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
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 @defmac LINK_ELIMINATE_DUPLICATE_LDIRECTORIES
391 A nonzero value causes @command{collect2} to remove duplicate @option{-L@var{directory}} search
392 directories from linking commands. Do not give it a nonzero value if
393 removing duplicate search directories changes the linker's semantics.
396 @deftypevr {Common Target Hook} bool TARGET_ALWAYS_STRIP_DOTDOT
397 True if @file{..} components should always be removed from directory names computed relative to GCC's internal directories, false (default) if such components should be preserved and directory names containing them passed to other tools such as the linker.
400 @defmac MULTILIB_DEFAULTS
401 Define this macro as a C expression for the initializer of an array of
402 string to tell the driver program which options are defaults for this
403 target and thus do not need to be handled specially when using
404 @code{MULTILIB_OPTIONS}.
406 Do not define this macro if @code{MULTILIB_OPTIONS} is not defined in
407 the target makefile fragment or if none of the options listed in
408 @code{MULTILIB_OPTIONS} are set by default.
409 @xref{Target Fragment}.
412 @defmac RELATIVE_PREFIX_NOT_LINKDIR
413 Define this macro to tell @command{gcc} that it should only translate
414 a @option{-B} prefix into a @option{-L} linker option if the prefix
415 indicates an absolute file name.
418 @defmac MD_EXEC_PREFIX
419 If defined, this macro is an additional prefix to try after
420 @code{STANDARD_EXEC_PREFIX}. @code{MD_EXEC_PREFIX} is not searched
421 when the compiler is built as a cross
422 compiler. If you define @code{MD_EXEC_PREFIX}, then be sure to add it
423 to the list of directories used to find the assembler in @file{configure.in}.
426 @defmac STANDARD_STARTFILE_PREFIX
427 Define this macro as a C string constant if you wish to override the
428 standard choice of @code{libdir} as the default prefix to
429 try when searching for startup files such as @file{crt0.o}.
430 @code{STANDARD_STARTFILE_PREFIX} is not searched when the compiler
431 is built as a cross compiler.
434 @defmac STANDARD_STARTFILE_PREFIX_1
435 Define this macro as a C string constant if you wish to override the
436 standard choice of @code{/lib} as a prefix to try after the default prefix
437 when searching for startup files such as @file{crt0.o}.
438 @code{STANDARD_STARTFILE_PREFIX_1} is not searched when the compiler
439 is built as a cross compiler.
442 @defmac STANDARD_STARTFILE_PREFIX_2
443 Define this macro as a C string constant if you wish to override the
444 standard choice of @code{/lib} as yet another prefix to try after the
445 default prefix when searching for startup files such as @file{crt0.o}.
446 @code{STANDARD_STARTFILE_PREFIX_2} is not searched when the compiler
447 is built as a cross compiler.
450 @defmac MD_STARTFILE_PREFIX
451 If defined, this macro supplies an additional prefix to try after the
452 standard prefixes. @code{MD_EXEC_PREFIX} is not searched when the
453 compiler is built as a cross compiler.
456 @defmac MD_STARTFILE_PREFIX_1
457 If defined, this macro supplies yet another prefix to try after the
458 standard prefixes. It is not searched when the compiler is built as a
462 @defmac INIT_ENVIRONMENT
463 Define this macro as a C string constant if you wish to set environment
464 variables for programs called by the driver, such as the assembler and
465 loader. The driver passes the value of this macro to @code{putenv} to
466 initialize the necessary environment variables.
469 @defmac LOCAL_INCLUDE_DIR
470 Define this macro as a C string constant if you wish to override the
471 standard choice of @file{/usr/local/include} as the default prefix to
472 try when searching for local header files. @code{LOCAL_INCLUDE_DIR}
473 comes before @code{NATIVE_SYSTEM_HEADER_DIR} (set in
474 @file{config.gcc}, normally @file{/usr/include}) in the search order.
476 Cross compilers do not search either @file{/usr/local/include} or its
480 @defmac NATIVE_SYSTEM_HEADER_COMPONENT
481 The ``component'' corresponding to @code{NATIVE_SYSTEM_HEADER_DIR}.
482 See @code{INCLUDE_DEFAULTS}, below, for the description of components.
483 If you do not define this macro, no component is used.
486 @defmac INCLUDE_DEFAULTS
487 Define this macro if you wish to override the entire default search path
488 for include files. For a native compiler, the default search path
489 usually consists of @code{GCC_INCLUDE_DIR}, @code{LOCAL_INCLUDE_DIR},
490 @code{GPLUSPLUS_INCLUDE_DIR}, and
491 @code{NATIVE_SYSTEM_HEADER_DIR}. In addition, @code{GPLUSPLUS_INCLUDE_DIR}
492 and @code{GCC_INCLUDE_DIR} are defined automatically by @file{Makefile},
493 and specify private search areas for GCC@. The directory
494 @code{GPLUSPLUS_INCLUDE_DIR} is used only for C++ programs.
496 The definition should be an initializer for an array of structures.
497 Each array element should have four elements: the directory name (a
498 string constant), the component name (also a string constant), a flag
499 for C++-only directories,
500 and a flag showing that the includes in the directory don't need to be
501 wrapped in @code{extern @samp{C}} when compiling C++. Mark the end of
502 the array with a null element.
504 The component name denotes what GNU package the include file is part of,
505 if any, in all uppercase letters. For example, it might be @samp{GCC}
506 or @samp{BINUTILS}. If the package is part of a vendor-supplied
507 operating system, code the component name as @samp{0}.
509 For example, here is the definition used for VAX/VMS:
512 #define INCLUDE_DEFAULTS \
514 @{ "GNU_GXX_INCLUDE:", "G++", 1, 1@}, \
515 @{ "GNU_CC_INCLUDE:", "GCC", 0, 0@}, \
516 @{ "SYS$SYSROOT:[SYSLIB.]", 0, 0, 0@}, \
523 Here is the order of prefixes tried for exec files:
527 Any prefixes specified by the user with @option{-B}.
530 The environment variable @code{GCC_EXEC_PREFIX} or, if @code{GCC_EXEC_PREFIX}
531 is not set and the compiler has not been installed in the configure-time
532 @var{prefix}, the location in which the compiler has actually been installed.
535 The directories specified by the environment variable @code{COMPILER_PATH}.
538 The macro @code{STANDARD_EXEC_PREFIX}, if the compiler has been installed
539 in the configured-time @var{prefix}.
542 The location @file{/usr/libexec/gcc/}, but only if this is a native compiler.
545 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
548 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
552 Here is the order of prefixes tried for startfiles:
556 Any prefixes specified by the user with @option{-B}.
559 The environment variable @code{GCC_EXEC_PREFIX} or its automatically determined
560 value based on the installed toolchain location.
563 The directories specified by the environment variable @code{LIBRARY_PATH}
564 (or port-specific name; native only, cross compilers do not use this).
567 The macro @code{STANDARD_EXEC_PREFIX}, but only if the toolchain is installed
568 in the configured @var{prefix} or this is a native compiler.
571 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
574 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
578 The macro @code{MD_STARTFILE_PREFIX}, if defined, but only if this is a
579 native compiler, or we have a target system root.
582 The macro @code{MD_STARTFILE_PREFIX_1}, if defined, but only if this is a
583 native compiler, or we have a target system root.
586 The macro @code{STANDARD_STARTFILE_PREFIX}, with any sysroot modifications.
587 If this path is relative it will be prefixed by @code{GCC_EXEC_PREFIX} and
588 the machine suffix or @code{STANDARD_EXEC_PREFIX} and the machine suffix.
591 The macro @code{STANDARD_STARTFILE_PREFIX_1}, but only if this is a native
592 compiler, or we have a target system root. The default for this macro is
596 The macro @code{STANDARD_STARTFILE_PREFIX_2}, but only if this is a native
597 compiler, or we have a target system root. The default for this macro is
601 @node Run-time Target
602 @section Run-time Target Specification
603 @cindex run-time target specification
604 @cindex predefined macros
605 @cindex target specifications
607 @c prevent bad page break with this line
608 Here are run-time target specifications.
610 @defmac TARGET_CPU_CPP_BUILTINS ()
611 This function-like macro expands to a block of code that defines
612 built-in preprocessor macros and assertions for the target CPU, using
613 the functions @code{builtin_define}, @code{builtin_define_std} and
614 @code{builtin_assert}. When the front end
615 calls this macro it provides a trailing semicolon, and since it has
616 finished command line option processing your code can use those
619 @code{builtin_assert} takes a string in the form you pass to the
620 command-line option @option{-A}, such as @code{cpu=mips}, and creates
621 the assertion. @code{builtin_define} takes a string in the form
622 accepted by option @option{-D} and unconditionally defines the macro.
624 @code{builtin_define_std} takes a string representing the name of an
625 object-like macro. If it doesn't lie in the user's namespace,
626 @code{builtin_define_std} defines it unconditionally. Otherwise, it
627 defines a version with two leading underscores, and another version
628 with two leading and trailing underscores, and defines the original
629 only if an ISO standard was not requested on the command line. For
630 example, passing @code{unix} defines @code{__unix}, @code{__unix__}
631 and possibly @code{unix}; passing @code{_mips} defines @code{__mips},
632 @code{__mips__} and possibly @code{_mips}, and passing @code{_ABI64}
633 defines only @code{_ABI64}.
635 You can also test for the C dialect being compiled. The variable
636 @code{c_language} is set to one of @code{clk_c}, @code{clk_cplusplus}
637 or @code{clk_objective_c}. Note that if we are preprocessing
638 assembler, this variable will be @code{clk_c} but the function-like
639 macro @code{preprocessing_asm_p()} will return true, so you might want
640 to check for that first. If you need to check for strict ANSI, the
641 variable @code{flag_iso} can be used. The function-like macro
642 @code{preprocessing_trad_p()} can be used to check for traditional
646 @defmac TARGET_OS_CPP_BUILTINS ()
647 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
648 and is used for the target operating system instead.
651 @defmac TARGET_OBJFMT_CPP_BUILTINS ()
652 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
653 and is used for the target object format. @file{elfos.h} uses this
654 macro to define @code{__ELF__}, so you probably do not need to define
658 @deftypevar {extern int} target_flags
659 This variable is declared in @file{options.h}, which is included before
660 any target-specific headers.
663 @deftypevr {Common Target Hook} int TARGET_DEFAULT_TARGET_FLAGS
664 This variable specifies the initial value of @code{target_flags}.
665 Its default setting is 0.
668 @cindex optional hardware or system features
669 @cindex features, optional, in system conventions
671 @deftypefn {Common Target Hook} bool TARGET_HANDLE_OPTION (struct gcc_options *@var{opts}, struct gcc_options *@var{opts_set}, const struct cl_decoded_option *@var{decoded}, location_t @var{loc})
672 This hook is called whenever the user specifies one of the
673 target-specific options described by the @file{.opt} definition files
674 (@pxref{Options}). It has the opportunity to do some option-specific
675 processing and should return true if the option is valid. The default
676 definition does nothing but return true.
678 @var{decoded} specifies the option and its arguments. @var{opts} and
679 @var{opts_set} are the @code{gcc_options} structures to be used for
680 storing option state, and @var{loc} is the location at which the
681 option was passed (@code{UNKNOWN_LOCATION} except for options passed
685 @deftypefn {C Target Hook} bool TARGET_HANDLE_C_OPTION (size_t @var{code}, const char *@var{arg}, int @var{value})
686 This target hook is called whenever the user specifies one of the
687 target-specific C language family options described by the @file{.opt}
688 definition files(@pxref{Options}). It has the opportunity to do some
689 option-specific processing and should return true if the option is
690 valid. The arguments are like for @code{TARGET_HANDLE_OPTION}. The
691 default definition does nothing but return false.
693 In general, you should use @code{TARGET_HANDLE_OPTION} to handle
694 options. However, if processing an option requires routines that are
695 only available in the C (and related language) front ends, then you
696 should use @code{TARGET_HANDLE_C_OPTION} instead.
699 @deftypefn {C Target Hook} tree TARGET_OBJC_CONSTRUCT_STRING_OBJECT (tree @var{string})
700 Targets may provide a string object type that can be used within and between C, C++ and their respective Objective-C dialects. A string object might, for example, embed encoding and length information. These objects are considered opaque to the compiler and handled as references. An ideal implementation makes the composition of the string object match that of the Objective-C @code{NSString} (@code{NXString} for GNUStep), allowing efficient interworking between C-only and Objective-C code. If a target implements string objects then this hook should return a reference to such an object constructed from the normal `C' string representation provided in @var{string}. At present, the hook is used by Objective-C only, to obtain a common-format string object when the target provides one.
703 @deftypefn {C Target Hook} bool TARGET_STRING_OBJECT_REF_TYPE_P (const_tree @var{stringref})
704 If a target implements string objects then this hook should return @code{true} if @var{stringref} is a valid reference to such an object.
707 @deftypefn {C Target Hook} void TARGET_CHECK_STRING_OBJECT_FORMAT_ARG (tree @var{format_arg}, tree @var{args_list})
708 If a target implements string objects then this hook should should provide a facility to check the function arguments in @var{args_list} against the format specifiers in @var{format_arg} where the type of @var{format_arg} is one recognized as a valid string reference type.
711 @deftypefn {Target Hook} void TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE (void)
712 This target function is similar to the hook @code{TARGET_OPTION_OVERRIDE}
713 but is called when the optimize level is changed via an attribute or
714 pragma or when it is reset at the end of the code affected by the
715 attribute or pragma. It is not called at the beginning of compilation
716 when @code{TARGET_OPTION_OVERRIDE} is called so if you want to perform these
717 actions then, you should have @code{TARGET_OPTION_OVERRIDE} call
718 @code{TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE}.
721 @defmac C_COMMON_OVERRIDE_OPTIONS
722 This is similar to the @code{TARGET_OPTION_OVERRIDE} hook
723 but is only used in the C
724 language frontends (C, Objective-C, C++, Objective-C++) and so can be
725 used to alter option flag variables which only exist in those
729 @deftypevr {Common Target Hook} {const struct default_options *} TARGET_OPTION_OPTIMIZATION_TABLE
730 Some machines may desire to change what optimizations are performed for
731 various optimization levels. This variable, if defined, describes
732 options to enable at particular sets of optimization levels. These
733 options are processed once
734 just after the optimization level is determined and before the remainder
735 of the command options have been parsed, so may be overridden by other
736 options passed explicitly.
738 This processing is run once at program startup and when the optimization
739 options are changed via @code{#pragma GCC optimize} or by using the
740 @code{optimize} attribute.
743 @deftypefn {Common Target Hook} void TARGET_OPTION_INIT_STRUCT (struct gcc_options *@var{opts})
744 Set target-dependent initial values of fields in @var{opts}.
747 @deftypefn {Common Target Hook} void TARGET_OPTION_DEFAULT_PARAMS (void)
748 Set target-dependent default values for @option{--param} settings, using calls to @code{set_default_param_value}.
751 @defmac SWITCHABLE_TARGET
752 Some targets need to switch between substantially different subtargets
753 during compilation. For example, the MIPS target has one subtarget for
754 the traditional MIPS architecture and another for MIPS16. Source code
755 can switch between these two subarchitectures using the @code{mips16}
756 and @code{nomips16} attributes.
758 Such subtargets can differ in things like the set of available
759 registers, the set of available instructions, the costs of various
760 operations, and so on. GCC caches a lot of this type of information
761 in global variables, and recomputing them for each subtarget takes a
762 significant amount of time. The compiler therefore provides a facility
763 for maintaining several versions of the global variables and quickly
764 switching between them; see @file{target-globals.h} for details.
766 Define this macro to 1 if your target needs this facility. The default
770 @node Per-Function Data
771 @section Defining data structures for per-function information.
772 @cindex per-function data
773 @cindex data structures
775 If the target needs to store information on a per-function basis, GCC
776 provides a macro and a couple of variables to allow this. Note, just
777 using statics to store the information is a bad idea, since GCC supports
778 nested functions, so you can be halfway through encoding one function
779 when another one comes along.
781 GCC defines a data structure called @code{struct function} which
782 contains all of the data specific to an individual function. This
783 structure contains a field called @code{machine} whose type is
784 @code{struct machine_function *}, which can be used by targets to point
785 to their own specific data.
787 If a target needs per-function specific data it should define the type
788 @code{struct machine_function} and also the macro @code{INIT_EXPANDERS}.
789 This macro should be used to initialize the function pointer
790 @code{init_machine_status}. This pointer is explained below.
792 One typical use of per-function, target specific data is to create an
793 RTX to hold the register containing the function's return address. This
794 RTX can then be used to implement the @code{__builtin_return_address}
795 function, for level 0.
797 Note---earlier implementations of GCC used a single data area to hold
798 all of the per-function information. Thus when processing of a nested
799 function began the old per-function data had to be pushed onto a
800 stack, and when the processing was finished, it had to be popped off the
801 stack. GCC used to provide function pointers called
802 @code{save_machine_status} and @code{restore_machine_status} to handle
803 the saving and restoring of the target specific information. Since the
804 single data area approach is no longer used, these pointers are no
807 @defmac INIT_EXPANDERS
808 Macro called to initialize any target specific information. This macro
809 is called once per function, before generation of any RTL has begun.
810 The intention of this macro is to allow the initialization of the
811 function pointer @code{init_machine_status}.
814 @deftypevar {void (*)(struct function *)} init_machine_status
815 If this function pointer is non-@code{NULL} it will be called once per
816 function, before function compilation starts, in order to allow the
817 target to perform any target specific initialization of the
818 @code{struct function} structure. It is intended that this would be
819 used to initialize the @code{machine} of that structure.
821 @code{struct machine_function} structures are expected to be freed by GC@.
822 Generally, any memory that they reference must be allocated by using
823 GC allocation, including the structure itself.
827 @section Storage Layout
828 @cindex storage layout
830 Note that the definitions of the macros in this table which are sizes or
831 alignments measured in bits do not need to be constant. They can be C
832 expressions that refer to static variables, such as the @code{target_flags}.
833 @xref{Run-time Target}.
835 @defmac BITS_BIG_ENDIAN
836 Define this macro to have the value 1 if the most significant bit in a
837 byte has the lowest number; otherwise define it to have the value zero.
838 This means that bit-field instructions count from the most significant
839 bit. If the machine has no bit-field instructions, then this must still
840 be defined, but it doesn't matter which value it is defined to. This
841 macro need not be a constant.
843 This macro does not affect the way structure fields are packed into
844 bytes or words; that is controlled by @code{BYTES_BIG_ENDIAN}.
847 @defmac BYTES_BIG_ENDIAN
848 Define this macro to have the value 1 if the most significant byte in a
849 word has the lowest number. This macro need not be a constant.
852 @defmac WORDS_BIG_ENDIAN
853 Define this macro to have the value 1 if, in a multiword object, the
854 most significant word has the lowest number. This applies to both
855 memory locations and registers; see @code{REG_WORDS_BIG_ENDIAN} if the
856 order of words in memory is not the same as the order in registers. This
857 macro need not be a constant.
860 @defmac REG_WORDS_BIG_ENDIAN
861 On some machines, the order of words in a multiword object differs between
862 registers in memory. In such a situation, define this macro to describe
863 the order of words in a register. The macro @code{WORDS_BIG_ENDIAN} controls
864 the order of words in memory.
867 @defmac FLOAT_WORDS_BIG_ENDIAN
868 Define this macro to have the value 1 if @code{DFmode}, @code{XFmode} or
869 @code{TFmode} floating point numbers are stored in memory with the word
870 containing the sign bit at the lowest address; otherwise define it to
871 have the value 0. This macro need not be a constant.
873 You need not define this macro if the ordering is the same as for
877 @defmac BITS_PER_UNIT
878 Define this macro to be the number of bits in an addressable storage
879 unit (byte). If you do not define this macro the default is 8.
882 @defmac BITS_PER_WORD
883 Number of bits in a word. If you do not define this macro, the default
884 is @code{BITS_PER_UNIT * UNITS_PER_WORD}.
887 @defmac MAX_BITS_PER_WORD
888 Maximum number of bits in a word. If this is undefined, the default is
889 @code{BITS_PER_WORD}. Otherwise, it is the constant value that is the
890 largest value that @code{BITS_PER_WORD} can have at run-time.
893 @defmac UNITS_PER_WORD
894 Number of storage units in a word; normally the size of a general-purpose
895 register, a power of two from 1 or 8.
898 @defmac MIN_UNITS_PER_WORD
899 Minimum number of units in a word. If this is undefined, the default is
900 @code{UNITS_PER_WORD}. Otherwise, it is the constant value that is the
901 smallest value that @code{UNITS_PER_WORD} can have at run-time.
905 Width of a pointer, in bits. You must specify a value no wider than the
906 width of @code{Pmode}. If it is not equal to the width of @code{Pmode},
907 you must define @code{POINTERS_EXTEND_UNSIGNED}. If you do not specify
908 a value the default is @code{BITS_PER_WORD}.
911 @defmac POINTERS_EXTEND_UNSIGNED
912 A C expression that determines how pointers should be extended from
913 @code{ptr_mode} to either @code{Pmode} or @code{word_mode}. It is
914 greater than zero if pointers should be zero-extended, zero if they
915 should be sign-extended, and negative if some other sort of conversion
916 is needed. In the last case, the extension is done by the target's
917 @code{ptr_extend} instruction.
919 You need not define this macro if the @code{ptr_mode}, @code{Pmode}
920 and @code{word_mode} are all the same width.
923 @defmac PROMOTE_MODE (@var{m}, @var{unsignedp}, @var{type})
924 A macro to update @var{m} and @var{unsignedp} when an object whose type
925 is @var{type} and which has the specified mode and signedness is to be
926 stored in a register. This macro is only called when @var{type} is a
929 On most RISC machines, which only have operations that operate on a full
930 register, define this macro to set @var{m} to @code{word_mode} if
931 @var{m} is an integer mode narrower than @code{BITS_PER_WORD}. In most
932 cases, only integer modes should be widened because wider-precision
933 floating-point operations are usually more expensive than their narrower
936 For most machines, the macro definition does not change @var{unsignedp}.
937 However, some machines, have instructions that preferentially handle
938 either signed or unsigned quantities of certain modes. For example, on
939 the DEC Alpha, 32-bit loads from memory and 32-bit add instructions
940 sign-extend the result to 64 bits. On such machines, set
941 @var{unsignedp} according to which kind of extension is more efficient.
943 Do not define this macro if it would never modify @var{m}.
946 @deftypefn {Target Hook} {enum machine_mode} TARGET_PROMOTE_FUNCTION_MODE (const_tree @var{type}, enum machine_mode @var{mode}, int *@var{punsignedp}, const_tree @var{funtype}, int @var{for_return})
947 Like @code{PROMOTE_MODE}, but it is applied to outgoing function arguments or
948 function return values. The target hook should return the new mode
949 and possibly change @code{*@var{punsignedp}} if the promotion should
950 change signedness. This function is called only for scalar @emph{or
953 @var{for_return} allows to distinguish the promotion of arguments and
954 return values. If it is @code{1}, a return value is being promoted and
955 @code{TARGET_FUNCTION_VALUE} must perform the same promotions done here.
956 If it is @code{2}, the returned mode should be that of the register in
957 which an incoming parameter is copied, or the outgoing result is computed;
958 then the hook should return the same mode as @code{promote_mode}, though
959 the signedness may be different.
961 @var{type} can be NULL when promoting function arguments of libcalls.
963 The default is to not promote arguments and return values. You can
964 also define the hook to @code{default_promote_function_mode_always_promote}
965 if you would like to apply the same rules given by @code{PROMOTE_MODE}.
968 @defmac PARM_BOUNDARY
969 Normal alignment required for function parameters on the stack, in
970 bits. All stack parameters receive at least this much alignment
971 regardless of data type. On most machines, this is the same as the
975 @defmac STACK_BOUNDARY
976 Define this macro to the minimum alignment enforced by hardware for the
977 stack pointer on this machine. The definition is a C expression for the
978 desired alignment (measured in bits). This value is used as a default
979 if @code{PREFERRED_STACK_BOUNDARY} is not defined. On most machines,
980 this should be the same as @code{PARM_BOUNDARY}.
983 @defmac PREFERRED_STACK_BOUNDARY
984 Define this macro if you wish to preserve a certain alignment for the
985 stack pointer, greater than what the hardware enforces. The definition
986 is a C expression for the desired alignment (measured in bits). This
987 macro must evaluate to a value equal to or larger than
988 @code{STACK_BOUNDARY}.
991 @defmac INCOMING_STACK_BOUNDARY
992 Define this macro if the incoming stack boundary may be different
993 from @code{PREFERRED_STACK_BOUNDARY}. This macro must evaluate
994 to a value equal to or larger than @code{STACK_BOUNDARY}.
997 @defmac FUNCTION_BOUNDARY
998 Alignment required for a function entry point, in bits.
1001 @defmac BIGGEST_ALIGNMENT
1002 Biggest alignment that any data type can require on this machine, in
1003 bits. Note that this is not the biggest alignment that is supported,
1004 just the biggest alignment that, when violated, may cause a fault.
1007 @defmac MALLOC_ABI_ALIGNMENT
1008 Alignment, in bits, a C conformant malloc implementation has to
1009 provide. If not defined, the default value is @code{BITS_PER_WORD}.
1012 @defmac ATTRIBUTE_ALIGNED_VALUE
1013 Alignment used by the @code{__attribute__ ((aligned))} construct. If
1014 not defined, the default value is @code{BIGGEST_ALIGNMENT}.
1017 @defmac MINIMUM_ATOMIC_ALIGNMENT
1018 If defined, the smallest alignment, in bits, that can be given to an
1019 object that can be referenced in one operation, without disturbing any
1020 nearby object. Normally, this is @code{BITS_PER_UNIT}, but may be larger
1021 on machines that don't have byte or half-word store operations.
1024 @defmac BIGGEST_FIELD_ALIGNMENT
1025 Biggest alignment that any structure or union field can require on this
1026 machine, in bits. If defined, this overrides @code{BIGGEST_ALIGNMENT} for
1027 structure and union fields only, unless the field alignment has been set
1028 by the @code{__attribute__ ((aligned (@var{n})))} construct.
1031 @defmac ADJUST_FIELD_ALIGN (@var{field}, @var{computed})
1032 An expression for the alignment of a structure field @var{field} if the
1033 alignment computed in the usual way (including applying of
1034 @code{BIGGEST_ALIGNMENT} and @code{BIGGEST_FIELD_ALIGNMENT} to the
1035 alignment) is @var{computed}. It overrides alignment only if the
1036 field alignment has not been set by the
1037 @code{__attribute__ ((aligned (@var{n})))} construct.
1040 @defmac MAX_STACK_ALIGNMENT
1041 Biggest stack alignment guaranteed by the backend. Use this macro
1042 to specify the maximum alignment of a variable on stack.
1044 If not defined, the default value is @code{STACK_BOUNDARY}.
1046 @c FIXME: The default should be @code{PREFERRED_STACK_BOUNDARY}.
1047 @c But the fix for PR 32893 indicates that we can only guarantee
1048 @c maximum stack alignment on stack up to @code{STACK_BOUNDARY}, not
1049 @c @code{PREFERRED_STACK_BOUNDARY}, if stack alignment isn't supported.
1052 @defmac MAX_OFILE_ALIGNMENT
1053 Biggest alignment supported by the object file format of this machine.
1054 Use this macro to limit the alignment which can be specified using the
1055 @code{__attribute__ ((aligned (@var{n})))} construct. If not defined,
1056 the default value is @code{BIGGEST_ALIGNMENT}.
1058 On systems that use ELF, the default (in @file{config/elfos.h}) is
1059 the largest supported 32-bit ELF section alignment representable on
1060 a 32-bit host e.g. @samp{(((unsigned HOST_WIDEST_INT) 1 << 28) * 8)}.
1061 On 32-bit ELF the largest supported section alignment in bits is
1062 @samp{(0x80000000 * 8)}, but this is not representable on 32-bit hosts.
1065 @defmac DATA_ALIGNMENT (@var{type}, @var{basic-align})
1066 If defined, a C expression to compute the alignment for a variable in
1067 the static store. @var{type} is the data type, and @var{basic-align} is
1068 the alignment that the object would ordinarily have. The value of this
1069 macro is used instead of that alignment to align the object.
1071 If this macro is not defined, then @var{basic-align} is used.
1074 One use of this macro is to increase alignment of medium-size data to
1075 make it all fit in fewer cache lines. Another is to cause character
1076 arrays to be word-aligned so that @code{strcpy} calls that copy
1077 constants to character arrays can be done inline.
1080 @defmac CONSTANT_ALIGNMENT (@var{constant}, @var{basic-align})
1081 If defined, a C expression to compute the alignment given to a constant
1082 that is being placed in memory. @var{constant} is the constant and
1083 @var{basic-align} is the alignment that the object would ordinarily
1084 have. The value of this macro is used instead of that alignment to
1087 If this macro is not defined, then @var{basic-align} is used.
1089 The typical use of this macro is to increase alignment for string
1090 constants to be word aligned so that @code{strcpy} calls that copy
1091 constants can be done inline.
1094 @defmac LOCAL_ALIGNMENT (@var{type}, @var{basic-align})
1095 If defined, a C expression to compute the alignment for a variable in
1096 the local store. @var{type} is the data type, and @var{basic-align} is
1097 the alignment that the object would ordinarily have. The value of this
1098 macro is used instead of that alignment to align the object.
1100 If this macro is not defined, then @var{basic-align} is used.
1102 One use of this macro is to increase alignment of medium-size data to
1103 make it all fit in fewer cache lines.
1105 If the value of this macro has a type, it should be an unsigned type.
1108 @defmac STACK_SLOT_ALIGNMENT (@var{type}, @var{mode}, @var{basic-align})
1109 If defined, a C expression to compute the alignment for stack slot.
1110 @var{type} is the data type, @var{mode} is the widest mode available,
1111 and @var{basic-align} is the alignment that the slot would ordinarily
1112 have. The value of this macro is used instead of that alignment to
1115 If this macro is not defined, then @var{basic-align} is used when
1116 @var{type} is @code{NULL}. Otherwise, @code{LOCAL_ALIGNMENT} will
1119 This macro is to set alignment of stack slot to the maximum alignment
1120 of all possible modes which the slot may have.
1122 If the value of this macro has a type, it should be an unsigned type.
1125 @defmac LOCAL_DECL_ALIGNMENT (@var{decl})
1126 If defined, a C expression to compute the alignment for a local
1127 variable @var{decl}.
1129 If this macro is not defined, then
1130 @code{LOCAL_ALIGNMENT (TREE_TYPE (@var{decl}), DECL_ALIGN (@var{decl}))}
1133 One use of this macro is to increase alignment of medium-size data to
1134 make it all fit in fewer cache lines.
1136 If the value of this macro has a type, it should be an unsigned type.
1139 @defmac MINIMUM_ALIGNMENT (@var{exp}, @var{mode}, @var{align})
1140 If defined, a C expression to compute the minimum required alignment
1141 for dynamic stack realignment purposes for @var{exp} (a type or decl),
1142 @var{mode}, assuming normal alignment @var{align}.
1144 If this macro is not defined, then @var{align} will be used.
1147 @defmac EMPTY_FIELD_BOUNDARY
1148 Alignment in bits to be given to a structure bit-field that follows an
1149 empty field such as @code{int : 0;}.
1151 If @code{PCC_BITFIELD_TYPE_MATTERS} is true, it overrides this macro.
1154 @defmac STRUCTURE_SIZE_BOUNDARY
1155 Number of bits which any structure or union's size must be a multiple of.
1156 Each structure or union's size is rounded up to a multiple of this.
1158 If you do not define this macro, the default is the same as
1159 @code{BITS_PER_UNIT}.
1162 @defmac STRICT_ALIGNMENT
1163 Define this macro to be the value 1 if instructions will fail to work
1164 if given data not on the nominal alignment. If instructions will merely
1165 go slower in that case, define this macro as 0.
1168 @defmac PCC_BITFIELD_TYPE_MATTERS
1169 Define this if you wish to imitate the way many other C compilers handle
1170 alignment of bit-fields and the structures that contain them.
1172 The behavior is that the type written for a named bit-field (@code{int},
1173 @code{short}, or other integer type) imposes an alignment for the entire
1174 structure, as if the structure really did contain an ordinary field of
1175 that type. In addition, the bit-field is placed within the structure so
1176 that it would fit within such a field, not crossing a boundary for it.
1178 Thus, on most machines, a named bit-field whose type is written as
1179 @code{int} would not cross a four-byte boundary, and would force
1180 four-byte alignment for the whole structure. (The alignment used may
1181 not be four bytes; it is controlled by the other alignment parameters.)
1183 An unnamed bit-field will not affect the alignment of the containing
1186 If the macro is defined, its definition should be a C expression;
1187 a nonzero value for the expression enables this behavior.
1189 Note that if this macro is not defined, or its value is zero, some
1190 bit-fields may cross more than one alignment boundary. The compiler can
1191 support such references if there are @samp{insv}, @samp{extv}, and
1192 @samp{extzv} insns that can directly reference memory.
1194 The other known way of making bit-fields work is to define
1195 @code{STRUCTURE_SIZE_BOUNDARY} as large as @code{BIGGEST_ALIGNMENT}.
1196 Then every structure can be accessed with fullwords.
1198 Unless the machine has bit-field instructions or you define
1199 @code{STRUCTURE_SIZE_BOUNDARY} that way, you must define
1200 @code{PCC_BITFIELD_TYPE_MATTERS} to have a nonzero value.
1202 If your aim is to make GCC use the same conventions for laying out
1203 bit-fields as are used by another compiler, here is how to investigate
1204 what the other compiler does. Compile and run this program:
1223 printf ("Size of foo1 is %d\n",
1224 sizeof (struct foo1));
1225 printf ("Size of foo2 is %d\n",
1226 sizeof (struct foo2));
1231 If this prints 2 and 5, then the compiler's behavior is what you would
1232 get from @code{PCC_BITFIELD_TYPE_MATTERS}.
1235 @defmac BITFIELD_NBYTES_LIMITED
1236 Like @code{PCC_BITFIELD_TYPE_MATTERS} except that its effect is limited
1237 to aligning a bit-field within the structure.
1240 @deftypefn {Target Hook} bool TARGET_ALIGN_ANON_BITFIELD (void)
1241 When @code{PCC_BITFIELD_TYPE_MATTERS} is true this hook will determine
1242 whether unnamed bitfields affect the alignment of the containing
1243 structure. The hook should return true if the structure should inherit
1244 the alignment requirements of an unnamed bitfield's type.
1247 @deftypefn {Target Hook} bool TARGET_NARROW_VOLATILE_BITFIELD (void)
1248 This target hook should return @code{true} if accesses to volatile bitfields
1249 should use the narrowest mode possible. It should return @code{false} if
1250 these accesses should use the bitfield container type.
1252 The default is @code{!TARGET_STRICT_ALIGN}.
1255 @defmac MEMBER_TYPE_FORCES_BLK (@var{field}, @var{mode})
1256 Return 1 if a structure or array containing @var{field} should be accessed using
1259 If @var{field} is the only field in the structure, @var{mode} is its
1260 mode, otherwise @var{mode} is VOIDmode. @var{mode} is provided in the
1261 case where structures of one field would require the structure's mode to
1262 retain the field's mode.
1264 Normally, this is not needed.
1267 @defmac ROUND_TYPE_ALIGN (@var{type}, @var{computed}, @var{specified})
1268 Define this macro as an expression for the alignment of a type (given
1269 by @var{type} as a tree node) if the alignment computed in the usual
1270 way is @var{computed} and the alignment explicitly specified was
1273 The default is to use @var{specified} if it is larger; otherwise, use
1274 the smaller of @var{computed} and @code{BIGGEST_ALIGNMENT}
1277 @defmac MAX_FIXED_MODE_SIZE
1278 An integer expression for the size in bits of the largest integer
1279 machine mode that should actually be used. All integer machine modes of
1280 this size or smaller can be used for structures and unions with the
1281 appropriate sizes. If this macro is undefined, @code{GET_MODE_BITSIZE
1282 (DImode)} is assumed.
1285 @defmac STACK_SAVEAREA_MODE (@var{save_level})
1286 If defined, an expression of type @code{enum machine_mode} that
1287 specifies the mode of the save area operand of a
1288 @code{save_stack_@var{level}} named pattern (@pxref{Standard Names}).
1289 @var{save_level} is one of @code{SAVE_BLOCK}, @code{SAVE_FUNCTION}, or
1290 @code{SAVE_NONLOCAL} and selects which of the three named patterns is
1291 having its mode specified.
1293 You need not define this macro if it always returns @code{Pmode}. You
1294 would most commonly define this macro if the
1295 @code{save_stack_@var{level}} patterns need to support both a 32- and a
1299 @defmac STACK_SIZE_MODE
1300 If defined, an expression of type @code{enum machine_mode} that
1301 specifies the mode of the size increment operand of an
1302 @code{allocate_stack} named pattern (@pxref{Standard Names}).
1304 You need not define this macro if it always returns @code{word_mode}.
1305 You would most commonly define this macro if the @code{allocate_stack}
1306 pattern needs to support both a 32- and a 64-bit mode.
1309 @deftypefn {Target Hook} {enum machine_mode} TARGET_LIBGCC_CMP_RETURN_MODE (void)
1310 This target hook should return the mode to be used for the return value
1311 of compare instructions expanded to libgcc calls. If not defined
1312 @code{word_mode} is returned which is the right choice for a majority of
1316 @deftypefn {Target Hook} {enum machine_mode} TARGET_LIBGCC_SHIFT_COUNT_MODE (void)
1317 This target hook should return the mode to be used for the shift count operand
1318 of shift instructions expanded to libgcc calls. If not defined
1319 @code{word_mode} is returned which is the right choice for a majority of
1323 @deftypefn {Target Hook} {enum machine_mode} TARGET_UNWIND_WORD_MODE (void)
1324 Return machine mode to be used for @code{_Unwind_Word} type.
1325 The default is to use @code{word_mode}.
1328 @defmac ROUND_TOWARDS_ZERO
1329 If defined, this macro should be true if the prevailing rounding
1330 mode is towards zero.
1332 Defining this macro only affects the way @file{libgcc.a} emulates
1333 floating-point arithmetic.
1335 Not defining this macro is equivalent to returning zero.
1338 @defmac LARGEST_EXPONENT_IS_NORMAL (@var{size})
1339 This macro should return true if floats with @var{size}
1340 bits do not have a NaN or infinity representation, but use the largest
1341 exponent for normal numbers instead.
1343 Defining this macro only affects the way @file{libgcc.a} emulates
1344 floating-point arithmetic.
1346 The default definition of this macro returns false for all sizes.
1349 @deftypefn {Target Hook} bool TARGET_MS_BITFIELD_LAYOUT_P (const_tree @var{record_type})
1350 This target hook returns @code{true} if bit-fields in the given
1351 @var{record_type} are to be laid out following the rules of Microsoft
1352 Visual C/C++, namely: (i) a bit-field won't share the same storage
1353 unit with the previous bit-field if their underlying types have
1354 different sizes, and the bit-field will be aligned to the highest
1355 alignment of the underlying types of itself and of the previous
1356 bit-field; (ii) a zero-sized bit-field will affect the alignment of
1357 the whole enclosing structure, even if it is unnamed; except that
1358 (iii) a zero-sized bit-field will be disregarded unless it follows
1359 another bit-field of nonzero size. If this hook returns @code{true},
1360 other macros that control bit-field layout are ignored.
1362 When a bit-field is inserted into a packed record, the whole size
1363 of the underlying type is used by one or more same-size adjacent
1364 bit-fields (that is, if its long:3, 32 bits is used in the record,
1365 and any additional adjacent long bit-fields are packed into the same
1366 chunk of 32 bits. However, if the size changes, a new field of that
1367 size is allocated). In an unpacked record, this is the same as using
1368 alignment, but not equivalent when packing.
1370 If both MS bit-fields and @samp{__attribute__((packed))} are used,
1371 the latter will take precedence. If @samp{__attribute__((packed))} is
1372 used on a single field when MS bit-fields are in use, it will take
1373 precedence for that field, but the alignment of the rest of the structure
1374 may affect its placement.
1377 @deftypefn {Target Hook} bool TARGET_DECIMAL_FLOAT_SUPPORTED_P (void)
1378 Returns true if the target supports decimal floating point.
1381 @deftypefn {Target Hook} bool TARGET_FIXED_POINT_SUPPORTED_P (void)
1382 Returns true if the target supports fixed-point arithmetic.
1385 @deftypefn {Target Hook} void TARGET_EXPAND_TO_RTL_HOOK (void)
1386 This hook is called just before expansion into rtl, allowing the target
1387 to perform additional initializations or analysis before the expansion.
1388 For example, the rs6000 port uses it to allocate a scratch stack slot
1389 for use in copying SDmode values between memory and floating point
1390 registers whenever the function being expanded has any SDmode
1394 @deftypefn {Target Hook} void TARGET_INSTANTIATE_DECLS (void)
1395 This hook allows the backend to perform additional instantiations on rtl
1396 that are not actually in any insns yet, but will be later.
1399 @deftypefn {Target Hook} {const char *} TARGET_MANGLE_TYPE (const_tree @var{type})
1400 If your target defines any fundamental types, or any types your target
1401 uses should be mangled differently from the default, define this hook
1402 to return the appropriate encoding for these types as part of a C++
1403 mangled name. The @var{type} argument is the tree structure representing
1404 the type to be mangled. The hook may be applied to trees which are
1405 not target-specific fundamental types; it should return @code{NULL}
1406 for all such types, as well as arguments it does not recognize. If the
1407 return value is not @code{NULL}, it must point to a statically-allocated
1410 Target-specific fundamental types might be new fundamental types or
1411 qualified versions of ordinary fundamental types. Encode new
1412 fundamental types as @samp{@w{u @var{n} @var{name}}}, where @var{name}
1413 is the name used for the type in source code, and @var{n} is the
1414 length of @var{name} in decimal. Encode qualified versions of
1415 ordinary types as @samp{@w{U @var{n} @var{name} @var{code}}}, where
1416 @var{name} is the name used for the type qualifier in source code,
1417 @var{n} is the length of @var{name} as above, and @var{code} is the
1418 code used to represent the unqualified version of this type. (See
1419 @code{write_builtin_type} in @file{cp/mangle.c} for the list of
1420 codes.) In both cases the spaces are for clarity; do not include any
1421 spaces in your string.
1423 This hook is applied to types prior to typedef resolution. If the mangled
1424 name for a particular type depends only on that type's main variant, you
1425 can perform typedef resolution yourself using @code{TYPE_MAIN_VARIANT}
1428 The default version of this hook always returns @code{NULL}, which is
1429 appropriate for a target that does not define any new fundamental
1434 @section Layout of Source Language Data Types
1436 These macros define the sizes and other characteristics of the standard
1437 basic data types used in programs being compiled. Unlike the macros in
1438 the previous section, these apply to specific features of C and related
1439 languages, rather than to fundamental aspects of storage layout.
1441 @defmac INT_TYPE_SIZE
1442 A C expression for the size in bits of the type @code{int} on the
1443 target machine. If you don't define this, the default is one word.
1446 @defmac SHORT_TYPE_SIZE
1447 A C expression for the size in bits of the type @code{short} on the
1448 target machine. If you don't define this, the default is half a word.
1449 (If this would be less than one storage unit, it is rounded up to one
1453 @defmac LONG_TYPE_SIZE
1454 A C expression for the size in bits of the type @code{long} on the
1455 target machine. If you don't define this, the default is one word.
1458 @defmac ADA_LONG_TYPE_SIZE
1459 On some machines, the size used for the Ada equivalent of the type
1460 @code{long} by a native Ada compiler differs from that used by C@. In
1461 that situation, define this macro to be a C expression to be used for
1462 the size of that type. If you don't define this, the default is the
1463 value of @code{LONG_TYPE_SIZE}.
1466 @defmac LONG_LONG_TYPE_SIZE
1467 A C expression for the size in bits of the type @code{long long} on the
1468 target machine. If you don't define this, the default is two
1469 words. If you want to support GNU Ada on your machine, the value of this
1470 macro must be at least 64.
1473 @defmac CHAR_TYPE_SIZE
1474 A C expression for the size in bits of the type @code{char} on the
1475 target machine. If you don't define this, the default is
1476 @code{BITS_PER_UNIT}.
1479 @defmac BOOL_TYPE_SIZE
1480 A C expression for the size in bits of the C++ type @code{bool} and
1481 C99 type @code{_Bool} on the target machine. If you don't define
1482 this, and you probably shouldn't, the default is @code{CHAR_TYPE_SIZE}.
1485 @defmac FLOAT_TYPE_SIZE
1486 A C expression for the size in bits of the type @code{float} on the
1487 target machine. If you don't define this, the default is one word.
1490 @defmac DOUBLE_TYPE_SIZE
1491 A C expression for the size in bits of the type @code{double} on the
1492 target machine. If you don't define this, the default is two
1496 @defmac LONG_DOUBLE_TYPE_SIZE
1497 A C expression for the size in bits of the type @code{long double} on
1498 the target machine. If you don't define this, the default is two
1502 @defmac SHORT_FRACT_TYPE_SIZE
1503 A C expression for the size in bits of the type @code{short _Fract} on
1504 the target machine. If you don't define this, the default is
1505 @code{BITS_PER_UNIT}.
1508 @defmac FRACT_TYPE_SIZE
1509 A C expression for the size in bits of the type @code{_Fract} on
1510 the target machine. If you don't define this, the default is
1511 @code{BITS_PER_UNIT * 2}.
1514 @defmac LONG_FRACT_TYPE_SIZE
1515 A C expression for the size in bits of the type @code{long _Fract} on
1516 the target machine. If you don't define this, the default is
1517 @code{BITS_PER_UNIT * 4}.
1520 @defmac LONG_LONG_FRACT_TYPE_SIZE
1521 A C expression for the size in bits of the type @code{long long _Fract} on
1522 the target machine. If you don't define this, the default is
1523 @code{BITS_PER_UNIT * 8}.
1526 @defmac SHORT_ACCUM_TYPE_SIZE
1527 A C expression for the size in bits of the type @code{short _Accum} on
1528 the target machine. If you don't define this, the default is
1529 @code{BITS_PER_UNIT * 2}.
1532 @defmac ACCUM_TYPE_SIZE
1533 A C expression for the size in bits of the type @code{_Accum} on
1534 the target machine. If you don't define this, the default is
1535 @code{BITS_PER_UNIT * 4}.
1538 @defmac LONG_ACCUM_TYPE_SIZE
1539 A C expression for the size in bits of the type @code{long _Accum} on
1540 the target machine. If you don't define this, the default is
1541 @code{BITS_PER_UNIT * 8}.
1544 @defmac LONG_LONG_ACCUM_TYPE_SIZE
1545 A C expression for the size in bits of the type @code{long long _Accum} on
1546 the target machine. If you don't define this, the default is
1547 @code{BITS_PER_UNIT * 16}.
1550 @defmac LIBGCC2_LONG_DOUBLE_TYPE_SIZE
1551 Define this macro if @code{LONG_DOUBLE_TYPE_SIZE} is not constant or
1552 if you want routines in @file{libgcc2.a} for a size other than
1553 @code{LONG_DOUBLE_TYPE_SIZE}. If you don't define this, the
1554 default is @code{LONG_DOUBLE_TYPE_SIZE}.
1557 @defmac LIBGCC2_HAS_DF_MODE
1558 Define this macro if neither @code{DOUBLE_TYPE_SIZE} nor
1559 @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is
1560 @code{DFmode} but you want @code{DFmode} routines in @file{libgcc2.a}
1561 anyway. If you don't define this and either @code{DOUBLE_TYPE_SIZE}
1562 or @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is 64 then the default is 1,
1566 @defmac LIBGCC2_HAS_XF_MODE
1567 Define this macro if @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is not
1568 @code{XFmode} but you want @code{XFmode} routines in @file{libgcc2.a}
1569 anyway. If you don't define this and @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE}
1570 is 80 then the default is 1, otherwise it is 0.
1573 @defmac LIBGCC2_HAS_TF_MODE
1574 Define this macro if @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is not
1575 @code{TFmode} but you want @code{TFmode} routines in @file{libgcc2.a}
1576 anyway. If you don't define this and @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE}
1577 is 128 then the default is 1, otherwise it is 0.
1580 @defmac LIBGCC2_GNU_PREFIX
1581 This macro corresponds to the @code{TARGET_LIBFUNC_GNU_PREFIX} target
1582 hook and should be defined if that hook is overriden to be true. It
1583 causes function names in libgcc to be changed to use a @code{__gnu_}
1584 prefix for their name rather than the default @code{__}. A port which
1585 uses this macro should also arrange to use @file{t-gnu-prefix} in
1586 the libgcc @file{config.host}.
1593 Define these macros to be the size in bits of the mantissa of
1594 @code{SFmode}, @code{DFmode}, @code{XFmode} and @code{TFmode} values,
1595 if the defaults in @file{libgcc2.h} are inappropriate. By default,
1596 @code{FLT_MANT_DIG} is used for @code{SF_SIZE}, @code{LDBL_MANT_DIG}
1597 for @code{XF_SIZE} and @code{TF_SIZE}, and @code{DBL_MANT_DIG} or
1598 @code{LDBL_MANT_DIG} for @code{DF_SIZE} according to whether
1599 @code{DOUBLE_TYPE_SIZE} or
1600 @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is 64.
1603 @defmac TARGET_FLT_EVAL_METHOD
1604 A C expression for the value for @code{FLT_EVAL_METHOD} in @file{float.h},
1605 assuming, if applicable, that the floating-point control word is in its
1606 default state. If you do not define this macro the value of
1607 @code{FLT_EVAL_METHOD} will be zero.
1610 @defmac WIDEST_HARDWARE_FP_SIZE
1611 A C expression for the size in bits of the widest floating-point format
1612 supported by the hardware. If you define this macro, you must specify a
1613 value less than or equal to the value of @code{LONG_DOUBLE_TYPE_SIZE}.
1614 If you do not define this macro, the value of @code{LONG_DOUBLE_TYPE_SIZE}
1618 @defmac DEFAULT_SIGNED_CHAR
1619 An expression whose value is 1 or 0, according to whether the type
1620 @code{char} should be signed or unsigned by default. The user can
1621 always override this default with the options @option{-fsigned-char}
1622 and @option{-funsigned-char}.
1625 @deftypefn {Target Hook} bool TARGET_DEFAULT_SHORT_ENUMS (void)
1626 This target hook should return true if the compiler should give an
1627 @code{enum} type only as many bytes as it takes to represent the range
1628 of possible values of that type. It should return false if all
1629 @code{enum} types should be allocated like @code{int}.
1631 The default is to return false.
1635 A C expression for a string describing the name of the data type to use
1636 for size values. The typedef name @code{size_t} is defined using the
1637 contents of the string.
1639 The string can contain more than one keyword. If so, separate them with
1640 spaces, and write first any length keyword, then @code{unsigned} if
1641 appropriate, and finally @code{int}. The string must exactly match one
1642 of the data type names defined in the function
1643 @code{init_decl_processing} in the file @file{c-decl.c}. You may not
1644 omit @code{int} or change the order---that would cause the compiler to
1647 If you don't define this macro, the default is @code{"long unsigned
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 @deftypefn {Target Hook} void TARGET_CONDITIONAL_REGISTER_USAGE (void)
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 @deftypefn {Target Hook} bool TARGET_HARD_REGNO_SCRATCH_OK (unsigned int @var{regno})
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{outer_code}, @var{index_code})
2439 A C expression whose value is the register class to which a valid
2440 base register must belong. @var{outer_code} and @var{index_code} define the
2441 context in which the base register occurs. @var{outer_code} is the code of
2442 the immediately enclosing expression (@code{MEM} for the top level of an
2443 address, @code{ADDRESS} for something that occurs in an
2444 @code{address_operand}). @var{index_code} is the code of the corresponding
2445 index expression if @var{outer_code} is @code{PLUS}; @code{SCRATCH} otherwise.
2448 @defmac INDEX_REG_CLASS
2449 A macro whose definition is the name of the class to which a valid
2450 index register must belong. An index register is one used in an
2451 address where its value is either multiplied by a scale factor or
2452 added to another register (as well as added to a displacement).
2455 @defmac REGNO_OK_FOR_BASE_P (@var{num})
2456 A C expression which is nonzero if register number @var{num} is
2457 suitable for use as a base register in operand addresses.
2460 @defmac REGNO_MODE_OK_FOR_BASE_P (@var{num}, @var{mode})
2461 A C expression that is just like @code{REGNO_OK_FOR_BASE_P}, except that
2462 that expression may examine the mode of the memory reference in
2463 @var{mode}. You should define this macro if the mode of the memory
2464 reference affects whether a register may be used as a base register. If
2465 you define this macro, the compiler will use it instead of
2466 @code{REGNO_OK_FOR_BASE_P}. The mode may be @code{VOIDmode} for
2467 addresses that appear outside a @code{MEM}, i.e., as an
2468 @code{address_operand}.
2471 @defmac REGNO_MODE_OK_FOR_REG_BASE_P (@var{num}, @var{mode})
2472 A C expression which is nonzero if register number @var{num} is suitable for
2473 use as a base register in base plus index operand addresses, accessing
2474 memory in mode @var{mode}. It may be either a suitable hard register or a
2475 pseudo register that has been allocated such a hard register. You should
2476 define this macro if base plus index addresses have different requirements
2477 than other base register uses.
2479 Use of this macro is deprecated; please use the more general
2480 @code{REGNO_MODE_CODE_OK_FOR_BASE_P}.
2483 @defmac REGNO_MODE_CODE_OK_FOR_BASE_P (@var{num}, @var{mode}, @var{outer_code}, @var{index_code})
2484 A C expression that is just like @code{REGNO_MODE_OK_FOR_BASE_P}, except
2485 that that expression may examine the context in which the register
2486 appears in the memory reference. @var{outer_code} is the code of the
2487 immediately enclosing expression (@code{MEM} if at the top level of the
2488 address, @code{ADDRESS} for something that occurs in an
2489 @code{address_operand}). @var{index_code} is the code of the
2490 corresponding index expression if @var{outer_code} is @code{PLUS};
2491 @code{SCRATCH} otherwise. The mode may be @code{VOIDmode} for addresses
2492 that appear outside a @code{MEM}, i.e., as an @code{address_operand}.
2495 @defmac REGNO_OK_FOR_INDEX_P (@var{num})
2496 A C expression which is nonzero if register number @var{num} is
2497 suitable for use as an index register in operand addresses. It may be
2498 either a suitable hard register or a pseudo register that has been
2499 allocated such a hard register.
2501 The difference between an index register and a base register is that
2502 the index register may be scaled. If an address involves the sum of
2503 two registers, neither one of them scaled, then either one may be
2504 labeled the ``base'' and the other the ``index''; but whichever
2505 labeling is used must fit the machine's constraints of which registers
2506 may serve in each capacity. The compiler will try both labelings,
2507 looking for one that is valid, and will reload one or both registers
2508 only if neither labeling works.
2511 @deftypefn {Target Hook} reg_class_t TARGET_PREFERRED_RENAME_CLASS (reg_class_t @var{rclass})
2512 A target hook that places additional preference on the register class to use when it is necessary to rename a register in class @var{rclass} to another class, or perhaps @var{NO_REGS}, if no preferred register class is found or hook @code{preferred_rename_class} is not implemented. Sometimes returning a more restrictive class makes better code. For example, on ARM, thumb-2 instructions using @code{LO_REGS} may be smaller than instructions using @code{GENERIC_REGS}. By returning @code{LO_REGS} from @code{preferred_rename_class}, code size can be reduced.
2515 @deftypefn {Target Hook} reg_class_t TARGET_PREFERRED_RELOAD_CLASS (rtx @var{x}, reg_class_t @var{rclass})
2516 A target hook that places additional restrictions on the register class
2517 to use when it is necessary to copy value @var{x} into a register in class
2518 @var{rclass}. The value is a register class; perhaps @var{rclass}, or perhaps
2519 another, smaller class.
2521 The default version of this hook always returns value of @code{rclass} argument.
2523 Sometimes returning a more restrictive class makes better code. For
2524 example, on the 68000, when @var{x} is an integer constant that is in range
2525 for a @samp{moveq} instruction, the value of this macro is always
2526 @code{DATA_REGS} as long as @var{rclass} includes the data registers.
2527 Requiring a data register guarantees that a @samp{moveq} will be used.
2529 One case where @code{TARGET_PREFERRED_RELOAD_CLASS} must not return
2530 @var{rclass} is if @var{x} is a legitimate constant which cannot be
2531 loaded into some register class. By returning @code{NO_REGS} you can
2532 force @var{x} into a memory location. For example, rs6000 can load
2533 immediate values into general-purpose registers, but does not have an
2534 instruction for loading an immediate value into a floating-point
2535 register, so @code{TARGET_PREFERRED_RELOAD_CLASS} returns @code{NO_REGS} when
2536 @var{x} is a floating-point constant. If the constant can't be loaded
2537 into any kind of register, code generation will be better if
2538 @code{TARGET_LEGITIMATE_CONSTANT_P} makes the constant illegitimate instead
2539 of using @code{TARGET_PREFERRED_RELOAD_CLASS}.
2541 If an insn has pseudos in it after register allocation, reload will go
2542 through the alternatives and call repeatedly @code{TARGET_PREFERRED_RELOAD_CLASS}
2543 to find the best one. Returning @code{NO_REGS}, in this case, makes
2544 reload add a @code{!} in front of the constraint: the x86 back-end uses
2545 this feature to discourage usage of 387 registers when math is done in
2546 the SSE registers (and vice versa).
2549 @defmac PREFERRED_RELOAD_CLASS (@var{x}, @var{class})
2550 A C expression that places additional restrictions on the register class
2551 to use when it is necessary to copy value @var{x} into a register in class
2552 @var{class}. The value is a register class; perhaps @var{class}, or perhaps
2553 another, smaller class. On many machines, the following definition is
2557 #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS
2560 Sometimes returning a more restrictive class makes better code. For
2561 example, on the 68000, when @var{x} is an integer constant that is in range
2562 for a @samp{moveq} instruction, the value of this macro is always
2563 @code{DATA_REGS} as long as @var{class} includes the data registers.
2564 Requiring a data register guarantees that a @samp{moveq} will be used.
2566 One case where @code{PREFERRED_RELOAD_CLASS} must not return
2567 @var{class} is if @var{x} is a legitimate constant which cannot be
2568 loaded into some register class. By returning @code{NO_REGS} you can
2569 force @var{x} into a memory location. For example, rs6000 can load
2570 immediate values into general-purpose registers, but does not have an
2571 instruction for loading an immediate value into a floating-point
2572 register, so @code{PREFERRED_RELOAD_CLASS} returns @code{NO_REGS} when
2573 @var{x} is a floating-point constant. If the constant can't be loaded
2574 into any kind of register, code generation will be better if
2575 @code{TARGET_LEGITIMATE_CONSTANT_P} makes the constant illegitimate instead
2576 of using @code{TARGET_PREFERRED_RELOAD_CLASS}.
2578 If an insn has pseudos in it after register allocation, reload will go
2579 through the alternatives and call repeatedly @code{PREFERRED_RELOAD_CLASS}
2580 to find the best one. Returning @code{NO_REGS}, in this case, makes
2581 reload add a @code{!} in front of the constraint: the x86 back-end uses
2582 this feature to discourage usage of 387 registers when math is done in
2583 the SSE registers (and vice versa).
2586 @deftypefn {Target Hook} reg_class_t TARGET_PREFERRED_OUTPUT_RELOAD_CLASS (rtx @var{x}, reg_class_t @var{rclass})
2587 Like @code{TARGET_PREFERRED_RELOAD_CLASS}, but for output reloads instead of
2590 The default version of this hook always returns value of @code{rclass}
2593 You can also use @code{TARGET_PREFERRED_OUTPUT_RELOAD_CLASS} to discourage
2594 reload from using some alternatives, like @code{TARGET_PREFERRED_RELOAD_CLASS}.
2597 @defmac LIMIT_RELOAD_CLASS (@var{mode}, @var{class})
2598 A C expression that places additional restrictions on the register class
2599 to use when it is necessary to be able to hold a value of mode
2600 @var{mode} in a reload register for which class @var{class} would
2603 Unlike @code{PREFERRED_RELOAD_CLASS}, this macro should be used when
2604 there are certain modes that simply can't go in certain reload classes.
2606 The value is a register class; perhaps @var{class}, or perhaps another,
2609 Don't define this macro unless the target machine has limitations which
2610 require the macro to do something nontrivial.
2613 @deftypefn {Target Hook} reg_class_t TARGET_SECONDARY_RELOAD (bool @var{in_p}, rtx @var{x}, reg_class_t @var{reload_class}, enum machine_mode @var{reload_mode}, secondary_reload_info *@var{sri})
2614 Many machines have some registers that cannot be copied directly to or
2615 from memory or even from other types of registers. An example is the
2616 @samp{MQ} register, which on most machines, can only be copied to or
2617 from general registers, but not memory. Below, we shall be using the
2618 term 'intermediate register' when a move operation cannot be performed
2619 directly, but has to be done by copying the source into the intermediate
2620 register first, and then copying the intermediate register to the
2621 destination. An intermediate register always has the same mode as
2622 source and destination. Since it holds the actual value being copied,
2623 reload might apply optimizations to re-use an intermediate register
2624 and eliding the copy from the source when it can determine that the
2625 intermediate register still holds the required value.
2627 Another kind of secondary reload is required on some machines which
2628 allow copying all registers to and from memory, but require a scratch
2629 register for stores to some memory locations (e.g., those with symbolic
2630 address on the RT, and those with certain symbolic address on the SPARC
2631 when compiling PIC)@. Scratch registers need not have the same mode
2632 as the value being copied, and usually hold a different value than
2633 that being copied. Special patterns in the md file are needed to
2634 describe how the copy is performed with the help of the scratch register;
2635 these patterns also describe the number, register class(es) and mode(s)
2636 of the scratch register(s).
2638 In some cases, both an intermediate and a scratch register are required.
2640 For input reloads, this target hook is called with nonzero @var{in_p},
2641 and @var{x} is an rtx that needs to be copied to a register of class
2642 @var{reload_class} in @var{reload_mode}. For output reloads, this target
2643 hook is called with zero @var{in_p}, and a register of class @var{reload_class}
2644 needs to be copied to rtx @var{x} in @var{reload_mode}.
2646 If copying a register of @var{reload_class} from/to @var{x} requires
2647 an intermediate register, the hook @code{secondary_reload} should
2648 return the register class required for this intermediate register.
2649 If no intermediate register is required, it should return NO_REGS.
2650 If more than one intermediate register is required, describe the one
2651 that is closest in the copy chain to the reload register.
2653 If scratch registers are needed, you also have to describe how to
2654 perform the copy from/to the reload register to/from this
2655 closest intermediate register. Or if no intermediate register is
2656 required, but still a scratch register is needed, describe the
2657 copy from/to the reload register to/from the reload operand @var{x}.
2659 You do this by setting @code{sri->icode} to the instruction code of a pattern
2660 in the md file which performs the move. Operands 0 and 1 are the output
2661 and input of this copy, respectively. Operands from operand 2 onward are
2662 for scratch operands. These scratch operands must have a mode, and a
2663 single-register-class
2664 @c [later: or memory]
2667 When an intermediate register is used, the @code{secondary_reload}
2668 hook will be called again to determine how to copy the intermediate
2669 register to/from the reload operand @var{x}, so your hook must also
2670 have code to handle the register class of the intermediate operand.
2672 @c [For later: maybe we'll allow multi-alternative reload patterns -
2673 @c the port maintainer could name a mov<mode> pattern that has clobbers -
2674 @c and match the constraints of input and output to determine the required
2675 @c alternative. A restriction would be that constraints used to match
2676 @c against reloads registers would have to be written as register class
2677 @c constraints, or we need a new target macro / hook that tells us if an
2678 @c arbitrary constraint can match an unknown register of a given class.
2679 @c Such a macro / hook would also be useful in other places.]
2682 @var{x} might be a pseudo-register or a @code{subreg} of a
2683 pseudo-register, which could either be in a hard register or in memory.
2684 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2685 in memory and the hard register number if it is in a register.
2687 Scratch operands in memory (constraint @code{"=m"} / @code{"=&m"}) are
2688 currently not supported. For the time being, you will have to continue
2689 to use @code{SECONDARY_MEMORY_NEEDED} for that purpose.
2691 @code{copy_cost} also uses this target hook to find out how values are
2692 copied. If you want it to include some extra cost for the need to allocate
2693 (a) scratch register(s), set @code{sri->extra_cost} to the additional cost.
2694 Or if two dependent moves are supposed to have a lower cost than the sum
2695 of the individual moves due to expected fortuitous scheduling and/or special
2696 forwarding logic, you can set @code{sri->extra_cost} to a negative amount.
2699 @defmac SECONDARY_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2700 @defmacx SECONDARY_INPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2701 @defmacx SECONDARY_OUTPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2702 These macros are obsolete, new ports should use the target hook
2703 @code{TARGET_SECONDARY_RELOAD} instead.
2705 These are obsolete macros, replaced by the @code{TARGET_SECONDARY_RELOAD}
2706 target hook. Older ports still define these macros to indicate to the
2707 reload phase that it may
2708 need to allocate at least one register for a reload in addition to the
2709 register to contain the data. Specifically, if copying @var{x} to a
2710 register @var{class} in @var{mode} requires an intermediate register,
2711 you were supposed to define @code{SECONDARY_INPUT_RELOAD_CLASS} to return the
2712 largest register class all of whose registers can be used as
2713 intermediate registers or scratch registers.
2715 If copying a register @var{class} in @var{mode} to @var{x} requires an
2716 intermediate or scratch register, @code{SECONDARY_OUTPUT_RELOAD_CLASS}
2717 was supposed to be defined be defined to return the largest register
2718 class required. If the
2719 requirements for input and output reloads were the same, the macro
2720 @code{SECONDARY_RELOAD_CLASS} should have been used instead of defining both
2723 The values returned by these macros are often @code{GENERAL_REGS}.
2724 Return @code{NO_REGS} if no spare register is needed; i.e., if @var{x}
2725 can be directly copied to or from a register of @var{class} in
2726 @var{mode} without requiring a scratch register. Do not define this
2727 macro if it would always return @code{NO_REGS}.
2729 If a scratch register is required (either with or without an
2730 intermediate register), you were supposed to define patterns for
2731 @samp{reload_in@var{m}} or @samp{reload_out@var{m}}, as required
2732 (@pxref{Standard Names}. These patterns, which were normally
2733 implemented with a @code{define_expand}, should be similar to the
2734 @samp{mov@var{m}} patterns, except that operand 2 is the scratch
2737 These patterns need constraints for the reload register and scratch
2739 contain a single register class. If the original reload register (whose
2740 class is @var{class}) can meet the constraint given in the pattern, the
2741 value returned by these macros is used for the class of the scratch
2742 register. Otherwise, two additional reload registers are required.
2743 Their classes are obtained from the constraints in the insn pattern.
2745 @var{x} might be a pseudo-register or a @code{subreg} of a
2746 pseudo-register, which could either be in a hard register or in memory.
2747 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2748 in memory and the hard register number if it is in a register.
2750 These macros should not be used in the case where a particular class of
2751 registers can only be copied to memory and not to another class of
2752 registers. In that case, secondary reload registers are not needed and
2753 would not be helpful. Instead, a stack location must be used to perform
2754 the copy and the @code{mov@var{m}} pattern should use memory as an
2755 intermediate storage. This case often occurs between floating-point and
2759 @defmac SECONDARY_MEMORY_NEEDED (@var{class1}, @var{class2}, @var{m})
2760 Certain machines have the property that some registers cannot be copied
2761 to some other registers without using memory. Define this macro on
2762 those machines to be a C expression that is nonzero if objects of mode
2763 @var{m} in registers of @var{class1} can only be copied to registers of
2764 class @var{class2} by storing a register of @var{class1} into memory
2765 and loading that memory location into a register of @var{class2}.
2767 Do not define this macro if its value would always be zero.
2770 @defmac SECONDARY_MEMORY_NEEDED_RTX (@var{mode})
2771 Normally when @code{SECONDARY_MEMORY_NEEDED} is defined, the compiler
2772 allocates a stack slot for a memory location needed for register copies.
2773 If this macro is defined, the compiler instead uses the memory location
2774 defined by this macro.
2776 Do not define this macro if you do not define
2777 @code{SECONDARY_MEMORY_NEEDED}.
2780 @defmac SECONDARY_MEMORY_NEEDED_MODE (@var{mode})
2781 When the compiler needs a secondary memory location to copy between two
2782 registers of mode @var{mode}, it normally allocates sufficient memory to
2783 hold a quantity of @code{BITS_PER_WORD} bits and performs the store and
2784 load operations in a mode that many bits wide and whose class is the
2785 same as that of @var{mode}.
2787 This is right thing to do on most machines because it ensures that all
2788 bits of the register are copied and prevents accesses to the registers
2789 in a narrower mode, which some machines prohibit for floating-point
2792 However, this default behavior is not correct on some machines, such as
2793 the DEC Alpha, that store short integers in floating-point registers
2794 differently than in integer registers. On those machines, the default
2795 widening will not work correctly and you must define this macro to
2796 suppress that widening in some cases. See the file @file{alpha.h} for
2799 Do not define this macro if you do not define
2800 @code{SECONDARY_MEMORY_NEEDED} or if widening @var{mode} to a mode that
2801 is @code{BITS_PER_WORD} bits wide is correct for your machine.
2804 @deftypefn {Target Hook} bool TARGET_CLASS_LIKELY_SPILLED_P (reg_class_t @var{rclass})
2805 A target hook which returns @code{true} if pseudos that have been assigned
2806 to registers of class @var{rclass} would likely be spilled because
2807 registers of @var{rclass} are needed for spill registers.
2809 The default version of this target hook returns @code{true} if @var{rclass}
2810 has exactly one register and @code{false} otherwise. On most machines, this
2811 default should be used. Only use this target hook to some other expression
2812 if pseudos allocated by @file{local-alloc.c} end up in memory because their
2813 hard registers were needed for spill registers. If this target hook returns
2814 @code{false} for those classes, those pseudos will only be allocated by
2815 @file{global.c}, which knows how to reallocate the pseudo to another
2816 register. If there would not be another register available for reallocation,
2817 you should not change the implementation of this target hook since
2818 the only effect of such implementation would be to slow down register
2822 @deftypefn {Target Hook} {unsigned char} TARGET_CLASS_MAX_NREGS (reg_class_t @var{rclass}, enum machine_mode @var{mode})
2823 A target hook returns the maximum number of consecutive registers
2824 of class @var{rclass} needed to hold a value of mode @var{mode}.
2826 This is closely related to the macro @code{HARD_REGNO_NREGS}. In fact,
2827 the value returned by @code{TARGET_CLASS_MAX_NREGS (@var{rclass},
2828 @var{mode})} target hook should be the maximum value of
2829 @code{HARD_REGNO_NREGS (@var{regno}, @var{mode})} for all @var{regno}
2830 values in the class @var{rclass}.
2832 This target hook helps control the handling of multiple-word values
2835 The default version of this target hook returns the size of @var{mode}
2839 @defmac CLASS_MAX_NREGS (@var{class}, @var{mode})
2840 A C expression for the maximum number of consecutive registers
2841 of class @var{class} needed to hold a value of mode @var{mode}.
2843 This is closely related to the macro @code{HARD_REGNO_NREGS}. In fact,
2844 the value of the macro @code{CLASS_MAX_NREGS (@var{class}, @var{mode})}
2845 should be the maximum value of @code{HARD_REGNO_NREGS (@var{regno},
2846 @var{mode})} for all @var{regno} values in the class @var{class}.
2848 This macro helps control the handling of multiple-word values
2852 @defmac CANNOT_CHANGE_MODE_CLASS (@var{from}, @var{to}, @var{class})
2853 If defined, a C expression that returns nonzero for a @var{class} for which
2854 a change from mode @var{from} to mode @var{to} is invalid.
2856 For the example, loading 32-bit integer or floating-point objects into
2857 floating-point registers on the Alpha extends them to 64 bits.
2858 Therefore loading a 64-bit object and then storing it as a 32-bit object
2859 does not store the low-order 32 bits, as would be the case for a normal
2860 register. Therefore, @file{alpha.h} defines @code{CANNOT_CHANGE_MODE_CLASS}
2864 #define CANNOT_CHANGE_MODE_CLASS(FROM, TO, CLASS) \
2865 (GET_MODE_SIZE (FROM) != GET_MODE_SIZE (TO) \
2866 ? reg_classes_intersect_p (FLOAT_REGS, (CLASS)) : 0)
2870 @node Old Constraints
2871 @section Obsolete Macros for Defining Constraints
2872 @cindex defining constraints, obsolete method
2873 @cindex constraints, defining, obsolete method
2875 Machine-specific constraints can be defined with these macros instead
2876 of the machine description constructs described in @ref{Define
2877 Constraints}. This mechanism is obsolete. New ports should not use
2878 it; old ports should convert to the new mechanism.
2880 @defmac CONSTRAINT_LEN (@var{char}, @var{str})
2881 For the constraint at the start of @var{str}, which starts with the letter
2882 @var{c}, return the length. This allows you to have register class /
2883 constant / extra constraints that are longer than a single letter;
2884 you don't need to define this macro if you can do with single-letter
2885 constraints only. The definition of this macro should use
2886 DEFAULT_CONSTRAINT_LEN for all the characters that you don't want
2887 to handle specially.
2888 There are some sanity checks in genoutput.c that check the constraint lengths
2889 for the md file, so you can also use this macro to help you while you are
2890 transitioning from a byzantine single-letter-constraint scheme: when you
2891 return a negative length for a constraint you want to re-use, genoutput
2892 will complain about every instance where it is used in the md file.
2895 @defmac REG_CLASS_FROM_LETTER (@var{char})
2896 A C expression which defines the machine-dependent operand constraint
2897 letters for register classes. If @var{char} is such a letter, the
2898 value should be the register class corresponding to it. Otherwise,
2899 the value should be @code{NO_REGS}. The register letter @samp{r},
2900 corresponding to class @code{GENERAL_REGS}, will not be passed
2901 to this macro; you do not need to handle it.
2904 @defmac REG_CLASS_FROM_CONSTRAINT (@var{char}, @var{str})
2905 Like @code{REG_CLASS_FROM_LETTER}, but you also get the constraint string
2906 passed in @var{str}, so that you can use suffixes to distinguish between
2910 @defmac CONST_OK_FOR_LETTER_P (@var{value}, @var{c})
2911 A C expression that defines the machine-dependent operand constraint
2912 letters (@samp{I}, @samp{J}, @samp{K}, @dots{} @samp{P}) that specify
2913 particular ranges of integer values. If @var{c} is one of those
2914 letters, the expression should check that @var{value}, an integer, is in
2915 the appropriate range and return 1 if so, 0 otherwise. If @var{c} is
2916 not one of those letters, the value should be 0 regardless of
2920 @defmac CONST_OK_FOR_CONSTRAINT_P (@var{value}, @var{c}, @var{str})
2921 Like @code{CONST_OK_FOR_LETTER_P}, but you also get the constraint
2922 string passed in @var{str}, so that you can use suffixes to distinguish
2923 between different variants.
2926 @defmac CONST_DOUBLE_OK_FOR_LETTER_P (@var{value}, @var{c})
2927 A C expression that defines the machine-dependent operand constraint
2928 letters that specify particular ranges of @code{const_double} values
2929 (@samp{G} or @samp{H}).
2931 If @var{c} is one of those letters, the expression should check that
2932 @var{value}, an RTX of code @code{const_double}, is in the appropriate
2933 range and return 1 if so, 0 otherwise. If @var{c} is not one of those
2934 letters, the value should be 0 regardless of @var{value}.
2936 @code{const_double} is used for all floating-point constants and for
2937 @code{DImode} fixed-point constants. A given letter can accept either
2938 or both kinds of values. It can use @code{GET_MODE} to distinguish
2939 between these kinds.
2942 @defmac CONST_DOUBLE_OK_FOR_CONSTRAINT_P (@var{value}, @var{c}, @var{str})
2943 Like @code{CONST_DOUBLE_OK_FOR_LETTER_P}, but you also get the constraint
2944 string passed in @var{str}, so that you can use suffixes to distinguish
2945 between different variants.
2948 @defmac EXTRA_CONSTRAINT (@var{value}, @var{c})
2949 A C expression that defines the optional machine-dependent constraint
2950 letters that can be used to segregate specific types of operands, usually
2951 memory references, for the target machine. Any letter that is not
2952 elsewhere defined and not matched by @code{REG_CLASS_FROM_LETTER} /
2953 @code{REG_CLASS_FROM_CONSTRAINT}
2954 may be used. Normally this macro will not be defined.
2956 If it is required for a particular target machine, it should return 1
2957 if @var{value} corresponds to the operand type represented by the
2958 constraint letter @var{c}. If @var{c} is not defined as an extra
2959 constraint, the value returned should be 0 regardless of @var{value}.
2961 For example, on the ROMP, load instructions cannot have their output
2962 in r0 if the memory reference contains a symbolic address. Constraint
2963 letter @samp{Q} is defined as representing a memory address that does
2964 @emph{not} contain a symbolic address. An alternative is specified with
2965 a @samp{Q} constraint on the input and @samp{r} on the output. The next
2966 alternative specifies @samp{m} on the input and a register class that
2967 does not include r0 on the output.
2970 @defmac EXTRA_CONSTRAINT_STR (@var{value}, @var{c}, @var{str})
2971 Like @code{EXTRA_CONSTRAINT}, but you also get the constraint string passed
2972 in @var{str}, so that you can use suffixes to distinguish between different
2976 @defmac EXTRA_MEMORY_CONSTRAINT (@var{c}, @var{str})
2977 A C expression that defines the optional machine-dependent constraint
2978 letters, amongst those accepted by @code{EXTRA_CONSTRAINT}, that should
2979 be treated like memory constraints by the reload pass.
2981 It should return 1 if the operand type represented by the constraint
2982 at the start of @var{str}, the first letter of which is the letter @var{c},
2983 comprises a subset of all memory references including
2984 all those whose address is simply a base register. This allows the reload
2985 pass to reload an operand, if it does not directly correspond to the operand
2986 type of @var{c}, by copying its address into a base register.
2988 For example, on the S/390, some instructions do not accept arbitrary
2989 memory references, but only those that do not make use of an index
2990 register. The constraint letter @samp{Q} is defined via
2991 @code{EXTRA_CONSTRAINT} as representing a memory address of this type.
2992 If the letter @samp{Q} is marked as @code{EXTRA_MEMORY_CONSTRAINT},
2993 a @samp{Q} constraint can handle any memory operand, because the
2994 reload pass knows it can be reloaded by copying the memory address
2995 into a base register if required. This is analogous to the way
2996 an @samp{o} constraint can handle any memory operand.
2999 @defmac EXTRA_ADDRESS_CONSTRAINT (@var{c}, @var{str})
3000 A C expression that defines the optional machine-dependent constraint
3001 letters, amongst those accepted by @code{EXTRA_CONSTRAINT} /
3002 @code{EXTRA_CONSTRAINT_STR}, that should
3003 be treated like address constraints by the reload pass.
3005 It should return 1 if the operand type represented by the constraint
3006 at the start of @var{str}, which starts with the letter @var{c}, comprises
3007 a subset of all memory addresses including
3008 all those that consist of just a base register. This allows the reload
3009 pass to reload an operand, if it does not directly correspond to the operand
3010 type of @var{str}, by copying it into a base register.
3012 Any constraint marked as @code{EXTRA_ADDRESS_CONSTRAINT} can only
3013 be used with the @code{address_operand} predicate. It is treated
3014 analogously to the @samp{p} constraint.
3017 @node Stack and Calling
3018 @section Stack Layout and Calling Conventions
3019 @cindex calling conventions
3021 @c prevent bad page break with this line
3022 This describes the stack layout and calling conventions.
3026 * Exception Handling::
3031 * Register Arguments::
3033 * Aggregate Return::
3038 * Stack Smashing Protection::
3042 @subsection Basic Stack Layout
3043 @cindex stack frame layout
3044 @cindex frame layout
3046 @c prevent bad page break with this line
3047 Here is the basic stack layout.
3049 @defmac STACK_GROWS_DOWNWARD
3050 Define this macro if pushing a word onto the stack moves the stack
3051 pointer to a smaller address.
3053 When we say, ``define this macro if @dots{}'', it means that the
3054 compiler checks this macro only with @code{#ifdef} so the precise
3055 definition used does not matter.
3058 @defmac STACK_PUSH_CODE
3059 This macro defines the operation used when something is pushed
3060 on the stack. In RTL, a push operation will be
3061 @code{(set (mem (STACK_PUSH_CODE (reg sp))) @dots{})}
3063 The choices are @code{PRE_DEC}, @code{POST_DEC}, @code{PRE_INC},
3064 and @code{POST_INC}. Which of these is correct depends on
3065 the stack direction and on whether the stack pointer points
3066 to the last item on the stack or whether it points to the
3067 space for the next item on the stack.
3069 The default is @code{PRE_DEC} when @code{STACK_GROWS_DOWNWARD} is
3070 defined, which is almost always right, and @code{PRE_INC} otherwise,
3071 which is often wrong.
3074 @defmac FRAME_GROWS_DOWNWARD
3075 Define this macro to nonzero value if the addresses of local variable slots
3076 are at negative offsets from the frame pointer.
3079 @defmac ARGS_GROW_DOWNWARD
3080 Define this macro if successive arguments to a function occupy decreasing
3081 addresses on the stack.
3084 @defmac STARTING_FRAME_OFFSET
3085 Offset from the frame pointer to the first local variable slot to be allocated.
3087 If @code{FRAME_GROWS_DOWNWARD}, find the next slot's offset by
3088 subtracting the first slot's length from @code{STARTING_FRAME_OFFSET}.
3089 Otherwise, it is found by adding the length of the first slot to the
3090 value @code{STARTING_FRAME_OFFSET}.
3091 @c i'm not sure if the above is still correct.. had to change it to get
3092 @c rid of an overfull. --mew 2feb93
3095 @defmac STACK_ALIGNMENT_NEEDED
3096 Define to zero to disable final alignment of the stack during reload.
3097 The nonzero default for this macro is suitable for most ports.
3099 On ports where @code{STARTING_FRAME_OFFSET} is nonzero or where there
3100 is a register save block following the local block that doesn't require
3101 alignment to @code{STACK_BOUNDARY}, it may be beneficial to disable
3102 stack alignment and do it in the backend.
3105 @defmac STACK_POINTER_OFFSET
3106 Offset from the stack pointer register to the first location at which
3107 outgoing arguments are placed. If not specified, the default value of
3108 zero is used. This is the proper value for most machines.
3110 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
3111 the first location at which outgoing arguments are placed.
3114 @defmac FIRST_PARM_OFFSET (@var{fundecl})
3115 Offset from the argument pointer register to the first argument's
3116 address. On some machines it may depend on the data type of the
3119 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
3120 the first argument's address.
3123 @defmac STACK_DYNAMIC_OFFSET (@var{fundecl})
3124 Offset from the stack pointer register to an item dynamically allocated
3125 on the stack, e.g., by @code{alloca}.
3127 The default value for this macro is @code{STACK_POINTER_OFFSET} plus the
3128 length of the outgoing arguments. The default is correct for most
3129 machines. See @file{function.c} for details.
3132 @defmac INITIAL_FRAME_ADDRESS_RTX
3133 A C expression whose value is RTL representing the address of the initial
3134 stack frame. This address is passed to @code{RETURN_ADDR_RTX} and
3135 @code{DYNAMIC_CHAIN_ADDRESS}. If you don't define this macro, a reasonable
3136 default value will be used. Define this macro in order to make frame pointer
3137 elimination work in the presence of @code{__builtin_frame_address (count)} and
3138 @code{__builtin_return_address (count)} for @code{count} not equal to zero.
3141 @defmac DYNAMIC_CHAIN_ADDRESS (@var{frameaddr})
3142 A C expression whose value is RTL representing the address in a stack
3143 frame where the pointer to the caller's frame is stored. Assume that
3144 @var{frameaddr} is an RTL expression for the address of the stack frame
3147 If you don't define this macro, the default is to return the value
3148 of @var{frameaddr}---that is, the stack frame address is also the
3149 address of the stack word that points to the previous frame.
3152 @defmac SETUP_FRAME_ADDRESSES
3153 If defined, a C expression that produces the machine-specific code to
3154 setup the stack so that arbitrary frames can be accessed. For example,
3155 on the SPARC, we must flush all of the register windows to the stack
3156 before we can access arbitrary stack frames. You will seldom need to
3160 @deftypefn {Target Hook} rtx TARGET_BUILTIN_SETJMP_FRAME_VALUE (void)
3161 This target hook should return an rtx that is used to store
3162 the address of the current frame into the built in @code{setjmp} buffer.
3163 The default value, @code{virtual_stack_vars_rtx}, is correct for most
3164 machines. One reason you may need to define this target hook is if
3165 @code{hard_frame_pointer_rtx} is the appropriate value on your machine.
3168 @defmac FRAME_ADDR_RTX (@var{frameaddr})
3169 A C expression whose value is RTL representing the value of the frame
3170 address for the current frame. @var{frameaddr} is the frame pointer
3171 of the current frame. This is used for __builtin_frame_address.
3172 You need only define this macro if the frame address is not the same
3173 as the frame pointer. Most machines do not need to define it.
3176 @defmac RETURN_ADDR_RTX (@var{count}, @var{frameaddr})
3177 A C expression whose value is RTL representing the value of the return
3178 address for the frame @var{count} steps up from the current frame, after
3179 the prologue. @var{frameaddr} is the frame pointer of the @var{count}
3180 frame, or the frame pointer of the @var{count} @minus{} 1 frame if
3181 @code{RETURN_ADDR_IN_PREVIOUS_FRAME} is defined.
3183 The value of the expression must always be the correct address when
3184 @var{count} is zero, but may be @code{NULL_RTX} if there is no way to
3185 determine the return address of other frames.
3188 @defmac RETURN_ADDR_IN_PREVIOUS_FRAME
3189 Define this if the return address of a particular stack frame is accessed
3190 from the frame pointer of the previous stack frame.
3193 @defmac INCOMING_RETURN_ADDR_RTX
3194 A C expression whose value is RTL representing the location of the
3195 incoming return address at the beginning of any function, before the
3196 prologue. This RTL is either a @code{REG}, indicating that the return
3197 value is saved in @samp{REG}, or a @code{MEM} representing a location in
3200 You only need to define this macro if you want to support call frame
3201 debugging information like that provided by DWARF 2.
3203 If this RTL is a @code{REG}, you should also define
3204 @code{DWARF_FRAME_RETURN_COLUMN} to @code{DWARF_FRAME_REGNUM (REGNO)}.
3207 @defmac DWARF_ALT_FRAME_RETURN_COLUMN
3208 A C expression whose value is an integer giving a DWARF 2 column
3209 number that may be used as an alternative return column. The column
3210 must not correspond to any gcc hard register (that is, it must not
3211 be in the range of @code{DWARF_FRAME_REGNUM}).
3213 This macro can be useful if @code{DWARF_FRAME_RETURN_COLUMN} is set to a
3214 general register, but an alternative column needs to be used for signal
3215 frames. Some targets have also used different frame return columns
3219 @defmac DWARF_ZERO_REG
3220 A C expression whose value is an integer giving a DWARF 2 register
3221 number that is considered to always have the value zero. This should
3222 only be defined if the target has an architected zero register, and
3223 someone decided it was a good idea to use that register number to
3224 terminate the stack backtrace. New ports should avoid this.
3227 @deftypefn {Target Hook} void TARGET_DWARF_HANDLE_FRAME_UNSPEC (const char *@var{label}, rtx @var{pattern}, int @var{index})
3228 This target hook allows the backend to emit frame-related insns that
3229 contain UNSPECs or UNSPEC_VOLATILEs. The DWARF 2 call frame debugging
3230 info engine will invoke it on insns of the form
3232 (set (reg) (unspec [@dots{}] UNSPEC_INDEX))
3236 (set (reg) (unspec_volatile [@dots{}] UNSPECV_INDEX)).
3238 to let the backend emit the call frame instructions. @var{label} is
3239 the CFI label attached to the insn, @var{pattern} is the pattern of
3240 the insn and @var{index} is @code{UNSPEC_INDEX} or @code{UNSPECV_INDEX}.
3243 @defmac INCOMING_FRAME_SP_OFFSET
3244 A C expression whose value is an integer giving the offset, in bytes,
3245 from the value of the stack pointer register to the top of the stack
3246 frame at the beginning of any function, before the prologue. The top of
3247 the frame is defined to be the value of the stack pointer in the
3248 previous frame, just before the call instruction.
3250 You only need to define this macro if you want to support call frame
3251 debugging information like that provided by DWARF 2.
3254 @defmac ARG_POINTER_CFA_OFFSET (@var{fundecl})
3255 A C expression whose value is an integer giving the offset, in bytes,
3256 from the argument pointer to the canonical frame address (cfa). The
3257 final value should coincide with that calculated by
3258 @code{INCOMING_FRAME_SP_OFFSET}. Which is unfortunately not usable
3259 during virtual register instantiation.
3261 The default value for this macro is
3262 @code{FIRST_PARM_OFFSET (fundecl) + crtl->args.pretend_args_size},
3263 which is correct for most machines; in general, the arguments are found
3264 immediately before the stack frame. Note that this is not the case on
3265 some targets that save registers into the caller's frame, such as SPARC
3266 and rs6000, and so such targets need to define this macro.
3268 You only need to define this macro if the default is incorrect, and you
3269 want to support call frame debugging information like that provided by
3273 @defmac FRAME_POINTER_CFA_OFFSET (@var{fundecl})
3274 If defined, a C expression whose value is an integer giving the offset
3275 in bytes from the frame pointer to the canonical frame address (cfa).
3276 The final value should coincide with that calculated by
3277 @code{INCOMING_FRAME_SP_OFFSET}.
3279 Normally the CFA is calculated as an offset from the argument pointer,
3280 via @code{ARG_POINTER_CFA_OFFSET}, but if the argument pointer is
3281 variable due to the ABI, this may not be possible. If this macro is
3282 defined, it implies that the virtual register instantiation should be
3283 based on the frame pointer instead of the argument pointer. Only one
3284 of @code{FRAME_POINTER_CFA_OFFSET} and @code{ARG_POINTER_CFA_OFFSET}
3288 @defmac CFA_FRAME_BASE_OFFSET (@var{fundecl})
3289 If defined, a C expression whose value is an integer giving the offset
3290 in bytes from the canonical frame address (cfa) to the frame base used
3291 in DWARF 2 debug information. The default is zero. A different value
3292 may reduce the size of debug information on some ports.
3295 @node Exception Handling
3296 @subsection Exception Handling Support
3297 @cindex exception handling
3299 @defmac EH_RETURN_DATA_REGNO (@var{N})
3300 A C expression whose value is the @var{N}th register number used for
3301 data by exception handlers, or @code{INVALID_REGNUM} if fewer than
3302 @var{N} registers are usable.
3304 The exception handling library routines communicate with the exception
3305 handlers via a set of agreed upon registers. Ideally these registers
3306 should be call-clobbered; it is possible to use call-saved registers,
3307 but may negatively impact code size. The target must support at least
3308 2 data registers, but should define 4 if there are enough free registers.
3310 You must define this macro if you want to support call frame exception
3311 handling like that provided by DWARF 2.
3314 @defmac EH_RETURN_STACKADJ_RTX
3315 A C expression whose value is RTL representing a location in which
3316 to store a stack adjustment to be applied before function return.
3317 This is used to unwind the stack to an exception handler's call frame.
3318 It will be assigned zero on code paths that return normally.
3320 Typically this is a call-clobbered hard register that is otherwise
3321 untouched by the epilogue, but could also be a stack slot.
3323 Do not define this macro if the stack pointer is saved and restored
3324 by the regular prolog and epilog code in the call frame itself; in
3325 this case, the exception handling library routines will update the
3326 stack location to be restored in place. Otherwise, you must define
3327 this macro if you want to support call frame exception handling like
3328 that provided by DWARF 2.
3331 @defmac EH_RETURN_HANDLER_RTX
3332 A C expression whose value is RTL representing a location in which
3333 to store the address of an exception handler to which we should
3334 return. It will not be assigned on code paths that return normally.
3336 Typically this is the location in the call frame at which the normal
3337 return address is stored. For targets that return by popping an
3338 address off the stack, this might be a memory address just below
3339 the @emph{target} call frame rather than inside the current call
3340 frame. If defined, @code{EH_RETURN_STACKADJ_RTX} will have already
3341 been assigned, so it may be used to calculate the location of the
3344 Some targets have more complex requirements than storing to an
3345 address calculable during initial code generation. In that case
3346 the @code{eh_return} instruction pattern should be used instead.
3348 If you want to support call frame exception handling, you must
3349 define either this macro or the @code{eh_return} instruction pattern.
3352 @defmac RETURN_ADDR_OFFSET
3353 If defined, an integer-valued C expression for which rtl will be generated
3354 to add it to the exception handler address before it is searched in the
3355 exception handling tables, and to subtract it again from the address before
3356 using it to return to the exception handler.
3359 @defmac ASM_PREFERRED_EH_DATA_FORMAT (@var{code}, @var{global})
3360 This macro chooses the encoding of pointers embedded in the exception
3361 handling sections. If at all possible, this should be defined such
3362 that the exception handling section will not require dynamic relocations,
3363 and so may be read-only.
3365 @var{code} is 0 for data, 1 for code labels, 2 for function pointers.
3366 @var{global} is true if the symbol may be affected by dynamic relocations.
3367 The macro should return a combination of the @code{DW_EH_PE_*} defines
3368 as found in @file{dwarf2.h}.
3370 If this macro is not defined, pointers will not be encoded but
3371 represented directly.
3374 @defmac ASM_MAYBE_OUTPUT_ENCODED_ADDR_RTX (@var{file}, @var{encoding}, @var{size}, @var{addr}, @var{done})
3375 This macro allows the target to emit whatever special magic is required
3376 to represent the encoding chosen by @code{ASM_PREFERRED_EH_DATA_FORMAT}.
3377 Generic code takes care of pc-relative and indirect encodings; this must
3378 be defined if the target uses text-relative or data-relative encodings.
3380 This is a C statement that branches to @var{done} if the format was
3381 handled. @var{encoding} is the format chosen, @var{size} is the number
3382 of bytes that the format occupies, @var{addr} is the @code{SYMBOL_REF}
3386 @defmac MD_FALLBACK_FRAME_STATE_FOR (@var{context}, @var{fs})
3387 This macro allows the target to add CPU and operating system specific
3388 code to the call-frame unwinder for use when there is no unwind data
3389 available. The most common reason to implement this macro is to unwind
3390 through signal frames.
3392 This macro is called from @code{uw_frame_state_for} in
3393 @file{unwind-dw2.c}, @file{unwind-dw2-xtensa.c} and
3394 @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context};
3395 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{context->ra}
3396 for the address of the code being executed and @code{context->cfa} for
3397 the stack pointer value. If the frame can be decoded, the register
3398 save addresses should be updated in @var{fs} and the macro should
3399 evaluate to @code{_URC_NO_REASON}. If the frame cannot be decoded,
3400 the macro should evaluate to @code{_URC_END_OF_STACK}.
3402 For proper signal handling in Java this macro is accompanied by
3403 @code{MAKE_THROW_FRAME}, defined in @file{libjava/include/*-signal.h} headers.
3406 @defmac MD_HANDLE_UNWABI (@var{context}, @var{fs})
3407 This macro allows the target to add operating system specific code to the
3408 call-frame unwinder to handle the IA-64 @code{.unwabi} unwinding directive,
3409 usually used for signal or interrupt frames.
3411 This macro is called from @code{uw_update_context} in @file{unwind-ia64.c}.
3412 @var{context} is an @code{_Unwind_Context};
3413 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{fs->unwabi}
3414 for the abi and context in the @code{.unwabi} directive. If the
3415 @code{.unwabi} directive can be handled, the register save addresses should
3416 be updated in @var{fs}.
3419 @defmac TARGET_USES_WEAK_UNWIND_INFO
3420 A C expression that evaluates to true if the target requires unwind
3421 info to be given comdat linkage. Define it to be @code{1} if comdat
3422 linkage is necessary. The default is @code{0}.
3425 @node Stack Checking
3426 @subsection Specifying How Stack Checking is Done
3428 GCC will check that stack references are within the boundaries of the
3429 stack, if the option @option{-fstack-check} is specified, in one of
3434 If the value of the @code{STACK_CHECK_BUILTIN} macro is nonzero, GCC
3435 will assume that you have arranged for full stack checking to be done
3436 at appropriate places in the configuration files. GCC will not do
3437 other special processing.
3440 If @code{STACK_CHECK_BUILTIN} is zero and the value of the
3441 @code{STACK_CHECK_STATIC_BUILTIN} macro is nonzero, GCC will assume
3442 that you have arranged for static stack checking (checking of the
3443 static stack frame of functions) to be done at appropriate places
3444 in the configuration files. GCC will only emit code to do dynamic
3445 stack checking (checking on dynamic stack allocations) using the third
3449 If neither of the above are true, GCC will generate code to periodically
3450 ``probe'' the stack pointer using the values of the macros defined below.
3453 If neither STACK_CHECK_BUILTIN nor STACK_CHECK_STATIC_BUILTIN is defined,
3454 GCC will change its allocation strategy for large objects if the option
3455 @option{-fstack-check} is specified: they will always be allocated
3456 dynamically if their size exceeds @code{STACK_CHECK_MAX_VAR_SIZE} bytes.
3458 @defmac STACK_CHECK_BUILTIN
3459 A nonzero value if stack checking is done by the configuration files in a
3460 machine-dependent manner. You should define this macro if stack checking
3461 is required by the ABI of your machine or if you would like to do stack
3462 checking in some more efficient way than the generic approach. The default
3463 value of this macro is zero.
3466 @defmac STACK_CHECK_STATIC_BUILTIN
3467 A nonzero value if static stack checking is done by the configuration files
3468 in a machine-dependent manner. You should define this macro if you would
3469 like to do static stack checking in some more efficient way than the generic
3470 approach. The default value of this macro is zero.
3473 @defmac STACK_CHECK_PROBE_INTERVAL_EXP
3474 An integer specifying the interval at which GCC must generate stack probe
3475 instructions, defined as 2 raised to this integer. You will normally
3476 define this macro so that the interval be no larger than the size of
3477 the ``guard pages'' at the end of a stack area. The default value
3478 of 12 (4096-byte interval) is suitable for most systems.
3481 @defmac STACK_CHECK_MOVING_SP
3482 An integer which is nonzero if GCC should move the stack pointer page by page
3483 when doing probes. This can be necessary on systems where the stack pointer
3484 contains the bottom address of the memory area accessible to the executing
3485 thread at any point in time. In this situation an alternate signal stack
3486 is required in order to be able to recover from a stack overflow. The
3487 default value of this macro is zero.
3490 @defmac STACK_CHECK_PROTECT
3491 The number of bytes of stack needed to recover from a stack overflow, for
3492 languages where such a recovery is supported. The default value of 75 words
3493 with the @code{setjmp}/@code{longjmp}-based exception handling mechanism and
3494 8192 bytes with other exception handling mechanisms should be adequate for
3498 The following macros are relevant only if neither STACK_CHECK_BUILTIN
3499 nor STACK_CHECK_STATIC_BUILTIN is defined; you can omit them altogether
3500 in the opposite case.
3502 @defmac STACK_CHECK_MAX_FRAME_SIZE
3503 The maximum size of a stack frame, in bytes. GCC will generate probe
3504 instructions in non-leaf functions to ensure at least this many bytes of
3505 stack are available. If a stack frame is larger than this size, stack
3506 checking will not be reliable and GCC will issue a warning. The
3507 default is chosen so that GCC only generates one instruction on most
3508 systems. You should normally not change the default value of this macro.
3511 @defmac STACK_CHECK_FIXED_FRAME_SIZE
3512 GCC uses this value to generate the above warning message. It
3513 represents the amount of fixed frame used by a function, not including
3514 space for any callee-saved registers, temporaries and user variables.
3515 You need only specify an upper bound for this amount and will normally
3516 use the default of four words.
3519 @defmac STACK_CHECK_MAX_VAR_SIZE
3520 The maximum size, in bytes, of an object that GCC will place in the
3521 fixed area of the stack frame when the user specifies
3522 @option{-fstack-check}.
3523 GCC computed the default from the values of the above macros and you will
3524 normally not need to override that default.
3528 @node Frame Registers
3529 @subsection Registers That Address the Stack Frame
3531 @c prevent bad page break with this line
3532 This discusses registers that address the stack frame.
3534 @defmac STACK_POINTER_REGNUM
3535 The register number of the stack pointer register, which must also be a
3536 fixed register according to @code{FIXED_REGISTERS}. On most machines,
3537 the hardware determines which register this is.
3540 @defmac FRAME_POINTER_REGNUM
3541 The register number of the frame pointer register, which is used to
3542 access automatic variables in the stack frame. On some machines, the
3543 hardware determines which register this is. On other machines, you can
3544 choose any register you wish for this purpose.
3547 @defmac HARD_FRAME_POINTER_REGNUM
3548 On some machines the offset between the frame pointer and starting
3549 offset of the automatic variables is not known until after register
3550 allocation has been done (for example, because the saved registers are
3551 between these two locations). On those machines, define
3552 @code{FRAME_POINTER_REGNUM} the number of a special, fixed register to
3553 be used internally until the offset is known, and define
3554 @code{HARD_FRAME_POINTER_REGNUM} to be the actual hard register number
3555 used for the frame pointer.
3557 You should define this macro only in the very rare circumstances when it
3558 is not possible to calculate the offset between the frame pointer and
3559 the automatic variables until after register allocation has been
3560 completed. When this macro is defined, you must also indicate in your
3561 definition of @code{ELIMINABLE_REGS} how to eliminate
3562 @code{FRAME_POINTER_REGNUM} into either @code{HARD_FRAME_POINTER_REGNUM}
3563 or @code{STACK_POINTER_REGNUM}.
3565 Do not define this macro if it would be the same as
3566 @code{FRAME_POINTER_REGNUM}.
3569 @defmac ARG_POINTER_REGNUM
3570 The register number of the arg pointer register, which is used to access
3571 the function's argument list. On some machines, this is the same as the
3572 frame pointer register. On some machines, the hardware determines which
3573 register this is. On other machines, you can choose any register you
3574 wish for this purpose. If this is not the same register as the frame
3575 pointer register, then you must mark it as a fixed register according to
3576 @code{FIXED_REGISTERS}, or arrange to be able to eliminate it
3577 (@pxref{Elimination}).
3580 @defmac HARD_FRAME_POINTER_IS_FRAME_POINTER
3581 Define this to a preprocessor constant that is nonzero if
3582 @code{hard_frame_pointer_rtx} and @code{frame_pointer_rtx} should be
3583 the same. The default definition is @samp{(HARD_FRAME_POINTER_REGNUM
3584 == FRAME_POINTER_REGNUM)}; you only need to define this macro if that
3585 definition is not suitable for use in preprocessor conditionals.
3588 @defmac HARD_FRAME_POINTER_IS_ARG_POINTER
3589 Define this to a preprocessor constant that is nonzero if
3590 @code{hard_frame_pointer_rtx} and @code{arg_pointer_rtx} should be the
3591 same. The default definition is @samp{(HARD_FRAME_POINTER_REGNUM ==
3592 ARG_POINTER_REGNUM)}; you only need to define this macro if that
3593 definition is not suitable for use in preprocessor conditionals.
3596 @defmac RETURN_ADDRESS_POINTER_REGNUM
3597 The register number of the return address pointer register, which is used to
3598 access the current function's return address from the stack. On some
3599 machines, the return address is not at a fixed offset from the frame
3600 pointer or stack pointer or argument pointer. This register can be defined
3601 to point to the return address on the stack, and then be converted by
3602 @code{ELIMINABLE_REGS} into either the frame pointer or stack pointer.
3604 Do not define this macro unless there is no other way to get the return
3605 address from the stack.
3608 @defmac STATIC_CHAIN_REGNUM
3609 @defmacx STATIC_CHAIN_INCOMING_REGNUM
3610 Register numbers used for passing a function's static chain pointer. If
3611 register windows are used, the register number as seen by the called
3612 function is @code{STATIC_CHAIN_INCOMING_REGNUM}, while the register
3613 number as seen by the calling function is @code{STATIC_CHAIN_REGNUM}. If
3614 these registers are the same, @code{STATIC_CHAIN_INCOMING_REGNUM} need
3617 The static chain register need not be a fixed register.
3619 If the static chain is passed in memory, these macros should not be
3620 defined; instead, the @code{TARGET_STATIC_CHAIN} hook should be used.
3623 @deftypefn {Target Hook} rtx TARGET_STATIC_CHAIN (const_tree @var{fndecl}, bool @var{incoming_p})
3624 This hook replaces the use of @code{STATIC_CHAIN_REGNUM} et al for
3625 targets that may use different static chain locations for different
3626 nested functions. This may be required if the target has function
3627 attributes that affect the calling conventions of the function and
3628 those calling conventions use different static chain locations.
3630 The default version of this hook uses @code{STATIC_CHAIN_REGNUM} et al.
3632 If the static chain is passed in memory, this hook should be used to
3633 provide rtx giving @code{mem} expressions that denote where they are stored.
3634 Often the @code{mem} expression as seen by the caller will be at an offset
3635 from the stack pointer and the @code{mem} expression as seen by the callee
3636 will be at an offset from the frame pointer.
3637 @findex stack_pointer_rtx
3638 @findex frame_pointer_rtx
3639 @findex arg_pointer_rtx
3640 The variables @code{stack_pointer_rtx}, @code{frame_pointer_rtx}, and
3641 @code{arg_pointer_rtx} will have been initialized and should be used
3642 to refer to those items.
3645 @defmac DWARF_FRAME_REGISTERS
3646 This macro specifies the maximum number of hard registers that can be
3647 saved in a call frame. This is used to size data structures used in
3648 DWARF2 exception handling.
3650 Prior to GCC 3.0, this macro was needed in order to establish a stable
3651 exception handling ABI in the face of adding new hard registers for ISA
3652 extensions. In GCC 3.0 and later, the EH ABI is insulated from changes
3653 in the number of hard registers. Nevertheless, this macro can still be
3654 used to reduce the runtime memory requirements of the exception handling
3655 routines, which can be substantial if the ISA contains a lot of
3656 registers that are not call-saved.
3658 If this macro is not defined, it defaults to
3659 @code{FIRST_PSEUDO_REGISTER}.
3662 @defmac PRE_GCC3_DWARF_FRAME_REGISTERS
3664 This macro is similar to @code{DWARF_FRAME_REGISTERS}, but is provided
3665 for backward compatibility in pre GCC 3.0 compiled code.
3667 If this macro is not defined, it defaults to
3668 @code{DWARF_FRAME_REGISTERS}.
3671 @defmac DWARF_REG_TO_UNWIND_COLUMN (@var{regno})
3673 Define this macro if the target's representation for dwarf registers
3674 is different than the internal representation for unwind column.
3675 Given a dwarf register, this macro should return the internal unwind
3676 column number to use instead.
3678 See the PowerPC's SPE target for an example.
3681 @defmac DWARF_FRAME_REGNUM (@var{regno})
3683 Define this macro if the target's representation for dwarf registers
3684 used in .eh_frame or .debug_frame is different from that used in other
3685 debug info sections. Given a GCC hard register number, this macro
3686 should return the .eh_frame register number. The default is
3687 @code{DBX_REGISTER_NUMBER (@var{regno})}.
3691 @defmac DWARF2_FRAME_REG_OUT (@var{regno}, @var{for_eh})
3693 Define this macro to map register numbers held in the call frame info
3694 that GCC has collected using @code{DWARF_FRAME_REGNUM} to those that
3695 should be output in .debug_frame (@code{@var{for_eh}} is zero) and
3696 .eh_frame (@code{@var{for_eh}} is nonzero). The default is to
3697 return @code{@var{regno}}.
3701 @defmac REG_VALUE_IN_UNWIND_CONTEXT
3703 Define this macro if the target stores register values as
3704 @code{_Unwind_Word} type in unwind context. It should be defined if
3705 target register size is larger than the size of @code{void *}. The
3706 default is to store register values as @code{void *} type.
3710 @defmac ASSUME_EXTENDED_UNWIND_CONTEXT
3712 Define this macro to be 1 if the target always uses extended unwind
3713 context with version, args_size and by_value fields. If it is undefined,
3714 it will be defined to 1 when @code{REG_VALUE_IN_UNWIND_CONTEXT} is
3715 defined and 0 otherwise.
3720 @subsection Eliminating Frame Pointer and Arg Pointer
3722 @c prevent bad page break with this line
3723 This is about eliminating the frame pointer and arg pointer.
3725 @deftypefn {Target Hook} bool TARGET_FRAME_POINTER_REQUIRED (void)
3726 This target hook should return @code{true} if a function must have and use
3727 a frame pointer. This target hook is called in the reload pass. If its return
3728 value is @code{true} the function will have a frame pointer.
3730 This target hook can in principle examine the current function and decide
3731 according to the facts, but on most machines the constant @code{false} or the
3732 constant @code{true} suffices. Use @code{false} when the machine allows code
3733 to be generated with no frame pointer, and doing so saves some time or space.
3734 Use @code{true} when there is no possible advantage to avoiding a frame
3737 In certain cases, the compiler does not know how to produce valid code
3738 without a frame pointer. The compiler recognizes those cases and
3739 automatically gives the function a frame pointer regardless of what
3740 @code{TARGET_FRAME_POINTER_REQUIRED} returns. You don't need to worry about
3743 In a function that does not require a frame pointer, the frame pointer
3744 register can be allocated for ordinary usage, unless you mark it as a
3745 fixed register. See @code{FIXED_REGISTERS} for more information.
3747 Default return value is @code{false}.
3750 @findex get_frame_size
3751 @defmac INITIAL_FRAME_POINTER_OFFSET (@var{depth-var})
3752 A C statement to store in the variable @var{depth-var} the difference
3753 between the frame pointer and the stack pointer values immediately after
3754 the function prologue. The value would be computed from information
3755 such as the result of @code{get_frame_size ()} and the tables of
3756 registers @code{regs_ever_live} and @code{call_used_regs}.
3758 If @code{ELIMINABLE_REGS} is defined, this macro will be not be used and
3759 need not be defined. Otherwise, it must be defined even if
3760 @code{TARGET_FRAME_POINTER_REQUIRED} always returns true; in that
3761 case, you may set @var{depth-var} to anything.
3764 @defmac ELIMINABLE_REGS
3765 If defined, this macro specifies a table of register pairs used to
3766 eliminate unneeded registers that point into the stack frame. If it is not
3767 defined, the only elimination attempted by the compiler is to replace
3768 references to the frame pointer with references to the stack pointer.
3770 The definition of this macro is a list of structure initializations, each
3771 of which specifies an original and replacement register.
3773 On some machines, the position of the argument pointer is not known until
3774 the compilation is completed. In such a case, a separate hard register
3775 must be used for the argument pointer. This register can be eliminated by
3776 replacing it with either the frame pointer or the argument pointer,
3777 depending on whether or not the frame pointer has been eliminated.
3779 In this case, you might specify:
3781 #define ELIMINABLE_REGS \
3782 @{@{ARG_POINTER_REGNUM, STACK_POINTER_REGNUM@}, \
3783 @{ARG_POINTER_REGNUM, FRAME_POINTER_REGNUM@}, \
3784 @{FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM@}@}
3787 Note that the elimination of the argument pointer with the stack pointer is
3788 specified first since that is the preferred elimination.
3791 @deftypefn {Target Hook} bool TARGET_CAN_ELIMINATE (const int @var{from_reg}, const int @var{to_reg})
3792 This target hook should returns @code{true} if the compiler is allowed to
3793 try to replace register number @var{from_reg} with register number
3794 @var{to_reg}. This target hook need only be defined if @code{ELIMINABLE_REGS}
3795 is defined, and will usually be @code{true}, since most of the cases
3796 preventing register elimination are things that the compiler already
3799 Default return value is @code{true}.
3802 @defmac INITIAL_ELIMINATION_OFFSET (@var{from-reg}, @var{to-reg}, @var{offset-var})
3803 This macro is similar to @code{INITIAL_FRAME_POINTER_OFFSET}. It
3804 specifies the initial difference between the specified pair of
3805 registers. This macro must be defined if @code{ELIMINABLE_REGS} is
3809 @node Stack Arguments
3810 @subsection Passing Function Arguments on the Stack
3811 @cindex arguments on stack
3812 @cindex stack arguments
3814 The macros in this section control how arguments are passed
3815 on the stack. See the following section for other macros that
3816 control passing certain arguments in registers.
3818 @deftypefn {Target Hook} bool TARGET_PROMOTE_PROTOTYPES (const_tree @var{fntype})
3819 This target hook returns @code{true} if an argument declared in a
3820 prototype as an integral type smaller than @code{int} should actually be
3821 passed as an @code{int}. In addition to avoiding errors in certain
3822 cases of mismatch, it also makes for better code on certain machines.
3823 The default is to not promote prototypes.
3827 A C expression. If nonzero, push insns will be used to pass
3829 If the target machine does not have a push instruction, set it to zero.
3830 That directs GCC to use an alternate strategy: to
3831 allocate the entire argument block and then store the arguments into
3832 it. When @code{PUSH_ARGS} is nonzero, @code{PUSH_ROUNDING} must be defined too.
3835 @defmac PUSH_ARGS_REVERSED
3836 A C expression. If nonzero, function arguments will be evaluated from
3837 last to first, rather than from first to last. If this macro is not
3838 defined, it defaults to @code{PUSH_ARGS} on targets where the stack
3839 and args grow in opposite directions, and 0 otherwise.
3842 @defmac PUSH_ROUNDING (@var{npushed})
3843 A C expression that is the number of bytes actually pushed onto the
3844 stack when an instruction attempts to push @var{npushed} bytes.
3846 On some machines, the definition
3849 #define PUSH_ROUNDING(BYTES) (BYTES)
3853 will suffice. But on other machines, instructions that appear
3854 to push one byte actually push two bytes in an attempt to maintain
3855 alignment. Then the definition should be
3858 #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1)
3861 If the value of this macro has a type, it should be an unsigned type.
3864 @findex current_function_outgoing_args_size
3865 @defmac ACCUMULATE_OUTGOING_ARGS
3866 A C expression. If nonzero, the maximum amount of space required for outgoing arguments
3867 will be computed and placed into the variable
3868 @code{current_function_outgoing_args_size}. No space will be pushed
3869 onto the stack for each call; instead, the function prologue should
3870 increase the stack frame size by this amount.
3872 Setting both @code{PUSH_ARGS} and @code{ACCUMULATE_OUTGOING_ARGS}
3876 @defmac REG_PARM_STACK_SPACE (@var{fndecl})
3877 Define this macro if functions should assume that stack space has been
3878 allocated for arguments even when their values are passed in
3881 The value of this macro is the size, in bytes, of the area reserved for
3882 arguments passed in registers for the function represented by @var{fndecl},
3883 which can be zero if GCC is calling a library function.
3884 The argument @var{fndecl} can be the FUNCTION_DECL, or the type itself
3887 This space can be allocated by the caller, or be a part of the
3888 machine-dependent stack frame: @code{OUTGOING_REG_PARM_STACK_SPACE} says
3891 @c above is overfull. not sure what to do. --mew 5feb93 did
3892 @c something, not sure if it looks good. --mew 10feb93
3894 @defmac OUTGOING_REG_PARM_STACK_SPACE (@var{fntype})
3895 Define this to a nonzero value if it is the responsibility of the
3896 caller to allocate the area reserved for arguments passed in registers
3897 when calling a function of @var{fntype}. @var{fntype} may be NULL
3898 if the function called is a library function.
3900 If @code{ACCUMULATE_OUTGOING_ARGS} is defined, this macro controls
3901 whether the space for these arguments counts in the value of
3902 @code{current_function_outgoing_args_size}.
3905 @defmac STACK_PARMS_IN_REG_PARM_AREA
3906 Define this macro if @code{REG_PARM_STACK_SPACE} is defined, but the
3907 stack parameters don't skip the area specified by it.
3908 @c i changed this, makes more sens and it should have taken care of the
3909 @c overfull.. not as specific, tho. --mew 5feb93
3911 Normally, when a parameter is not passed in registers, it is placed on the
3912 stack beyond the @code{REG_PARM_STACK_SPACE} area. Defining this macro
3913 suppresses this behavior and causes the parameter to be passed on the
3914 stack in its natural location.
3917 @deftypefn {Target Hook} int TARGET_RETURN_POPS_ARGS (tree @var{fundecl}, tree @var{funtype}, int @var{size})
3918 This target hook returns the number of bytes of its own arguments that
3919 a function pops on returning, or 0 if the function pops no arguments
3920 and the caller must therefore pop them all after the function returns.
3922 @var{fundecl} is a C variable whose value is a tree node that describes
3923 the function in question. Normally it is a node of type
3924 @code{FUNCTION_DECL} that describes the declaration of the function.
3925 From this you can obtain the @code{DECL_ATTRIBUTES} of the function.
3927 @var{funtype} is a C variable whose value is a tree node that
3928 describes the function in question. Normally it is a node of type
3929 @code{FUNCTION_TYPE} that describes the data type of the function.
3930 From this it is possible to obtain the data types of the value and
3931 arguments (if known).
3933 When a call to a library function is being considered, @var{fundecl}
3934 will contain an identifier node for the library function. Thus, if
3935 you need to distinguish among various library functions, you can do so
3936 by their names. Note that ``library function'' in this context means
3937 a function used to perform arithmetic, whose name is known specially
3938 in the compiler and was not mentioned in the C code being compiled.
3940 @var{size} is the number of bytes of arguments passed on the
3941 stack. If a variable number of bytes is passed, it is zero, and
3942 argument popping will always be the responsibility of the calling function.
3944 On the VAX, all functions always pop their arguments, so the definition
3945 of this macro is @var{size}. On the 68000, using the standard
3946 calling convention, no functions pop their arguments, so the value of
3947 the macro is always 0 in this case. But an alternative calling
3948 convention is available in which functions that take a fixed number of
3949 arguments pop them but other functions (such as @code{printf}) pop
3950 nothing (the caller pops all). When this convention is in use,
3951 @var{funtype} is examined to determine whether a function takes a fixed
3952 number of arguments.
3955 @defmac CALL_POPS_ARGS (@var{cum})
3956 A C expression that should indicate the number of bytes a call sequence
3957 pops off the stack. It is added to the value of @code{RETURN_POPS_ARGS}
3958 when compiling a function call.
3960 @var{cum} is the variable in which all arguments to the called function
3961 have been accumulated.
3963 On certain architectures, such as the SH5, a call trampoline is used
3964 that pops certain registers off the stack, depending on the arguments
3965 that have been passed to the function. Since this is a property of the
3966 call site, not of the called function, @code{RETURN_POPS_ARGS} is not
3970 @node Register Arguments
3971 @subsection Passing Arguments in Registers
3972 @cindex arguments in registers
3973 @cindex registers arguments
3975 This section describes the macros which let you control how various
3976 types of arguments are passed in registers or how they are arranged in
3979 @deftypefn {Target Hook} rtx TARGET_FUNCTION_ARG (cumulative_args_t @var{ca}, enum machine_mode @var{mode}, const_tree @var{type}, bool @var{named})
3980 Return an RTX indicating whether a function argument is passed in a
3981 register and if so, which register.
3983 The arguments are @var{ca}, which summarizes all the previous
3984 arguments; @var{mode}, the machine mode of the argument; @var{type},
3985 the data type of the argument as a tree node or 0 if that is not known
3986 (which happens for C support library functions); and @var{named},
3987 which is @code{true} for an ordinary argument and @code{false} for
3988 nameless arguments that correspond to @samp{@dots{}} in the called
3989 function's prototype. @var{type} can be an incomplete type if a
3990 syntax error has previously occurred.
3992 The return value is usually either a @code{reg} RTX for the hard
3993 register in which to pass the argument, or zero to pass the argument
3996 The value of the expression can also be a @code{parallel} RTX@. This is
3997 used when an argument is passed in multiple locations. The mode of the
3998 @code{parallel} should be the mode of the entire argument. The
3999 @code{parallel} holds any number of @code{expr_list} pairs; each one
4000 describes where part of the argument is passed. In each
4001 @code{expr_list} the first operand must be a @code{reg} RTX for the hard
4002 register in which to pass this part of the argument, and the mode of the
4003 register RTX indicates how large this part of the argument is. The
4004 second operand of the @code{expr_list} is a @code{const_int} which gives
4005 the offset in bytes into the entire argument of where this part starts.
4006 As a special exception the first @code{expr_list} in the @code{parallel}
4007 RTX may have a first operand of zero. This indicates that the entire
4008 argument is also stored on the stack.
4010 The last time this hook is called, it is called with @code{MODE ==
4011 VOIDmode}, and its result is passed to the @code{call} or @code{call_value}
4012 pattern as operands 2 and 3 respectively.
4014 @cindex @file{stdarg.h} and register arguments
4015 The usual way to make the ISO library @file{stdarg.h} work on a
4016 machine where some arguments are usually passed in registers, is to
4017 cause nameless arguments to be passed on the stack instead. This is
4018 done by making @code{TARGET_FUNCTION_ARG} return 0 whenever
4019 @var{named} is @code{false}.
4021 @cindex @code{TARGET_MUST_PASS_IN_STACK}, and @code{TARGET_FUNCTION_ARG}
4022 @cindex @code{REG_PARM_STACK_SPACE}, and @code{TARGET_FUNCTION_ARG}
4023 You may use the hook @code{targetm.calls.must_pass_in_stack}
4024 in the definition of this macro to determine if this argument is of a
4025 type that must be passed in the stack. If @code{REG_PARM_STACK_SPACE}
4026 is not defined and @code{TARGET_FUNCTION_ARG} returns nonzero for such an
4027 argument, the compiler will abort. If @code{REG_PARM_STACK_SPACE} is
4028 defined, the argument will be computed in the stack and then loaded into
4032 @deftypefn {Target Hook} bool TARGET_MUST_PASS_IN_STACK (enum machine_mode @var{mode}, const_tree @var{type})
4033 This target hook should return @code{true} if we should not pass @var{type}
4034 solely in registers. The file @file{expr.h} defines a
4035 definition that is usually appropriate, refer to @file{expr.h} for additional
4039 @deftypefn {Target Hook} rtx TARGET_FUNCTION_INCOMING_ARG (cumulative_args_t @var{ca}, enum machine_mode @var{mode}, const_tree @var{type}, bool @var{named})
4040 Define this hook if the target machine has ``register windows'', so
4041 that the register in which a function sees an arguments is not
4042 necessarily the same as the one in which the caller passed the
4045 For such machines, @code{TARGET_FUNCTION_ARG} computes the register in
4046 which the caller passes the value, and
4047 @code{TARGET_FUNCTION_INCOMING_ARG} should be defined in a similar
4048 fashion to tell the function being called where the arguments will
4051 If @code{TARGET_FUNCTION_INCOMING_ARG} is not defined,
4052 @code{TARGET_FUNCTION_ARG} serves both purposes.
4055 @deftypefn {Target Hook} int TARGET_ARG_PARTIAL_BYTES (cumulative_args_t @var{cum}, enum machine_mode @var{mode}, tree @var{type}, bool @var{named})
4056 This target hook returns the number of bytes at the beginning of an
4057 argument that must be put in registers. The value must be zero for
4058 arguments that are passed entirely in registers or that are entirely
4059 pushed on the stack.
4061 On some machines, certain arguments must be passed partially in
4062 registers and partially in memory. On these machines, typically the
4063 first few words of arguments are passed in registers, and the rest
4064 on the stack. If a multi-word argument (a @code{double} or a
4065 structure) crosses that boundary, its first few words must be passed
4066 in registers and the rest must be pushed. This macro tells the
4067 compiler when this occurs, and how many bytes should go in registers.
4069 @code{TARGET_FUNCTION_ARG} for these arguments should return the first
4070 register to be used by the caller for this argument; likewise
4071 @code{TARGET_FUNCTION_INCOMING_ARG}, for the called function.
4074 @deftypefn {Target Hook} bool TARGET_PASS_BY_REFERENCE (cumulative_args_t @var{cum}, enum machine_mode @var{mode}, const_tree @var{type}, bool @var{named})
4075 This target hook should return @code{true} if an argument at the
4076 position indicated by @var{cum} should be passed by reference. This
4077 predicate is queried after target independent reasons for being
4078 passed by reference, such as @code{TREE_ADDRESSABLE (type)}.
4080 If the hook returns true, a copy of that argument is made in memory and a
4081 pointer to the argument is passed instead of the argument itself.
4082 The pointer is passed in whatever way is appropriate for passing a pointer
4086 @deftypefn {Target Hook} bool TARGET_CALLEE_COPIES (cumulative_args_t @var{cum}, enum machine_mode @var{mode}, const_tree @var{type}, bool @var{named})
4087 The function argument described by the parameters to this hook is
4088 known to be passed by reference. The hook should return true if the
4089 function argument should be copied by the callee instead of copied
4092 For any argument for which the hook returns true, if it can be
4093 determined that the argument is not modified, then a copy need
4096 The default version of this hook always returns false.
4099 @defmac CUMULATIVE_ARGS
4100 A C type for declaring a variable that is used as the first argument
4101 of @code{TARGET_FUNCTION_ARG} and other related values. For some
4102 target machines, the type @code{int} suffices and can hold the number
4103 of bytes of argument so far.
4105 There is no need to record in @code{CUMULATIVE_ARGS} anything about the
4106 arguments that have been passed on the stack. The compiler has other
4107 variables to keep track of that. For target machines on which all
4108 arguments are passed on the stack, there is no need to store anything in
4109 @code{CUMULATIVE_ARGS}; however, the data structure must exist and
4110 should not be empty, so use @code{int}.
4113 @defmac OVERRIDE_ABI_FORMAT (@var{fndecl})
4114 If defined, this macro is called before generating any code for a
4115 function, but after the @var{cfun} descriptor for the function has been
4116 created. The back end may use this macro to update @var{cfun} to
4117 reflect an ABI other than that which would normally be used by default.
4118 If the compiler is generating code for a compiler-generated function,
4119 @var{fndecl} may be @code{NULL}.
4122 @defmac INIT_CUMULATIVE_ARGS (@var{cum}, @var{fntype}, @var{libname}, @var{fndecl}, @var{n_named_args})
4123 A C statement (sans semicolon) for initializing the variable
4124 @var{cum} for the state at the beginning of the argument list. The
4125 variable has type @code{CUMULATIVE_ARGS}. The value of @var{fntype}
4126 is the tree node for the data type of the function which will receive
4127 the args, or 0 if the args are to a compiler support library function.
4128 For direct calls that are not libcalls, @var{fndecl} contain the
4129 declaration node of the function. @var{fndecl} is also set when
4130 @code{INIT_CUMULATIVE_ARGS} is used to find arguments for the function
4131 being compiled. @var{n_named_args} is set to the number of named
4132 arguments, including a structure return address if it is passed as a
4133 parameter, when making a call. When processing incoming arguments,
4134 @var{n_named_args} is set to @minus{}1.
4136 When processing a call to a compiler support library function,
4137 @var{libname} identifies which one. It is a @code{symbol_ref} rtx which
4138 contains the name of the function, as a string. @var{libname} is 0 when
4139 an ordinary C function call is being processed. Thus, each time this
4140 macro is called, either @var{libname} or @var{fntype} is nonzero, but
4141 never both of them at once.
4144 @defmac INIT_CUMULATIVE_LIBCALL_ARGS (@var{cum}, @var{mode}, @var{libname})
4145 Like @code{INIT_CUMULATIVE_ARGS} but only used for outgoing libcalls,
4146 it gets a @code{MODE} argument instead of @var{fntype}, that would be
4147 @code{NULL}. @var{indirect} would always be zero, too. If this macro
4148 is not defined, @code{INIT_CUMULATIVE_ARGS (cum, NULL_RTX, libname,
4149 0)} is used instead.
4152 @defmac INIT_CUMULATIVE_INCOMING_ARGS (@var{cum}, @var{fntype}, @var{libname})
4153 Like @code{INIT_CUMULATIVE_ARGS} but overrides it for the purposes of
4154 finding the arguments for the function being compiled. If this macro is
4155 undefined, @code{INIT_CUMULATIVE_ARGS} is used instead.
4157 The value passed for @var{libname} is always 0, since library routines
4158 with special calling conventions are never compiled with GCC@. The
4159 argument @var{libname} exists for symmetry with
4160 @code{INIT_CUMULATIVE_ARGS}.
4161 @c could use "this macro" in place of @code{INIT_CUMULATIVE_ARGS}, maybe.
4162 @c --mew 5feb93 i switched the order of the sentences. --mew 10feb93
4165 @deftypefn {Target Hook} void TARGET_FUNCTION_ARG_ADVANCE (cumulative_args_t @var{ca}, enum machine_mode @var{mode}, const_tree @var{type}, bool @var{named})
4166 This hook updates the summarizer variable pointed to by @var{ca} to
4167 advance past an argument in the argument list. The values @var{mode},
4168 @var{type} and @var{named} describe that argument. Once this is done,
4169 the variable @var{cum} is suitable for analyzing the @emph{following}
4170 argument with @code{TARGET_FUNCTION_ARG}, etc.
4172 This hook need not do anything if the argument in question was passed
4173 on the stack. The compiler knows how to track the amount of stack space
4174 used for arguments without any special help.
4177 @defmac FUNCTION_ARG_OFFSET (@var{mode}, @var{type})
4178 If defined, a C expression that is the number of bytes to add to the
4179 offset of the argument passed in memory. This is needed for the SPU,
4180 which passes @code{char} and @code{short} arguments in the preferred
4181 slot that is in the middle of the quad word instead of starting at the
4185 @defmac FUNCTION_ARG_PADDING (@var{mode}, @var{type})
4186 If defined, a C expression which determines whether, and in which direction,
4187 to pad out an argument with extra space. The value should be of type
4188 @code{enum direction}: either @code{upward} to pad above the argument,
4189 @code{downward} to pad below, or @code{none} to inhibit padding.
4191 The @emph{amount} of padding is not controlled by this macro, but by the
4192 target hook @code{TARGET_FUNCTION_ARG_ROUND_BOUNDARY}. It is
4193 always just enough to reach the next multiple of that boundary.
4195 This macro has a default definition which is right for most systems.
4196 For little-endian machines, the default is to pad upward. For
4197 big-endian machines, the default is to pad downward for an argument of
4198 constant size shorter than an @code{int}, and upward otherwise.
4201 @defmac PAD_VARARGS_DOWN
4202 If defined, a C expression which determines whether the default
4203 implementation of va_arg will attempt to pad down before reading the
4204 next argument, if that argument is smaller than its aligned space as
4205 controlled by @code{PARM_BOUNDARY}. If this macro is not defined, all such
4206 arguments are padded down if @code{BYTES_BIG_ENDIAN} is true.
4209 @defmac BLOCK_REG_PADDING (@var{mode}, @var{type}, @var{first})
4210 Specify padding for the last element of a block move between registers and
4211 memory. @var{first} is nonzero if this is the only element. Defining this
4212 macro allows better control of register function parameters on big-endian
4213 machines, without using @code{PARALLEL} rtl. In particular,
4214 @code{MUST_PASS_IN_STACK} need not test padding and mode of types in
4215 registers, as there is no longer a "wrong" part of a register; For example,
4216 a three byte aggregate may be passed in the high part of a register if so
4220 @deftypefn {Target Hook} {unsigned int} TARGET_FUNCTION_ARG_BOUNDARY (enum machine_mode @var{mode}, const_tree @var{type})
4221 This hook returns the alignment boundary, in bits, of an argument
4222 with the specified mode and type. The default hook returns
4223 @code{PARM_BOUNDARY} for all arguments.
4226 @deftypefn {Target Hook} {unsigned int} TARGET_FUNCTION_ARG_ROUND_BOUNDARY (enum machine_mode @var{mode}, const_tree @var{type})
4227 Normally, the size of an argument is rounded up to @code{PARM_BOUNDARY},
4228 which is the default value for this hook. You can define this hook to
4229 return a different value if an argument size must be rounded to a larger
4233 @defmac FUNCTION_ARG_REGNO_P (@var{regno})
4234 A C expression that is nonzero if @var{regno} is the number of a hard
4235 register in which function arguments are sometimes passed. This does
4236 @emph{not} include implicit arguments such as the static chain and
4237 the structure-value address. On many machines, no registers can be
4238 used for this purpose since all function arguments are pushed on the
4242 @deftypefn {Target Hook} bool TARGET_SPLIT_COMPLEX_ARG (const_tree @var{type})
4243 This hook should return true if parameter of type @var{type} are passed
4244 as two scalar parameters. By default, GCC will attempt to pack complex
4245 arguments into the target's word size. Some ABIs require complex arguments
4246 to be split and treated as their individual components. For example, on
4247 AIX64, complex floats should be passed in a pair of floating point
4248 registers, even though a complex float would fit in one 64-bit floating
4251 The default value of this hook is @code{NULL}, which is treated as always
4255 @deftypefn {Target Hook} tree TARGET_BUILD_BUILTIN_VA_LIST (void)
4256 This hook returns a type node for @code{va_list} for the target.
4257 The default version of the hook returns @code{void*}.
4260 @deftypefn {Target Hook} int TARGET_ENUM_VA_LIST_P (int @var{idx}, const char **@var{pname}, tree *@var{ptree})
4261 This target hook is used in function @code{c_common_nodes_and_builtins}
4262 to iterate through the target specific builtin types for va_list. The
4263 variable @var{idx} is used as iterator. @var{pname} has to be a pointer
4264 to a @code{const char *} and @var{ptree} a pointer to a @code{tree} typed
4266 The arguments @var{pname} and @var{ptree} are used to store the result of
4267 this macro and are set to the name of the va_list builtin type and its
4269 If the return value of this macro is zero, then there is no more element.
4270 Otherwise the @var{IDX} should be increased for the next call of this
4271 macro to iterate through all types.
4274 @deftypefn {Target Hook} tree TARGET_FN_ABI_VA_LIST (tree @var{fndecl})
4275 This hook returns the va_list type of the calling convention specified by
4277 The default version of this hook returns @code{va_list_type_node}.
4280 @deftypefn {Target Hook} tree TARGET_CANONICAL_VA_LIST_TYPE (tree @var{type})
4281 This hook returns the va_list type of the calling convention specified by the
4282 type of @var{type}. If @var{type} is not a valid va_list type, it returns
4286 @deftypefn {Target Hook} tree TARGET_GIMPLIFY_VA_ARG_EXPR (tree @var{valist}, tree @var{type}, gimple_seq *@var{pre_p}, gimple_seq *@var{post_p})
4287 This hook performs target-specific gimplification of
4288 @code{VA_ARG_EXPR}. The first two parameters correspond to the
4289 arguments to @code{va_arg}; the latter two are as in
4290 @code{gimplify.c:gimplify_expr}.
4293 @deftypefn {Target Hook} bool TARGET_VALID_POINTER_MODE (enum machine_mode @var{mode})
4294 Define this to return nonzero if the port can handle pointers
4295 with machine mode @var{mode}. The default version of this
4296 hook returns true for both @code{ptr_mode} and @code{Pmode}.
4299 @deftypefn {Target Hook} bool TARGET_REF_MAY_ALIAS_ERRNO (struct ao_ref_s *@var{ref})
4300 Define this to return nonzero if the memory reference @var{ref} may alias with the system C library errno location. The default version of this hook assumes the system C library errno location is either a declaration of type int or accessed by dereferencing a pointer to int.
4303 @deftypefn {Target Hook} bool TARGET_SCALAR_MODE_SUPPORTED_P (enum machine_mode @var{mode})
4304 Define this to return nonzero if the port is prepared to handle
4305 insns involving scalar mode @var{mode}. For a scalar mode to be
4306 considered supported, all the basic arithmetic and comparisons
4309 The default version of this hook returns true for any mode
4310 required to handle the basic C types (as defined by the port).
4311 Included here are the double-word arithmetic supported by the
4312 code in @file{optabs.c}.
4315 @deftypefn {Target Hook} bool TARGET_VECTOR_MODE_SUPPORTED_P (enum machine_mode @var{mode})
4316 Define this to return nonzero if the port is prepared to handle
4317 insns involving vector mode @var{mode}. At the very least, it
4318 must have move patterns for this mode.
4321 @deftypefn {Target Hook} bool TARGET_ARRAY_MODE_SUPPORTED_P (enum machine_mode @var{mode}, unsigned HOST_WIDE_INT @var{nelems})
4322 Return true if GCC should try to use a scalar mode to store an array
4323 of @var{nelems} elements, given that each element has mode @var{mode}.
4324 Returning true here overrides the usual @code{MAX_FIXED_MODE} limit
4325 and allows GCC to use any defined integer mode.
4327 One use of this hook is to support vector load and store operations
4328 that operate on several homogeneous vectors. For example, ARM NEON
4329 has operations like:
4332 int8x8x3_t vld3_s8 (const int8_t *)
4335 where the return type is defined as:
4338 typedef struct int8x8x3_t
4344 If this hook allows @code{val} to have a scalar mode, then
4345 @code{int8x8x3_t} can have the same mode. GCC can then store
4346 @code{int8x8x3_t}s in registers rather than forcing them onto the stack.
4349 @deftypefn {Target Hook} bool TARGET_SMALL_REGISTER_CLASSES_FOR_MODE_P (enum machine_mode @var{mode})
4350 Define this to return nonzero for machine modes for which the port has
4351 small register classes. If this target hook returns nonzero for a given
4352 @var{mode}, the compiler will try to minimize the lifetime of registers
4353 in @var{mode}. The hook may be called with @code{VOIDmode} as argument.
4354 In this case, the hook is expected to return nonzero if it returns nonzero
4357 On some machines, it is risky to let hard registers live across arbitrary
4358 insns. Typically, these machines have instructions that require values
4359 to be in specific registers (like an accumulator), and reload will fail
4360 if the required hard register is used for another purpose across such an
4363 Passes before reload do not know which hard registers will be used
4364 in an instruction, but the machine modes of the registers set or used in
4365 the instruction are already known. And for some machines, register
4366 classes are small for, say, integer registers but not for floating point
4367 registers. For example, the AMD x86-64 architecture requires specific
4368 registers for the legacy x86 integer instructions, but there are many
4369 SSE registers for floating point operations. On such targets, a good
4370 strategy may be to return nonzero from this hook for @code{INTEGRAL_MODE_P}
4371 machine modes but zero for the SSE register classes.
4373 The default version of this hook returns false for any mode. It is always
4374 safe to redefine this hook to return with a nonzero value. But if you
4375 unnecessarily define it, you will reduce the amount of optimizations
4376 that can be performed in some cases. If you do not define this hook
4377 to return a nonzero value when it is required, the compiler will run out
4378 of spill registers and print a fatal error message.
4381 @deftypevr {Target Hook} {unsigned int} TARGET_FLAGS_REGNUM
4382 If the target has a dedicated flags register, and it needs to use the post-reload comparison elimination pass, then this value should be set appropriately.
4386 @subsection How Scalar Function Values Are Returned
4387 @cindex return values in registers
4388 @cindex values, returned by functions
4389 @cindex scalars, returned as values
4391 This section discusses the macros that control returning scalars as
4392 values---values that can fit in registers.
4394 @deftypefn {Target Hook} rtx TARGET_FUNCTION_VALUE (const_tree @var{ret_type}, const_tree @var{fn_decl_or_type}, bool @var{outgoing})
4396 Define this to return an RTX representing the place where a function
4397 returns or receives a value of data type @var{ret_type}, a tree node
4398 representing a data type. @var{fn_decl_or_type} is a tree node
4399 representing @code{FUNCTION_DECL} or @code{FUNCTION_TYPE} of a
4400 function being called. If @var{outgoing} is false, the hook should
4401 compute the register in which the caller will see the return value.
4402 Otherwise, the hook should return an RTX representing the place where
4403 a function returns a value.
4405 On many machines, only @code{TYPE_MODE (@var{ret_type})} is relevant.
4406 (Actually, on most machines, scalar values are returned in the same
4407 place regardless of mode.) The value of the expression is usually a
4408 @code{reg} RTX for the hard register where the return value is stored.
4409 The value can also be a @code{parallel} RTX, if the return value is in
4410 multiple places. See @code{TARGET_FUNCTION_ARG} for an explanation of the
4411 @code{parallel} form. Note that the callee will populate every
4412 location specified in the @code{parallel}, but if the first element of
4413 the @code{parallel} contains the whole return value, callers will use
4414 that element as the canonical location and ignore the others. The m68k
4415 port uses this type of @code{parallel} to return pointers in both
4416 @samp{%a0} (the canonical location) and @samp{%d0}.
4418 If @code{TARGET_PROMOTE_FUNCTION_RETURN} returns true, you must apply
4419 the same promotion rules specified in @code{PROMOTE_MODE} if
4420 @var{valtype} is a scalar type.
4422 If the precise function being called is known, @var{func} is a tree
4423 node (@code{FUNCTION_DECL}) for it; otherwise, @var{func} is a null
4424 pointer. This makes it possible to use a different value-returning
4425 convention for specific functions when all their calls are
4428 Some target machines have ``register windows'' so that the register in
4429 which a function returns its value is not the same as the one in which
4430 the caller sees the value. For such machines, you should return
4431 different RTX depending on @var{outgoing}.
4433 @code{TARGET_FUNCTION_VALUE} is not used for return values with
4434 aggregate data types, because these are returned in another way. See
4435 @code{TARGET_STRUCT_VALUE_RTX} and related macros, below.
4438 @defmac FUNCTION_VALUE (@var{valtype}, @var{func})
4439 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE} for
4440 a new target instead.
4443 @defmac LIBCALL_VALUE (@var{mode})
4444 A C expression to create an RTX representing the place where a library
4445 function returns a value of mode @var{mode}.
4447 Note that ``library function'' in this context means a compiler
4448 support routine, used to perform arithmetic, whose name is known
4449 specially by the compiler and was not mentioned in the C code being
4453 @deftypefn {Target Hook} rtx TARGET_LIBCALL_VALUE (enum machine_mode @var{mode}, const_rtx @var{fun})
4454 Define this hook if the back-end needs to know the name of the libcall
4455 function in order to determine where the result should be returned.
4457 The mode of the result is given by @var{mode} and the name of the called
4458 library function is given by @var{fun}. The hook should return an RTX
4459 representing the place where the library function result will be returned.
4461 If this hook is not defined, then LIBCALL_VALUE will be used.
4464 @defmac FUNCTION_VALUE_REGNO_P (@var{regno})
4465 A C expression that is nonzero if @var{regno} is the number of a hard
4466 register in which the values of called function may come back.
4468 A register whose use for returning values is limited to serving as the
4469 second of a pair (for a value of type @code{double}, say) need not be
4470 recognized by this macro. So for most machines, this definition
4474 #define FUNCTION_VALUE_REGNO_P(N) ((N) == 0)
4477 If the machine has register windows, so that the caller and the called
4478 function use different registers for the return value, this macro
4479 should recognize only the caller's register numbers.
4481 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE_REGNO_P}
4482 for a new target instead.
4485 @deftypefn {Target Hook} bool TARGET_FUNCTION_VALUE_REGNO_P (const unsigned int @var{regno})
4486 A target hook that return @code{true} if @var{regno} is the number of a hard
4487 register in which the values of called function may come back.
4489 A register whose use for returning values is limited to serving as the
4490 second of a pair (for a value of type @code{double}, say) need not be
4491 recognized by this target hook.
4493 If the machine has register windows, so that the caller and the called
4494 function use different registers for the return value, this target hook
4495 should recognize only the caller's register numbers.
4497 If this hook is not defined, then FUNCTION_VALUE_REGNO_P will be used.
4500 @defmac APPLY_RESULT_SIZE
4501 Define this macro if @samp{untyped_call} and @samp{untyped_return}
4502 need more space than is implied by @code{FUNCTION_VALUE_REGNO_P} for
4503 saving and restoring an arbitrary return value.
4506 @deftypefn {Target Hook} bool TARGET_RETURN_IN_MSB (const_tree @var{type})
4507 This hook should return true if values of type @var{type} are returned
4508 at the most significant end of a register (in other words, if they are
4509 padded at the least significant end). You can assume that @var{type}
4510 is returned in a register; the caller is required to check this.
4512 Note that the register provided by @code{TARGET_FUNCTION_VALUE} must
4513 be able to hold the complete return value. For example, if a 1-, 2-
4514 or 3-byte structure is returned at the most significant end of a
4515 4-byte register, @code{TARGET_FUNCTION_VALUE} should provide an
4519 @node Aggregate Return
4520 @subsection How Large Values Are Returned
4521 @cindex aggregates as return values
4522 @cindex large return values
4523 @cindex returning aggregate values
4524 @cindex structure value address
4526 When a function value's mode is @code{BLKmode} (and in some other
4527 cases), the value is not returned according to
4528 @code{TARGET_FUNCTION_VALUE} (@pxref{Scalar Return}). Instead, the
4529 caller passes the address of a block of memory in which the value
4530 should be stored. This address is called the @dfn{structure value
4533 This section describes how to control returning structure values in
4536 @deftypefn {Target Hook} bool TARGET_RETURN_IN_MEMORY (const_tree @var{type}, const_tree @var{fntype})
4537 This target hook should return a nonzero value to say to return the
4538 function value in memory, just as large structures are always returned.
4539 Here @var{type} will be the data type of the value, and @var{fntype}
4540 will be the type of the function doing the returning, or @code{NULL} for
4543 Note that values of mode @code{BLKmode} must be explicitly handled
4544 by this function. Also, the option @option{-fpcc-struct-return}
4545 takes effect regardless of this macro. On most systems, it is
4546 possible to leave the hook undefined; this causes a default
4547 definition to be used, whose value is the constant 1 for @code{BLKmode}
4548 values, and 0 otherwise.
4550 Do not use this hook to indicate that structures and unions should always
4551 be returned in memory. You should instead use @code{DEFAULT_PCC_STRUCT_RETURN}
4555 @defmac DEFAULT_PCC_STRUCT_RETURN
4556 Define this macro to be 1 if all structure and union return values must be
4557 in memory. Since this results in slower code, this should be defined
4558 only if needed for compatibility with other compilers or with an ABI@.
4559 If you define this macro to be 0, then the conventions used for structure
4560 and union return values are decided by the @code{TARGET_RETURN_IN_MEMORY}
4563 If not defined, this defaults to the value 1.
4566 @deftypefn {Target Hook} rtx TARGET_STRUCT_VALUE_RTX (tree @var{fndecl}, int @var{incoming})
4567 This target hook should return the location of the structure value
4568 address (normally a @code{mem} or @code{reg}), or 0 if the address is
4569 passed as an ``invisible'' first argument. Note that @var{fndecl} may
4570 be @code{NULL}, for libcalls. You do not need to define this target
4571 hook if the address is always passed as an ``invisible'' first
4574 On some architectures the place where the structure value address
4575 is found by the called function is not the same place that the
4576 caller put it. This can be due to register windows, or it could
4577 be because the function prologue moves it to a different place.
4578 @var{incoming} is @code{1} or @code{2} when the location is needed in
4579 the context of the called function, and @code{0} in the context of
4582 If @var{incoming} is nonzero and the address is to be found on the
4583 stack, return a @code{mem} which refers to the frame pointer. If
4584 @var{incoming} is @code{2}, the result is being used to fetch the
4585 structure value address at the beginning of a function. If you need
4586 to emit adjusting code, you should do it at this point.
4589 @defmac PCC_STATIC_STRUCT_RETURN
4590 Define this macro if the usual system convention on the target machine
4591 for returning structures and unions is for the called function to return
4592 the address of a static variable containing the value.
4594 Do not define this if the usual system convention is for the caller to
4595 pass an address to the subroutine.
4597 This macro has effect in @option{-fpcc-struct-return} mode, but it does
4598 nothing when you use @option{-freg-struct-return} mode.
4601 @deftypefn {Target Hook} {enum machine_mode} TARGET_GET_RAW_RESULT_MODE (int @var{regno})
4602 This target hook returns the mode to be used when accessing raw return registers in @code{__builtin_return}. Define this macro if the value in @var{reg_raw_mode} is not correct.
4605 @deftypefn {Target Hook} {enum machine_mode} TARGET_GET_RAW_ARG_MODE (int @var{regno})
4606 This target hook returns the mode to be used when accessing raw argument registers in @code{__builtin_apply_args}. Define this macro if the value in @var{reg_raw_mode} is not correct.
4610 @subsection Caller-Saves Register Allocation
4612 If you enable it, GCC can save registers around function calls. This
4613 makes it possible to use call-clobbered registers to hold variables that
4614 must live across calls.
4616 @defmac CALLER_SAVE_PROFITABLE (@var{refs}, @var{calls})
4617 A C expression to determine whether it is worthwhile to consider placing
4618 a pseudo-register in a call-clobbered hard register and saving and
4619 restoring it around each function call. The expression should be 1 when
4620 this is worth doing, and 0 otherwise.
4622 If you don't define this macro, a default is used which is good on most
4623 machines: @code{4 * @var{calls} < @var{refs}}.
4626 @defmac HARD_REGNO_CALLER_SAVE_MODE (@var{regno}, @var{nregs})
4627 A C expression specifying which mode is required for saving @var{nregs}
4628 of a pseudo-register in call-clobbered hard register @var{regno}. If
4629 @var{regno} is unsuitable for caller save, @code{VOIDmode} should be
4630 returned. For most machines this macro need not be defined since GCC
4631 will select the smallest suitable mode.
4634 @node Function Entry
4635 @subsection Function Entry and Exit
4636 @cindex function entry and exit
4640 This section describes the macros that output function entry
4641 (@dfn{prologue}) and exit (@dfn{epilogue}) code.
4643 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_PROLOGUE (FILE *@var{file}, HOST_WIDE_INT @var{size})
4644 If defined, a function that outputs the assembler code for entry to a
4645 function. The prologue is responsible for setting up the stack frame,
4646 initializing the frame pointer register, saving registers that must be
4647 saved, and allocating @var{size} additional bytes of storage for the
4648 local variables. @var{size} is an integer. @var{file} is a stdio
4649 stream to which the assembler code should be output.
4651 The label for the beginning of the function need not be output by this
4652 macro. That has already been done when the macro is run.
4654 @findex regs_ever_live
4655 To determine which registers to save, the macro can refer to the array
4656 @code{regs_ever_live}: element @var{r} is nonzero if hard register
4657 @var{r} is used anywhere within the function. This implies the function
4658 prologue should save register @var{r}, provided it is not one of the
4659 call-used registers. (@code{TARGET_ASM_FUNCTION_EPILOGUE} must likewise use
4660 @code{regs_ever_live}.)
4662 On machines that have ``register windows'', the function entry code does
4663 not save on the stack the registers that are in the windows, even if
4664 they are supposed to be preserved by function calls; instead it takes
4665 appropriate steps to ``push'' the register stack, if any non-call-used
4666 registers are used in the function.
4668 @findex frame_pointer_needed
4669 On machines where functions may or may not have frame-pointers, the
4670 function entry code must vary accordingly; it must set up the frame
4671 pointer if one is wanted, and not otherwise. To determine whether a
4672 frame pointer is in wanted, the macro can refer to the variable
4673 @code{frame_pointer_needed}. The variable's value will be 1 at run
4674 time in a function that needs a frame pointer. @xref{Elimination}.
4676 The function entry code is responsible for allocating any stack space
4677 required for the function. This stack space consists of the regions
4678 listed below. In most cases, these regions are allocated in the
4679 order listed, with the last listed region closest to the top of the
4680 stack (the lowest address if @code{STACK_GROWS_DOWNWARD} is defined, and
4681 the highest address if it is not defined). You can use a different order
4682 for a machine if doing so is more convenient or required for
4683 compatibility reasons. Except in cases where required by standard
4684 or by a debugger, there is no reason why the stack layout used by GCC
4685 need agree with that used by other compilers for a machine.
4688 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_END_PROLOGUE (FILE *@var{file})
4689 If defined, a function that outputs assembler code at the end of a
4690 prologue. This should be used when the function prologue is being
4691 emitted as RTL, and you have some extra assembler that needs to be
4692 emitted. @xref{prologue instruction pattern}.
4695 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_BEGIN_EPILOGUE (FILE *@var{file})
4696 If defined, a function that outputs assembler code at the start of an
4697 epilogue. This should be used when the function epilogue is being
4698 emitted as RTL, and you have some extra assembler that needs to be
4699 emitted. @xref{epilogue instruction pattern}.
4702 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_EPILOGUE (FILE *@var{file}, HOST_WIDE_INT @var{size})
4703 If defined, a function that outputs the assembler code for exit from a
4704 function. The epilogue is responsible for restoring the saved
4705 registers and stack pointer to their values when the function was
4706 called, and returning control to the caller. This macro takes the
4707 same arguments as the macro @code{TARGET_ASM_FUNCTION_PROLOGUE}, and the
4708 registers to restore are determined from @code{regs_ever_live} and
4709 @code{CALL_USED_REGISTERS} in the same way.
4711 On some machines, there is a single instruction that does all the work
4712 of returning from the function. On these machines, give that
4713 instruction the name @samp{return} and do not define the macro
4714 @code{TARGET_ASM_FUNCTION_EPILOGUE} at all.
4716 Do not define a pattern named @samp{return} if you want the
4717 @code{TARGET_ASM_FUNCTION_EPILOGUE} to be used. If you want the target
4718 switches to control whether return instructions or epilogues are used,
4719 define a @samp{return} pattern with a validity condition that tests the
4720 target switches appropriately. If the @samp{return} pattern's validity
4721 condition is false, epilogues will be used.
4723 On machines where functions may or may not have frame-pointers, the
4724 function exit code must vary accordingly. Sometimes the code for these
4725 two cases is completely different. To determine whether a frame pointer
4726 is wanted, the macro can refer to the variable
4727 @code{frame_pointer_needed}. The variable's value will be 1 when compiling
4728 a function that needs a frame pointer.
4730 Normally, @code{TARGET_ASM_FUNCTION_PROLOGUE} and
4731 @code{TARGET_ASM_FUNCTION_EPILOGUE} must treat leaf functions specially.
4732 The C variable @code{current_function_is_leaf} is nonzero for such a
4733 function. @xref{Leaf Functions}.
4735 On some machines, some functions pop their arguments on exit while
4736 others leave that for the caller to do. For example, the 68020 when
4737 given @option{-mrtd} pops arguments in functions that take a fixed
4738 number of arguments.
4740 @findex current_function_pops_args
4741 Your definition of the macro @code{RETURN_POPS_ARGS} decides which
4742 functions pop their own arguments. @code{TARGET_ASM_FUNCTION_EPILOGUE}
4743 needs to know what was decided. The number of bytes of the current
4744 function's arguments that this function should pop is available in
4745 @code{crtl->args.pops_args}. @xref{Scalar Return}.
4750 @findex current_function_pretend_args_size
4751 A region of @code{current_function_pretend_args_size} bytes of
4752 uninitialized space just underneath the first argument arriving on the
4753 stack. (This may not be at the very start of the allocated stack region
4754 if the calling sequence has pushed anything else since pushing the stack
4755 arguments. But usually, on such machines, nothing else has been pushed
4756 yet, because the function prologue itself does all the pushing.) This
4757 region is used on machines where an argument may be passed partly in
4758 registers and partly in memory, and, in some cases to support the
4759 features in @code{<stdarg.h>}.
4762 An area of memory used to save certain registers used by the function.
4763 The size of this area, which may also include space for such things as
4764 the return address and pointers to previous stack frames, is
4765 machine-specific and usually depends on which registers have been used
4766 in the function. Machines with register windows often do not require
4770 A region of at least @var{size} bytes, possibly rounded up to an allocation
4771 boundary, to contain the local variables of the function. On some machines,
4772 this region and the save area may occur in the opposite order, with the
4773 save area closer to the top of the stack.
4776 @cindex @code{ACCUMULATE_OUTGOING_ARGS} and stack frames
4777 Optionally, when @code{ACCUMULATE_OUTGOING_ARGS} is defined, a region of
4778 @code{current_function_outgoing_args_size} bytes to be used for outgoing
4779 argument lists of the function. @xref{Stack Arguments}.
4782 @defmac EXIT_IGNORE_STACK
4783 Define this macro as a C expression that is nonzero if the return
4784 instruction or the function epilogue ignores the value of the stack
4785 pointer; in other words, if it is safe to delete an instruction to
4786 adjust the stack pointer before a return from the function. The
4789 Note that this macro's value is relevant only for functions for which
4790 frame pointers are maintained. It is never safe to delete a final
4791 stack adjustment in a function that has no frame pointer, and the
4792 compiler knows this regardless of @code{EXIT_IGNORE_STACK}.
4795 @defmac EPILOGUE_USES (@var{regno})
4796 Define this macro as a C expression that is nonzero for registers that are
4797 used by the epilogue or the @samp{return} pattern. The stack and frame
4798 pointer registers are already assumed to be used as needed.
4801 @defmac EH_USES (@var{regno})
4802 Define this macro as a C expression that is nonzero for registers that are
4803 used by the exception handling mechanism, and so should be considered live
4804 on entry to an exception edge.
4807 @defmac DELAY_SLOTS_FOR_EPILOGUE
4808 Define this macro if the function epilogue contains delay slots to which
4809 instructions from the rest of the function can be ``moved''. The
4810 definition should be a C expression whose value is an integer
4811 representing the number of delay slots there.
4814 @defmac ELIGIBLE_FOR_EPILOGUE_DELAY (@var{insn}, @var{n})
4815 A C expression that returns 1 if @var{insn} can be placed in delay
4816 slot number @var{n} of the epilogue.
4818 The argument @var{n} is an integer which identifies the delay slot now
4819 being considered (since different slots may have different rules of
4820 eligibility). It is never negative and is always less than the number
4821 of epilogue delay slots (what @code{DELAY_SLOTS_FOR_EPILOGUE} returns).
4822 If you reject a particular insn for a given delay slot, in principle, it
4823 may be reconsidered for a subsequent delay slot. Also, other insns may
4824 (at least in principle) be considered for the so far unfilled delay
4827 @findex current_function_epilogue_delay_list
4828 @findex final_scan_insn
4829 The insns accepted to fill the epilogue delay slots are put in an RTL
4830 list made with @code{insn_list} objects, stored in the variable
4831 @code{current_function_epilogue_delay_list}. The insn for the first
4832 delay slot comes first in the list. Your definition of the macro
4833 @code{TARGET_ASM_FUNCTION_EPILOGUE} should fill the delay slots by
4834 outputting the insns in this list, usually by calling
4835 @code{final_scan_insn}.
4837 You need not define this macro if you did not define
4838 @code{DELAY_SLOTS_FOR_EPILOGUE}.
4841 @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_MI_THUNK (FILE *@var{file}, tree @var{thunk_fndecl}, HOST_WIDE_INT @var{delta}, HOST_WIDE_INT @var{vcall_offset}, tree @var{function})
4842 A function that outputs the assembler code for a thunk
4843 function, used to implement C++ virtual function calls with multiple
4844 inheritance. The thunk acts as a wrapper around a virtual function,
4845 adjusting the implicit object parameter before handing control off to
4848 First, emit code to add the integer @var{delta} to the location that
4849 contains the incoming first argument. Assume that this argument
4850 contains a pointer, and is the one used to pass the @code{this} pointer
4851 in C++. This is the incoming argument @emph{before} the function prologue,
4852 e.g.@: @samp{%o0} on a sparc. The addition must preserve the values of
4853 all other incoming arguments.
4855 Then, if @var{vcall_offset} is nonzero, an additional adjustment should be
4856 made after adding @code{delta}. In particular, if @var{p} is the
4857 adjusted pointer, the following adjustment should be made:
4860 p += (*((ptrdiff_t **)p))[vcall_offset/sizeof(ptrdiff_t)]
4863 After the additions, emit code to jump to @var{function}, which is a
4864 @code{FUNCTION_DECL}. This is a direct pure jump, not a call, and does
4865 not touch the return address. Hence returning from @var{FUNCTION} will
4866 return to whoever called the current @samp{thunk}.
4868 The effect must be as if @var{function} had been called directly with
4869 the adjusted first argument. This macro is responsible for emitting all
4870 of the code for a thunk function; @code{TARGET_ASM_FUNCTION_PROLOGUE}
4871 and @code{TARGET_ASM_FUNCTION_EPILOGUE} are not invoked.
4873 The @var{thunk_fndecl} is redundant. (@var{delta} and @var{function}
4874 have already been extracted from it.) It might possibly be useful on
4875 some targets, but probably not.
4877 If you do not define this macro, the target-independent code in the C++
4878 front end will generate a less efficient heavyweight thunk that calls
4879 @var{function} instead of jumping to it. The generic approach does
4880 not support varargs.
4883 @deftypefn {Target Hook} bool TARGET_ASM_CAN_OUTPUT_MI_THUNK (const_tree @var{thunk_fndecl}, HOST_WIDE_INT @var{delta}, HOST_WIDE_INT @var{vcall_offset}, const_tree @var{function})
4884 A function that returns true if TARGET_ASM_OUTPUT_MI_THUNK would be able
4885 to output the assembler code for the thunk function specified by the
4886 arguments it is passed, and false otherwise. In the latter case, the
4887 generic approach will be used by the C++ front end, with the limitations
4892 @subsection Generating Code for Profiling
4893 @cindex profiling, code generation
4895 These macros will help you generate code for profiling.
4897 @defmac FUNCTION_PROFILER (@var{file}, @var{labelno})
4898 A C statement or compound statement to output to @var{file} some
4899 assembler code to call the profiling subroutine @code{mcount}.
4902 The details of how @code{mcount} expects to be called are determined by
4903 your operating system environment, not by GCC@. To figure them out,
4904 compile a small program for profiling using the system's installed C
4905 compiler and look at the assembler code that results.
4907 Older implementations of @code{mcount} expect the address of a counter
4908 variable to be loaded into some register. The name of this variable is
4909 @samp{LP} followed by the number @var{labelno}, so you would generate
4910 the name using @samp{LP%d} in a @code{fprintf}.
4913 @defmac PROFILE_HOOK
4914 A C statement or compound statement to output to @var{file} some assembly
4915 code to call the profiling subroutine @code{mcount} even the target does
4916 not support profiling.
4919 @defmac NO_PROFILE_COUNTERS
4920 Define this macro to be an expression with a nonzero value if the
4921 @code{mcount} subroutine on your system does not need a counter variable
4922 allocated for each function. This is true for almost all modern
4923 implementations. If you define this macro, you must not use the
4924 @var{labelno} argument to @code{FUNCTION_PROFILER}.
4927 @defmac PROFILE_BEFORE_PROLOGUE
4928 Define this macro if the code for function profiling should come before
4929 the function prologue. Normally, the profiling code comes after.
4933 @subsection Permitting tail calls
4936 @deftypefn {Target Hook} bool TARGET_FUNCTION_OK_FOR_SIBCALL (tree @var{decl}, tree @var{exp})
4937 True if it is ok to do sibling call optimization for the specified
4938 call expression @var{exp}. @var{decl} will be the called function,
4939 or @code{NULL} if this is an indirect call.
4941 It is not uncommon for limitations of calling conventions to prevent
4942 tail calls to functions outside the current unit of translation, or
4943 during PIC compilation. The hook is used to enforce these restrictions,
4944 as the @code{sibcall} md pattern can not fail, or fall over to a
4945 ``normal'' call. The criteria for successful sibling call optimization
4946 may vary greatly between different architectures.
4949 @deftypefn {Target Hook} void TARGET_EXTRA_LIVE_ON_ENTRY (bitmap @var{regs})
4950 Add any hard registers to @var{regs} that are live on entry to the
4951 function. This hook only needs to be defined to provide registers that
4952 cannot be found by examination of FUNCTION_ARG_REGNO_P, the callee saved
4953 registers, STATIC_CHAIN_INCOMING_REGNUM, STATIC_CHAIN_REGNUM,
4954 TARGET_STRUCT_VALUE_RTX, FRAME_POINTER_REGNUM, EH_USES,
4955 FRAME_POINTER_REGNUM, ARG_POINTER_REGNUM, and the PIC_OFFSET_TABLE_REGNUM.
4958 @node Stack Smashing Protection
4959 @subsection Stack smashing protection
4960 @cindex stack smashing protection
4962 @deftypefn {Target Hook} tree TARGET_STACK_PROTECT_GUARD (void)
4963 This hook returns a @code{DECL} node for the external variable to use
4964 for the stack protection guard. This variable is initialized by the
4965 runtime to some random value and is used to initialize the guard value
4966 that is placed at the top of the local stack frame. The type of this
4967 variable must be @code{ptr_type_node}.
4969 The default version of this hook creates a variable called
4970 @samp{__stack_chk_guard}, which is normally defined in @file{libgcc2.c}.
4973 @deftypefn {Target Hook} tree TARGET_STACK_PROTECT_FAIL (void)
4974 This hook returns a tree expression that alerts the runtime that the
4975 stack protect guard variable has been modified. This expression should
4976 involve a call to a @code{noreturn} function.
4978 The default version of this hook invokes a function called
4979 @samp{__stack_chk_fail}, taking no arguments. This function is
4980 normally defined in @file{libgcc2.c}.
4983 @deftypefn {Common Target Hook} bool TARGET_SUPPORTS_SPLIT_STACK (bool @var{report}, struct gcc_options *@var{opts})
4984 Whether this target supports splitting the stack when the options described in @var{opts} have been passed. This is called after options have been parsed, so the target may reject splitting the stack in some configurations. The default version of this hook returns false. If @var{report} is true, this function may issue a warning or error; if @var{report} is false, it must simply return a value
4988 @section Implementing the Varargs Macros
4989 @cindex varargs implementation
4991 GCC comes with an implementation of @code{<varargs.h>} and
4992 @code{<stdarg.h>} that work without change on machines that pass arguments
4993 on the stack. Other machines require their own implementations of
4994 varargs, and the two machine independent header files must have
4995 conditionals to include it.
4997 ISO @code{<stdarg.h>} differs from traditional @code{<varargs.h>} mainly in
4998 the calling convention for @code{va_start}. The traditional
4999 implementation takes just one argument, which is the variable in which
5000 to store the argument pointer. The ISO implementation of
5001 @code{va_start} takes an additional second argument. The user is
5002 supposed to write the last named argument of the function here.
5004 However, @code{va_start} should not use this argument. The way to find
5005 the end of the named arguments is with the built-in functions described
5008 @defmac __builtin_saveregs ()
5009 Use this built-in function to save the argument registers in memory so
5010 that the varargs mechanism can access them. Both ISO and traditional
5011 versions of @code{va_start} must use @code{__builtin_saveregs}, unless
5012 you use @code{TARGET_SETUP_INCOMING_VARARGS} (see below) instead.
5014 On some machines, @code{__builtin_saveregs} is open-coded under the
5015 control of the target hook @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. On
5016 other machines, it calls a routine written in assembler language,
5017 found in @file{libgcc2.c}.
5019 Code generated for the call to @code{__builtin_saveregs} appears at the
5020 beginning of the function, as opposed to where the call to
5021 @code{__builtin_saveregs} is written, regardless of what the code is.
5022 This is because the registers must be saved before the function starts
5023 to use them for its own purposes.
5024 @c i rewrote the first sentence above to fix an overfull hbox. --mew
5028 @defmac __builtin_next_arg (@var{lastarg})
5029 This builtin returns the address of the first anonymous stack
5030 argument, as type @code{void *}. If @code{ARGS_GROW_DOWNWARD}, it
5031 returns the address of the location above the first anonymous stack
5032 argument. Use it in @code{va_start} to initialize the pointer for
5033 fetching arguments from the stack. Also use it in @code{va_start} to
5034 verify that the second parameter @var{lastarg} is the last named argument
5035 of the current function.
5038 @defmac __builtin_classify_type (@var{object})
5039 Since each machine has its own conventions for which data types are
5040 passed in which kind of register, your implementation of @code{va_arg}
5041 has to embody these conventions. The easiest way to categorize the
5042 specified data type is to use @code{__builtin_classify_type} together
5043 with @code{sizeof} and @code{__alignof__}.
5045 @code{__builtin_classify_type} ignores the value of @var{object},
5046 considering only its data type. It returns an integer describing what
5047 kind of type that is---integer, floating, pointer, structure, and so on.
5049 The file @file{typeclass.h} defines an enumeration that you can use to
5050 interpret the values of @code{__builtin_classify_type}.
5053 These machine description macros help implement varargs:
5055 @deftypefn {Target Hook} rtx TARGET_EXPAND_BUILTIN_SAVEREGS (void)
5056 If defined, this hook produces the machine-specific code for a call to
5057 @code{__builtin_saveregs}. This code will be moved to the very
5058 beginning of the function, before any parameter access are made. The
5059 return value of this function should be an RTX that contains the value
5060 to use as the return of @code{__builtin_saveregs}.
5063 @deftypefn {Target Hook} void TARGET_SETUP_INCOMING_VARARGS (cumulative_args_t @var{args_so_far}, enum machine_mode @var{mode}, tree @var{type}, int *@var{pretend_args_size}, int @var{second_time})
5064 This target hook offers an alternative to using
5065 @code{__builtin_saveregs} and defining the hook
5066 @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. Use it to store the anonymous
5067 register arguments into the stack so that all the arguments appear to
5068 have been passed consecutively on the stack. Once this is done, you can
5069 use the standard implementation of varargs that works for machines that
5070 pass all their arguments on the stack.
5072 The argument @var{args_so_far} points to the @code{CUMULATIVE_ARGS} data
5073 structure, containing the values that are obtained after processing the
5074 named arguments. The arguments @var{mode} and @var{type} describe the
5075 last named argument---its machine mode and its data type as a tree node.
5077 The target hook should do two things: first, push onto the stack all the
5078 argument registers @emph{not} used for the named arguments, and second,
5079 store the size of the data thus pushed into the @code{int}-valued
5080 variable pointed to by @var{pretend_args_size}. The value that you
5081 store here will serve as additional offset for setting up the stack
5084 Because you must generate code to push the anonymous arguments at
5085 compile time without knowing their data types,
5086 @code{TARGET_SETUP_INCOMING_VARARGS} is only useful on machines that
5087 have just a single category of argument register and use it uniformly
5090 If the argument @var{second_time} is nonzero, it means that the
5091 arguments of the function are being analyzed for the second time. This
5092 happens for an inline function, which is not actually compiled until the
5093 end of the source file. The hook @code{TARGET_SETUP_INCOMING_VARARGS} should
5094 not generate any instructions in this case.
5097 @deftypefn {Target Hook} bool TARGET_STRICT_ARGUMENT_NAMING (cumulative_args_t @var{ca})
5098 Define this hook to return @code{true} if the location where a function
5099 argument is passed depends on whether or not it is a named argument.
5101 This hook controls how the @var{named} argument to @code{TARGET_FUNCTION_ARG}
5102 is set for varargs and stdarg functions. If this hook returns
5103 @code{true}, the @var{named} argument is always true for named
5104 arguments, and false for unnamed arguments. If it returns @code{false},
5105 but @code{TARGET_PRETEND_OUTGOING_VARARGS_NAMED} returns @code{true},
5106 then all arguments are treated as named. Otherwise, all named arguments
5107 except the last are treated as named.
5109 You need not define this hook if it always returns @code{false}.
5112 @deftypefn {Target Hook} bool TARGET_PRETEND_OUTGOING_VARARGS_NAMED (cumulative_args_t @var{ca})
5113 If you need to conditionally change ABIs so that one works with
5114 @code{TARGET_SETUP_INCOMING_VARARGS}, but the other works like neither
5115 @code{TARGET_SETUP_INCOMING_VARARGS} nor @code{TARGET_STRICT_ARGUMENT_NAMING} was
5116 defined, then define this hook to return @code{true} if
5117 @code{TARGET_SETUP_INCOMING_VARARGS} is used, @code{false} otherwise.
5118 Otherwise, you should not define this hook.
5122 @section Trampolines for Nested Functions
5123 @cindex trampolines for nested functions
5124 @cindex nested functions, trampolines for
5126 A @dfn{trampoline} is a small piece of code that is created at run time
5127 when the address of a nested function is taken. It normally resides on
5128 the stack, in the stack frame of the containing function. These macros
5129 tell GCC how to generate code to allocate and initialize a
5132 The instructions in the trampoline must do two things: load a constant
5133 address into the static chain register, and jump to the real address of
5134 the nested function. On CISC machines such as the m68k, this requires
5135 two instructions, a move immediate and a jump. Then the two addresses
5136 exist in the trampoline as word-long immediate operands. On RISC
5137 machines, it is often necessary to load each address into a register in
5138 two parts. Then pieces of each address form separate immediate
5141 The code generated to initialize the trampoline must store the variable
5142 parts---the static chain value and the function address---into the
5143 immediate operands of the instructions. On a CISC machine, this is
5144 simply a matter of copying each address to a memory reference at the
5145 proper offset from the start of the trampoline. On a RISC machine, it
5146 may be necessary to take out pieces of the address and store them
5149 @deftypefn {Target Hook} void TARGET_ASM_TRAMPOLINE_TEMPLATE (FILE *@var{f})
5150 This hook is called by @code{assemble_trampoline_template} to output,
5151 on the stream @var{f}, assembler code for a block of data that contains
5152 the constant parts of a trampoline. This code should not include a
5153 label---the label is taken care of automatically.
5155 If you do not define this hook, it means no template is needed
5156 for the target. Do not define this hook on systems where the block move
5157 code to copy the trampoline into place would be larger than the code
5158 to generate it on the spot.
5161 @defmac TRAMPOLINE_SECTION
5162 Return the section into which the trampoline template is to be placed
5163 (@pxref{Sections}). The default value is @code{readonly_data_section}.
5166 @defmac TRAMPOLINE_SIZE
5167 A C expression for the size in bytes of the trampoline, as an integer.
5170 @defmac TRAMPOLINE_ALIGNMENT
5171 Alignment required for trampolines, in bits.
5173 If you don't define this macro, the value of @code{FUNCTION_ALIGNMENT}
5174 is used for aligning trampolines.
5177 @deftypefn {Target Hook} void TARGET_TRAMPOLINE_INIT (rtx @var{m_tramp}, tree @var{fndecl}, rtx @var{static_chain})
5178 This hook is called to initialize a trampoline.
5179 @var{m_tramp} is an RTX for the memory block for the trampoline; @var{fndecl}
5180 is the @code{FUNCTION_DECL} for the nested function; @var{static_chain} is an
5181 RTX for the static chain value that should be passed to the function
5184 If the target defines @code{TARGET_ASM_TRAMPOLINE_TEMPLATE}, then the
5185 first thing this hook should do is emit a block move into @var{m_tramp}
5186 from the memory block returned by @code{assemble_trampoline_template}.
5187 Note that the block move need only cover the constant parts of the
5188 trampoline. If the target isolates the variable parts of the trampoline
5189 to the end, not all @code{TRAMPOLINE_SIZE} bytes need be copied.
5191 If the target requires any other actions, such as flushing caches or
5192 enabling stack execution, these actions should be performed after
5193 initializing the trampoline proper.
5196 @deftypefn {Target Hook} rtx TARGET_TRAMPOLINE_ADJUST_ADDRESS (rtx @var{addr})
5197 This hook should perform any machine-specific adjustment in
5198 the address of the trampoline. Its argument contains the address of the
5199 memory block that was passed to @code{TARGET_TRAMPOLINE_INIT}. In case
5200 the address to be used for a function call should be different from the
5201 address at which the template was stored, the different address should
5202 be returned; otherwise @var{addr} should be returned unchanged.
5203 If this hook is not defined, @var{addr} will be used for function calls.
5206 Implementing trampolines is difficult on many machines because they have
5207 separate instruction and data caches. Writing into a stack location
5208 fails to clear the memory in the instruction cache, so when the program
5209 jumps to that location, it executes the old contents.
5211 Here are two possible solutions. One is to clear the relevant parts of
5212 the instruction cache whenever a trampoline is set up. The other is to
5213 make all trampolines identical, by having them jump to a standard
5214 subroutine. The former technique makes trampoline execution faster; the
5215 latter makes initialization faster.
5217 To clear the instruction cache when a trampoline is initialized, define
5218 the following macro.
5220 @defmac CLEAR_INSN_CACHE (@var{beg}, @var{end})
5221 If defined, expands to a C expression clearing the @emph{instruction
5222 cache} in the specified interval. The definition of this macro would
5223 typically be a series of @code{asm} statements. Both @var{beg} and
5224 @var{end} are both pointer expressions.
5227 To use a standard subroutine, define the following macro. In addition,
5228 you must make sure that the instructions in a trampoline fill an entire
5229 cache line with identical instructions, or else ensure that the
5230 beginning of the trampoline code is always aligned at the same point in
5231 its cache line. Look in @file{m68k.h} as a guide.
5233 @defmac TRANSFER_FROM_TRAMPOLINE
5234 Define this macro if trampolines need a special subroutine to do their
5235 work. The macro should expand to a series of @code{asm} statements
5236 which will be compiled with GCC@. They go in a library function named
5237 @code{__transfer_from_trampoline}.
5239 If you need to avoid executing the ordinary prologue code of a compiled
5240 C function when you jump to the subroutine, you can do so by placing a
5241 special label of your own in the assembler code. Use one @code{asm}
5242 statement to generate an assembler label, and another to make the label
5243 global. Then trampolines can use that label to jump directly to your
5244 special assembler code.
5248 @section Implicit Calls to Library Routines
5249 @cindex library subroutine names
5250 @cindex @file{libgcc.a}
5252 @c prevent bad page break with this line
5253 Here is an explanation of implicit calls to library routines.
5255 @defmac DECLARE_LIBRARY_RENAMES
5256 This macro, if defined, should expand to a piece of C code that will get
5257 expanded when compiling functions for libgcc.a. It can be used to
5258 provide alternate names for GCC's internal library functions if there
5259 are ABI-mandated names that the compiler should provide.
5262 @findex set_optab_libfunc
5263 @findex init_one_libfunc
5264 @deftypefn {Target Hook} void TARGET_INIT_LIBFUNCS (void)
5265 This hook should declare additional library routines or rename
5266 existing ones, using the functions @code{set_optab_libfunc} and
5267 @code{init_one_libfunc} defined in @file{optabs.c}.
5268 @code{init_optabs} calls this macro after initializing all the normal
5271 The default is to do nothing. Most ports don't need to define this hook.
5274 @deftypevr {Target Hook} bool TARGET_LIBFUNC_GNU_PREFIX
5275 If false (the default), internal library routines start with two
5276 underscores. If set to true, these routines start with @code{__gnu_}
5277 instead. E.g., @code{__muldi3} changes to @code{__gnu_muldi3}. This
5278 currently only affects functions defined in @file{libgcc2.c}. If this
5279 is set to true, the @file{tm.h} file must also
5280 @code{#define LIBGCC2_GNU_PREFIX}.
5283 @defmac FLOAT_LIB_COMPARE_RETURNS_BOOL (@var{mode}, @var{comparison})
5284 This macro should return @code{true} if the library routine that
5285 implements the floating point comparison operator @var{comparison} in
5286 mode @var{mode} will return a boolean, and @var{false} if it will
5289 GCC's own floating point libraries return tristates from the
5290 comparison operators, so the default returns false always. Most ports
5291 don't need to define this macro.
5294 @defmac TARGET_LIB_INT_CMP_BIASED
5295 This macro should evaluate to @code{true} if the integer comparison
5296 functions (like @code{__cmpdi2}) return 0 to indicate that the first
5297 operand is smaller than the second, 1 to indicate that they are equal,
5298 and 2 to indicate that the first operand is greater than the second.
5299 If this macro evaluates to @code{false} the comparison functions return
5300 @minus{}1, 0, and 1 instead of 0, 1, and 2. If the target uses the routines
5301 in @file{libgcc.a}, you do not need to define this macro.
5304 @cindex @code{EDOM}, implicit usage
5307 The value of @code{EDOM} on the target machine, as a C integer constant
5308 expression. If you don't define this macro, GCC does not attempt to
5309 deposit the value of @code{EDOM} into @code{errno} directly. Look in
5310 @file{/usr/include/errno.h} to find the value of @code{EDOM} on your
5313 If you do not define @code{TARGET_EDOM}, then compiled code reports
5314 domain errors by calling the library function and letting it report the
5315 error. If mathematical functions on your system use @code{matherr} when
5316 there is an error, then you should leave @code{TARGET_EDOM} undefined so
5317 that @code{matherr} is used normally.
5320 @cindex @code{errno}, implicit usage
5321 @defmac GEN_ERRNO_RTX
5322 Define this macro as a C expression to create an rtl expression that
5323 refers to the global ``variable'' @code{errno}. (On certain systems,
5324 @code{errno} may not actually be a variable.) If you don't define this
5325 macro, a reasonable default is used.
5328 @cindex C99 math functions, implicit usage
5329 @defmac TARGET_C99_FUNCTIONS
5330 When this macro is nonzero, GCC will implicitly optimize @code{sin} calls into
5331 @code{sinf} and similarly for other functions defined by C99 standard. The
5332 default is zero because a number of existing systems lack support for these
5333 functions in their runtime so this macro needs to be redefined to one on
5334 systems that do support the C99 runtime.
5337 @cindex sincos math function, implicit usage
5338 @defmac TARGET_HAS_SINCOS
5339 When this macro is nonzero, GCC will implicitly optimize calls to @code{sin}
5340 and @code{cos} with the same argument to a call to @code{sincos}. The
5341 default is zero. The target has to provide the following functions:
5343 void sincos(double x, double *sin, double *cos);
5344 void sincosf(float x, float *sin, float *cos);
5345 void sincosl(long double x, long double *sin, long double *cos);
5349 @defmac NEXT_OBJC_RUNTIME
5350 Define this macro to generate code for Objective-C message sending using
5351 the calling convention of the NeXT system. This calling convention
5352 involves passing the object, the selector and the method arguments all
5353 at once to the method-lookup library function.
5355 The default calling convention passes just the object and the selector
5356 to the lookup function, which returns a pointer to the method.
5359 @node Addressing Modes
5360 @section Addressing Modes
5361 @cindex addressing modes
5363 @c prevent bad page break with this line
5364 This is about addressing modes.
5366 @defmac HAVE_PRE_INCREMENT
5367 @defmacx HAVE_PRE_DECREMENT
5368 @defmacx HAVE_POST_INCREMENT
5369 @defmacx HAVE_POST_DECREMENT
5370 A C expression that is nonzero if the machine supports pre-increment,
5371 pre-decrement, post-increment, or post-decrement addressing respectively.
5374 @defmac HAVE_PRE_MODIFY_DISP
5375 @defmacx HAVE_POST_MODIFY_DISP
5376 A C expression that is nonzero if the machine supports pre- or
5377 post-address side-effect generation involving constants other than
5378 the size of the memory operand.
5381 @defmac HAVE_PRE_MODIFY_REG
5382 @defmacx HAVE_POST_MODIFY_REG
5383 A C expression that is nonzero if the machine supports pre- or
5384 post-address side-effect generation involving a register displacement.
5387 @defmac CONSTANT_ADDRESS_P (@var{x})
5388 A C expression that is 1 if the RTX @var{x} is a constant which
5389 is a valid address. On most machines the default definition of
5390 @code{(CONSTANT_P (@var{x}) && GET_CODE (@var{x}) != CONST_DOUBLE)}
5391 is acceptable, but a few machines are more restrictive as to which
5392 constant addresses are supported.
5395 @defmac CONSTANT_P (@var{x})
5396 @code{CONSTANT_P}, which is defined by target-independent code,
5397 accepts integer-values expressions whose values are not explicitly
5398 known, such as @code{symbol_ref}, @code{label_ref}, and @code{high}
5399 expressions and @code{const} arithmetic expressions, in addition to
5400 @code{const_int} and @code{const_double} expressions.
5403 @defmac MAX_REGS_PER_ADDRESS
5404 A number, the maximum number of registers that can appear in a valid
5405 memory address. Note that it is up to you to specify a value equal to
5406 the maximum number that @code{TARGET_LEGITIMATE_ADDRESS_P} would ever
5410 @deftypefn {Target Hook} bool TARGET_LEGITIMATE_ADDRESS_P (enum machine_mode @var{mode}, rtx @var{x}, bool @var{strict})
5411 A function that returns whether @var{x} (an RTX) is a legitimate memory
5412 address on the target machine for a memory operand of mode @var{mode}.
5414 Legitimate addresses are defined in two variants: a strict variant and a
5415 non-strict one. The @var{strict} parameter chooses which variant is
5416 desired by the caller.
5418 The strict variant is used in the reload pass. It must be defined so
5419 that any pseudo-register that has not been allocated a hard register is
5420 considered a memory reference. This is because in contexts where some
5421 kind of register is required, a pseudo-register with no hard register
5422 must be rejected. For non-hard registers, the strict variant should look
5423 up the @code{reg_renumber} array; it should then proceed using the hard
5424 register number in the array, or treat the pseudo as a memory reference
5425 if the array holds @code{-1}.
5427 The non-strict variant is used in other passes. It must be defined to
5428 accept all pseudo-registers in every context where some kind of
5429 register is required.
5431 Normally, constant addresses which are the sum of a @code{symbol_ref}
5432 and an integer are stored inside a @code{const} RTX to mark them as
5433 constant. Therefore, there is no need to recognize such sums
5434 specifically as legitimate addresses. Normally you would simply
5435 recognize any @code{const} as legitimate.
5437 Usually @code{PRINT_OPERAND_ADDRESS} is not prepared to handle constant
5438 sums that are not marked with @code{const}. It assumes that a naked
5439 @code{plus} indicates indexing. If so, then you @emph{must} reject such
5440 naked constant sums as illegitimate addresses, so that none of them will
5441 be given to @code{PRINT_OPERAND_ADDRESS}.
5443 @cindex @code{TARGET_ENCODE_SECTION_INFO} and address validation
5444 On some machines, whether a symbolic address is legitimate depends on
5445 the section that the address refers to. On these machines, define the
5446 target hook @code{TARGET_ENCODE_SECTION_INFO} to store the information
5447 into the @code{symbol_ref}, and then check for it here. When you see a
5448 @code{const}, you will have to look inside it to find the
5449 @code{symbol_ref} in order to determine the section. @xref{Assembler
5452 @cindex @code{GO_IF_LEGITIMATE_ADDRESS}
5453 Some ports are still using a deprecated legacy substitute for
5454 this hook, the @code{GO_IF_LEGITIMATE_ADDRESS} macro. This macro
5458 #define GO_IF_LEGITIMATE_ADDRESS (@var{mode}, @var{x}, @var{label})
5462 and should @code{goto @var{label}} if the address @var{x} is a valid
5463 address on the target machine for a memory operand of mode @var{mode}.
5465 @findex REG_OK_STRICT
5466 Compiler source files that want to use the strict variant of this
5467 macro define the macro @code{REG_OK_STRICT}. You should use an
5468 @code{#ifdef REG_OK_STRICT} conditional to define the strict variant in
5469 that case and the non-strict variant otherwise.
5471 Using the hook is usually simpler because it limits the number of
5472 files that are recompiled when changes are made.
5475 @defmac TARGET_MEM_CONSTRAINT
5476 A single character to be used instead of the default @code{'m'}
5477 character for general memory addresses. This defines the constraint
5478 letter which matches the memory addresses accepted by
5479 @code{TARGET_LEGITIMATE_ADDRESS_P}. Define this macro if you want to
5480 support new address formats in your back end without changing the
5481 semantics of the @code{'m'} constraint. This is necessary in order to
5482 preserve functionality of inline assembly constructs using the
5483 @code{'m'} constraint.
5486 @defmac FIND_BASE_TERM (@var{x})
5487 A C expression to determine the base term of address @var{x},
5488 or to provide a simplified version of @var{x} from which @file{alias.c}
5489 can easily find the base term. This macro is used in only two places:
5490 @code{find_base_value} and @code{find_base_term} in @file{alias.c}.
5492 It is always safe for this macro to not be defined. It exists so
5493 that alias analysis can understand machine-dependent addresses.
5495 The typical use of this macro is to handle addresses containing
5496 a label_ref or symbol_ref within an UNSPEC@.
5499 @deftypefn {Target Hook} rtx TARGET_LEGITIMIZE_ADDRESS (rtx @var{x}, rtx @var{oldx}, enum machine_mode @var{mode})
5500 This hook is given an invalid memory address @var{x} for an
5501 operand of mode @var{mode} and should try to return a valid memory
5504 @findex break_out_memory_refs
5505 @var{x} will always be the result of a call to @code{break_out_memory_refs},
5506 and @var{oldx} will be the operand that was given to that function to produce
5509 The code of the hook should not alter the substructure of
5510 @var{x}. If it transforms @var{x} into a more legitimate form, it
5511 should return the new @var{x}.
5513 It is not necessary for this hook to come up with a legitimate address.
5514 The compiler has standard ways of doing so in all cases. In fact, it
5515 is safe to omit this hook or make it return @var{x} if it cannot find
5516 a valid way to legitimize the address. But often a machine-dependent
5517 strategy can generate better code.
5520 @defmac LEGITIMIZE_RELOAD_ADDRESS (@var{x}, @var{mode}, @var{opnum}, @var{type}, @var{ind_levels}, @var{win})
5521 A C compound statement that attempts to replace @var{x}, which is an address
5522 that needs reloading, with a valid memory address for an operand of mode
5523 @var{mode}. @var{win} will be a C statement label elsewhere in the code.
5524 It is not necessary to define this macro, but it might be useful for
5525 performance reasons.
5527 For example, on the i386, it is sometimes possible to use a single
5528 reload register instead of two by reloading a sum of two pseudo
5529 registers into a register. On the other hand, for number of RISC
5530 processors offsets are limited so that often an intermediate address
5531 needs to be generated in order to address a stack slot. By defining
5532 @code{LEGITIMIZE_RELOAD_ADDRESS} appropriately, the intermediate addresses
5533 generated for adjacent some stack slots can be made identical, and thus
5536 @emph{Note}: This macro should be used with caution. It is necessary
5537 to know something of how reload works in order to effectively use this,
5538 and it is quite easy to produce macros that build in too much knowledge
5539 of reload internals.
5541 @emph{Note}: This macro must be able to reload an address created by a
5542 previous invocation of this macro. If it fails to handle such addresses
5543 then the compiler may generate incorrect code or abort.
5546 The macro definition should use @code{push_reload} to indicate parts that
5547 need reloading; @var{opnum}, @var{type} and @var{ind_levels} are usually
5548 suitable to be passed unaltered to @code{push_reload}.
5550 The code generated by this macro must not alter the substructure of
5551 @var{x}. If it transforms @var{x} into a more legitimate form, it
5552 should assign @var{x} (which will always be a C variable) a new value.
5553 This also applies to parts that you change indirectly by calling
5556 @findex strict_memory_address_p
5557 The macro definition may use @code{strict_memory_address_p} to test if
5558 the address has become legitimate.
5561 If you want to change only a part of @var{x}, one standard way of doing
5562 this is to use @code{copy_rtx}. Note, however, that it unshares only a
5563 single level of rtl. Thus, if the part to be changed is not at the
5564 top level, you'll need to replace first the top level.
5565 It is not necessary for this macro to come up with a legitimate
5566 address; but often a machine-dependent strategy can generate better code.
5569 @deftypefn {Target Hook} bool TARGET_MODE_DEPENDENT_ADDRESS_P (const_rtx @var{addr})
5570 This hook returns @code{true} if memory address @var{addr} can have
5571 different meanings depending on the machine mode of the memory
5572 reference it is used for or if the address is valid for some modes
5575 Autoincrement and autodecrement addresses typically have mode-dependent
5576 effects because the amount of the increment or decrement is the size
5577 of the operand being addressed. Some machines have other mode-dependent
5578 addresses. Many RISC machines have no mode-dependent addresses.
5580 You may assume that @var{addr} is a valid address for the machine.
5582 The default version of this hook returns @code{false}.
5585 @defmac GO_IF_MODE_DEPENDENT_ADDRESS (@var{addr}, @var{label})
5586 A C statement or compound statement with a conditional @code{goto
5587 @var{label};} executed if memory address @var{x} (an RTX) can have
5588 different meanings depending on the machine mode of the memory
5589 reference it is used for or if the address is valid for some modes
5592 Autoincrement and autodecrement addresses typically have mode-dependent
5593 effects because the amount of the increment or decrement is the size
5594 of the operand being addressed. Some machines have other mode-dependent
5595 addresses. Many RISC machines have no mode-dependent addresses.
5597 You may assume that @var{addr} is a valid address for the machine.
5599 These are obsolete macros, replaced by the
5600 @code{TARGET_MODE_DEPENDENT_ADDRESS_P} target hook.
5603 @deftypefn {Target Hook} bool TARGET_LEGITIMATE_CONSTANT_P (enum machine_mode @var{mode}, rtx @var{x})
5604 This hook returns true if @var{x} is a legitimate constant for a
5605 @var{mode}-mode immediate operand on the target machine. You can assume that
5606 @var{x} satisfies @code{CONSTANT_P}, so you need not check this.
5608 The default definition returns true.
5611 @deftypefn {Target Hook} rtx TARGET_DELEGITIMIZE_ADDRESS (rtx @var{x})
5612 This hook is used to undo the possibly obfuscating effects of the
5613 @code{LEGITIMIZE_ADDRESS} and @code{LEGITIMIZE_RELOAD_ADDRESS} target
5614 macros. Some backend implementations of these macros wrap symbol
5615 references inside an @code{UNSPEC} rtx to represent PIC or similar
5616 addressing modes. This target hook allows GCC's optimizers to understand
5617 the semantics of these opaque @code{UNSPEC}s by converting them back
5618 into their original form.
5621 @deftypefn {Target Hook} bool TARGET_CANNOT_FORCE_CONST_MEM (enum machine_mode @var{mode}, rtx @var{x})
5622 This hook should return true if @var{x} is of a form that cannot (or
5623 should not) be spilled to the constant pool. @var{mode} is the mode
5626 The default version of this hook returns false.
5628 The primary reason to define this hook is to prevent reload from
5629 deciding that a non-legitimate constant would be better reloaded
5630 from the constant pool instead of spilling and reloading a register
5631 holding the constant. This restriction is often true of addresses
5632 of TLS symbols for various targets.
5635 @deftypefn {Target Hook} bool TARGET_USE_BLOCKS_FOR_CONSTANT_P (enum machine_mode @var{mode}, const_rtx @var{x})
5636 This hook should return true if pool entries for constant @var{x} can
5637 be placed in an @code{object_block} structure. @var{mode} is the mode
5640 The default version returns false for all constants.
5643 @deftypefn {Target Hook} tree TARGET_BUILTIN_RECIPROCAL (unsigned @var{fn}, bool @var{md_fn}, bool @var{sqrt})
5644 This hook should return the DECL of a function that implements reciprocal of
5645 the builtin function with builtin function code @var{fn}, or
5646 @code{NULL_TREE} if such a function is not available. @var{md_fn} is true
5647 when @var{fn} is a code of a machine-dependent builtin function. When
5648 @var{sqrt} is true, additional optimizations that apply only to the reciprocal
5649 of a square root function are performed, and only reciprocals of @code{sqrt}
5653 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_MASK_FOR_LOAD (void)
5654 This hook should return the DECL of a function @var{f} that given an
5655 address @var{addr} as an argument returns a mask @var{m} that can be
5656 used to extract from two vectors the relevant data that resides in
5657 @var{addr} in case @var{addr} is not properly aligned.
5659 The autovectorizer, when vectorizing a load operation from an address
5660 @var{addr} that may be unaligned, will generate two vector loads from
5661 the two aligned addresses around @var{addr}. It then generates a
5662 @code{REALIGN_LOAD} operation to extract the relevant data from the
5663 two loaded vectors. The first two arguments to @code{REALIGN_LOAD},
5664 @var{v1} and @var{v2}, are the two vectors, each of size @var{VS}, and
5665 the third argument, @var{OFF}, defines how the data will be extracted
5666 from these two vectors: if @var{OFF} is 0, then the returned vector is
5667 @var{v2}; otherwise, the returned vector is composed from the last
5668 @var{VS}-@var{OFF} elements of @var{v1} concatenated to the first
5669 @var{OFF} elements of @var{v2}.
5671 If this hook is defined, the autovectorizer will generate a call
5672 to @var{f} (using the DECL tree that this hook returns) and will
5673 use the return value of @var{f} as the argument @var{OFF} to
5674 @code{REALIGN_LOAD}. Therefore, the mask @var{m} returned by @var{f}
5675 should comply with the semantics expected by @code{REALIGN_LOAD}
5677 If this hook is not defined, then @var{addr} will be used as
5678 the argument @var{OFF} to @code{REALIGN_LOAD}, in which case the low
5679 log2(@var{VS}) @minus{} 1 bits of @var{addr} will be considered.
5682 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_EVEN (tree @var{x})
5683 This hook should return the DECL of a function @var{f} that implements
5684 widening multiplication of the even elements of two input vectors of type @var{x}.
5686 If this hook is defined, the autovectorizer will use it along with the
5687 @code{TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_ODD} target hook when vectorizing
5688 widening multiplication in cases that the order of the results does not have to be
5689 preserved (e.g.@: used only by a reduction computation). Otherwise, the
5690 @code{widen_mult_hi/lo} idioms will be used.
5693 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_ODD (tree @var{x})
5694 This hook should return the DECL of a function @var{f} that implements
5695 widening multiplication of the odd elements of two input vectors of type @var{x}.
5697 If this hook is defined, the autovectorizer will use it along with the
5698 @code{TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_EVEN} target hook when vectorizing
5699 widening multiplication in cases that the order of the results does not have to be
5700 preserved (e.g.@: used only by a reduction computation). Otherwise, the
5701 @code{widen_mult_hi/lo} idioms will be used.
5704 @deftypefn {Target Hook} int TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST (enum vect_cost_for_stmt @var{type_of_cost}, tree @var{vectype}, int @var{misalign})
5705 Returns cost of different scalar or vector statements for vectorization cost model.
5706 For vector memory operations the cost may depend on type (@var{vectype}) and
5707 misalignment value (@var{misalign}).
5710 @deftypefn {Target Hook} bool TARGET_VECTORIZE_VECTOR_ALIGNMENT_REACHABLE (const_tree @var{type}, bool @var{is_packed})
5711 Return true if vector alignment is reachable (by peeling N iterations) for the given type.
5714 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_VEC_PERM (tree @var{type}, tree *@var{mask_element_type})
5715 Target builtin that implements vector permute.
5718 @deftypefn {Target Hook} bool TARGET_VECTORIZE_BUILTIN_VEC_PERM_OK (tree @var{vec_type}, tree @var{mask})
5719 Return true if a vector created for @code{builtin_vec_perm} is valid.
5722 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_CONVERSION (unsigned @var{code}, tree @var{dest_type}, tree @var{src_type})
5723 This hook should return the DECL of a function that implements conversion of the
5724 input vector of type @var{src_type} to type @var{dest_type}.
5725 The value of @var{code} is one of the enumerators in @code{enum tree_code} and
5726 specifies how the conversion is to be applied
5727 (truncation, rounding, etc.).
5729 If this hook is defined, the autovectorizer will use the
5730 @code{TARGET_VECTORIZE_BUILTIN_CONVERSION} target hook when vectorizing
5731 conversion. Otherwise, it will return @code{NULL_TREE}.
5734 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION (tree @var{fndecl}, tree @var{vec_type_out}, tree @var{vec_type_in})
5735 This hook should return the decl of a function that implements the
5736 vectorized variant of the builtin function with builtin function code
5737 @var{code} or @code{NULL_TREE} if such a function is not available.
5738 The value of @var{fndecl} is the builtin function declaration. The
5739 return type of the vectorized function shall be of vector type
5740 @var{vec_type_out} and the argument types should be @var{vec_type_in}.
5743 @deftypefn {Target Hook} bool TARGET_VECTORIZE_SUPPORT_VECTOR_MISALIGNMENT (enum machine_mode @var{mode}, const_tree @var{type}, int @var{misalignment}, bool @var{is_packed})
5744 This hook should return true if the target supports misaligned vector
5745 store/load of a specific factor denoted in the @var{misalignment}
5746 parameter. The vector store/load should be of machine mode @var{mode} and
5747 the elements in the vectors should be of type @var{type}. @var{is_packed}
5748 parameter is true if the memory access is defined in a packed struct.
5751 @deftypefn {Target Hook} {enum machine_mode} TARGET_VECTORIZE_PREFERRED_SIMD_MODE (enum machine_mode @var{mode})
5752 This hook should return the preferred mode for vectorizing scalar
5753 mode @var{mode}. The default is
5754 equal to @code{word_mode}, because the vectorizer can do some
5755 transformations even in absence of specialized @acronym{SIMD} hardware.
5758 @deftypefn {Target Hook} {unsigned int} TARGET_VECTORIZE_AUTOVECTORIZE_VECTOR_SIZES (void)
5759 This hook should return a mask of sizes that should be iterated over
5760 after trying to autovectorize using the vector size derived from the
5761 mode returned by @code{TARGET_VECTORIZE_PREFERRED_SIMD_MODE}.
5762 The default is zero which means to not iterate over other vector sizes.
5765 @node Anchored Addresses
5766 @section Anchored Addresses
5767 @cindex anchored addresses
5768 @cindex @option{-fsection-anchors}
5770 GCC usually addresses every static object as a separate entity.
5771 For example, if we have:
5775 int foo (void) @{ return a + b + c; @}
5778 the code for @code{foo} will usually calculate three separate symbolic
5779 addresses: those of @code{a}, @code{b} and @code{c}. On some targets,
5780 it would be better to calculate just one symbolic address and access
5781 the three variables relative to it. The equivalent pseudocode would
5787 register int *xr = &x;
5788 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
5792 (which isn't valid C). We refer to shared addresses like @code{x} as
5793 ``section anchors''. Their use is controlled by @option{-fsection-anchors}.
5795 The hooks below describe the target properties that GCC needs to know
5796 in order to make effective use of section anchors. It won't use
5797 section anchors at all unless either @code{TARGET_MIN_ANCHOR_OFFSET}
5798 or @code{TARGET_MAX_ANCHOR_OFFSET} is set to a nonzero value.
5800 @deftypevr {Target Hook} HOST_WIDE_INT TARGET_MIN_ANCHOR_OFFSET
5801 The minimum offset that should be applied to a section anchor.
5802 On most targets, it should be the smallest offset that can be
5803 applied to a base register while still giving a legitimate address
5804 for every mode. The default value is 0.
5807 @deftypevr {Target Hook} HOST_WIDE_INT TARGET_MAX_ANCHOR_OFFSET
5808 Like @code{TARGET_MIN_ANCHOR_OFFSET}, but the maximum (inclusive)
5809 offset that should be applied to section anchors. The default
5813 @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_ANCHOR (rtx @var{x})
5814 Write the assembly code to define section anchor @var{x}, which is a
5815 @code{SYMBOL_REF} for which @samp{SYMBOL_REF_ANCHOR_P (@var{x})} is true.
5816 The hook is called with the assembly output position set to the beginning
5817 of @code{SYMBOL_REF_BLOCK (@var{x})}.
5819 If @code{ASM_OUTPUT_DEF} is available, the hook's default definition uses
5820 it to define the symbol as @samp{. + SYMBOL_REF_BLOCK_OFFSET (@var{x})}.
5821 If @code{ASM_OUTPUT_DEF} is not available, the hook's default definition
5822 is @code{NULL}, which disables the use of section anchors altogether.
5825 @deftypefn {Target Hook} bool TARGET_USE_ANCHORS_FOR_SYMBOL_P (const_rtx @var{x})
5826 Return true if GCC should attempt to use anchors to access @code{SYMBOL_REF}
5827 @var{x}. You can assume @samp{SYMBOL_REF_HAS_BLOCK_INFO_P (@var{x})} and
5828 @samp{!SYMBOL_REF_ANCHOR_P (@var{x})}.
5830 The default version is correct for most targets, but you might need to
5831 intercept this hook to handle things like target-specific attributes
5832 or target-specific sections.
5835 @node Condition Code
5836 @section Condition Code Status
5837 @cindex condition code status
5839 The macros in this section can be split in two families, according to the
5840 two ways of representing condition codes in GCC.
5842 The first representation is the so called @code{(cc0)} representation
5843 (@pxref{Jump Patterns}), where all instructions can have an implicit
5844 clobber of the condition codes. The second is the condition code
5845 register representation, which provides better schedulability for
5846 architectures that do have a condition code register, but on which
5847 most instructions do not affect it. The latter category includes
5850 The implicit clobbering poses a strong restriction on the placement of
5851 the definition and use of the condition code, which need to be in adjacent
5852 insns for machines using @code{(cc0)}. This can prevent important
5853 optimizations on some machines. For example, on the IBM RS/6000, there
5854 is a delay for taken branches unless the condition code register is set
5855 three instructions earlier than the conditional branch. The instruction
5856 scheduler cannot perform this optimization if it is not permitted to
5857 separate the definition and use of the condition code register.
5859 For this reason, it is possible and suggested to use a register to
5860 represent the condition code for new ports. If there is a specific
5861 condition code register in the machine, use a hard register. If the
5862 condition code or comparison result can be placed in any general register,
5863 or if there are multiple condition registers, use a pseudo register.
5864 Registers used to store the condition code value will usually have a mode
5865 that is in class @code{MODE_CC}.
5867 Alternatively, you can use @code{BImode} if the comparison operator is
5868 specified already in the compare instruction. In this case, you are not
5869 interested in most macros in this section.
5872 * CC0 Condition Codes:: Old style representation of condition codes.
5873 * MODE_CC Condition Codes:: Modern representation of condition codes.
5874 * Cond Exec Macros:: Macros to control conditional execution.
5877 @node CC0 Condition Codes
5878 @subsection Representation of condition codes using @code{(cc0)}
5882 The file @file{conditions.h} defines a variable @code{cc_status} to
5883 describe how the condition code was computed (in case the interpretation of
5884 the condition code depends on the instruction that it was set by). This
5885 variable contains the RTL expressions on which the condition code is
5886 currently based, and several standard flags.
5888 Sometimes additional machine-specific flags must be defined in the machine
5889 description header file. It can also add additional machine-specific
5890 information by defining @code{CC_STATUS_MDEP}.
5892 @defmac CC_STATUS_MDEP
5893 C code for a data type which is used for declaring the @code{mdep}
5894 component of @code{cc_status}. It defaults to @code{int}.
5896 This macro is not used on machines that do not use @code{cc0}.
5899 @defmac CC_STATUS_MDEP_INIT
5900 A C expression to initialize the @code{mdep} field to ``empty''.
5901 The default definition does nothing, since most machines don't use
5902 the field anyway. If you want to use the field, you should probably
5903 define this macro to initialize it.
5905 This macro is not used on machines that do not use @code{cc0}.
5908 @defmac NOTICE_UPDATE_CC (@var{exp}, @var{insn})
5909 A C compound statement to set the components of @code{cc_status}
5910 appropriately for an insn @var{insn} whose body is @var{exp}. It is
5911 this macro's responsibility to recognize insns that set the condition
5912 code as a byproduct of other activity as well as those that explicitly
5915 This macro is not used on machines that do not use @code{cc0}.
5917 If there are insns that do not set the condition code but do alter
5918 other machine registers, this macro must check to see whether they
5919 invalidate the expressions that the condition code is recorded as
5920 reflecting. For example, on the 68000, insns that store in address
5921 registers do not set the condition code, which means that usually
5922 @code{NOTICE_UPDATE_CC} can leave @code{cc_status} unaltered for such
5923 insns. But suppose that the previous insn set the condition code
5924 based on location @samp{a4@@(102)} and the current insn stores a new
5925 value in @samp{a4}. Although the condition code is not changed by
5926 this, it will no longer be true that it reflects the contents of
5927 @samp{a4@@(102)}. Therefore, @code{NOTICE_UPDATE_CC} must alter
5928 @code{cc_status} in this case to say that nothing is known about the
5929 condition code value.
5931 The definition of @code{NOTICE_UPDATE_CC} must be prepared to deal
5932 with the results of peephole optimization: insns whose patterns are
5933 @code{parallel} RTXs containing various @code{reg}, @code{mem} or
5934 constants which are just the operands. The RTL structure of these
5935 insns is not sufficient to indicate what the insns actually do. What
5936 @code{NOTICE_UPDATE_CC} should do when it sees one is just to run
5937 @code{CC_STATUS_INIT}.
5939 A possible definition of @code{NOTICE_UPDATE_CC} is to call a function
5940 that looks at an attribute (@pxref{Insn Attributes}) named, for example,
5941 @samp{cc}. This avoids having detailed information about patterns in
5942 two places, the @file{md} file and in @code{NOTICE_UPDATE_CC}.
5945 @node MODE_CC Condition Codes
5946 @subsection Representation of condition codes using registers
5950 @defmac SELECT_CC_MODE (@var{op}, @var{x}, @var{y})
5951 On many machines, the condition code may be produced by other instructions
5952 than compares, for example the branch can use directly the condition
5953 code set by a subtract instruction. However, on some machines
5954 when the condition code is set this way some bits (such as the overflow
5955 bit) are not set in the same way as a test instruction, so that a different
5956 branch instruction must be used for some conditional branches. When
5957 this happens, use the machine mode of the condition code register to
5958 record different formats of the condition code register. Modes can
5959 also be used to record which compare instruction (e.g. a signed or an
5960 unsigned comparison) produced the condition codes.
5962 If other modes than @code{CCmode} are required, add them to
5963 @file{@var{machine}-modes.def} and define @code{SELECT_CC_MODE} to choose
5964 a mode given an operand of a compare. This is needed because the modes
5965 have to be chosen not only during RTL generation but also, for example,
5966 by instruction combination. The result of @code{SELECT_CC_MODE} should
5967 be consistent with the mode used in the patterns; for example to support
5968 the case of the add on the SPARC discussed above, we have the pattern
5972 [(set (reg:CC_NOOV 0)
5974 (plus:SI (match_operand:SI 0 "register_operand" "%r")
5975 (match_operand:SI 1 "arith_operand" "rI"))
5982 together with a @code{SELECT_CC_MODE} that returns @code{CC_NOOVmode}
5983 for comparisons whose argument is a @code{plus}:
5986 #define SELECT_CC_MODE(OP,X,Y) \
5987 (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \
5988 ? ((OP == EQ || OP == NE) ? CCFPmode : CCFPEmode) \
5989 : ((GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \
5990 || GET_CODE (X) == NEG) \
5991 ? CC_NOOVmode : CCmode))
5994 Another reason to use modes is to retain information on which operands
5995 were used by the comparison; see @code{REVERSIBLE_CC_MODE} later in
5998 You should define this macro if and only if you define extra CC modes
5999 in @file{@var{machine}-modes.def}.
6002 @defmac CANONICALIZE_COMPARISON (@var{code}, @var{op0}, @var{op1})
6003 On some machines not all possible comparisons are defined, but you can
6004 convert an invalid comparison into a valid one. For example, the Alpha
6005 does not have a @code{GT} comparison, but you can use an @code{LT}
6006 comparison instead and swap the order of the operands.
6008 On such machines, define this macro to be a C statement to do any
6009 required conversions. @var{code} is the initial comparison code
6010 and @var{op0} and @var{op1} are the left and right operands of the
6011 comparison, respectively. You should modify @var{code}, @var{op0}, and
6012 @var{op1} as required.
6014 GCC will not assume that the comparison resulting from this macro is
6015 valid but will see if the resulting insn matches a pattern in the
6018 You need not define this macro if it would never change the comparison
6022 @defmac REVERSIBLE_CC_MODE (@var{mode})
6023 A C expression whose value is one if it is always safe to reverse a
6024 comparison whose mode is @var{mode}. If @code{SELECT_CC_MODE}
6025 can ever return @var{mode} for a floating-point inequality comparison,
6026 then @code{REVERSIBLE_CC_MODE (@var{mode})} must be zero.
6028 You need not define this macro if it would always returns zero or if the
6029 floating-point format is anything other than @code{IEEE_FLOAT_FORMAT}.
6030 For example, here is the definition used on the SPARC, where floating-point
6031 inequality comparisons are always given @code{CCFPEmode}:
6034 #define REVERSIBLE_CC_MODE(MODE) ((MODE) != CCFPEmode)
6038 @defmac REVERSE_CONDITION (@var{code}, @var{mode})
6039 A C expression whose value is reversed condition code of the @var{code} for
6040 comparison done in CC_MODE @var{mode}. The macro is used only in case
6041 @code{REVERSIBLE_CC_MODE (@var{mode})} is nonzero. Define this macro in case
6042 machine has some non-standard way how to reverse certain conditionals. For
6043 instance in case all floating point conditions are non-trapping, compiler may
6044 freely convert unordered compares to ordered one. Then definition may look
6048 #define REVERSE_CONDITION(CODE, MODE) \
6049 ((MODE) != CCFPmode ? reverse_condition (CODE) \
6050 : reverse_condition_maybe_unordered (CODE))
6054 @deftypefn {Target Hook} bool TARGET_FIXED_CONDITION_CODE_REGS (unsigned int *@var{p1}, unsigned int *@var{p2})
6055 On targets which do not use @code{(cc0)}, and which use a hard
6056 register rather than a pseudo-register to hold condition codes, the
6057 regular CSE passes are often not able to identify cases in which the
6058 hard register is set to a common value. Use this hook to enable a
6059 small pass which optimizes such cases. This hook should return true
6060 to enable this pass, and it should set the integers to which its
6061 arguments point to the hard register numbers used for condition codes.
6062 When there is only one such register, as is true on most systems, the
6063 integer pointed to by @var{p2} should be set to
6064 @code{INVALID_REGNUM}.
6066 The default version of this hook returns false.
6069 @deftypefn {Target Hook} {enum machine_mode} TARGET_CC_MODES_COMPATIBLE (enum machine_mode @var{m1}, enum machine_mode @var{m2})
6070 On targets which use multiple condition code modes in class
6071 @code{MODE_CC}, it is sometimes the case that a comparison can be
6072 validly done in more than one mode. On such a system, define this
6073 target hook to take two mode arguments and to return a mode in which
6074 both comparisons may be validly done. If there is no such mode,
6075 return @code{VOIDmode}.
6077 The default version of this hook checks whether the modes are the
6078 same. If they are, it returns that mode. If they are different, it
6079 returns @code{VOIDmode}.
6082 @node Cond Exec Macros
6083 @subsection Macros to control conditional execution
6084 @findex conditional execution
6087 There is one macro that may need to be defined for targets
6088 supporting conditional execution, independent of how they
6089 represent conditional branches.
6091 @defmac REVERSE_CONDEXEC_PREDICATES_P (@var{op1}, @var{op2})
6092 A C expression that returns true if the conditional execution predicate
6093 @var{op1}, a comparison operation, is the inverse of @var{op2} and vice
6094 versa. Define this to return 0 if the target has conditional execution
6095 predicates that cannot be reversed safely. There is no need to validate
6096 that the arguments of op1 and op2 are the same, this is done separately.
6097 If no expansion is specified, this macro is defined as follows:
6100 #define REVERSE_CONDEXEC_PREDICATES_P (x, y) \
6101 (GET_CODE ((x)) == reversed_comparison_code ((y), NULL))
6106 @section Describing Relative Costs of Operations
6107 @cindex costs of instructions
6108 @cindex relative costs
6109 @cindex speed of instructions
6111 These macros let you describe the relative speed of various operations
6112 on the target machine.
6114 @defmac REGISTER_MOVE_COST (@var{mode}, @var{from}, @var{to})
6115 A C expression for the cost of moving data of mode @var{mode} from a
6116 register in class @var{from} to one in class @var{to}. The classes are
6117 expressed using the enumeration values such as @code{GENERAL_REGS}. A
6118 value of 2 is the default; other values are interpreted relative to
6121 It is not required that the cost always equal 2 when @var{from} is the
6122 same as @var{to}; on some machines it is expensive to move between
6123 registers if they are not general registers.
6125 If reload sees an insn consisting of a single @code{set} between two
6126 hard registers, and if @code{REGISTER_MOVE_COST} applied to their
6127 classes returns a value of 2, reload does not check to ensure that the
6128 constraints of the insn are met. Setting a cost of other than 2 will
6129 allow reload to verify that the constraints are met. You should do this
6130 if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
6132 These macros are obsolete, new ports should use the target hook
6133 @code{TARGET_REGISTER_MOVE_COST} instead.
6136 @deftypefn {Target Hook} int TARGET_REGISTER_MOVE_COST (enum machine_mode @var{mode}, reg_class_t @var{from}, reg_class_t @var{to})
6137 This target hook should return the cost of moving data of mode @var{mode}
6138 from a register in class @var{from} to one in class @var{to}. The classes
6139 are expressed using the enumeration values such as @code{GENERAL_REGS}.
6140 A value of 2 is the default; other values are interpreted relative to
6143 It is not required that the cost always equal 2 when @var{from} is the
6144 same as @var{to}; on some machines it is expensive to move between
6145 registers if they are not general registers.
6147 If reload sees an insn consisting of a single @code{set} between two
6148 hard registers, and if @code{TARGET_REGISTER_MOVE_COST} applied to their
6149 classes returns a value of 2, reload does not check to ensure that the
6150 constraints of the insn are met. Setting a cost of other than 2 will
6151 allow reload to verify that the constraints are met. You should do this
6152 if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
6154 The default version of this function returns 2.
6157 @defmac MEMORY_MOVE_COST (@var{mode}, @var{class}, @var{in})
6158 A C expression for the cost of moving data of mode @var{mode} between a
6159 register of class @var{class} and memory; @var{in} is zero if the value
6160 is to be written to memory, nonzero if it is to be read in. This cost
6161 is relative to those in @code{REGISTER_MOVE_COST}. If moving between
6162 registers and memory is more expensive than between two registers, you
6163 should define this macro to express the relative cost.
6165 If you do not define this macro, GCC uses a default cost of 4 plus
6166 the cost of copying via a secondary reload register, if one is
6167 needed. If your machine requires a secondary reload register to copy
6168 between memory and a register of @var{class} but the reload mechanism is
6169 more complex than copying via an intermediate, define this macro to
6170 reflect the actual cost of the move.
6172 GCC defines the function @code{memory_move_secondary_cost} if
6173 secondary reloads are needed. It computes the costs due to copying via
6174 a secondary register. If your machine copies from memory using a
6175 secondary register in the conventional way but the default base value of
6176 4 is not correct for your machine, define this macro to add some other
6177 value to the result of that function. The arguments to that function
6178 are the same as to this macro.
6180 These macros are obsolete, new ports should use the target hook
6181 @code{TARGET_MEMORY_MOVE_COST} instead.
6184 @deftypefn {Target Hook} int TARGET_MEMORY_MOVE_COST (enum machine_mode @var{mode}, reg_class_t @var{rclass}, bool @var{in})
6185 This target hook should return the cost of moving data of mode @var{mode}
6186 between a register of class @var{rclass} and memory; @var{in} is @code{false}
6187 if the value is to be written to memory, @code{true} if it is to be read in.
6188 This cost is relative to those in @code{TARGET_REGISTER_MOVE_COST}.
6189 If moving between registers and memory is more expensive than between two
6190 registers, you should add this target hook to express the relative cost.
6192 If you do not add this target hook, GCC uses a default cost of 4 plus
6193 the cost of copying via a secondary reload register, if one is
6194 needed. If your machine requires a secondary reload register to copy
6195 between memory and a register of @var{rclass} but the reload mechanism is
6196 more complex than copying via an intermediate, use this target hook to
6197 reflect the actual cost of the move.
6199 GCC defines the function @code{memory_move_secondary_cost} if
6200 secondary reloads are needed. It computes the costs due to copying via
6201 a secondary register. If your machine copies from memory using a
6202 secondary register in the conventional way but the default base value of
6203 4 is not correct for your machine, use this target hook to add some other
6204 value to the result of that function. The arguments to that function
6205 are the same as to this target hook.
6208 @defmac BRANCH_COST (@var{speed_p}, @var{predictable_p})
6209 A C expression for the cost of a branch instruction. A value of 1 is
6210 the default; other values are interpreted relative to that. Parameter
6211 @var{speed_p} is true when the branch in question should be optimized
6212 for speed. When it is false, @code{BRANCH_COST} should return a value
6213 optimal for code size rather than performance. @var{predictable_p} is
6214 true for well-predicted branches. On many architectures the
6215 @code{BRANCH_COST} can be reduced then.
6218 Here are additional macros which do not specify precise relative costs,
6219 but only that certain actions are more expensive than GCC would
6222 @defmac SLOW_BYTE_ACCESS
6223 Define this macro as a C expression which is nonzero if accessing less
6224 than a word of memory (i.e.@: a @code{char} or a @code{short}) is no
6225 faster than accessing a word of memory, i.e., if such access
6226 require more than one instruction or if there is no difference in cost
6227 between byte and (aligned) word loads.
6229 When this macro is not defined, the compiler will access a field by
6230 finding the smallest containing object; when it is defined, a fullword
6231 load will be used if alignment permits. Unless bytes accesses are
6232 faster than word accesses, using word accesses is preferable since it
6233 may eliminate subsequent memory access if subsequent accesses occur to
6234 other fields in the same word of the structure, but to different bytes.
6237 @defmac SLOW_UNALIGNED_ACCESS (@var{mode}, @var{alignment})
6238 Define this macro to be the value 1 if memory accesses described by the
6239 @var{mode} and @var{alignment} parameters have a cost many times greater
6240 than aligned accesses, for example if they are emulated in a trap
6243 When this macro is nonzero, the compiler will act as if
6244 @code{STRICT_ALIGNMENT} were nonzero when generating code for block
6245 moves. This can cause significantly more instructions to be produced.
6246 Therefore, do not set this macro nonzero if unaligned accesses only add a
6247 cycle or two to the time for a memory access.
6249 If the value of this macro is always zero, it need not be defined. If
6250 this macro is defined, it should produce a nonzero value when
6251 @code{STRICT_ALIGNMENT} is nonzero.
6254 @defmac MOVE_RATIO (@var{speed})
6255 The threshold of number of scalar memory-to-memory move insns, @emph{below}
6256 which a sequence of insns should be generated instead of a
6257 string move insn or a library call. Increasing the value will always
6258 make code faster, but eventually incurs high cost in increased code size.
6260 Note that on machines where the corresponding move insn is a
6261 @code{define_expand} that emits a sequence of insns, this macro counts
6262 the number of such sequences.
6264 The parameter @var{speed} is true if the code is currently being
6265 optimized for speed rather than size.
6267 If you don't define this, a reasonable default is used.
6270 @defmac MOVE_BY_PIECES_P (@var{size}, @var{alignment})
6271 A C expression used to determine whether @code{move_by_pieces} will be used to
6272 copy a chunk of memory, or whether some other block move mechanism
6273 will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6274 than @code{MOVE_RATIO}.
6277 @defmac MOVE_MAX_PIECES
6278 A C expression used by @code{move_by_pieces} to determine the largest unit
6279 a load or store used to copy memory is. Defaults to @code{MOVE_MAX}.
6282 @defmac CLEAR_RATIO (@var{speed})
6283 The threshold of number of scalar move insns, @emph{below} which a sequence
6284 of insns should be generated to clear memory instead of a string clear insn
6285 or a library call. Increasing the value will always make code faster, but
6286 eventually incurs high cost in increased code size.
6288 The parameter @var{speed} is true if the code is currently being
6289 optimized for speed rather than size.
6291 If you don't define this, a reasonable default is used.
6294 @defmac CLEAR_BY_PIECES_P (@var{size}, @var{alignment})
6295 A C expression used to determine whether @code{clear_by_pieces} will be used
6296 to clear a chunk of memory, or whether some other block clear mechanism
6297 will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6298 than @code{CLEAR_RATIO}.
6301 @defmac SET_RATIO (@var{speed})
6302 The threshold of number of scalar move insns, @emph{below} which a sequence
6303 of insns should be generated to set memory to a constant value, instead of
6304 a block set insn or a library call.
6305 Increasing the value will always make code faster, but
6306 eventually incurs high cost in increased code size.
6308 The parameter @var{speed} is true if the code is currently being
6309 optimized for speed rather than size.
6311 If you don't define this, it defaults to the value of @code{MOVE_RATIO}.
6314 @defmac SET_BY_PIECES_P (@var{size}, @var{alignment})
6315 A C expression used to determine whether @code{store_by_pieces} will be
6316 used to set a chunk of memory to a constant value, or whether some
6317 other mechanism will be used. Used by @code{__builtin_memset} when
6318 storing values other than constant zero.
6319 Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6320 than @code{SET_RATIO}.
6323 @defmac STORE_BY_PIECES_P (@var{size}, @var{alignment})
6324 A C expression used to determine whether @code{store_by_pieces} will be
6325 used to set a chunk of memory to a constant string value, or whether some
6326 other mechanism will be used. Used by @code{__builtin_strcpy} when
6327 called with a constant source string.
6328 Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6329 than @code{MOVE_RATIO}.
6332 @defmac USE_LOAD_POST_INCREMENT (@var{mode})
6333 A C expression used to determine whether a load postincrement is a good
6334 thing to use for a given mode. Defaults to the value of
6335 @code{HAVE_POST_INCREMENT}.
6338 @defmac USE_LOAD_POST_DECREMENT (@var{mode})
6339 A C expression used to determine whether a load postdecrement is a good
6340 thing to use for a given mode. Defaults to the value of
6341 @code{HAVE_POST_DECREMENT}.
6344 @defmac USE_LOAD_PRE_INCREMENT (@var{mode})
6345 A C expression used to determine whether a load preincrement is a good
6346 thing to use for a given mode. Defaults to the value of
6347 @code{HAVE_PRE_INCREMENT}.
6350 @defmac USE_LOAD_PRE_DECREMENT (@var{mode})
6351 A C expression used to determine whether a load predecrement is a good
6352 thing to use for a given mode. Defaults to the value of
6353 @code{HAVE_PRE_DECREMENT}.
6356 @defmac USE_STORE_POST_INCREMENT (@var{mode})
6357 A C expression used to determine whether a store postincrement is a good
6358 thing to use for a given mode. Defaults to the value of
6359 @code{HAVE_POST_INCREMENT}.
6362 @defmac USE_STORE_POST_DECREMENT (@var{mode})
6363 A C expression used to determine whether a store postdecrement is a good
6364 thing to use for a given mode. Defaults to the value of
6365 @code{HAVE_POST_DECREMENT}.
6368 @defmac USE_STORE_PRE_INCREMENT (@var{mode})
6369 This macro is used to determine whether a store preincrement is a good
6370 thing to use for a given mode. Defaults to the value of
6371 @code{HAVE_PRE_INCREMENT}.
6374 @defmac USE_STORE_PRE_DECREMENT (@var{mode})
6375 This macro is used to determine whether a store predecrement is a good
6376 thing to use for a given mode. Defaults to the value of
6377 @code{HAVE_PRE_DECREMENT}.
6380 @defmac NO_FUNCTION_CSE
6381 Define this macro if it is as good or better to call a constant
6382 function address than to call an address kept in a register.
6385 @defmac RANGE_TEST_NON_SHORT_CIRCUIT
6386 Define this macro if a non-short-circuit operation produced by
6387 @samp{fold_range_test ()} is optimal. This macro defaults to true if
6388 @code{BRANCH_COST} is greater than or equal to the value 2.
6391 @deftypefn {Target Hook} bool TARGET_RTX_COSTS (rtx @var{x}, int @var{code}, int @var{outer_code}, int @var{opno}, int *@var{total}, bool @var{speed})
6392 This target hook describes the relative costs of RTL expressions.
6394 The cost may depend on the precise form of the expression, which is
6395 available for examination in @var{x}, and the fact that @var{x} appears
6396 as operand @var{opno} of an expression with rtx code @var{outer_code}.
6397 That is, the hook can assume that there is some rtx @var{y} such
6398 that @samp{GET_CODE (@var{y}) == @var{outer_code}} and such that
6399 either (a) @samp{XEXP (@var{y}, @var{opno}) == @var{x}} or
6400 (b) @samp{XVEC (@var{y}, @var{opno})} contains @var{x}.
6402 @var{code} is @var{x}'s expression code---redundant, since it can be
6403 obtained with @code{GET_CODE (@var{x})}.
6405 In implementing this hook, you can use the construct
6406 @code{COSTS_N_INSNS (@var{n})} to specify a cost equal to @var{n} fast
6409 On entry to the hook, @code{*@var{total}} contains a default estimate
6410 for the cost of the expression. The hook should modify this value as
6411 necessary. Traditionally, the default costs are @code{COSTS_N_INSNS (5)}
6412 for multiplications, @code{COSTS_N_INSNS (7)} for division and modulus
6413 operations, and @code{COSTS_N_INSNS (1)} for all other operations.
6415 When optimizing for code size, i.e.@: when @code{speed} is
6416 false, this target hook should be used to estimate the relative
6417 size cost of an expression, again relative to @code{COSTS_N_INSNS}.
6419 The hook returns true when all subexpressions of @var{x} have been
6420 processed, and false when @code{rtx_cost} should recurse.
6423 @deftypefn {Target Hook} int TARGET_ADDRESS_COST (rtx @var{address}, bool @var{speed})
6424 This hook computes the cost of an addressing mode that contains
6425 @var{address}. If not defined, the cost is computed from
6426 the @var{address} expression and the @code{TARGET_RTX_COST} hook.
6428 For most CISC machines, the default cost is a good approximation of the
6429 true cost of the addressing mode. However, on RISC machines, all
6430 instructions normally have the same length and execution time. Hence
6431 all addresses will have equal costs.
6433 In cases where more than one form of an address is known, the form with
6434 the lowest cost will be used. If multiple forms have the same, lowest,
6435 cost, the one that is the most complex will be used.
6437 For example, suppose an address that is equal to the sum of a register
6438 and a constant is used twice in the same basic block. When this macro
6439 is not defined, the address will be computed in a register and memory
6440 references will be indirect through that register. On machines where
6441 the cost of the addressing mode containing the sum is no higher than
6442 that of a simple indirect reference, this will produce an additional
6443 instruction and possibly require an additional register. Proper
6444 specification of this macro eliminates this overhead for such machines.
6446 This hook is never called with an invalid address.
6448 On machines where an address involving more than one register is as
6449 cheap as an address computation involving only one register, defining
6450 @code{TARGET_ADDRESS_COST} to reflect this can cause two registers to
6451 be live over a region of code where only one would have been if
6452 @code{TARGET_ADDRESS_COST} were not defined in that manner. This effect
6453 should be considered in the definition of this macro. Equivalent costs
6454 should probably only be given to addresses with different numbers of
6455 registers on machines with lots of registers.
6459 @section Adjusting the Instruction Scheduler
6461 The instruction scheduler may need a fair amount of machine-specific
6462 adjustment in order to produce good code. GCC provides several target
6463 hooks for this purpose. It is usually enough to define just a few of
6464 them: try the first ones in this list first.
6466 @deftypefn {Target Hook} int TARGET_SCHED_ISSUE_RATE (void)
6467 This hook returns the maximum number of instructions that can ever
6468 issue at the same time on the target machine. The default is one.
6469 Although the insn scheduler can define itself the possibility of issue
6470 an insn on the same cycle, the value can serve as an additional
6471 constraint to issue insns on the same simulated processor cycle (see
6472 hooks @samp{TARGET_SCHED_REORDER} and @samp{TARGET_SCHED_REORDER2}).
6473 This value must be constant over the entire compilation. If you need
6474 it to vary depending on what the instructions are, you must use
6475 @samp{TARGET_SCHED_VARIABLE_ISSUE}.
6478 @deftypefn {Target Hook} int TARGET_SCHED_VARIABLE_ISSUE (FILE *@var{file}, int @var{verbose}, rtx @var{insn}, int @var{more})
6479 This hook is executed by the scheduler after it has scheduled an insn
6480 from the ready list. It should return the number of insns which can
6481 still be issued in the current cycle. The default is
6482 @samp{@w{@var{more} - 1}} for insns other than @code{CLOBBER} and
6483 @code{USE}, which normally are not counted against the issue rate.
6484 You should define this hook if some insns take more machine resources
6485 than others, so that fewer insns can follow them in the same cycle.
6486 @var{file} is either a null pointer, or a stdio stream to write any
6487 debug output to. @var{verbose} is the verbose level provided by
6488 @option{-fsched-verbose-@var{n}}. @var{insn} is the instruction that
6492 @deftypefn {Target Hook} int TARGET_SCHED_ADJUST_COST (rtx @var{insn}, rtx @var{link}, rtx @var{dep_insn}, int @var{cost})
6493 This function corrects the value of @var{cost} based on the
6494 relationship between @var{insn} and @var{dep_insn} through the
6495 dependence @var{link}. It should return the new value. The default
6496 is to make no adjustment to @var{cost}. This can be used for example
6497 to specify to the scheduler using the traditional pipeline description
6498 that an output- or anti-dependence does not incur the same cost as a
6499 data-dependence. If the scheduler using the automaton based pipeline
6500 description, the cost of anti-dependence is zero and the cost of
6501 output-dependence is maximum of one and the difference of latency
6502 times of the first and the second insns. If these values are not
6503 acceptable, you could use the hook to modify them too. See also
6504 @pxref{Processor pipeline description}.
6507 @deftypefn {Target Hook} int TARGET_SCHED_ADJUST_PRIORITY (rtx @var{insn}, int @var{priority})
6508 This hook adjusts the integer scheduling priority @var{priority} of
6509 @var{insn}. It should return the new priority. Increase the priority to
6510 execute @var{insn} earlier, reduce the priority to execute @var{insn}
6511 later. Do not define this hook if you do not need to adjust the
6512 scheduling priorities of insns.
6515 @deftypefn {Target Hook} int TARGET_SCHED_REORDER (FILE *@var{file}, int @var{verbose}, rtx *@var{ready}, int *@var{n_readyp}, int @var{clock})
6516 This hook is executed by the scheduler after it has scheduled the ready
6517 list, to allow the machine description to reorder it (for example to
6518 combine two small instructions together on @samp{VLIW} machines).
6519 @var{file} is either a null pointer, or a stdio stream to write any
6520 debug output to. @var{verbose} is the verbose level provided by
6521 @option{-fsched-verbose-@var{n}}. @var{ready} is a pointer to the ready
6522 list of instructions that are ready to be scheduled. @var{n_readyp} is
6523 a pointer to the number of elements in the ready list. The scheduler
6524 reads the ready list in reverse order, starting with
6525 @var{ready}[@var{*n_readyp} @minus{} 1] and going to @var{ready}[0]. @var{clock}
6526 is the timer tick of the scheduler. You may modify the ready list and
6527 the number of ready insns. The return value is the number of insns that
6528 can issue this cycle; normally this is just @code{issue_rate}. See also
6529 @samp{TARGET_SCHED_REORDER2}.
6532 @deftypefn {Target Hook} int TARGET_SCHED_REORDER2 (FILE *@var{file}, int @var{verbose}, rtx *@var{ready}, int *@var{n_readyp}, int @var{clock})
6533 Like @samp{TARGET_SCHED_REORDER}, but called at a different time. That
6534 function is called whenever the scheduler starts a new cycle. This one
6535 is called once per iteration over a cycle, immediately after
6536 @samp{TARGET_SCHED_VARIABLE_ISSUE}; it can reorder the ready list and
6537 return the number of insns to be scheduled in the same cycle. Defining
6538 this hook can be useful if there are frequent situations where
6539 scheduling one insn causes other insns to become ready in the same
6540 cycle. These other insns can then be taken into account properly.
6543 @deftypefn {Target Hook} void TARGET_SCHED_DEPENDENCIES_EVALUATION_HOOK (rtx @var{head}, rtx @var{tail})
6544 This hook is called after evaluation forward dependencies of insns in
6545 chain given by two parameter values (@var{head} and @var{tail}
6546 correspondingly) but before insns scheduling of the insn chain. For
6547 example, it can be used for better insn classification if it requires
6548 analysis of dependencies. This hook can use backward and forward
6549 dependencies of the insn scheduler because they are already
6553 @deftypefn {Target Hook} void TARGET_SCHED_INIT (FILE *@var{file}, int @var{verbose}, int @var{max_ready})
6554 This hook is executed by the scheduler at the beginning of each block of
6555 instructions that are to be scheduled. @var{file} is either a null
6556 pointer, or a stdio stream to write any debug output to. @var{verbose}
6557 is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6558 @var{max_ready} is the maximum number of insns in the current scheduling
6559 region that can be live at the same time. This can be used to allocate
6560 scratch space if it is needed, e.g.@: by @samp{TARGET_SCHED_REORDER}.
6563 @deftypefn {Target Hook} void TARGET_SCHED_FINISH (FILE *@var{file}, int @var{verbose})
6564 This hook is executed by the scheduler at the end of each block of
6565 instructions that are to be scheduled. It can be used to perform
6566 cleanup of any actions done by the other scheduling hooks. @var{file}
6567 is either a null pointer, or a stdio stream to write any debug output
6568 to. @var{verbose} is the verbose level provided by
6569 @option{-fsched-verbose-@var{n}}.
6572 @deftypefn {Target Hook} void TARGET_SCHED_INIT_GLOBAL (FILE *@var{file}, int @var{verbose}, int @var{old_max_uid})
6573 This hook is executed by the scheduler after function level initializations.
6574 @var{file} is either a null pointer, or a stdio stream to write any debug output to.
6575 @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6576 @var{old_max_uid} is the maximum insn uid when scheduling begins.
6579 @deftypefn {Target Hook} void TARGET_SCHED_FINISH_GLOBAL (FILE *@var{file}, int @var{verbose})
6580 This is the cleanup hook corresponding to @code{TARGET_SCHED_INIT_GLOBAL}.
6581 @var{file} is either a null pointer, or a stdio stream to write any debug output to.
6582 @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6585 @deftypefn {Target Hook} rtx TARGET_SCHED_DFA_PRE_CYCLE_INSN (void)
6586 The hook returns an RTL insn. The automaton state used in the
6587 pipeline hazard recognizer is changed as if the insn were scheduled
6588 when the new simulated processor cycle starts. Usage of the hook may
6589 simplify the automaton pipeline description for some @acronym{VLIW}
6590 processors. If the hook is defined, it is used only for the automaton
6591 based pipeline description. The default is not to change the state
6592 when the new simulated processor cycle starts.
6595 @deftypefn {Target Hook} void TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN (void)
6596 The hook can be used to initialize data used by the previous hook.
6599 @deftypefn {Target Hook} rtx TARGET_SCHED_DFA_POST_CYCLE_INSN (void)
6600 The hook is analogous to @samp{TARGET_SCHED_DFA_PRE_CYCLE_INSN} but used
6601 to changed the state as if the insn were scheduled when the new
6602 simulated processor cycle finishes.
6605 @deftypefn {Target Hook} void TARGET_SCHED_INIT_DFA_POST_CYCLE_INSN (void)
6606 The hook is analogous to @samp{TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN} but
6607 used to initialize data used by the previous hook.
6610 @deftypefn {Target Hook} void TARGET_SCHED_DFA_PRE_ADVANCE_CYCLE (void)
6611 The hook to notify target that the current simulated cycle is about to finish.
6612 The hook is analogous to @samp{TARGET_SCHED_DFA_PRE_CYCLE_INSN} but used
6613 to change the state in more complicated situations - e.g., when advancing
6614 state on a single insn is not enough.
6617 @deftypefn {Target Hook} void TARGET_SCHED_DFA_POST_ADVANCE_CYCLE (void)
6618 The hook to notify target that new simulated cycle has just started.
6619 The hook is analogous to @samp{TARGET_SCHED_DFA_POST_CYCLE_INSN} but used
6620 to change the state in more complicated situations - e.g., when advancing
6621 state on a single insn is not enough.
6624 @deftypefn {Target Hook} int TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD (void)
6625 This hook controls better choosing an insn from the ready insn queue
6626 for the @acronym{DFA}-based insn scheduler. Usually the scheduler
6627 chooses the first insn from the queue. If the hook returns a positive
6628 value, an additional scheduler code tries all permutations of
6629 @samp{TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD ()}
6630 subsequent ready insns to choose an insn whose issue will result in
6631 maximal number of issued insns on the same cycle. For the
6632 @acronym{VLIW} processor, the code could actually solve the problem of
6633 packing simple insns into the @acronym{VLIW} insn. Of course, if the
6634 rules of @acronym{VLIW} packing are described in the automaton.
6636 This code also could be used for superscalar @acronym{RISC}
6637 processors. Let us consider a superscalar @acronym{RISC} processor
6638 with 3 pipelines. Some insns can be executed in pipelines @var{A} or
6639 @var{B}, some insns can be executed only in pipelines @var{B} or
6640 @var{C}, and one insn can be executed in pipeline @var{B}. The
6641 processor may issue the 1st insn into @var{A} and the 2nd one into
6642 @var{B}. In this case, the 3rd insn will wait for freeing @var{B}
6643 until the next cycle. If the scheduler issues the 3rd insn the first,
6644 the processor could issue all 3 insns per cycle.
6646 Actually this code demonstrates advantages of the automaton based
6647 pipeline hazard recognizer. We try quickly and easy many insn
6648 schedules to choose the best one.
6650 The default is no multipass scheduling.
6653 @deftypefn {Target Hook} int TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD (rtx @var{insn})
6655 This hook controls what insns from the ready insn queue will be
6656 considered for the multipass insn scheduling. If the hook returns
6657 zero for @var{insn}, the insn will be not chosen to
6660 The default is that any ready insns can be chosen to be issued.
6663 @deftypefn {Target Hook} void TARGET_SCHED_FIRST_CYCLE_MULTIPASS_BEGIN (void *@var{data}, char *@var{ready_try}, int @var{n_ready}, bool @var{first_cycle_insn_p})
6664 This hook prepares the target backend for a new round of multipass
6668 @deftypefn {Target Hook} void TARGET_SCHED_FIRST_CYCLE_MULTIPASS_ISSUE (void *@var{data}, char *@var{ready_try}, int @var{n_ready}, rtx @var{insn}, const void *@var{prev_data})
6669 This hook is called when multipass scheduling evaluates instruction INSN.
6672 @deftypefn {Target Hook} void TARGET_SCHED_FIRST_CYCLE_MULTIPASS_BACKTRACK (const void *@var{data}, char *@var{ready_try}, int @var{n_ready})
6673 This is called when multipass scheduling backtracks from evaluation of
6677 @deftypefn {Target Hook} void TARGET_SCHED_FIRST_CYCLE_MULTIPASS_END (const void *@var{data})
6678 This hook notifies the target about the result of the concluded current
6679 round of multipass scheduling.
6682 @deftypefn {Target Hook} void TARGET_SCHED_FIRST_CYCLE_MULTIPASS_INIT (void *@var{data})
6683 This hook initializes target-specific data used in multipass scheduling.
6686 @deftypefn {Target Hook} void TARGET_SCHED_FIRST_CYCLE_MULTIPASS_FINI (void *@var{data})
6687 This hook finalizes target-specific data used in multipass scheduling.
6690 @deftypefn {Target Hook} int TARGET_SCHED_DFA_NEW_CYCLE (FILE *@var{dump}, int @var{verbose}, rtx @var{insn}, int @var{last_clock}, int @var{clock}, int *@var{sort_p})
6691 This hook is called by the insn scheduler before issuing @var{insn}
6692 on cycle @var{clock}. If the hook returns nonzero,
6693 @var{insn} is not issued on this processor cycle. Instead,
6694 the processor cycle is advanced. If *@var{sort_p}
6695 is zero, the insn ready queue is not sorted on the new cycle
6696 start as usually. @var{dump} and @var{verbose} specify the file and
6697 verbosity level to use for debugging output.
6698 @var{last_clock} and @var{clock} are, respectively, the
6699 processor cycle on which the previous insn has been issued,
6700 and the current processor cycle.
6703 @deftypefn {Target Hook} bool TARGET_SCHED_IS_COSTLY_DEPENDENCE (struct _dep *@var{_dep}, int @var{cost}, int @var{distance})
6704 This hook is used to define which dependences are considered costly by
6705 the target, so costly that it is not advisable to schedule the insns that
6706 are involved in the dependence too close to one another. The parameters
6707 to this hook are as follows: The first parameter @var{_dep} is the dependence
6708 being evaluated. The second parameter @var{cost} is the cost of the
6709 dependence as estimated by the scheduler, and the third
6710 parameter @var{distance} is the distance in cycles between the two insns.
6711 The hook returns @code{true} if considering the distance between the two
6712 insns the dependence between them is considered costly by the target,
6713 and @code{false} otherwise.
6715 Defining this hook can be useful in multiple-issue out-of-order machines,
6716 where (a) it's practically hopeless to predict the actual data/resource
6717 delays, however: (b) there's a better chance to predict the actual grouping
6718 that will be formed, and (c) correctly emulating the grouping can be very
6719 important. In such targets one may want to allow issuing dependent insns
6720 closer to one another---i.e., closer than the dependence distance; however,
6721 not in cases of ``costly dependences'', which this hooks allows to define.
6724 @deftypefn {Target Hook} void TARGET_SCHED_H_I_D_EXTENDED (void)
6725 This hook is called by the insn scheduler after emitting a new instruction to
6726 the instruction stream. The hook notifies a target backend to extend its
6727 per instruction data structures.
6730 @deftypefn {Target Hook} {void *} TARGET_SCHED_ALLOC_SCHED_CONTEXT (void)
6731 Return a pointer to a store large enough to hold target scheduling context.
6734 @deftypefn {Target Hook} void TARGET_SCHED_INIT_SCHED_CONTEXT (void *@var{tc}, bool @var{clean_p})
6735 Initialize store pointed to by @var{tc} to hold target scheduling context.
6736 It @var{clean_p} is true then initialize @var{tc} as if scheduler is at the
6737 beginning of the block. Otherwise, copy the current context into @var{tc}.
6740 @deftypefn {Target Hook} void TARGET_SCHED_SET_SCHED_CONTEXT (void *@var{tc})
6741 Copy target scheduling context pointed to by @var{tc} to the current context.
6744 @deftypefn {Target Hook} void TARGET_SCHED_CLEAR_SCHED_CONTEXT (void *@var{tc})
6745 Deallocate internal data in target scheduling context pointed to by @var{tc}.
6748 @deftypefn {Target Hook} void TARGET_SCHED_FREE_SCHED_CONTEXT (void *@var{tc})
6749 Deallocate a store for target scheduling context pointed to by @var{tc}.
6752 @deftypefn {Target Hook} int TARGET_SCHED_SPECULATE_INSN (rtx @var{insn}, int @var{request}, rtx *@var{new_pat})
6753 This hook is called by the insn scheduler when @var{insn} has only
6754 speculative dependencies and therefore can be scheduled speculatively.
6755 The hook is used to check if the pattern of @var{insn} has a speculative
6756 version and, in case of successful check, to generate that speculative
6757 pattern. The hook should return 1, if the instruction has a speculative form,
6758 or @minus{}1, if it doesn't. @var{request} describes the type of requested
6759 speculation. If the return value equals 1 then @var{new_pat} is assigned
6760 the generated speculative pattern.
6763 @deftypefn {Target Hook} bool TARGET_SCHED_NEEDS_BLOCK_P (int @var{dep_status})
6764 This hook is called by the insn scheduler during generation of recovery code
6765 for @var{insn}. It should return @code{true}, if the corresponding check
6766 instruction should branch to recovery code, or @code{false} otherwise.
6769 @deftypefn {Target Hook} rtx TARGET_SCHED_GEN_SPEC_CHECK (rtx @var{insn}, rtx @var{label}, int @var{mutate_p})
6770 This hook is called by the insn scheduler to generate a pattern for recovery
6771 check instruction. If @var{mutate_p} is zero, then @var{insn} is a
6772 speculative instruction for which the check should be generated.
6773 @var{label} is either a label of a basic block, where recovery code should
6774 be emitted, or a null pointer, when requested check doesn't branch to
6775 recovery code (a simple check). If @var{mutate_p} is nonzero, then
6776 a pattern for a branchy check corresponding to a simple check denoted by
6777 @var{insn} should be generated. In this case @var{label} can't be null.
6780 @deftypefn {Target Hook} bool TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD_SPEC (const_rtx @var{insn})
6781 This hook is used as a workaround for
6782 @samp{TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD} not being
6783 called on the first instruction of the ready list. The hook is used to
6784 discard speculative instructions that stand first in the ready list from
6785 being scheduled on the current cycle. If the hook returns @code{false},
6786 @var{insn} will not be chosen to be issued.
6787 For non-speculative instructions,
6788 the hook should always return @code{true}. For example, in the ia64 backend
6789 the hook is used to cancel data speculative insns when the ALAT table
6793 @deftypefn {Target Hook} void TARGET_SCHED_SET_SCHED_FLAGS (struct spec_info_def *@var{spec_info})
6794 This hook is used by the insn scheduler to find out what features should be
6796 The structure *@var{spec_info} should be filled in by the target.
6797 The structure describes speculation types that can be used in the scheduler.
6800 @deftypefn {Target Hook} int TARGET_SCHED_SMS_RES_MII (struct ddg *@var{g})
6801 This hook is called by the swing modulo scheduler to calculate a
6802 resource-based lower bound which is based on the resources available in
6803 the machine and the resources required by each instruction. The target
6804 backend can use @var{g} to calculate such bound. A very simple lower
6805 bound will be used in case this hook is not implemented: the total number
6806 of instructions divided by the issue rate.
6809 @deftypefn {Target Hook} bool TARGET_SCHED_DISPATCH (rtx @var{insn}, int @var{x})
6810 This hook is called by Haifa Scheduler. It returns true if dispatch scheduling
6811 is supported in hardware and the condition specified in the parameter is true.
6814 @deftypefn {Target Hook} void TARGET_SCHED_DISPATCH_DO (rtx @var{insn}, int @var{x})
6815 This hook is called by Haifa Scheduler. It performs the operation specified
6816 in its second parameter.
6819 @deftypevr {Target Hook} bool TARGET_SCHED_EXPOSED_PIPELINE
6820 True if the processor has an exposed pipeline, which means that not just
6821 the order of instructions is important for correctness when scheduling, but
6822 also the latencies of operations.
6825 @deftypefn {Target Hook} int TARGET_SCHED_REASSOCIATION_WIDTH (unsigned int @var{opc}, enum machine_mode @var{mode})
6826 This hook is called by tree reassociator to determine a level of
6827 parallelism required in output calculations chain.
6831 @section Dividing the Output into Sections (Texts, Data, @dots{})
6832 @c the above section title is WAY too long. maybe cut the part between
6833 @c the (...)? --mew 10feb93
6835 An object file is divided into sections containing different types of
6836 data. In the most common case, there are three sections: the @dfn{text
6837 section}, which holds instructions and read-only data; the @dfn{data
6838 section}, which holds initialized writable data; and the @dfn{bss
6839 section}, which holds uninitialized data. Some systems have other kinds
6842 @file{varasm.c} provides several well-known sections, such as
6843 @code{text_section}, @code{data_section} and @code{bss_section}.
6844 The normal way of controlling a @code{@var{foo}_section} variable
6845 is to define the associated @code{@var{FOO}_SECTION_ASM_OP} macro,
6846 as described below. The macros are only read once, when @file{varasm.c}
6847 initializes itself, so their values must be run-time constants.
6848 They may however depend on command-line flags.
6850 @emph{Note:} Some run-time files, such @file{crtstuff.c}, also make
6851 use of the @code{@var{FOO}_SECTION_ASM_OP} macros, and expect them
6852 to be string literals.
6854 Some assemblers require a different string to be written every time a
6855 section is selected. If your assembler falls into this category, you
6856 should define the @code{TARGET_ASM_INIT_SECTIONS} hook and use
6857 @code{get_unnamed_section} to set up the sections.
6859 You must always create a @code{text_section}, either by defining
6860 @code{TEXT_SECTION_ASM_OP} or by initializing @code{text_section}
6861 in @code{TARGET_ASM_INIT_SECTIONS}. The same is true of
6862 @code{data_section} and @code{DATA_SECTION_ASM_OP}. If you do not
6863 create a distinct @code{readonly_data_section}, the default is to
6864 reuse @code{text_section}.
6866 All the other @file{varasm.c} sections are optional, and are null
6867 if the target does not provide them.
6869 @defmac TEXT_SECTION_ASM_OP
6870 A C expression whose value is a string, including spacing, containing the
6871 assembler operation that should precede instructions and read-only data.
6872 Normally @code{"\t.text"} is right.
6875 @defmac HOT_TEXT_SECTION_NAME
6876 If defined, a C string constant for the name of the section containing most
6877 frequently executed functions of the program. If not defined, GCC will provide
6878 a default definition if the target supports named sections.
6881 @defmac UNLIKELY_EXECUTED_TEXT_SECTION_NAME
6882 If defined, a C string constant for the name of the section containing unlikely
6883 executed functions in the program.
6886 @defmac DATA_SECTION_ASM_OP
6887 A C expression whose value is a string, including spacing, containing the
6888 assembler operation to identify the following data as writable initialized
6889 data. Normally @code{"\t.data"} is right.
6892 @defmac SDATA_SECTION_ASM_OP
6893 If defined, a C expression whose value is a string, including spacing,
6894 containing the assembler operation to identify the following data as
6895 initialized, writable small data.
6898 @defmac READONLY_DATA_SECTION_ASM_OP
6899 A C expression whose value is a string, including spacing, containing the
6900 assembler operation to identify the following data as read-only initialized
6904 @defmac BSS_SECTION_ASM_OP
6905 If defined, a C expression whose value is a string, including spacing,
6906 containing the assembler operation to identify the following data as
6907 uninitialized global data. If not defined, and
6908 @code{ASM_OUTPUT_ALIGNED_BSS} not defined,
6909 uninitialized global data will be output in the data section if
6910 @option{-fno-common} is passed, otherwise @code{ASM_OUTPUT_COMMON} will be
6914 @defmac SBSS_SECTION_ASM_OP
6915 If defined, a C expression whose value is a string, including spacing,
6916 containing the assembler operation to identify the following data as
6917 uninitialized, writable small data.
6920 @defmac TLS_COMMON_ASM_OP
6921 If defined, a C expression whose value is a string containing the
6922 assembler operation to identify the following data as thread-local
6923 common data. The default is @code{".tls_common"}.
6926 @defmac TLS_SECTION_ASM_FLAG
6927 If defined, a C expression whose value is a character constant
6928 containing the flag used to mark a section as a TLS section. The
6929 default is @code{'T'}.
6932 @defmac INIT_SECTION_ASM_OP
6933 If defined, a C expression whose value is a string, including spacing,
6934 containing the assembler operation to identify the following data as
6935 initialization code. If not defined, GCC will assume such a section does
6936 not exist. This section has no corresponding @code{init_section}
6937 variable; it is used entirely in runtime code.
6940 @defmac FINI_SECTION_ASM_OP
6941 If defined, a C expression whose value is a string, including spacing,
6942 containing the assembler operation to identify the following data as
6943 finalization code. If not defined, GCC will assume such a section does
6944 not exist. This section has no corresponding @code{fini_section}
6945 variable; it is used entirely in runtime code.
6948 @defmac INIT_ARRAY_SECTION_ASM_OP
6949 If defined, a C expression whose value is a string, including spacing,
6950 containing the assembler operation to identify the following data as
6951 part of the @code{.init_array} (or equivalent) section. If not
6952 defined, GCC will assume such a section does not exist. Do not define
6953 both this macro and @code{INIT_SECTION_ASM_OP}.
6956 @defmac FINI_ARRAY_SECTION_ASM_OP
6957 If defined, a C expression whose value is a string, including spacing,
6958 containing the assembler operation to identify the following data as
6959 part of the @code{.fini_array} (or equivalent) section. If not
6960 defined, GCC will assume such a section does not exist. Do not define
6961 both this macro and @code{FINI_SECTION_ASM_OP}.
6964 @defmac CRT_CALL_STATIC_FUNCTION (@var{section_op}, @var{function})
6965 If defined, an ASM statement that switches to a different section
6966 via @var{section_op}, calls @var{function}, and switches back to
6967 the text section. This is used in @file{crtstuff.c} if
6968 @code{INIT_SECTION_ASM_OP} or @code{FINI_SECTION_ASM_OP} to calls
6969 to initialization and finalization functions from the init and fini
6970 sections. By default, this macro uses a simple function call. Some
6971 ports need hand-crafted assembly code to avoid dependencies on
6972 registers initialized in the function prologue or to ensure that
6973 constant pools don't end up too far way in the text section.
6976 @defmac TARGET_LIBGCC_SDATA_SECTION
6977 If defined, a string which names the section into which small
6978 variables defined in crtstuff and libgcc should go. This is useful
6979 when the target has options for optimizing access to small data, and
6980 you want the crtstuff and libgcc routines to be conservative in what
6981 they expect of your application yet liberal in what your application
6982 expects. For example, for targets with a @code{.sdata} section (like
6983 MIPS), you could compile crtstuff with @code{-G 0} so that it doesn't
6984 require small data support from your application, but use this macro
6985 to put small data into @code{.sdata} so that your application can
6986 access these variables whether it uses small data or not.
6989 @defmac FORCE_CODE_SECTION_ALIGN
6990 If defined, an ASM statement that aligns a code section to some
6991 arbitrary boundary. This is used to force all fragments of the
6992 @code{.init} and @code{.fini} sections to have to same alignment
6993 and thus prevent the linker from having to add any padding.
6996 @defmac JUMP_TABLES_IN_TEXT_SECTION
6997 Define this macro to be an expression with a nonzero value if jump
6998 tables (for @code{tablejump} insns) should be output in the text
6999 section, along with the assembler instructions. Otherwise, the
7000 readonly data section is used.
7002 This macro is irrelevant if there is no separate readonly data section.
7005 @deftypefn {Target Hook} void TARGET_ASM_INIT_SECTIONS (void)
7006 Define this hook if you need to do something special to set up the
7007 @file{varasm.c} sections, or if your target has some special sections
7008 of its own that you need to create.
7010 GCC calls this hook after processing the command line, but before writing
7011 any assembly code, and before calling any of the section-returning hooks
7015 @deftypefn {Target Hook} int TARGET_ASM_RELOC_RW_MASK (void)
7016 Return a mask describing how relocations should be treated when
7017 selecting sections. Bit 1 should be set if global relocations
7018 should be placed in a read-write section; bit 0 should be set if
7019 local relocations should be placed in a read-write section.
7021 The default version of this function returns 3 when @option{-fpic}
7022 is in effect, and 0 otherwise. The hook is typically redefined
7023 when the target cannot support (some kinds of) dynamic relocations
7024 in read-only sections even in executables.
7027 @deftypefn {Target Hook} {section *} TARGET_ASM_SELECT_SECTION (tree @var{exp}, int @var{reloc}, unsigned HOST_WIDE_INT @var{align})
7028 Return the section into which @var{exp} should be placed. You can
7029 assume that @var{exp} is either a @code{VAR_DECL} node or a constant of
7030 some sort. @var{reloc} indicates whether the initial value of @var{exp}
7031 requires link-time relocations. Bit 0 is set when variable contains
7032 local relocations only, while bit 1 is set for global relocations.
7033 @var{align} is the constant alignment in bits.
7035 The default version of this function takes care of putting read-only
7036 variables in @code{readonly_data_section}.
7038 See also @var{USE_SELECT_SECTION_FOR_FUNCTIONS}.
7041 @defmac USE_SELECT_SECTION_FOR_FUNCTIONS
7042 Define this macro if you wish TARGET_ASM_SELECT_SECTION to be called
7043 for @code{FUNCTION_DECL}s as well as for variables and constants.
7045 In the case of a @code{FUNCTION_DECL}, @var{reloc} will be zero if the
7046 function has been determined to be likely to be called, and nonzero if
7047 it is unlikely to be called.
7050 @deftypefn {Target Hook} void TARGET_ASM_UNIQUE_SECTION (tree @var{decl}, int @var{reloc})
7051 Build up a unique section name, expressed as a @code{STRING_CST} node,
7052 and assign it to @samp{DECL_SECTION_NAME (@var{decl})}.
7053 As with @code{TARGET_ASM_SELECT_SECTION}, @var{reloc} indicates whether
7054 the initial value of @var{exp} requires link-time relocations.
7056 The default version of this function appends the symbol name to the
7057 ELF section name that would normally be used for the symbol. For
7058 example, the function @code{foo} would be placed in @code{.text.foo}.
7059 Whatever the actual target object format, this is often good enough.
7062 @deftypefn {Target Hook} {section *} TARGET_ASM_FUNCTION_RODATA_SECTION (tree @var{decl})
7063 Return the readonly data section associated with
7064 @samp{DECL_SECTION_NAME (@var{decl})}.
7065 The default version of this function selects @code{.gnu.linkonce.r.name} if
7066 the function's section is @code{.gnu.linkonce.t.name}, @code{.rodata.name}
7067 if function is in @code{.text.name}, and the normal readonly-data section
7071 @deftypevr {Target Hook} {const char *} TARGET_ASM_MERGEABLE_RODATA_PREFIX
7072 Usually, the compiler uses the prefix @code{".rodata"} to construct
7073 section names for mergeable constant data. Define this macro to override
7074 the string if a different section name should be used.
7077 @deftypefn {Target Hook} {section *} TARGET_ASM_SELECT_RTX_SECTION (enum machine_mode @var{mode}, rtx @var{x}, unsigned HOST_WIDE_INT @var{align})
7078 Return the section into which a constant @var{x}, of mode @var{mode},
7079 should be placed. You can assume that @var{x} is some kind of
7080 constant in RTL@. The argument @var{mode} is redundant except in the
7081 case of a @code{const_int} rtx. @var{align} is the constant alignment
7084 The default version of this function takes care of putting symbolic
7085 constants in @code{flag_pic} mode in @code{data_section} and everything
7086 else in @code{readonly_data_section}.
7089 @deftypefn {Target Hook} tree TARGET_MANGLE_DECL_ASSEMBLER_NAME (tree @var{decl}, tree @var{id})
7090 Define this hook if you need to postprocess the assembler name generated
7091 by target-independent code. The @var{id} provided to this hook will be
7092 the computed name (e.g., the macro @code{DECL_NAME} of the @var{decl} in C,
7093 or the mangled name of the @var{decl} in C++). The return value of the
7094 hook is an @code{IDENTIFIER_NODE} for the appropriate mangled name on
7095 your target system. The default implementation of this hook just
7096 returns the @var{id} provided.
7099 @deftypefn {Target Hook} void TARGET_ENCODE_SECTION_INFO (tree @var{decl}, rtx @var{rtl}, int @var{new_decl_p})
7100 Define this hook if references to a symbol or a constant must be
7101 treated differently depending on something about the variable or
7102 function named by the symbol (such as what section it is in).
7104 The hook is executed immediately after rtl has been created for
7105 @var{decl}, which may be a variable or function declaration or
7106 an entry in the constant pool. In either case, @var{rtl} is the
7107 rtl in question. Do @emph{not} use @code{DECL_RTL (@var{decl})}
7108 in this hook; that field may not have been initialized yet.
7110 In the case of a constant, it is safe to assume that the rtl is
7111 a @code{mem} whose address is a @code{symbol_ref}. Most decls
7112 will also have this form, but that is not guaranteed. Global
7113 register variables, for instance, will have a @code{reg} for their
7114 rtl. (Normally the right thing to do with such unusual rtl is
7117 The @var{new_decl_p} argument will be true if this is the first time
7118 that @code{TARGET_ENCODE_SECTION_INFO} has been invoked on this decl. It will
7119 be false for subsequent invocations, which will happen for duplicate
7120 declarations. Whether or not anything must be done for the duplicate
7121 declaration depends on whether the hook examines @code{DECL_ATTRIBUTES}.
7122 @var{new_decl_p} is always true when the hook is called for a constant.
7124 @cindex @code{SYMBOL_REF_FLAG}, in @code{TARGET_ENCODE_SECTION_INFO}
7125 The usual thing for this hook to do is to record flags in the
7126 @code{symbol_ref}, using @code{SYMBOL_REF_FLAG} or @code{SYMBOL_REF_FLAGS}.
7127 Historically, the name string was modified if it was necessary to
7128 encode more than one bit of information, but this practice is now
7129 discouraged; use @code{SYMBOL_REF_FLAGS}.
7131 The default definition of this hook, @code{default_encode_section_info}
7132 in @file{varasm.c}, sets a number of commonly-useful bits in
7133 @code{SYMBOL_REF_FLAGS}. Check whether the default does what you need
7134 before overriding it.
7137 @deftypefn {Target Hook} {const char *} TARGET_STRIP_NAME_ENCODING (const char *@var{name})
7138 Decode @var{name} and return the real name part, sans
7139 the characters that @code{TARGET_ENCODE_SECTION_INFO}
7143 @deftypefn {Target Hook} bool TARGET_IN_SMALL_DATA_P (const_tree @var{exp})
7144 Returns true if @var{exp} should be placed into a ``small data'' section.
7145 The default version of this hook always returns false.
7148 @deftypevr {Target Hook} bool TARGET_HAVE_SRODATA_SECTION
7149 Contains the value true if the target places read-only
7150 ``small data'' into a separate section. The default value is false.
7153 @deftypefn {Target Hook} bool TARGET_PROFILE_BEFORE_PROLOGUE (void)
7154 It returns true if target wants profile code emitted before prologue.
7156 The default version of this hook use the target macro
7157 @code{PROFILE_BEFORE_PROLOGUE}.
7160 @deftypefn {Target Hook} bool TARGET_BINDS_LOCAL_P (const_tree @var{exp})
7161 Returns true if @var{exp} names an object for which name resolution
7162 rules must resolve to the current ``module'' (dynamic shared library
7163 or executable image).
7165 The default version of this hook implements the name resolution rules
7166 for ELF, which has a looser model of global name binding than other
7167 currently supported object file formats.
7170 @deftypevr {Target Hook} bool TARGET_HAVE_TLS
7171 Contains the value true if the target supports thread-local storage.
7172 The default value is false.
7177 @section Position Independent Code
7178 @cindex position independent code
7181 This section describes macros that help implement generation of position
7182 independent code. Simply defining these macros is not enough to
7183 generate valid PIC; you must also add support to the hook
7184 @code{TARGET_LEGITIMATE_ADDRESS_P} and to the macro
7185 @code{PRINT_OPERAND_ADDRESS}, as well as @code{LEGITIMIZE_ADDRESS}. You
7186 must modify the definition of @samp{movsi} to do something appropriate
7187 when the source operand contains a symbolic address. You may also
7188 need to alter the handling of switch statements so that they use
7190 @c i rearranged the order of the macros above to try to force one of
7191 @c them to the next line, to eliminate an overfull hbox. --mew 10feb93
7193 @defmac PIC_OFFSET_TABLE_REGNUM
7194 The register number of the register used to address a table of static
7195 data addresses in memory. In some cases this register is defined by a
7196 processor's ``application binary interface'' (ABI)@. When this macro
7197 is defined, RTL is generated for this register once, as with the stack
7198 pointer and frame pointer registers. If this macro is not defined, it
7199 is up to the machine-dependent files to allocate such a register (if
7200 necessary). Note that this register must be fixed when in use (e.g.@:
7201 when @code{flag_pic} is true).
7204 @defmac PIC_OFFSET_TABLE_REG_CALL_CLOBBERED
7205 A C expression that is nonzero if the register defined by
7206 @code{PIC_OFFSET_TABLE_REGNUM} is clobbered by calls. If not defined,
7207 the default is zero. Do not define
7208 this macro if @code{PIC_OFFSET_TABLE_REGNUM} is not defined.
7211 @defmac LEGITIMATE_PIC_OPERAND_P (@var{x})
7212 A C expression that is nonzero if @var{x} is a legitimate immediate
7213 operand on the target machine when generating position independent code.
7214 You can assume that @var{x} satisfies @code{CONSTANT_P}, so you need not
7215 check this. You can also assume @var{flag_pic} is true, so you need not
7216 check it either. You need not define this macro if all constants
7217 (including @code{SYMBOL_REF}) can be immediate operands when generating
7218 position independent code.
7221 @node Assembler Format
7222 @section Defining the Output Assembler Language
7224 This section describes macros whose principal purpose is to describe how
7225 to write instructions in assembler language---rather than what the
7229 * File Framework:: Structural information for the assembler file.
7230 * Data Output:: Output of constants (numbers, strings, addresses).
7231 * Uninitialized Data:: Output of uninitialized variables.
7232 * Label Output:: Output and generation of labels.
7233 * Initialization:: General principles of initialization
7234 and termination routines.
7235 * Macros for Initialization::
7236 Specific macros that control the handling of
7237 initialization and termination routines.
7238 * Instruction Output:: Output of actual instructions.
7239 * Dispatch Tables:: Output of jump tables.
7240 * Exception Region Output:: Output of exception region code.
7241 * Alignment Output:: Pseudo ops for alignment and skipping data.
7244 @node File Framework
7245 @subsection The Overall Framework of an Assembler File
7246 @cindex assembler format
7247 @cindex output of assembler code
7249 @c prevent bad page break with this line
7250 This describes the overall framework of an assembly file.
7252 @findex default_file_start
7253 @deftypefn {Target Hook} void TARGET_ASM_FILE_START (void)
7254 Output to @code{asm_out_file} any text which the assembler expects to
7255 find at the beginning of a file. The default behavior is controlled
7256 by two flags, documented below. Unless your target's assembler is
7257 quite unusual, if you override the default, you should call
7258 @code{default_file_start} at some point in your target hook. This
7259 lets other target files rely on these variables.
7262 @deftypevr {Target Hook} bool TARGET_ASM_FILE_START_APP_OFF
7263 If this flag is true, the text of the macro @code{ASM_APP_OFF} will be
7264 printed as the very first line in the assembly file, unless
7265 @option{-fverbose-asm} is in effect. (If that macro has been defined
7266 to the empty string, this variable has no effect.) With the normal
7267 definition of @code{ASM_APP_OFF}, the effect is to notify the GNU
7268 assembler that it need not bother stripping comments or extra
7269 whitespace from its input. This allows it to work a bit faster.
7271 The default is false. You should not set it to true unless you have
7272 verified that your port does not generate any extra whitespace or
7273 comments that will cause GAS to issue errors in NO_APP mode.
7276 @deftypevr {Target Hook} bool TARGET_ASM_FILE_START_FILE_DIRECTIVE
7277 If this flag is true, @code{output_file_directive} will be called
7278 for the primary source file, immediately after printing
7279 @code{ASM_APP_OFF} (if that is enabled). Most ELF assemblers expect
7280 this to be done. The default is false.
7283 @deftypefn {Target Hook} void TARGET_ASM_FILE_END (void)
7284 Output to @code{asm_out_file} any text which the assembler expects
7285 to find at the end of a file. The default is to output nothing.
7288 @deftypefun void file_end_indicate_exec_stack ()
7289 Some systems use a common convention, the @samp{.note.GNU-stack}
7290 special section, to indicate whether or not an object file relies on
7291 the stack being executable. If your system uses this convention, you
7292 should define @code{TARGET_ASM_FILE_END} to this function. If you
7293 need to do other things in that hook, have your hook function call
7297 @deftypefn {Target Hook} void TARGET_ASM_LTO_START (void)
7298 Output to @code{asm_out_file} any text which the assembler expects
7299 to find at the start of an LTO section. The default is to output
7303 @deftypefn {Target Hook} void TARGET_ASM_LTO_END (void)
7304 Output to @code{asm_out_file} any text which the assembler expects
7305 to find at the end of an LTO section. The default is to output
7309 @deftypefn {Target Hook} void TARGET_ASM_CODE_END (void)
7310 Output to @code{asm_out_file} any text which is needed before emitting
7311 unwind info and debug info at the end of a file. Some targets emit
7312 here PIC setup thunks that cannot be emitted at the end of file,
7313 because they couldn't have unwind info then. The default is to output
7317 @defmac ASM_COMMENT_START
7318 A C string constant describing how to begin a comment in the target
7319 assembler language. The compiler assumes that the comment will end at
7320 the end of the line.
7324 A C string constant for text to be output before each @code{asm}
7325 statement or group of consecutive ones. Normally this is
7326 @code{"#APP"}, which is a comment that has no effect on most
7327 assemblers but tells the GNU assembler that it must check the lines
7328 that follow for all valid assembler constructs.
7332 A C string constant for text to be output after each @code{asm}
7333 statement or group of consecutive ones. Normally this is
7334 @code{"#NO_APP"}, which tells the GNU assembler to resume making the
7335 time-saving assumptions that are valid for ordinary compiler output.
7338 @defmac ASM_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name})
7339 A C statement to output COFF information or DWARF debugging information
7340 which indicates that filename @var{name} is the current source file to
7341 the stdio stream @var{stream}.
7343 This macro need not be defined if the standard form of output
7344 for the file format in use is appropriate.
7347 @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_SOURCE_FILENAME (FILE *@var{file}, const char *@var{name})
7348 Output COFF information or DWARF debugging information which indicates that filename @var{name} is the current source file to the stdio stream @var{file}.
7350 This target hook need not be defined if the standard form of output for the file format in use is appropriate.
7353 @defmac OUTPUT_QUOTED_STRING (@var{stream}, @var{string})
7354 A C statement to output the string @var{string} to the stdio stream
7355 @var{stream}. If you do not call the function @code{output_quoted_string}
7356 in your config files, GCC will only call it to output filenames to
7357 the assembler source. So you can use it to canonicalize the format
7358 of the filename using this macro.
7361 @defmac ASM_OUTPUT_IDENT (@var{stream}, @var{string})
7362 A C statement to output something to the assembler file to handle a
7363 @samp{#ident} directive containing the text @var{string}. If this
7364 macro is not defined, nothing is output for a @samp{#ident} directive.
7367 @deftypefn {Target Hook} void TARGET_ASM_NAMED_SECTION (const char *@var{name}, unsigned int @var{flags}, tree @var{decl})
7368 Output assembly directives to switch to section @var{name}. The section
7369 should have attributes as specified by @var{flags}, which is a bit mask
7370 of the @code{SECTION_*} flags defined in @file{output.h}. If @var{decl}
7371 is non-NULL, it is the @code{VAR_DECL} or @code{FUNCTION_DECL} with which
7372 this section is associated.
7375 @deftypefn {Target Hook} {section *} TARGET_ASM_FUNCTION_SECTION (tree @var{decl}, enum node_frequency @var{freq}, bool @var{startup}, bool @var{exit})
7376 Return preferred text (sub)section for function @var{decl}.
7377 Main purpose of this function is to separate cold, normal and hot
7378 functions. @var{startup} is true when function is known to be used only
7379 at startup (from static constructors or it is @code{main()}).
7380 @var{exit} is true when function is known to be used only at exit
7381 (from static destructors).
7382 Return NULL if function should go to default text section.
7385 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS (FILE *@var{file}, tree @var{decl}, bool @var{new_is_cold})
7386 Used by the target to emit any assembler directives or additional labels needed when a function is partitioned between different sections. Output should be written to @var{file}. The function decl is available as @var{decl} and the new section is `cold' if @var{new_is_cold} is @code{true}.
7389 @deftypevr {Common Target Hook} bool TARGET_HAVE_NAMED_SECTIONS
7390 This flag is true if the target supports @code{TARGET_ASM_NAMED_SECTION}.
7391 It must not be modified by command-line option processing.
7394 @anchor{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}
7395 @deftypevr {Target Hook} bool TARGET_HAVE_SWITCHABLE_BSS_SECTIONS
7396 This flag is true if we can create zeroed data by switching to a BSS
7397 section and then using @code{ASM_OUTPUT_SKIP} to allocate the space.
7398 This is true on most ELF targets.
7401 @deftypefn {Target Hook} {unsigned int} TARGET_SECTION_TYPE_FLAGS (tree @var{decl}, const char *@var{name}, int @var{reloc})
7402 Choose a set of section attributes for use by @code{TARGET_ASM_NAMED_SECTION}
7403 based on a variable or function decl, a section name, and whether or not the
7404 declaration's initializer may contain runtime relocations. @var{decl} may be
7405 null, in which case read-write data should be assumed.
7407 The default version of this function handles choosing code vs data,
7408 read-only vs read-write data, and @code{flag_pic}. You should only
7409 need to override this if your target has special flags that might be
7410 set via @code{__attribute__}.
7413 @deftypefn {Target Hook} int TARGET_ASM_RECORD_GCC_SWITCHES (print_switch_type @var{type}, const char *@var{text})
7414 Provides the target with the ability to record the gcc command line
7415 switches that have been passed to the compiler, and options that are
7416 enabled. The @var{type} argument specifies what is being recorded.
7417 It can take the following values:
7420 @item SWITCH_TYPE_PASSED
7421 @var{text} is a command line switch that has been set by the user.
7423 @item SWITCH_TYPE_ENABLED
7424 @var{text} is an option which has been enabled. This might be as a
7425 direct result of a command line switch, or because it is enabled by
7426 default or because it has been enabled as a side effect of a different
7427 command line switch. For example, the @option{-O2} switch enables
7428 various different individual optimization passes.
7430 @item SWITCH_TYPE_DESCRIPTIVE
7431 @var{text} is either NULL or some descriptive text which should be
7432 ignored. If @var{text} is NULL then it is being used to warn the
7433 target hook that either recording is starting or ending. The first
7434 time @var{type} is SWITCH_TYPE_DESCRIPTIVE and @var{text} is NULL, the
7435 warning is for start up and the second time the warning is for
7436 wind down. This feature is to allow the target hook to make any
7437 necessary preparations before it starts to record switches and to
7438 perform any necessary tidying up after it has finished recording
7441 @item SWITCH_TYPE_LINE_START
7442 This option can be ignored by this target hook.
7444 @item SWITCH_TYPE_LINE_END
7445 This option can be ignored by this target hook.
7448 The hook's return value must be zero. Other return values may be
7449 supported in the future.
7451 By default this hook is set to NULL, but an example implementation is
7452 provided for ELF based targets. Called @var{elf_record_gcc_switches},
7453 it records the switches as ASCII text inside a new, string mergeable
7454 section in the assembler output file. The name of the new section is
7455 provided by the @code{TARGET_ASM_RECORD_GCC_SWITCHES_SECTION} target
7459 @deftypevr {Target Hook} {const char *} TARGET_ASM_RECORD_GCC_SWITCHES_SECTION
7460 This is the name of the section that will be created by the example
7461 ELF implementation of the @code{TARGET_ASM_RECORD_GCC_SWITCHES} target
7467 @subsection Output of Data
7470 @deftypevr {Target Hook} {const char *} TARGET_ASM_BYTE_OP
7471 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_HI_OP
7472 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_SI_OP
7473 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_DI_OP
7474 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_TI_OP
7475 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_HI_OP
7476 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_SI_OP
7477 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_DI_OP
7478 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_TI_OP
7479 These hooks specify assembly directives for creating certain kinds
7480 of integer object. The @code{TARGET_ASM_BYTE_OP} directive creates a
7481 byte-sized object, the @code{TARGET_ASM_ALIGNED_HI_OP} one creates an
7482 aligned two-byte object, and so on. Any of the hooks may be
7483 @code{NULL}, indicating that no suitable directive is available.
7485 The compiler will print these strings at the start of a new line,
7486 followed immediately by the object's initial value. In most cases,
7487 the string should contain a tab, a pseudo-op, and then another tab.
7490 @deftypefn {Target Hook} bool TARGET_ASM_INTEGER (rtx @var{x}, unsigned int @var{size}, int @var{aligned_p})
7491 The @code{assemble_integer} function uses this hook to output an
7492 integer object. @var{x} is the object's value, @var{size} is its size
7493 in bytes and @var{aligned_p} indicates whether it is aligned. The
7494 function should return @code{true} if it was able to output the
7495 object. If it returns false, @code{assemble_integer} will try to
7496 split the object into smaller parts.
7498 The default implementation of this hook will use the
7499 @code{TARGET_ASM_BYTE_OP} family of strings, returning @code{false}
7500 when the relevant string is @code{NULL}.
7503 @deftypefn {Target Hook} bool TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA (FILE *@var{file}, rtx @var{x})
7504 A target hook to recognize @var{rtx} patterns that @code{output_addr_const}
7505 can't deal with, and output assembly code to @var{file} corresponding to
7506 the pattern @var{x}. This may be used to allow machine-dependent
7507 @code{UNSPEC}s to appear within constants.
7509 If target hook fails to recognize a pattern, it must return @code{false},
7510 so that a standard error message is printed. If it prints an error message
7511 itself, by calling, for example, @code{output_operand_lossage}, it may just
7515 @defmac ASM_OUTPUT_ASCII (@var{stream}, @var{ptr}, @var{len})
7516 A C statement to output to the stdio stream @var{stream} an assembler
7517 instruction to assemble a string constant containing the @var{len}
7518 bytes at @var{ptr}. @var{ptr} will be a C expression of type
7519 @code{char *} and @var{len} a C expression of type @code{int}.
7521 If the assembler has a @code{.ascii} pseudo-op as found in the
7522 Berkeley Unix assembler, do not define the macro
7523 @code{ASM_OUTPUT_ASCII}.
7526 @defmac ASM_OUTPUT_FDESC (@var{stream}, @var{decl}, @var{n})
7527 A C statement to output word @var{n} of a function descriptor for
7528 @var{decl}. This must be defined if @code{TARGET_VTABLE_USES_DESCRIPTORS}
7529 is defined, and is otherwise unused.
7532 @defmac CONSTANT_POOL_BEFORE_FUNCTION
7533 You may define this macro as a C expression. You should define the
7534 expression to have a nonzero value if GCC should output the constant
7535 pool for a function before the code for the function, or a zero value if
7536 GCC should output the constant pool after the function. If you do
7537 not define this macro, the usual case, GCC will output the constant
7538 pool before the function.
7541 @defmac ASM_OUTPUT_POOL_PROLOGUE (@var{file}, @var{funname}, @var{fundecl}, @var{size})
7542 A C statement to output assembler commands to define the start of the
7543 constant pool for a function. @var{funname} is a string giving
7544 the name of the function. Should the return type of the function
7545 be required, it can be obtained via @var{fundecl}. @var{size}
7546 is the size, in bytes, of the constant pool that will be written
7547 immediately after this call.
7549 If no constant-pool prefix is required, the usual case, this macro need
7553 @defmac ASM_OUTPUT_SPECIAL_POOL_ENTRY (@var{file}, @var{x}, @var{mode}, @var{align}, @var{labelno}, @var{jumpto})
7554 A C statement (with or without semicolon) to output a constant in the
7555 constant pool, if it needs special treatment. (This macro need not do
7556 anything for RTL expressions that can be output normally.)
7558 The argument @var{file} is the standard I/O stream to output the
7559 assembler code on. @var{x} is the RTL expression for the constant to
7560 output, and @var{mode} is the machine mode (in case @var{x} is a
7561 @samp{const_int}). @var{align} is the required alignment for the value
7562 @var{x}; you should output an assembler directive to force this much
7565 The argument @var{labelno} is a number to use in an internal label for
7566 the address of this pool entry. The definition of this macro is
7567 responsible for outputting the label definition at the proper place.
7568 Here is how to do this:
7571 @code{(*targetm.asm_out.internal_label)} (@var{file}, "LC", @var{labelno});
7574 When you output a pool entry specially, you should end with a
7575 @code{goto} to the label @var{jumpto}. This will prevent the same pool
7576 entry from being output a second time in the usual manner.
7578 You need not define this macro if it would do nothing.
7581 @defmac ASM_OUTPUT_POOL_EPILOGUE (@var{file} @var{funname} @var{fundecl} @var{size})
7582 A C statement to output assembler commands to at the end of the constant
7583 pool for a function. @var{funname} is a string giving the name of the
7584 function. Should the return type of the function be required, you can
7585 obtain it via @var{fundecl}. @var{size} is the size, in bytes, of the
7586 constant pool that GCC wrote immediately before this call.
7588 If no constant-pool epilogue is required, the usual case, you need not
7592 @defmac IS_ASM_LOGICAL_LINE_SEPARATOR (@var{C}, @var{STR})
7593 Define this macro as a C expression which is nonzero if @var{C} is
7594 used as a logical line separator by the assembler. @var{STR} points
7595 to the position in the string where @var{C} was found; this can be used if
7596 a line separator uses multiple characters.
7598 If you do not define this macro, the default is that only
7599 the character @samp{;} is treated as a logical line separator.
7602 @deftypevr {Target Hook} {const char *} TARGET_ASM_OPEN_PAREN
7603 @deftypevrx {Target Hook} {const char *} TARGET_ASM_CLOSE_PAREN
7604 These target hooks are C string constants, describing the syntax in the
7605 assembler for grouping arithmetic expressions. If not overridden, they
7606 default to normal parentheses, which is correct for most assemblers.
7609 These macros are provided by @file{real.h} for writing the definitions
7610 of @code{ASM_OUTPUT_DOUBLE} and the like:
7612 @defmac REAL_VALUE_TO_TARGET_SINGLE (@var{x}, @var{l})
7613 @defmacx REAL_VALUE_TO_TARGET_DOUBLE (@var{x}, @var{l})
7614 @defmacx REAL_VALUE_TO_TARGET_LONG_DOUBLE (@var{x}, @var{l})
7615 @defmacx REAL_VALUE_TO_TARGET_DECIMAL32 (@var{x}, @var{l})
7616 @defmacx REAL_VALUE_TO_TARGET_DECIMAL64 (@var{x}, @var{l})
7617 @defmacx REAL_VALUE_TO_TARGET_DECIMAL128 (@var{x}, @var{l})
7618 These translate @var{x}, of type @code{REAL_VALUE_TYPE}, to the
7619 target's floating point representation, and store its bit pattern in
7620 the variable @var{l}. For @code{REAL_VALUE_TO_TARGET_SINGLE} and
7621 @code{REAL_VALUE_TO_TARGET_DECIMAL32}, this variable should be a
7622 simple @code{long int}. For the others, it should be an array of
7623 @code{long int}. The number of elements in this array is determined
7624 by the size of the desired target floating point data type: 32 bits of
7625 it go in each @code{long int} array element. Each array element holds
7626 32 bits of the result, even if @code{long int} is wider than 32 bits
7627 on the host machine.
7629 The array element values are designed so that you can print them out
7630 using @code{fprintf} in the order they should appear in the target
7634 @node Uninitialized Data
7635 @subsection Output of Uninitialized Variables
7637 Each of the macros in this section is used to do the whole job of
7638 outputting a single uninitialized variable.
7640 @defmac ASM_OUTPUT_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
7641 A C statement (sans semicolon) to output to the stdio stream
7642 @var{stream} the assembler definition of a common-label named
7643 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7644 is the size rounded up to whatever alignment the caller wants. It is
7645 possible that @var{size} may be zero, for instance if a struct with no
7646 other member than a zero-length array is defined. In this case, the
7647 backend must output a symbol definition that allocates at least one
7648 byte, both so that the address of the resulting object does not compare
7649 equal to any other, and because some object formats cannot even express
7650 the concept of a zero-sized common symbol, as that is how they represent
7651 an ordinary undefined external.
7653 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7654 output the name itself; before and after that, output the additional
7655 assembler syntax for defining the name, and a newline.
7657 This macro controls how the assembler definitions of uninitialized
7658 common global variables are output.
7661 @defmac ASM_OUTPUT_ALIGNED_COMMON (@var{stream}, @var{name}, @var{size}, @var{alignment})
7662 Like @code{ASM_OUTPUT_COMMON} except takes the required alignment as a
7663 separate, explicit argument. If you define this macro, it is used in
7664 place of @code{ASM_OUTPUT_COMMON}, and gives you more flexibility in
7665 handling the required alignment of the variable. The alignment is specified
7666 as the number of bits.
7669 @defmac ASM_OUTPUT_ALIGNED_DECL_COMMON (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7670 Like @code{ASM_OUTPUT_ALIGNED_COMMON} except that @var{decl} of the
7671 variable to be output, if there is one, or @code{NULL_TREE} if there
7672 is no corresponding variable. If you define this macro, GCC will use it
7673 in place of both @code{ASM_OUTPUT_COMMON} and
7674 @code{ASM_OUTPUT_ALIGNED_COMMON}. Define this macro when you need to see
7675 the variable's decl in order to chose what to output.
7678 @defmac ASM_OUTPUT_ALIGNED_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7679 A C statement (sans semicolon) to output to the stdio stream
7680 @var{stream} the assembler definition of uninitialized global @var{decl} named
7681 @var{name} whose size is @var{size} bytes. The variable @var{alignment}
7682 is the alignment specified as the number of bits.
7684 Try to use function @code{asm_output_aligned_bss} defined in file
7685 @file{varasm.c} when defining this macro. If unable, use the expression
7686 @code{assemble_name (@var{stream}, @var{name})} to output the name itself;
7687 before and after that, output the additional assembler syntax for defining
7688 the name, and a newline.
7690 There are two ways of handling global BSS@. One is to define this macro.
7691 The other is to have @code{TARGET_ASM_SELECT_SECTION} return a
7692 switchable BSS section (@pxref{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}).
7693 You do not need to do both.
7695 Some languages do not have @code{common} data, and require a
7696 non-common form of global BSS in order to handle uninitialized globals
7697 efficiently. C++ is one example of this. However, if the target does
7698 not support global BSS, the front end may choose to make globals
7699 common in order to save space in the object file.
7702 @defmac ASM_OUTPUT_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded})
7703 A C statement (sans semicolon) to output to the stdio stream
7704 @var{stream} the assembler definition of a local-common-label named
7705 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7706 is the size rounded up to whatever alignment the caller wants.
7708 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7709 output the name itself; before and after that, output the additional
7710 assembler syntax for defining the name, and a newline.
7712 This macro controls how the assembler definitions of uninitialized
7713 static variables are output.
7716 @defmac ASM_OUTPUT_ALIGNED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{alignment})
7717 Like @code{ASM_OUTPUT_LOCAL} except takes the required alignment as a
7718 separate, explicit argument. If you define this macro, it is used in
7719 place of @code{ASM_OUTPUT_LOCAL}, and gives you more flexibility in
7720 handling the required alignment of the variable. The alignment is specified
7721 as the number of bits.
7724 @defmac ASM_OUTPUT_ALIGNED_DECL_LOCAL (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7725 Like @code{ASM_OUTPUT_ALIGNED_DECL} except that @var{decl} of the
7726 variable to be output, if there is one, or @code{NULL_TREE} if there
7727 is no corresponding variable. If you define this macro, GCC will use it
7728 in place of both @code{ASM_OUTPUT_DECL} and
7729 @code{ASM_OUTPUT_ALIGNED_DECL}. Define this macro when you need to see
7730 the variable's decl in order to chose what to output.
7734 @subsection Output and Generation of Labels
7736 @c prevent bad page break with this line
7737 This is about outputting labels.
7739 @findex assemble_name
7740 @defmac ASM_OUTPUT_LABEL (@var{stream}, @var{name})
7741 A C statement (sans semicolon) to output to the stdio stream
7742 @var{stream} the assembler definition of a label named @var{name}.
7743 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7744 output the name itself; before and after that, output the additional
7745 assembler syntax for defining the name, and a newline. A default
7746 definition of this macro is provided which is correct for most systems.
7749 @defmac ASM_OUTPUT_FUNCTION_LABEL (@var{stream}, @var{name}, @var{decl})
7750 A C statement (sans semicolon) to output to the stdio stream
7751 @var{stream} the assembler definition of a label named @var{name} of
7753 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7754 output the name itself; before and after that, output the additional
7755 assembler syntax for defining the name, and a newline. A default
7756 definition of this macro is provided which is correct for most systems.
7758 If this macro is not defined, then the function name is defined in the
7759 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
7762 @findex assemble_name_raw
7763 @defmac ASM_OUTPUT_INTERNAL_LABEL (@var{stream}, @var{name})
7764 Identical to @code{ASM_OUTPUT_LABEL}, except that @var{name} is known
7765 to refer to a compiler-generated label. The default definition uses
7766 @code{assemble_name_raw}, which is like @code{assemble_name} except
7767 that it is more efficient.
7771 A C string containing the appropriate assembler directive to specify the
7772 size of a symbol, without any arguments. On systems that use ELF, the
7773 default (in @file{config/elfos.h}) is @samp{"\t.size\t"}; on other
7774 systems, the default is not to define this macro.
7776 Define this macro only if it is correct to use the default definitions
7777 of @code{ASM_OUTPUT_SIZE_DIRECTIVE} and @code{ASM_OUTPUT_MEASURED_SIZE}
7778 for your system. If you need your own custom definitions of those
7779 macros, or if you do not need explicit symbol sizes at all, do not
7783 @defmac ASM_OUTPUT_SIZE_DIRECTIVE (@var{stream}, @var{name}, @var{size})
7784 A C statement (sans semicolon) to output to the stdio stream
7785 @var{stream} a directive telling the assembler that the size of the
7786 symbol @var{name} is @var{size}. @var{size} is a @code{HOST_WIDE_INT}.
7787 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
7791 @defmac ASM_OUTPUT_MEASURED_SIZE (@var{stream}, @var{name})
7792 A C statement (sans semicolon) to output to the stdio stream
7793 @var{stream} a directive telling the assembler to calculate the size of
7794 the symbol @var{name} by subtracting its address from the current
7797 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
7798 provided. The default assumes that the assembler recognizes a special
7799 @samp{.} symbol as referring to the current address, and can calculate
7800 the difference between this and another symbol. If your assembler does
7801 not recognize @samp{.} or cannot do calculations with it, you will need
7802 to redefine @code{ASM_OUTPUT_MEASURED_SIZE} to use some other technique.
7806 A C string containing the appropriate assembler directive to specify the
7807 type of a symbol, without any arguments. On systems that use ELF, the
7808 default (in @file{config/elfos.h}) is @samp{"\t.type\t"}; on other
7809 systems, the default is not to define this macro.
7811 Define this macro only if it is correct to use the default definition of
7812 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
7813 custom definition of this macro, or if you do not need explicit symbol
7814 types at all, do not define this macro.
7817 @defmac TYPE_OPERAND_FMT
7818 A C string which specifies (using @code{printf} syntax) the format of
7819 the second operand to @code{TYPE_ASM_OP}. On systems that use ELF, the
7820 default (in @file{config/elfos.h}) is @samp{"@@%s"}; on other systems,
7821 the default is not to define this macro.
7823 Define this macro only if it is correct to use the default definition of
7824 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
7825 custom definition of this macro, or if you do not need explicit symbol
7826 types at all, do not define this macro.
7829 @defmac ASM_OUTPUT_TYPE_DIRECTIVE (@var{stream}, @var{type})
7830 A C statement (sans semicolon) to output to the stdio stream
7831 @var{stream} a directive telling the assembler that the type of the
7832 symbol @var{name} is @var{type}. @var{type} is a C string; currently,
7833 that string is always either @samp{"function"} or @samp{"object"}, but
7834 you should not count on this.
7836 If you define @code{TYPE_ASM_OP} and @code{TYPE_OPERAND_FMT}, a default
7837 definition of this macro is provided.
7840 @defmac ASM_DECLARE_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl})
7841 A C statement (sans semicolon) to output to the stdio stream
7842 @var{stream} any text necessary for declaring the name @var{name} of a
7843 function which is being defined. This macro is responsible for
7844 outputting the label definition (perhaps using
7845 @code{ASM_OUTPUT_FUNCTION_LABEL}). The argument @var{decl} is the
7846 @code{FUNCTION_DECL} tree node representing the function.
7848 If this macro is not defined, then the function name is defined in the
7849 usual manner as a label (by means of @code{ASM_OUTPUT_FUNCTION_LABEL}).
7851 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition
7855 @defmac ASM_DECLARE_FUNCTION_SIZE (@var{stream}, @var{name}, @var{decl})
7856 A C statement (sans semicolon) to output to the stdio stream
7857 @var{stream} any text necessary for declaring the size of a function
7858 which is being defined. The argument @var{name} is the name of the
7859 function. The argument @var{decl} is the @code{FUNCTION_DECL} tree node
7860 representing the function.
7862 If this macro is not defined, then the function size is not defined.
7864 You may wish to use @code{ASM_OUTPUT_MEASURED_SIZE} in the definition
7868 @defmac ASM_DECLARE_OBJECT_NAME (@var{stream}, @var{name}, @var{decl})
7869 A C statement (sans semicolon) to output to the stdio stream
7870 @var{stream} any text necessary for declaring the name @var{name} of an
7871 initialized variable which is being defined. This macro must output the
7872 label definition (perhaps using @code{ASM_OUTPUT_LABEL}). The argument
7873 @var{decl} is the @code{VAR_DECL} tree node representing the variable.
7875 If this macro is not defined, then the variable name is defined in the
7876 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
7878 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} and/or
7879 @code{ASM_OUTPUT_SIZE_DIRECTIVE} in the definition of this macro.
7882 @deftypefn {Target Hook} void TARGET_ASM_DECLARE_CONSTANT_NAME (FILE *@var{file}, const char *@var{name}, const_tree @var{expr}, HOST_WIDE_INT @var{size})
7883 A target hook to output to the stdio stream @var{file} any text necessary
7884 for declaring the name @var{name} of a constant which is being defined. This
7885 target hook is responsible for outputting the label definition (perhaps using
7886 @code{assemble_label}). The argument @var{exp} is the value of the constant,
7887 and @var{size} is the size of the constant in bytes. The @var{name}
7888 will be an internal label.
7890 The default version of this target hook, define the @var{name} in the
7891 usual manner as a label (by means of @code{assemble_label}).
7893 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in this target hook.
7896 @defmac ASM_DECLARE_REGISTER_GLOBAL (@var{stream}, @var{decl}, @var{regno}, @var{name})
7897 A C statement (sans semicolon) to output to the stdio stream
7898 @var{stream} any text necessary for claiming a register @var{regno}
7899 for a global variable @var{decl} with name @var{name}.
7901 If you don't define this macro, that is equivalent to defining it to do
7905 @defmac ASM_FINISH_DECLARE_OBJECT (@var{stream}, @var{decl}, @var{toplevel}, @var{atend})
7906 A C statement (sans semicolon) to finish up declaring a variable name
7907 once the compiler has processed its initializer fully and thus has had a
7908 chance to determine the size of an array when controlled by an
7909 initializer. This is used on systems where it's necessary to declare
7910 something about the size of the object.
7912 If you don't define this macro, that is equivalent to defining it to do
7915 You may wish to use @code{ASM_OUTPUT_SIZE_DIRECTIVE} and/or
7916 @code{ASM_OUTPUT_MEASURED_SIZE} in the definition of this macro.
7919 @deftypefn {Target Hook} void TARGET_ASM_GLOBALIZE_LABEL (FILE *@var{stream}, const char *@var{name})
7920 This target hook is a function to output to the stdio stream
7921 @var{stream} some commands that will make the label @var{name} global;
7922 that is, available for reference from other files.
7924 The default implementation relies on a proper definition of
7925 @code{GLOBAL_ASM_OP}.
7928 @deftypefn {Target Hook} void TARGET_ASM_GLOBALIZE_DECL_NAME (FILE *@var{stream}, tree @var{decl})
7929 This target hook is a function to output to the stdio stream
7930 @var{stream} some commands that will make the name associated with @var{decl}
7931 global; that is, available for reference from other files.
7933 The default implementation uses the TARGET_ASM_GLOBALIZE_LABEL target hook.
7936 @defmac ASM_WEAKEN_LABEL (@var{stream}, @var{name})
7937 A C statement (sans semicolon) to output to the stdio stream
7938 @var{stream} some commands that will make the label @var{name} weak;
7939 that is, available for reference from other files but only used if
7940 no other definition is available. Use the expression
7941 @code{assemble_name (@var{stream}, @var{name})} to output the name
7942 itself; before and after that, output the additional assembler syntax
7943 for making that name weak, and a newline.
7945 If you don't define this macro or @code{ASM_WEAKEN_DECL}, GCC will not
7946 support weak symbols and you should not define the @code{SUPPORTS_WEAK}
7950 @defmac ASM_WEAKEN_DECL (@var{stream}, @var{decl}, @var{name}, @var{value})
7951 Combines (and replaces) the function of @code{ASM_WEAKEN_LABEL} and
7952 @code{ASM_OUTPUT_WEAK_ALIAS}, allowing access to the associated function
7953 or variable decl. If @var{value} is not @code{NULL}, this C statement
7954 should output to the stdio stream @var{stream} assembler code which
7955 defines (equates) the weak symbol @var{name} to have the value
7956 @var{value}. If @var{value} is @code{NULL}, it should output commands
7957 to make @var{name} weak.
7960 @defmac ASM_OUTPUT_WEAKREF (@var{stream}, @var{decl}, @var{name}, @var{value})
7961 Outputs a directive that enables @var{name} to be used to refer to
7962 symbol @var{value} with weak-symbol semantics. @code{decl} is the
7963 declaration of @code{name}.
7966 @defmac SUPPORTS_WEAK
7967 A preprocessor constant expression which evaluates to true if the target
7968 supports weak symbols.
7970 If you don't define this macro, @file{defaults.h} provides a default
7971 definition. If either @code{ASM_WEAKEN_LABEL} or @code{ASM_WEAKEN_DECL}
7972 is defined, the default definition is @samp{1}; otherwise, it is @samp{0}.
7975 @defmac TARGET_SUPPORTS_WEAK
7976 A C expression which evaluates to true if the target supports weak symbols.
7978 If you don't define this macro, @file{defaults.h} provides a default
7979 definition. The default definition is @samp{(SUPPORTS_WEAK)}. Define
7980 this macro if you want to control weak symbol support with a compiler
7981 flag such as @option{-melf}.
7984 @defmac MAKE_DECL_ONE_ONLY (@var{decl})
7985 A C statement (sans semicolon) to mark @var{decl} to be emitted as a
7986 public symbol such that extra copies in multiple translation units will
7987 be discarded by the linker. Define this macro if your object file
7988 format provides support for this concept, such as the @samp{COMDAT}
7989 section flags in the Microsoft Windows PE/COFF format, and this support
7990 requires changes to @var{decl}, such as putting it in a separate section.
7993 @defmac SUPPORTS_ONE_ONLY
7994 A C expression which evaluates to true if the target supports one-only
7997 If you don't define this macro, @file{varasm.c} provides a default
7998 definition. If @code{MAKE_DECL_ONE_ONLY} is defined, the default
7999 definition is @samp{1}; otherwise, it is @samp{0}. Define this macro if
8000 you want to control one-only symbol support with a compiler flag, or if
8001 setting the @code{DECL_ONE_ONLY} flag is enough to mark a declaration to
8002 be emitted as one-only.
8005 @deftypefn {Target Hook} void TARGET_ASM_ASSEMBLE_VISIBILITY (tree @var{decl}, int @var{visibility})
8006 This target hook is a function to output to @var{asm_out_file} some
8007 commands that will make the symbol(s) associated with @var{decl} have
8008 hidden, protected or internal visibility as specified by @var{visibility}.
8011 @defmac TARGET_WEAK_NOT_IN_ARCHIVE_TOC
8012 A C expression that evaluates to true if the target's linker expects
8013 that weak symbols do not appear in a static archive's table of contents.
8014 The default is @code{0}.
8016 Leaving weak symbols out of an archive's table of contents means that,
8017 if a symbol will only have a definition in one translation unit and
8018 will have undefined references from other translation units, that
8019 symbol should not be weak. Defining this macro to be nonzero will
8020 thus have the effect that certain symbols that would normally be weak
8021 (explicit template instantiations, and vtables for polymorphic classes
8022 with noninline key methods) will instead be nonweak.
8024 The C++ ABI requires this macro to be zero. Define this macro for
8025 targets where full C++ ABI compliance is impossible and where linker
8026 restrictions require weak symbols to be left out of a static archive's
8030 @defmac ASM_OUTPUT_EXTERNAL (@var{stream}, @var{decl}, @var{name})
8031 A C statement (sans semicolon) to output to the stdio stream
8032 @var{stream} any text necessary for declaring the name of an external
8033 symbol named @var{name} which is referenced in this compilation but
8034 not defined. The value of @var{decl} is the tree node for the
8037 This macro need not be defined if it does not need to output anything.
8038 The GNU assembler and most Unix assemblers don't require anything.
8041 @deftypefn {Target Hook} void TARGET_ASM_EXTERNAL_LIBCALL (rtx @var{symref})
8042 This target hook is a function to output to @var{asm_out_file} an assembler
8043 pseudo-op to declare a library function name external. The name of the
8044 library function is given by @var{symref}, which is a @code{symbol_ref}.
8047 @deftypefn {Target Hook} void TARGET_ASM_MARK_DECL_PRESERVED (const char *@var{symbol})
8048 This target hook is a function to output to @var{asm_out_file} an assembler
8049 directive to annotate @var{symbol} as used. The Darwin target uses the
8050 .no_dead_code_strip directive.
8053 @defmac ASM_OUTPUT_LABELREF (@var{stream}, @var{name})
8054 A C statement (sans semicolon) to output to the stdio stream
8055 @var{stream} a reference in assembler syntax to a label named
8056 @var{name}. This should add @samp{_} to the front of the name, if that
8057 is customary on your operating system, as it is in most Berkeley Unix
8058 systems. This macro is used in @code{assemble_name}.
8061 @deftypefn {Target Hook} tree TARGET_MANGLE_ASSEMBLER_NAME (const char *@var{name})
8062 Given a symbol @var{name}, perform same mangling as @code{varasm.c}'s @code{assemble_name}, but in memory rather than to a file stream, returning result as an @code{IDENTIFIER_NODE}. Required for correct LTO symtabs. The default implementation calls the @code{TARGET_STRIP_NAME_ENCODING} hook and then prepends the @code{USER_LABEL_PREFIX}, if any.
8065 @defmac ASM_OUTPUT_SYMBOL_REF (@var{stream}, @var{sym})
8066 A C statement (sans semicolon) to output a reference to
8067 @code{SYMBOL_REF} @var{sym}. If not defined, @code{assemble_name}
8068 will be used to output the name of the symbol. This macro may be used
8069 to modify the way a symbol is referenced depending on information
8070 encoded by @code{TARGET_ENCODE_SECTION_INFO}.
8073 @defmac ASM_OUTPUT_LABEL_REF (@var{stream}, @var{buf})
8074 A C statement (sans semicolon) to output a reference to @var{buf}, the
8075 result of @code{ASM_GENERATE_INTERNAL_LABEL}. If not defined,
8076 @code{assemble_name} will be used to output the name of the symbol.
8077 This macro is not used by @code{output_asm_label}, or the @code{%l}
8078 specifier that calls it; the intention is that this macro should be set
8079 when it is necessary to output a label differently when its address is
8083 @deftypefn {Target Hook} void TARGET_ASM_INTERNAL_LABEL (FILE *@var{stream}, const char *@var{prefix}, unsigned long @var{labelno})
8084 A function to output to the stdio stream @var{stream} a label whose
8085 name is made from the string @var{prefix} and the number @var{labelno}.
8087 It is absolutely essential that these labels be distinct from the labels
8088 used for user-level functions and variables. Otherwise, certain programs
8089 will have name conflicts with internal labels.
8091 It is desirable to exclude internal labels from the symbol table of the
8092 object file. Most assemblers have a naming convention for labels that
8093 should be excluded; on many systems, the letter @samp{L} at the
8094 beginning of a label has this effect. You should find out what
8095 convention your system uses, and follow it.
8097 The default version of this function utilizes @code{ASM_GENERATE_INTERNAL_LABEL}.
8100 @defmac ASM_OUTPUT_DEBUG_LABEL (@var{stream}, @var{prefix}, @var{num})
8101 A C statement to output to the stdio stream @var{stream} a debug info
8102 label whose name is made from the string @var{prefix} and the number
8103 @var{num}. This is useful for VLIW targets, where debug info labels
8104 may need to be treated differently than branch target labels. On some
8105 systems, branch target labels must be at the beginning of instruction
8106 bundles, but debug info labels can occur in the middle of instruction
8109 If this macro is not defined, then @code{(*targetm.asm_out.internal_label)} will be
8113 @defmac ASM_GENERATE_INTERNAL_LABEL (@var{string}, @var{prefix}, @var{num})
8114 A C statement to store into the string @var{string} a label whose name
8115 is made from the string @var{prefix} and the number @var{num}.
8117 This string, when output subsequently by @code{assemble_name}, should
8118 produce the output that @code{(*targetm.asm_out.internal_label)} would produce
8119 with the same @var{prefix} and @var{num}.
8121 If the string begins with @samp{*}, then @code{assemble_name} will
8122 output the rest of the string unchanged. It is often convenient for
8123 @code{ASM_GENERATE_INTERNAL_LABEL} to use @samp{*} in this way. If the
8124 string doesn't start with @samp{*}, then @code{ASM_OUTPUT_LABELREF} gets
8125 to output the string, and may change it. (Of course,
8126 @code{ASM_OUTPUT_LABELREF} is also part of your machine description, so
8127 you should know what it does on your machine.)
8130 @defmac ASM_FORMAT_PRIVATE_NAME (@var{outvar}, @var{name}, @var{number})
8131 A C expression to assign to @var{outvar} (which is a variable of type
8132 @code{char *}) a newly allocated string made from the string
8133 @var{name} and the number @var{number}, with some suitable punctuation
8134 added. Use @code{alloca} to get space for the string.
8136 The string will be used as an argument to @code{ASM_OUTPUT_LABELREF} to
8137 produce an assembler label for an internal static variable whose name is
8138 @var{name}. Therefore, the string must be such as to result in valid
8139 assembler code. The argument @var{number} is different each time this
8140 macro is executed; it prevents conflicts between similarly-named
8141 internal static variables in different scopes.
8143 Ideally this string should not be a valid C identifier, to prevent any
8144 conflict with the user's own symbols. Most assemblers allow periods
8145 or percent signs in assembler symbols; putting at least one of these
8146 between the name and the number will suffice.
8148 If this macro is not defined, a default definition will be provided
8149 which is correct for most systems.
8152 @defmac ASM_OUTPUT_DEF (@var{stream}, @var{name}, @var{value})
8153 A C statement to output to the stdio stream @var{stream} assembler code
8154 which defines (equates) the symbol @var{name} to have the value @var{value}.
8157 If @code{SET_ASM_OP} is defined, a default definition is provided which is
8158 correct for most systems.
8161 @defmac ASM_OUTPUT_DEF_FROM_DECLS (@var{stream}, @var{decl_of_name}, @var{decl_of_value})
8162 A C statement to output to the stdio stream @var{stream} assembler code
8163 which defines (equates) the symbol whose tree node is @var{decl_of_name}
8164 to have the value of the tree node @var{decl_of_value}. This macro will
8165 be used in preference to @samp{ASM_OUTPUT_DEF} if it is defined and if
8166 the tree nodes are available.
8169 If @code{SET_ASM_OP} is defined, a default definition is provided which is
8170 correct for most systems.
8173 @defmac TARGET_DEFERRED_OUTPUT_DEFS (@var{decl_of_name}, @var{decl_of_value})
8174 A C statement that evaluates to true if the assembler code which defines
8175 (equates) the symbol whose tree node is @var{decl_of_name} to have the value
8176 of the tree node @var{decl_of_value} should be emitted near the end of the
8177 current compilation unit. The default is to not defer output of defines.
8178 This macro affects defines output by @samp{ASM_OUTPUT_DEF} and
8179 @samp{ASM_OUTPUT_DEF_FROM_DECLS}.
8182 @defmac ASM_OUTPUT_WEAK_ALIAS (@var{stream}, @var{name}, @var{value})
8183 A C statement to output to the stdio stream @var{stream} assembler code
8184 which defines (equates) the weak symbol @var{name} to have the value
8185 @var{value}. If @var{value} is @code{NULL}, it defines @var{name} as
8186 an undefined weak symbol.
8188 Define this macro if the target only supports weak aliases; define
8189 @code{ASM_OUTPUT_DEF} instead if possible.
8192 @defmac OBJC_GEN_METHOD_LABEL (@var{buf}, @var{is_inst}, @var{class_name}, @var{cat_name}, @var{sel_name})
8193 Define this macro to override the default assembler names used for
8194 Objective-C methods.
8196 The default name is a unique method number followed by the name of the
8197 class (e.g.@: @samp{_1_Foo}). For methods in categories, the name of
8198 the category is also included in the assembler name (e.g.@:
8201 These names are safe on most systems, but make debugging difficult since
8202 the method's selector is not present in the name. Therefore, particular
8203 systems define other ways of computing names.
8205 @var{buf} is an expression of type @code{char *} which gives you a
8206 buffer in which to store the name; its length is as long as
8207 @var{class_name}, @var{cat_name} and @var{sel_name} put together, plus
8208 50 characters extra.
8210 The argument @var{is_inst} specifies whether the method is an instance
8211 method or a class method; @var{class_name} is the name of the class;
8212 @var{cat_name} is the name of the category (or @code{NULL} if the method is not
8213 in a category); and @var{sel_name} is the name of the selector.
8215 On systems where the assembler can handle quoted names, you can use this
8216 macro to provide more human-readable names.
8219 @defmac ASM_DECLARE_CLASS_REFERENCE (@var{stream}, @var{name})
8220 A C statement (sans semicolon) to output to the stdio stream
8221 @var{stream} commands to declare that the label @var{name} is an
8222 Objective-C class reference. This is only needed for targets whose
8223 linkers have special support for NeXT-style runtimes.
8226 @defmac ASM_DECLARE_UNRESOLVED_REFERENCE (@var{stream}, @var{name})
8227 A C statement (sans semicolon) to output to the stdio stream
8228 @var{stream} commands to declare that the label @var{name} is an
8229 unresolved Objective-C class reference. This is only needed for targets
8230 whose linkers have special support for NeXT-style runtimes.
8233 @node Initialization
8234 @subsection How Initialization Functions Are Handled
8235 @cindex initialization routines
8236 @cindex termination routines
8237 @cindex constructors, output of
8238 @cindex destructors, output of
8240 The compiled code for certain languages includes @dfn{constructors}
8241 (also called @dfn{initialization routines})---functions to initialize
8242 data in the program when the program is started. These functions need
8243 to be called before the program is ``started''---that is to say, before
8244 @code{main} is called.
8246 Compiling some languages generates @dfn{destructors} (also called
8247 @dfn{termination routines}) that should be called when the program
8250 To make the initialization and termination functions work, the compiler
8251 must output something in the assembler code to cause those functions to
8252 be called at the appropriate time. When you port the compiler to a new
8253 system, you need to specify how to do this.
8255 There are two major ways that GCC currently supports the execution of
8256 initialization and termination functions. Each way has two variants.
8257 Much of the structure is common to all four variations.
8259 @findex __CTOR_LIST__
8260 @findex __DTOR_LIST__
8261 The linker must build two lists of these functions---a list of
8262 initialization functions, called @code{__CTOR_LIST__}, and a list of
8263 termination functions, called @code{__DTOR_LIST__}.
8265 Each list always begins with an ignored function pointer (which may hold
8266 0, @minus{}1, or a count of the function pointers after it, depending on
8267 the environment). This is followed by a series of zero or more function
8268 pointers to constructors (or destructors), followed by a function
8269 pointer containing zero.
8271 Depending on the operating system and its executable file format, either
8272 @file{crtstuff.c} or @file{libgcc2.c} traverses these lists at startup
8273 time and exit time. Constructors are called in reverse order of the
8274 list; destructors in forward order.
8276 The best way to handle static constructors works only for object file
8277 formats which provide arbitrarily-named sections. A section is set
8278 aside for a list of constructors, and another for a list of destructors.
8279 Traditionally these are called @samp{.ctors} and @samp{.dtors}. Each
8280 object file that defines an initialization function also puts a word in
8281 the constructor section to point to that function. The linker
8282 accumulates all these words into one contiguous @samp{.ctors} section.
8283 Termination functions are handled similarly.
8285 This method will be chosen as the default by @file{target-def.h} if
8286 @code{TARGET_ASM_NAMED_SECTION} is defined. A target that does not
8287 support arbitrary sections, but does support special designated
8288 constructor and destructor sections may define @code{CTORS_SECTION_ASM_OP}
8289 and @code{DTORS_SECTION_ASM_OP} to achieve the same effect.
8291 When arbitrary sections are available, there are two variants, depending
8292 upon how the code in @file{crtstuff.c} is called. On systems that
8293 support a @dfn{.init} section which is executed at program startup,
8294 parts of @file{crtstuff.c} are compiled into that section. The
8295 program is linked by the @command{gcc} driver like this:
8298 ld -o @var{output_file} crti.o crtbegin.o @dots{} -lgcc crtend.o crtn.o
8301 The prologue of a function (@code{__init}) appears in the @code{.init}
8302 section of @file{crti.o}; the epilogue appears in @file{crtn.o}. Likewise
8303 for the function @code{__fini} in the @dfn{.fini} section. Normally these
8304 files are provided by the operating system or by the GNU C library, but
8305 are provided by GCC for a few targets.
8307 The objects @file{crtbegin.o} and @file{crtend.o} are (for most targets)
8308 compiled from @file{crtstuff.c}. They contain, among other things, code
8309 fragments within the @code{.init} and @code{.fini} sections that branch
8310 to routines in the @code{.text} section. The linker will pull all parts
8311 of a section together, which results in a complete @code{__init} function
8312 that invokes the routines we need at startup.
8314 To use this variant, you must define the @code{INIT_SECTION_ASM_OP}
8317 If no init section is available, when GCC compiles any function called
8318 @code{main} (or more accurately, any function designated as a program
8319 entry point by the language front end calling @code{expand_main_function}),
8320 it inserts a procedure call to @code{__main} as the first executable code
8321 after the function prologue. The @code{__main} function is defined
8322 in @file{libgcc2.c} and runs the global constructors.
8324 In file formats that don't support arbitrary sections, there are again
8325 two variants. In the simplest variant, the GNU linker (GNU @code{ld})
8326 and an `a.out' format must be used. In this case,
8327 @code{TARGET_ASM_CONSTRUCTOR} is defined to produce a @code{.stabs}
8328 entry of type @samp{N_SETT}, referencing the name @code{__CTOR_LIST__},
8329 and with the address of the void function containing the initialization
8330 code as its value. The GNU linker recognizes this as a request to add
8331 the value to a @dfn{set}; the values are accumulated, and are eventually
8332 placed in the executable as a vector in the format described above, with
8333 a leading (ignored) count and a trailing zero element.
8334 @code{TARGET_ASM_DESTRUCTOR} is handled similarly. Since no init
8335 section is available, the absence of @code{INIT_SECTION_ASM_OP} causes
8336 the compilation of @code{main} to call @code{__main} as above, starting
8337 the initialization process.
8339 The last variant uses neither arbitrary sections nor the GNU linker.
8340 This is preferable when you want to do dynamic linking and when using
8341 file formats which the GNU linker does not support, such as `ECOFF'@. In
8342 this case, @code{TARGET_HAVE_CTORS_DTORS} is false, initialization and
8343 termination functions are recognized simply by their names. This requires
8344 an extra program in the linkage step, called @command{collect2}. This program
8345 pretends to be the linker, for use with GCC; it does its job by running
8346 the ordinary linker, but also arranges to include the vectors of
8347 initialization and termination functions. These functions are called
8348 via @code{__main} as described above. In order to use this method,
8349 @code{use_collect2} must be defined in the target in @file{config.gcc}.
8352 The following section describes the specific macros that control and
8353 customize the handling of initialization and termination functions.
8356 @node Macros for Initialization
8357 @subsection Macros Controlling Initialization Routines
8359 Here are the macros that control how the compiler handles initialization
8360 and termination functions:
8362 @defmac INIT_SECTION_ASM_OP
8363 If defined, a C string constant, including spacing, for the assembler
8364 operation to identify the following data as initialization code. If not
8365 defined, GCC will assume such a section does not exist. When you are
8366 using special sections for initialization and termination functions, this
8367 macro also controls how @file{crtstuff.c} and @file{libgcc2.c} arrange to
8368 run the initialization functions.
8371 @defmac HAS_INIT_SECTION
8372 If defined, @code{main} will not call @code{__main} as described above.
8373 This macro should be defined for systems that control start-up code
8374 on a symbol-by-symbol basis, such as OSF/1, and should not
8375 be defined explicitly for systems that support @code{INIT_SECTION_ASM_OP}.
8378 @defmac LD_INIT_SWITCH
8379 If defined, a C string constant for a switch that tells the linker that
8380 the following symbol is an initialization routine.
8383 @defmac LD_FINI_SWITCH
8384 If defined, a C string constant for a switch that tells the linker that
8385 the following symbol is a finalization routine.
8388 @defmac COLLECT_SHARED_INIT_FUNC (@var{stream}, @var{func})
8389 If defined, a C statement that will write a function that can be
8390 automatically called when a shared library is loaded. The function
8391 should call @var{func}, which takes no arguments. If not defined, and
8392 the object format requires an explicit initialization function, then a
8393 function called @code{_GLOBAL__DI} will be generated.
8395 This function and the following one are used by collect2 when linking a
8396 shared library that needs constructors or destructors, or has DWARF2
8397 exception tables embedded in the code.
8400 @defmac COLLECT_SHARED_FINI_FUNC (@var{stream}, @var{func})
8401 If defined, a C statement that will write a function that can be
8402 automatically called when a shared library is unloaded. The function
8403 should call @var{func}, which takes no arguments. If not defined, and
8404 the object format requires an explicit finalization function, then a
8405 function called @code{_GLOBAL__DD} will be generated.
8408 @defmac INVOKE__main
8409 If defined, @code{main} will call @code{__main} despite the presence of
8410 @code{INIT_SECTION_ASM_OP}. This macro should be defined for systems
8411 where the init section is not actually run automatically, but is still
8412 useful for collecting the lists of constructors and destructors.
8415 @defmac SUPPORTS_INIT_PRIORITY
8416 If nonzero, the C++ @code{init_priority} attribute is supported and the
8417 compiler should emit instructions to control the order of initialization
8418 of objects. If zero, the compiler will issue an error message upon
8419 encountering an @code{init_priority} attribute.
8422 @deftypevr {Target Hook} bool TARGET_HAVE_CTORS_DTORS
8423 This value is true if the target supports some ``native'' method of
8424 collecting constructors and destructors to be run at startup and exit.
8425 It is false if we must use @command{collect2}.
8428 @deftypefn {Target Hook} void TARGET_ASM_CONSTRUCTOR (rtx @var{symbol}, int @var{priority})
8429 If defined, a function that outputs assembler code to arrange to call
8430 the function referenced by @var{symbol} at initialization time.
8432 Assume that @var{symbol} is a @code{SYMBOL_REF} for a function taking
8433 no arguments and with no return value. If the target supports initialization
8434 priorities, @var{priority} is a value between 0 and @code{MAX_INIT_PRIORITY};
8435 otherwise it must be @code{DEFAULT_INIT_PRIORITY}.
8437 If this macro is not defined by the target, a suitable default will
8438 be chosen if (1) the target supports arbitrary section names, (2) the
8439 target defines @code{CTORS_SECTION_ASM_OP}, or (3) @code{USE_COLLECT2}
8443 @deftypefn {Target Hook} void TARGET_ASM_DESTRUCTOR (rtx @var{symbol}, int @var{priority})
8444 This is like @code{TARGET_ASM_CONSTRUCTOR} but used for termination
8445 functions rather than initialization functions.
8448 If @code{TARGET_HAVE_CTORS_DTORS} is true, the initialization routine
8449 generated for the generated object file will have static linkage.
8451 If your system uses @command{collect2} as the means of processing
8452 constructors, then that program normally uses @command{nm} to scan
8453 an object file for constructor functions to be called.
8455 On certain kinds of systems, you can define this macro to make
8456 @command{collect2} work faster (and, in some cases, make it work at all):
8458 @defmac OBJECT_FORMAT_COFF
8459 Define this macro if the system uses COFF (Common Object File Format)
8460 object files, so that @command{collect2} can assume this format and scan
8461 object files directly for dynamic constructor/destructor functions.
8463 This macro is effective only in a native compiler; @command{collect2} as
8464 part of a cross compiler always uses @command{nm} for the target machine.
8467 @defmac REAL_NM_FILE_NAME
8468 Define this macro as a C string constant containing the file name to use
8469 to execute @command{nm}. The default is to search the path normally for
8474 @command{collect2} calls @command{nm} to scan object files for static
8475 constructors and destructors and LTO info. By default, @option{-n} is
8476 passed. Define @code{NM_FLAGS} to a C string constant if other options
8477 are needed to get the same output format as GNU @command{nm -n}
8481 If your system supports shared libraries and has a program to list the
8482 dynamic dependencies of a given library or executable, you can define
8483 these macros to enable support for running initialization and
8484 termination functions in shared libraries:
8487 Define this macro to a C string constant containing the name of the program
8488 which lists dynamic dependencies, like @command{ldd} under SunOS 4.
8491 @defmac PARSE_LDD_OUTPUT (@var{ptr})
8492 Define this macro to be C code that extracts filenames from the output
8493 of the program denoted by @code{LDD_SUFFIX}. @var{ptr} is a variable
8494 of type @code{char *} that points to the beginning of a line of output
8495 from @code{LDD_SUFFIX}. If the line lists a dynamic dependency, the
8496 code must advance @var{ptr} to the beginning of the filename on that
8497 line. Otherwise, it must set @var{ptr} to @code{NULL}.
8500 @defmac SHLIB_SUFFIX
8501 Define this macro to a C string constant containing the default shared
8502 library extension of the target (e.g., @samp{".so"}). @command{collect2}
8503 strips version information after this suffix when generating global
8504 constructor and destructor names. This define is only needed on targets
8505 that use @command{collect2} to process constructors and destructors.
8508 @node Instruction Output
8509 @subsection Output of Assembler Instructions
8511 @c prevent bad page break with this line
8512 This describes assembler instruction output.
8514 @defmac REGISTER_NAMES
8515 A C initializer containing the assembler's names for the machine
8516 registers, each one as a C string constant. This is what translates
8517 register numbers in the compiler into assembler language.
8520 @defmac ADDITIONAL_REGISTER_NAMES
8521 If defined, a C initializer for an array of structures containing a name
8522 and a register number. This macro defines additional names for hard
8523 registers, thus allowing the @code{asm} option in declarations to refer
8524 to registers using alternate names.
8527 @defmac OVERLAPPING_REGISTER_NAMES
8528 If defined, a C initializer for an array of structures containing a
8529 name, a register number and a count of the number of consecutive
8530 machine registers the name overlaps. This macro defines additional
8531 names for hard registers, thus allowing the @code{asm} option in
8532 declarations to refer to registers using alternate names. Unlike
8533 @code{ADDITIONAL_REGISTER_NAMES}, this macro should be used when the
8534 register name implies multiple underlying registers.
8536 This macro should be used when it is important that a clobber in an
8537 @code{asm} statement clobbers all the underlying values implied by the
8538 register name. For example, on ARM, clobbering the double-precision
8539 VFP register ``d0'' implies clobbering both single-precision registers
8543 @defmac ASM_OUTPUT_OPCODE (@var{stream}, @var{ptr})
8544 Define this macro if you are using an unusual assembler that
8545 requires different names for the machine instructions.
8547 The definition is a C statement or statements which output an
8548 assembler instruction opcode to the stdio stream @var{stream}. The
8549 macro-operand @var{ptr} is a variable of type @code{char *} which
8550 points to the opcode name in its ``internal'' form---the form that is
8551 written in the machine description. The definition should output the
8552 opcode name to @var{stream}, performing any translation you desire, and
8553 increment the variable @var{ptr} to point at the end of the opcode
8554 so that it will not be output twice.
8556 In fact, your macro definition may process less than the entire opcode
8557 name, or more than the opcode name; but if you want to process text
8558 that includes @samp{%}-sequences to substitute operands, you must take
8559 care of the substitution yourself. Just be sure to increment
8560 @var{ptr} over whatever text should not be output normally.
8562 @findex recog_data.operand
8563 If you need to look at the operand values, they can be found as the
8564 elements of @code{recog_data.operand}.
8566 If the macro definition does nothing, the instruction is output
8570 @defmac FINAL_PRESCAN_INSN (@var{insn}, @var{opvec}, @var{noperands})
8571 If defined, a C statement to be executed just prior to the output of
8572 assembler code for @var{insn}, to modify the extracted operands so
8573 they will be output differently.
8575 Here the argument @var{opvec} is the vector containing the operands
8576 extracted from @var{insn}, and @var{noperands} is the number of
8577 elements of the vector which contain meaningful data for this insn.
8578 The contents of this vector are what will be used to convert the insn
8579 template into assembler code, so you can change the assembler output
8580 by changing the contents of the vector.
8582 This macro is useful when various assembler syntaxes share a single
8583 file of instruction patterns; by defining this macro differently, you
8584 can cause a large class of instructions to be output differently (such
8585 as with rearranged operands). Naturally, variations in assembler
8586 syntax affecting individual insn patterns ought to be handled by
8587 writing conditional output routines in those patterns.
8589 If this macro is not defined, it is equivalent to a null statement.
8592 @deftypefn {Target Hook} void TARGET_ASM_FINAL_POSTSCAN_INSN (FILE *@var{file}, rtx @var{insn}, rtx *@var{opvec}, int @var{noperands})
8593 If defined, this target hook is a function which is executed just after the
8594 output of assembler code for @var{insn}, to change the mode of the assembler
8597 Here the argument @var{opvec} is the vector containing the operands
8598 extracted from @var{insn}, and @var{noperands} is the number of
8599 elements of the vector which contain meaningful data for this insn.
8600 The contents of this vector are what was used to convert the insn
8601 template into assembler code, so you can change the assembler mode
8602 by checking the contents of the vector.
8605 @defmac PRINT_OPERAND (@var{stream}, @var{x}, @var{code})
8606 A C compound statement to output to stdio stream @var{stream} the
8607 assembler syntax for an instruction operand @var{x}. @var{x} is an
8610 @var{code} is a value that can be used to specify one of several ways
8611 of printing the operand. It is used when identical operands must be
8612 printed differently depending on the context. @var{code} comes from
8613 the @samp{%} specification that was used to request printing of the
8614 operand. If the specification was just @samp{%@var{digit}} then
8615 @var{code} is 0; if the specification was @samp{%@var{ltr}
8616 @var{digit}} then @var{code} is the ASCII code for @var{ltr}.
8619 If @var{x} is a register, this macro should print the register's name.
8620 The names can be found in an array @code{reg_names} whose type is
8621 @code{char *[]}. @code{reg_names} is initialized from
8622 @code{REGISTER_NAMES}.
8624 When the machine description has a specification @samp{%@var{punct}}
8625 (a @samp{%} followed by a punctuation character), this macro is called
8626 with a null pointer for @var{x} and the punctuation character for
8630 @defmac PRINT_OPERAND_PUNCT_VALID_P (@var{code})
8631 A C expression which evaluates to true if @var{code} is a valid
8632 punctuation character for use in the @code{PRINT_OPERAND} macro. If
8633 @code{PRINT_OPERAND_PUNCT_VALID_P} is not defined, it means that no
8634 punctuation characters (except for the standard one, @samp{%}) are used
8638 @defmac PRINT_OPERAND_ADDRESS (@var{stream}, @var{x})
8639 A C compound statement to output to stdio stream @var{stream} the
8640 assembler syntax for an instruction operand that is a memory reference
8641 whose address is @var{x}. @var{x} is an RTL expression.
8643 @cindex @code{TARGET_ENCODE_SECTION_INFO} usage
8644 On some machines, the syntax for a symbolic address depends on the
8645 section that the address refers to. On these machines, define the hook
8646 @code{TARGET_ENCODE_SECTION_INFO} to store the information into the
8647 @code{symbol_ref}, and then check for it here. @xref{Assembler
8651 @findex dbr_sequence_length
8652 @defmac DBR_OUTPUT_SEQEND (@var{file})
8653 A C statement, to be executed after all slot-filler instructions have
8654 been output. If necessary, call @code{dbr_sequence_length} to
8655 determine the number of slots filled in a sequence (zero if not
8656 currently outputting a sequence), to decide how many no-ops to output,
8659 Don't define this macro if it has nothing to do, but it is helpful in
8660 reading assembly output if the extent of the delay sequence is made
8661 explicit (e.g.@: with white space).
8664 @findex final_sequence
8665 Note that output routines for instructions with delay slots must be
8666 prepared to deal with not being output as part of a sequence
8667 (i.e.@: when the scheduling pass is not run, or when no slot fillers could be
8668 found.) The variable @code{final_sequence} is null when not
8669 processing a sequence, otherwise it contains the @code{sequence} rtx
8673 @defmac REGISTER_PREFIX
8674 @defmacx LOCAL_LABEL_PREFIX
8675 @defmacx USER_LABEL_PREFIX
8676 @defmacx IMMEDIATE_PREFIX
8677 If defined, C string expressions to be used for the @samp{%R}, @samp{%L},
8678 @samp{%U}, and @samp{%I} options of @code{asm_fprintf} (see
8679 @file{final.c}). These are useful when a single @file{md} file must
8680 support multiple assembler formats. In that case, the various @file{tm.h}
8681 files can define these macros differently.
8684 @defmac ASM_FPRINTF_EXTENSIONS (@var{file}, @var{argptr}, @var{format})
8685 If defined this macro should expand to a series of @code{case}
8686 statements which will be parsed inside the @code{switch} statement of
8687 the @code{asm_fprintf} function. This allows targets to define extra
8688 printf formats which may useful when generating their assembler
8689 statements. Note that uppercase letters are reserved for future
8690 generic extensions to asm_fprintf, and so are not available to target
8691 specific code. The output file is given by the parameter @var{file}.
8692 The varargs input pointer is @var{argptr} and the rest of the format
8693 string, starting the character after the one that is being switched
8694 upon, is pointed to by @var{format}.
8697 @defmac ASSEMBLER_DIALECT
8698 If your target supports multiple dialects of assembler language (such as
8699 different opcodes), define this macro as a C expression that gives the
8700 numeric index of the assembler language dialect to use, with zero as the
8703 If this macro is defined, you may use constructs of the form
8705 @samp{@{option0|option1|option2@dots{}@}}
8708 in the output templates of patterns (@pxref{Output Template}) or in the
8709 first argument of @code{asm_fprintf}. This construct outputs
8710 @samp{option0}, @samp{option1}, @samp{option2}, etc., if the value of
8711 @code{ASSEMBLER_DIALECT} is zero, one, two, etc. Any special characters
8712 within these strings retain their usual meaning. If there are fewer
8713 alternatives within the braces than the value of
8714 @code{ASSEMBLER_DIALECT}, the construct outputs nothing.
8716 If you do not define this macro, the characters @samp{@{}, @samp{|} and
8717 @samp{@}} do not have any special meaning when used in templates or
8718 operands to @code{asm_fprintf}.
8720 Define the macros @code{REGISTER_PREFIX}, @code{LOCAL_LABEL_PREFIX},
8721 @code{USER_LABEL_PREFIX} and @code{IMMEDIATE_PREFIX} if you can express
8722 the variations in assembler language syntax with that mechanism. Define
8723 @code{ASSEMBLER_DIALECT} and use the @samp{@{option0|option1@}} syntax
8724 if the syntax variant are larger and involve such things as different
8725 opcodes or operand order.
8728 @defmac ASM_OUTPUT_REG_PUSH (@var{stream}, @var{regno})
8729 A C expression to output to @var{stream} some assembler code
8730 which will push hard register number @var{regno} onto the stack.
8731 The code need not be optimal, since this macro is used only when
8735 @defmac ASM_OUTPUT_REG_POP (@var{stream}, @var{regno})
8736 A C expression to output to @var{stream} some assembler code
8737 which will pop hard register number @var{regno} off of the stack.
8738 The code need not be optimal, since this macro is used only when
8742 @node Dispatch Tables
8743 @subsection Output of Dispatch Tables
8745 @c prevent bad page break with this line
8746 This concerns dispatch tables.
8748 @cindex dispatch table
8749 @defmac ASM_OUTPUT_ADDR_DIFF_ELT (@var{stream}, @var{body}, @var{value}, @var{rel})
8750 A C statement to output to the stdio stream @var{stream} an assembler
8751 pseudo-instruction to generate a difference between two labels.
8752 @var{value} and @var{rel} are the numbers of two internal labels. The
8753 definitions of these labels are output using
8754 @code{(*targetm.asm_out.internal_label)}, and they must be printed in the same
8755 way here. For example,
8758 fprintf (@var{stream}, "\t.word L%d-L%d\n",
8759 @var{value}, @var{rel})
8762 You must provide this macro on machines where the addresses in a
8763 dispatch table are relative to the table's own address. If defined, GCC
8764 will also use this macro on all machines when producing PIC@.
8765 @var{body} is the body of the @code{ADDR_DIFF_VEC}; it is provided so that the
8766 mode and flags can be read.
8769 @defmac ASM_OUTPUT_ADDR_VEC_ELT (@var{stream}, @var{value})
8770 This macro should be provided on machines where the addresses
8771 in a dispatch table are absolute.
8773 The definition should be a C statement to output to the stdio stream
8774 @var{stream} an assembler pseudo-instruction to generate a reference to
8775 a label. @var{value} is the number of an internal label whose
8776 definition is output using @code{(*targetm.asm_out.internal_label)}.
8780 fprintf (@var{stream}, "\t.word L%d\n", @var{value})
8784 @defmac ASM_OUTPUT_CASE_LABEL (@var{stream}, @var{prefix}, @var{num}, @var{table})
8785 Define this if the label before a jump-table needs to be output
8786 specially. The first three arguments are the same as for
8787 @code{(*targetm.asm_out.internal_label)}; the fourth argument is the
8788 jump-table which follows (a @code{jump_insn} containing an
8789 @code{addr_vec} or @code{addr_diff_vec}).
8791 This feature is used on system V to output a @code{swbeg} statement
8794 If this macro is not defined, these labels are output with
8795 @code{(*targetm.asm_out.internal_label)}.
8798 @defmac ASM_OUTPUT_CASE_END (@var{stream}, @var{num}, @var{table})
8799 Define this if something special must be output at the end of a
8800 jump-table. The definition should be a C statement to be executed
8801 after the assembler code for the table is written. It should write
8802 the appropriate code to stdio stream @var{stream}. The argument
8803 @var{table} is the jump-table insn, and @var{num} is the label-number
8804 of the preceding label.
8806 If this macro is not defined, nothing special is output at the end of
8810 @deftypefn {Target Hook} void TARGET_ASM_EMIT_UNWIND_LABEL (FILE *@var{stream}, tree @var{decl}, int @var{for_eh}, int @var{empty})
8811 This target hook emits a label at the beginning of each FDE@. It
8812 should be defined on targets where FDEs need special labels, and it
8813 should write the appropriate label, for the FDE associated with the
8814 function declaration @var{decl}, to the stdio stream @var{stream}.
8815 The third argument, @var{for_eh}, is a boolean: true if this is for an
8816 exception table. The fourth argument, @var{empty}, is a boolean:
8817 true if this is a placeholder label for an omitted FDE@.
8819 The default is that FDEs are not given nonlocal labels.
8822 @deftypefn {Target Hook} void TARGET_ASM_EMIT_EXCEPT_TABLE_LABEL (FILE *@var{stream})
8823 This target hook emits a label at the beginning of the exception table.
8824 It should be defined on targets where it is desirable for the table
8825 to be broken up according to function.
8827 The default is that no label is emitted.
8830 @deftypefn {Target Hook} void TARGET_ASM_EMIT_EXCEPT_PERSONALITY (rtx @var{personality})
8831 If the target implements @code{TARGET_ASM_UNWIND_EMIT}, this hook may be used to emit a directive to install a personality hook into the unwind info. This hook should not be used if dwarf2 unwind info is used.
8834 @deftypefn {Target Hook} void TARGET_ASM_UNWIND_EMIT (FILE *@var{stream}, rtx @var{insn})
8835 This target hook emits assembly directives required to unwind the
8836 given instruction. This is only used when @code{TARGET_EXCEPT_UNWIND_INFO}
8837 returns @code{UI_TARGET}.
8840 @deftypevr {Target Hook} bool TARGET_ASM_UNWIND_EMIT_BEFORE_INSN
8841 True if the @code{TARGET_ASM_UNWIND_EMIT} hook should be called before the assembly for @var{insn} has been emitted, false if the hook should be called afterward.
8844 @node Exception Region Output
8845 @subsection Assembler Commands for Exception Regions
8847 @c prevent bad page break with this line
8849 This describes commands marking the start and the end of an exception
8852 @defmac EH_FRAME_SECTION_NAME
8853 If defined, a C string constant for the name of the section containing
8854 exception handling frame unwind information. If not defined, GCC will
8855 provide a default definition if the target supports named sections.
8856 @file{crtstuff.c} uses this macro to switch to the appropriate section.
8858 You should define this symbol if your target supports DWARF 2 frame
8859 unwind information and the default definition does not work.
8862 @defmac EH_FRAME_IN_DATA_SECTION
8863 If defined, DWARF 2 frame unwind information will be placed in the
8864 data section even though the target supports named sections. This
8865 might be necessary, for instance, if the system linker does garbage
8866 collection and sections cannot be marked as not to be collected.
8868 Do not define this macro unless @code{TARGET_ASM_NAMED_SECTION} is
8872 @defmac EH_TABLES_CAN_BE_READ_ONLY
8873 Define this macro to 1 if your target is such that no frame unwind
8874 information encoding used with non-PIC code will ever require a
8875 runtime relocation, but the linker may not support merging read-only
8876 and read-write sections into a single read-write section.
8879 @defmac MASK_RETURN_ADDR
8880 An rtx used to mask the return address found via @code{RETURN_ADDR_RTX}, so
8881 that it does not contain any extraneous set bits in it.
8884 @defmac DWARF2_UNWIND_INFO
8885 Define this macro to 0 if your target supports DWARF 2 frame unwind
8886 information, but it does not yet work with exception handling.
8887 Otherwise, if your target supports this information (if it defines
8888 @code{INCOMING_RETURN_ADDR_RTX} and either @code{UNALIGNED_INT_ASM_OP}
8889 or @code{OBJECT_FORMAT_ELF}), GCC will provide a default definition of 1.
8892 @deftypefn {Common Target Hook} {enum unwind_info_type} TARGET_EXCEPT_UNWIND_INFO (struct gcc_options *@var{opts})
8893 This hook defines the mechanism that will be used for exception handling
8894 by the target. If the target has ABI specified unwind tables, the hook
8895 should return @code{UI_TARGET}. If the target is to use the
8896 @code{setjmp}/@code{longjmp}-based exception handling scheme, the hook
8897 should return @code{UI_SJLJ}. If the target supports DWARF 2 frame unwind
8898 information, the hook should return @code{UI_DWARF2}.
8900 A target may, if exceptions are disabled, choose to return @code{UI_NONE}.
8901 This may end up simplifying other parts of target-specific code. The
8902 default implementation of this hook never returns @code{UI_NONE}.
8904 Note that the value returned by this hook should be constant. It should
8905 not depend on anything except the command-line switches described by
8906 @var{opts}. In particular, the
8907 setting @code{UI_SJLJ} must be fixed at compiler start-up as C pre-processor
8908 macros and builtin functions related to exception handling are set up
8909 depending on this setting.
8911 The default implementation of the hook first honors the
8912 @option{--enable-sjlj-exceptions} configure option, then
8913 @code{DWARF2_UNWIND_INFO}, and finally defaults to @code{UI_SJLJ}. If
8914 @code{DWARF2_UNWIND_INFO} depends on command-line options, the target
8915 must define this hook so that @var{opts} is used correctly.
8918 @deftypevr {Common Target Hook} bool TARGET_UNWIND_TABLES_DEFAULT
8919 This variable should be set to @code{true} if the target ABI requires unwinding
8920 tables even when exceptions are not used. It must not be modified by
8921 command-line option processing.
8924 @defmac DONT_USE_BUILTIN_SETJMP
8925 Define this macro to 1 if the @code{setjmp}/@code{longjmp}-based scheme
8926 should use the @code{setjmp}/@code{longjmp} functions from the C library
8927 instead of the @code{__builtin_setjmp}/@code{__builtin_longjmp} machinery.
8930 @defmac DWARF_CIE_DATA_ALIGNMENT
8931 This macro need only be defined if the target might save registers in the
8932 function prologue at an offset to the stack pointer that is not aligned to
8933 @code{UNITS_PER_WORD}. The definition should be the negative minimum
8934 alignment if @code{STACK_GROWS_DOWNWARD} is defined, and the positive
8935 minimum alignment otherwise. @xref{SDB and DWARF}. Only applicable if
8936 the target supports DWARF 2 frame unwind information.
8939 @deftypevr {Target Hook} bool TARGET_TERMINATE_DW2_EH_FRAME_INFO
8940 Contains the value true if the target should add a zero word onto the
8941 end of a Dwarf-2 frame info section when used for exception handling.
8942 Default value is false if @code{EH_FRAME_SECTION_NAME} is defined, and
8946 @deftypefn {Target Hook} rtx TARGET_DWARF_REGISTER_SPAN (rtx @var{reg})
8947 Given a register, this hook should return a parallel of registers to
8948 represent where to find the register pieces. Define this hook if the
8949 register and its mode are represented in Dwarf in non-contiguous
8950 locations, or if the register should be represented in more than one
8951 register in Dwarf. Otherwise, this hook should return @code{NULL_RTX}.
8952 If not defined, the default is to return @code{NULL_RTX}.
8955 @deftypefn {Target Hook} void TARGET_INIT_DWARF_REG_SIZES_EXTRA (tree @var{address})
8956 If some registers are represented in Dwarf-2 unwind information in
8957 multiple pieces, define this hook to fill in information about the
8958 sizes of those pieces in the table used by the unwinder at runtime.
8959 It will be called by @code{expand_builtin_init_dwarf_reg_sizes} after
8960 filling in a single size corresponding to each hard register;
8961 @var{address} is the address of the table.
8964 @deftypefn {Target Hook} bool TARGET_ASM_TTYPE (rtx @var{sym})
8965 This hook is used to output a reference from a frame unwinding table to
8966 the type_info object identified by @var{sym}. It should return @code{true}
8967 if the reference was output. Returning @code{false} will cause the
8968 reference to be output using the normal Dwarf2 routines.
8971 @deftypevr {Target Hook} bool TARGET_ARM_EABI_UNWINDER
8972 This flag should be set to @code{true} on targets that use an ARM EABI
8973 based unwinding library, and @code{false} on other targets. This effects
8974 the format of unwinding tables, and how the unwinder in entered after
8975 running a cleanup. The default is @code{false}.
8978 @node Alignment Output
8979 @subsection Assembler Commands for Alignment
8981 @c prevent bad page break with this line
8982 This describes commands for alignment.
8984 @defmac JUMP_ALIGN (@var{label})
8985 The alignment (log base 2) to put in front of @var{label}, which is
8986 a common destination of jumps and has no fallthru incoming edge.
8988 This macro need not be defined if you don't want any special alignment
8989 to be done at such a time. Most machine descriptions do not currently
8992 Unless it's necessary to inspect the @var{label} parameter, it is better
8993 to set the variable @var{align_jumps} in the target's
8994 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
8995 selection in @var{align_jumps} in a @code{JUMP_ALIGN} implementation.
8998 @deftypefn {Target Hook} int TARGET_ASM_JUMP_ALIGN_MAX_SKIP (rtx @var{label})
8999 The maximum number of bytes to skip before @var{label} when applying
9000 @code{JUMP_ALIGN}. This works only if
9001 @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
9004 @defmac LABEL_ALIGN_AFTER_BARRIER (@var{label})
9005 The alignment (log base 2) to put in front of @var{label}, which follows
9008 This macro need not be defined if you don't want any special alignment
9009 to be done at such a time. Most machine descriptions do not currently
9013 @deftypefn {Target Hook} int TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP (rtx @var{label})
9014 The maximum number of bytes to skip before @var{label} when applying
9015 @code{LABEL_ALIGN_AFTER_BARRIER}. This works only if
9016 @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
9019 @defmac LOOP_ALIGN (@var{label})
9020 The alignment (log base 2) to put in front of @var{label}, which follows
9021 a @code{NOTE_INSN_LOOP_BEG} note.
9023 This macro need not be defined if you don't want any special alignment
9024 to be done at such a time. Most machine descriptions do not currently
9027 Unless it's necessary to inspect the @var{label} parameter, it is better
9028 to set the variable @code{align_loops} in the target's
9029 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
9030 selection in @code{align_loops} in a @code{LOOP_ALIGN} implementation.
9033 @deftypefn {Target Hook} int TARGET_ASM_LOOP_ALIGN_MAX_SKIP (rtx @var{label})
9034 The maximum number of bytes to skip when applying @code{LOOP_ALIGN} to
9035 @var{label}. This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is
9039 @defmac LABEL_ALIGN (@var{label})
9040 The alignment (log base 2) to put in front of @var{label}.
9041 If @code{LABEL_ALIGN_AFTER_BARRIER} / @code{LOOP_ALIGN} specify a different alignment,
9042 the maximum of the specified values is used.
9044 Unless it's necessary to inspect the @var{label} parameter, it is better
9045 to set the variable @code{align_labels} in the target's
9046 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
9047 selection in @code{align_labels} in a @code{LABEL_ALIGN} implementation.
9050 @deftypefn {Target Hook} int TARGET_ASM_LABEL_ALIGN_MAX_SKIP (rtx @var{label})
9051 The maximum number of bytes to skip when applying @code{LABEL_ALIGN}
9052 to @var{label}. This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN}
9056 @defmac ASM_OUTPUT_SKIP (@var{stream}, @var{nbytes})
9057 A C statement to output to the stdio stream @var{stream} an assembler
9058 instruction to advance the location counter by @var{nbytes} bytes.
9059 Those bytes should be zero when loaded. @var{nbytes} will be a C
9060 expression of type @code{unsigned HOST_WIDE_INT}.
9063 @defmac ASM_NO_SKIP_IN_TEXT
9064 Define this macro if @code{ASM_OUTPUT_SKIP} should not be used in the
9065 text section because it fails to put zeros in the bytes that are skipped.
9066 This is true on many Unix systems, where the pseudo--op to skip bytes
9067 produces no-op instructions rather than zeros when used in the text
9071 @defmac ASM_OUTPUT_ALIGN (@var{stream}, @var{power})
9072 A C statement to output to the stdio stream @var{stream} an assembler
9073 command to advance the location counter to a multiple of 2 to the
9074 @var{power} bytes. @var{power} will be a C expression of type @code{int}.
9077 @defmac ASM_OUTPUT_ALIGN_WITH_NOP (@var{stream}, @var{power})
9078 Like @code{ASM_OUTPUT_ALIGN}, except that the ``nop'' instruction is used
9079 for padding, if necessary.
9082 @defmac ASM_OUTPUT_MAX_SKIP_ALIGN (@var{stream}, @var{power}, @var{max_skip})
9083 A C statement to output to the stdio stream @var{stream} an assembler
9084 command to advance the location counter to a multiple of 2 to the
9085 @var{power} bytes, but only if @var{max_skip} or fewer bytes are needed to
9086 satisfy the alignment request. @var{power} and @var{max_skip} will be
9087 a C expression of type @code{int}.
9091 @node Debugging Info
9092 @section Controlling Debugging Information Format
9094 @c prevent bad page break with this line
9095 This describes how to specify debugging information.
9098 * All Debuggers:: Macros that affect all debugging formats uniformly.
9099 * DBX Options:: Macros enabling specific options in DBX format.
9100 * DBX Hooks:: Hook macros for varying DBX format.
9101 * File Names and DBX:: Macros controlling output of file names in DBX format.
9102 * SDB and DWARF:: Macros for SDB (COFF) and DWARF formats.
9103 * VMS Debug:: Macros for VMS debug format.
9107 @subsection Macros Affecting All Debugging Formats
9109 @c prevent bad page break with this line
9110 These macros affect all debugging formats.
9112 @defmac DBX_REGISTER_NUMBER (@var{regno})
9113 A C expression that returns the DBX register number for the compiler
9114 register number @var{regno}. In the default macro provided, the value
9115 of this expression will be @var{regno} itself. But sometimes there are
9116 some registers that the compiler knows about and DBX does not, or vice
9117 versa. In such cases, some register may need to have one number in the
9118 compiler and another for DBX@.
9120 If two registers have consecutive numbers inside GCC, and they can be
9121 used as a pair to hold a multiword value, then they @emph{must} have
9122 consecutive numbers after renumbering with @code{DBX_REGISTER_NUMBER}.
9123 Otherwise, debuggers will be unable to access such a pair, because they
9124 expect register pairs to be consecutive in their own numbering scheme.
9126 If you find yourself defining @code{DBX_REGISTER_NUMBER} in way that
9127 does not preserve register pairs, then what you must do instead is
9128 redefine the actual register numbering scheme.
9131 @defmac DEBUGGER_AUTO_OFFSET (@var{x})
9132 A C expression that returns the integer offset value for an automatic
9133 variable having address @var{x} (an RTL expression). The default
9134 computation assumes that @var{x} is based on the frame-pointer and
9135 gives the offset from the frame-pointer. This is required for targets
9136 that produce debugging output for DBX or COFF-style debugging output
9137 for SDB and allow the frame-pointer to be eliminated when the
9138 @option{-g} options is used.
9141 @defmac DEBUGGER_ARG_OFFSET (@var{offset}, @var{x})
9142 A C expression that returns the integer offset value for an argument
9143 having address @var{x} (an RTL expression). The nominal offset is
9147 @defmac PREFERRED_DEBUGGING_TYPE
9148 A C expression that returns the type of debugging output GCC should
9149 produce when the user specifies just @option{-g}. Define
9150 this if you have arranged for GCC to support more than one format of
9151 debugging output. Currently, the allowable values are @code{DBX_DEBUG},
9152 @code{SDB_DEBUG}, @code{DWARF_DEBUG}, @code{DWARF2_DEBUG},
9153 @code{XCOFF_DEBUG}, @code{VMS_DEBUG}, and @code{VMS_AND_DWARF2_DEBUG}.
9155 When the user specifies @option{-ggdb}, GCC normally also uses the
9156 value of this macro to select the debugging output format, but with two
9157 exceptions. If @code{DWARF2_DEBUGGING_INFO} is defined, GCC uses the
9158 value @code{DWARF2_DEBUG}. Otherwise, if @code{DBX_DEBUGGING_INFO} is
9159 defined, GCC uses @code{DBX_DEBUG}.
9161 The value of this macro only affects the default debugging output; the
9162 user can always get a specific type of output by using @option{-gstabs},
9163 @option{-gcoff}, @option{-gdwarf-2}, @option{-gxcoff}, or @option{-gvms}.
9167 @subsection Specific Options for DBX Output
9169 @c prevent bad page break with this line
9170 These are specific options for DBX output.
9172 @defmac DBX_DEBUGGING_INFO
9173 Define this macro if GCC should produce debugging output for DBX
9174 in response to the @option{-g} option.
9177 @defmac XCOFF_DEBUGGING_INFO
9178 Define this macro if GCC should produce XCOFF format debugging output
9179 in response to the @option{-g} option. This is a variant of DBX format.
9182 @defmac DEFAULT_GDB_EXTENSIONS
9183 Define this macro to control whether GCC should by default generate
9184 GDB's extended version of DBX debugging information (assuming DBX-format
9185 debugging information is enabled at all). If you don't define the
9186 macro, the default is 1: always generate the extended information
9187 if there is any occasion to.
9190 @defmac DEBUG_SYMS_TEXT
9191 Define this macro if all @code{.stabs} commands should be output while
9192 in the text section.
9195 @defmac ASM_STABS_OP
9196 A C string constant, including spacing, naming the assembler pseudo op to
9197 use instead of @code{"\t.stabs\t"} to define an ordinary debugging symbol.
9198 If you don't define this macro, @code{"\t.stabs\t"} is used. This macro
9199 applies only to DBX debugging information format.
9202 @defmac ASM_STABD_OP
9203 A C string constant, including spacing, naming the assembler pseudo op to
9204 use instead of @code{"\t.stabd\t"} to define a debugging symbol whose
9205 value is the current location. If you don't define this macro,
9206 @code{"\t.stabd\t"} is used. This macro applies only to DBX debugging
9210 @defmac ASM_STABN_OP
9211 A C string constant, including spacing, naming the assembler pseudo op to
9212 use instead of @code{"\t.stabn\t"} to define a debugging symbol with no
9213 name. If you don't define this macro, @code{"\t.stabn\t"} is used. This
9214 macro applies only to DBX debugging information format.
9217 @defmac DBX_NO_XREFS
9218 Define this macro if DBX on your system does not support the construct
9219 @samp{xs@var{tagname}}. On some systems, this construct is used to
9220 describe a forward reference to a structure named @var{tagname}.
9221 On other systems, this construct is not supported at all.
9224 @defmac DBX_CONTIN_LENGTH
9225 A symbol name in DBX-format debugging information is normally
9226 continued (split into two separate @code{.stabs} directives) when it
9227 exceeds a certain length (by default, 80 characters). On some
9228 operating systems, DBX requires this splitting; on others, splitting
9229 must not be done. You can inhibit splitting by defining this macro
9230 with the value zero. You can override the default splitting-length by
9231 defining this macro as an expression for the length you desire.
9234 @defmac DBX_CONTIN_CHAR
9235 Normally continuation is indicated by adding a @samp{\} character to
9236 the end of a @code{.stabs} string when a continuation follows. To use
9237 a different character instead, define this macro as a character
9238 constant for the character you want to use. Do not define this macro
9239 if backslash is correct for your system.
9242 @defmac DBX_STATIC_STAB_DATA_SECTION
9243 Define this macro if it is necessary to go to the data section before
9244 outputting the @samp{.stabs} pseudo-op for a non-global static
9248 @defmac DBX_TYPE_DECL_STABS_CODE
9249 The value to use in the ``code'' field of the @code{.stabs} directive
9250 for a typedef. The default is @code{N_LSYM}.
9253 @defmac DBX_STATIC_CONST_VAR_CODE
9254 The value to use in the ``code'' field of the @code{.stabs} directive
9255 for a static variable located in the text section. DBX format does not
9256 provide any ``right'' way to do this. The default is @code{N_FUN}.
9259 @defmac DBX_REGPARM_STABS_CODE
9260 The value to use in the ``code'' field of the @code{.stabs} directive
9261 for a parameter passed in registers. DBX format does not provide any
9262 ``right'' way to do this. The default is @code{N_RSYM}.
9265 @defmac DBX_REGPARM_STABS_LETTER
9266 The letter to use in DBX symbol data to identify a symbol as a parameter
9267 passed in registers. DBX format does not customarily provide any way to
9268 do this. The default is @code{'P'}.
9271 @defmac DBX_FUNCTION_FIRST
9272 Define this macro if the DBX information for a function and its
9273 arguments should precede the assembler code for the function. Normally,
9274 in DBX format, the debugging information entirely follows the assembler
9278 @defmac DBX_BLOCKS_FUNCTION_RELATIVE
9279 Define this macro, with value 1, if the value of a symbol describing
9280 the scope of a block (@code{N_LBRAC} or @code{N_RBRAC}) should be
9281 relative to the start of the enclosing function. Normally, GCC uses
9282 an absolute address.
9285 @defmac DBX_LINES_FUNCTION_RELATIVE
9286 Define this macro, with value 1, if the value of a symbol indicating
9287 the current line number (@code{N_SLINE}) should be relative to the
9288 start of the enclosing function. Normally, GCC uses an absolute address.
9291 @defmac DBX_USE_BINCL
9292 Define this macro if GCC should generate @code{N_BINCL} and
9293 @code{N_EINCL} stabs for included header files, as on Sun systems. This
9294 macro also directs GCC to output a type number as a pair of a file
9295 number and a type number within the file. Normally, GCC does not
9296 generate @code{N_BINCL} or @code{N_EINCL} stabs, and it outputs a single
9297 number for a type number.
9301 @subsection Open-Ended Hooks for DBX Format
9303 @c prevent bad page break with this line
9304 These are hooks for DBX format.
9306 @defmac DBX_OUTPUT_LBRAC (@var{stream}, @var{name})
9307 Define this macro to say how to output to @var{stream} the debugging
9308 information for the start of a scope level for variable names. The
9309 argument @var{name} is the name of an assembler symbol (for use with
9310 @code{assemble_name}) whose value is the address where the scope begins.
9313 @defmac DBX_OUTPUT_RBRAC (@var{stream}, @var{name})
9314 Like @code{DBX_OUTPUT_LBRAC}, but for the end of a scope level.
9317 @defmac DBX_OUTPUT_NFUN (@var{stream}, @var{lscope_label}, @var{decl})
9318 Define this macro if the target machine requires special handling to
9319 output an @code{N_FUN} entry for the function @var{decl}.
9322 @defmac DBX_OUTPUT_SOURCE_LINE (@var{stream}, @var{line}, @var{counter})
9323 A C statement to output DBX debugging information before code for line
9324 number @var{line} of the current source file to the stdio stream
9325 @var{stream}. @var{counter} is the number of time the macro was
9326 invoked, including the current invocation; it is intended to generate
9327 unique labels in the assembly output.
9329 This macro should not be defined if the default output is correct, or
9330 if it can be made correct by defining @code{DBX_LINES_FUNCTION_RELATIVE}.
9333 @defmac NO_DBX_FUNCTION_END
9334 Some stabs encapsulation formats (in particular ECOFF), cannot handle the
9335 @code{.stabs "",N_FUN,,0,0,Lscope-function-1} gdb dbx extension construct.
9336 On those machines, define this macro to turn this feature off without
9337 disturbing the rest of the gdb extensions.
9340 @defmac NO_DBX_BNSYM_ENSYM
9341 Some assemblers cannot handle the @code{.stabd BNSYM/ENSYM,0,0} gdb dbx
9342 extension construct. On those machines, define this macro to turn this
9343 feature off without disturbing the rest of the gdb extensions.
9346 @node File Names and DBX
9347 @subsection File Names in DBX Format
9349 @c prevent bad page break with this line
9350 This describes file names in DBX format.
9352 @defmac DBX_OUTPUT_MAIN_SOURCE_FILENAME (@var{stream}, @var{name})
9353 A C statement to output DBX debugging information to the stdio stream
9354 @var{stream}, which indicates that file @var{name} is the main source
9355 file---the file specified as the input file for compilation.
9356 This macro is called only once, at the beginning of compilation.
9358 This macro need not be defined if the standard form of output
9359 for DBX debugging information is appropriate.
9361 It may be necessary to refer to a label equal to the beginning of the
9362 text section. You can use @samp{assemble_name (stream, ltext_label_name)}
9363 to do so. If you do this, you must also set the variable
9364 @var{used_ltext_label_name} to @code{true}.
9367 @defmac NO_DBX_MAIN_SOURCE_DIRECTORY
9368 Define this macro, with value 1, if GCC should not emit an indication
9369 of the current directory for compilation and current source language at
9370 the beginning of the file.
9373 @defmac NO_DBX_GCC_MARKER
9374 Define this macro, with value 1, if GCC should not emit an indication
9375 that this object file was compiled by GCC@. The default is to emit
9376 an @code{N_OPT} stab at the beginning of every source file, with
9377 @samp{gcc2_compiled.} for the string and value 0.
9380 @defmac DBX_OUTPUT_MAIN_SOURCE_FILE_END (@var{stream}, @var{name})
9381 A C statement to output DBX debugging information at the end of
9382 compilation of the main source file @var{name}. Output should be
9383 written to the stdio stream @var{stream}.
9385 If you don't define this macro, nothing special is output at the end
9386 of compilation, which is correct for most machines.
9389 @defmac DBX_OUTPUT_NULL_N_SO_AT_MAIN_SOURCE_FILE_END
9390 Define this macro @emph{instead of} defining
9391 @code{DBX_OUTPUT_MAIN_SOURCE_FILE_END}, if what needs to be output at
9392 the end of compilation is an @code{N_SO} stab with an empty string,
9393 whose value is the highest absolute text address in the file.
9398 @subsection Macros for SDB and DWARF Output
9400 @c prevent bad page break with this line
9401 Here are macros for SDB and DWARF output.
9403 @defmac SDB_DEBUGGING_INFO
9404 Define this macro if GCC should produce COFF-style debugging output
9405 for SDB in response to the @option{-g} option.
9408 @defmac DWARF2_DEBUGGING_INFO
9409 Define this macro if GCC should produce dwarf version 2 format
9410 debugging output in response to the @option{-g} option.
9412 @deftypefn {Target Hook} int TARGET_DWARF_CALLING_CONVENTION (const_tree @var{function})
9413 Define this to enable the dwarf attribute @code{DW_AT_calling_convention} to
9414 be emitted for each function. Instead of an integer return the enum
9415 value for the @code{DW_CC_} tag.
9418 To support optional call frame debugging information, you must also
9419 define @code{INCOMING_RETURN_ADDR_RTX} and either set
9420 @code{RTX_FRAME_RELATED_P} on the prologue insns if you use RTL for the
9421 prologue, or call @code{dwarf2out_def_cfa} and @code{dwarf2out_reg_save}
9422 as appropriate from @code{TARGET_ASM_FUNCTION_PROLOGUE} if you don't.
9425 @defmac DWARF2_FRAME_INFO
9426 Define this macro to a nonzero value if GCC should always output
9427 Dwarf 2 frame information. If @code{TARGET_EXCEPT_UNWIND_INFO}
9428 (@pxref{Exception Region Output}) returns @code{UI_DWARF2}, and
9429 exceptions are enabled, GCC will output this information not matter
9430 how you define @code{DWARF2_FRAME_INFO}.
9433 @deftypefn {Target Hook} {enum unwind_info_type} TARGET_DEBUG_UNWIND_INFO (void)
9434 This hook defines the mechanism that will be used for describing frame
9435 unwind information to the debugger. Normally the hook will return
9436 @code{UI_DWARF2} if DWARF 2 debug information is enabled, and
9437 return @code{UI_NONE} otherwise.
9439 A target may return @code{UI_DWARF2} even when DWARF 2 debug information
9440 is disabled in order to always output DWARF 2 frame information.
9442 A target may return @code{UI_TARGET} if it has ABI specified unwind tables.
9443 This will suppress generation of the normal debug frame unwind information.
9446 @defmac DWARF2_ASM_LINE_DEBUG_INFO
9447 Define this macro to be a nonzero value if the assembler can generate Dwarf 2
9448 line debug info sections. This will result in much more compact line number
9449 tables, and hence is desirable if it works.
9452 @deftypevr {Target Hook} bool TARGET_WANT_DEBUG_PUB_SECTIONS
9453 True if the @code{.debug_pubtypes} and @code{.debug_pubnames} sections should be emitted. These sections are not used on most platforms, and in particular GDB does not use them.
9456 @deftypevr {Target Hook} bool TARGET_DELAY_SCHED2
9457 True if sched2 is not to be run at its normal place. This usually means it will be run as part of machine-specific reorg.
9460 @deftypevr {Target Hook} bool TARGET_DELAY_VARTRACK
9461 True if vartrack is not to be run at its normal place. This usually means it will be run as part of machine-specific reorg.
9464 @defmac ASM_OUTPUT_DWARF_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
9465 A C statement to issue assembly directives that create a difference
9466 @var{lab1} minus @var{lab2}, using an integer of the given @var{size}.
9469 @defmac ASM_OUTPUT_DWARF_VMS_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
9470 A C statement to issue assembly directives that create a difference
9471 between the two given labels in system defined units, e.g. instruction
9472 slots on IA64 VMS, using an integer of the given size.
9475 @defmac ASM_OUTPUT_DWARF_OFFSET (@var{stream}, @var{size}, @var{label}, @var{section})
9476 A C statement to issue assembly directives that create a
9477 section-relative reference to the given @var{label}, using an integer of the
9478 given @var{size}. The label is known to be defined in the given @var{section}.
9481 @defmac ASM_OUTPUT_DWARF_PCREL (@var{stream}, @var{size}, @var{label})
9482 A C statement to issue assembly directives that create a self-relative
9483 reference to the given @var{label}, using an integer of the given @var{size}.
9486 @defmac ASM_OUTPUT_DWARF_TABLE_REF (@var{label})
9487 A C statement to issue assembly directives that create a reference to
9488 the DWARF table identifier @var{label} from the current section. This
9489 is used on some systems to avoid garbage collecting a DWARF table which
9490 is referenced by a function.
9493 @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_DWARF_DTPREL (FILE *@var{file}, int @var{size}, rtx @var{x})
9494 If defined, this target hook is a function which outputs a DTP-relative
9495 reference to the given TLS symbol of the specified size.
9498 @defmac PUT_SDB_@dots{}
9499 Define these macros to override the assembler syntax for the special
9500 SDB assembler directives. See @file{sdbout.c} for a list of these
9501 macros and their arguments. If the standard syntax is used, you need
9502 not define them yourself.
9506 Some assemblers do not support a semicolon as a delimiter, even between
9507 SDB assembler directives. In that case, define this macro to be the
9508 delimiter to use (usually @samp{\n}). It is not necessary to define
9509 a new set of @code{PUT_SDB_@var{op}} macros if this is the only change
9513 @defmac SDB_ALLOW_UNKNOWN_REFERENCES
9514 Define this macro to allow references to unknown structure,
9515 union, or enumeration tags to be emitted. Standard COFF does not
9516 allow handling of unknown references, MIPS ECOFF has support for
9520 @defmac SDB_ALLOW_FORWARD_REFERENCES
9521 Define this macro to allow references to structure, union, or
9522 enumeration tags that have not yet been seen to be handled. Some
9523 assemblers choke if forward tags are used, while some require it.
9526 @defmac SDB_OUTPUT_SOURCE_LINE (@var{stream}, @var{line})
9527 A C statement to output SDB debugging information before code for line
9528 number @var{line} of the current source file to the stdio stream
9529 @var{stream}. The default is to emit an @code{.ln} directive.
9534 @subsection Macros for VMS Debug Format
9536 @c prevent bad page break with this line
9537 Here are macros for VMS debug format.
9539 @defmac VMS_DEBUGGING_INFO
9540 Define this macro if GCC should produce debugging output for VMS
9541 in response to the @option{-g} option. The default behavior for VMS
9542 is to generate minimal debug info for a traceback in the absence of
9543 @option{-g} unless explicitly overridden with @option{-g0}. This
9544 behavior is controlled by @code{TARGET_OPTION_OPTIMIZATION} and
9545 @code{TARGET_OPTION_OVERRIDE}.
9548 @node Floating Point
9549 @section Cross Compilation and Floating Point
9550 @cindex cross compilation and floating point
9551 @cindex floating point and cross compilation
9553 While all modern machines use twos-complement representation for integers,
9554 there are a variety of representations for floating point numbers. This
9555 means that in a cross-compiler the representation of floating point numbers
9556 in the compiled program may be different from that used in the machine
9557 doing the compilation.
9559 Because different representation systems may offer different amounts of
9560 range and precision, all floating point constants must be represented in
9561 the target machine's format. Therefore, the cross compiler cannot
9562 safely use the host machine's floating point arithmetic; it must emulate
9563 the target's arithmetic. To ensure consistency, GCC always uses
9564 emulation to work with floating point values, even when the host and
9565 target floating point formats are identical.
9567 The following macros are provided by @file{real.h} for the compiler to
9568 use. All parts of the compiler which generate or optimize
9569 floating-point calculations must use these macros. They may evaluate
9570 their operands more than once, so operands must not have side effects.
9572 @defmac REAL_VALUE_TYPE
9573 The C data type to be used to hold a floating point value in the target
9574 machine's format. Typically this is a @code{struct} containing an
9575 array of @code{HOST_WIDE_INT}, but all code should treat it as an opaque
9579 @deftypefn Macro int REAL_VALUES_EQUAL (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
9580 Compares for equality the two values, @var{x} and @var{y}. If the target
9581 floating point format supports negative zeroes and/or NaNs,
9582 @samp{REAL_VALUES_EQUAL (-0.0, 0.0)} is true, and
9583 @samp{REAL_VALUES_EQUAL (NaN, NaN)} is false.
9586 @deftypefn Macro int REAL_VALUES_LESS (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
9587 Tests whether @var{x} is less than @var{y}.
9590 @deftypefn Macro HOST_WIDE_INT REAL_VALUE_FIX (REAL_VALUE_TYPE @var{x})
9591 Truncates @var{x} to a signed integer, rounding toward zero.
9594 @deftypefn Macro {unsigned HOST_WIDE_INT} REAL_VALUE_UNSIGNED_FIX (REAL_VALUE_TYPE @var{x})
9595 Truncates @var{x} to an unsigned integer, rounding toward zero. If
9596 @var{x} is negative, returns zero.
9599 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ATOF (const char *@var{string}, enum machine_mode @var{mode})
9600 Converts @var{string} into a floating point number in the target machine's
9601 representation for mode @var{mode}. This routine can handle both
9602 decimal and hexadecimal floating point constants, using the syntax
9603 defined by the C language for both.
9606 @deftypefn Macro int REAL_VALUE_NEGATIVE (REAL_VALUE_TYPE @var{x})
9607 Returns 1 if @var{x} is negative (including negative zero), 0 otherwise.
9610 @deftypefn Macro int REAL_VALUE_ISINF (REAL_VALUE_TYPE @var{x})
9611 Determines whether @var{x} represents infinity (positive or negative).
9614 @deftypefn Macro int REAL_VALUE_ISNAN (REAL_VALUE_TYPE @var{x})
9615 Determines whether @var{x} represents a ``NaN'' (not-a-number).
9618 @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})
9619 Calculates an arithmetic operation on the two floating point values
9620 @var{x} and @var{y}, storing the result in @var{output} (which must be a
9623 The operation to be performed is specified by @var{code}. Only the
9624 following codes are supported: @code{PLUS_EXPR}, @code{MINUS_EXPR},
9625 @code{MULT_EXPR}, @code{RDIV_EXPR}, @code{MAX_EXPR}, @code{MIN_EXPR}.
9627 If @code{REAL_ARITHMETIC} is asked to evaluate division by zero and the
9628 target's floating point format cannot represent infinity, it will call
9629 @code{abort}. Callers should check for this situation first, using
9630 @code{MODE_HAS_INFINITIES}. @xref{Storage Layout}.
9633 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_NEGATE (REAL_VALUE_TYPE @var{x})
9634 Returns the negative of the floating point value @var{x}.
9637 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ABS (REAL_VALUE_TYPE @var{x})
9638 Returns the absolute value of @var{x}.
9641 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_TRUNCATE (REAL_VALUE_TYPE @var{mode}, enum machine_mode @var{x})
9642 Truncates the floating point value @var{x} to fit in @var{mode}. The
9643 return value is still a full-size @code{REAL_VALUE_TYPE}, but it has an
9644 appropriate bit pattern to be output as a floating constant whose
9645 precision accords with mode @var{mode}.
9648 @deftypefn Macro void REAL_VALUE_TO_INT (HOST_WIDE_INT @var{low}, HOST_WIDE_INT @var{high}, REAL_VALUE_TYPE @var{x})
9649 Converts a floating point value @var{x} into a double-precision integer
9650 which is then stored into @var{low} and @var{high}. If the value is not
9651 integral, it is truncated.
9654 @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})
9655 Converts a double-precision integer found in @var{low} and @var{high},
9656 into a floating point value which is then stored into @var{x}. The
9657 value is truncated to fit in mode @var{mode}.
9660 @node Mode Switching
9661 @section Mode Switching Instructions
9662 @cindex mode switching
9663 The following macros control mode switching optimizations:
9665 @defmac OPTIMIZE_MODE_SWITCHING (@var{entity})
9666 Define this macro if the port needs extra instructions inserted for mode
9667 switching in an optimizing compilation.
9669 For an example, the SH4 can perform both single and double precision
9670 floating point operations, but to perform a single precision operation,
9671 the FPSCR PR bit has to be cleared, while for a double precision
9672 operation, this bit has to be set. Changing the PR bit requires a general
9673 purpose register as a scratch register, hence these FPSCR sets have to
9674 be inserted before reload, i.e.@: you can't put this into instruction emitting
9675 or @code{TARGET_MACHINE_DEPENDENT_REORG}.
9677 You can have multiple entities that are mode-switched, and select at run time
9678 which entities actually need it. @code{OPTIMIZE_MODE_SWITCHING} should
9679 return nonzero for any @var{entity} that needs mode-switching.
9680 If you define this macro, you also have to define
9681 @code{NUM_MODES_FOR_MODE_SWITCHING}, @code{MODE_NEEDED},
9682 @code{MODE_PRIORITY_TO_MODE} and @code{EMIT_MODE_SET}.
9683 @code{MODE_AFTER}, @code{MODE_ENTRY}, and @code{MODE_EXIT}
9687 @defmac NUM_MODES_FOR_MODE_SWITCHING
9688 If you define @code{OPTIMIZE_MODE_SWITCHING}, you have to define this as
9689 initializer for an array of integers. Each initializer element
9690 N refers to an entity that needs mode switching, and specifies the number
9691 of different modes that might need to be set for this entity.
9692 The position of the initializer in the initializer---starting counting at
9693 zero---determines the integer that is used to refer to the mode-switched
9695 In macros that take mode arguments / yield a mode result, modes are
9696 represented as numbers 0 @dots{} N @minus{} 1. N is used to specify that no mode
9697 switch is needed / supplied.
9700 @defmac MODE_NEEDED (@var{entity}, @var{insn})
9701 @var{entity} is an integer specifying a mode-switched entity. If
9702 @code{OPTIMIZE_MODE_SWITCHING} is defined, you must define this macro to
9703 return an integer value not larger than the corresponding element in
9704 @code{NUM_MODES_FOR_MODE_SWITCHING}, to denote the mode that @var{entity} must
9705 be switched into prior to the execution of @var{insn}.
9708 @defmac MODE_AFTER (@var{mode}, @var{insn})
9709 If this macro is defined, it is evaluated for every @var{insn} during
9710 mode switching. It determines the mode that an insn results in (if
9711 different from the incoming mode).
9714 @defmac MODE_ENTRY (@var{entity})
9715 If this macro is defined, it is evaluated for every @var{entity} that needs
9716 mode switching. It should evaluate to an integer, which is a mode that
9717 @var{entity} is assumed to be switched to at function entry. If @code{MODE_ENTRY}
9718 is defined then @code{MODE_EXIT} must be defined.
9721 @defmac MODE_EXIT (@var{entity})
9722 If this macro is defined, it is evaluated for every @var{entity} that needs
9723 mode switching. It should evaluate to an integer, which is a mode that
9724 @var{entity} is assumed to be switched to at function exit. If @code{MODE_EXIT}
9725 is defined then @code{MODE_ENTRY} must be defined.
9728 @defmac MODE_PRIORITY_TO_MODE (@var{entity}, @var{n})
9729 This macro specifies the order in which modes for @var{entity} are processed.
9730 0 is the highest priority, @code{NUM_MODES_FOR_MODE_SWITCHING[@var{entity}] - 1} the
9731 lowest. The value of the macro should be an integer designating a mode
9732 for @var{entity}. For any fixed @var{entity}, @code{mode_priority_to_mode}
9733 (@var{entity}, @var{n}) shall be a bijection in 0 @dots{}
9734 @code{num_modes_for_mode_switching[@var{entity}] - 1}.
9737 @defmac EMIT_MODE_SET (@var{entity}, @var{mode}, @var{hard_regs_live})
9738 Generate one or more insns to set @var{entity} to @var{mode}.
9739 @var{hard_reg_live} is the set of hard registers live at the point where
9740 the insn(s) are to be inserted.
9743 @node Target Attributes
9744 @section Defining target-specific uses of @code{__attribute__}
9745 @cindex target attributes
9746 @cindex machine attributes
9747 @cindex attributes, target-specific
9749 Target-specific attributes may be defined for functions, data and types.
9750 These are described using the following target hooks; they also need to
9751 be documented in @file{extend.texi}.
9753 @deftypevr {Target Hook} {const struct attribute_spec *} TARGET_ATTRIBUTE_TABLE
9754 If defined, this target hook points to an array of @samp{struct
9755 attribute_spec} (defined in @file{tree.h}) specifying the machine
9756 specific attributes for this target and some of the restrictions on the
9757 entities to which these attributes are applied and the arguments they
9761 @deftypefn {Target Hook} bool TARGET_ATTRIBUTE_TAKES_IDENTIFIER_P (const_tree @var{name})
9762 If defined, this target hook is a function which returns true if the
9763 machine-specific attribute named @var{name} expects an identifier
9764 given as its first argument to be passed on as a plain identifier, not
9765 subjected to name lookup. If this is not defined, the default is
9766 false for all machine-specific attributes.
9769 @deftypefn {Target Hook} int TARGET_COMP_TYPE_ATTRIBUTES (const_tree @var{type1}, const_tree @var{type2})
9770 If defined, this target hook is a function which returns zero if the attributes on
9771 @var{type1} and @var{type2} are incompatible, one if they are compatible,
9772 and two if they are nearly compatible (which causes a warning to be
9773 generated). If this is not defined, machine-specific attributes are
9774 supposed always to be compatible.
9777 @deftypefn {Target Hook} void TARGET_SET_DEFAULT_TYPE_ATTRIBUTES (tree @var{type})
9778 If defined, this target hook is a function which assigns default attributes to
9779 the newly defined @var{type}.
9782 @deftypefn {Target Hook} tree TARGET_MERGE_TYPE_ATTRIBUTES (tree @var{type1}, tree @var{type2})
9783 Define this target hook if the merging of type attributes needs special
9784 handling. If defined, the result is a list of the combined
9785 @code{TYPE_ATTRIBUTES} of @var{type1} and @var{type2}. It is assumed
9786 that @code{comptypes} has already been called and returned 1. This
9787 function may call @code{merge_attributes} to handle machine-independent
9791 @deftypefn {Target Hook} tree TARGET_MERGE_DECL_ATTRIBUTES (tree @var{olddecl}, tree @var{newdecl})
9792 Define this target hook if the merging of decl attributes needs special
9793 handling. If defined, the result is a list of the combined
9794 @code{DECL_ATTRIBUTES} of @var{olddecl} and @var{newdecl}.
9795 @var{newdecl} is a duplicate declaration of @var{olddecl}. Examples of
9796 when this is needed are when one attribute overrides another, or when an
9797 attribute is nullified by a subsequent definition. This function may
9798 call @code{merge_attributes} to handle machine-independent merging.
9800 @findex TARGET_DLLIMPORT_DECL_ATTRIBUTES
9801 If the only target-specific handling you require is @samp{dllimport}
9802 for Microsoft Windows targets, you should define the macro
9803 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES} to @code{1}. The compiler
9804 will then define a function called
9805 @code{merge_dllimport_decl_attributes} which can then be defined as
9806 the expansion of @code{TARGET_MERGE_DECL_ATTRIBUTES}. You can also
9807 add @code{handle_dll_attribute} in the attribute table for your port
9808 to perform initial processing of the @samp{dllimport} and
9809 @samp{dllexport} attributes. This is done in @file{i386/cygwin.h} and
9810 @file{i386/i386.c}, for example.
9813 @deftypefn {Target Hook} bool TARGET_VALID_DLLIMPORT_ATTRIBUTE_P (const_tree @var{decl})
9814 @var{decl} is a variable or function with @code{__attribute__((dllimport))} specified. Use this hook if the target needs to add extra validation checks to @code{handle_dll_attribute}.
9817 @defmac TARGET_DECLSPEC
9818 Define this macro to a nonzero value if you want to treat
9819 @code{__declspec(X)} as equivalent to @code{__attribute((X))}. By
9820 default, this behavior is enabled only for targets that define
9821 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES}. The current implementation
9822 of @code{__declspec} is via a built-in macro, but you should not rely
9823 on this implementation detail.
9826 @deftypefn {Target Hook} void TARGET_INSERT_ATTRIBUTES (tree @var{node}, tree *@var{attr_ptr})
9827 Define this target hook if you want to be able to add attributes to a decl
9828 when it is being created. This is normally useful for back ends which
9829 wish to implement a pragma by using the attributes which correspond to
9830 the pragma's effect. The @var{node} argument is the decl which is being
9831 created. The @var{attr_ptr} argument is a pointer to the attribute list
9832 for this decl. The list itself should not be modified, since it may be
9833 shared with other decls, but attributes may be chained on the head of
9834 the list and @code{*@var{attr_ptr}} modified to point to the new
9835 attributes, or a copy of the list may be made if further changes are
9839 @deftypefn {Target Hook} bool TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P (const_tree @var{fndecl})
9841 This target hook returns @code{true} if it is ok to inline @var{fndecl}
9842 into the current function, despite its having target-specific
9843 attributes, @code{false} otherwise. By default, if a function has a
9844 target specific attribute attached to it, it will not be inlined.
9847 @deftypefn {Target Hook} bool TARGET_OPTION_VALID_ATTRIBUTE_P (tree @var{fndecl}, tree @var{name}, tree @var{args}, int @var{flags})
9848 This hook is called to parse the @code{attribute(option("..."))}, and
9849 it allows the function to set different target machine compile time
9850 options for the current function that might be different than the
9851 options specified on the command line. The hook should return
9852 @code{true} if the options are valid.
9854 The hook should set the @var{DECL_FUNCTION_SPECIFIC_TARGET} field in
9855 the function declaration to hold a pointer to a target specific
9856 @var{struct cl_target_option} structure.
9859 @deftypefn {Target Hook} void TARGET_OPTION_SAVE (struct cl_target_option *@var{ptr})
9860 This hook is called to save any additional target specific information
9861 in the @var{struct cl_target_option} structure for function specific
9863 @xref{Option file format}.
9866 @deftypefn {Target Hook} void TARGET_OPTION_RESTORE (struct cl_target_option *@var{ptr})
9867 This hook is called to restore any additional target specific
9868 information in the @var{struct cl_target_option} structure for
9869 function specific options.
9872 @deftypefn {Target Hook} void TARGET_OPTION_PRINT (FILE *@var{file}, int @var{indent}, struct cl_target_option *@var{ptr})
9873 This hook is called to print any additional target specific
9874 information in the @var{struct cl_target_option} structure for
9875 function specific options.
9878 @deftypefn {Target Hook} bool TARGET_OPTION_PRAGMA_PARSE (tree @var{args}, tree @var{pop_target})
9879 This target hook parses the options for @code{#pragma GCC option} to
9880 set the machine specific options for functions that occur later in the
9881 input stream. The options should be the same as handled by the
9882 @code{TARGET_OPTION_VALID_ATTRIBUTE_P} hook.
9885 @deftypefn {Target Hook} void TARGET_OPTION_OVERRIDE (void)
9886 Sometimes certain combinations of command options do not make sense on
9887 a particular target machine. You can override the hook
9888 @code{TARGET_OPTION_OVERRIDE} to take account of this. This hooks is called
9889 once just after all the command options have been parsed.
9891 Don't use this hook to turn on various extra optimizations for
9892 @option{-O}. That is what @code{TARGET_OPTION_OPTIMIZATION} is for.
9894 If you need to do something whenever the optimization level is
9895 changed via the optimize attribute or pragma, see
9896 @code{TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE}
9899 @deftypefn {Target Hook} bool TARGET_CAN_INLINE_P (tree @var{caller}, tree @var{callee})
9900 This target hook returns @code{false} if the @var{caller} function
9901 cannot inline @var{callee}, based on target specific information. By
9902 default, inlining is not allowed if the callee function has function
9903 specific target options and the caller does not use the same options.
9907 @section Emulating TLS
9908 @cindex Emulated TLS
9910 For targets whose psABI does not provide Thread Local Storage via
9911 specific relocations and instruction sequences, an emulation layer is
9912 used. A set of target hooks allows this emulation layer to be
9913 configured for the requirements of a particular target. For instance
9914 the psABI may in fact specify TLS support in terms of an emulation
9917 The emulation layer works by creating a control object for every TLS
9918 object. To access the TLS object, a lookup function is provided
9919 which, when given the address of the control object, will return the
9920 address of the current thread's instance of the TLS object.
9922 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_GET_ADDRESS
9923 Contains the name of the helper function that uses a TLS control
9924 object to locate a TLS instance. The default causes libgcc's
9925 emulated TLS helper function to be used.
9928 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_REGISTER_COMMON
9929 Contains the name of the helper function that should be used at
9930 program startup to register TLS objects that are implicitly
9931 initialized to zero. If this is @code{NULL}, all TLS objects will
9932 have explicit initializers. The default causes libgcc's emulated TLS
9933 registration function to be used.
9936 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_VAR_SECTION
9937 Contains the name of the section in which TLS control variables should
9938 be placed. The default of @code{NULL} allows these to be placed in
9942 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_TMPL_SECTION
9943 Contains the name of the section in which TLS initializers should be
9944 placed. The default of @code{NULL} allows these to be placed in any
9948 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_VAR_PREFIX
9949 Contains the prefix to be prepended to TLS control variable names.
9950 The default of @code{NULL} uses a target-specific prefix.
9953 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_TMPL_PREFIX
9954 Contains the prefix to be prepended to TLS initializer objects. The
9955 default of @code{NULL} uses a target-specific prefix.
9958 @deftypefn {Target Hook} tree TARGET_EMUTLS_VAR_FIELDS (tree @var{type}, tree *@var{name})
9959 Specifies a function that generates the FIELD_DECLs for a TLS control
9960 object type. @var{type} is the RECORD_TYPE the fields are for and
9961 @var{name} should be filled with the structure tag, if the default of
9962 @code{__emutls_object} is unsuitable. The default creates a type suitable
9963 for libgcc's emulated TLS function.
9966 @deftypefn {Target Hook} tree TARGET_EMUTLS_VAR_INIT (tree @var{var}, tree @var{decl}, tree @var{tmpl_addr})
9967 Specifies a function that generates the CONSTRUCTOR to initialize a
9968 TLS control object. @var{var} is the TLS control object, @var{decl}
9969 is the TLS object and @var{tmpl_addr} is the address of the
9970 initializer. The default initializes libgcc's emulated TLS control object.
9973 @deftypevr {Target Hook} bool TARGET_EMUTLS_VAR_ALIGN_FIXED
9974 Specifies whether the alignment of TLS control variable objects is
9975 fixed and should not be increased as some backends may do to optimize
9976 single objects. The default is false.
9979 @deftypevr {Target Hook} bool TARGET_EMUTLS_DEBUG_FORM_TLS_ADDRESS
9980 Specifies whether a DWARF @code{DW_OP_form_tls_address} location descriptor
9981 may be used to describe emulated TLS control objects.
9984 @node MIPS Coprocessors
9985 @section Defining coprocessor specifics for MIPS targets.
9986 @cindex MIPS coprocessor-definition macros
9988 The MIPS specification allows MIPS implementations to have as many as 4
9989 coprocessors, each with as many as 32 private registers. GCC supports
9990 accessing these registers and transferring values between the registers
9991 and memory using asm-ized variables. For example:
9994 register unsigned int cp0count asm ("c0r1");
10000 (``c0r1'' is the default name of register 1 in coprocessor 0; alternate
10001 names may be added as described below, or the default names may be
10002 overridden entirely in @code{SUBTARGET_CONDITIONAL_REGISTER_USAGE}.)
10004 Coprocessor registers are assumed to be epilogue-used; sets to them will
10005 be preserved even if it does not appear that the register is used again
10006 later in the function.
10008 Another note: according to the MIPS spec, coprocessor 1 (if present) is
10009 the FPU@. One accesses COP1 registers through standard mips
10010 floating-point support; they are not included in this mechanism.
10012 There is one macro used in defining the MIPS coprocessor interface which
10013 you may want to override in subtargets; it is described below.
10015 @defmac ALL_COP_ADDITIONAL_REGISTER_NAMES
10016 A comma-separated list (with leading comma) of pairs describing the
10017 alternate names of coprocessor registers. The format of each entry should be
10019 @{ @var{alternatename}, @var{register_number}@}
10025 @section Parameters for Precompiled Header Validity Checking
10026 @cindex parameters, precompiled headers
10028 @deftypefn {Target Hook} {void *} TARGET_GET_PCH_VALIDITY (size_t *@var{sz})
10029 This hook returns a pointer to the data needed by
10030 @code{TARGET_PCH_VALID_P} and sets
10031 @samp{*@var{sz}} to the size of the data in bytes.
10034 @deftypefn {Target Hook} {const char *} TARGET_PCH_VALID_P (const void *@var{data}, size_t @var{sz})
10035 This hook checks whether the options used to create a PCH file are
10036 compatible with the current settings. It returns @code{NULL}
10037 if so and a suitable error message if not. Error messages will
10038 be presented to the user and must be localized using @samp{_(@var{msg})}.
10040 @var{data} is the data that was returned by @code{TARGET_GET_PCH_VALIDITY}
10041 when the PCH file was created and @var{sz} is the size of that data in bytes.
10042 It's safe to assume that the data was created by the same version of the
10043 compiler, so no format checking is needed.
10045 The default definition of @code{default_pch_valid_p} should be
10046 suitable for most targets.
10049 @deftypefn {Target Hook} {const char *} TARGET_CHECK_PCH_TARGET_FLAGS (int @var{pch_flags})
10050 If this hook is nonnull, the default implementation of
10051 @code{TARGET_PCH_VALID_P} will use it to check for compatible values
10052 of @code{target_flags}. @var{pch_flags} specifies the value that
10053 @code{target_flags} had when the PCH file was created. The return
10054 value is the same as for @code{TARGET_PCH_VALID_P}.
10058 @section C++ ABI parameters
10059 @cindex parameters, c++ abi
10061 @deftypefn {Target Hook} tree TARGET_CXX_GUARD_TYPE (void)
10062 Define this hook to override the integer type used for guard variables.
10063 These are used to implement one-time construction of static objects. The
10064 default is long_long_integer_type_node.
10067 @deftypefn {Target Hook} bool TARGET_CXX_GUARD_MASK_BIT (void)
10068 This hook determines how guard variables are used. It should return
10069 @code{false} (the default) if the first byte should be used. A return value of
10070 @code{true} indicates that only the least significant bit should be used.
10073 @deftypefn {Target Hook} tree TARGET_CXX_GET_COOKIE_SIZE (tree @var{type})
10074 This hook returns the size of the cookie to use when allocating an array
10075 whose elements have the indicated @var{type}. Assumes that it is already
10076 known that a cookie is needed. The default is
10077 @code{max(sizeof (size_t), alignof(type))}, as defined in section 2.7 of the
10078 IA64/Generic C++ ABI@.
10081 @deftypefn {Target Hook} bool TARGET_CXX_COOKIE_HAS_SIZE (void)
10082 This hook should return @code{true} if the element size should be stored in
10083 array cookies. The default is to return @code{false}.
10086 @deftypefn {Target Hook} int TARGET_CXX_IMPORT_EXPORT_CLASS (tree @var{type}, int @var{import_export})
10087 If defined by a backend this hook allows the decision made to export
10088 class @var{type} to be overruled. Upon entry @var{import_export}
10089 will contain 1 if the class is going to be exported, @minus{}1 if it is going
10090 to be imported and 0 otherwise. This function should return the
10091 modified value and perform any other actions necessary to support the
10092 backend's targeted operating system.
10095 @deftypefn {Target Hook} bool TARGET_CXX_CDTOR_RETURNS_THIS (void)
10096 This hook should return @code{true} if constructors and destructors return
10097 the address of the object created/destroyed. The default is to return
10101 @deftypefn {Target Hook} bool TARGET_CXX_KEY_METHOD_MAY_BE_INLINE (void)
10102 This hook returns true if the key method for a class (i.e., the method
10103 which, if defined in the current translation unit, causes the virtual
10104 table to be emitted) may be an inline function. Under the standard
10105 Itanium C++ ABI the key method may be an inline function so long as
10106 the function is not declared inline in the class definition. Under
10107 some variants of the ABI, an inline function can never be the key
10108 method. The default is to return @code{true}.
10111 @deftypefn {Target Hook} void TARGET_CXX_DETERMINE_CLASS_DATA_VISIBILITY (tree @var{decl})
10112 @var{decl} is a virtual table, virtual table table, typeinfo object, or other similar implicit class data object that will be emitted with external linkage in this translation unit. No ELF visibility has been explicitly specified. If the target needs to specify a visibility other than that of the containing class, use this hook to set @code{DECL_VISIBILITY} and @code{DECL_VISIBILITY_SPECIFIED}.
10115 @deftypefn {Target Hook} bool TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT (void)
10116 This hook returns true (the default) if virtual tables and other
10117 similar implicit class data objects are always COMDAT if they have
10118 external linkage. If this hook returns false, then class data for
10119 classes whose virtual table will be emitted in only one translation
10120 unit will not be COMDAT.
10123 @deftypefn {Target Hook} bool TARGET_CXX_LIBRARY_RTTI_COMDAT (void)
10124 This hook returns true (the default) if the RTTI information for
10125 the basic types which is defined in the C++ runtime should always
10126 be COMDAT, false if it should not be COMDAT.
10129 @deftypefn {Target Hook} bool TARGET_CXX_USE_AEABI_ATEXIT (void)
10130 This hook returns true if @code{__aeabi_atexit} (as defined by the ARM EABI)
10131 should be used to register static destructors when @option{-fuse-cxa-atexit}
10132 is in effect. The default is to return false to use @code{__cxa_atexit}.
10135 @deftypefn {Target Hook} bool TARGET_CXX_USE_ATEXIT_FOR_CXA_ATEXIT (void)
10136 This hook returns true if the target @code{atexit} function can be used
10137 in the same manner as @code{__cxa_atexit} to register C++ static
10138 destructors. This requires that @code{atexit}-registered functions in
10139 shared libraries are run in the correct order when the libraries are
10140 unloaded. The default is to return false.
10143 @deftypefn {Target Hook} void TARGET_CXX_ADJUST_CLASS_AT_DEFINITION (tree @var{type})
10144 @var{type} is a C++ class (i.e., RECORD_TYPE or UNION_TYPE) that has just been defined. Use this hook to make adjustments to the class (eg, tweak visibility or perform any other required target modifications).
10147 @deftypefn {Target Hook} tree TARGET_CXX_DECL_MANGLING_CONTEXT (const_tree @var{decl})
10148 Return target-specific mangling context of @var{decl} or @code{NULL_TREE}.
10151 @node Named Address Spaces
10152 @section Adding support for named address spaces
10153 @cindex named address spaces
10155 The draft technical report of the ISO/IEC JTC1 S22 WG14 N1275
10156 standards committee, @cite{Programming Languages - C - Extensions to
10157 support embedded processors}, specifies a syntax for embedded
10158 processors to specify alternate address spaces. You can configure a
10159 GCC port to support section 5.1 of the draft report to add support for
10160 address spaces other than the default address space. These address
10161 spaces are new keywords that are similar to the @code{volatile} and
10162 @code{const} type attributes.
10164 Pointers to named address spaces can have a different size than
10165 pointers to the generic address space.
10167 For example, the SPU port uses the @code{__ea} address space to refer
10168 to memory in the host processor, rather than memory local to the SPU
10169 processor. Access to memory in the @code{__ea} address space involves
10170 issuing DMA operations to move data between the host processor and the
10171 local processor memory address space. Pointers in the @code{__ea}
10172 address space are either 32 bits or 64 bits based on the
10173 @option{-mea32} or @option{-mea64} switches (native SPU pointers are
10176 Internally, address spaces are represented as a small integer in the
10177 range 0 to 15 with address space 0 being reserved for the generic
10180 To register a named address space qualifier keyword with the C front end,
10181 the target may call the @code{c_register_addr_space} routine. For example,
10182 the SPU port uses the following to declare @code{__ea} as the keyword for
10183 named address space #1:
10185 #define ADDR_SPACE_EA 1
10186 c_register_addr_space ("__ea", ADDR_SPACE_EA);
10189 @deftypefn {Target Hook} {enum machine_mode} TARGET_ADDR_SPACE_POINTER_MODE (addr_space_t @var{address_space})
10190 Define this to return the machine mode to use for pointers to
10191 @var{address_space} if the target supports named address spaces.
10192 The default version of this hook returns @code{ptr_mode} for the
10193 generic address space only.
10196 @deftypefn {Target Hook} {enum machine_mode} TARGET_ADDR_SPACE_ADDRESS_MODE (addr_space_t @var{address_space})
10197 Define this to return the machine mode to use for addresses in
10198 @var{address_space} if the target supports named address spaces.
10199 The default version of this hook returns @code{Pmode} for the
10200 generic address space only.
10203 @deftypefn {Target Hook} bool TARGET_ADDR_SPACE_VALID_POINTER_MODE (enum machine_mode @var{mode}, addr_space_t @var{as})
10204 Define this to return nonzero if the port can handle pointers
10205 with machine mode @var{mode} to address space @var{as}. This target
10206 hook is the same as the @code{TARGET_VALID_POINTER_MODE} target hook,
10207 except that it includes explicit named address space support. The default
10208 version of this hook returns true for the modes returned by either the
10209 @code{TARGET_ADDR_SPACE_POINTER_MODE} or @code{TARGET_ADDR_SPACE_ADDRESS_MODE}
10210 target hooks for the given address space.
10213 @deftypefn {Target Hook} bool TARGET_ADDR_SPACE_LEGITIMATE_ADDRESS_P (enum machine_mode @var{mode}, rtx @var{exp}, bool @var{strict}, addr_space_t @var{as})
10214 Define this to return true if @var{exp} is a valid address for mode
10215 @var{mode} in the named address space @var{as}. The @var{strict}
10216 parameter says whether strict addressing is in effect after reload has
10217 finished. This target hook is the same as the
10218 @code{TARGET_LEGITIMATE_ADDRESS_P} target hook, except that it includes
10219 explicit named address space support.
10222 @deftypefn {Target Hook} rtx TARGET_ADDR_SPACE_LEGITIMIZE_ADDRESS (rtx @var{x}, rtx @var{oldx}, enum machine_mode @var{mode}, addr_space_t @var{as})
10223 Define this to modify an invalid address @var{x} to be a valid address
10224 with mode @var{mode} in the named address space @var{as}. This target
10225 hook is the same as the @code{TARGET_LEGITIMIZE_ADDRESS} target hook,
10226 except that it includes explicit named address space support.
10229 @deftypefn {Target Hook} bool TARGET_ADDR_SPACE_SUBSET_P (addr_space_t @var{subset}, addr_space_t @var{superset})
10230 Define this to return whether the @var{subset} named address space is
10231 contained within the @var{superset} named address space. Pointers to
10232 a named address space that is a subset of another named address space
10233 will be converted automatically without a cast if used together in
10234 arithmetic operations. Pointers to a superset address space can be
10235 converted to pointers to a subset address space via explicit casts.
10238 @deftypefn {Target Hook} rtx TARGET_ADDR_SPACE_CONVERT (rtx @var{op}, tree @var{from_type}, tree @var{to_type})
10239 Define this to convert the pointer expression represented by the RTL
10240 @var{op} with type @var{from_type} that points to a named address
10241 space to a new pointer expression with type @var{to_type} that points
10242 to a different named address space. When this hook it called, it is
10243 guaranteed that one of the two address spaces is a subset of the other,
10244 as determined by the @code{TARGET_ADDR_SPACE_SUBSET_P} target hook.
10248 @section Miscellaneous Parameters
10249 @cindex parameters, miscellaneous
10251 @c prevent bad page break with this line
10252 Here are several miscellaneous parameters.
10254 @defmac HAS_LONG_COND_BRANCH
10255 Define this boolean macro to indicate whether or not your architecture
10256 has conditional branches that can span all of memory. It is used in
10257 conjunction with an optimization that partitions hot and cold basic
10258 blocks into separate sections of the executable. If this macro is
10259 set to false, gcc will convert any conditional branches that attempt
10260 to cross between sections into unconditional branches or indirect jumps.
10263 @defmac HAS_LONG_UNCOND_BRANCH
10264 Define this boolean macro to indicate whether or not your architecture
10265 has unconditional branches that can span all of memory. It is used in
10266 conjunction with an optimization that partitions hot and cold basic
10267 blocks into separate sections of the executable. If this macro is
10268 set to false, gcc will convert any unconditional branches that attempt
10269 to cross between sections into indirect jumps.
10272 @defmac CASE_VECTOR_MODE
10273 An alias for a machine mode name. This is the machine mode that
10274 elements of a jump-table should have.
10277 @defmac CASE_VECTOR_SHORTEN_MODE (@var{min_offset}, @var{max_offset}, @var{body})
10278 Optional: return the preferred mode for an @code{addr_diff_vec}
10279 when the minimum and maximum offset are known. If you define this,
10280 it enables extra code in branch shortening to deal with @code{addr_diff_vec}.
10281 To make this work, you also have to define @code{INSN_ALIGN} and
10282 make the alignment for @code{addr_diff_vec} explicit.
10283 The @var{body} argument is provided so that the offset_unsigned and scale
10284 flags can be updated.
10287 @defmac CASE_VECTOR_PC_RELATIVE
10288 Define this macro to be a C expression to indicate when jump-tables
10289 should contain relative addresses. You need not define this macro if
10290 jump-tables never contain relative addresses, or jump-tables should
10291 contain relative addresses only when @option{-fPIC} or @option{-fPIC}
10295 @deftypefn {Target Hook} {unsigned int} TARGET_CASE_VALUES_THRESHOLD (void)
10296 This function return the smallest number of different values for which it
10297 is best to use a jump-table instead of a tree of conditional branches.
10298 The default is four for machines with a @code{casesi} instruction and
10299 five otherwise. This is best for most machines.
10302 @defmac CASE_USE_BIT_TESTS
10303 Define this macro to be a C expression to indicate whether C switch
10304 statements may be implemented by a sequence of bit tests. This is
10305 advantageous on processors that can efficiently implement left shift
10306 of 1 by the number of bits held in a register, but inappropriate on
10307 targets that would require a loop. By default, this macro returns
10308 @code{true} if the target defines an @code{ashlsi3} pattern, and
10309 @code{false} otherwise.
10312 @defmac WORD_REGISTER_OPERATIONS
10313 Define this macro if operations between registers with integral mode
10314 smaller than a word are always performed on the entire register.
10315 Most RISC machines have this property and most CISC machines do not.
10318 @defmac LOAD_EXTEND_OP (@var{mem_mode})
10319 Define this macro to be a C expression indicating when insns that read
10320 memory in @var{mem_mode}, an integral mode narrower than a word, set the
10321 bits outside of @var{mem_mode} to be either the sign-extension or the
10322 zero-extension of the data read. Return @code{SIGN_EXTEND} for values
10323 of @var{mem_mode} for which the
10324 insn sign-extends, @code{ZERO_EXTEND} for which it zero-extends, and
10325 @code{UNKNOWN} for other modes.
10327 This macro is not called with @var{mem_mode} non-integral or with a width
10328 greater than or equal to @code{BITS_PER_WORD}, so you may return any
10329 value in this case. Do not define this macro if it would always return
10330 @code{UNKNOWN}. On machines where this macro is defined, you will normally
10331 define it as the constant @code{SIGN_EXTEND} or @code{ZERO_EXTEND}.
10333 You may return a non-@code{UNKNOWN} value even if for some hard registers
10334 the sign extension is not performed, if for the @code{REGNO_REG_CLASS}
10335 of these hard registers @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero
10336 when the @var{from} mode is @var{mem_mode} and the @var{to} mode is any
10337 integral mode larger than this but not larger than @code{word_mode}.
10339 You must return @code{UNKNOWN} if for some hard registers that allow this
10340 mode, @code{CANNOT_CHANGE_MODE_CLASS} says that they cannot change to
10341 @code{word_mode}, but that they can change to another integral mode that
10342 is larger then @var{mem_mode} but still smaller than @code{word_mode}.
10345 @defmac SHORT_IMMEDIATES_SIGN_EXTEND
10346 Define this macro if loading short immediate values into registers sign
10350 @defmac FIXUNS_TRUNC_LIKE_FIX_TRUNC
10351 Define this macro if the same instructions that convert a floating
10352 point number to a signed fixed point number also convert validly to an
10356 @deftypefn {Target Hook} {unsigned int} TARGET_MIN_DIVISIONS_FOR_RECIP_MUL (enum machine_mode @var{mode})
10357 When @option{-ffast-math} is in effect, GCC tries to optimize
10358 divisions by the same divisor, by turning them into multiplications by
10359 the reciprocal. This target hook specifies the minimum number of divisions
10360 that should be there for GCC to perform the optimization for a variable
10361 of mode @var{mode}. The default implementation returns 3 if the machine
10362 has an instruction for the division, and 2 if it does not.
10366 The maximum number of bytes that a single instruction can move quickly
10367 between memory and registers or between two memory locations.
10370 @defmac MAX_MOVE_MAX
10371 The maximum number of bytes that a single instruction can move quickly
10372 between memory and registers or between two memory locations. If this
10373 is undefined, the default is @code{MOVE_MAX}. Otherwise, it is the
10374 constant value that is the largest value that @code{MOVE_MAX} can have
10378 @defmac SHIFT_COUNT_TRUNCATED
10379 A C expression that is nonzero if on this machine the number of bits
10380 actually used for the count of a shift operation is equal to the number
10381 of bits needed to represent the size of the object being shifted. When
10382 this macro is nonzero, the compiler will assume that it is safe to omit
10383 a sign-extend, zero-extend, and certain bitwise `and' instructions that
10384 truncates the count of a shift operation. On machines that have
10385 instructions that act on bit-fields at variable positions, which may
10386 include `bit test' instructions, a nonzero @code{SHIFT_COUNT_TRUNCATED}
10387 also enables deletion of truncations of the values that serve as
10388 arguments to bit-field instructions.
10390 If both types of instructions truncate the count (for shifts) and
10391 position (for bit-field operations), or if no variable-position bit-field
10392 instructions exist, you should define this macro.
10394 However, on some machines, such as the 80386 and the 680x0, truncation
10395 only applies to shift operations and not the (real or pretended)
10396 bit-field operations. Define @code{SHIFT_COUNT_TRUNCATED} to be zero on
10397 such machines. Instead, add patterns to the @file{md} file that include
10398 the implied truncation of the shift instructions.
10400 You need not define this macro if it would always have the value of zero.
10403 @anchor{TARGET_SHIFT_TRUNCATION_MASK}
10404 @deftypefn {Target Hook} {unsigned HOST_WIDE_INT} TARGET_SHIFT_TRUNCATION_MASK (enum machine_mode @var{mode})
10405 This function describes how the standard shift patterns for @var{mode}
10406 deal with shifts by negative amounts or by more than the width of the mode.
10407 @xref{shift patterns}.
10409 On many machines, the shift patterns will apply a mask @var{m} to the
10410 shift count, meaning that a fixed-width shift of @var{x} by @var{y} is
10411 equivalent to an arbitrary-width shift of @var{x} by @var{y & m}. If
10412 this is true for mode @var{mode}, the function should return @var{m},
10413 otherwise it should return 0. A return value of 0 indicates that no
10414 particular behavior is guaranteed.
10416 Note that, unlike @code{SHIFT_COUNT_TRUNCATED}, this function does
10417 @emph{not} apply to general shift rtxes; it applies only to instructions
10418 that are generated by the named shift patterns.
10420 The default implementation of this function returns
10421 @code{GET_MODE_BITSIZE (@var{mode}) - 1} if @code{SHIFT_COUNT_TRUNCATED}
10422 and 0 otherwise. This definition is always safe, but if
10423 @code{SHIFT_COUNT_TRUNCATED} is false, and some shift patterns
10424 nevertheless truncate the shift count, you may get better code
10428 @defmac TRULY_NOOP_TRUNCATION (@var{outprec}, @var{inprec})
10429 A C expression which is nonzero if on this machine it is safe to
10430 ``convert'' an integer of @var{inprec} bits to one of @var{outprec}
10431 bits (where @var{outprec} is smaller than @var{inprec}) by merely
10432 operating on it as if it had only @var{outprec} bits.
10434 On many machines, this expression can be 1.
10436 @c rearranged this, removed the phrase "it is reported that". this was
10437 @c to fix an overfull hbox. --mew 10feb93
10438 When @code{TRULY_NOOP_TRUNCATION} returns 1 for a pair of sizes for
10439 modes for which @code{MODES_TIEABLE_P} is 0, suboptimal code can result.
10440 If this is the case, making @code{TRULY_NOOP_TRUNCATION} return 0 in
10441 such cases may improve things.
10444 @deftypefn {Target Hook} int TARGET_MODE_REP_EXTENDED (enum machine_mode @var{mode}, enum machine_mode @var{rep_mode})
10445 The representation of an integral mode can be such that the values
10446 are always extended to a wider integral mode. Return
10447 @code{SIGN_EXTEND} if values of @var{mode} are represented in
10448 sign-extended form to @var{rep_mode}. Return @code{UNKNOWN}
10449 otherwise. (Currently, none of the targets use zero-extended
10450 representation this way so unlike @code{LOAD_EXTEND_OP},
10451 @code{TARGET_MODE_REP_EXTENDED} is expected to return either
10452 @code{SIGN_EXTEND} or @code{UNKNOWN}. Also no target extends
10453 @var{mode} to @var{rep_mode} so that @var{rep_mode} is not the next
10454 widest integral mode and currently we take advantage of this fact.)
10456 Similarly to @code{LOAD_EXTEND_OP} you may return a non-@code{UNKNOWN}
10457 value even if the extension is not performed on certain hard registers
10458 as long as for the @code{REGNO_REG_CLASS} of these hard registers
10459 @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero.
10461 Note that @code{TARGET_MODE_REP_EXTENDED} and @code{LOAD_EXTEND_OP}
10462 describe two related properties. If you define
10463 @code{TARGET_MODE_REP_EXTENDED (mode, word_mode)} you probably also want
10464 to define @code{LOAD_EXTEND_OP (mode)} to return the same type of
10467 In order to enforce the representation of @code{mode},
10468 @code{TRULY_NOOP_TRUNCATION} should return false when truncating to
10472 @defmac STORE_FLAG_VALUE
10473 A C expression describing the value returned by a comparison operator
10474 with an integral mode and stored by a store-flag instruction
10475 (@samp{cstore@var{mode}4}) when the condition is true. This description must
10476 apply to @emph{all} the @samp{cstore@var{mode}4} patterns and all the
10477 comparison operators whose results have a @code{MODE_INT} mode.
10479 A value of 1 or @minus{}1 means that the instruction implementing the
10480 comparison operator returns exactly 1 or @minus{}1 when the comparison is true
10481 and 0 when the comparison is false. Otherwise, the value indicates
10482 which bits of the result are guaranteed to be 1 when the comparison is
10483 true. This value is interpreted in the mode of the comparison
10484 operation, which is given by the mode of the first operand in the
10485 @samp{cstore@var{mode}4} pattern. Either the low bit or the sign bit of
10486 @code{STORE_FLAG_VALUE} be on. Presently, only those bits are used by
10489 If @code{STORE_FLAG_VALUE} is neither 1 or @minus{}1, the compiler will
10490 generate code that depends only on the specified bits. It can also
10491 replace comparison operators with equivalent operations if they cause
10492 the required bits to be set, even if the remaining bits are undefined.
10493 For example, on a machine whose comparison operators return an
10494 @code{SImode} value and where @code{STORE_FLAG_VALUE} is defined as
10495 @samp{0x80000000}, saying that just the sign bit is relevant, the
10499 (ne:SI (and:SI @var{x} (const_int @var{power-of-2})) (const_int 0))
10503 can be converted to
10506 (ashift:SI @var{x} (const_int @var{n}))
10510 where @var{n} is the appropriate shift count to move the bit being
10511 tested into the sign bit.
10513 There is no way to describe a machine that always sets the low-order bit
10514 for a true value, but does not guarantee the value of any other bits,
10515 but we do not know of any machine that has such an instruction. If you
10516 are trying to port GCC to such a machine, include an instruction to
10517 perform a logical-and of the result with 1 in the pattern for the
10518 comparison operators and let us know at @email{gcc@@gcc.gnu.org}.
10520 Often, a machine will have multiple instructions that obtain a value
10521 from a comparison (or the condition codes). Here are rules to guide the
10522 choice of value for @code{STORE_FLAG_VALUE}, and hence the instructions
10527 Use the shortest sequence that yields a valid definition for
10528 @code{STORE_FLAG_VALUE}. It is more efficient for the compiler to
10529 ``normalize'' the value (convert it to, e.g., 1 or 0) than for the
10530 comparison operators to do so because there may be opportunities to
10531 combine the normalization with other operations.
10534 For equal-length sequences, use a value of 1 or @minus{}1, with @minus{}1 being
10535 slightly preferred on machines with expensive jumps and 1 preferred on
10539 As a second choice, choose a value of @samp{0x80000001} if instructions
10540 exist that set both the sign and low-order bits but do not define the
10544 Otherwise, use a value of @samp{0x80000000}.
10547 Many machines can produce both the value chosen for
10548 @code{STORE_FLAG_VALUE} and its negation in the same number of
10549 instructions. On those machines, you should also define a pattern for
10550 those cases, e.g., one matching
10553 (set @var{A} (neg:@var{m} (ne:@var{m} @var{B} @var{C})))
10556 Some machines can also perform @code{and} or @code{plus} operations on
10557 condition code values with less instructions than the corresponding
10558 @samp{cstore@var{mode}4} insn followed by @code{and} or @code{plus}. On those
10559 machines, define the appropriate patterns. Use the names @code{incscc}
10560 and @code{decscc}, respectively, for the patterns which perform
10561 @code{plus} or @code{minus} operations on condition code values. See
10562 @file{rs6000.md} for some examples. The GNU Superoptimizer can be used to
10563 find such instruction sequences on other machines.
10565 If this macro is not defined, the default value, 1, is used. You need
10566 not define @code{STORE_FLAG_VALUE} if the machine has no store-flag
10567 instructions, or if the value generated by these instructions is 1.
10570 @defmac FLOAT_STORE_FLAG_VALUE (@var{mode})
10571 A C expression that gives a nonzero @code{REAL_VALUE_TYPE} value that is
10572 returned when comparison operators with floating-point results are true.
10573 Define this macro on machines that have comparison operations that return
10574 floating-point values. If there are no such operations, do not define
10578 @defmac VECTOR_STORE_FLAG_VALUE (@var{mode})
10579 A C expression that gives a rtx representing the nonzero true element
10580 for vector comparisons. The returned rtx should be valid for the inner
10581 mode of @var{mode} which is guaranteed to be a vector mode. Define
10582 this macro on machines that have vector comparison operations that
10583 return a vector result. If there are no such operations, do not define
10584 this macro. Typically, this macro is defined as @code{const1_rtx} or
10585 @code{constm1_rtx}. This macro may return @code{NULL_RTX} to prevent
10586 the compiler optimizing such vector comparison operations for the
10590 @defmac CLZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
10591 @defmacx CTZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
10592 A C expression that indicates whether the architecture defines a value
10593 for @code{clz} or @code{ctz} with a zero operand.
10594 A result of @code{0} indicates the value is undefined.
10595 If the value is defined for only the RTL expression, the macro should
10596 evaluate to @code{1}; if the value applies also to the corresponding optab
10597 entry (which is normally the case if it expands directly into
10598 the corresponding RTL), then the macro should evaluate to @code{2}.
10599 In the cases where the value is defined, @var{value} should be set to
10602 If this macro is not defined, the value of @code{clz} or
10603 @code{ctz} at zero is assumed to be undefined.
10605 This macro must be defined if the target's expansion for @code{ffs}
10606 relies on a particular value to get correct results. Otherwise it
10607 is not necessary, though it may be used to optimize some corner cases, and
10608 to provide a default expansion for the @code{ffs} optab.
10610 Note that regardless of this macro the ``definedness'' of @code{clz}
10611 and @code{ctz} at zero do @emph{not} extend to the builtin functions
10612 visible to the user. Thus one may be free to adjust the value at will
10613 to match the target expansion of these operations without fear of
10618 An alias for the machine mode for pointers. On most machines, define
10619 this to be the integer mode corresponding to the width of a hardware
10620 pointer; @code{SImode} on 32-bit machine or @code{DImode} on 64-bit machines.
10621 On some machines you must define this to be one of the partial integer
10622 modes, such as @code{PSImode}.
10624 The width of @code{Pmode} must be at least as large as the value of
10625 @code{POINTER_SIZE}. If it is not equal, you must define the macro
10626 @code{POINTERS_EXTEND_UNSIGNED} to specify how pointers are extended
10630 @defmac FUNCTION_MODE
10631 An alias for the machine mode used for memory references to functions
10632 being called, in @code{call} RTL expressions. On most CISC machines,
10633 where an instruction can begin at any byte address, this should be
10634 @code{QImode}. On most RISC machines, where all instructions have fixed
10635 size and alignment, this should be a mode with the same size and alignment
10636 as the machine instruction words - typically @code{SImode} or @code{HImode}.
10639 @defmac STDC_0_IN_SYSTEM_HEADERS
10640 In normal operation, the preprocessor expands @code{__STDC__} to the
10641 constant 1, to signify that GCC conforms to ISO Standard C@. On some
10642 hosts, like Solaris, the system compiler uses a different convention,
10643 where @code{__STDC__} is normally 0, but is 1 if the user specifies
10644 strict conformance to the C Standard.
10646 Defining @code{STDC_0_IN_SYSTEM_HEADERS} makes GNU CPP follows the host
10647 convention when processing system header files, but when processing user
10648 files @code{__STDC__} will always expand to 1.
10651 @defmac NO_IMPLICIT_EXTERN_C
10652 Define this macro if the system header files support C++ as well as C@.
10653 This macro inhibits the usual method of using system header files in
10654 C++, which is to pretend that the file's contents are enclosed in
10655 @samp{extern "C" @{@dots{}@}}.
10660 @defmac REGISTER_TARGET_PRAGMAS ()
10661 Define this macro if you want to implement any target-specific pragmas.
10662 If defined, it is a C expression which makes a series of calls to
10663 @code{c_register_pragma} or @code{c_register_pragma_with_expansion}
10664 for each pragma. The macro may also do any
10665 setup required for the pragmas.
10667 The primary reason to define this macro is to provide compatibility with
10668 other compilers for the same target. In general, we discourage
10669 definition of target-specific pragmas for GCC@.
10671 If the pragma can be implemented by attributes then you should consider
10672 defining the target hook @samp{TARGET_INSERT_ATTRIBUTES} as well.
10674 Preprocessor macros that appear on pragma lines are not expanded. All
10675 @samp{#pragma} directives that do not match any registered pragma are
10676 silently ignored, unless the user specifies @option{-Wunknown-pragmas}.
10679 @deftypefun void c_register_pragma (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
10680 @deftypefunx void c_register_pragma_with_expansion (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
10682 Each call to @code{c_register_pragma} or
10683 @code{c_register_pragma_with_expansion} establishes one pragma. The
10684 @var{callback} routine will be called when the preprocessor encounters a
10688 #pragma [@var{space}] @var{name} @dots{}
10691 @var{space} is the case-sensitive namespace of the pragma, or
10692 @code{NULL} to put the pragma in the global namespace. The callback
10693 routine receives @var{pfile} as its first argument, which can be passed
10694 on to cpplib's functions if necessary. You can lex tokens after the
10695 @var{name} by calling @code{pragma_lex}. Tokens that are not read by the
10696 callback will be silently ignored. The end of the line is indicated by
10697 a token of type @code{CPP_EOF}. Macro expansion occurs on the
10698 arguments of pragmas registered with
10699 @code{c_register_pragma_with_expansion} but not on the arguments of
10700 pragmas registered with @code{c_register_pragma}.
10702 Note that the use of @code{pragma_lex} is specific to the C and C++
10703 compilers. It will not work in the Java or Fortran compilers, or any
10704 other language compilers for that matter. Thus if @code{pragma_lex} is going
10705 to be called from target-specific code, it must only be done so when
10706 building the C and C++ compilers. This can be done by defining the
10707 variables @code{c_target_objs} and @code{cxx_target_objs} in the
10708 target entry in the @file{config.gcc} file. These variables should name
10709 the target-specific, language-specific object file which contains the
10710 code that uses @code{pragma_lex}. Note it will also be necessary to add a
10711 rule to the makefile fragment pointed to by @code{tmake_file} that shows
10712 how to build this object file.
10715 @defmac HANDLE_PRAGMA_PACK_WITH_EXPANSION
10716 Define this macro if macros should be expanded in the
10717 arguments of @samp{#pragma pack}.
10720 @deftypevr {Target Hook} bool TARGET_HANDLE_PRAGMA_EXTERN_PREFIX
10721 True if @code{#pragma extern_prefix} is to be supported.
10724 @defmac TARGET_DEFAULT_PACK_STRUCT
10725 If your target requires a structure packing default other than 0 (meaning
10726 the machine default), define this macro to the necessary value (in bytes).
10727 This must be a value that would also be valid to use with
10728 @samp{#pragma pack()} (that is, a small power of two).
10731 @defmac DOLLARS_IN_IDENTIFIERS
10732 Define this macro to control use of the character @samp{$} in
10733 identifier names for the C family of languages. 0 means @samp{$} is
10734 not allowed by default; 1 means it is allowed. 1 is the default;
10735 there is no need to define this macro in that case.
10738 @defmac NO_DOLLAR_IN_LABEL
10739 Define this macro if the assembler does not accept the character
10740 @samp{$} in label names. By default constructors and destructors in
10741 G++ have @samp{$} in the identifiers. If this macro is defined,
10742 @samp{.} is used instead.
10745 @defmac NO_DOT_IN_LABEL
10746 Define this macro if the assembler does not accept the character
10747 @samp{.} in label names. By default constructors and destructors in G++
10748 have names that use @samp{.}. If this macro is defined, these names
10749 are rewritten to avoid @samp{.}.
10752 @defmac INSN_SETS_ARE_DELAYED (@var{insn})
10753 Define this macro as a C expression that is nonzero if it is safe for the
10754 delay slot scheduler to place instructions in the delay slot of @var{insn},
10755 even if they appear to use a resource set or clobbered in @var{insn}.
10756 @var{insn} is always a @code{jump_insn} or an @code{insn}; GCC knows that
10757 every @code{call_insn} has this behavior. On machines where some @code{insn}
10758 or @code{jump_insn} is really a function call and hence has this behavior,
10759 you should define this macro.
10761 You need not define this macro if it would always return zero.
10764 @defmac INSN_REFERENCES_ARE_DELAYED (@var{insn})
10765 Define this macro as a C expression that is nonzero if it is safe for the
10766 delay slot scheduler to place instructions in the delay slot of @var{insn},
10767 even if they appear to set or clobber a resource referenced in @var{insn}.
10768 @var{insn} is always a @code{jump_insn} or an @code{insn}. On machines where
10769 some @code{insn} or @code{jump_insn} is really a function call and its operands
10770 are registers whose use is actually in the subroutine it calls, you should
10771 define this macro. Doing so allows the delay slot scheduler to move
10772 instructions which copy arguments into the argument registers into the delay
10773 slot of @var{insn}.
10775 You need not define this macro if it would always return zero.
10778 @defmac MULTIPLE_SYMBOL_SPACES
10779 Define this macro as a C expression that is nonzero if, in some cases,
10780 global symbols from one translation unit may not be bound to undefined
10781 symbols in another translation unit without user intervention. For
10782 instance, under Microsoft Windows symbols must be explicitly imported
10783 from shared libraries (DLLs).
10785 You need not define this macro if it would always evaluate to zero.
10788 @deftypefn {Target Hook} tree TARGET_MD_ASM_CLOBBERS (tree @var{outputs}, tree @var{inputs}, tree @var{clobbers})
10789 This target hook should add to @var{clobbers} @code{STRING_CST} trees for
10790 any hard regs the port wishes to automatically clobber for an asm.
10791 It should return the result of the last @code{tree_cons} used to add a
10792 clobber. The @var{outputs}, @var{inputs} and @var{clobber} lists are the
10793 corresponding parameters to the asm and may be inspected to avoid
10794 clobbering a register that is an input or output of the asm. You can use
10795 @code{tree_overlaps_hard_reg_set}, declared in @file{tree.h}, to test
10796 for overlap with regards to asm-declared registers.
10799 @defmac MATH_LIBRARY
10800 Define this macro as a C string constant for the linker argument to link
10801 in the system math library, minus the initial @samp{"-l"}, or
10802 @samp{""} if the target does not have a
10803 separate math library.
10805 You need only define this macro if the default of @samp{"m"} is wrong.
10808 @defmac LIBRARY_PATH_ENV
10809 Define this macro as a C string constant for the environment variable that
10810 specifies where the linker should look for libraries.
10812 You need only define this macro if the default of @samp{"LIBRARY_PATH"}
10816 @defmac TARGET_POSIX_IO
10817 Define this macro if the target supports the following POSIX@ file
10818 functions, access, mkdir and file locking with fcntl / F_SETLKW@.
10819 Defining @code{TARGET_POSIX_IO} will enable the test coverage code
10820 to use file locking when exiting a program, which avoids race conditions
10821 if the program has forked. It will also create directories at run-time
10822 for cross-profiling.
10825 @defmac MAX_CONDITIONAL_EXECUTE
10827 A C expression for the maximum number of instructions to execute via
10828 conditional execution instructions instead of a branch. A value of
10829 @code{BRANCH_COST}+1 is the default if the machine does not use cc0, and
10830 1 if it does use cc0.
10833 @defmac IFCVT_MODIFY_TESTS (@var{ce_info}, @var{true_expr}, @var{false_expr})
10834 Used if the target needs to perform machine-dependent modifications on the
10835 conditionals used for turning basic blocks into conditionally executed code.
10836 @var{ce_info} points to a data structure, @code{struct ce_if_block}, which
10837 contains information about the currently processed blocks. @var{true_expr}
10838 and @var{false_expr} are the tests that are used for converting the
10839 then-block and the else-block, respectively. Set either @var{true_expr} or
10840 @var{false_expr} to a null pointer if the tests cannot be converted.
10843 @defmac IFCVT_MODIFY_MULTIPLE_TESTS (@var{ce_info}, @var{bb}, @var{true_expr}, @var{false_expr})
10844 Like @code{IFCVT_MODIFY_TESTS}, but used when converting more complicated
10845 if-statements into conditions combined by @code{and} and @code{or} operations.
10846 @var{bb} contains the basic block that contains the test that is currently
10847 being processed and about to be turned into a condition.
10850 @defmac IFCVT_MODIFY_INSN (@var{ce_info}, @var{pattern}, @var{insn})
10851 A C expression to modify the @var{PATTERN} of an @var{INSN} that is to
10852 be converted to conditional execution format. @var{ce_info} points to
10853 a data structure, @code{struct ce_if_block}, which contains information
10854 about the currently processed blocks.
10857 @defmac IFCVT_MODIFY_FINAL (@var{ce_info})
10858 A C expression to perform any final machine dependent modifications in
10859 converting code to conditional execution. The involved basic blocks
10860 can be found in the @code{struct ce_if_block} structure that is pointed
10861 to by @var{ce_info}.
10864 @defmac IFCVT_MODIFY_CANCEL (@var{ce_info})
10865 A C expression to cancel any machine dependent modifications in
10866 converting code to conditional execution. The involved basic blocks
10867 can be found in the @code{struct ce_if_block} structure that is pointed
10868 to by @var{ce_info}.
10871 @defmac IFCVT_INIT_EXTRA_FIELDS (@var{ce_info})
10872 A C expression to initialize any extra fields in a @code{struct ce_if_block}
10873 structure, which are defined by the @code{IFCVT_EXTRA_FIELDS} macro.
10876 @defmac IFCVT_EXTRA_FIELDS
10877 If defined, it should expand to a set of field declarations that will be
10878 added to the @code{struct ce_if_block} structure. These should be initialized
10879 by the @code{IFCVT_INIT_EXTRA_FIELDS} macro.
10882 @deftypefn {Target Hook} void TARGET_MACHINE_DEPENDENT_REORG (void)
10883 If non-null, this hook performs a target-specific pass over the
10884 instruction stream. The compiler will run it at all optimization levels,
10885 just before the point at which it normally does delayed-branch scheduling.
10887 The exact purpose of the hook varies from target to target. Some use
10888 it to do transformations that are necessary for correctness, such as
10889 laying out in-function constant pools or avoiding hardware hazards.
10890 Others use it as an opportunity to do some machine-dependent optimizations.
10892 You need not implement the hook if it has nothing to do. The default
10893 definition is null.
10896 @deftypefn {Target Hook} void TARGET_INIT_BUILTINS (void)
10897 Define this hook if you have any machine-specific built-in functions
10898 that need to be defined. It should be a function that performs the
10901 Machine specific built-in functions can be useful to expand special machine
10902 instructions that would otherwise not normally be generated because
10903 they have no equivalent in the source language (for example, SIMD vector
10904 instructions or prefetch instructions).
10906 To create a built-in function, call the function
10907 @code{lang_hooks.builtin_function}
10908 which is defined by the language front end. You can use any type nodes set
10909 up by @code{build_common_tree_nodes};
10910 only language front ends that use those two functions will call
10911 @samp{TARGET_INIT_BUILTINS}.
10914 @deftypefn {Target Hook} tree TARGET_BUILTIN_DECL (unsigned @var{code}, bool @var{initialize_p})
10915 Define this hook if you have any machine-specific built-in functions
10916 that need to be defined. It should be a function that returns the
10917 builtin function declaration for the builtin function code @var{code}.
10918 If there is no such builtin and it cannot be initialized at this time
10919 if @var{initialize_p} is true the function should return @code{NULL_TREE}.
10920 If @var{code} is out of range the function should return
10921 @code{error_mark_node}.
10924 @deftypefn {Target Hook} rtx TARGET_EXPAND_BUILTIN (tree @var{exp}, rtx @var{target}, rtx @var{subtarget}, enum machine_mode @var{mode}, int @var{ignore})
10926 Expand a call to a machine specific built-in function that was set up by
10927 @samp{TARGET_INIT_BUILTINS}. @var{exp} is the expression for the
10928 function call; the result should go to @var{target} if that is
10929 convenient, and have mode @var{mode} if that is convenient.
10930 @var{subtarget} may be used as the target for computing one of
10931 @var{exp}'s operands. @var{ignore} is nonzero if the value is to be
10932 ignored. This function should return the result of the call to the
10936 @deftypefn {Target Hook} tree TARGET_RESOLVE_OVERLOADED_BUILTIN (unsigned int @var{loc}, tree @var{fndecl}, void *@var{arglist})
10937 Select a replacement for a machine specific built-in function that
10938 was set up by @samp{TARGET_INIT_BUILTINS}. This is done
10939 @emph{before} regular type checking, and so allows the target to
10940 implement a crude form of function overloading. @var{fndecl} is the
10941 declaration of the built-in function. @var{arglist} is the list of
10942 arguments passed to the built-in function. The result is a
10943 complete expression that implements the operation, usually
10944 another @code{CALL_EXPR}.
10945 @var{arglist} really has type @samp{VEC(tree,gc)*}
10948 @deftypefn {Target Hook} tree TARGET_FOLD_BUILTIN (tree @var{fndecl}, int @var{n_args}, tree *@var{argp}, bool @var{ignore})
10949 Fold a call to a machine specific built-in function that was set up by
10950 @samp{TARGET_INIT_BUILTINS}. @var{fndecl} is the declaration of the
10951 built-in function. @var{n_args} is the number of arguments passed to
10952 the function; the arguments themselves are pointed to by @var{argp}.
10953 The result is another tree containing a simplified expression for the
10954 call's result. If @var{ignore} is true the value will be ignored.
10957 @deftypefn {Target Hook} {const char *} TARGET_INVALID_WITHIN_DOLOOP (const_rtx @var{insn})
10959 Take an instruction in @var{insn} and return NULL if it is valid within a
10960 low-overhead loop, otherwise return a string explaining why doloop
10961 could not be applied.
10963 Many targets use special registers for low-overhead looping. For any
10964 instruction that clobbers these this function should return a string indicating
10965 the reason why the doloop could not be applied.
10966 By default, the RTL loop optimizer does not use a present doloop pattern for
10967 loops containing function calls or branch on table instructions.
10970 @defmac MD_CAN_REDIRECT_BRANCH (@var{branch1}, @var{branch2})
10972 Take a branch insn in @var{branch1} and another in @var{branch2}.
10973 Return true if redirecting @var{branch1} to the destination of
10974 @var{branch2} is possible.
10976 On some targets, branches may have a limited range. Optimizing the
10977 filling of delay slots can result in branches being redirected, and this
10978 may in turn cause a branch offset to overflow.
10981 @deftypefn {Target Hook} bool TARGET_COMMUTATIVE_P (const_rtx @var{x}, int @var{outer_code})
10982 This target hook returns @code{true} if @var{x} is considered to be commutative.
10983 Usually, this is just COMMUTATIVE_P (@var{x}), but the HP PA doesn't consider
10984 PLUS to be commutative inside a MEM@. @var{outer_code} is the rtx code
10985 of the enclosing rtl, if known, otherwise it is UNKNOWN.
10988 @deftypefn {Target Hook} rtx TARGET_ALLOCATE_INITIAL_VALUE (rtx @var{hard_reg})
10990 When the initial value of a hard register has been copied in a pseudo
10991 register, it is often not necessary to actually allocate another register
10992 to this pseudo register, because the original hard register or a stack slot
10993 it has been saved into can be used. @code{TARGET_ALLOCATE_INITIAL_VALUE}
10994 is called at the start of register allocation once for each hard register
10995 that had its initial value copied by using
10996 @code{get_func_hard_reg_initial_val} or @code{get_hard_reg_initial_val}.
10997 Possible values are @code{NULL_RTX}, if you don't want
10998 to do any special allocation, a @code{REG} rtx---that would typically be
10999 the hard register itself, if it is known not to be clobbered---or a
11001 If you are returning a @code{MEM}, this is only a hint for the allocator;
11002 it might decide to use another register anyways.
11003 You may use @code{current_function_leaf_function} in the hook, functions
11004 that use @code{REG_N_SETS}, to determine if the hard
11005 register in question will not be clobbered.
11006 The default value of this hook is @code{NULL}, which disables any special
11010 @deftypefn {Target Hook} int TARGET_UNSPEC_MAY_TRAP_P (const_rtx @var{x}, unsigned @var{flags})
11011 This target hook returns nonzero if @var{x}, an @code{unspec} or
11012 @code{unspec_volatile} operation, might cause a trap. Targets can use
11013 this hook to enhance precision of analysis for @code{unspec} and
11014 @code{unspec_volatile} operations. You may call @code{may_trap_p_1}
11015 to analyze inner elements of @var{x} in which case @var{flags} should be
11019 @deftypefn {Target Hook} void TARGET_SET_CURRENT_FUNCTION (tree @var{decl})
11020 The compiler invokes this hook whenever it changes its current function
11021 context (@code{cfun}). You can define this function if
11022 the back end needs to perform any initialization or reset actions on a
11023 per-function basis. For example, it may be used to implement function
11024 attributes that affect register usage or code generation patterns.
11025 The argument @var{decl} is the declaration for the new function context,
11026 and may be null to indicate that the compiler has left a function context
11027 and is returning to processing at the top level.
11028 The default hook function does nothing.
11030 GCC sets @code{cfun} to a dummy function context during initialization of
11031 some parts of the back end. The hook function is not invoked in this
11032 situation; you need not worry about the hook being invoked recursively,
11033 or when the back end is in a partially-initialized state.
11034 @code{cfun} might be @code{NULL} to indicate processing at top level,
11035 outside of any function scope.
11038 @defmac TARGET_OBJECT_SUFFIX
11039 Define this macro to be a C string representing the suffix for object
11040 files on your target machine. If you do not define this macro, GCC will
11041 use @samp{.o} as the suffix for object files.
11044 @defmac TARGET_EXECUTABLE_SUFFIX
11045 Define this macro to be a C string representing the suffix to be
11046 automatically added to executable files on your target machine. If you
11047 do not define this macro, GCC will use the null string as the suffix for
11051 @defmac COLLECT_EXPORT_LIST
11052 If defined, @code{collect2} will scan the individual object files
11053 specified on its command line and create an export list for the linker.
11054 Define this macro for systems like AIX, where the linker discards
11055 object files that are not referenced from @code{main} and uses export
11059 @defmac MODIFY_JNI_METHOD_CALL (@var{mdecl})
11060 Define this macro to a C expression representing a variant of the
11061 method call @var{mdecl}, if Java Native Interface (JNI) methods
11062 must be invoked differently from other methods on your target.
11063 For example, on 32-bit Microsoft Windows, JNI methods must be invoked using
11064 the @code{stdcall} calling convention and this macro is then
11065 defined as this expression:
11068 build_type_attribute_variant (@var{mdecl},
11070 (get_identifier ("stdcall"),
11075 @deftypefn {Target Hook} bool TARGET_CANNOT_MODIFY_JUMPS_P (void)
11076 This target hook returns @code{true} past the point in which new jump
11077 instructions could be created. On machines that require a register for
11078 every jump such as the SHmedia ISA of SH5, this point would typically be
11079 reload, so this target hook should be defined to a function such as:
11083 cannot_modify_jumps_past_reload_p ()
11085 return (reload_completed || reload_in_progress);
11090 @deftypefn {Target Hook} reg_class_t TARGET_BRANCH_TARGET_REGISTER_CLASS (void)
11091 This target hook returns a register class for which branch target register
11092 optimizations should be applied. All registers in this class should be
11093 usable interchangeably. After reload, registers in this class will be
11094 re-allocated and loads will be hoisted out of loops and be subjected
11095 to inter-block scheduling.
11098 @deftypefn {Target Hook} bool TARGET_BRANCH_TARGET_REGISTER_CALLEE_SAVED (bool @var{after_prologue_epilogue_gen})
11099 Branch target register optimization will by default exclude callee-saved
11101 that are not already live during the current function; if this target hook
11102 returns true, they will be included. The target code must than make sure
11103 that all target registers in the class returned by
11104 @samp{TARGET_BRANCH_TARGET_REGISTER_CLASS} that might need saving are
11105 saved. @var{after_prologue_epilogue_gen} indicates if prologues and
11106 epilogues have already been generated. Note, even if you only return
11107 true when @var{after_prologue_epilogue_gen} is false, you still are likely
11108 to have to make special provisions in @code{INITIAL_ELIMINATION_OFFSET}
11109 to reserve space for caller-saved target registers.
11112 @deftypefn {Target Hook} bool TARGET_HAVE_CONDITIONAL_EXECUTION (void)
11113 This target hook returns true if the target supports conditional execution.
11114 This target hook is required only when the target has several different
11115 modes and they have different conditional execution capability, such as ARM.
11118 @deftypefn {Target Hook} unsigned TARGET_LOOP_UNROLL_ADJUST (unsigned @var{nunroll}, struct loop *@var{loop})
11119 This target hook returns a new value for the number of times @var{loop}
11120 should be unrolled. The parameter @var{nunroll} is the number of times
11121 the loop is to be unrolled. The parameter @var{loop} is a pointer to
11122 the loop, which is going to be checked for unrolling. This target hook
11123 is required only when the target has special constraints like maximum
11124 number of memory accesses.
11127 @defmac POWI_MAX_MULTS
11128 If defined, this macro is interpreted as a signed integer C expression
11129 that specifies the maximum number of floating point multiplications
11130 that should be emitted when expanding exponentiation by an integer
11131 constant inline. When this value is defined, exponentiation requiring
11132 more than this number of multiplications is implemented by calling the
11133 system library's @code{pow}, @code{powf} or @code{powl} routines.
11134 The default value places no upper bound on the multiplication count.
11137 @deftypefn Macro void TARGET_EXTRA_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
11138 This target hook should register any extra include files for the
11139 target. The parameter @var{stdinc} indicates if normal include files
11140 are present. The parameter @var{sysroot} is the system root directory.
11141 The parameter @var{iprefix} is the prefix for the gcc directory.
11144 @deftypefn Macro void TARGET_EXTRA_PRE_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
11145 This target hook should register any extra include files for the
11146 target before any standard headers. The parameter @var{stdinc}
11147 indicates if normal include files are present. The parameter
11148 @var{sysroot} is the system root directory. The parameter
11149 @var{iprefix} is the prefix for the gcc directory.
11152 @deftypefn Macro void TARGET_OPTF (char *@var{path})
11153 This target hook should register special include paths for the target.
11154 The parameter @var{path} is the include to register. On Darwin
11155 systems, this is used for Framework includes, which have semantics
11156 that are different from @option{-I}.
11159 @defmac bool TARGET_USE_LOCAL_THUNK_ALIAS_P (tree @var{fndecl})
11160 This target macro returns @code{true} if it is safe to use a local alias
11161 for a virtual function @var{fndecl} when constructing thunks,
11162 @code{false} otherwise. By default, the macro returns @code{true} for all
11163 functions, if a target supports aliases (i.e.@: defines
11164 @code{ASM_OUTPUT_DEF}), @code{false} otherwise,
11167 @defmac TARGET_FORMAT_TYPES
11168 If defined, this macro is the name of a global variable containing
11169 target-specific format checking information for the @option{-Wformat}
11170 option. The default is to have no target-specific format checks.
11173 @defmac TARGET_N_FORMAT_TYPES
11174 If defined, this macro is the number of entries in
11175 @code{TARGET_FORMAT_TYPES}.
11178 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES
11179 If defined, this macro is the name of a global variable containing
11180 target-specific format overrides for the @option{-Wformat} option. The
11181 default is to have no target-specific format overrides. If defined,
11182 @code{TARGET_FORMAT_TYPES} must be defined, too.
11185 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES_COUNT
11186 If defined, this macro specifies the number of entries in
11187 @code{TARGET_OVERRIDES_FORMAT_ATTRIBUTES}.
11190 @defmac TARGET_OVERRIDES_FORMAT_INIT
11191 If defined, this macro specifies the optional initialization
11192 routine for target specific customizations of the system printf
11193 and scanf formatter settings.
11196 @deftypevr {Target Hook} bool TARGET_RELAXED_ORDERING
11197 If set to @code{true}, means that the target's memory model does not
11198 guarantee that loads which do not depend on one another will access
11199 main memory in the order of the instruction stream; if ordering is
11200 important, an explicit memory barrier must be used. This is true of
11201 many recent processors which implement a policy of ``relaxed,''
11202 ``weak,'' or ``release'' memory consistency, such as Alpha, PowerPC,
11203 and ia64. The default is @code{false}.
11206 @deftypefn {Target Hook} {const char *} TARGET_INVALID_ARG_FOR_UNPROTOTYPED_FN (const_tree @var{typelist}, const_tree @var{funcdecl}, const_tree @var{val})
11207 If defined, this macro returns the diagnostic message when it is
11208 illegal to pass argument @var{val} to function @var{funcdecl}
11209 with prototype @var{typelist}.
11212 @deftypefn {Target Hook} {const char *} TARGET_INVALID_CONVERSION (const_tree @var{fromtype}, const_tree @var{totype})
11213 If defined, this macro returns the diagnostic message when it is
11214 invalid to convert from @var{fromtype} to @var{totype}, or @code{NULL}
11215 if validity should be determined by the front end.
11218 @deftypefn {Target Hook} {const char *} TARGET_INVALID_UNARY_OP (int @var{op}, const_tree @var{type})
11219 If defined, this macro returns the diagnostic message when it is
11220 invalid to apply operation @var{op} (where unary plus is denoted by
11221 @code{CONVERT_EXPR}) to an operand of type @var{type}, or @code{NULL}
11222 if validity should be determined by the front end.
11225 @deftypefn {Target Hook} {const char *} TARGET_INVALID_BINARY_OP (int @var{op}, const_tree @var{type1}, const_tree @var{type2})
11226 If defined, this macro returns the diagnostic message when it is
11227 invalid to apply operation @var{op} to operands of types @var{type1}
11228 and @var{type2}, or @code{NULL} if validity should be determined by
11232 @deftypefn {Target Hook} {const char *} TARGET_INVALID_PARAMETER_TYPE (const_tree @var{type})
11233 If defined, this macro returns the diagnostic message when it is
11234 invalid for functions to include parameters of type @var{type},
11235 or @code{NULL} if validity should be determined by
11236 the front end. This is currently used only by the C and C++ front ends.
11239 @deftypefn {Target Hook} {const char *} TARGET_INVALID_RETURN_TYPE (const_tree @var{type})
11240 If defined, this macro returns the diagnostic message when it is
11241 invalid for functions to have return type @var{type},
11242 or @code{NULL} if validity should be determined by
11243 the front end. This is currently used only by the C and C++ front ends.
11246 @deftypefn {Target Hook} tree TARGET_PROMOTED_TYPE (const_tree @var{type})
11247 If defined, this target hook returns the type to which values of
11248 @var{type} should be promoted when they appear in expressions,
11249 analogous to the integer promotions, or @code{NULL_TREE} to use the
11250 front end's normal promotion rules. This hook is useful when there are
11251 target-specific types with special promotion rules.
11252 This is currently used only by the C and C++ front ends.
11255 @deftypefn {Target Hook} tree TARGET_CONVERT_TO_TYPE (tree @var{type}, tree @var{expr})
11256 If defined, this hook returns the result of converting @var{expr} to
11257 @var{type}. It should return the converted expression,
11258 or @code{NULL_TREE} to apply the front end's normal conversion rules.
11259 This hook is useful when there are target-specific types with special
11261 This is currently used only by the C and C++ front ends.
11264 @defmac TARGET_USE_JCR_SECTION
11265 This macro determines whether to use the JCR section to register Java
11266 classes. By default, TARGET_USE_JCR_SECTION is defined to 1 if both
11267 SUPPORTS_WEAK and TARGET_HAVE_NAMED_SECTIONS are true, else 0.
11271 This macro determines the size of the objective C jump buffer for the
11272 NeXT runtime. By default, OBJC_JBLEN is defined to an innocuous value.
11275 @defmac LIBGCC2_UNWIND_ATTRIBUTE
11276 Define this macro if any target-specific attributes need to be attached
11277 to the functions in @file{libgcc} that provide low-level support for
11278 call stack unwinding. It is used in declarations in @file{unwind-generic.h}
11279 and the associated definitions of those functions.
11282 @deftypefn {Target Hook} void TARGET_UPDATE_STACK_BOUNDARY (void)
11283 Define this macro to update the current function stack boundary if
11287 @deftypefn {Target Hook} rtx TARGET_GET_DRAP_RTX (void)
11288 This hook should return an rtx for Dynamic Realign Argument Pointer (DRAP) if a
11289 different argument pointer register is needed to access the function's
11290 argument list due to stack realignment. Return @code{NULL} if no DRAP
11294 @deftypefn {Target Hook} bool TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS (void)
11295 When optimization is disabled, this hook indicates whether or not
11296 arguments should be allocated to stack slots. Normally, GCC allocates
11297 stacks slots for arguments when not optimizing in order to make
11298 debugging easier. However, when a function is declared with
11299 @code{__attribute__((naked))}, there is no stack frame, and the compiler
11300 cannot safely move arguments from the registers in which they are passed
11301 to the stack. Therefore, this hook should return true in general, but
11302 false for naked functions. The default implementation always returns true.
11305 @deftypevr {Target Hook} {unsigned HOST_WIDE_INT} TARGET_CONST_ANCHOR
11306 On some architectures it can take multiple instructions to synthesize
11307 a constant. If there is another constant already in a register that
11308 is close enough in value then it is preferable that the new constant
11309 is computed from this register using immediate addition or
11310 subtraction. We accomplish this through CSE. Besides the value of
11311 the constant we also add a lower and an upper constant anchor to the
11312 available expressions. These are then queried when encountering new
11313 constants. The anchors are computed by rounding the constant up and
11314 down to a multiple of the value of @code{TARGET_CONST_ANCHOR}.
11315 @code{TARGET_CONST_ANCHOR} should be the maximum positive value
11316 accepted by immediate-add plus one. We currently assume that the
11317 value of @code{TARGET_CONST_ANCHOR} is a power of 2. For example, on
11318 MIPS, where add-immediate takes a 16-bit signed value,
11319 @code{TARGET_CONST_ANCHOR} is set to @samp{0x8000}. The default value
11320 is zero, which disables this optimization. @end deftypevr