LOAD}
5634 If this hook is not defined, then @var{addr} will be used as
5635 the argument @var{OFF} to @code{REALIGN_LOAD}, in which case the low
5636 log2(@var{VS}) @minus{} 1 bits of @var{addr} will be considered.
5639 @hook TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST
5640 Returns cost of different scalar or vector statements for vectorization cost model.
5641 For vector memory operations the cost may depend on type (@var{vectype}) and
5642 misalignment value (@var{misalign}).
5645 @hook TARGET_VECTORIZE_VECTOR_ALIGNMENT_REACHABLE
5646 Return true if vector alignment is reachable (by peeling N iterations) for the given type.
5649 @hook TARGET_VECTORIZE_VEC_PERM_CONST_OK
5650 Return true if a vector created for @code{vec_perm_const} is valid.
5653 @hook TARGET_VECTORIZE_BUILTIN_CONVERSION
5654 This hook should return the DECL of a function that implements conversion of the
5655 input vector of type @var{src_type} to type @var{dest_type}.
5656 The value of @var{code} is one of the enumerators in @code{enum tree_code} and
5657 specifies how the conversion is to be applied
5658 (truncation, rounding, etc.).
5660 If this hook is defined, the autovectorizer will use the
5661 @code{TARGET_VECTORIZE_BUILTIN_CONVERSION} target hook when vectorizing
5662 conversion. Otherwise, it will return @code{NULL_TREE}.
5665 @hook TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION
5666 This hook should return the decl of a function that implements the
5667 vectorized variant of the builtin function with builtin function code
5668 @var{code} or @code{NULL_TREE} if such a function is not available.
5669 The value of @var{fndecl} is the builtin function declaration. The
5670 return type of the vectorized function shall be of vector type
5671 @var{vec_type_out} and the argument types should be @var{vec_type_in}.
5674 @hook TARGET_VECTORIZE_SUPPORT_VECTOR_MISALIGNMENT
5675 This hook should return true if the target supports misaligned vector
5676 store/load of a specific factor denoted in the @var{misalignment}
5677 parameter. The vector store/load should be of machine mode @var{mode} and
5678 the elements in the vectors should be of type @var{type}. @var{is_packed}
5679 parameter is true if the memory access is defined in a packed struct.
5682 @hook TARGET_VECTORIZE_PREFERRED_SIMD_MODE
5683 This hook should return the preferred mode for vectorizing scalar
5684 mode @var{mode}. The default is
5685 equal to @code{word_mode}, because the vectorizer can do some
5686 transformations even in absence of specialized @acronym{SIMD} hardware.
5689 @hook TARGET_VECTORIZE_AUTOVECTORIZE_VECTOR_SIZES
5690 This hook should return a mask of sizes that should be iterated over
5691 after trying to autovectorize using the vector size derived from the
5692 mode returned by @code{TARGET_VECTORIZE_PREFERRED_SIMD_MODE}.
5693 The default is zero which means to not iterate over other vector sizes.
5696 @hook TARGET_VECTORIZE_INIT_COST
5698 @hook TARGET_VECTORIZE_ADD_STMT_COST
5700 @hook TARGET_VECTORIZE_FINISH_COST
5702 @hook TARGET_VECTORIZE_DESTROY_COST_DATA
5704 @hook TARGET_VECTORIZE_BUILTIN_TM_LOAD
5706 @hook TARGET_VECTORIZE_BUILTIN_TM_STORE
5708 @hook TARGET_VECTORIZE_BUILTIN_GATHER
5709 Target builtin that implements vector gather operation. @var{mem_vectype}
5710 is the vector type of the load and @var{index_type} is scalar type of
5711 the index, scaled by @var{scale}.
5712 The default is @code{NULL_TREE} which means to not vectorize gather
5716 @node Anchored Addresses
5717 @section Anchored Addresses
5718 @cindex anchored addresses
5719 @cindex @option{-fsection-anchors}
5721 GCC usually addresses every static object as a separate entity.
5722 For example, if we have:
5726 int foo (void) @{ return a + b + c; @}
5729 the code for @code{foo} will usually calculate three separate symbolic
5730 addresses: those of @code{a}, @code{b} and @code{c}. On some targets,
5731 it would be better to calculate just one symbolic address and access
5732 the three variables relative to it. The equivalent pseudocode would
5738 register int *xr = &x;
5739 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
5743 (which isn't valid C). We refer to shared addresses like @code{x} as
5744 ``section anchors''. Their use is controlled by @option{-fsection-anchors}.
5746 The hooks below describe the target properties that GCC needs to know
5747 in order to make effective use of section anchors. It won't use
5748 section anchors at all unless either @code{TARGET_MIN_ANCHOR_OFFSET}
5749 or @code{TARGET_MAX_ANCHOR_OFFSET} is set to a nonzero value.
5751 @hook TARGET_MIN_ANCHOR_OFFSET
5752 The minimum offset that should be applied to a section anchor.
5753 On most targets, it should be the smallest offset that can be
5754 applied to a base register while still giving a legitimate address
5755 for every mode. The default value is 0.
5758 @hook TARGET_MAX_ANCHOR_OFFSET
5759 Like @code{TARGET_MIN_ANCHOR_OFFSET}, but the maximum (inclusive)
5760 offset that should be applied to section anchors. The default
5764 @hook TARGET_ASM_OUTPUT_ANCHOR
5765 Write the assembly code to define section anchor @var{x}, which is a
5766 @code{SYMBOL_REF} for which @samp{SYMBOL_REF_ANCHOR_P (@var{x})} is true.
5767 The hook is called with the assembly output position set to the beginning
5768 of @code{SYMBOL_REF_BLOCK (@var{x})}.
5770 If @code{ASM_OUTPUT_DEF} is available, the hook's default definition uses
5771 it to define the symbol as @samp{. + SYMBOL_REF_BLOCK_OFFSET (@var{x})}.
5772 If @code{ASM_OUTPUT_DEF} is not available, the hook's default definition
5773 is @code{NULL}, which disables the use of section anchors altogether.
5776 @hook TARGET_USE_ANCHORS_FOR_SYMBOL_P
5777 Return true if GCC should attempt to use anchors to access @code{SYMBOL_REF}
5778 @var{x}. You can assume @samp{SYMBOL_REF_HAS_BLOCK_INFO_P (@var{x})} and
5779 @samp{!SYMBOL_REF_ANCHOR_P (@var{x})}.
5781 The default version is correct for most targets, but you might need to
5782 intercept this hook to handle things like target-specific attributes
5783 or target-specific sections.
5786 @node Condition Code
5787 @section Condition Code Status
5788 @cindex condition code status
5790 The macros in this section can be split in two families, according to the
5791 two ways of representing condition codes in GCC.
5793 The first representation is the so called @code{(cc0)} representation
5794 (@pxref{Jump Patterns}), where all instructions can have an implicit
5795 clobber of the condition codes. The second is the condition code
5796 register representation, which provides better schedulability for
5797 architectures that do have a condition code register, but on which
5798 most instructions do not affect it. The latter category includes
5801 The implicit clobbering poses a strong restriction on the placement of
5802 the definition and use of the condition code, which need to be in adjacent
5803 insns for machines using @code{(cc0)}. This can prevent important
5804 optimizations on some machines. For example, on the IBM RS/6000, there
5805 is a delay for taken branches unless the condition code register is set
5806 three instructions earlier than the conditional branch. The instruction
5807 scheduler cannot perform this optimization if it is not permitted to
5808 separate the definition and use of the condition code register.
5810 For this reason, it is possible and suggested to use a register to
5811 represent the condition code for new ports. If there is a specific
5812 condition code register in the machine, use a hard register. If the
5813 condition code or comparison result can be placed in any general register,
5814 or if there are multiple condition registers, use a pseudo register.
5815 Registers used to store the condition code value will usually have a mode
5816 that is in class @code{MODE_CC}.
5818 Alternatively, you can use @code{BImode} if the comparison operator is
5819 specified already in the compare instruction. In this case, you are not
5820 interested in most macros in this section.
5823 * CC0 Condition Codes:: Old style representation of condition codes.
5824 * MODE_CC Condition Codes:: Modern representation of condition codes.
5825 * Cond Exec Macros:: Macros to control conditional execution.
5828 @node CC0 Condition Codes
5829 @subsection Representation of condition codes using @code{(cc0)}
5833 The file @file{conditions.h} defines a variable @code{cc_status} to
5834 describe how the condition code was computed (in case the interpretation of
5835 the condition code depends on the instruction that it was set by). This
5836 variable contains the RTL expressions on which the condition code is
5837 currently based, and several standard flags.
5839 Sometimes additional machine-specific flags must be defined in the machine
5840 description header file. It can also add additional machine-specific
5841 information by defining @code{CC_STATUS_MDEP}.
5843 @defmac CC_STATUS_MDEP
5844 C code for a data type which is used for declaring the @code{mdep}
5845 component of @code{cc_status}. It defaults to @code{int}.
5847 This macro is not used on machines that do not use @code{cc0}.
5850 @defmac CC_STATUS_MDEP_INIT
5851 A C expression to initialize the @code{mdep} field to ``empty''.
5852 The default definition does nothing, since most machines don't use
5853 the field anyway. If you want to use the field, you should probably
5854 define this macro to initialize it.
5856 This macro is not used on machines that do not use @code{cc0}.
5859 @defmac NOTICE_UPDATE_CC (@var{exp}, @var{insn})
5860 A C compound statement to set the components of @code{cc_status}
5861 appropriately for an insn @var{insn} whose body is @var{exp}. It is
5862 this macro's responsibility to recognize insns that set the condition
5863 code as a byproduct of other activity as well as those that explicitly
5866 This macro is not used on machines that do not use @code{cc0}.
5868 If there are insns that do not set the condition code but do alter
5869 other machine registers, this macro must check to see whether they
5870 invalidate the expressions that the condition code is recorded as
5871 reflecting. For example, on the 68000, insns that store in address
5872 registers do not set the condition code, which means that usually
5873 @code{NOTICE_UPDATE_CC} can leave @code{cc_status} unaltered for such
5874 insns. But suppose that the previous insn set the condition code
5875 based on location @samp{a4@@(102)} and the current insn stores a new
5876 value in @samp{a4}. Although the condition code is not changed by
5877 this, it will no longer be true that it reflects the contents of
5878 @samp{a4@@(102)}. Therefore, @code{NOTICE_UPDATE_CC} must alter
5879 @code{cc_status} in this case to say that nothing is known about the
5880 condition code value.
5882 The definition of @code{NOTICE_UPDATE_CC} must be prepared to deal
5883 with the results of peephole optimization: insns whose patterns are
5884 @code{parallel} RTXs containing various @code{reg}, @code{mem} or
5885 constants which are just the operands. The RTL structure of these
5886 insns is not sufficient to indicate what the insns actually do. What
5887 @code{NOTICE_UPDATE_CC} should do when it sees one is just to run
5888 @code{CC_STATUS_INIT}.
5890 A possible definition of @code{NOTICE_UPDATE_CC} is to call a function
5891 that looks at an attribute (@pxref{Insn Attributes}) named, for example,
5892 @samp{cc}. This avoids having detailed information about patterns in
5893 two places, the @file{md} file and in @code{NOTICE_UPDATE_CC}.
5896 @node MODE_CC Condition Codes
5897 @subsection Representation of condition codes using registers
5901 @defmac SELECT_CC_MODE (@var{op}, @var{x}, @var{y})
5902 On many machines, the condition code may be produced by other instructions
5903 than compares, for example the branch can use directly the condition
5904 code set by a subtract instruction. However, on some machines
5905 when the condition code is set this way some bits (such as the overflow
5906 bit) are not set in the same way as a test instruction, so that a different
5907 branch instruction must be used for some conditional branches. When
5908 this happens, use the machine mode of the condition code register to
5909 record different formats of the condition code register. Modes can
5910 also be used to record which compare instruction (e.g. a signed or an
5911 unsigned comparison) produced the condition codes.
5913 If other modes than @code{CCmode} are required, add them to
5914 @file{@var{machine}-modes.def} and define @code{SELECT_CC_MODE} to choose
5915 a mode given an operand of a compare. This is needed because the modes
5916 have to be chosen not only during RTL generation but also, for example,
5917 by instruction combination. The result of @code{SELECT_CC_MODE} should
5918 be consistent with the mode used in the patterns; for example to support
5919 the case of the add on the SPARC discussed above, we have the pattern
5923 [(set (reg:CC_NOOV 0)
5925 (plus:SI (match_operand:SI 0 "register_operand" "%r")
5926 (match_operand:SI 1 "arith_operand" "rI"))
5933 together with a @code{SELECT_CC_MODE} that returns @code{CC_NOOVmode}
5934 for comparisons whose argument is a @code{plus}:
5937 #define SELECT_CC_MODE(OP,X,Y) \
5938 (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \
5939 ? ((OP == EQ || OP == NE) ? CCFPmode : CCFPEmode) \
5940 : ((GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \
5941 || GET_CODE (X) == NEG) \
5942 ? CC_NOOVmode : CCmode))
5945 Another reason to use modes is to retain information on which operands
5946 were used by the comparison; see @code{REVERSIBLE_CC_MODE} later in
5949 You should define this macro if and only if you define extra CC modes
5950 in @file{@var{machine}-modes.def}.
5953 @defmac CANONICALIZE_COMPARISON (@var{code}, @var{op0}, @var{op1})
5954 On some machines not all possible comparisons are defined, but you can
5955 convert an invalid comparison into a valid one. For example, the Alpha
5956 does not have a @code{GT} comparison, but you can use an @code{LT}
5957 comparison instead and swap the order of the operands.
5959 On such machines, define this macro to be a C statement to do any
5960 required conversions. @var{code} is the initial comparison code
5961 and @var{op0} and @var{op1} are the left and right operands of the
5962 comparison, respectively. You should modify @var{code}, @var{op0}, and
5963 @var{op1} as required.
5965 GCC will not assume that the comparison resulting from this macro is
5966 valid but will see if the resulting insn matches a pattern in the
5969 You need not define this macro if it would never change the comparison
5973 @defmac REVERSIBLE_CC_MODE (@var{mode})
5974 A C expression whose value is one if it is always safe to reverse a
5975 comparison whose mode is @var{mode}. If @code{SELECT_CC_MODE}
5976 can ever return @var{mode} for a floating-point inequality comparison,
5977 then @code{REVERSIBLE_CC_MODE (@var{mode})} must be zero.
5979 You need not define this macro if it would always returns zero or if the
5980 floating-point format is anything other than @code{IEEE_FLOAT_FORMAT}.
5981 For example, here is the definition used on the SPARC, where floating-point
5982 inequality comparisons are always given @code{CCFPEmode}:
5985 #define REVERSIBLE_CC_MODE(MODE) ((MODE) != CCFPEmode)
5989 @defmac REVERSE_CONDITION (@var{code}, @var{mode})
5990 A C expression whose value is reversed condition code of the @var{code} for
5991 comparison done in CC_MODE @var{mode}. The macro is used only in case
5992 @code{REVERSIBLE_CC_MODE (@var{mode})} is nonzero. Define this macro in case
5993 machine has some non-standard way how to reverse certain conditionals. For
5994 instance in case all floating point conditions are non-trapping, compiler may
5995 freely convert unordered compares to ordered one. Then definition may look
5999 #define REVERSE_CONDITION(CODE, MODE) \
6000 ((MODE) != CCFPmode ? reverse_condition (CODE) \
6001 : reverse_condition_maybe_unordered (CODE))
6005 @hook TARGET_FIXED_CONDITION_CODE_REGS
6006 On targets which do not use @code{(cc0)}, and which use a hard
6007 register rather than a pseudo-register to hold condition codes, the
6008 regular CSE passes are often not able to identify cases in which the
6009 hard register is set to a common value. Use this hook to enable a
6010 small pass which optimizes such cases. This hook should return true
6011 to enable this pass, and it should set the integers to which its
6012 arguments point to the hard register numbers used for condition codes.
6013 When there is only one such register, as is true on most systems, the
6014 integer pointed to by @var{p2} should be set to
6015 @code{INVALID_REGNUM}.
6017 The default version of this hook returns false.
6020 @hook TARGET_CC_MODES_COMPATIBLE
6021 On targets which use multiple condition code modes in class
6022 @code{MODE_CC}, it is sometimes the case that a comparison can be
6023 validly done in more than one mode. On such a system, define this
6024 target hook to take two mode arguments and to return a mode in which
6025 both comparisons may be validly done. If there is no such mode,
6026 return @code{VOIDmode}.
6028 The default version of this hook checks whether the modes are the
6029 same. If they are, it returns that mode. If they are different, it
6030 returns @code{VOIDmode}.
6033 @node Cond Exec Macros
6034 @subsection Macros to control conditional execution
6035 @findex conditional execution
6038 There is one macro that may need to be defined for targets
6039 supporting conditional execution, independent of how they
6040 represent conditional branches.
6043 @section Describing Relative Costs of Operations
6044 @cindex costs of instructions
6045 @cindex relative costs
6046 @cindex speed of instructions
6048 These macros let you describe the relative speed of various operations
6049 on the target machine.
6051 @defmac REGISTER_MOVE_COST (@var{mode}, @var{from}, @var{to})
6052 A C expression for the cost of moving data of mode @var{mode} from a
6053 register in class @var{from} to one in class @var{to}. The classes are
6054 expressed using the enumeration values such as @code{GENERAL_REGS}. A
6055 value of 2 is the default; other values are interpreted relative to
6058 It is not required that the cost always equal 2 when @var{from} is the
6059 same as @var{to}; on some machines it is expensive to move between
6060 registers if they are not general registers.
6062 If reload sees an insn consisting of a single @code{set} between two
6063 hard registers, and if @code{REGISTER_MOVE_COST} applied to their
6064 classes returns a value of 2, reload does not check to ensure that the
6065 constraints of the insn are met. Setting a cost of other than 2 will
6066 allow reload to verify that the constraints are met. You should do this
6067 if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
6069 These macros are obsolete, new ports should use the target hook
6070 @code{TARGET_REGISTER_MOVE_COST} instead.
6073 @hook TARGET_REGISTER_MOVE_COST
6074 This target hook should return the cost of moving data of mode @var{mode}
6075 from a register in class @var{from} to one in class @var{to}. The classes
6076 are expressed using the enumeration values such as @code{GENERAL_REGS}.
6077 A value of 2 is the default; other values are interpreted relative to
6080 It is not required that the cost always equal 2 when @var{from} is the
6081 same as @var{to}; on some machines it is expensive to move between
6082 registers if they are not general registers.
6084 If reload sees an insn consisting of a single @code{set} between two
6085 hard registers, and if @code{TARGET_REGISTER_MOVE_COST} applied to their
6086 classes returns a value of 2, reload does not check to ensure that the
6087 constraints of the insn are met. Setting a cost of other than 2 will
6088 allow reload to verify that the constraints are met. You should do this
6089 if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
6091 The default version of this function returns 2.
6094 @defmac MEMORY_MOVE_COST (@var{mode}, @var{class}, @var{in})
6095 A C expression for the cost of moving data of mode @var{mode} between a
6096 register of class @var{class} and memory; @var{in} is zero if the value
6097 is to be written to memory, nonzero if it is to be read in. This cost
6098 is relative to those in @code{REGISTER_MOVE_COST}. If moving between
6099 registers and memory is more expensive than between two registers, you
6100 should define this macro to express the relative cost.
6102 If you do not define this macro, GCC uses a default cost of 4 plus
6103 the cost of copying via a secondary reload register, if one is
6104 needed. If your machine requires a secondary reload register to copy
6105 between memory and a register of @var{class} but the reload mechanism is
6106 more complex than copying via an intermediate, define this macro to
6107 reflect the actual cost of the move.
6109 GCC defines the function @code{memory_move_secondary_cost} if
6110 secondary reloads are needed. It computes the costs due to copying via
6111 a secondary register. If your machine copies from memory using a
6112 secondary register in the conventional way but the default base value of
6113 4 is not correct for your machine, define this macro to add some other
6114 value to the result of that function. The arguments to that function
6115 are the same as to this macro.
6117 These macros are obsolete, new ports should use the target hook
6118 @code{TARGET_MEMORY_MOVE_COST} instead.
6121 @hook TARGET_MEMORY_MOVE_COST
6122 This target hook should return the cost of moving data of mode @var{mode}
6123 between a register of class @var{rclass} and memory; @var{in} is @code{false}
6124 if the value is to be written to memory, @code{true} if it is to be read in.
6125 This cost is relative to those in @code{TARGET_REGISTER_MOVE_COST}.
6126 If moving between registers and memory is more expensive than between two
6127 registers, you should add this target hook to express the relative cost.
6129 If you do not add this target hook, GCC uses a default cost of 4 plus
6130 the cost of copying via a secondary reload register, if one is
6131 needed. If your machine requires a secondary reload register to copy
6132 between memory and a register of @var{rclass} but the reload mechanism is
6133 more complex than copying via an intermediate, use this target hook to
6134 reflect the actual cost of the move.
6136 GCC defines the function @code{memory_move_secondary_cost} if
6137 secondary reloads are needed. It computes the costs due to copying via
6138 a secondary register. If your machine copies from memory using a
6139 secondary register in the conventional way but the default base value of
6140 4 is not correct for your machine, use this target hook to add some other
6141 value to the result of that function. The arguments to that function
6142 are the same as to this target hook.
6145 @defmac BRANCH_COST (@var{speed_p}, @var{predictable_p})
6146 A C expression for the cost of a branch instruction. A value of 1 is
6147 the default; other values are interpreted relative to that. Parameter
6148 @var{speed_p} is true when the branch in question should be optimized
6149 for speed. When it is false, @code{BRANCH_COST} should return a value
6150 optimal for code size rather than performance. @var{predictable_p} is
6151 true for well-predicted branches. On many architectures the
6152 @code{BRANCH_COST} can be reduced then.
6155 Here are additional macros which do not specify precise relative costs,
6156 but only that certain actions are more expensive than GCC would
6159 @defmac SLOW_BYTE_ACCESS
6160 Define this macro as a C expression which is nonzero if accessing less
6161 than a word of memory (i.e.@: a @code{char} or a @code{short}) is no
6162 faster than accessing a word of memory, i.e., if such access
6163 require more than one instruction or if there is no difference in cost
6164 between byte and (aligned) word loads.
6166 When this macro is not defined, the compiler will access a field by
6167 finding the smallest containing object; when it is defined, a fullword
6168 load will be used if alignment permits. Unless bytes accesses are
6169 faster than word accesses, using word accesses is preferable since it
6170 may eliminate subsequent memory access if subsequent accesses occur to
6171 other fields in the same word of the structure, but to different bytes.
6174 @defmac SLOW_UNALIGNED_ACCESS (@var{mode}, @var{alignment})
6175 Define this macro to be the value 1 if memory accesses described by the
6176 @var{mode} and @var{alignment} parameters have a cost many times greater
6177 than aligned accesses, for example if they are emulated in a trap
6180 When this macro is nonzero, the compiler will act as if
6181 @code{STRICT_ALIGNMENT} were nonzero when generating code for block
6182 moves. This can cause significantly more instructions to be produced.
6183 Therefore, do not set this macro nonzero if unaligned accesses only add a
6184 cycle or two to the time for a memory access.
6186 If the value of this macro is always zero, it need not be defined. If
6187 this macro is defined, it should produce a nonzero value when
6188 @code{STRICT_ALIGNMENT} is nonzero.
6191 @defmac MOVE_RATIO (@var{speed})
6192 The threshold of number of scalar memory-to-memory move insns, @emph{below}
6193 which a sequence of insns should be generated instead of a
6194 string move insn or a library call. Increasing the value will always
6195 make code faster, but eventually incurs high cost in increased code size.
6197 Note that on machines where the corresponding move insn is a
6198 @code{define_expand} that emits a sequence of insns, this macro counts
6199 the number of such sequences.
6201 The parameter @var{speed} is true if the code is currently being
6202 optimized for speed rather than size.
6204 If you don't define this, a reasonable default is used.
6207 @defmac MOVE_BY_PIECES_P (@var{size}, @var{alignment})
6208 A C expression used to determine whether @code{move_by_pieces} will be used to
6209 copy a chunk of memory, or whether some other block move mechanism
6210 will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6211 than @code{MOVE_RATIO}.
6214 @defmac MOVE_MAX_PIECES
6215 A C expression used by @code{move_by_pieces} to determine the largest unit
6216 a load or store used to copy memory is. Defaults to @code{MOVE_MAX}.
6219 @defmac CLEAR_RATIO (@var{speed})
6220 The threshold of number of scalar move insns, @emph{below} which a sequence
6221 of insns should be generated to clear memory instead of a string clear insn
6222 or a library call. Increasing the value will always make code faster, but
6223 eventually incurs high cost in increased code size.
6225 The parameter @var{speed} is true if the code is currently being
6226 optimized for speed rather than size.
6228 If you don't define this, a reasonable default is used.
6231 @defmac CLEAR_BY_PIECES_P (@var{size}, @var{alignment})
6232 A C expression used to determine whether @code{clear_by_pieces} will be used
6233 to clear a chunk of memory, or whether some other block clear mechanism
6234 will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6235 than @code{CLEAR_RATIO}.
6238 @defmac SET_RATIO (@var{speed})
6239 The threshold of number of scalar move insns, @emph{below} which a sequence
6240 of insns should be generated to set memory to a constant value, instead of
6241 a block set insn or a library call.
6242 Increasing the value will always make code faster, but
6243 eventually incurs high cost in increased code size.
6245 The parameter @var{speed} is true if the code is currently being
6246 optimized for speed rather than size.
6248 If you don't define this, it defaults to the value of @code{MOVE_RATIO}.
6251 @defmac SET_BY_PIECES_P (@var{size}, @var{alignment})
6252 A C expression used to determine whether @code{store_by_pieces} will be
6253 used to set a chunk of memory to a constant value, or whether some
6254 other mechanism will be used. Used by @code{__builtin_memset} when
6255 storing values other than constant zero.
6256 Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6257 than @code{SET_RATIO}.
6260 @defmac STORE_BY_PIECES_P (@var{size}, @var{alignment})
6261 A C expression used to determine whether @code{store_by_pieces} will be
6262 used to set a chunk of memory to a constant string value, or whether some
6263 other mechanism will be used. Used by @code{__builtin_strcpy} when
6264 called with a constant source string.
6265 Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6266 than @code{MOVE_RATIO}.
6269 @defmac USE_LOAD_POST_INCREMENT (@var{mode})
6270 A C expression used to determine whether a load postincrement is a good
6271 thing to use for a given mode. Defaults to the value of
6272 @code{HAVE_POST_INCREMENT}.
6275 @defmac USE_LOAD_POST_DECREMENT (@var{mode})
6276 A C expression used to determine whether a load postdecrement is a good
6277 thing to use for a given mode. Defaults to the value of
6278 @code{HAVE_POST_DECREMENT}.
6281 @defmac USE_LOAD_PRE_INCREMENT (@var{mode})
6282 A C expression used to determine whether a load preincrement is a good
6283 thing to use for a given mode. Defaults to the value of
6284 @code{HAVE_PRE_INCREMENT}.
6287 @defmac USE_LOAD_PRE_DECREMENT (@var{mode})
6288 A C expression used to determine whether a load predecrement is a good
6289 thing to use for a given mode. Defaults to the value of
6290 @code{HAVE_PRE_DECREMENT}.
6293 @defmac USE_STORE_POST_INCREMENT (@var{mode})
6294 A C expression used to determine whether a store postincrement is a good
6295 thing to use for a given mode. Defaults to the value of
6296 @code{HAVE_POST_INCREMENT}.
6299 @defmac USE_STORE_POST_DECREMENT (@var{mode})
6300 A C expression used to determine whether a store postdecrement is a good
6301 thing to use for a given mode. Defaults to the value of
6302 @code{HAVE_POST_DECREMENT}.
6305 @defmac USE_STORE_PRE_INCREMENT (@var{mode})
6306 This macro is used to determine whether a store preincrement is a good
6307 thing to use for a given mode. Defaults to the value of
6308 @code{HAVE_PRE_INCREMENT}.
6311 @defmac USE_STORE_PRE_DECREMENT (@var{mode})
6312 This macro is used to determine whether a store predecrement is a good
6313 thing to use for a given mode. Defaults to the value of
6314 @code{HAVE_PRE_DECREMENT}.
6317 @defmac NO_FUNCTION_CSE
6318 Define this macro if it is as good or better to call a constant
6319 function address than to call an address kept in a register.
6322 @defmac LOGICAL_OP_NON_SHORT_CIRCUIT
6323 Define this macro if a non-short-circuit operation produced by
6324 @samp{fold_range_test ()} is optimal. This macro defaults to true if
6325 @code{BRANCH_COST} is greater than or equal to the value 2.
6328 @hook TARGET_RTX_COSTS
6329 This target hook describes the relative costs of RTL expressions.
6331 The cost may depend on the precise form of the expression, which is
6332 available for examination in @var{x}, and the fact that @var{x} appears
6333 as operand @var{opno} of an expression with rtx code @var{outer_code}.
6334 That is, the hook can assume that there is some rtx @var{y} such
6335 that @samp{GET_CODE (@var{y}) == @var{outer_code}} and such that
6336 either (a) @samp{XEXP (@var{y}, @var{opno}) == @var{x}} or
6337 (b) @samp{XVEC (@var{y}, @var{opno})} contains @var{x}.
6339 @var{code} is @var{x}'s expression code---redundant, since it can be
6340 obtained with @code{GET_CODE (@var{x})}.
6342 In implementing this hook, you can use the construct
6343 @code{COSTS_N_INSNS (@var{n})} to specify a cost equal to @var{n} fast
6346 On entry to the hook, @code{*@var{total}} contains a default estimate
6347 for the cost of the expression. The hook should modify this value as
6348 necessary. Traditionally, the default costs are @code{COSTS_N_INSNS (5)}
6349 for multiplications, @code{COSTS_N_INSNS (7)} for division and modulus
6350 operations, and @code{COSTS_N_INSNS (1)} for all other operations.
6352 When optimizing for code size, i.e.@: when @code{speed} is
6353 false, this target hook should be used to estimate the relative
6354 size cost of an expression, again relative to @code{COSTS_N_INSNS}.
6356 The hook returns true when all subexpressions of @var{x} have been
6357 processed, and false when @code{rtx_cost} should recurse.
6360 @hook TARGET_ADDRESS_COST
6361 This hook computes the cost of an addressing mode that contains
6362 @var{address}. If not defined, the cost is computed from
6363 the @var{address} expression and the @code{TARGET_RTX_COST} hook.
6365 For most CISC machines, the default cost is a good approximation of the
6366 true cost of the addressing mode. However, on RISC machines, all
6367 instructions normally have the same length and execution time. Hence
6368 all addresses will have equal costs.
6370 In cases where more than one form of an address is known, the form with
6371 the lowest cost will be used. If multiple forms have the same, lowest,
6372 cost, the one that is the most complex will be used.
6374 For example, suppose an address that is equal to the sum of a register
6375 and a constant is used twice in the same basic block. When this macro
6376 is not defined, the address will be computed in a register and memory
6377 references will be indirect through that register. On machines where
6378 the cost of the addressing mode containing the sum is no higher than
6379 that of a simple indirect reference, this will produce an additional
6380 instruction and possibly require an additional register. Proper
6381 specification of this macro eliminates this overhead for such machines.
6383 This hook is never called with an invalid address.
6385 On machines where an address involving more than one register is as
6386 cheap as an address computation involving only one register, defining
6387 @code{TARGET_ADDRESS_COST} to reflect this can cause two registers to
6388 be live over a region of code where only one would have been if
6389 @code{TARGET_ADDRESS_COST} were not defined in that manner. This effect
6390 should be considered in the definition of this macro. Equivalent costs
6391 should probably only be given to addresses with different numbers of
6392 registers on machines with lots of registers.
6396 @section Adjusting the Instruction Scheduler
6398 The instruction scheduler may need a fair amount of machine-specific
6399 adjustment in order to produce good code. GCC provides several target
6400 hooks for this purpose. It is usually enough to define just a few of
6401 them: try the first ones in this list first.
6403 @hook TARGET_SCHED_ISSUE_RATE
6404 This hook returns the maximum number of instructions that can ever
6405 issue at the same time on the target machine. The default is one.
6406 Although the insn scheduler can define itself the possibility of issue
6407 an insn on the same cycle, the value can serve as an additional
6408 constraint to issue insns on the same simulated processor cycle (see
6409 hooks @samp{TARGET_SCHED_REORDER} and @samp{TARGET_SCHED_REORDER2}).
6410 This value must be constant over the entire compilation. If you need
6411 it to vary depending on what the instructions are, you must use
6412 @samp{TARGET_SCHED_VARIABLE_ISSUE}.
6415 @hook TARGET_SCHED_VARIABLE_ISSUE
6416 This hook is executed by the scheduler after it has scheduled an insn
6417 from the ready list. It should return the number of insns which can
6418 still be issued in the current cycle. The default is
6419 @samp{@w{@var{more} - 1}} for insns other than @code{CLOBBER} and
6420 @code{USE}, which normally are not counted against the issue rate.
6421 You should define this hook if some insns take more machine resources
6422 than others, so that fewer insns can follow them in the same cycle.
6423 @var{file} is either a null pointer, or a stdio stream to write any
6424 debug output to. @var{verbose} is the verbose level provided by
6425 @option{-fsched-verbose-@var{n}}. @var{insn} is the instruction that
6429 @hook TARGET_SCHED_ADJUST_COST
6430 This function corrects the value of @var{cost} based on the
6431 relationship between @var{insn} and @var{dep_insn} through the
6432 dependence @var{link}. It should return the new value. The default
6433 is to make no adjustment to @var{cost}. This can be used for example
6434 to specify to the scheduler using the traditional pipeline description
6435 that an output- or anti-dependence does not incur the same cost as a
6436 data-dependence. If the scheduler using the automaton based pipeline
6437 description, the cost of anti-dependence is zero and the cost of
6438 output-dependence is maximum of one and the difference of latency
6439 times of the first and the second insns. If these values are not
6440 acceptable, you could use the hook to modify them too. See also
6441 @pxref{Processor pipeline description}.
6444 @hook TARGET_SCHED_ADJUST_PRIORITY
6445 This hook adjusts the integer scheduling priority @var{priority} of
6446 @var{insn}. It should return the new priority. Increase the priority to
6447 execute @var{insn} earlier, reduce the priority to execute @var{insn}
6448 later. Do not define this hook if you do not need to adjust the
6449 scheduling priorities of insns.
6452 @hook TARGET_SCHED_REORDER
6453 This hook is executed by the scheduler after it has scheduled the ready
6454 list, to allow the machine description to reorder it (for example to
6455 combine two small instructions together on @samp{VLIW} machines).
6456 @var{file} is either a null pointer, or a stdio stream to write any
6457 debug output to. @var{verbose} is the verbose level provided by
6458 @option{-fsched-verbose-@var{n}}. @var{ready} is a pointer to the ready
6459 list of instructions that are ready to be scheduled. @var{n_readyp} is
6460 a pointer to the number of elements in the ready list. The scheduler
6461 reads the ready list in reverse order, starting with
6462 @var{ready}[@var{*n_readyp} @minus{} 1] and going to @var{ready}[0]. @var{clock}
6463 is the timer tick of the scheduler. You may modify the ready list and
6464 the number of ready insns. The return value is the number of insns that
6465 can issue this cycle; normally this is just @code{issue_rate}. See also
6466 @samp{TARGET_SCHED_REORDER2}.
6469 @hook TARGET_SCHED_REORDER2
6470 Like @samp{TARGET_SCHED_REORDER}, but called at a different time. That
6471 function is called whenever the scheduler starts a new cycle. This one
6472 is called once per iteration over a cycle, immediately after
6473 @samp{TARGET_SCHED_VARIABLE_ISSUE}; it can reorder the ready list and
6474 return the number of insns to be scheduled in the same cycle. Defining
6475 this hook can be useful if there are frequent situations where
6476 scheduling one insn causes other insns to become ready in the same
6477 cycle. These other insns can then be taken into account properly.
6480 @hook TARGET_SCHED_DEPENDENCIES_EVALUATION_HOOK
6481 This hook is called after evaluation forward dependencies of insns in
6482 chain given by two parameter values (@var{head} and @var{tail}
6483 correspondingly) but before insns scheduling of the insn chain. For
6484 example, it can be used for better insn classification if it requires
6485 analysis of dependencies. This hook can use backward and forward
6486 dependencies of the insn scheduler because they are already
6490 @hook TARGET_SCHED_INIT
6491 This hook is executed by the scheduler at the beginning of each block of
6492 instructions that are to be scheduled. @var{file} is either a null
6493 pointer, or a stdio stream to write any debug output to. @var{verbose}
6494 is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6495 @var{max_ready} is the maximum number of insns in the current scheduling
6496 region that can be live at the same time. This can be used to allocate
6497 scratch space if it is needed, e.g.@: by @samp{TARGET_SCHED_REORDER}.
6500 @hook TARGET_SCHED_FINISH
6501 This hook is executed by the scheduler at the end of each block of
6502 instructions that are to be scheduled. It can be used to perform
6503 cleanup of any actions done by the other scheduling hooks. @var{file}
6504 is either a null pointer, or a stdio stream to write any debug output
6505 to. @var{verbose} is the verbose level provided by
6506 @option{-fsched-verbose-@var{n}}.
6509 @hook TARGET_SCHED_INIT_GLOBAL
6510 This hook is executed by the scheduler after function level initializations.
6511 @var{file} is either a null pointer, or a stdio stream to write any debug output to.
6512 @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6513 @var{old_max_uid} is the maximum insn uid when scheduling begins.
6516 @hook TARGET_SCHED_FINISH_GLOBAL
6517 This is the cleanup hook corresponding to @code{TARGET_SCHED_INIT_GLOBAL}.
6518 @var{file} is either a null pointer, or a stdio stream to write any debug output to.
6519 @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6522 @hook TARGET_SCHED_DFA_PRE_CYCLE_INSN
6523 The hook returns an RTL insn. The automaton state used in the
6524 pipeline hazard recognizer is changed as if the insn were scheduled
6525 when the new simulated processor cycle starts. Usage of the hook may
6526 simplify the automaton pipeline description for some @acronym{VLIW}
6527 processors. If the hook is defined, it is used only for the automaton
6528 based pipeline description. The default is not to change the state
6529 when the new simulated processor cycle starts.
6532 @hook TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN
6533 The hook can be used to initialize data used by the previous hook.
6536 @hook TARGET_SCHED_DFA_POST_CYCLE_INSN
6537 The hook is analogous to @samp{TARGET_SCHED_DFA_PRE_CYCLE_INSN} but used
6538 to changed the state as if the insn were scheduled when the new
6539 simulated processor cycle finishes.
6542 @hook TARGET_SCHED_INIT_DFA_POST_CYCLE_INSN
6543 The hook is analogous to @samp{TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN} but
6544 used to initialize data used by the previous hook.
6547 @hook TARGET_SCHED_DFA_PRE_ADVANCE_CYCLE
6548 The hook to notify target that the current simulated cycle is about to finish.
6549 The hook is analogous to @samp{TARGET_SCHED_DFA_PRE_CYCLE_INSN} but used
6550 to change the state in more complicated situations - e.g., when advancing
6551 state on a single insn is not enough.
6554 @hook TARGET_SCHED_DFA_POST_ADVANCE_CYCLE
6555 The hook to notify target that new simulated cycle has just started.
6556 The hook is analogous to @samp{TARGET_SCHED_DFA_POST_CYCLE_INSN} but used
6557 to change the state in more complicated situations - e.g., when advancing
6558 state on a single insn is not enough.
6561 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD
6562 This hook controls better choosing an insn from the ready insn queue
6563 for the @acronym{DFA}-based insn scheduler. Usually the scheduler
6564 chooses the first insn from the queue. If the hook returns a positive
6565 value, an additional scheduler code tries all permutations of
6566 @samp{TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD ()}
6567 subsequent ready insns to choose an insn whose issue will result in
6568 maximal number of issued insns on the same cycle. For the
6569 @acronym{VLIW} processor, the code could actually solve the problem of
6570 packing simple insns into the @acronym{VLIW} insn. Of course, if the
6571 rules of @acronym{VLIW} packing are described in the automaton.
6573 This code also could be used for superscalar @acronym{RISC}
6574 processors. Let us consider a superscalar @acronym{RISC} processor
6575 with 3 pipelines. Some insns can be executed in pipelines @var{A} or
6576 @var{B}, some insns can be executed only in pipelines @var{B} or
6577 @var{C}, and one insn can be executed in pipeline @var{B}. The
6578 processor may issue the 1st insn into @var{A} and the 2nd one into
6579 @var{B}. In this case, the 3rd insn will wait for freeing @var{B}
6580 until the next cycle. If the scheduler issues the 3rd insn the first,
6581 the processor could issue all 3 insns per cycle.
6583 Actually this code demonstrates advantages of the automaton based
6584 pipeline hazard recognizer. We try quickly and easy many insn
6585 schedules to choose the best one.
6587 The default is no multipass scheduling.
6590 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD
6592 This hook controls what insns from the ready insn queue will be
6593 considered for the multipass insn scheduling. If the hook returns
6594 zero for @var{insn}, the insn will be not chosen to
6597 The default is that any ready insns can be chosen to be issued.
6600 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_BEGIN
6601 This hook prepares the target backend for a new round of multipass
6605 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_ISSUE
6606 This hook is called when multipass scheduling evaluates instruction INSN.
6609 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_BACKTRACK
6610 This is called when multipass scheduling backtracks from evaluation of
6614 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_END
6615 This hook notifies the target about the result of the concluded current
6616 round of multipass scheduling.
6619 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_INIT
6620 This hook initializes target-specific data used in multipass scheduling.
6623 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_FINI
6624 This hook finalizes target-specific data used in multipass scheduling.
6627 @hook TARGET_SCHED_DFA_NEW_CYCLE
6628 This hook is called by the insn scheduler before issuing @var{insn}
6629 on cycle @var{clock}. If the hook returns nonzero,
6630 @var{insn} is not issued on this processor cycle. Instead,
6631 the processor cycle is advanced. If *@var{sort_p}
6632 is zero, the insn ready queue is not sorted on the new cycle
6633 start as usually. @var{dump} and @var{verbose} specify the file and
6634 verbosity level to use for debugging output.
6635 @var{last_clock} and @var{clock} are, respectively, the
6636 processor cycle on which the previous insn has been issued,
6637 and the current processor cycle.
6640 @hook TARGET_SCHED_IS_COSTLY_DEPENDENCE
6641 This hook is used to define which dependences are considered costly by
6642 the target, so costly that it is not advisable to schedule the insns that
6643 are involved in the dependence too close to one another. The parameters
6644 to this hook are as follows: The first parameter @var{_dep} is the dependence
6645 being evaluated. The second parameter @var{cost} is the cost of the
6646 dependence as estimated by the scheduler, and the third
6647 parameter @var{distance} is the distance in cycles between the two insns.
6648 The hook returns @code{true} if considering the distance between the two
6649 insns the dependence between them is considered costly by the target,
6650 and @code{false} otherwise.
6652 Defining this hook can be useful in multiple-issue out-of-order machines,
6653 where (a) it's practically hopeless to predict the actual data/resource
6654 delays, however: (b) there's a better chance to predict the actual grouping
6655 that will be formed, and (c) correctly emulating the grouping can be very
6656 important. In such targets one may want to allow issuing dependent insns
6657 closer to one another---i.e., closer than the dependence distance; however,
6658 not in cases of ``costly dependences'', which this hooks allows to define.
6661 @hook TARGET_SCHED_H_I_D_EXTENDED
6662 This hook is called by the insn scheduler after emitting a new instruction to
6663 the instruction stream. The hook notifies a target backend to extend its
6664 per instruction data structures.
6667 @hook TARGET_SCHED_ALLOC_SCHED_CONTEXT
6668 Return a pointer to a store large enough to hold target scheduling context.
6671 @hook TARGET_SCHED_INIT_SCHED_CONTEXT
6672 Initialize store pointed to by @var{tc} to hold target scheduling context.
6673 It @var{clean_p} is true then initialize @var{tc} as if scheduler is at the
6674 beginning of the block. Otherwise, copy the current context into @var{tc}.
6677 @hook TARGET_SCHED_SET_SCHED_CONTEXT
6678 Copy target scheduling context pointed to by @var{tc} to the current context.
6681 @hook TARGET_SCHED_CLEAR_SCHED_CONTEXT
6682 Deallocate internal data in target scheduling context pointed to by @var{tc}.
6685 @hook TARGET_SCHED_FREE_SCHED_CONTEXT
6686 Deallocate a store for target scheduling context pointed to by @var{tc}.
6689 @hook TARGET_SCHED_SPECULATE_INSN
6690 This hook is called by the insn scheduler when @var{insn} has only
6691 speculative dependencies and therefore can be scheduled speculatively.
6692 The hook is used to check if the pattern of @var{insn} has a speculative
6693 version and, in case of successful check, to generate that speculative
6694 pattern. The hook should return 1, if the instruction has a speculative form,
6695 or @minus{}1, if it doesn't. @var{request} describes the type of requested
6696 speculation. If the return value equals 1 then @var{new_pat} is assigned
6697 the generated speculative pattern.
6700 @hook TARGET_SCHED_NEEDS_BLOCK_P
6701 This hook is called by the insn scheduler during generation of recovery code
6702 for @var{insn}. It should return @code{true}, if the corresponding check
6703 instruction should branch to recovery code, or @code{false} otherwise.
6706 @hook TARGET_SCHED_GEN_SPEC_CHECK
6707 This hook is called by the insn scheduler to generate a pattern for recovery
6708 check instruction. If @var{mutate_p} is zero, then @var{insn} is a
6709 speculative instruction for which the check should be generated.
6710 @var{label} is either a label of a basic block, where recovery code should
6711 be emitted, or a null pointer, when requested check doesn't branch to
6712 recovery code (a simple check). If @var{mutate_p} is nonzero, then
6713 a pattern for a branchy check corresponding to a simple check denoted by
6714 @var{insn} should be generated. In this case @var{label} can't be null.
6717 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD_SPEC
6718 This hook is used as a workaround for
6719 @samp{TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD} not being
6720 called on the first instruction of the ready list. The hook is used to
6721 discard speculative instructions that stand first in the ready list from
6722 being scheduled on the current cycle. If the hook returns @code{false},
6723 @var{insn} will not be chosen to be issued.
6724 For non-speculative instructions,
6725 the hook should always return @code{true}. For example, in the ia64 backend
6726 the hook is used to cancel data speculative insns when the ALAT table
6730 @hook TARGET_SCHED_SET_SCHED_FLAGS
6731 This hook is used by the insn scheduler to find out what features should be
6733 The structure *@var{spec_info} should be filled in by the target.
6734 The structure describes speculation types that can be used in the scheduler.
6737 @hook TARGET_SCHED_SMS_RES_MII
6738 This hook is called by the swing modulo scheduler to calculate a
6739 resource-based lower bound which is based on the resources available in
6740 the machine and the resources required by each instruction. The target
6741 backend can use @var{g} to calculate such bound. A very simple lower
6742 bound will be used in case this hook is not implemented: the total number
6743 of instructions divided by the issue rate.
6746 @hook TARGET_SCHED_DISPATCH
6747 This hook is called by Haifa Scheduler. It returns true if dispatch scheduling
6748 is supported in hardware and the condition specified in the parameter is true.
6751 @hook TARGET_SCHED_DISPATCH_DO
6752 This hook is called by Haifa Scheduler. It performs the operation specified
6753 in its second parameter.
6756 @hook TARGET_SCHED_EXPOSED_PIPELINE
6758 @hook TARGET_SCHED_REASSOCIATION_WIDTH
6761 @section Dividing the Output into Sections (Texts, Data, @dots{})
6762 @c the above section title is WAY too long. maybe cut the part between
6763 @c the (...)? --mew 10feb93
6765 An object file is divided into sections containing different types of
6766 data. In the most common case, there are three sections: the @dfn{text
6767 section}, which holds instructions and read-only data; the @dfn{data
6768 section}, which holds initialized writable data; and the @dfn{bss
6769 section}, which holds uninitialized data. Some systems have other kinds
6772 @file{varasm.c} provides several well-known sections, such as
6773 @code{text_section}, @code{data_section} and @code{bss_section}.
6774 The normal way of controlling a @code{@var{foo}_section} variable
6775 is to define the associated @code{@var{FOO}_SECTION_ASM_OP} macro,
6776 as described below. The macros are only read once, when @file{varasm.c}
6777 initializes itself, so their values must be run-time constants.
6778 They may however depend on command-line flags.
6780 @emph{Note:} Some run-time files, such @file{crtstuff.c}, also make
6781 use of the @code{@var{FOO}_SECTION_ASM_OP} macros, and expect them
6782 to be string literals.
6784 Some assemblers require a different string to be written every time a
6785 section is selected. If your assembler falls into this category, you
6786 should define the @code{TARGET_ASM_INIT_SECTIONS} hook and use
6787 @code{get_unnamed_section} to set up the sections.
6789 You must always create a @code{text_section}, either by defining
6790 @code{TEXT_SECTION_ASM_OP} or by initializing @code{text_section}
6791 in @code{TARGET_ASM_INIT_SECTIONS}. The same is true of
6792 @code{data_section} and @code{DATA_SECTION_ASM_OP}. If you do not
6793 create a distinct @code{readonly_data_section}, the default is to
6794 reuse @code{text_section}.
6796 All the other @file{varasm.c} sections are optional, and are null
6797 if the target does not provide them.
6799 @defmac TEXT_SECTION_ASM_OP
6800 A C expression whose value is a string, including spacing, containing the
6801 assembler operation that should precede instructions and read-only data.
6802 Normally @code{"\t.text"} is right.
6805 @defmac HOT_TEXT_SECTION_NAME
6806 If defined, a C string constant for the name of the section containing most
6807 frequently executed functions of the program. If not defined, GCC will provide
6808 a default definition if the target supports named sections.
6811 @defmac UNLIKELY_EXECUTED_TEXT_SECTION_NAME
6812 If defined, a C string constant for the name of the section containing unlikely
6813 executed functions in the program.
6816 @defmac DATA_SECTION_ASM_OP
6817 A C expression whose value is a string, including spacing, containing the
6818 assembler operation to identify the following data as writable initialized
6819 data. Normally @code{"\t.data"} is right.
6822 @defmac SDATA_SECTION_ASM_OP
6823 If defined, a C expression whose value is a string, including spacing,
6824 containing the assembler operation to identify the following data as
6825 initialized, writable small data.
6828 @defmac READONLY_DATA_SECTION_ASM_OP
6829 A C expression whose value is a string, including spacing, containing the
6830 assembler operation to identify the following data as read-only initialized
6834 @defmac BSS_SECTION_ASM_OP
6835 If defined, a C expression whose value is a string, including spacing,
6836 containing the assembler operation to identify the following data as
6837 uninitialized global data. If not defined, and
6838 @code{ASM_OUTPUT_ALIGNED_BSS} not defined,
6839 uninitialized global data will be output in the data section if
6840 @option{-fno-common} is passed, otherwise @code{ASM_OUTPUT_COMMON} will be
6844 @defmac SBSS_SECTION_ASM_OP
6845 If defined, a C expression whose value is a string, including spacing,
6846 containing the assembler operation to identify the following data as
6847 uninitialized, writable small data.
6850 @defmac TLS_COMMON_ASM_OP
6851 If defined, a C expression whose value is a string containing the
6852 assembler operation to identify the following data as thread-local
6853 common data. The default is @code{".tls_common"}.
6856 @defmac TLS_SECTION_ASM_FLAG
6857 If defined, a C expression whose value is a character constant
6858 containing the flag used to mark a section as a TLS section. The
6859 default is @code{'T'}.
6862 @defmac INIT_SECTION_ASM_OP
6863 If defined, a C expression whose value is a string, including spacing,
6864 containing the assembler operation to identify the following data as
6865 initialization code. If not defined, GCC will assume such a section does
6866 not exist. This section has no corresponding @code{init_section}
6867 variable; it is used entirely in runtime code.
6870 @defmac FINI_SECTION_ASM_OP
6871 If defined, a C expression whose value is a string, including spacing,
6872 containing the assembler operation to identify the following data as
6873 finalization code. If not defined, GCC will assume such a section does
6874 not exist. This section has no corresponding @code{fini_section}
6875 variable; it is used entirely in runtime code.
6878 @defmac INIT_ARRAY_SECTION_ASM_OP
6879 If defined, a C expression whose value is a string, including spacing,
6880 containing the assembler operation to identify the following data as
6881 part of the @code{.init_array} (or equivalent) section. If not
6882 defined, GCC will assume such a section does not exist. Do not define
6883 both this macro and @code{INIT_SECTION_ASM_OP}.
6886 @defmac FINI_ARRAY_SECTION_ASM_OP
6887 If defined, a C expression whose value is a string, including spacing,
6888 containing the assembler operation to identify the following data as
6889 part of the @code{.fini_array} (or equivalent) section. If not
6890 defined, GCC will assume such a section does not exist. Do not define
6891 both this macro and @code{FINI_SECTION_ASM_OP}.
6894 @defmac CRT_CALL_STATIC_FUNCTION (@var{section_op}, @var{function})
6895 If defined, an ASM statement that switches to a different section
6896 via @var{section_op}, calls @var{function}, and switches back to
6897 the text section. This is used in @file{crtstuff.c} if
6898 @code{INIT_SECTION_ASM_OP} or @code{FINI_SECTION_ASM_OP} to calls
6899 to initialization and finalization functions from the init and fini
6900 sections. By default, this macro uses a simple function call. Some
6901 ports need hand-crafted assembly code to avoid dependencies on
6902 registers initialized in the function prologue or to ensure that
6903 constant pools don't end up too far way in the text section.
6906 @defmac TARGET_LIBGCC_SDATA_SECTION
6907 If defined, a string which names the section into which small
6908 variables defined in crtstuff and libgcc should go. This is useful
6909 when the target has options for optimizing access to small data, and
6910 you want the crtstuff and libgcc routines to be conservative in what
6911 they expect of your application yet liberal in what your application
6912 expects. For example, for targets with a @code{.sdata} section (like
6913 MIPS), you could compile crtstuff with @code{-G 0} so that it doesn't
6914 require small data support from your application, but use this macro
6915 to put small data into @code{.sdata} so that your application can
6916 access these variables whether it uses small data or not.
6919 @defmac FORCE_CODE_SECTION_ALIGN
6920 If defined, an ASM statement that aligns a code section to some
6921 arbitrary boundary. This is used to force all fragments of the
6922 @code{.init} and @code{.fini} sections to have to same alignment
6923 and thus prevent the linker from having to add any padding.
6926 @defmac JUMP_TABLES_IN_TEXT_SECTION
6927 Define this macro to be an expression with a nonzero value if jump
6928 tables (for @code{tablejump} insns) should be output in the text
6929 section, along with the assembler instructions. Otherwise, the
6930 readonly data section is used.
6932 This macro is irrelevant if there is no separate readonly data section.
6935 @hook TARGET_ASM_INIT_SECTIONS
6936 Define this hook if you need to do something special to set up the
6937 @file{varasm.c} sections, or if your target has some special sections
6938 of its own that you need to create.
6940 GCC calls this hook after processing the command line, but before writing
6941 any assembly code, and before calling any of the section-returning hooks
6945 @hook TARGET_ASM_RELOC_RW_MASK
6946 Return a mask describing how relocations should be treated when
6947 selecting sections. Bit 1 should be set if global relocations
6948 should be placed in a read-write section; bit 0 should be set if
6949 local relocations should be placed in a read-write section.
6951 The default version of this function returns 3 when @option{-fpic}
6952 is in effect, and 0 otherwise. The hook is typically redefined
6953 when the target cannot support (some kinds of) dynamic relocations
6954 in read-only sections even in executables.
6957 @hook TARGET_ASM_SELECT_SECTION
6958 Return the section into which @var{exp} should be placed. You can
6959 assume that @var{exp} is either a @code{VAR_DECL} node or a constant of
6960 some sort. @var{reloc} indicates whether the initial value of @var{exp}
6961 requires link-time relocations. Bit 0 is set when variable contains
6962 local relocations only, while bit 1 is set for global relocations.
6963 @var{align} is the constant alignment in bits.
6965 The default version of this function takes care of putting read-only
6966 variables in @code{readonly_data_section}.
6968 See also @var{USE_SELECT_SECTION_FOR_FUNCTIONS}.
6971 @defmac USE_SELECT_SECTION_FOR_FUNCTIONS
6972 Define this macro if you wish TARGET_ASM_SELECT_SECTION to be called
6973 for @code{FUNCTION_DECL}s as well as for variables and constants.
6975 In the case of a @code{FUNCTION_DECL}, @var{reloc} will be zero if the
6976 function has been determined to be likely to be called, and nonzero if
6977 it is unlikely to be called.
6980 @hook TARGET_ASM_UNIQUE_SECTION
6981 Build up a unique section name, expressed as a @code{STRING_CST} node,
6982 and assign it to @samp{DECL_SECTION_NAME (@var{decl})}.
6983 As with @code{TARGET_ASM_SELECT_SECTION}, @var{reloc} indicates whether
6984 the initial value of @var{exp} requires link-time relocations.
6986 The default version of this function appends the symbol name to the
6987 ELF section name that would normally be used for the symbol. For
6988 example, the function @code{foo} would be placed in @code{.text.foo}.
6989 Whatever the actual target object format, this is often good enough.
6992 @hook TARGET_ASM_FUNCTION_RODATA_SECTION
6993 Return the readonly data section associated with
6994 @samp{DECL_SECTION_NAME (@var{decl})}.
6995 The default version of this function selects @code{.gnu.linkonce.r.name} if
6996 the function's section is @code{.gnu.linkonce.t.name}, @code{.rodata.name}
6997 if function is in @code{.text.name}, and the normal readonly-data section
7001 @hook TARGET_ASM_MERGEABLE_RODATA_PREFIX
7003 @hook TARGET_ASM_TM_CLONE_TABLE_SECTION
7005 @hook TARGET_ASM_SELECT_RTX_SECTION
7006 Return the section into which a constant @var{x}, of mode @var{mode},
7007 should be placed. You can assume that @var{x} is some kind of
7008 constant in RTL@. The argument @var{mode} is redundant except in the
7009 case of a @code{const_int} rtx. @var{align} is the constant alignment
7012 The default version of this function takes care of putting symbolic
7013 constants in @code{flag_pic} mode in @code{data_section} and everything
7014 else in @code{readonly_data_section}.
7017 @hook TARGET_MANGLE_DECL_ASSEMBLER_NAME
7018 Define this hook if you need to postprocess the assembler name generated
7019 by target-independent code. The @var{id} provided to this hook will be
7020 the computed name (e.g., the macro @code{DECL_NAME} of the @var{decl} in C,
7021 or the mangled name of the @var{decl} in C++). The return value of the
7022 hook is an @code{IDENTIFIER_NODE} for the appropriate mangled name on
7023 your target system. The default implementation of this hook just
7024 returns the @var{id} provided.
7027 @hook TARGET_ENCODE_SECTION_INFO
7028 Define this hook if references to a symbol or a constant must be
7029 treated differently depending on something about the variable or
7030 function named by the symbol (such as what section it is in).
7032 The hook is executed immediately after rtl has been created for
7033 @var{decl}, which may be a variable or function declaration or
7034 an entry in the constant pool. In either case, @var{rtl} is the
7035 rtl in question. Do @emph{not} use @code{DECL_RTL (@var{decl})}
7036 in this hook; that field may not have been initialized yet.
7038 In the case of a constant, it is safe to assume that the rtl is
7039 a @code{mem} whose address is a @code{symbol_ref}. Most decls
7040 will also have this form, but that is not guaranteed. Global
7041 register variables, for instance, will have a @code{reg} for their
7042 rtl. (Normally the right thing to do with such unusual rtl is
7045 The @var{new_decl_p} argument will be true if this is the first time
7046 that @code{TARGET_ENCODE_SECTION_INFO} has been invoked on this decl. It will
7047 be false for subsequent invocations, which will happen for duplicate
7048 declarations. Whether or not anything must be done for the duplicate
7049 declaration depends on whether the hook examines @code{DECL_ATTRIBUTES}.
7050 @var{new_decl_p} is always true when the hook is called for a constant.
7052 @cindex @code{SYMBOL_REF_FLAG}, in @code{TARGET_ENCODE_SECTION_INFO}
7053 The usual thing for this hook to do is to record flags in the
7054 @code{symbol_ref}, using @code{SYMBOL_REF_FLAG} or @code{SYMBOL_REF_FLAGS}.
7055 Historically, the name string was modified if it was necessary to
7056 encode more than one bit of information, but this practice is now
7057 discouraged; use @code{SYMBOL_REF_FLAGS}.
7059 The default definition of this hook, @code{default_encode_section_info}
7060 in @file{varasm.c}, sets a number of commonly-useful bits in
7061 @code{SYMBOL_REF_FLAGS}. Check whether the default does what you need
7062 before overriding it.
7065 @hook TARGET_STRIP_NAME_ENCODING
7066 Decode @var{name} and return the real name part, sans
7067 the characters that @code{TARGET_ENCODE_SECTION_INFO}
7071 @hook TARGET_IN_SMALL_DATA_P
7072 Returns true if @var{exp} should be placed into a ``small data'' section.
7073 The default version of this hook always returns false.
7076 @hook TARGET_HAVE_SRODATA_SECTION
7077 Contains the value true if the target places read-only
7078 ``small data'' into a separate section. The default value is false.
7081 @hook TARGET_PROFILE_BEFORE_PROLOGUE
7083 @hook TARGET_BINDS_LOCAL_P
7084 Returns true if @var{exp} names an object for which name resolution
7085 rules must resolve to the current ``module'' (dynamic shared library
7086 or executable image).
7088 The default version of this hook implements the name resolution rules
7089 for ELF, which has a looser model of global name binding than other
7090 currently supported object file formats.
7093 @hook TARGET_HAVE_TLS
7094 Contains the value true if the target supports thread-local storage.
7095 The default value is false.
7100 @section Position Independent Code
7101 @cindex position independent code
7104 This section describes macros that help implement generation of position
7105 independent code. Simply defining these macros is not enough to
7106 generate valid PIC; you must also add support to the hook
7107 @code{TARGET_LEGITIMATE_ADDRESS_P} and to the macro
7108 @code{PRINT_OPERAND_ADDRESS}, as well as @code{LEGITIMIZE_ADDRESS}. You
7109 must modify the definition of @samp{movsi} to do something appropriate
7110 when the source operand contains a symbolic address. You may also
7111 need to alter the handling of switch statements so that they use
7113 @c i rearranged the order of the macros above to try to force one of
7114 @c them to the next line, to eliminate an overfull hbox. --mew 10feb93
7116 @defmac PIC_OFFSET_TABLE_REGNUM
7117 The register number of the register used to address a table of static
7118 data addresses in memory. In some cases this register is defined by a
7119 processor's ``application binary interface'' (ABI)@. When this macro
7120 is defined, RTL is generated for this register once, as with the stack
7121 pointer and frame pointer registers. If this macro is not defined, it
7122 is up to the machine-dependent files to allocate such a register (if
7123 necessary). Note that this register must be fixed when in use (e.g.@:
7124 when @code{flag_pic} is true).
7127 @defmac PIC_OFFSET_TABLE_REG_CALL_CLOBBERED
7128 A C expression that is nonzero if the register defined by
7129 @code{PIC_OFFSET_TABLE_REGNUM} is clobbered by calls. If not defined,
7130 the default is zero. Do not define
7131 this macro if @code{PIC_OFFSET_TABLE_REGNUM} is not defined.
7134 @defmac LEGITIMATE_PIC_OPERAND_P (@var{x})
7135 A C expression that is nonzero if @var{x} is a legitimate immediate
7136 operand on the target machine when generating position independent code.
7137 You can assume that @var{x} satisfies @code{CONSTANT_P}, so you need not
7138 check this. You can also assume @var{flag_pic} is true, so you need not
7139 check it either. You need not define this macro if all constants
7140 (including @code{SYMBOL_REF}) can be immediate operands when generating
7141 position independent code.
7144 @node Assembler Format
7145 @section Defining the Output Assembler Language
7147 This section describes macros whose principal purpose is to describe how
7148 to write instructions in assembler language---rather than what the
7152 * File Framework:: Structural information for the assembler file.
7153 * Data Output:: Output of constants (numbers, strings, addresses).
7154 * Uninitialized Data:: Output of uninitialized variables.
7155 * Label Output:: Output and generation of labels.
7156 * Initialization:: General principles of initialization
7157 and termination routines.
7158 * Macros for Initialization::
7159 Specific macros that control the handling of
7160 initialization and termination routines.
7161 * Instruction Output:: Output of actual instructions.
7162 * Dispatch Tables:: Output of jump tables.
7163 * Exception Region Output:: Output of exception region code.
7164 * Alignment Output:: Pseudo ops for alignment and skipping data.
7167 @node File Framework
7168 @subsection The Overall Framework of an Assembler File
7169 @cindex assembler format
7170 @cindex output of assembler code
7172 @c prevent bad page break with this line
7173 This describes the overall framework of an assembly file.
7175 @findex default_file_start
7176 @hook TARGET_ASM_FILE_START
7177 Output to @code{asm_out_file} any text which the assembler expects to
7178 find at the beginning of a file. The default behavior is controlled
7179 by two flags, documented below. Unless your target's assembler is
7180 quite unusual, if you override the default, you should call
7181 @code{default_file_start} at some point in your target hook. This
7182 lets other target files rely on these variables.
7185 @hook TARGET_ASM_FILE_START_APP_OFF
7186 If this flag is true, the text of the macro @code{ASM_APP_OFF} will be
7187 printed as the very first line in the assembly file, unless
7188 @option{-fverbose-asm} is in effect. (If that macro has been defined
7189 to the empty string, this variable has no effect.) With the normal
7190 definition of @code{ASM_APP_OFF}, the effect is to notify the GNU
7191 assembler that it need not bother stripping comments or extra
7192 whitespace from its input. This allows it to work a bit faster.
7194 The default is false. You should not set it to true unless you have
7195 verified that your port does not generate any extra whitespace or
7196 comments that will cause GAS to issue errors in NO_APP mode.
7199 @hook TARGET_ASM_FILE_START_FILE_DIRECTIVE
7200 If this flag is true, @code{output_file_directive} will be called
7201 for the primary source file, immediately after printing
7202 @code{ASM_APP_OFF} (if that is enabled). Most ELF assemblers expect
7203 this to be done. The default is false.
7206 @hook TARGET_ASM_FILE_END
7207 Output to @code{asm_out_file} any text which the assembler expects
7208 to find at the end of a file. The default is to output nothing.
7211 @deftypefun void file_end_indicate_exec_stack ()
7212 Some systems use a common convention, the @samp{.note.GNU-stack}
7213 special section, to indicate whether or not an object file relies on
7214 the stack being executable. If your system uses this convention, you
7215 should define @code{TARGET_ASM_FILE_END} to this function. If you
7216 need to do other things in that hook, have your hook function call
7220 @hook TARGET_ASM_LTO_START
7221 Output to @code{asm_out_file} any text which the assembler expects
7222 to find at the start of an LTO section. The default is to output
7226 @hook TARGET_ASM_LTO_END
7227 Output to @code{asm_out_file} any text which the assembler expects
7228 to find at the end of an LTO section. The default is to output
7232 @hook TARGET_ASM_CODE_END
7233 Output to @code{asm_out_file} any text which is needed before emitting
7234 unwind info and debug info at the end of a file. Some targets emit
7235 here PIC setup thunks that cannot be emitted at the end of file,
7236 because they couldn't have unwind info then. The default is to output
7240 @defmac ASM_COMMENT_START
7241 A C string constant describing how to begin a comment in the target
7242 assembler language. The compiler assumes that the comment will end at
7243 the end of the line.
7247 A C string constant for text to be output before each @code{asm}
7248 statement or group of consecutive ones. Normally this is
7249 @code{"#APP"}, which is a comment that has no effect on most
7250 assemblers but tells the GNU assembler that it must check the lines
7251 that follow for all valid assembler constructs.
7255 A C string constant for text to be output after each @code{asm}
7256 statement or group of consecutive ones. Normally this is
7257 @code{"#NO_APP"}, which tells the GNU assembler to resume making the
7258 time-saving assumptions that are valid for ordinary compiler output.
7261 @defmac ASM_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name})
7262 A C statement to output COFF information or DWARF debugging information
7263 which indicates that filename @var{name} is the current source file to
7264 the stdio stream @var{stream}.
7266 This macro need not be defined if the standard form of output
7267 for the file format in use is appropriate.
7270 @hook TARGET_ASM_OUTPUT_SOURCE_FILENAME
7272 @hook TARGET_ASM_OUTPUT_IDENT
7274 @defmac OUTPUT_QUOTED_STRING (@var{stream}, @var{string})
7275 A C statement to output the string @var{string} to the stdio stream
7276 @var{stream}. If you do not call the function @code{output_quoted_string}
7277 in your config files, GCC will only call it to output filenames to
7278 the assembler source. So you can use it to canonicalize the format
7279 of the filename using this macro.
7282 @hook TARGET_ASM_NAMED_SECTION
7283 Output assembly directives to switch to section @var{name}. The section
7284 should have attributes as specified by @var{flags}, which is a bit mask
7285 of the @code{SECTION_*} flags defined in @file{output.h}. If @var{decl}
7286 is non-NULL, it is the @code{VAR_DECL} or @code{FUNCTION_DECL} with which
7287 this section is associated.
7290 @hook TARGET_ASM_FUNCTION_SECTION
7291 Return preferred text (sub)section for function @var{decl}.
7292 Main purpose of this function is to separate cold, normal and hot
7293 functions. @var{startup} is true when function is known to be used only
7294 at startup (from static constructors or it is @code{main()}).
7295 @var{exit} is true when function is known to be used only at exit
7296 (from static destructors).
7297 Return NULL if function should go to default text section.
7300 @hook TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS
7302 @hook TARGET_HAVE_NAMED_SECTIONS
7303 This flag is true if the target supports @code{TARGET_ASM_NAMED_SECTION}.
7304 It must not be modified by command-line option processing.
7307 @anchor{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}
7308 @hook TARGET_HAVE_SWITCHABLE_BSS_SECTIONS
7309 This flag is true if we can create zeroed data by switching to a BSS
7310 section and then using @code{ASM_OUTPUT_SKIP} to allocate the space.
7311 This is true on most ELF targets.
7314 @hook TARGET_SECTION_TYPE_FLAGS
7315 Choose a set of section attributes for use by @code{TARGET_ASM_NAMED_SECTION}
7316 based on a variable or function decl, a section name, and whether or not the
7317 declaration's initializer may contain runtime relocations. @var{decl} may be
7318 null, in which case read-write data should be assumed.
7320 The default version of this function handles choosing code vs data,
7321 read-only vs read-write data, and @code{flag_pic}. You should only
7322 need to override this if your target has special flags that might be
7323 set via @code{__attribute__}.
7326 @hook TARGET_ASM_RECORD_GCC_SWITCHES
7327 Provides the target with the ability to record the gcc command line
7328 switches that have been passed to the compiler, and options that are
7329 enabled. The @var{type} argument specifies what is being recorded.
7330 It can take the following values:
7333 @item SWITCH_TYPE_PASSED
7334 @var{text} is a command line switch that has been set by the user.
7336 @item SWITCH_TYPE_ENABLED
7337 @var{text} is an option which has been enabled. This might be as a
7338 direct result of a command line switch, or because it is enabled by
7339 default or because it has been enabled as a side effect of a different
7340 command line switch. For example, the @option{-O2} switch enables
7341 various different individual optimization passes.
7343 @item SWITCH_TYPE_DESCRIPTIVE
7344 @var{text} is either NULL or some descriptive text which should be
7345 ignored. If @var{text} is NULL then it is being used to warn the
7346 target hook that either recording is starting or ending. The first
7347 time @var{type} is SWITCH_TYPE_DESCRIPTIVE and @var{text} is NULL, the
7348 warning is for start up and the second time the warning is for
7349 wind down. This feature is to allow the target hook to make any
7350 necessary preparations before it starts to record switches and to
7351 perform any necessary tidying up after it has finished recording
7354 @item SWITCH_TYPE_LINE_START
7355 This option can be ignored by this target hook.
7357 @item SWITCH_TYPE_LINE_END
7358 This option can be ignored by this target hook.
7361 The hook's return value must be zero. Other return values may be
7362 supported in the future.
7364 By default this hook is set to NULL, but an example implementation is
7365 provided for ELF based targets. Called @var{elf_record_gcc_switches},
7366 it records the switches as ASCII text inside a new, string mergeable
7367 section in the assembler output file. The name of the new section is
7368 provided by the @code{TARGET_ASM_RECORD_GCC_SWITCHES_SECTION} target
7372 @hook TARGET_ASM_RECORD_GCC_SWITCHES_SECTION
7373 This is the name of the section that will be created by the example
7374 ELF implementation of the @code{TARGET_ASM_RECORD_GCC_SWITCHES} target
7380 @subsection Output of Data
7383 @hook TARGET_ASM_BYTE_OP
7384 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_HI_OP
7385 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_SI_OP
7386 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_DI_OP
7387 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_TI_OP
7388 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_HI_OP
7389 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_SI_OP
7390 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_DI_OP
7391 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_TI_OP
7392 These hooks specify assembly directives for creating certain kinds
7393 of integer object. The @code{TARGET_ASM_BYTE_OP} directive creates a
7394 byte-sized object, the @code{TARGET_ASM_ALIGNED_HI_OP} one creates an
7395 aligned two-byte object, and so on. Any of the hooks may be
7396 @code{NULL}, indicating that no suitable directive is available.
7398 The compiler will print these strings at the start of a new line,
7399 followed immediately by the object's initial value. In most cases,
7400 the string should contain a tab, a pseudo-op, and then another tab.
7403 @hook TARGET_ASM_INTEGER
7404 The @code{assemble_integer} function uses this hook to output an
7405 integer object. @var{x} is the object's value, @var{size} is its size
7406 in bytes and @var{aligned_p} indicates whether it is aligned. The
7407 function should return @code{true} if it was able to output the
7408 object. If it returns false, @code{assemble_integer} will try to
7409 split the object into smaller parts.
7411 The default implementation of this hook will use the
7412 @code{TARGET_ASM_BYTE_OP} family of strings, returning @code{false}
7413 when the relevant string is @code{NULL}.
7416 @hook TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA
7417 A target hook to recognize @var{rtx} patterns that @code{output_addr_const}
7418 can't deal with, and output assembly code to @var{file} corresponding to
7419 the pattern @var{x}. This may be used to allow machine-dependent
7420 @code{UNSPEC}s to appear within constants.
7422 If target hook fails to recognize a pattern, it must return @code{false},
7423 so that a standard error message is printed. If it prints an error message
7424 itself, by calling, for example, @code{output_operand_lossage}, it may just
7428 @defmac ASM_OUTPUT_ASCII (@var{stream}, @var{ptr}, @var{len})
7429 A C statement to output to the stdio stream @var{stream} an assembler
7430 instruction to assemble a string constant containing the @var{len}
7431 bytes at @var{ptr}. @var{ptr} will be a C expression of type
7432 @code{char *} and @var{len} a C expression of type @code{int}.
7434 If the assembler has a @code{.ascii} pseudo-op as found in the
7435 Berkeley Unix assembler, do not define the macro
7436 @code{ASM_OUTPUT_ASCII}.
7439 @defmac ASM_OUTPUT_FDESC (@var{stream}, @var{decl}, @var{n})
7440 A C statement to output word @var{n} of a function descriptor for
7441 @var{decl}. This must be defined if @code{TARGET_VTABLE_USES_DESCRIPTORS}
7442 is defined, and is otherwise unused.
7445 @defmac CONSTANT_POOL_BEFORE_FUNCTION
7446 You may define this macro as a C expression. You should define the
7447 expression to have a nonzero value if GCC should output the constant
7448 pool for a function before the code for the function, or a zero value if
7449 GCC should output the constant pool after the function. If you do
7450 not define this macro, the usual case, GCC will output the constant
7451 pool before the function.
7454 @defmac ASM_OUTPUT_POOL_PROLOGUE (@var{file}, @var{funname}, @var{fundecl}, @var{size})
7455 A C statement to output assembler commands to define the start of the
7456 constant pool for a function. @var{funname} is a string giving
7457 the name of the function. Should the return type of the function
7458 be required, it can be obtained via @var{fundecl}. @var{size}
7459 is the size, in bytes, of the constant pool that will be written
7460 immediately after this call.
7462 If no constant-pool prefix is required, the usual case, this macro need
7466 @defmac ASM_OUTPUT_SPECIAL_POOL_ENTRY (@var{file}, @var{x}, @var{mode}, @var{align}, @var{labelno}, @var{jumpto})
7467 A C statement (with or without semicolon) to output a constant in the
7468 constant pool, if it needs special treatment. (This macro need not do
7469 anything for RTL expressions that can be output normally.)
7471 The argument @var{file} is the standard I/O stream to output the
7472 assembler code on. @var{x} is the RTL expression for the constant to
7473 output, and @var{mode} is the machine mode (in case @var{x} is a
7474 @samp{const_int}). @var{align} is the required alignment for the value
7475 @var{x}; you should output an assembler directive to force this much
7478 The argument @var{labelno} is a number to use in an internal label for
7479 the address of this pool entry. The definition of this macro is
7480 responsible for outputting the label definition at the proper place.
7481 Here is how to do this:
7484 @code{(*targetm.asm_out.internal_label)} (@var{file}, "LC", @var{labelno});
7487 When you output a pool entry specially, you should end with a
7488 @code{goto} to the label @var{jumpto}. This will prevent the same pool
7489 entry from being output a second time in the usual manner.
7491 You need not define this macro if it would do nothing.
7494 @defmac ASM_OUTPUT_POOL_EPILOGUE (@var{file} @var{funname} @var{fundecl} @var{size})
7495 A C statement to output assembler commands to at the end of the constant
7496 pool for a function. @var{funname} is a string giving the name of the
7497 function. Should the return type of the function be required, you can
7498 obtain it via @var{fundecl}. @var{size} is the size, in bytes, of the
7499 constant pool that GCC wrote immediately before this call.
7501 If no constant-pool epilogue is required, the usual case, you need not
7505 @defmac IS_ASM_LOGICAL_LINE_SEPARATOR (@var{C}, @var{STR})
7506 Define this macro as a C expression which is nonzero if @var{C} is
7507 used as a logical line separator by the assembler. @var{STR} points
7508 to the position in the string where @var{C} was found; this can be used if
7509 a line separator uses multiple characters.
7511 If you do not define this macro, the default is that only
7512 the character @samp{;} is treated as a logical line separator.
7515 @hook TARGET_ASM_OPEN_PAREN
7516 These target hooks are C string constants, describing the syntax in the
7517 assembler for grouping arithmetic expressions. If not overridden, they
7518 default to normal parentheses, which is correct for most assemblers.
7521 These macros are provided by @file{real.h} for writing the definitions
7522 of @code{ASM_OUTPUT_DOUBLE} and the like:
7524 @defmac REAL_VALUE_TO_TARGET_SINGLE (@var{x}, @var{l})
7525 @defmacx REAL_VALUE_TO_TARGET_DOUBLE (@var{x}, @var{l})
7526 @defmacx REAL_VALUE_TO_TARGET_LONG_DOUBLE (@var{x}, @var{l})
7527 @defmacx REAL_VALUE_TO_TARGET_DECIMAL32 (@var{x}, @var{l})
7528 @defmacx REAL_VALUE_TO_TARGET_DECIMAL64 (@var{x}, @var{l})
7529 @defmacx REAL_VALUE_TO_TARGET_DECIMAL128 (@var{x}, @var{l})
7530 These translate @var{x}, of type @code{REAL_VALUE_TYPE}, to the
7531 target's floating point representation, and store its bit pattern in
7532 the variable @var{l}. For @code{REAL_VALUE_TO_TARGET_SINGLE} and
7533 @code{REAL_VALUE_TO_TARGET_DECIMAL32}, this variable should be a
7534 simple @code{long int}. For the others, it should be an array of
7535 @code{long int}. The number of elements in this array is determined
7536 by the size of the desired target floating point data type: 32 bits of
7537 it go in each @code{long int} array element. Each array element holds
7538 32 bits of the result, even if @code{long int} is wider than 32 bits
7539 on the host machine.
7541 The array element values are designed so that you can print them out
7542 using @code{fprintf} in the order they should appear in the target
7546 @node Uninitialized Data
7547 @subsection Output of Uninitialized Variables
7549 Each of the macros in this section is used to do the whole job of
7550 outputting a single uninitialized variable.
7552 @defmac ASM_OUTPUT_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
7553 A C statement (sans semicolon) to output to the stdio stream
7554 @var{stream} the assembler definition of a common-label named
7555 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7556 is the size rounded up to whatever alignment the caller wants. It is
7557 possible that @var{size} may be zero, for instance if a struct with no
7558 other member than a zero-length array is defined. In this case, the
7559 backend must output a symbol definition that allocates at least one
7560 byte, both so that the address of the resulting object does not compare
7561 equal to any other, and because some object formats cannot even express
7562 the concept of a zero-sized common symbol, as that is how they represent
7563 an ordinary undefined external.
7565 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7566 output the name itself; before and after that, output the additional
7567 assembler syntax for defining the name, and a newline.
7569 This macro controls how the assembler definitions of uninitialized
7570 common global variables are output.
7573 @defmac ASM_OUTPUT_ALIGNED_COMMON (@var{stream}, @var{name}, @var{size}, @var{alignment})
7574 Like @code{ASM_OUTPUT_COMMON} except takes the required alignment as a
7575 separate, explicit argument. If you define this macro, it is used in
7576 place of @code{ASM_OUTPUT_COMMON}, and gives you more flexibility in
7577 handling the required alignment of the variable. The alignment is specified
7578 as the number of bits.
7581 @defmac ASM_OUTPUT_ALIGNED_DECL_COMMON (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7582 Like @code{ASM_OUTPUT_ALIGNED_COMMON} except that @var{decl} of the
7583 variable to be output, if there is one, or @code{NULL_TREE} if there
7584 is no corresponding variable. If you define this macro, GCC will use it
7585 in place of both @code{ASM_OUTPUT_COMMON} and
7586 @code{ASM_OUTPUT_ALIGNED_COMMON}. Define this macro when you need to see
7587 the variable's decl in order to chose what to output.
7590 @defmac ASM_OUTPUT_ALIGNED_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7591 A C statement (sans semicolon) to output to the stdio stream
7592 @var{stream} the assembler definition of uninitialized global @var{decl} named
7593 @var{name} whose size is @var{size} bytes. The variable @var{alignment}
7594 is the alignment specified as the number of bits.
7596 Try to use function @code{asm_output_aligned_bss} defined in file
7597 @file{varasm.c} when defining this macro. If unable, use the expression
7598 @code{assemble_name (@var{stream}, @var{name})} to output the name itself;
7599 before and after that, output the additional assembler syntax for defining
7600 the name, and a newline.
7602 There are two ways of handling global BSS@. One is to define this macro.
7603 The other is to have @code{TARGET_ASM_SELECT_SECTION} return a
7604 switchable BSS section (@pxref{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}).
7605 You do not need to do both.
7607 Some languages do not have @code{common} data, and require a
7608 non-common form of global BSS in order to handle uninitialized globals
7609 efficiently. C++ is one example of this. However, if the target does
7610 not support global BSS, the front end may choose to make globals
7611 common in order to save space in the object file.
7614 @defmac ASM_OUTPUT_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded})
7615 A C statement (sans semicolon) to output to the stdio stream
7616 @var{stream} the assembler definition of a local-common-label named
7617 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7618 is the size rounded up to whatever alignment the caller wants.
7620 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7621 output the name itself; before and after that, output the additional
7622 assembler syntax for defining the name, and a newline.
7624 This macro controls how the assembler definitions of uninitialized
7625 static variables are output.
7628 @defmac ASM_OUTPUT_ALIGNED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{alignment})
7629 Like @code{ASM_OUTPUT_LOCAL} except takes the required alignment as a
7630 separate, explicit argument. If you define this macro, it is used in
7631 place of @code{ASM_OUTPUT_LOCAL}, and gives you more flexibility in
7632 handling the required alignment of the variable. The alignment is specified
7633 as the number of bits.
7636 @defmac ASM_OUTPUT_ALIGNED_DECL_LOCAL (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7637 Like @code{ASM_OUTPUT_ALIGNED_DECL} except that @var{decl} of the
7638 variable to be output, if there is one, or @code{NULL_TREE} if there
7639 is no corresponding variable. If you define this macro, GCC will use it
7640 in place of both @code{ASM_OUTPUT_DECL} and
7641 @code{ASM_OUTPUT_ALIGNED_DECL}. Define this macro when you need to see
7642 the variable's decl in order to chose what to output.
7646 @subsection Output and Generation of Labels
7648 @c prevent bad page break with this line
7649 This is about outputting labels.
7651 @findex assemble_name
7652 @defmac ASM_OUTPUT_LABEL (@var{stream}, @var{name})
7653 A C statement (sans semicolon) to output to the stdio stream
7654 @var{stream} the assembler definition of a label named @var{name}.
7655 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7656 output the name itself; before and after that, output the additional
7657 assembler syntax for defining the name, and a newline. A default
7658 definition of this macro is provided which is correct for most systems.
7661 @defmac ASM_OUTPUT_FUNCTION_LABEL (@var{stream}, @var{name}, @var{decl})
7662 A C statement (sans semicolon) to output to the stdio stream
7663 @var{stream} the assembler definition of a label named @var{name} of
7665 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7666 output the name itself; before and after that, output the additional
7667 assembler syntax for defining the name, and a newline. A default
7668 definition of this macro is provided which is correct for most systems.
7670 If this macro is not defined, then the function name is defined in the
7671 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
7674 @findex assemble_name_raw
7675 @defmac ASM_OUTPUT_INTERNAL_LABEL (@var{stream}, @var{name})
7676 Identical to @code{ASM_OUTPUT_LABEL}, except that @var{name} is known
7677 to refer to a compiler-generated label. The default definition uses
7678 @code{assemble_name_raw}, which is like @code{assemble_name} except
7679 that it is more efficient.
7683 A C string containing the appropriate assembler directive to specify the
7684 size of a symbol, without any arguments. On systems that use ELF, the
7685 default (in @file{config/elfos.h}) is @samp{"\t.size\t"}; on other
7686 systems, the default is not to define this macro.
7688 Define this macro only if it is correct to use the default definitions
7689 of @code{ASM_OUTPUT_SIZE_DIRECTIVE} and @code{ASM_OUTPUT_MEASURED_SIZE}
7690 for your system. If you need your own custom definitions of those
7691 macros, or if you do not need explicit symbol sizes at all, do not
7695 @defmac ASM_OUTPUT_SIZE_DIRECTIVE (@var{stream}, @var{name}, @var{size})
7696 A C statement (sans semicolon) to output to the stdio stream
7697 @var{stream} a directive telling the assembler that the size of the
7698 symbol @var{name} is @var{size}. @var{size} is a @code{HOST_WIDE_INT}.
7699 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
7703 @defmac ASM_OUTPUT_MEASURED_SIZE (@var{stream}, @var{name})
7704 A C statement (sans semicolon) to output to the stdio stream
7705 @var{stream} a directive telling the assembler to calculate the size of
7706 the symbol @var{name} by subtracting its address from the current
7709 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
7710 provided. The default assumes that the assembler recognizes a special
7711 @samp{.} symbol as referring to the current address, and can calculate
7712 the difference between this and another symbol. If your assembler does
7713 not recognize @samp{.} or cannot do calculations with it, you will need
7714 to redefine @code{ASM_OUTPUT_MEASURED_SIZE} to use some other technique.
7717 @defmac NO_DOLLAR_IN_LABEL
7718 Define this macro if the assembler does not accept the character
7719 @samp{$} in label names. By default constructors and destructors in
7720 G++ have @samp{$} in the identifiers. If this macro is defined,
7721 @samp{.} is used instead.
7724 @defmac NO_DOT_IN_LABEL
7725 Define this macro if the assembler does not accept the character
7726 @samp{.} in label names. By default constructors and destructors in G++
7727 have names that use @samp{.}. If this macro is defined, these names
7728 are rewritten to avoid @samp{.}.
7732 A C string containing the appropriate assembler directive to specify the
7733 type of a symbol, without any arguments. On systems that use ELF, the
7734 default (in @file{config/elfos.h}) is @samp{"\t.type\t"}; on other
7735 systems, the default is not to define this macro.
7737 Define this macro only if it is correct to use the default definition of
7738 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
7739 custom definition of this macro, or if you do not need explicit symbol
7740 types at all, do not define this macro.
7743 @defmac TYPE_OPERAND_FMT
7744 A C string which specifies (using @code{printf} syntax) the format of
7745 the second operand to @code{TYPE_ASM_OP}. On systems that use ELF, the
7746 default (in @file{config/elfos.h}) is @samp{"@@%s"}; on other systems,
7747 the default is not to define this macro.
7749 Define this macro only if it is correct to use the default definition of
7750 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
7751 custom definition of this macro, or if you do not need explicit symbol
7752 types at all, do not define this macro.
7755 @defmac ASM_OUTPUT_TYPE_DIRECTIVE (@var{stream}, @var{type})
7756 A C statement (sans semicolon) to output to the stdio stream
7757 @var{stream} a directive telling the assembler that the type of the
7758 symbol @var{name} is @var{type}. @var{type} is a C string; currently,
7759 that string is always either @samp{"function"} or @samp{"object"}, but
7760 you should not count on this.
7762 If you define @code{TYPE_ASM_OP} and @code{TYPE_OPERAND_FMT}, a default
7763 definition of this macro is provided.
7766 @defmac ASM_DECLARE_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl})
7767 A C statement (sans semicolon) to output to the stdio stream
7768 @var{stream} any text necessary for declaring the name @var{name} of a
7769 function which is being defined. This macro is responsible for
7770 outputting the label definition (perhaps using
7771 @code{ASM_OUTPUT_FUNCTION_LABEL}). The argument @var{decl} is the
7772 @code{FUNCTION_DECL} tree node representing the function.
7774 If this macro is not defined, then the function name is defined in the
7775 usual manner as a label (by means of @code{ASM_OUTPUT_FUNCTION_LABEL}).
7777 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition
7781 @defmac ASM_DECLARE_FUNCTION_SIZE (@var{stream}, @var{name}, @var{decl})
7782 A C statement (sans semicolon) to output to the stdio stream
7783 @var{stream} any text necessary for declaring the size of a function
7784 which is being defined. The argument @var{name} is the name of the
7785 function. The argument @var{decl} is the @code{FUNCTION_DECL} tree node
7786 representing the function.
7788 If this macro is not defined, then the function size is not defined.
7790 You may wish to use @code{ASM_OUTPUT_MEASURED_SIZE} in the definition
7794 @defmac ASM_DECLARE_OBJECT_NAME (@var{stream}, @var{name}, @var{decl})
7795 A C statement (sans semicolon) to output to the stdio stream
7796 @var{stream} any text necessary for declaring the name @var{name} of an
7797 initialized variable which is being defined. This macro must output the
7798 label definition (perhaps using @code{ASM_OUTPUT_LABEL}). The argument
7799 @var{decl} is the @code{VAR_DECL} tree node representing the variable.
7801 If this macro is not defined, then the variable name is defined in the
7802 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
7804 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} and/or
7805 @code{ASM_OUTPUT_SIZE_DIRECTIVE} in the definition of this macro.
7808 @hook TARGET_ASM_DECLARE_CONSTANT_NAME
7809 A target hook to output to the stdio stream @var{file} any text necessary
7810 for declaring the name @var{name} of a constant which is being defined. This
7811 target hook is responsible for outputting the label definition (perhaps using
7812 @code{assemble_label}). The argument @var{exp} is the value of the constant,
7813 and @var{size} is the size of the constant in bytes. The @var{name}
7814 will be an internal label.
7816 The default version of this target hook, define the @var{name} in the
7817 usual manner as a label (by means of @code{assemble_label}).
7819 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in this target hook.
7822 @defmac ASM_DECLARE_REGISTER_GLOBAL (@var{stream}, @var{decl}, @var{regno}, @var{name})
7823 A C statement (sans semicolon) to output to the stdio stream
7824 @var{stream} any text necessary for claiming a register @var{regno}
7825 for a global variable @var{decl} with name @var{name}.
7827 If you don't define this macro, that is equivalent to defining it to do
7831 @defmac ASM_FINISH_DECLARE_OBJECT (@var{stream}, @var{decl}, @var{toplevel}, @var{atend})
7832 A C statement (sans semicolon) to finish up declaring a variable name
7833 once the compiler has processed its initializer fully and thus has had a
7834 chance to determine the size of an array when controlled by an
7835 initializer. This is used on systems where it's necessary to declare
7836 something about the size of the object.
7838 If you don't define this macro, that is equivalent to defining it to do
7841 You may wish to use @code{ASM_OUTPUT_SIZE_DIRECTIVE} and/or
7842 @code{ASM_OUTPUT_MEASURED_SIZE} in the definition of this macro.
7845 @hook TARGET_ASM_GLOBALIZE_LABEL
7846 This target hook is a function to output to the stdio stream
7847 @var{stream} some commands that will make the label @var{name} global;
7848 that is, available for reference from other files.
7850 The default implementation relies on a proper definition of
7851 @code{GLOBAL_ASM_OP}.
7854 @hook TARGET_ASM_GLOBALIZE_DECL_NAME
7855 This target hook is a function to output to the stdio stream
7856 @var{stream} some commands that will make the name associated with @var{decl}
7857 global; that is, available for reference from other files.
7859 The default implementation uses the TARGET_ASM_GLOBALIZE_LABEL target hook.
7862 @defmac ASM_WEAKEN_LABEL (@var{stream}, @var{name})
7863 A C statement (sans semicolon) to output to the stdio stream
7864 @var{stream} some commands that will make the label @var{name} weak;
7865 that is, available for reference from other files but only used if
7866 no other definition is available. Use the expression
7867 @code{assemble_name (@var{stream}, @var{name})} to output the name
7868 itself; before and after that, output the additional assembler syntax
7869 for making that name weak, and a newline.
7871 If you don't define this macro or @code{ASM_WEAKEN_DECL}, GCC will not
7872 support weak symbols and you should not define the @code{SUPPORTS_WEAK}
7876 @defmac ASM_WEAKEN_DECL (@var{stream}, @var{decl}, @var{name}, @var{value})
7877 Combines (and replaces) the function of @code{ASM_WEAKEN_LABEL} and
7878 @code{ASM_OUTPUT_WEAK_ALIAS}, allowing access to the associated function
7879 or variable decl. If @var{value} is not @code{NULL}, this C statement
7880 should output to the stdio stream @var{stream} assembler code which
7881 defines (equates) the weak symbol @var{name} to have the value
7882 @var{value}. If @var{value} is @code{NULL}, it should output commands
7883 to make @var{name} weak.
7886 @defmac ASM_OUTPUT_WEAKREF (@var{stream}, @var{decl}, @var{name}, @var{value})
7887 Outputs a directive that enables @var{name} to be used to refer to
7888 symbol @var{value} with weak-symbol semantics. @code{decl} is the
7889 declaration of @code{name}.
7892 @defmac SUPPORTS_WEAK
7893 A preprocessor constant expression which evaluates to true if the target
7894 supports weak symbols.
7896 If you don't define this macro, @file{defaults.h} provides a default
7897 definition. If either @code{ASM_WEAKEN_LABEL} or @code{ASM_WEAKEN_DECL}
7898 is defined, the default definition is @samp{1}; otherwise, it is @samp{0}.
7901 @defmac TARGET_SUPPORTS_WEAK
7902 A C expression which evaluates to true if the target supports weak symbols.
7904 If you don't define this macro, @file{defaults.h} provides a default
7905 definition. The default definition is @samp{(SUPPORTS_WEAK)}. Define
7906 this macro if you want to control weak symbol support with a compiler
7907 flag such as @option{-melf}.
7910 @defmac MAKE_DECL_ONE_ONLY (@var{decl})
7911 A C statement (sans semicolon) to mark @var{decl} to be emitted as a
7912 public symbol such that extra copies in multiple translation units will
7913 be discarded by the linker. Define this macro if your object file
7914 format provides support for this concept, such as the @samp{COMDAT}
7915 section flags in the Microsoft Windows PE/COFF format, and this support
7916 requires changes to @var{decl}, such as putting it in a separate section.
7919 @defmac SUPPORTS_ONE_ONLY
7920 A C expression which evaluates to true if the target supports one-only
7923 If you don't define this macro, @file{varasm.c} provides a default
7924 definition. If @code{MAKE_DECL_ONE_ONLY} is defined, the default
7925 definition is @samp{1}; otherwise, it is @samp{0}. Define this macro if
7926 you want to control one-only symbol support with a compiler flag, or if
7927 setting the @code{DECL_ONE_ONLY} flag is enough to mark a declaration to
7928 be emitted as one-only.
7931 @hook TARGET_ASM_ASSEMBLE_VISIBILITY
7932 This target hook is a function to output to @var{asm_out_file} some
7933 commands that will make the symbol(s) associated with @var{decl} have
7934 hidden, protected or internal visibility as specified by @var{visibility}.
7937 @defmac TARGET_WEAK_NOT_IN_ARCHIVE_TOC
7938 A C expression that evaluates to true if the target's linker expects
7939 that weak symbols do not appear in a static archive's table of contents.
7940 The default is @code{0}.
7942 Leaving weak symbols out of an archive's table of contents means that,
7943 if a symbol will only have a definition in one translation unit and
7944 will have undefined references from other translation units, that
7945 symbol should not be weak. Defining this macro to be nonzero will
7946 thus have the effect that certain symbols that would normally be weak
7947 (explicit template instantiations, and vtables for polymorphic classes
7948 with noninline key methods) will instead be nonweak.
7950 The C++ ABI requires this macro to be zero. Define this macro for
7951 targets where full C++ ABI compliance is impossible and where linker
7952 restrictions require weak symbols to be left out of a static archive's
7956 @defmac ASM_OUTPUT_EXTERNAL (@var{stream}, @var{decl}, @var{name})
7957 A C statement (sans semicolon) to output to the stdio stream
7958 @var{stream} any text necessary for declaring the name of an external
7959 symbol named @var{name} which is referenced in this compilation but
7960 not defined. The value of @var{decl} is the tree node for the
7963 This macro need not be defined if it does not need to output anything.
7964 The GNU assembler and most Unix assemblers don't require anything.
7967 @hook TARGET_ASM_EXTERNAL_LIBCALL
7968 This target hook is a function to output to @var{asm_out_file} an assembler
7969 pseudo-op to declare a library function name external. The name of the
7970 library function is given by @var{symref}, which is a @code{symbol_ref}.
7973 @hook TARGET_ASM_MARK_DECL_PRESERVED
7974 This target hook is a function to output to @var{asm_out_file} an assembler
7975 directive to annotate @var{symbol} as used. The Darwin target uses the
7976 .no_dead_code_strip directive.
7979 @defmac ASM_OUTPUT_LABELREF (@var{stream}, @var{name})
7980 A C statement (sans semicolon) to output to the stdio stream
7981 @var{stream} a reference in assembler syntax to a label named
7982 @var{name}. This should add @samp{_} to the front of the name, if that
7983 is customary on your operating system, as it is in most Berkeley Unix
7984 systems. This macro is used in @code{assemble_name}.
7987 @hook TARGET_MANGLE_ASSEMBLER_NAME
7989 @defmac ASM_OUTPUT_SYMBOL_REF (@var{stream}, @var{sym})
7990 A C statement (sans semicolon) to output a reference to
7991 @code{SYMBOL_REF} @var{sym}. If not defined, @code{assemble_name}
7992 will be used to output the name of the symbol. This macro may be used
7993 to modify the way a symbol is referenced depending on information
7994 encoded by @code{TARGET_ENCODE_SECTION_INFO}.
7997 @defmac ASM_OUTPUT_LABEL_REF (@var{stream}, @var{buf})
7998 A C statement (sans semicolon) to output a reference to @var{buf}, the
7999 result of @code{ASM_GENERATE_INTERNAL_LABEL}. If not defined,
8000 @code{assemble_name} will be used to output the name of the symbol.
8001 This macro is not used by @code{output_asm_label}, or the @code{%l}
8002 specifier that calls it; the intention is that this macro should be set
8003 when it is necessary to output a label differently when its address is
8007 @hook TARGET_ASM_INTERNAL_LABEL
8008 A function to output to the stdio stream @var{stream} a label whose
8009 name is made from the string @var{prefix} and the number @var{labelno}.
8011 It is absolutely essential that these labels be distinct from the labels
8012 used for user-level functions and variables. Otherwise, certain programs
8013 will have name conflicts with internal labels.
8015 It is desirable to exclude internal labels from the symbol table of the
8016 object file. Most assemblers have a naming convention for labels that
8017 should be excluded; on many systems, the letter @samp{L} at the
8018 beginning of a label has this effect. You should find out what
8019 convention your system uses, and follow it.
8021 The default version of this function utilizes @code{ASM_GENERATE_INTERNAL_LABEL}.
8024 @defmac ASM_OUTPUT_DEBUG_LABEL (@var{stream}, @var{prefix}, @var{num})
8025 A C statement to output to the stdio stream @var{stream} a debug info
8026 label whose name is made from the string @var{prefix} and the number
8027 @var{num}. This is useful for VLIW targets, where debug info labels
8028 may need to be treated differently than branch target labels. On some
8029 systems, branch target labels must be at the beginning of instruction
8030 bundles, but debug info labels can occur in the middle of instruction
8033 If this macro is not defined, then @code{(*targetm.asm_out.internal_label)} will be
8037 @defmac ASM_GENERATE_INTERNAL_LABEL (@var{string}, @var{prefix}, @var{num})
8038 A C statement to store into the string @var{string} a label whose name
8039 is made from the string @var{prefix} and the number @var{num}.
8041 This string, when output subsequently by @code{assemble_name}, should
8042 produce the output that @code{(*targetm.asm_out.internal_label)} would produce
8043 with the same @var{prefix} and @var{num}.
8045 If the string begins with @samp{*}, then @code{assemble_name} will
8046 output the rest of the string unchanged. It is often convenient for
8047 @code{ASM_GENERATE_INTERNAL_LABEL} to use @samp{*} in this way. If the
8048 string doesn't start with @samp{*}, then @code{ASM_OUTPUT_LABELREF} gets
8049 to output the string, and may change it. (Of course,
8050 @code{ASM_OUTPUT_LABELREF} is also part of your machine description, so
8051 you should know what it does on your machine.)
8054 @defmac ASM_FORMAT_PRIVATE_NAME (@var{outvar}, @var{name}, @var{number})
8055 A C expression to assign to @var{outvar} (which is a variable of type
8056 @code{char *}) a newly allocated string made from the string
8057 @var{name} and the number @var{number}, with some suitable punctuation
8058 added. Use @code{alloca} to get space for the string.
8060 The string will be used as an argument to @code{ASM_OUTPUT_LABELREF} to
8061 produce an assembler label for an internal static variable whose name is
8062 @var{name}. Therefore, the string must be such as to result in valid
8063 assembler code. The argument @var{number} is different each time this
8064 macro is executed; it prevents conflicts between similarly-named
8065 internal static variables in different scopes.
8067 Ideally this string should not be a valid C identifier, to prevent any
8068 conflict with the user's own symbols. Most assemblers allow periods
8069 or percent signs in assembler symbols; putting at least one of these
8070 between the name and the number will suffice.
8072 If this macro is not defined, a default definition will be provided
8073 which is correct for most systems.
8076 @defmac ASM_OUTPUT_DEF (@var{stream}, @var{name}, @var{value})
8077 A C statement to output to the stdio stream @var{stream} assembler code
8078 which defines (equates) the symbol @var{name} to have the value @var{value}.
8081 If @code{SET_ASM_OP} is defined, a default definition is provided which is
8082 correct for most systems.
8085 @defmac ASM_OUTPUT_DEF_FROM_DECLS (@var{stream}, @var{decl_of_name}, @var{decl_of_value})
8086 A C statement to output to the stdio stream @var{stream} assembler code
8087 which defines (equates) the symbol whose tree node is @var{decl_of_name}
8088 to have the value of the tree node @var{decl_of_value}. This macro will
8089 be used in preference to @samp{ASM_OUTPUT_DEF} if it is defined and if
8090 the tree nodes are available.
8093 If @code{SET_ASM_OP} is defined, a default definition is provided which is
8094 correct for most systems.
8097 @defmac TARGET_DEFERRED_OUTPUT_DEFS (@var{decl_of_name}, @var{decl_of_value})
8098 A C statement that evaluates to true if the assembler code which defines
8099 (equates) the symbol whose tree node is @var{decl_of_name} to have the value
8100 of the tree node @var{decl_of_value} should be emitted near the end of the
8101 current compilation unit. The default is to not defer output of defines.
8102 This macro affects defines output by @samp{ASM_OUTPUT_DEF} and
8103 @samp{ASM_OUTPUT_DEF_FROM_DECLS}.
8106 @defmac ASM_OUTPUT_WEAK_ALIAS (@var{stream}, @var{name}, @var{value})
8107 A C statement to output to the stdio stream @var{stream} assembler code
8108 which defines (equates) the weak symbol @var{name} to have the value
8109 @var{value}. If @var{value} is @code{NULL}, it defines @var{name} as
8110 an undefined weak symbol.
8112 Define this macro if the target only supports weak aliases; define
8113 @code{ASM_OUTPUT_DEF} instead if possible.
8116 @defmac OBJC_GEN_METHOD_LABEL (@var{buf}, @var{is_inst}, @var{class_name}, @var{cat_name}, @var{sel_name})
8117 Define this macro to override the default assembler names used for
8118 Objective-C methods.
8120 The default name is a unique method number followed by the name of the
8121 class (e.g.@: @samp{_1_Foo}). For methods in categories, the name of
8122 the category is also included in the assembler name (e.g.@:
8125 These names are safe on most systems, but make debugging difficult since
8126 the method's selector is not present in the name. Therefore, particular
8127 systems define other ways of computing names.
8129 @var{buf} is an expression of type @code{char *} which gives you a
8130 buffer in which to store the name; its length is as long as
8131 @var{class_name}, @var{cat_name} and @var{sel_name} put together, plus
8132 50 characters extra.
8134 The argument @var{is_inst} specifies whether the method is an instance
8135 method or a class method; @var{class_name} is the name of the class;
8136 @var{cat_name} is the name of the category (or @code{NULL} if the method is not
8137 in a category); and @var{sel_name} is the name of the selector.
8139 On systems where the assembler can handle quoted names, you can use this
8140 macro to provide more human-readable names.
8143 @node Initialization
8144 @subsection How Initialization Functions Are Handled
8145 @cindex initialization routines
8146 @cindex termination routines
8147 @cindex constructors, output of
8148 @cindex destructors, output of
8150 The compiled code for certain languages includes @dfn{constructors}
8151 (also called @dfn{initialization routines})---functions to initialize
8152 data in the program when the program is started. These functions need
8153 to be called before the program is ``started''---that is to say, before
8154 @code{main} is called.
8156 Compiling some languages generates @dfn{destructors} (also called
8157 @dfn{termination routines}) that should be called when the program
8160 To make the initialization and termination functions work, the compiler
8161 must output something in the assembler code to cause those functions to
8162 be called at the appropriate time. When you port the compiler to a new
8163 system, you need to specify how to do this.
8165 There are two major ways that GCC currently supports the execution of
8166 initialization and termination functions. Each way has two variants.
8167 Much of the structure is common to all four variations.
8169 @findex __CTOR_LIST__
8170 @findex __DTOR_LIST__
8171 The linker must build two lists of these functions---a list of
8172 initialization functions, called @code{__CTOR_LIST__}, and a list of
8173 termination functions, called @code{__DTOR_LIST__}.
8175 Each list always begins with an ignored function pointer (which may hold
8176 0, @minus{}1, or a count of the function pointers after it, depending on
8177 the environment). This is followed by a series of zero or more function
8178 pointers to constructors (or destructors), followed by a function
8179 pointer containing zero.
8181 Depending on the operating system and its executable file format, either
8182 @file{crtstuff.c} or @file{libgcc2.c} traverses these lists at startup
8183 time and exit time. Constructors are called in reverse order of the
8184 list; destructors in forward order.
8186 The best way to handle static constructors works only for object file
8187 formats which provide arbitrarily-named sections. A section is set
8188 aside for a list of constructors, and another for a list of destructors.
8189 Traditionally these are called @samp{.ctors} and @samp{.dtors}. Each
8190 object file that defines an initialization function also puts a word in
8191 the constructor section to point to that function. The linker
8192 accumulates all these words into one contiguous @samp{.ctors} section.
8193 Termination functions are handled similarly.
8195 This method will be chosen as the default by @file{target-def.h} if
8196 @code{TARGET_ASM_NAMED_SECTION} is defined. A target that does not
8197 support arbitrary sections, but does support special designated
8198 constructor and destructor sections may define @code{CTORS_SECTION_ASM_OP}
8199 and @code{DTORS_SECTION_ASM_OP} to achieve the same effect.
8201 When arbitrary sections are available, there are two variants, depending
8202 upon how the code in @file{crtstuff.c} is called. On systems that
8203 support a @dfn{.init} section which is executed at program startup,
8204 parts of @file{crtstuff.c} are compiled into that section. The
8205 program is linked by the @command{gcc} driver like this:
8208 ld -o @var{output_file} crti.o crtbegin.o @dots{} -lgcc crtend.o crtn.o
8211 The prologue of a function (@code{__init}) appears in the @code{.init}
8212 section of @file{crti.o}; the epilogue appears in @file{crtn.o}. Likewise
8213 for the function @code{__fini} in the @dfn{.fini} section. Normally these
8214 files are provided by the operating system or by the GNU C library, but
8215 are provided by GCC for a few targets.
8217 The objects @file{crtbegin.o} and @file{crtend.o} are (for most targets)
8218 compiled from @file{crtstuff.c}. They contain, among other things, code
8219 fragments within the @code{.init} and @code{.fini} sections that branch
8220 to routines in the @code{.text} section. The linker will pull all parts
8221 of a section together, which results in a complete @code{__init} function
8222 that invokes the routines we need at startup.
8224 To use this variant, you must define the @code{INIT_SECTION_ASM_OP}
8227 If no init section is available, when GCC compiles any function called
8228 @code{main} (or more accurately, any function designated as a program
8229 entry point by the language front end calling @code{expand_main_function}),
8230 it inserts a procedure call to @code{__main} as the first executable code
8231 after the function prologue. The @code{__main} function is defined
8232 in @file{libgcc2.c} and runs the global constructors.
8234 In file formats that don't support arbitrary sections, there are again
8235 two variants. In the simplest variant, the GNU linker (GNU @code{ld})
8236 and an `a.out' format must be used. In this case,
8237 @code{TARGET_ASM_CONSTRUCTOR} is defined to produce a @code{.stabs}
8238 entry of type @samp{N_SETT}, referencing the name @code{__CTOR_LIST__},
8239 and with the address of the void function containing the initialization
8240 code as its value. The GNU linker recognizes this as a request to add
8241 the value to a @dfn{set}; the values are accumulated, and are eventually
8242 placed in the executable as a vector in the format described above, with
8243 a leading (ignored) count and a trailing zero element.
8244 @code{TARGET_ASM_DESTRUCTOR} is handled similarly. Since no init
8245 section is available, the absence of @code{INIT_SECTION_ASM_OP} causes
8246 the compilation of @code{main} to call @code{__main} as above, starting
8247 the initialization process.
8249 The last variant uses neither arbitrary sections nor the GNU linker.
8250 This is preferable when you want to do dynamic linking and when using
8251 file formats which the GNU linker does not support, such as `ECOFF'@. In
8252 this case, @code{TARGET_HAVE_CTORS_DTORS} is false, initialization and
8253 termination functions are recognized simply by their names. This requires
8254 an extra program in the linkage step, called @command{collect2}. This program
8255 pretends to be the linker, for use with GCC; it does its job by running
8256 the ordinary linker, but also arranges to include the vectors of
8257 initialization and termination functions. These functions are called
8258 via @code{__main} as described above. In order to use this method,
8259 @code{use_collect2} must be defined in the target in @file{config.gcc}.
8262 The following section describes the specific macros that control and
8263 customize the handling of initialization and termination functions.
8266 @node Macros for Initialization
8267 @subsection Macros Controlling Initialization Routines
8269 Here are the macros that control how the compiler handles initialization
8270 and termination functions:
8272 @defmac INIT_SECTION_ASM_OP
8273 If defined, a C string constant, including spacing, for the assembler
8274 operation to identify the following data as initialization code. If not
8275 defined, GCC will assume such a section does not exist. When you are
8276 using special sections for initialization and termination functions, this
8277 macro also controls how @file{crtstuff.c} and @file{libgcc2.c} arrange to
8278 run the initialization functions.
8281 @defmac HAS_INIT_SECTION
8282 If defined, @code{main} will not call @code{__main} as described above.
8283 This macro should be defined for systems that control start-up code
8284 on a symbol-by-symbol basis, such as OSF/1, and should not
8285 be defined explicitly for systems that support @code{INIT_SECTION_ASM_OP}.
8288 @defmac LD_INIT_SWITCH
8289 If defined, a C string constant for a switch that tells the linker that
8290 the following symbol is an initialization routine.
8293 @defmac LD_FINI_SWITCH
8294 If defined, a C string constant for a switch that tells the linker that
8295 the following symbol is a finalization routine.
8298 @defmac COLLECT_SHARED_INIT_FUNC (@var{stream}, @var{func})
8299 If defined, a C statement that will write a function that can be
8300 automatically called when a shared library is loaded. The function
8301 should call @var{func}, which takes no arguments. If not defined, and
8302 the object format requires an explicit initialization function, then a
8303 function called @code{_GLOBAL__DI} will be generated.
8305 This function and the following one are used by collect2 when linking a
8306 shared library that needs constructors or destructors, or has DWARF2
8307 exception tables embedded in the code.
8310 @defmac COLLECT_SHARED_FINI_FUNC (@var{stream}, @var{func})
8311 If defined, a C statement that will write a function that can be
8312 automatically called when a shared library is unloaded. The function
8313 should call @var{func}, which takes no arguments. If not defined, and
8314 the object format requires an explicit finalization function, then a
8315 function called @code{_GLOBAL__DD} will be generated.
8318 @defmac INVOKE__main
8319 If defined, @code{main} will call @code{__main} despite the presence of
8320 @code{INIT_SECTION_ASM_OP}. This macro should be defined for systems
8321 where the init section is not actually run automatically, but is still
8322 useful for collecting the lists of constructors and destructors.
8325 @defmac SUPPORTS_INIT_PRIORITY
8326 If nonzero, the C++ @code{init_priority} attribute is supported and the
8327 compiler should emit instructions to control the order of initialization
8328 of objects. If zero, the compiler will issue an error message upon
8329 encountering an @code{init_priority} attribute.
8332 @hook TARGET_HAVE_CTORS_DTORS
8333 This value is true if the target supports some ``native'' method of
8334 collecting constructors and destructors to be run at startup and exit.
8335 It is false if we must use @command{collect2}.
8338 @hook TARGET_ASM_CONSTRUCTOR
8339 If defined, a function that outputs assembler code to arrange to call
8340 the function referenced by @var{symbol} at initialization time.
8342 Assume that @var{symbol} is a @code{SYMBOL_REF} for a function taking
8343 no arguments and with no return value. If the target supports initialization
8344 priorities, @var{priority} is a value between 0 and @code{MAX_INIT_PRIORITY};
8345 otherwise it must be @code{DEFAULT_INIT_PRIORITY}.
8347 If this macro is not defined by the target, a suitable default will
8348 be chosen if (1) the target supports arbitrary section names, (2) the
8349 target defines @code{CTORS_SECTION_ASM_OP}, or (3) @code{USE_COLLECT2}
8353 @hook TARGET_ASM_DESTRUCTOR
8354 This is like @code{TARGET_ASM_CONSTRUCTOR} but used for termination
8355 functions rather than initialization functions.
8358 If @code{TARGET_HAVE_CTORS_DTORS} is true, the initialization routine
8359 generated for the generated object file will have static linkage.
8361 If your system uses @command{collect2} as the means of processing
8362 constructors, then that program normally uses @command{nm} to scan
8363 an object file for constructor functions to be called.
8365 On certain kinds of systems, you can define this macro to make
8366 @command{collect2} work faster (and, in some cases, make it work at all):
8368 @defmac OBJECT_FORMAT_COFF
8369 Define this macro if the system uses COFF (Common Object File Format)
8370 object files, so that @command{collect2} can assume this format and scan
8371 object files directly for dynamic constructor/destructor functions.
8373 This macro is effective only in a native compiler; @command{collect2} as
8374 part of a cross compiler always uses @command{nm} for the target machine.
8377 @defmac REAL_NM_FILE_NAME
8378 Define this macro as a C string constant containing the file name to use
8379 to execute @command{nm}. The default is to search the path normally for
8384 @command{collect2} calls @command{nm} to scan object files for static
8385 constructors and destructors and LTO info. By default, @option{-n} is
8386 passed. Define @code{NM_FLAGS} to a C string constant if other options
8387 are needed to get the same output format as GNU @command{nm -n}
8391 If your system supports shared libraries and has a program to list the
8392 dynamic dependencies of a given library or executable, you can define
8393 these macros to enable support for running initialization and
8394 termination functions in shared libraries:
8397 Define this macro to a C string constant containing the name of the program
8398 which lists dynamic dependencies, like @command{ldd} under SunOS 4.
8401 @defmac PARSE_LDD_OUTPUT (@var{ptr})
8402 Define this macro to be C code that extracts filenames from the output
8403 of the program denoted by @code{LDD_SUFFIX}. @var{ptr} is a variable
8404 of type @code{char *} that points to the beginning of a line of output
8405 from @code{LDD_SUFFIX}. If the line lists a dynamic dependency, the
8406 code must advance @var{ptr} to the beginning of the filename on that
8407 line. Otherwise, it must set @var{ptr} to @code{NULL}.
8410 @defmac SHLIB_SUFFIX
8411 Define this macro to a C string constant containing the default shared
8412 library extension of the target (e.g., @samp{".so"}). @command{collect2}
8413 strips version information after this suffix when generating global
8414 constructor and destructor names. This define is only needed on targets
8415 that use @command{collect2} to process constructors and destructors.
8418 @node Instruction Output
8419 @subsection Output of Assembler Instructions
8421 @c prevent bad page break with this line
8422 This describes assembler instruction output.
8424 @defmac REGISTER_NAMES
8425 A C initializer containing the assembler's names for the machine
8426 registers, each one as a C string constant. This is what translates
8427 register numbers in the compiler into assembler language.
8430 @defmac ADDITIONAL_REGISTER_NAMES
8431 If defined, a C initializer for an array of structures containing a name
8432 and a register number. This macro defines additional names for hard
8433 registers, thus allowing the @code{asm} option in declarations to refer
8434 to registers using alternate names.
8437 @defmac OVERLAPPING_REGISTER_NAMES
8438 If defined, a C initializer for an array of structures containing a
8439 name, a register number and a count of the number of consecutive
8440 machine registers the name overlaps. This macro defines additional
8441 names for hard registers, thus allowing the @code{asm} option in
8442 declarations to refer to registers using alternate names. Unlike
8443 @code{ADDITIONAL_REGISTER_NAMES}, this macro should be used when the
8444 register name implies multiple underlying registers.
8446 This macro should be used when it is important that a clobber in an
8447 @code{asm} statement clobbers all the underlying values implied by the
8448 register name. For example, on ARM, clobbering the double-precision
8449 VFP register ``d0'' implies clobbering both single-precision registers
8453 @defmac ASM_OUTPUT_OPCODE (@var{stream}, @var{ptr})
8454 Define this macro if you are using an unusual assembler that
8455 requires different names for the machine instructions.
8457 The definition is a C statement or statements which output an
8458 assembler instruction opcode to the stdio stream @var{stream}. The
8459 macro-operand @var{ptr} is a variable of type @code{char *} which
8460 points to the opcode name in its ``internal'' form---the form that is
8461 written in the machine description. The definition should output the
8462 opcode name to @var{stream}, performing any translation you desire, and
8463 increment the variable @var{ptr} to point at the end of the opcode
8464 so that it will not be output twice.
8466 In fact, your macro definition may process less than the entire opcode
8467 name, or more than the opcode name; but if you want to process text
8468 that includes @samp{%}-sequences to substitute operands, you must take
8469 care of the substitution yourself. Just be sure to increment
8470 @var{ptr} over whatever text should not be output normally.
8472 @findex recog_data.operand
8473 If you need to look at the operand values, they can be found as the
8474 elements of @code{recog_data.operand}.
8476 If the macro definition does nothing, the instruction is output
8480 @defmac FINAL_PRESCAN_INSN (@var{insn}, @var{opvec}, @var{noperands})
8481 If defined, a C statement to be executed just prior to the output of
8482 assembler code for @var{insn}, to modify the extracted operands so
8483 they will be output differently.
8485 Here the argument @var{opvec} is the vector containing the operands
8486 extracted from @var{insn}, and @var{noperands} is the number of
8487 elements of the vector which contain meaningful data for this insn.
8488 The contents of this vector are what will be used to convert the insn
8489 template into assembler code, so you can change the assembler output
8490 by changing the contents of the vector.
8492 This macro is useful when various assembler syntaxes share a single
8493 file of instruction patterns; by defining this macro differently, you
8494 can cause a large class of instructions to be output differently (such
8495 as with rearranged operands). Naturally, variations in assembler
8496 syntax affecting individual insn patterns ought to be handled by
8497 writing conditional output routines in those patterns.
8499 If this macro is not defined, it is equivalent to a null statement.
8502 @hook TARGET_ASM_FINAL_POSTSCAN_INSN
8503 If defined, this target hook is a function which is executed just after the
8504 output of assembler code for @var{insn}, to change the mode of the assembler
8507 Here the argument @var{opvec} is the vector containing the operands
8508 extracted from @var{insn}, and @var{noperands} is the number of
8509 elements of the vector which contain meaningful data for this insn.
8510 The contents of this vector are what was used to convert the insn
8511 template into assembler code, so you can change the assembler mode
8512 by checking the contents of the vector.
8515 @defmac PRINT_OPERAND (@var{stream}, @var{x}, @var{code})
8516 A C compound statement to output to stdio stream @var{stream} the
8517 assembler syntax for an instruction operand @var{x}. @var{x} is an
8520 @var{code} is a value that can be used to specify one of several ways
8521 of printing the operand. It is used when identical operands must be
8522 printed differently depending on the context. @var{code} comes from
8523 the @samp{%} specification that was used to request printing of the
8524 operand. If the specification was just @samp{%@var{digit}} then
8525 @var{code} is 0; if the specification was @samp{%@var{ltr}
8526 @var{digit}} then @var{code} is the ASCII code for @var{ltr}.
8529 If @var{x} is a register, this macro should print the register's name.
8530 The names can be found in an array @code{reg_names} whose type is
8531 @code{char *[]}. @code{reg_names} is initialized from
8532 @code{REGISTER_NAMES}.
8534 When the machine description has a specification @samp{%@var{punct}}
8535 (a @samp{%} followed by a punctuation character), this macro is called
8536 with a null pointer for @var{x} and the punctuation character for
8540 @defmac PRINT_OPERAND_PUNCT_VALID_P (@var{code})
8541 A C expression which evaluates to true if @var{code} is a valid
8542 punctuation character for use in the @code{PRINT_OPERAND} macro. If
8543 @code{PRINT_OPERAND_PUNCT_VALID_P} is not defined, it means that no
8544 punctuation characters (except for the standard one, @samp{%}) are used
8548 @defmac PRINT_OPERAND_ADDRESS (@var{stream}, @var{x})
8549 A C compound statement to output to stdio stream @var{stream} the
8550 assembler syntax for an instruction operand that is a memory reference
8551 whose address is @var{x}. @var{x} is an RTL expression.
8553 @cindex @code{TARGET_ENCODE_SECTION_INFO} usage
8554 On some machines, the syntax for a symbolic address depends on the
8555 section that the address refers to. On these machines, define the hook
8556 @code{TARGET_ENCODE_SECTION_INFO} to store the information into the
8557 @code{symbol_ref}, and then check for it here. @xref{Assembler
8561 @findex dbr_sequence_length
8562 @defmac DBR_OUTPUT_SEQEND (@var{file})
8563 A C statement, to be executed after all slot-filler instructions have
8564 been output. If necessary, call @code{dbr_sequence_length} to
8565 determine the number of slots filled in a sequence (zero if not
8566 currently outputting a sequence), to decide how many no-ops to output,
8569 Don't define this macro if it has nothing to do, but it is helpful in
8570 reading assembly output if the extent of the delay sequence is made
8571 explicit (e.g.@: with white space).
8574 @findex final_sequence
8575 Note that output routines for instructions with delay slots must be
8576 prepared to deal with not being output as part of a sequence
8577 (i.e.@: when the scheduling pass is not run, or when no slot fillers could be
8578 found.) The variable @code{final_sequence} is null when not
8579 processing a sequence, otherwise it contains the @code{sequence} rtx
8583 @defmac REGISTER_PREFIX
8584 @defmacx LOCAL_LABEL_PREFIX
8585 @defmacx USER_LABEL_PREFIX
8586 @defmacx IMMEDIATE_PREFIX
8587 If defined, C string expressions to be used for the @samp{%R}, @samp{%L},
8588 @samp{%U}, and @samp{%I} options of @code{asm_fprintf} (see
8589 @file{final.c}). These are useful when a single @file{md} file must
8590 support multiple assembler formats. In that case, the various @file{tm.h}
8591 files can define these macros differently.
8594 @defmac ASM_FPRINTF_EXTENSIONS (@var{file}, @var{argptr}, @var{format})
8595 If defined this macro should expand to a series of @code{case}
8596 statements which will be parsed inside the @code{switch} statement of
8597 the @code{asm_fprintf} function. This allows targets to define extra
8598 printf formats which may useful when generating their assembler
8599 statements. Note that uppercase letters are reserved for future
8600 generic extensions to asm_fprintf, and so are not available to target
8601 specific code. The output file is given by the parameter @var{file}.
8602 The varargs input pointer is @var{argptr} and the rest of the format
8603 string, starting the character after the one that is being switched
8604 upon, is pointed to by @var{format}.
8607 @defmac ASSEMBLER_DIALECT
8608 If your target supports multiple dialects of assembler language (such as
8609 different opcodes), define this macro as a C expression that gives the
8610 numeric index of the assembler language dialect to use, with zero as the
8613 If this macro is defined, you may use constructs of the form
8615 @samp{@{option0|option1|option2@dots{}@}}
8618 in the output templates of patterns (@pxref{Output Template}) or in the
8619 first argument of @code{asm_fprintf}. This construct outputs
8620 @samp{option0}, @samp{option1}, @samp{option2}, etc., if the value of
8621 @code{ASSEMBLER_DIALECT} is zero, one, two, etc. Any special characters
8622 within these strings retain their usual meaning. If there are fewer
8623 alternatives within the braces than the value of
8624 @code{ASSEMBLER_DIALECT}, the construct outputs nothing.
8626 If you do not define this macro, the characters @samp{@{}, @samp{|} and
8627 @samp{@}} do not have any special meaning when used in templates or
8628 operands to @code{asm_fprintf}.
8630 Define the macros @code{REGISTER_PREFIX}, @code{LOCAL_LABEL_PREFIX},
8631 @code{USER_LABEL_PREFIX} and @code{IMMEDIATE_PREFIX} if you can express
8632 the variations in assembler language syntax with that mechanism. Define
8633 @code{ASSEMBLER_DIALECT} and use the @samp{@{option0|option1@}} syntax
8634 if the syntax variant are larger and involve such things as different
8635 opcodes or operand order.
8638 @defmac ASM_OUTPUT_REG_PUSH (@var{stream}, @var{regno})
8639 A C expression to output to @var{stream} some assembler code
8640 which will push hard register number @var{regno} onto the stack.
8641 The code need not be optimal, since this macro is used only when
8645 @defmac ASM_OUTPUT_REG_POP (@var{stream}, @var{regno})
8646 A C expression to output to @var{stream} some assembler code
8647 which will pop hard register number @var{regno} off of the stack.
8648 The code need not be optimal, since this macro is used only when
8652 @node Dispatch Tables
8653 @subsection Output of Dispatch Tables
8655 @c prevent bad page break with this line
8656 This concerns dispatch tables.
8658 @cindex dispatch table
8659 @defmac ASM_OUTPUT_ADDR_DIFF_ELT (@var{stream}, @var{body}, @var{value}, @var{rel})
8660 A C statement to output to the stdio stream @var{stream} an assembler
8661 pseudo-instruction to generate a difference between two labels.
8662 @var{value} and @var{rel} are the numbers of two internal labels. The
8663 definitions of these labels are output using
8664 @code{(*targetm.asm_out.internal_label)}, and they must be printed in the same
8665 way here. For example,
8668 fprintf (@var{stream}, "\t.word L%d-L%d\n",
8669 @var{value}, @var{rel})
8672 You must provide this macro on machines where the addresses in a
8673 dispatch table are relative to the table's own address. If defined, GCC
8674 will also use this macro on all machines when producing PIC@.
8675 @var{body} is the body of the @code{ADDR_DIFF_VEC}; it is provided so that the
8676 mode and flags can be read.
8679 @defmac ASM_OUTPUT_ADDR_VEC_ELT (@var{stream}, @var{value})
8680 This macro should be provided on machines where the addresses
8681 in a dispatch table are absolute.
8683 The definition should be a C statement to output to the stdio stream
8684 @var{stream} an assembler pseudo-instruction to generate a reference to
8685 a label. @var{value} is the number of an internal label whose
8686 definition is output using @code{(*targetm.asm_out.internal_label)}.
8690 fprintf (@var{stream}, "\t.word L%d\n", @var{value})
8694 @defmac ASM_OUTPUT_CASE_LABEL (@var{stream}, @var{prefix}, @var{num}, @var{table})
8695 Define this if the label before a jump-table needs to be output
8696 specially. The first three arguments are the same as for
8697 @code{(*targetm.asm_out.internal_label)}; the fourth argument is the
8698 jump-table which follows (a @code{jump_insn} containing an
8699 @code{addr_vec} or @code{addr_diff_vec}).
8701 This feature is used on system V to output a @code{swbeg} statement
8704 If this macro is not defined, these labels are output with
8705 @code{(*targetm.asm_out.internal_label)}.
8708 @defmac ASM_OUTPUT_CASE_END (@var{stream}, @var{num}, @var{table})
8709 Define this if something special must be output at the end of a
8710 jump-table. The definition should be a C statement to be executed
8711 after the assembler code for the table is written. It should write
8712 the appropriate code to stdio stream @var{stream}. The argument
8713 @var{table} is the jump-table insn, and @var{num} is the label-number
8714 of the preceding label.
8716 If this macro is not defined, nothing special is output at the end of
8720 @hook TARGET_ASM_EMIT_UNWIND_LABEL
8721 This target hook emits a label at the beginning of each FDE@. It
8722 should be defined on targets where FDEs need special labels, and it
8723 should write the appropriate label, for the FDE associated with the
8724 function declaration @var{decl}, to the stdio stream @var{stream}.
8725 The third argument, @var{for_eh}, is a boolean: true if this is for an
8726 exception table. The fourth argument, @var{empty}, is a boolean:
8727 true if this is a placeholder label for an omitted FDE@.
8729 The default is that FDEs are not given nonlocal labels.
8732 @hook TARGET_ASM_EMIT_EXCEPT_TABLE_LABEL
8733 This target hook emits a label at the beginning of the exception table.
8734 It should be defined on targets where it is desirable for the table
8735 to be broken up according to function.
8737 The default is that no label is emitted.
8740 @hook TARGET_ASM_EMIT_EXCEPT_PERSONALITY
8742 @hook TARGET_ASM_UNWIND_EMIT
8743 This target hook emits assembly directives required to unwind the
8744 given instruction. This is only used when @code{TARGET_EXCEPT_UNWIND_INFO}
8745 returns @code{UI_TARGET}.
8748 @hook TARGET_ASM_UNWIND_EMIT_BEFORE_INSN
8750 @node Exception Region Output
8751 @subsection Assembler Commands for Exception Regions
8753 @c prevent bad page break with this line
8755 This describes commands marking the start and the end of an exception
8758 @defmac EH_FRAME_SECTION_NAME
8759 If defined, a C string constant for the name of the section containing
8760 exception handling frame unwind information. If not defined, GCC will
8761 provide a default definition if the target supports named sections.
8762 @file{crtstuff.c} uses this macro to switch to the appropriate section.
8764 You should define this symbol if your target supports DWARF 2 frame
8765 unwind information and the default definition does not work.
8768 @defmac EH_FRAME_IN_DATA_SECTION
8769 If defined, DWARF 2 frame unwind information will be placed in the
8770 data section even though the target supports named sections. This
8771 might be necessary, for instance, if the system linker does garbage
8772 collection and sections cannot be marked as not to be collected.
8774 Do not define this macro unless @code{TARGET_ASM_NAMED_SECTION} is
8778 @defmac EH_TABLES_CAN_BE_READ_ONLY
8779 Define this macro to 1 if your target is such that no frame unwind
8780 information encoding used with non-PIC code will ever require a
8781 runtime relocation, but the linker may not support merging read-only
8782 and read-write sections into a single read-write section.
8785 @defmac MASK_RETURN_ADDR
8786 An rtx used to mask the return address found via @code{RETURN_ADDR_RTX}, so
8787 that it does not contain any extraneous set bits in it.
8790 @defmac DWARF2_UNWIND_INFO
8791 Define this macro to 0 if your target supports DWARF 2 frame unwind
8792 information, but it does not yet work with exception handling.
8793 Otherwise, if your target supports this information (if it defines
8794 @code{INCOMING_RETURN_ADDR_RTX} and @code{OBJECT_FORMAT_ELF}),
8795 GCC will provide a default definition of 1.
8798 @hook TARGET_EXCEPT_UNWIND_INFO
8799 This hook defines the mechanism that will be used for exception handling
8800 by the target. If the target has ABI specified unwind tables, the hook
8801 should return @code{UI_TARGET}. If the target is to use the
8802 @code{setjmp}/@code{longjmp}-based exception handling scheme, the hook
8803 should return @code{UI_SJLJ}. If the target supports DWARF 2 frame unwind
8804 information, the hook should return @code{UI_DWARF2}.
8806 A target may, if exceptions are disabled, choose to return @code{UI_NONE}.
8807 This may end up simplifying other parts of target-specific code. The
8808 default implementation of this hook never returns @code{UI_NONE}.
8810 Note that the value returned by this hook should be constant. It should
8811 not depend on anything except the command-line switches described by
8812 @var{opts}. In particular, the
8813 setting @code{UI_SJLJ} must be fixed at compiler start-up as C pre-processor
8814 macros and builtin functions related to exception handling are set up
8815 depending on this setting.
8817 The default implementation of the hook first honors the
8818 @option{--enable-sjlj-exceptions} configure option, then
8819 @code{DWARF2_UNWIND_INFO}, and finally defaults to @code{UI_SJLJ}. If
8820 @code{DWARF2_UNWIND_INFO} depends on command-line options, the target
8821 must define this hook so that @var{opts} is used correctly.
8824 @hook TARGET_UNWIND_TABLES_DEFAULT
8825 This variable should be set to @code{true} if the target ABI requires unwinding
8826 tables even when exceptions are not used. It must not be modified by
8827 command-line option processing.
8830 @defmac DONT_USE_BUILTIN_SETJMP
8831 Define this macro to 1 if the @code{setjmp}/@code{longjmp}-based scheme
8832 should use the @code{setjmp}/@code{longjmp} functions from the C library
8833 instead of the @code{__builtin_setjmp}/@code{__builtin_longjmp} machinery.
8836 @defmac JMP_BUF_SIZE
8837 This macro has no effect unless @code{DONT_USE_BUILTIN_SETJMP} is also
8838 defined. Define this macro if the default size of @code{jmp_buf} buffer
8839 for the @code{setjmp}/@code{longjmp}-based exception handling mechanism
8840 is not large enough, or if it is much too large.
8841 The default size is @code{FIRST_PSEUDO_REGISTER * sizeof(void *)}.
8844 @defmac DWARF_CIE_DATA_ALIGNMENT
8845 This macro need only be defined if the target might save registers in the
8846 function prologue at an offset to the stack pointer that is not aligned to
8847 @code{UNITS_PER_WORD}. The definition should be the negative minimum
8848 alignment if @code{STACK_GROWS_DOWNWARD} is defined, and the positive
8849 minimum alignment otherwise. @xref{SDB and DWARF}. Only applicable if
8850 the target supports DWARF 2 frame unwind information.
8853 @hook TARGET_TERMINATE_DW2_EH_FRAME_INFO
8854 Contains the value true if the target should add a zero word onto the
8855 end of a Dwarf-2 frame info section when used for exception handling.
8856 Default value is false if @code{EH_FRAME_SECTION_NAME} is defined, and
8860 @hook TARGET_DWARF_REGISTER_SPAN
8861 Given a register, this hook should return a parallel of registers to
8862 represent where to find the register pieces. Define this hook if the
8863 register and its mode are represented in Dwarf in non-contiguous
8864 locations, or if the register should be represented in more than one
8865 register in Dwarf. Otherwise, this hook should return @code{NULL_RTX}.
8866 If not defined, the default is to return @code{NULL_RTX}.
8869 @hook TARGET_INIT_DWARF_REG_SIZES_EXTRA
8870 If some registers are represented in Dwarf-2 unwind information in
8871 multiple pieces, define this hook to fill in information about the
8872 sizes of those pieces in the table used by the unwinder at runtime.
8873 It will be called by @code{expand_builtin_init_dwarf_reg_sizes} after
8874 filling in a single size corresponding to each hard register;
8875 @var{address} is the address of the table.
8878 @hook TARGET_ASM_TTYPE
8879 This hook is used to output a reference from a frame unwinding table to
8880 the type_info object identified by @var{sym}. It should return @code{true}
8881 if the reference was output. Returning @code{false} will cause the
8882 reference to be output using the normal Dwarf2 routines.
8885 @hook TARGET_ARM_EABI_UNWINDER
8886 This flag should be set to @code{true} on targets that use an ARM EABI
8887 based unwinding library, and @code{false} on other targets. This effects
8888 the format of unwinding tables, and how the unwinder in entered after
8889 running a cleanup. The default is @code{false}.
8892 @node Alignment Output
8893 @subsection Assembler Commands for Alignment
8895 @c prevent bad page break with this line
8896 This describes commands for alignment.
8898 @defmac JUMP_ALIGN (@var{label})
8899 The alignment (log base 2) to put in front of @var{label}, which is
8900 a common destination of jumps and has no fallthru incoming edge.
8902 This macro need not be defined if you don't want any special alignment
8903 to be done at such a time. Most machine descriptions do not currently
8906 Unless it's necessary to inspect the @var{label} parameter, it is better
8907 to set the variable @var{align_jumps} in the target's
8908 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
8909 selection in @var{align_jumps} in a @code{JUMP_ALIGN} implementation.
8912 @hook TARGET_ASM_JUMP_ALIGN_MAX_SKIP
8913 The maximum number of bytes to skip before @var{label} when applying
8914 @code{JUMP_ALIGN}. This works only if
8915 @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
8918 @defmac LABEL_ALIGN_AFTER_BARRIER (@var{label})
8919 The alignment (log base 2) to put in front of @var{label}, which follows
8922 This macro need not be defined if you don't want any special alignment
8923 to be done at such a time. Most machine descriptions do not currently
8927 @hook TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP
8928 The maximum number of bytes to skip before @var{label} when applying
8929 @code{LABEL_ALIGN_AFTER_BARRIER}. This works only if
8930 @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
8933 @defmac LOOP_ALIGN (@var{label})
8934 The alignment (log base 2) to put in front of @var{label}, which follows
8935 a @code{NOTE_INSN_LOOP_BEG} note.
8937 This macro need not be defined if you don't want any special alignment
8938 to be done at such a time. Most machine descriptions do not currently
8941 Unless it's necessary to inspect the @var{label} parameter, it is better
8942 to set the variable @code{align_loops} in the target's
8943 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
8944 selection in @code{align_loops} in a @code{LOOP_ALIGN} implementation.
8947 @hook TARGET_ASM_LOOP_ALIGN_MAX_SKIP
8948 The maximum number of bytes to skip when applying @code{LOOP_ALIGN} to
8949 @var{label}. This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is
8953 @defmac LABEL_ALIGN (@var{label})
8954 The alignment (log base 2) to put in front of @var{label}.
8955 If @code{LABEL_ALIGN_AFTER_BARRIER} / @code{LOOP_ALIGN} specify a different alignment,
8956 the maximum of the specified values is used.
8958 Unless it's necessary to inspect the @var{label} parameter, it is better
8959 to set the variable @code{align_labels} in the target's
8960 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
8961 selection in @code{align_labels} in a @code{LABEL_ALIGN} implementation.
8964 @hook TARGET_ASM_LABEL_ALIGN_MAX_SKIP
8965 The maximum number of bytes to skip when applying @code{LABEL_ALIGN}
8966 to @var{label}. This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN}
8970 @defmac ASM_OUTPUT_SKIP (@var{stream}, @var{nbytes})
8971 A C statement to output to the stdio stream @var{stream} an assembler
8972 instruction to advance the location counter by @var{nbytes} bytes.
8973 Those bytes should be zero when loaded. @var{nbytes} will be a C
8974 expression of type @code{unsigned HOST_WIDE_INT}.
8977 @defmac ASM_NO_SKIP_IN_TEXT
8978 Define this macro if @code{ASM_OUTPUT_SKIP} should not be used in the
8979 text section because it fails to put zeros in the bytes that are skipped.
8980 This is true on many Unix systems, where the pseudo--op to skip bytes
8981 produces no-op instructions rather than zeros when used in the text
8985 @defmac ASM_OUTPUT_ALIGN (@var{stream}, @var{power})
8986 A C statement to output to the stdio stream @var{stream} an assembler
8987 command to advance the location counter to a multiple of 2 to the
8988 @var{power} bytes. @var{power} will be a C expression of type @code{int}.
8991 @defmac ASM_OUTPUT_ALIGN_WITH_NOP (@var{stream}, @var{power})
8992 Like @code{ASM_OUTPUT_ALIGN}, except that the ``nop'' instruction is used
8993 for padding, if necessary.
8996 @defmac ASM_OUTPUT_MAX_SKIP_ALIGN (@var{stream}, @var{power}, @var{max_skip})
8997 A C statement to output to the stdio stream @var{stream} an assembler
8998 command to advance the location counter to a multiple of 2 to the
8999 @var{power} bytes, but only if @var{max_skip} or fewer bytes are needed to
9000 satisfy the alignment request. @var{power} and @var{max_skip} will be
9001 a C expression of type @code{int}.
9005 @node Debugging Info
9006 @section Controlling Debugging Information Format
9008 @c prevent bad page break with this line
9009 This describes how to specify debugging information.
9012 * All Debuggers:: Macros that affect all debugging formats uniformly.
9013 * DBX Options:: Macros enabling specific options in DBX format.
9014 * DBX Hooks:: Hook macros for varying DBX format.
9015 * File Names and DBX:: Macros controlling output of file names in DBX format.
9016 * SDB and DWARF:: Macros for SDB (COFF) and DWARF formats.
9017 * VMS Debug:: Macros for VMS debug format.
9021 @subsection Macros Affecting All Debugging Formats
9023 @c prevent bad page break with this line
9024 These macros affect all debugging formats.
9026 @defmac DBX_REGISTER_NUMBER (@var{regno})
9027 A C expression that returns the DBX register number for the compiler
9028 register number @var{regno}. In the default macro provided, the value
9029 of this expression will be @var{regno} itself. But sometimes there are
9030 some registers that the compiler knows about and DBX does not, or vice
9031 versa. In such cases, some register may need to have one number in the
9032 compiler and another for DBX@.
9034 If two registers have consecutive numbers inside GCC, and they can be
9035 used as a pair to hold a multiword value, then they @emph{must} have
9036 consecutive numbers after renumbering with @code{DBX_REGISTER_NUMBER}.
9037 Otherwise, debuggers will be unable to access such a pair, because they
9038 expect register pairs to be consecutive in their own numbering scheme.
9040 If you find yourself defining @code{DBX_REGISTER_NUMBER} in way that
9041 does not preserve register pairs, then what you must do instead is
9042 redefine the actual register numbering scheme.
9045 @defmac DEBUGGER_AUTO_OFFSET (@var{x})
9046 A C expression that returns the integer offset value for an automatic
9047 variable having address @var{x} (an RTL expression). The default
9048 computation assumes that @var{x} is based on the frame-pointer and
9049 gives the offset from the frame-pointer. This is required for targets
9050 that produce debugging output for DBX or COFF-style debugging output
9051 for SDB and allow the frame-pointer to be eliminated when the
9052 @option{-g} options is used.
9055 @defmac DEBUGGER_ARG_OFFSET (@var{offset}, @var{x})
9056 A C expression that returns the integer offset value for an argument
9057 having address @var{x} (an RTL expression). The nominal offset is
9061 @defmac PREFERRED_DEBUGGING_TYPE
9062 A C expression that returns the type of debugging output GCC should
9063 produce when the user specifies just @option{-g}. Define
9064 this if you have arranged for GCC to support more than one format of
9065 debugging output. Currently, the allowable values are @code{DBX_DEBUG},
9066 @code{SDB_DEBUG}, @code{DWARF_DEBUG}, @code{DWARF2_DEBUG},
9067 @code{XCOFF_DEBUG}, @code{VMS_DEBUG}, and @code{VMS_AND_DWARF2_DEBUG}.
9069 When the user specifies @option{-ggdb}, GCC normally also uses the
9070 value of this macro to select the debugging output format, but with two
9071 exceptions. If @code{DWARF2_DEBUGGING_INFO} is defined, GCC uses the
9072 value @code{DWARF2_DEBUG}. Otherwise, if @code{DBX_DEBUGGING_INFO} is
9073 defined, GCC uses @code{DBX_DEBUG}.
9075 The value of this macro only affects the default debugging output; the
9076 user can always get a specific type of output by using @option{-gstabs},
9077 @option{-gcoff}, @option{-gdwarf-2}, @option{-gxcoff}, or @option{-gvms}.
9081 @subsection Specific Options for DBX Output
9083 @c prevent bad page break with this line
9084 These are specific options for DBX output.
9086 @defmac DBX_DEBUGGING_INFO
9087 Define this macro if GCC should produce debugging output for DBX
9088 in response to the @option{-g} option.
9091 @defmac XCOFF_DEBUGGING_INFO
9092 Define this macro if GCC should produce XCOFF format debugging output
9093 in response to the @option{-g} option. This is a variant of DBX format.
9096 @defmac DEFAULT_GDB_EXTENSIONS
9097 Define this macro to control whether GCC should by default generate
9098 GDB's extended version of DBX debugging information (assuming DBX-format
9099 debugging information is enabled at all). If you don't define the
9100 macro, the default is 1: always generate the extended information
9101 if there is any occasion to.
9104 @defmac DEBUG_SYMS_TEXT
9105 Define this macro if all @code{.stabs} commands should be output while
9106 in the text section.
9109 @defmac ASM_STABS_OP
9110 A C string constant, including spacing, naming the assembler pseudo op to
9111 use instead of @code{"\t.stabs\t"} to define an ordinary debugging symbol.
9112 If you don't define this macro, @code{"\t.stabs\t"} is used. This macro
9113 applies only to DBX debugging information format.
9116 @defmac ASM_STABD_OP
9117 A C string constant, including spacing, naming the assembler pseudo op to
9118 use instead of @code{"\t.stabd\t"} to define a debugging symbol whose
9119 value is the current location. If you don't define this macro,
9120 @code{"\t.stabd\t"} is used. This macro applies only to DBX debugging
9124 @defmac ASM_STABN_OP
9125 A C string constant, including spacing, naming the assembler pseudo op to
9126 use instead of @code{"\t.stabn\t"} to define a debugging symbol with no
9127 name. If you don't define this macro, @code{"\t.stabn\t"} is used. This
9128 macro applies only to DBX debugging information format.
9131 @defmac DBX_NO_XREFS
9132 Define this macro if DBX on your system does not support the construct
9133 @samp{xs@var{tagname}}. On some systems, this construct is used to
9134 describe a forward reference to a structure named @var{tagname}.
9135 On other systems, this construct is not supported at all.
9138 @defmac DBX_CONTIN_LENGTH
9139 A symbol name in DBX-format debugging information is normally
9140 continued (split into two separate @code{.stabs} directives) when it
9141 exceeds a certain length (by default, 80 characters). On some
9142 operating systems, DBX requires this splitting; on others, splitting
9143 must not be done. You can inhibit splitting by defining this macro
9144 with the value zero. You can override the default splitting-length by
9145 defining this macro as an expression for the length you desire.
9148 @defmac DBX_CONTIN_CHAR
9149 Normally continuation is indicated by adding a @samp{\} character to
9150 the end of a @code{.stabs} string when a continuation follows. To use
9151 a different character instead, define this macro as a character
9152 constant for the character you want to use. Do not define this macro
9153 if backslash is correct for your system.
9156 @defmac DBX_STATIC_STAB_DATA_SECTION
9157 Define this macro if it is necessary to go to the data section before
9158 outputting the @samp{.stabs} pseudo-op for a non-global static
9162 @defmac DBX_TYPE_DECL_STABS_CODE
9163 The value to use in the ``code'' field of the @code{.stabs} directive
9164 for a typedef. The default is @code{N_LSYM}.
9167 @defmac DBX_STATIC_CONST_VAR_CODE
9168 The value to use in the ``code'' field of the @code{.stabs} directive
9169 for a static variable located in the text section. DBX format does not
9170 provide any ``right'' way to do this. The default is @code{N_FUN}.
9173 @defmac DBX_REGPARM_STABS_CODE
9174 The value to use in the ``code'' field of the @code{.stabs} directive
9175 for a parameter passed in registers. DBX format does not provide any
9176 ``right'' way to do this. The default is @code{N_RSYM}.
9179 @defmac DBX_REGPARM_STABS_LETTER
9180 The letter to use in DBX symbol data to identify a symbol as a parameter
9181 passed in registers. DBX format does not customarily provide any way to
9182 do this. The default is @code{'P'}.
9185 @defmac DBX_FUNCTION_FIRST
9186 Define this macro if the DBX information for a function and its
9187 arguments should precede the assembler code for the function. Normally,
9188 in DBX format, the debugging information entirely follows the assembler
9192 @defmac DBX_BLOCKS_FUNCTION_RELATIVE
9193 Define this macro, with value 1, if the value of a symbol describing
9194 the scope of a block (@code{N_LBRAC} or @code{N_RBRAC}) should be
9195 relative to the start of the enclosing function. Normally, GCC uses
9196 an absolute address.
9199 @defmac DBX_LINES_FUNCTION_RELATIVE
9200 Define this macro, with value 1, if the value of a symbol indicating
9201 the current line number (@code{N_SLINE}) should be relative to the
9202 start of the enclosing function. Normally, GCC uses an absolute address.
9205 @defmac DBX_USE_BINCL
9206 Define this macro if GCC should generate @code{N_BINCL} and
9207 @code{N_EINCL} stabs for included header files, as on Sun systems. This
9208 macro also directs GCC to output a type number as a pair of a file
9209 number and a type number within the file. Normally, GCC does not
9210 generate @code{N_BINCL} or @code{N_EINCL} stabs, and it outputs a single
9211 number for a type number.
9215 @subsection Open-Ended Hooks for DBX Format
9217 @c prevent bad page break with this line
9218 These are hooks for DBX format.
9220 @defmac DBX_OUTPUT_SOURCE_LINE (@var{stream}, @var{line}, @var{counter})
9221 A C statement to output DBX debugging information before code for line
9222 number @var{line} of the current source file to the stdio stream
9223 @var{stream}. @var{counter} is the number of time the macro was
9224 invoked, including the current invocation; it is intended to generate
9225 unique labels in the assembly output.
9227 This macro should not be defined if the default output is correct, or
9228 if it can be made correct by defining @code{DBX_LINES_FUNCTION_RELATIVE}.
9231 @defmac NO_DBX_FUNCTION_END
9232 Some stabs encapsulation formats (in particular ECOFF), cannot handle the
9233 @code{.stabs "",N_FUN,,0,0,Lscope-function-1} gdb dbx extension construct.
9234 On those machines, define this macro to turn this feature off without
9235 disturbing the rest of the gdb extensions.
9238 @defmac NO_DBX_BNSYM_ENSYM
9239 Some assemblers cannot handle the @code{.stabd BNSYM/ENSYM,0,0} gdb dbx
9240 extension construct. On those machines, define this macro to turn this
9241 feature off without disturbing the rest of the gdb extensions.
9244 @node File Names and DBX
9245 @subsection File Names in DBX Format
9247 @c prevent bad page break with this line
9248 This describes file names in DBX format.
9250 @defmac DBX_OUTPUT_MAIN_SOURCE_FILENAME (@var{stream}, @var{name})
9251 A C statement to output DBX debugging information to the stdio stream
9252 @var{stream}, which indicates that file @var{name} is the main source
9253 file---the file specified as the input file for compilation.
9254 This macro is called only once, at the beginning of compilation.
9256 This macro need not be defined if the standard form of output
9257 for DBX debugging information is appropriate.
9259 It may be necessary to refer to a label equal to the beginning of the
9260 text section. You can use @samp{assemble_name (stream, ltext_label_name)}
9261 to do so. If you do this, you must also set the variable
9262 @var{used_ltext_label_name} to @code{true}.
9265 @defmac NO_DBX_MAIN_SOURCE_DIRECTORY
9266 Define this macro, with value 1, if GCC should not emit an indication
9267 of the current directory for compilation and current source language at
9268 the beginning of the file.
9271 @defmac NO_DBX_GCC_MARKER
9272 Define this macro, with value 1, if GCC should not emit an indication
9273 that this object file was compiled by GCC@. The default is to emit
9274 an @code{N_OPT} stab at the beginning of every source file, with
9275 @samp{gcc2_compiled.} for the string and value 0.
9278 @defmac DBX_OUTPUT_MAIN_SOURCE_FILE_END (@var{stream}, @var{name})
9279 A C statement to output DBX debugging information at the end of
9280 compilation of the main source file @var{name}. Output should be
9281 written to the stdio stream @var{stream}.
9283 If you don't define this macro, nothing special is output at the end
9284 of compilation, which is correct for most machines.
9287 @defmac DBX_OUTPUT_NULL_N_SO_AT_MAIN_SOURCE_FILE_END
9288 Define this macro @emph{instead of} defining
9289 @code{DBX_OUTPUT_MAIN_SOURCE_FILE_END}, if what needs to be output at
9290 the end of compilation is an @code{N_SO} stab with an empty string,
9291 whose value is the highest absolute text address in the file.
9296 @subsection Macros for SDB and DWARF Output
9298 @c prevent bad page break with this line
9299 Here are macros for SDB and DWARF output.
9301 @defmac SDB_DEBUGGING_INFO
9302 Define this macro if GCC should produce COFF-style debugging output
9303 for SDB in response to the @option{-g} option.
9306 @defmac DWARF2_DEBUGGING_INFO
9307 Define this macro if GCC should produce dwarf version 2 format
9308 debugging output in response to the @option{-g} option.
9310 @hook TARGET_DWARF_CALLING_CONVENTION
9311 Define this to enable the dwarf attribute @code{DW_AT_calling_convention} to
9312 be emitted for each function. Instead of an integer return the enum
9313 value for the @code{DW_CC_} tag.
9316 To support optional call frame debugging information, you must also
9317 define @code{INCOMING_RETURN_ADDR_RTX} and either set
9318 @code{RTX_FRAME_RELATED_P} on the prologue insns if you use RTL for the
9319 prologue, or call @code{dwarf2out_def_cfa} and @code{dwarf2out_reg_save}
9320 as appropriate from @code{TARGET_ASM_FUNCTION_PROLOGUE} if you don't.
9323 @defmac DWARF2_FRAME_INFO
9324 Define this macro to a nonzero value if GCC should always output
9325 Dwarf 2 frame information. If @code{TARGET_EXCEPT_UNWIND_INFO}
9326 (@pxref{Exception Region Output}) returns @code{UI_DWARF2}, and
9327 exceptions are enabled, GCC will output this information not matter
9328 how you define @code{DWARF2_FRAME_INFO}.
9331 @hook TARGET_DEBUG_UNWIND_INFO
9332 This hook defines the mechanism that will be used for describing frame
9333 unwind information to the debugger. Normally the hook will return
9334 @code{UI_DWARF2} if DWARF 2 debug information is enabled, and
9335 return @code{UI_NONE} otherwise.
9337 A target may return @code{UI_DWARF2} even when DWARF 2 debug information
9338 is disabled in order to always output DWARF 2 frame information.
9340 A target may return @code{UI_TARGET} if it has ABI specified unwind tables.
9341 This will suppress generation of the normal debug frame unwind information.
9344 @defmac DWARF2_ASM_LINE_DEBUG_INFO
9345 Define this macro to be a nonzero value if the assembler can generate Dwarf 2
9346 line debug info sections. This will result in much more compact line number
9347 tables, and hence is desirable if it works.
9350 @hook TARGET_WANT_DEBUG_PUB_SECTIONS
9352 @hook TARGET_FORCE_AT_COMP_DIR
9354 @hook TARGET_DELAY_SCHED2
9356 @hook TARGET_DELAY_VARTRACK
9358 @defmac ASM_OUTPUT_DWARF_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
9359 A C statement to issue assembly directives that create a difference
9360 @var{lab1} minus @var{lab2}, using an integer of the given @var{size}.
9363 @defmac ASM_OUTPUT_DWARF_VMS_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
9364 A C statement to issue assembly directives that create a difference
9365 between the two given labels in system defined units, e.g. instruction
9366 slots on IA64 VMS, using an integer of the given size.
9369 @defmac ASM_OUTPUT_DWARF_OFFSET (@var{stream}, @var{size}, @var{label}, @var{section})
9370 A C statement to issue assembly directives that create a
9371 section-relative reference to the given @var{label}, using an integer of the
9372 given @var{size}. The label is known to be defined in the given @var{section}.
9375 @defmac ASM_OUTPUT_DWARF_PCREL (@var{stream}, @var{size}, @var{label})
9376 A C statement to issue assembly directives that create a self-relative
9377 reference to the given @var{label}, using an integer of the given @var{size}.
9380 @defmac ASM_OUTPUT_DWARF_TABLE_REF (@var{label})
9381 A C statement to issue assembly directives that create a reference to
9382 the DWARF table identifier @var{label} from the current section. This
9383 is used on some systems to avoid garbage collecting a DWARF table which
9384 is referenced by a function.
9387 @hook TARGET_ASM_OUTPUT_DWARF_DTPREL
9388 If defined, this target hook is a function which outputs a DTP-relative
9389 reference to the given TLS symbol of the specified size.
9392 @defmac PUT_SDB_@dots{}
9393 Define these macros to override the assembler syntax for the special
9394 SDB assembler directives. See @file{sdbout.c} for a list of these
9395 macros and their arguments. If the standard syntax is used, you need
9396 not define them yourself.
9400 Some assemblers do not support a semicolon as a delimiter, even between
9401 SDB assembler directives. In that case, define this macro to be the
9402 delimiter to use (usually @samp{\n}). It is not necessary to define
9403 a new set of @code{PUT_SDB_@var{op}} macros if this is the only change
9407 @defmac SDB_ALLOW_UNKNOWN_REFERENCES
9408 Define this macro to allow references to unknown structure,
9409 union, or enumeration tags to be emitted. Standard COFF does not
9410 allow handling of unknown references, MIPS ECOFF has support for
9414 @defmac SDB_ALLOW_FORWARD_REFERENCES
9415 Define this macro to allow references to structure, union, or
9416 enumeration tags that have not yet been seen to be handled. Some
9417 assemblers choke if forward tags are used, while some require it.
9420 @defmac SDB_OUTPUT_SOURCE_LINE (@var{stream}, @var{line})
9421 A C statement to output SDB debugging information before code for line
9422 number @var{line} of the current source file to the stdio stream
9423 @var{stream}. The default is to emit an @code{.ln} directive.
9428 @subsection Macros for VMS Debug Format
9430 @c prevent bad page break with this line
9431 Here are macros for VMS debug format.
9433 @defmac VMS_DEBUGGING_INFO
9434 Define this macro if GCC should produce debugging output for VMS
9435 in response to the @option{-g} option. The default behavior for VMS
9436 is to generate minimal debug info for a traceback in the absence of
9437 @option{-g} unless explicitly overridden with @option{-g0}. This
9438 behavior is controlled by @code{TARGET_OPTION_OPTIMIZATION} and
9439 @code{TARGET_OPTION_OVERRIDE}.
9442 @node Floating Point
9443 @section Cross Compilation and Floating Point
9444 @cindex cross compilation and floating point
9445 @cindex floating point and cross compilation
9447 While all modern machines use twos-complement representation for integers,
9448 there are a variety of representations for floating point numbers. This
9449 means that in a cross-compiler the representation of floating point numbers
9450 in the compiled program may be different from that used in the machine
9451 doing the compilation.
9453 Because different representation systems may offer different amounts of
9454 range and precision, all floating point constants must be represented in
9455 the target machine's format. Therefore, the cross compiler cannot
9456 safely use the host machine's floating point arithmetic; it must emulate
9457 the target's arithmetic. To ensure consistency, GCC always uses
9458 emulation to work with floating point values, even when the host and
9459 target floating point formats are identical.
9461 The following macros are provided by @file{real.h} for the compiler to
9462 use. All parts of the compiler which generate or optimize
9463 floating-point calculations must use these macros. They may evaluate
9464 their operands more than once, so operands must not have side effects.
9466 @defmac REAL_VALUE_TYPE
9467 The C data type to be used to hold a floating point value in the target
9468 machine's format. Typically this is a @code{struct} containing an
9469 array of @code{HOST_WIDE_INT}, but all code should treat it as an opaque
9473 @deftypefn Macro int REAL_VALUES_EQUAL (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
9474 Compares for equality the two values, @var{x} and @var{y}. If the target
9475 floating point format supports negative zeroes and/or NaNs,
9476 @samp{REAL_VALUES_EQUAL (-0.0, 0.0)} is true, and
9477 @samp{REAL_VALUES_EQUAL (NaN, NaN)} is false.
9480 @deftypefn Macro int REAL_VALUES_LESS (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
9481 Tests whether @var{x} is less than @var{y}.
9484 @deftypefn Macro HOST_WIDE_INT REAL_VALUE_FIX (REAL_VALUE_TYPE @var{x})
9485 Truncates @var{x} to a signed integer, rounding toward zero.
9488 @deftypefn Macro {unsigned HOST_WIDE_INT} REAL_VALUE_UNSIGNED_FIX (REAL_VALUE_TYPE @var{x})
9489 Truncates @var{x} to an unsigned integer, rounding toward zero. If
9490 @var{x} is negative, returns zero.
9493 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ATOF (const char *@var{string}, enum machine_mode @var{mode})
9494 Converts @var{string} into a floating point number in the target machine's
9495 representation for mode @var{mode}. This routine can handle both
9496 decimal and hexadecimal floating point constants, using the syntax
9497 defined by the C language for both.
9500 @deftypefn Macro int REAL_VALUE_NEGATIVE (REAL_VALUE_TYPE @var{x})
9501 Returns 1 if @var{x} is negative (including negative zero), 0 otherwise.
9504 @deftypefn Macro int REAL_VALUE_ISINF (REAL_VALUE_TYPE @var{x})
9505 Determines whether @var{x} represents infinity (positive or negative).
9508 @deftypefn Macro int REAL_VALUE_ISNAN (REAL_VALUE_TYPE @var{x})
9509 Determines whether @var{x} represents a ``NaN'' (not-a-number).
9512 @deftypefn Macro void REAL_ARITHMETIC (REAL_VALUE_TYPE @var{output}, enum tree_code @var{code}, REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
9513 Calculates an arithmetic operation on the two floating point values
9514 @var{x} and @var{y}, storing the result in @var{output} (which must be a
9517 The operation to be performed is specified by @var{code}. Only the
9518 following codes are supported: @code{PLUS_EXPR}, @code{MINUS_EXPR},
9519 @code{MULT_EXPR}, @code{RDIV_EXPR}, @code{MAX_EXPR}, @code{MIN_EXPR}.
9521 If @code{REAL_ARITHMETIC} is asked to evaluate division by zero and the
9522 target's floating point format cannot represent infinity, it will call
9523 @code{abort}. Callers should check for this situation first, using
9524 @code{MODE_HAS_INFINITIES}. @xref{Storage Layout}.
9527 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_NEGATE (REAL_VALUE_TYPE @var{x})
9528 Returns the negative of the floating point value @var{x}.
9531 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ABS (REAL_VALUE_TYPE @var{x})
9532 Returns the absolute value of @var{x}.
9535 @deftypefn Macro void REAL_VALUE_TO_INT (HOST_WIDE_INT @var{low}, HOST_WIDE_INT @var{high}, REAL_VALUE_TYPE @var{x})
9536 Converts a floating point value @var{x} into a double-precision integer
9537 which is then stored into @var{low} and @var{high}. If the value is not
9538 integral, it is truncated.
9541 @deftypefn Macro void REAL_VALUE_FROM_INT (REAL_VALUE_TYPE @var{x}, HOST_WIDE_INT @var{low}, HOST_WIDE_INT @var{high}, enum machine_mode @var{mode})
9542 Converts a double-precision integer found in @var{low} and @var{high},
9543 into a floating point value which is then stored into @var{x}. The
9544 value is truncated to fit in mode @var{mode}.
9547 @node Mode Switching
9548 @section Mode Switching Instructions
9549 @cindex mode switching
9550 The following macros control mode switching optimizations:
9552 @defmac OPTIMIZE_MODE_SWITCHING (@var{entity})
9553 Define this macro if the port needs extra instructions inserted for mode
9554 switching in an optimizing compilation.
9556 For an example, the SH4 can perform both single and double precision
9557 floating point operations, but to perform a single precision operation,
9558 the FPSCR PR bit has to be cleared, while for a double precision
9559 operation, this bit has to be set. Changing the PR bit requires a general
9560 purpose register as a scratch register, hence these FPSCR sets have to
9561 be inserted before reload, i.e.@: you can't put this into instruction emitting
9562 or @code{TARGET_MACHINE_DEPENDENT_REORG}.
9564 You can have multiple entities that are mode-switched, and select at run time
9565 which entities actually need it. @code{OPTIMIZE_MODE_SWITCHING} should
9566 return nonzero for any @var{entity} that needs mode-switching.
9567 If you define this macro, you also have to define
9568 @code{NUM_MODES_FOR_MODE_SWITCHING}, @code{MODE_NEEDED},
9569 @code{MODE_PRIORITY_TO_MODE} and @code{EMIT_MODE_SET}.
9570 @code{MODE_AFTER}, @code{MODE_ENTRY}, and @code{MODE_EXIT}
9574 @defmac NUM_MODES_FOR_MODE_SWITCHING
9575 If you define @code{OPTIMIZE_MODE_SWITCHING}, you have to define this as
9576 initializer for an array of integers. Each initializer element
9577 N refers to an entity that needs mode switching, and specifies the number
9578 of different modes that might need to be set for this entity.
9579 The position of the initializer in the initializer---starting counting at
9580 zero---determines the integer that is used to refer to the mode-switched
9582 In macros that take mode arguments / yield a mode result, modes are
9583 represented as numbers 0 @dots{} N @minus{} 1. N is used to specify that no mode
9584 switch is needed / supplied.
9587 @defmac MODE_NEEDED (@var{entity}, @var{insn})
9588 @var{entity} is an integer specifying a mode-switched entity. If
9589 @code{OPTIMIZE_MODE_SWITCHING} is defined, you must define this macro to
9590 return an integer value not larger than the corresponding element in
9591 @code{NUM_MODES_FOR_MODE_SWITCHING}, to denote the mode that @var{entity} must
9592 be switched into prior to the execution of @var{insn}.
9595 @defmac MODE_AFTER (@var{entity}, @var{mode}, @var{insn})
9596 @var{entity} is an integer specifying a mode-switched entity. If
9597 this macro is defined, it is evaluated for every @var{insn} during
9598 mode switching. It determines the mode that an insn results in (if
9599 different from the incoming mode).
9602 @defmac MODE_ENTRY (@var{entity})
9603 If this macro is defined, it is evaluated for every @var{entity} that needs
9604 mode switching. It should evaluate to an integer, which is a mode that
9605 @var{entity} is assumed to be switched to at function entry. If @code{MODE_ENTRY}
9606 is defined then @code{MODE_EXIT} must be defined.
9609 @defmac MODE_EXIT (@var{entity})
9610 If this macro is defined, it is evaluated for every @var{entity} that needs
9611 mode switching. It should evaluate to an integer, which is a mode that
9612 @var{entity} is assumed to be switched to at function exit. If @code{MODE_EXIT}
9613 is defined then @code{MODE_ENTRY} must be defined.
9616 @defmac MODE_PRIORITY_TO_MODE (@var{entity}, @var{n})
9617 This macro specifies the order in which modes for @var{entity} are processed.
9618 0 is the highest priority, @code{NUM_MODES_FOR_MODE_SWITCHING[@var{entity}] - 1} the
9619 lowest. The value of the macro should be an integer designating a mode
9620 for @var{entity}. For any fixed @var{entity}, @code{mode_priority_to_mode}
9621 (@var{entity}, @var{n}) shall be a bijection in 0 @dots{}
9622 @code{num_modes_for_mode_switching[@var{entity}] - 1}.
9625 @defmac EMIT_MODE_SET (@var{entity}, @var{mode}, @var{hard_regs_live})
9626 Generate one or more insns to set @var{entity} to @var{mode}.
9627 @var{hard_reg_live} is the set of hard registers live at the point where
9628 the insn(s) are to be inserted.
9631 @node Target Attributes
9632 @section Defining target-specific uses of @code{__attribute__}
9633 @cindex target attributes
9634 @cindex machine attributes
9635 @cindex attributes, target-specific
9637 Target-specific attributes may be defined for functions, data and types.
9638 These are described using the following target hooks; they also need to
9639 be documented in @file{extend.texi}.
9641 @hook TARGET_ATTRIBUTE_TABLE
9642 If defined, this target hook points to an array of @samp{struct
9643 attribute_spec} (defined in @file{tree.h}) specifying the machine
9644 specific attributes for this target and some of the restrictions on the
9645 entities to which these attributes are applied and the arguments they
9649 @hook TARGET_ATTRIBUTE_TAKES_IDENTIFIER_P
9650 If defined, this target hook is a function which returns true if the
9651 machine-specific attribute named @var{name} expects an identifier
9652 given as its first argument to be passed on as a plain identifier, not
9653 subjected to name lookup. If this is not defined, the default is
9654 false for all machine-specific attributes.
9657 @hook TARGET_COMP_TYPE_ATTRIBUTES
9658 If defined, this target hook is a function which returns zero if the attributes on
9659 @var{type1} and @var{type2} are incompatible, one if they are compatible,
9660 and two if they are nearly compatible (which causes a warning to be
9661 generated). If this is not defined, machine-specific attributes are
9662 supposed always to be compatible.
9665 @hook TARGET_SET_DEFAULT_TYPE_ATTRIBUTES
9666 If defined, this target hook is a function which assigns default attributes to
9667 the newly defined @var{type}.
9670 @hook TARGET_MERGE_TYPE_ATTRIBUTES
9671 Define this target hook if the merging of type attributes needs special
9672 handling. If defined, the result is a list of the combined
9673 @code{TYPE_ATTRIBUTES} of @var{type1} and @var{type2}. It is assumed
9674 that @code{comptypes} has already been called and returned 1. This
9675 function may call @code{merge_attributes} to handle machine-independent
9679 @hook TARGET_MERGE_DECL_ATTRIBUTES
9680 Define this target hook if the merging of decl attributes needs special
9681 handling. If defined, the result is a list of the combined
9682 @code{DECL_ATTRIBUTES} of @var{olddecl} and @var{newdecl}.
9683 @var{newdecl} is a duplicate declaration of @var{olddecl}. Examples of
9684 when this is needed are when one attribute overrides another, or when an
9685 attribute is nullified by a subsequent definition. This function may
9686 call @code{merge_attributes} to handle machine-independent merging.
9688 @findex TARGET_DLLIMPORT_DECL_ATTRIBUTES
9689 If the only target-specific handling you require is @samp{dllimport}
9690 for Microsoft Windows targets, you should define the macro
9691 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES} to @code{1}. The compiler
9692 will then define a function called
9693 @code{merge_dllimport_decl_attributes} which can then be defined as
9694 the expansion of @code{TARGET_MERGE_DECL_ATTRIBUTES}. You can also
9695 add @code{handle_dll_attribute} in the attribute table for your port
9696 to perform initial processing of the @samp{dllimport} and
9697 @samp{dllexport} attributes. This is done in @file{i386/cygwin.h} and
9698 @file{i386/i386.c}, for example.
9701 @hook TARGET_VALID_DLLIMPORT_ATTRIBUTE_P
9703 @defmac TARGET_DECLSPEC
9704 Define this macro to a nonzero value if you want to treat
9705 @code{__declspec(X)} as equivalent to @code{__attribute((X))}. By
9706 default, this behavior is enabled only for targets that define
9707 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES}. The current implementation
9708 of @code{__declspec} is via a built-in macro, but you should not rely
9709 on this implementation detail.
9712 @hook TARGET_INSERT_ATTRIBUTES
9713 Define this target hook if you want to be able to add attributes to a decl
9714 when it is being created. This is normally useful for back ends which
9715 wish to implement a pragma by using the attributes which correspond to
9716 the pragma's effect. The @var{node} argument is the decl which is being
9717 created. The @var{attr_ptr} argument is a pointer to the attribute list
9718 for this decl. The list itself should not be modified, since it may be
9719 shared with other decls, but attributes may be chained on the head of
9720 the list and @code{*@var{attr_ptr}} modified to point to the new
9721 attributes, or a copy of the list may be made if further changes are
9725 @hook TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P
9727 This target hook returns @code{true} if it is ok to inline @var{fndecl}
9728 into the current function, despite its having target-specific
9729 attributes, @code{false} otherwise. By default, if a function has a
9730 target specific attribute attached to it, it will not be inlined.
9733 @hook TARGET_OPTION_VALID_ATTRIBUTE_P
9734 This hook is called to parse the @code{attribute(option("..."))}, and
9735 it allows the function to set different target machine compile time
9736 options for the current function that might be different than the
9737 options specified on the command line. The hook should return
9738 @code{true} if the options are valid.
9740 The hook should set the @var{DECL_FUNCTION_SPECIFIC_TARGET} field in
9741 the function declaration to hold a pointer to a target specific
9742 @var{struct cl_target_option} structure.
9745 @hook TARGET_OPTION_SAVE
9746 This hook is called to save any additional target specific information
9747 in the @var{struct cl_target_option} structure for function specific
9749 @xref{Option file format}.
9752 @hook TARGET_OPTION_RESTORE
9753 This hook is called to restore any additional target specific
9754 information in the @var{struct cl_target_option} structure for
9755 function specific options.
9758 @hook TARGET_OPTION_PRINT
9759 This hook is called to print any additional target specific
9760 information in the @var{struct cl_target_option} structure for
9761 function specific options.
9764 @hook TARGET_OPTION_PRAGMA_PARSE
9765 This target hook parses the options for @code{#pragma GCC option} to
9766 set the machine specific options for functions that occur later in the
9767 input stream. The options should be the same as handled by the
9768 @code{TARGET_OPTION_VALID_ATTRIBUTE_P} hook.
9771 @hook TARGET_OPTION_OVERRIDE
9772 Sometimes certain combinations of command options do not make sense on
9773 a particular target machine. You can override the hook
9774 @code{TARGET_OPTION_OVERRIDE} to take account of this. This hooks is called
9775 once just after all the command options have been parsed.
9777 Don't use this hook to turn on various extra optimizations for
9778 @option{-O}. That is what @code{TARGET_OPTION_OPTIMIZATION} is for.
9780 If you need to do something whenever the optimization level is
9781 changed via the optimize attribute or pragma, see
9782 @code{TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE}
9785 @hook TARGET_CAN_INLINE_P
9786 This target hook returns @code{false} if the @var{caller} function
9787 cannot inline @var{callee}, based on target specific information. By
9788 default, inlining is not allowed if the callee function has function
9789 specific target options and the caller does not use the same options.
9793 @section Emulating TLS
9794 @cindex Emulated TLS
9796 For targets whose psABI does not provide Thread Local Storage via
9797 specific relocations and instruction sequences, an emulation layer is
9798 used. A set of target hooks allows this emulation layer to be
9799 configured for the requirements of a particular target. For instance
9800 the psABI may in fact specify TLS support in terms of an emulation
9803 The emulation layer works by creating a control object for every TLS
9804 object. To access the TLS object, a lookup function is provided
9805 which, when given the address of the control object, will return the
9806 address of the current thread's instance of the TLS object.
9808 @hook TARGET_EMUTLS_GET_ADDRESS
9809 Contains the name of the helper function that uses a TLS control
9810 object to locate a TLS instance. The default causes libgcc's
9811 emulated TLS helper function to be used.
9814 @hook TARGET_EMUTLS_REGISTER_COMMON
9815 Contains the name of the helper function that should be used at
9816 program startup to register TLS objects that are implicitly
9817 initialized to zero. If this is @code{NULL}, all TLS objects will
9818 have explicit initializers. The default causes libgcc's emulated TLS
9819 registration function to be used.
9822 @hook TARGET_EMUTLS_VAR_SECTION
9823 Contains the name of the section in which TLS control variables should
9824 be placed. The default of @code{NULL} allows these to be placed in
9828 @hook TARGET_EMUTLS_TMPL_SECTION
9829 Contains the name of the section in which TLS initializers should be
9830 placed. The default of @code{NULL} allows these to be placed in any
9834 @hook TARGET_EMUTLS_VAR_PREFIX
9835 Contains the prefix to be prepended to TLS control variable names.
9836 The default of @code{NULL} uses a target-specific prefix.
9839 @hook TARGET_EMUTLS_TMPL_PREFIX
9840 Contains the prefix to be prepended to TLS initializer objects. The
9841 default of @code{NULL} uses a target-specific prefix.
9844 @hook TARGET_EMUTLS_VAR_FIELDS
9845 Specifies a function that generates the FIELD_DECLs for a TLS control
9846 object type. @var{type} is the RECORD_TYPE the fields are for and
9847 @var{name} should be filled with the structure tag, if the default of
9848 @code{__emutls_object} is unsuitable. The default creates a type suitable
9849 for libgcc's emulated TLS function.
9852 @hook TARGET_EMUTLS_VAR_INIT
9853 Specifies a function that generates the CONSTRUCTOR to initialize a
9854 TLS control object. @var{var} is the TLS control object, @var{decl}
9855 is the TLS object and @var{tmpl_addr} is the address of the
9856 initializer. The default initializes libgcc's emulated TLS control object.
9859 @hook TARGET_EMUTLS_VAR_ALIGN_FIXED
9860 Specifies whether the alignment of TLS control variable objects is
9861 fixed and should not be increased as some backends may do to optimize
9862 single objects. The default is false.
9865 @hook TARGET_EMUTLS_DEBUG_FORM_TLS_ADDRESS
9866 Specifies whether a DWARF @code{DW_OP_form_tls_address} location descriptor
9867 may be used to describe emulated TLS control objects.
9870 @node MIPS Coprocessors
9871 @section Defining coprocessor specifics for MIPS targets.
9872 @cindex MIPS coprocessor-definition macros
9874 The MIPS specification allows MIPS implementations to have as many as 4
9875 coprocessors, each with as many as 32 private registers. GCC supports
9876 accessing these registers and transferring values between the registers
9877 and memory using asm-ized variables. For example:
9880 register unsigned int cp0count asm ("c0r1");
9886 (``c0r1'' is the default name of register 1 in coprocessor 0; alternate
9887 names may be added as described below, or the default names may be
9888 overridden entirely in @code{SUBTARGET_CONDITIONAL_REGISTER_USAGE}.)
9890 Coprocessor registers are assumed to be epilogue-used; sets to them will
9891 be preserved even if it does not appear that the register is used again
9892 later in the function.
9894 Another note: according to the MIPS spec, coprocessor 1 (if present) is
9895 the FPU@. One accesses COP1 registers through standard mips
9896 floating-point support; they are not included in this mechanism.
9898 There is one macro used in defining the MIPS coprocessor interface which
9899 you may want to override in subtargets; it is described below.
9902 @section Parameters for Precompiled Header Validity Checking
9903 @cindex parameters, precompiled headers
9905 @hook TARGET_GET_PCH_VALIDITY
9906 This hook returns a pointer to the data needed by
9907 @code{TARGET_PCH_VALID_P} and sets
9908 @samp{*@var{sz}} to the size of the data in bytes.
9911 @hook TARGET_PCH_VALID_P
9912 This hook checks whether the options used to create a PCH file are
9913 compatible with the current settings. It returns @code{NULL}
9914 if so and a suitable error message if not. Error messages will
9915 be presented to the user and must be localized using @samp{_(@var{msg})}.
9917 @var{data} is the data that was returned by @code{TARGET_GET_PCH_VALIDITY}
9918 when the PCH file was created and @var{sz} is the size of that data in bytes.
9919 It's safe to assume that the data was created by the same version of the
9920 compiler, so no format checking is needed.
9922 The default definition of @code{default_pch_valid_p} should be
9923 suitable for most targets.
9926 @hook TARGET_CHECK_PCH_TARGET_FLAGS
9927 If this hook is nonnull, the default implementation of
9928 @code{TARGET_PCH_VALID_P} will use it to check for compatible values
9929 of @code{target_flags}. @var{pch_flags} specifies the value that
9930 @code{target_flags} had when the PCH file was created. The return
9931 value is the same as for @code{TARGET_PCH_VALID_P}.
9934 @hook TARGET_PREPARE_PCH_SAVE
9937 @section C++ ABI parameters
9938 @cindex parameters, c++ abi
9940 @hook TARGET_CXX_GUARD_TYPE
9941 Define this hook to override the integer type used for guard variables.
9942 These are used to implement one-time construction of static objects. The
9943 default is long_long_integer_type_node.
9946 @hook TARGET_CXX_GUARD_MASK_BIT
9947 This hook determines how guard variables are used. It should return
9948 @code{false} (the default) if the first byte should be used. A return value of
9949 @code{true} indicates that only the least significant bit should be used.
9952 @hook TARGET_CXX_GET_COOKIE_SIZE
9953 This hook returns the size of the cookie to use when allocating an array
9954 whose elements have the indicated @var{type}. Assumes that it is already
9955 known that a cookie is needed. The default is
9956 @code{max(sizeof (size_t), alignof(type))}, as defined in section 2.7 of the
9957 IA64/Generic C++ ABI@.
9960 @hook TARGET_CXX_COOKIE_HAS_SIZE
9961 This hook should return @code{true} if the element size should be stored in
9962 array cookies. The default is to return @code{false}.
9965 @hook TARGET_CXX_IMPORT_EXPORT_CLASS
9966 If defined by a backend this hook allows the decision made to export
9967 class @var{type} to be overruled. Upon entry @var{import_export}
9968 will contain 1 if the class is going to be exported, @minus{}1 if it is going
9969 to be imported and 0 otherwise. This function should return the
9970 modified value and perform any other actions necessary to support the
9971 backend's targeted operating system.
9974 @hook TARGET_CXX_CDTOR_RETURNS_THIS
9975 This hook should return @code{true} if constructors and destructors return
9976 the address of the object created/destroyed. The default is to return
9980 @hook TARGET_CXX_KEY_METHOD_MAY_BE_INLINE
9981 This hook returns true if the key method for a class (i.e., the method
9982 which, if defined in the current translation unit, causes the virtual
9983 table to be emitted) may be an inline function. Under the standard
9984 Itanium C++ ABI the key method may be an inline function so long as
9985 the function is not declared inline in the class definition. Under
9986 some variants of the ABI, an inline function can never be the key
9987 method. The default is to return @code{true}.
9990 @hook TARGET_CXX_DETERMINE_CLASS_DATA_VISIBILITY
9992 @hook TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT
9993 This hook returns true (the default) if virtual tables and other
9994 similar implicit class data objects are always COMDAT if they have
9995 external linkage. If this hook returns false, then class data for
9996 classes whose virtual table will be emitted in only one translation
9997 unit will not be COMDAT.
10000 @hook TARGET_CXX_LIBRARY_RTTI_COMDAT
10001 This hook returns true (the default) if the RTTI information for
10002 the basic types which is defined in the C++ runtime should always
10003 be COMDAT, false if it should not be COMDAT.
10006 @hook TARGET_CXX_USE_AEABI_ATEXIT
10007 This hook returns true if @code{__aeabi_atexit} (as defined by the ARM EABI)
10008 should be used to register static destructors when @option{-fuse-cxa-atexit}
10009 is in effect. The default is to return false to use @code{__cxa_atexit}.
10012 @hook TARGET_CXX_USE_ATEXIT_FOR_CXA_ATEXIT
10013 This hook returns true if the target @code{atexit} function can be used
10014 in the same manner as @code{__cxa_atexit} to register C++ static
10015 destructors. This requires that @code{atexit}-registered functions in
10016 shared libraries are run in the correct order when the libraries are
10017 unloaded. The default is to return false.
10020 @hook TARGET_CXX_ADJUST_CLASS_AT_DEFINITION
10022 @hook TARGET_CXX_DECL_MANGLING_CONTEXT
10024 @node Named Address Spaces
10025 @section Adding support for named address spaces
10026 @cindex named address spaces
10028 The draft technical report of the ISO/IEC JTC1 S22 WG14 N1275
10029 standards committee, @cite{Programming Languages - C - Extensions to
10030 support embedded processors}, specifies a syntax for embedded
10031 processors to specify alternate address spaces. You can configure a
10032 GCC port to support section 5.1 of the draft report to add support for
10033 address spaces other than the default address space. These address
10034 spaces are new keywords that are similar to the @code{volatile} and
10035 @code{const} type attributes.
10037 Pointers to named address spaces can have a different size than
10038 pointers to the generic address space.
10040 For example, the SPU port uses the @code{__ea} address space to refer
10041 to memory in the host processor, rather than memory local to the SPU
10042 processor. Access to memory in the @code{__ea} address space involves
10043 issuing DMA operations to move data between the host processor and the
10044 local processor memory address space. Pointers in the @code{__ea}
10045 address space are either 32 bits or 64 bits based on the
10046 @option{-mea32} or @option{-mea64} switches (native SPU pointers are
10049 Internally, address spaces are represented as a small integer in the
10050 range 0 to 15 with address space 0 being reserved for the generic
10053 To register a named address space qualifier keyword with the C front end,
10054 the target may call the @code{c_register_addr_space} routine. For example,
10055 the SPU port uses the following to declare @code{__ea} as the keyword for
10056 named address space #1:
10058 #define ADDR_SPACE_EA 1
10059 c_register_addr_space ("__ea", ADDR_SPACE_EA);
10062 @hook TARGET_ADDR_SPACE_POINTER_MODE
10063 Define this to return the machine mode to use for pointers to
10064 @var{address_space} if the target supports named address spaces.
10065 The default version of this hook returns @code{ptr_mode} for the
10066 generic address space only.
10069 @hook TARGET_ADDR_SPACE_ADDRESS_MODE
10070 Define this to return the machine mode to use for addresses in
10071 @var{address_space} if the target supports named address spaces.
10072 The default version of this hook returns @code{Pmode} for the
10073 generic address space only.
10076 @hook TARGET_ADDR_SPACE_VALID_POINTER_MODE
10077 Define this to return nonzero if the port can handle pointers
10078 with machine mode @var{mode} to address space @var{as}. This target
10079 hook is the same as the @code{TARGET_VALID_POINTER_MODE} target hook,
10080 except that it includes explicit named address space support. The default
10081 version of this hook returns true for the modes returned by either the
10082 @code{TARGET_ADDR_SPACE_POINTER_MODE} or @code{TARGET_ADDR_SPACE_ADDRESS_MODE}
10083 target hooks for the given address space.
10086 @hook TARGET_ADDR_SPACE_LEGITIMATE_ADDRESS_P
10087 Define this to return true if @var{exp} is a valid address for mode
10088 @var{mode} in the named address space @var{as}. The @var{strict}
10089 parameter says whether strict addressing is in effect after reload has
10090 finished. This target hook is the same as the
10091 @code{TARGET_LEGITIMATE_ADDRESS_P} target hook, except that it includes
10092 explicit named address space support.
10095 @hook TARGET_ADDR_SPACE_LEGITIMIZE_ADDRESS
10096 Define this to modify an invalid address @var{x} to be a valid address
10097 with mode @var{mode} in the named address space @var{as}. This target
10098 hook is the same as the @code{TARGET_LEGITIMIZE_ADDRESS} target hook,
10099 except that it includes explicit named address space support.
10102 @hook TARGET_ADDR_SPACE_SUBSET_P
10103 Define this to return whether the @var{subset} named address space is
10104 contained within the @var{superset} named address space. Pointers to
10105 a named address space that is a subset of another named address space
10106 will be converted automatically without a cast if used together in
10107 arithmetic operations. Pointers to a superset address space can be
10108 converted to pointers to a subset address space via explicit casts.
10111 @hook TARGET_ADDR_SPACE_CONVERT
10112 Define this to convert the pointer expression represented by the RTL
10113 @var{op} with type @var{from_type} that points to a named address
10114 space to a new pointer expression with type @var{to_type} that points
10115 to a different named address space. When this hook it called, it is
10116 guaranteed that one of the two address spaces is a subset of the other,
10117 as determined by the @code{TARGET_ADDR_SPACE_SUBSET_P} target hook.
10121 @section Miscellaneous Parameters
10122 @cindex parameters, miscellaneous
10124 @c prevent bad page break with this line
10125 Here are several miscellaneous parameters.
10127 @defmac HAS_LONG_COND_BRANCH
10128 Define this boolean macro to indicate whether or not your architecture
10129 has conditional branches that can span all of memory. It is used in
10130 conjunction with an optimization that partitions hot and cold basic
10131 blocks into separate sections of the executable. If this macro is
10132 set to false, gcc will convert any conditional branches that attempt
10133 to cross between sections into unconditional branches or indirect jumps.
10136 @defmac HAS_LONG_UNCOND_BRANCH
10137 Define this boolean macro to indicate whether or not your architecture
10138 has unconditional branches that can span all of memory. It is used in
10139 conjunction with an optimization that partitions hot and cold basic
10140 blocks into separate sections of the executable. If this macro is
10141 set to false, gcc will convert any unconditional branches that attempt
10142 to cross between sections into indirect jumps.
10145 @defmac CASE_VECTOR_MODE
10146 An alias for a machine mode name. This is the machine mode that
10147 elements of a jump-table should have.
10150 @defmac CASE_VECTOR_SHORTEN_MODE (@var{min_offset}, @var{max_offset}, @var{body})
10151 Optional: return the preferred mode for an @code{addr_diff_vec}
10152 when the minimum and maximum offset are known. If you define this,
10153 it enables extra code in branch shortening to deal with @code{addr_diff_vec}.
10154 To make this work, you also have to define @code{INSN_ALIGN} and
10155 make the alignment for @code{addr_diff_vec} explicit.
10156 The @var{body} argument is provided so that the offset_unsigned and scale
10157 flags can be updated.
10160 @defmac CASE_VECTOR_PC_RELATIVE
10161 Define this macro to be a C expression to indicate when jump-tables
10162 should contain relative addresses. You need not define this macro if
10163 jump-tables never contain relative addresses, or jump-tables should
10164 contain relative addresses only when @option{-fPIC} or @option{-fPIC}
10168 @hook TARGET_CASE_VALUES_THRESHOLD
10169 This function return the smallest number of different values for which it
10170 is best to use a jump-table instead of a tree of conditional branches.
10171 The default is four for machines with a @code{casesi} instruction and
10172 five otherwise. This is best for most machines.
10175 @defmac WORD_REGISTER_OPERATIONS
10176 Define this macro if operations between registers with integral mode
10177 smaller than a word are always performed on the entire register.
10178 Most RISC machines have this property and most CISC machines do not.
10181 @defmac LOAD_EXTEND_OP (@var{mem_mode})
10182 Define this macro to be a C expression indicating when insns that read
10183 memory in @var{mem_mode}, an integral mode narrower than a word, set the
10184 bits outside of @var{mem_mode} to be either the sign-extension or the
10185 zero-extension of the data read. Return @code{SIGN_EXTEND} for values
10186 of @var{mem_mode} for which the
10187 insn sign-extends, @code{ZERO_EXTEND} for which it zero-extends, and
10188 @code{UNKNOWN} for other modes.
10190 This macro is not called with @var{mem_mode} non-integral or with a width
10191 greater than or equal to @code{BITS_PER_WORD}, so you may return any
10192 value in this case. Do not define this macro if it would always return
10193 @code{UNKNOWN}. On machines where this macro is defined, you will normally
10194 define it as the constant @code{SIGN_EXTEND} or @code{ZERO_EXTEND}.
10196 You may return a non-@code{UNKNOWN} value even if for some hard registers
10197 the sign extension is not performed, if for the @code{REGNO_REG_CLASS}
10198 of these hard registers @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero
10199 when the @var{from} mode is @var{mem_mode} and the @var{to} mode is any
10200 integral mode larger than this but not larger than @code{word_mode}.
10202 You must return @code{UNKNOWN} if for some hard registers that allow this
10203 mode, @code{CANNOT_CHANGE_MODE_CLASS} says that they cannot change to
10204 @code{word_mode}, but that they can change to another integral mode that
10205 is larger then @var{mem_mode} but still smaller than @code{word_mode}.
10208 @defmac SHORT_IMMEDIATES_SIGN_EXTEND
10209 Define this macro if loading short immediate values into registers sign
10213 @hook TARGET_MIN_DIVISIONS_FOR_RECIP_MUL
10214 When @option{-ffast-math} is in effect, GCC tries to optimize
10215 divisions by the same divisor, by turning them into multiplications by
10216 the reciprocal. This target hook specifies the minimum number of divisions
10217 that should be there for GCC to perform the optimization for a variable
10218 of mode @var{mode}. The default implementation returns 3 if the machine
10219 has an instruction for the division, and 2 if it does not.
10223 The maximum number of bytes that a single instruction can move quickly
10224 between memory and registers or between two memory locations.
10227 @defmac MAX_MOVE_MAX
10228 The maximum number of bytes that a single instruction can move quickly
10229 between memory and registers or between two memory locations. If this
10230 is undefined, the default is @code{MOVE_MAX}. Otherwise, it is the
10231 constant value that is the largest value that @code{MOVE_MAX} can have
10235 @defmac SHIFT_COUNT_TRUNCATED
10236 A C expression that is nonzero if on this machine the number of bits
10237 actually used for the count of a shift operation is equal to the number
10238 of bits needed to represent the size of the object being shifted. When
10239 this macro is nonzero, the compiler will assume that it is safe to omit
10240 a sign-extend, zero-extend, and certain bitwise `and' instructions that
10241 truncates the count of a shift operation. On machines that have
10242 instructions that act on bit-fields at variable positions, which may
10243 include `bit test' instructions, a nonzero @code{SHIFT_COUNT_TRUNCATED}
10244 also enables deletion of truncations of the values that serve as
10245 arguments to bit-field instructions.
10247 If both types of instructions truncate the count (for shifts) and
10248 position (for bit-field operations), or if no variable-position bit-field
10249 instructions exist, you should define this macro.
10251 However, on some machines, such as the 80386 and the 680x0, truncation
10252 only applies to shift operations and not the (real or pretended)
10253 bit-field operations. Define @code{SHIFT_COUNT_TRUNCATED} to be zero on
10254 such machines. Instead, add patterns to the @file{md} file that include
10255 the implied truncation of the shift instructions.
10257 You need not define this macro if it would always have the value of zero.
10260 @anchor{TARGET_SHIFT_TRUNCATION_MASK}
10261 @hook TARGET_SHIFT_TRUNCATION_MASK
10262 This function describes how the standard shift patterns for @var{mode}
10263 deal with shifts by negative amounts or by more than the width of the mode.
10264 @xref{shift patterns}.
10266 On many machines, the shift patterns will apply a mask @var{m} to the
10267 shift count, meaning that a fixed-width shift of @var{x} by @var{y} is
10268 equivalent to an arbitrary-width shift of @var{x} by @var{y & m}. If
10269 this is true for mode @var{mode}, the function should return @var{m},
10270 otherwise it should return 0. A return value of 0 indicates that no
10271 particular behavior is guaranteed.
10273 Note that, unlike @code{SHIFT_COUNT_TRUNCATED}, this function does
10274 @emph{not} apply to general shift rtxes; it applies only to instructions
10275 that are generated by the named shift patterns.
10277 The default implementation of this function returns
10278 @code{GET_MODE_BITSIZE (@var{mode}) - 1} if @code{SHIFT_COUNT_TRUNCATED}
10279 and 0 otherwise. This definition is always safe, but if
10280 @code{SHIFT_COUNT_TRUNCATED} is false, and some shift patterns
10281 nevertheless truncate the shift count, you may get better code
10285 @defmac TRULY_NOOP_TRUNCATION (@var{outprec}, @var{inprec})
10286 A C expression which is nonzero if on this machine it is safe to
10287 ``convert'' an integer of @var{inprec} bits to one of @var{outprec}
10288 bits (where @var{outprec} is smaller than @var{inprec}) by merely
10289 operating on it as if it had only @var{outprec} bits.
10291 On many machines, this expression can be 1.
10293 @c rearranged this, removed the phrase "it is reported that". this was
10294 @c to fix an overfull hbox. --mew 10feb93
10295 When @code{TRULY_NOOP_TRUNCATION} returns 1 for a pair of sizes for
10296 modes for which @code{MODES_TIEABLE_P} is 0, suboptimal code can result.
10297 If this is the case, making @code{TRULY_NOOP_TRUNCATION} return 0 in
10298 such cases may improve things.
10301 @hook TARGET_MODE_REP_EXTENDED
10302 The representation of an integral mode can be such that the values
10303 are always extended to a wider integral mode. Return
10304 @code{SIGN_EXTEND} if values of @var{mode} are represented in
10305 sign-extended form to @var{rep_mode}. Return @code{UNKNOWN}
10306 otherwise. (Currently, none of the targets use zero-extended
10307 representation this way so unlike @code{LOAD_EXTEND_OP},
10308 @code{TARGET_MODE_REP_EXTENDED} is expected to return either
10309 @code{SIGN_EXTEND} or @code{UNKNOWN}. Also no target extends
10310 @var{mode} to @var{rep_mode} so that @var{rep_mode} is not the next
10311 widest integral mode and currently we take advantage of this fact.)
10313 Similarly to @code{LOAD_EXTEND_OP} you may return a non-@code{UNKNOWN}
10314 value even if the extension is not performed on certain hard registers
10315 as long as for the @code{REGNO_REG_CLASS} of these hard registers
10316 @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero.
10318 Note that @code{TARGET_MODE_REP_EXTENDED} and @code{LOAD_EXTEND_OP}
10319 describe two related properties. If you define
10320 @code{TARGET_MODE_REP_EXTENDED (mode, word_mode)} you probably also want
10321 to define @code{LOAD_EXTEND_OP (mode)} to return the same type of
10324 In order to enforce the representation of @code{mode},
10325 @code{TRULY_NOOP_TRUNCATION} should return false when truncating to
10329 @defmac STORE_FLAG_VALUE
10330 A C expression describing the value returned by a comparison operator
10331 with an integral mode and stored by a store-flag instruction
10332 (@samp{cstore@var{mode}4}) when the condition is true. This description must
10333 apply to @emph{all} the @samp{cstore@var{mode}4} patterns and all the
10334 comparison operators whose results have a @code{MODE_INT} mode.
10336 A value of 1 or @minus{}1 means that the instruction implementing the
10337 comparison operator returns exactly 1 or @minus{}1 when the comparison is true
10338 and 0 when the comparison is false. Otherwise, the value indicates
10339 which bits of the result are guaranteed to be 1 when the comparison is
10340 true. This value is interpreted in the mode of the comparison
10341 operation, which is given by the mode of the first operand in the
10342 @samp{cstore@var{mode}4} pattern. Either the low bit or the sign bit of
10343 @code{STORE_FLAG_VALUE} be on. Presently, only those bits are used by
10346 If @code{STORE_FLAG_VALUE} is neither 1 or @minus{}1, the compiler will
10347 generate code that depends only on the specified bits. It can also
10348 replace comparison operators with equivalent operations if they cause
10349 the required bits to be set, even if the remaining bits are undefined.
10350 For example, on a machine whose comparison operators return an
10351 @code{SImode} value and where @code{STORE_FLAG_VALUE} is defined as
10352 @samp{0x80000000}, saying that just the sign bit is relevant, the
10356 (ne:SI (and:SI @var{x} (const_int @var{power-of-2})) (const_int 0))
10360 can be converted to
10363 (ashift:SI @var{x} (const_int @var{n}))
10367 where @var{n} is the appropriate shift count to move the bit being
10368 tested into the sign bit.
10370 There is no way to describe a machine that always sets the low-order bit
10371 for a true value, but does not guarantee the value of any other bits,
10372 but we do not know of any machine that has such an instruction. If you
10373 are trying to port GCC to such a machine, include an instruction to
10374 perform a logical-and of the result with 1 in the pattern for the
10375 comparison operators and let us know at @email{gcc@@gcc.gnu.org}.
10377 Often, a machine will have multiple instructions that obtain a value
10378 from a comparison (or the condition codes). Here are rules to guide the
10379 choice of value for @code{STORE_FLAG_VALUE}, and hence the instructions
10384 Use the shortest sequence that yields a valid definition for
10385 @code{STORE_FLAG_VALUE}. It is more efficient for the compiler to
10386 ``normalize'' the value (convert it to, e.g., 1 or 0) than for the
10387 comparison operators to do so because there may be opportunities to
10388 combine the normalization with other operations.
10391 For equal-length sequences, use a value of 1 or @minus{}1, with @minus{}1 being
10392 slightly preferred on machines with expensive jumps and 1 preferred on
10396 As a second choice, choose a value of @samp{0x80000001} if instructions
10397 exist that set both the sign and low-order bits but do not define the
10401 Otherwise, use a value of @samp{0x80000000}.
10404 Many machines can produce both the value chosen for
10405 @code{STORE_FLAG_VALUE} and its negation in the same number of
10406 instructions. On those machines, you should also define a pattern for
10407 those cases, e.g., one matching
10410 (set @var{A} (neg:@var{m} (ne:@var{m} @var{B} @var{C})))
10413 Some machines can also perform @code{and} or @code{plus} operations on
10414 condition code values with less instructions than the corresponding
10415 @samp{cstore@var{mode}4} insn followed by @code{and} or @code{plus}. On those
10416 machines, define the appropriate patterns. Use the names @code{incscc}
10417 and @code{decscc}, respectively, for the patterns which perform
10418 @code{plus} or @code{minus} operations on condition code values. See
10419 @file{rs6000.md} for some examples. The GNU Superoptimizer can be used to
10420 find such instruction sequences on other machines.
10422 If this macro is not defined, the default value, 1, is used. You need
10423 not define @code{STORE_FLAG_VALUE} if the machine has no store-flag
10424 instructions, or if the value generated by these instructions is 1.
10427 @defmac FLOAT_STORE_FLAG_VALUE (@var{mode})
10428 A C expression that gives a nonzero @code{REAL_VALUE_TYPE} value that is
10429 returned when comparison operators with floating-point results are true.
10430 Define this macro on machines that have comparison operations that return
10431 floating-point values. If there are no such operations, do not define
10435 @defmac VECTOR_STORE_FLAG_VALUE (@var{mode})
10436 A C expression that gives a rtx representing the nonzero true element
10437 for vector comparisons. The returned rtx should be valid for the inner
10438 mode of @var{mode} which is guaranteed to be a vector mode. Define
10439 this macro on machines that have vector comparison operations that
10440 return a vector result. If there are no such operations, do not define
10441 this macro. Typically, this macro is defined as @code{const1_rtx} or
10442 @code{constm1_rtx}. This macro may return @code{NULL_RTX} to prevent
10443 the compiler optimizing such vector comparison operations for the
10447 @defmac CLZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
10448 @defmacx CTZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
10449 A C expression that indicates whether the architecture defines a value
10450 for @code{clz} or @code{ctz} with a zero operand.
10451 A result of @code{0} indicates the value is undefined.
10452 If the value is defined for only the RTL expression, the macro should
10453 evaluate to @code{1}; if the value applies also to the corresponding optab
10454 entry (which is normally the case if it expands directly into
10455 the corresponding RTL), then the macro should evaluate to @code{2}.
10456 In the cases where the value is defined, @var{value} should be set to
10459 If this macro is not defined, the value of @code{clz} or
10460 @code{ctz} at zero is assumed to be undefined.
10462 This macro must be defined if the target's expansion for @code{ffs}
10463 relies on a particular value to get correct results. Otherwise it
10464 is not necessary, though it may be used to optimize some corner cases, and
10465 to provide a default expansion for the @code{ffs} optab.
10467 Note that regardless of this macro the ``definedness'' of @code{clz}
10468 and @code{ctz} at zero do @emph{not} extend to the builtin functions
10469 visible to the user. Thus one may be free to adjust the value at will
10470 to match the target expansion of these operations without fear of
10475 An alias for the machine mode for pointers. On most machines, define
10476 this to be the integer mode corresponding to the width of a hardware
10477 pointer; @code{SImode} on 32-bit machine or @code{DImode} on 64-bit machines.
10478 On some machines you must define this to be one of the partial integer
10479 modes, such as @code{PSImode}.
10481 The width of @code{Pmode} must be at least as large as the value of
10482 @code{POINTER_SIZE}. If it is not equal, you must define the macro
10483 @code{POINTERS_EXTEND_UNSIGNED} to specify how pointers are extended
10487 @defmac FUNCTION_MODE
10488 An alias for the machine mode used for memory references to functions
10489 being called, in @code{call} RTL expressions. On most CISC machines,
10490 where an instruction can begin at any byte address, this should be
10491 @code{QImode}. On most RISC machines, where all instructions have fixed
10492 size and alignment, this should be a mode with the same size and alignment
10493 as the machine instruction words - typically @code{SImode} or @code{HImode}.
10496 @defmac STDC_0_IN_SYSTEM_HEADERS
10497 In normal operation, the preprocessor expands @code{__STDC__} to the
10498 constant 1, to signify that GCC conforms to ISO Standard C@. On some
10499 hosts, like Solaris, the system compiler uses a different convention,
10500 where @code{__STDC__} is normally 0, but is 1 if the user specifies
10501 strict conformance to the C Standard.
10503 Defining @code{STDC_0_IN_SYSTEM_HEADERS} makes GNU CPP follows the host
10504 convention when processing system header files, but when processing user
10505 files @code{__STDC__} will always expand to 1.
10508 @hook TARGET_C_PREINCLUDE
10510 @defmac NO_IMPLICIT_EXTERN_C
10511 Define this macro if the system header files support C++ as well as C@.
10512 This macro inhibits the usual method of using system header files in
10513 C++, which is to pretend that the file's contents are enclosed in
10514 @samp{extern "C" @{@dots{}@}}.
10519 @defmac REGISTER_TARGET_PRAGMAS ()
10520 Define this macro if you want to implement any target-specific pragmas.
10521 If defined, it is a C expression which makes a series of calls to
10522 @code{c_register_pragma} or @code{c_register_pragma_with_expansion}
10523 for each pragma. The macro may also do any
10524 setup required for the pragmas.
10526 The primary reason to define this macro is to provide compatibility with
10527 other compilers for the same target. In general, we discourage
10528 definition of target-specific pragmas for GCC@.
10530 If the pragma can be implemented by attributes then you should consider
10531 defining the target hook @samp{TARGET_INSERT_ATTRIBUTES} as well.
10533 Preprocessor macros that appear on pragma lines are not expanded. All
10534 @samp{#pragma} directives that do not match any registered pragma are
10535 silently ignored, unless the user specifies @option{-Wunknown-pragmas}.
10538 @deftypefun void c_register_pragma (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
10539 @deftypefunx void c_register_pragma_with_expansion (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
10541 Each call to @code{c_register_pragma} or
10542 @code{c_register_pragma_with_expansion} establishes one pragma. The
10543 @var{callback} routine will be called when the preprocessor encounters a
10547 #pragma [@var{space}] @var{name} @dots{}
10550 @var{space} is the case-sensitive namespace of the pragma, or
10551 @code{NULL} to put the pragma in the global namespace. The callback
10552 routine receives @var{pfile} as its first argument, which can be passed
10553 on to cpplib's functions if necessary. You can lex tokens after the
10554 @var{name} by calling @code{pragma_lex}. Tokens that are not read by the
10555 callback will be silently ignored. The end of the line is indicated by
10556 a token of type @code{CPP_EOF}. Macro expansion occurs on the
10557 arguments of pragmas registered with
10558 @code{c_register_pragma_with_expansion} but not on the arguments of
10559 pragmas registered with @code{c_register_pragma}.
10561 Note that the use of @code{pragma_lex} is specific to the C and C++
10562 compilers. It will not work in the Java or Fortran compilers, or any
10563 other language compilers for that matter. Thus if @code{pragma_lex} is going
10564 to be called from target-specific code, it must only be done so when
10565 building the C and C++ compilers. This can be done by defining the
10566 variables @code{c_target_objs} and @code{cxx_target_objs} in the
10567 target entry in the @file{config.gcc} file. These variables should name
10568 the target-specific, language-specific object file which contains the
10569 code that uses @code{pragma_lex}. Note it will also be necessary to add a
10570 rule to the makefile fragment pointed to by @code{tmake_file} that shows
10571 how to build this object file.
10574 @defmac HANDLE_PRAGMA_PACK_WITH_EXPANSION
10575 Define this macro if macros should be expanded in the
10576 arguments of @samp{#pragma pack}.
10579 @defmac TARGET_DEFAULT_PACK_STRUCT
10580 If your target requires a structure packing default other than 0 (meaning
10581 the machine default), define this macro to the necessary value (in bytes).
10582 This must be a value that would also be valid to use with
10583 @samp{#pragma pack()} (that is, a small power of two).
10586 @defmac DOLLARS_IN_IDENTIFIERS
10587 Define this macro to control use of the character @samp{$} in
10588 identifier names for the C family of languages. 0 means @samp{$} is
10589 not allowed by default; 1 means it is allowed. 1 is the default;
10590 there is no need to define this macro in that case.
10593 @defmac INSN_SETS_ARE_DELAYED (@var{insn})
10594 Define this macro as a C expression that is nonzero if it is safe for the
10595 delay slot scheduler to place instructions in the delay slot of @var{insn},
10596 even if they appear to use a resource set or clobbered in @var{insn}.
10597 @var{insn} is always a @code{jump_insn} or an @code{insn}; GCC knows that
10598 every @code{call_insn} has this behavior. On machines where some @code{insn}
10599 or @code{jump_insn} is really a function call and hence has this behavior,
10600 you should define this macro.
10602 You need not define this macro if it would always return zero.
10605 @defmac INSN_REFERENCES_ARE_DELAYED (@var{insn})
10606 Define this macro as a C expression that is nonzero if it is safe for the
10607 delay slot scheduler to place instructions in the delay slot of @var{insn},
10608 even if they appear to set or clobber a resource referenced in @var{insn}.
10609 @var{insn} is always a @code{jump_insn} or an @code{insn}. On machines where
10610 some @code{insn} or @code{jump_insn} is really a function call and its operands
10611 are registers whose use is actually in the subroutine it calls, you should
10612 define this macro. Doing so allows the delay slot scheduler to move
10613 instructions which copy arguments into the argument registers into the delay
10614 slot of @var{insn}.
10616 You need not define this macro if it would always return zero.
10619 @defmac MULTIPLE_SYMBOL_SPACES
10620 Define this macro as a C expression that is nonzero if, in some cases,
10621 global symbols from one translation unit may not be bound to undefined
10622 symbols in another translation unit without user intervention. For
10623 instance, under Microsoft Windows symbols must be explicitly imported
10624 from shared libraries (DLLs).
10626 You need not define this macro if it would always evaluate to zero.
10629 @hook TARGET_MD_ASM_CLOBBERS
10630 This target hook should add to @var{clobbers} @code{STRING_CST} trees for
10631 any hard regs the port wishes to automatically clobber for an asm.
10632 It should return the result of the last @code{tree_cons} used to add a
10633 clobber. The @var{outputs}, @var{inputs} and @var{clobber} lists are the
10634 corresponding parameters to the asm and may be inspected to avoid
10635 clobbering a register that is an input or output of the asm. You can use
10636 @code{tree_overlaps_hard_reg_set}, declared in @file{tree.h}, to test
10637 for overlap with regards to asm-declared registers.
10640 @defmac MATH_LIBRARY
10641 Define this macro as a C string constant for the linker argument to link
10642 in the system math library, minus the initial @samp{"-l"}, or
10643 @samp{""} if the target does not have a
10644 separate math library.
10646 You need only define this macro if the default of @samp{"m"} is wrong.
10649 @defmac LIBRARY_PATH_ENV
10650 Define this macro as a C string constant for the environment variable that
10651 specifies where the linker should look for libraries.
10653 You need only define this macro if the default of @samp{"LIBRARY_PATH"}
10657 @defmac TARGET_POSIX_IO
10658 Define this macro if the target supports the following POSIX@ file
10659 functions, access, mkdir and file locking with fcntl / F_SETLKW@.
10660 Defining @code{TARGET_POSIX_IO} will enable the test coverage code
10661 to use file locking when exiting a program, which avoids race conditions
10662 if the program has forked. It will also create directories at run-time
10663 for cross-profiling.
10666 @defmac MAX_CONDITIONAL_EXECUTE
10668 A C expression for the maximum number of instructions to execute via
10669 conditional execution instructions instead of a branch. A value of
10670 @code{BRANCH_COST}+1 is the default if the machine does not use cc0, and
10671 1 if it does use cc0.
10674 @defmac IFCVT_MODIFY_TESTS (@var{ce_info}, @var{true_expr}, @var{false_expr})
10675 Used if the target needs to perform machine-dependent modifications on the
10676 conditionals used for turning basic blocks into conditionally executed code.
10677 @var{ce_info} points to a data structure, @code{struct ce_if_block}, which
10678 contains information about the currently processed blocks. @var{true_expr}
10679 and @var{false_expr} are the tests that are used for converting the
10680 then-block and the else-block, respectively. Set either @var{true_expr} or
10681 @var{false_expr} to a null pointer if the tests cannot be converted.
10684 @defmac IFCVT_MODIFY_MULTIPLE_TESTS (@var{ce_info}, @var{bb}, @var{true_expr}, @var{false_expr})
10685 Like @code{IFCVT_MODIFY_TESTS}, but used when converting more complicated
10686 if-statements into conditions combined by @code{and} and @code{or} operations.
10687 @var{bb} contains the basic block that contains the test that is currently
10688 being processed and about to be turned into a condition.
10691 @defmac IFCVT_MODIFY_INSN (@var{ce_info}, @var{pattern}, @var{insn})
10692 A C expression to modify the @var{PATTERN} of an @var{INSN} that is to
10693 be converted to conditional execution format. @var{ce_info} points to
10694 a data structure, @code{struct ce_if_block}, which contains information
10695 about the currently processed blocks.
10698 @defmac IFCVT_MODIFY_FINAL (@var{ce_info})
10699 A C expression to perform any final machine dependent modifications in
10700 converting code to conditional execution. The involved basic blocks
10701 can be found in the @code{struct ce_if_block} structure that is pointed
10702 to by @var{ce_info}.
10705 @defmac IFCVT_MODIFY_CANCEL (@var{ce_info})
10706 A C expression to cancel any machine dependent modifications in
10707 converting code to conditional execution. The involved basic blocks
10708 can be found in the @code{struct ce_if_block} structure that is pointed
10709 to by @var{ce_info}.
10712 @defmac IFCVT_MACHDEP_INIT (@var{ce_info})
10713 A C expression to initialize any machine specific data for if-conversion
10714 of the if-block in the @code{struct ce_if_block} structure that is pointed
10715 to by @var{ce_info}.
10718 @hook TARGET_MACHINE_DEPENDENT_REORG
10719 If non-null, this hook performs a target-specific pass over the
10720 instruction stream. The compiler will run it at all optimization levels,
10721 just before the point at which it normally does delayed-branch scheduling.
10723 The exact purpose of the hook varies from target to target. Some use
10724 it to do transformations that are necessary for correctness, such as
10725 laying out in-function constant pools or avoiding hardware hazards.
10726 Others use it as an opportunity to do some machine-dependent optimizations.
10728 You need not implement the hook if it has nothing to do. The default
10729 definition is null.
10732 @hook TARGET_INIT_BUILTINS
10733 Define this hook if you have any machine-specific built-in functions
10734 that need to be defined. It should be a function that performs the
10737 Machine specific built-in functions can be useful to expand special machine
10738 instructions that would otherwise not normally be generated because
10739 they have no equivalent in the source language (for example, SIMD vector
10740 instructions or prefetch instructions).
10742 To create a built-in function, call the function
10743 @code{lang_hooks.builtin_function}
10744 which is defined by the language front end. You can use any type nodes set
10745 up by @code{build_common_tree_nodes};
10746 only language front ends that use those two functions will call
10747 @samp{TARGET_INIT_BUILTINS}.
10750 @hook TARGET_BUILTIN_DECL
10751 Define this hook if you have any machine-specific built-in functions
10752 that need to be defined. It should be a function that returns the
10753 builtin function declaration for the builtin function code @var{code}.
10754 If there is no such builtin and it cannot be initialized at this time
10755 if @var{initialize_p} is true the function should return @code{NULL_TREE}.
10756 If @var{code} is out of range the function should return
10757 @code{error_mark_node}.
10760 @hook TARGET_EXPAND_BUILTIN
10762 Expand a call to a machine specific built-in function that was set up by
10763 @samp{TARGET_INIT_BUILTINS}. @var{exp} is the expression for the
10764 function call; the result should go to @var{target} if that is
10765 convenient, and have mode @var{mode} if that is convenient.
10766 @var{subtarget} may be used as the target for computing one of
10767 @var{exp}'s operands. @var{ignore} is nonzero if the value is to be
10768 ignored. This function should return the result of the call to the
10772 @hook TARGET_RESOLVE_OVERLOADED_BUILTIN
10773 Select a replacement for a machine specific built-in function that
10774 was set up by @samp{TARGET_INIT_BUILTINS}. This is done
10775 @emph{before} regular type checking, and so allows the target to
10776 implement a crude form of function overloading. @var{fndecl} is the
10777 declaration of the built-in function. @var{arglist} is the list of
10778 arguments passed to the built-in function. The result is a
10779 complete expression that implements the operation, usually
10780 another @code{CALL_EXPR}.
10781 @var{arglist} really has type @samp{VEC(tree,gc)*}
10784 @hook TARGET_FOLD_BUILTIN
10785 Fold a call to a machine specific built-in function that was set up by
10786 @samp{TARGET_INIT_BUILTINS}. @var{fndecl} is the declaration of the
10787 built-in function. @var{n_args} is the number of arguments passed to
10788 the function; the arguments themselves are pointed to by @var{argp}.
10789 The result is another tree containing a simplified expression for the
10790 call's result. If @var{ignore} is true the value will be ignored.
10793 @hook TARGET_INVALID_WITHIN_DOLOOP
10795 Take an instruction in @var{insn} and return NULL if it is valid within a
10796 low-overhead loop, otherwise return a string explaining why doloop
10797 could not be applied.
10799 Many targets use special registers for low-overhead looping. For any
10800 instruction that clobbers these this function should return a string indicating
10801 the reason why the doloop could not be applied.
10802 By default, the RTL loop optimizer does not use a present doloop pattern for
10803 loops containing function calls or branch on table instructions.
10806 @hook TARGET_LEGITIMATE_COMBINED_INSN
10808 @defmac MD_CAN_REDIRECT_BRANCH (@var{branch1}, @var{branch2})
10810 Take a branch insn in @var{branch1} and another in @var{branch2}.
10811 Return true if redirecting @var{branch1} to the destination of
10812 @var{branch2} is possible.
10814 On some targets, branches may have a limited range. Optimizing the
10815 filling of delay slots can result in branches being redirected, and this
10816 may in turn cause a branch offset to overflow.
10819 @hook TARGET_CAN_FOLLOW_JUMP
10821 @hook TARGET_COMMUTATIVE_P
10822 This target hook returns @code{true} if @var{x} is considered to be commutative.
10823 Usually, this is just COMMUTATIVE_P (@var{x}), but the HP PA doesn't consider
10824 PLUS to be commutative inside a MEM@. @var{outer_code} is the rtx code
10825 of the enclosing rtl, if known, otherwise it is UNKNOWN.
10828 @hook TARGET_ALLOCATE_INITIAL_VALUE
10830 When the initial value of a hard register has been copied in a pseudo
10831 register, it is often not necessary to actually allocate another register
10832 to this pseudo register, because the original hard register or a stack slot
10833 it has been saved into can be used. @code{TARGET_ALLOCATE_INITIAL_VALUE}
10834 is called at the start of register allocation once for each hard register
10835 that had its initial value copied by using
10836 @code{get_func_hard_reg_initial_val} or @code{get_hard_reg_initial_val}.
10837 Possible values are @code{NULL_RTX}, if you don't want
10838 to do any special allocation, a @code{REG} rtx---that would typically be
10839 the hard register itself, if it is known not to be clobbered---or a
10841 If you are returning a @code{MEM}, this is only a hint for the allocator;
10842 it might decide to use another register anyways.
10843 You may use @code{current_function_is_leaf} or
10844 @code{REG_N_SETS} in the hook to determine if the hard
10845 register in question will not be clobbered.
10846 The default value of this hook is @code{NULL}, which disables any special
10850 @hook TARGET_UNSPEC_MAY_TRAP_P
10851 This target hook returns nonzero if @var{x}, an @code{unspec} or
10852 @code{unspec_volatile} operation, might cause a trap. Targets can use
10853 this hook to enhance precision of analysis for @code{unspec} and
10854 @code{unspec_volatile} operations. You may call @code{may_trap_p_1}
10855 to analyze inner elements of @var{x} in which case @var{flags} should be
10859 @hook TARGET_SET_CURRENT_FUNCTION
10860 The compiler invokes this hook whenever it changes its current function
10861 context (@code{cfun}). You can define this function if
10862 the back end needs to perform any initialization or reset actions on a
10863 per-function basis. For example, it may be used to implement function
10864 attributes that affect register usage or code generation patterns.
10865 The argument @var{decl} is the declaration for the new function context,
10866 and may be null to indicate that the compiler has left a function context
10867 and is returning to processing at the top level.
10868 The default hook function does nothing.
10870 GCC sets @code{cfun} to a dummy function context during initialization of
10871 some parts of the back end. The hook function is not invoked in this
10872 situation; you need not worry about the hook being invoked recursively,
10873 or when the back end is in a partially-initialized state.
10874 @code{cfun} might be @code{NULL} to indicate processing at top level,
10875 outside of any function scope.
10878 @defmac TARGET_OBJECT_SUFFIX
10879 Define this macro to be a C string representing the suffix for object
10880 files on your target machine. If you do not define this macro, GCC will
10881 use @samp{.o} as the suffix for object files.
10884 @defmac TARGET_EXECUTABLE_SUFFIX
10885 Define this macro to be a C string representing the suffix to be
10886 automatically added to executable files on your target machine. If you
10887 do not define this macro, GCC will use the null string as the suffix for
10891 @defmac COLLECT_EXPORT_LIST
10892 If defined, @code{collect2} will scan the individual object files
10893 specified on its command line and create an export list for the linker.
10894 Define this macro for systems like AIX, where the linker discards
10895 object files that are not referenced from @code{main} and uses export
10899 @defmac MODIFY_JNI_METHOD_CALL (@var{mdecl})
10900 Define this macro to a C expression representing a variant of the
10901 method call @var{mdecl}, if Java Native Interface (JNI) methods
10902 must be invoked differently from other methods on your target.
10903 For example, on 32-bit Microsoft Windows, JNI methods must be invoked using
10904 the @code{stdcall} calling convention and this macro is then
10905 defined as this expression:
10908 build_type_attribute_variant (@var{mdecl},
10910 (get_identifier ("stdcall"),
10915 @hook TARGET_CANNOT_MODIFY_JUMPS_P
10916 This target hook returns @code{true} past the point in which new jump
10917 instructions could be created. On machines that require a register for
10918 every jump such as the SHmedia ISA of SH5, this point would typically be
10919 reload, so this target hook should be defined to a function such as:
10923 cannot_modify_jumps_past_reload_p ()
10925 return (reload_completed || reload_in_progress);
10930 @hook TARGET_BRANCH_TARGET_REGISTER_CLASS
10931 This target hook returns a register class for which branch target register
10932 optimizations should be applied. All registers in this class should be
10933 usable interchangeably. After reload, registers in this class will be
10934 re-allocated and loads will be hoisted out of loops and be subjected
10935 to inter-block scheduling.
10938 @hook TARGET_BRANCH_TARGET_REGISTER_CALLEE_SAVED
10939 Branch target register optimization will by default exclude callee-saved
10941 that are not already live during the current function; if this target hook
10942 returns true, they will be included. The target code must than make sure
10943 that all target registers in the class returned by
10944 @samp{TARGET_BRANCH_TARGET_REGISTER_CLASS} that might need saving are
10945 saved. @var{after_prologue_epilogue_gen} indicates if prologues and
10946 epilogues have already been generated. Note, even if you only return
10947 true when @var{after_prologue_epilogue_gen} is false, you still are likely
10948 to have to make special provisions in @code{INITIAL_ELIMINATION_OFFSET}
10949 to reserve space for caller-saved target registers.
10952 @hook TARGET_HAVE_CONDITIONAL_EXECUTION
10953 This target hook returns true if the target supports conditional execution.
10954 This target hook is required only when the target has several different
10955 modes and they have different conditional execution capability, such as ARM.
10958 @hook TARGET_LOOP_UNROLL_ADJUST
10959 This target hook returns a new value for the number of times @var{loop}
10960 should be unrolled. The parameter @var{nunroll} is the number of times
10961 the loop is to be unrolled. The parameter @var{loop} is a pointer to
10962 the loop, which is going to be checked for unrolling. This target hook
10963 is required only when the target has special constraints like maximum
10964 number of memory accesses.
10967 @defmac POWI_MAX_MULTS
10968 If defined, this macro is interpreted as a signed integer C expression
10969 that specifies the maximum number of floating point multiplications
10970 that should be emitted when expanding exponentiation by an integer
10971 constant inline. When this value is defined, exponentiation requiring
10972 more than this number of multiplications is implemented by calling the
10973 system library's @code{pow}, @code{powf} or @code{powl} routines.
10974 The default value places no upper bound on the multiplication count.
10977 @deftypefn Macro void TARGET_EXTRA_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
10978 This target hook should register any extra include files for the
10979 target. The parameter @var{stdinc} indicates if normal include files
10980 are present. The parameter @var{sysroot} is the system root directory.
10981 The parameter @var{iprefix} is the prefix for the gcc directory.
10984 @deftypefn Macro void TARGET_EXTRA_PRE_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
10985 This target hook should register any extra include files for the
10986 target before any standard headers. The parameter @var{stdinc}
10987 indicates if normal include files are present. The parameter
10988 @var{sysroot} is the system root directory. The parameter
10989 @var{iprefix} is the prefix for the gcc directory.
10992 @deftypefn Macro void TARGET_OPTF (char *@var{path})
10993 This target hook should register special include paths for the target.
10994 The parameter @var{path} is the include to register. On Darwin
10995 systems, this is used for Framework includes, which have semantics
10996 that are different from @option{-I}.
10999 @defmac bool TARGET_USE_LOCAL_THUNK_ALIAS_P (tree @var{fndecl})
11000 This target macro returns @code{true} if it is safe to use a local alias
11001 for a virtual function @var{fndecl} when constructing thunks,
11002 @code{false} otherwise. By default, the macro returns @code{true} for all
11003 functions, if a target supports aliases (i.e.@: defines
11004 @code{ASM_OUTPUT_DEF}), @code{false} otherwise,
11007 @defmac TARGET_FORMAT_TYPES
11008 If defined, this macro is the name of a global variable containing
11009 target-specific format checking information for the @option{-Wformat}
11010 option. The default is to have no target-specific format checks.
11013 @defmac TARGET_N_FORMAT_TYPES
11014 If defined, this macro is the number of entries in
11015 @code{TARGET_FORMAT_TYPES}.
11018 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES
11019 If defined, this macro is the name of a global variable containing
11020 target-specific format overrides for the @option{-Wformat} option. The
11021 default is to have no target-specific format overrides. If defined,
11022 @code{TARGET_FORMAT_TYPES} must be defined, too.
11025 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES_COUNT
11026 If defined, this macro specifies the number of entries in
11027 @code{TARGET_OVERRIDES_FORMAT_ATTRIBUTES}.
11030 @defmac TARGET_OVERRIDES_FORMAT_INIT
11031 If defined, this macro specifies the optional initialization
11032 routine for target specific customizations of the system printf
11033 and scanf formatter settings.
11036 @hook TARGET_RELAXED_ORDERING
11037 If set to @code{true}, means that the target's memory model does not
11038 guarantee that loads which do not depend on one another will access
11039 main memory in the order of the instruction stream; if ordering is
11040 important, an explicit memory barrier must be used. This is true of
11041 many recent processors which implement a policy of ``relaxed,''
11042 ``weak,'' or ``release'' memory consistency, such as Alpha, PowerPC,
11043 and ia64. The default is @code{false}.
11046 @hook TARGET_INVALID_ARG_FOR_UNPROTOTYPED_FN
11047 If defined, this macro returns the diagnostic message when it is
11048 illegal to pass argument @var{val} to function @var{funcdecl}
11049 with prototype @var{typelist}.
11052 @hook TARGET_INVALID_CONVERSION
11053 If defined, this macro returns the diagnostic message when it is
11054 invalid to convert from @var{fromtype} to @var{totype}, or @code{NULL}
11055 if validity should be determined by the front end.
11058 @hook TARGET_INVALID_UNARY_OP
11059 If defined, this macro returns the diagnostic message when it is
11060 invalid to apply operation @var{op} (where unary plus is denoted by
11061 @code{CONVERT_EXPR}) to an operand of type @var{type}, or @code{NULL}
11062 if validity should be determined by the front end.
11065 @hook TARGET_INVALID_BINARY_OP
11066 If defined, this macro returns the diagnostic message when it is
11067 invalid to apply operation @var{op} to operands of types @var{type1}
11068 and @var{type2}, or @code{NULL} if validity should be determined by
11072 @hook TARGET_INVALID_PARAMETER_TYPE
11073 If defined, this macro returns the diagnostic message when it is
11074 invalid for functions to include parameters of type @var{type},
11075 or @code{NULL} if validity should be determined by
11076 the front end. This is currently used only by the C and C++ front ends.
11079 @hook TARGET_INVALID_RETURN_TYPE
11080 If defined, this macro returns the diagnostic message when it is
11081 invalid for functions to have return type @var{type},
11082 or @code{NULL} if validity should be determined by
11083 the front end. This is currently used only by the C and C++ front ends.
11086 @hook TARGET_PROMOTED_TYPE
11087 If defined, this target hook returns the type to which values of
11088 @var{type} should be promoted when they appear in expressions,
11089 analogous to the integer promotions, or @code{NULL_TREE} to use the
11090 front end's normal promotion rules. This hook is useful when there are
11091 target-specific types with special promotion rules.
11092 This is currently used only by the C and C++ front ends.
11095 @hook TARGET_CONVERT_TO_TYPE
11096 If defined, this hook returns the result of converting @var{expr} to
11097 @var{type}. It should return the converted expression,
11098 or @code{NULL_TREE} to apply the front end's normal conversion rules.
11099 This hook is useful when there are target-specific types with special
11101 This is currently used only by the C and C++ front ends.
11104 @defmac TARGET_USE_JCR_SECTION
11105 This macro determines whether to use the JCR section to register Java
11106 classes. By default, TARGET_USE_JCR_SECTION is defined to 1 if both
11107 SUPPORTS_WEAK and TARGET_HAVE_NAMED_SECTIONS are true, else 0.
11111 This macro determines the size of the objective C jump buffer for the
11112 NeXT runtime. By default, OBJC_JBLEN is defined to an innocuous value.
11115 @defmac LIBGCC2_UNWIND_ATTRIBUTE
11116 Define this macro if any target-specific attributes need to be attached
11117 to the functions in @file{libgcc} that provide low-level support for
11118 call stack unwinding. It is used in declarations in @file{unwind-generic.h}
11119 and the associated definitions of those functions.
11122 @hook TARGET_UPDATE_STACK_BOUNDARY
11123 Define this macro to update the current function stack boundary if
11127 @hook TARGET_GET_DRAP_RTX
11128 This hook should return an rtx for Dynamic Realign Argument Pointer (DRAP) if a
11129 different argument pointer register is needed to access the function's
11130 argument list due to stack realignment. Return @code{NULL} if no DRAP
11134 @hook TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS
11135 When optimization is disabled, this hook indicates whether or not
11136 arguments should be allocated to stack slots. Normally, GCC allocates
11137 stacks slots for arguments when not optimizing in order to make
11138 debugging easier. However, when a function is declared with
11139 @code{__attribute__((naked))}, there is no stack frame, and the compiler
11140 cannot safely move arguments from the registers in which they are passed
11141 to the stack. Therefore, this hook should return true in general, but
11142 false for naked functions. The default implementation always returns true.
11145 @hook TARGET_CONST_ANCHOR
11146 On some architectures it can take multiple instructions to synthesize
11147 a constant. If there is another constant already in a register that
11148 is close enough in value then it is preferable that the new constant
11149 is computed from this register using immediate addition or
11150 subtraction. We accomplish this through CSE. Besides the value of
11151 the constant we also add a lower and an upper constant anchor to the
11152 available expressions. These are then queried when encountering new
11153 constants. The anchors are computed by rounding the constant up and
11154 down to a multiple of the value of @code{TARGET_CONST_ANCHOR}.
11155 @code{TARGET_CONST_ANCHOR} should be the maximum positive value
11156 accepted by immediate-add plus one. We currently assume that the
11157 value of @code{TARGET_CONST_ANCHOR} is a power of 2. For example, on
11158 MIPS, where add-immediate takes a 16-bit signed value,
11159 @code{TARGET_CONST_ANCHOR} is set to @samp{0x8000}. The default value
11160 is zero, which disables this optimization. @end deftypevr
11162 @hook TARGET_MEMMODEL_CHECK
11163 Validate target specific memory model mask bits. When NULL no target specific
11164 memory model bits are allowed.
11167 @hook TARGET_ATOMIC_TEST_AND_SET_TRUEVAL