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 @hook TARGET_ALWAYS_STRIP_DOTDOT
398 @defmac MULTILIB_DEFAULTS
399 Define this macro as a C expression for the initializer of an array of
400 string to tell the driver program which options are defaults for this
401 target and thus do not need to be handled specially when using
402 @code{MULTILIB_OPTIONS}.
404 Do not define this macro if @code{MULTILIB_OPTIONS} is not defined in
405 the target makefile fragment or if none of the options listed in
406 @code{MULTILIB_OPTIONS} are set by default.
407 @xref{Target Fragment}.
410 @defmac RELATIVE_PREFIX_NOT_LINKDIR
411 Define this macro to tell @command{gcc} that it should only translate
412 a @option{-B} prefix into a @option{-L} linker option if the prefix
413 indicates an absolute file name.
416 @defmac MD_EXEC_PREFIX
417 If defined, this macro is an additional prefix to try after
418 @code{STANDARD_EXEC_PREFIX}. @code{MD_EXEC_PREFIX} is not searched
419 when the compiler is built as a cross
420 compiler. If you define @code{MD_EXEC_PREFIX}, then be sure to add it
421 to the list of directories used to find the assembler in @file{configure.in}.
424 @defmac STANDARD_STARTFILE_PREFIX
425 Define this macro as a C string constant if you wish to override the
426 standard choice of @code{libdir} as the default prefix to
427 try when searching for startup files such as @file{crt0.o}.
428 @code{STANDARD_STARTFILE_PREFIX} is not searched when the compiler
429 is built as a cross compiler.
432 @defmac STANDARD_STARTFILE_PREFIX_1
433 Define this macro as a C string constant if you wish to override the
434 standard choice of @code{/lib} as a prefix to try after the default prefix
435 when searching for startup files such as @file{crt0.o}.
436 @code{STANDARD_STARTFILE_PREFIX_1} is not searched when the compiler
437 is built as a cross compiler.
440 @defmac STANDARD_STARTFILE_PREFIX_2
441 Define this macro as a C string constant if you wish to override the
442 standard choice of @code{/lib} as yet another prefix to try after the
443 default prefix when searching for startup files such as @file{crt0.o}.
444 @code{STANDARD_STARTFILE_PREFIX_2} is not searched when the compiler
445 is built as a cross compiler.
448 @defmac MD_STARTFILE_PREFIX
449 If defined, this macro supplies an additional prefix to try after the
450 standard prefixes. @code{MD_EXEC_PREFIX} is not searched when the
451 compiler is built as a cross compiler.
454 @defmac MD_STARTFILE_PREFIX_1
455 If defined, this macro supplies yet another prefix to try after the
456 standard prefixes. It is not searched when the compiler is built as a
460 @defmac INIT_ENVIRONMENT
461 Define this macro as a C string constant if you wish to set environment
462 variables for programs called by the driver, such as the assembler and
463 loader. The driver passes the value of this macro to @code{putenv} to
464 initialize the necessary environment variables.
467 @defmac LOCAL_INCLUDE_DIR
468 Define this macro as a C string constant if you wish to override the
469 standard choice of @file{/usr/local/include} as the default prefix to
470 try when searching for local header files. @code{LOCAL_INCLUDE_DIR}
471 comes before @code{SYSTEM_INCLUDE_DIR} in the search order.
473 Cross compilers do not search either @file{/usr/local/include} or its
477 @defmac SYSTEM_INCLUDE_DIR
478 Define this macro as a C string constant if you wish to specify a
479 system-specific directory to search for header files before the standard
480 directory. @code{SYSTEM_INCLUDE_DIR} comes before
481 @code{STANDARD_INCLUDE_DIR} in the search order.
483 Cross compilers do not use this macro and do not search the directory
487 @defmac STANDARD_INCLUDE_DIR
488 Define this macro as a C string constant if you wish to override the
489 standard choice of @file{/usr/include} as the default prefix to
490 try when searching for header files.
492 Cross compilers ignore this macro and do not search either
493 @file{/usr/include} or its replacement.
496 @defmac STANDARD_INCLUDE_COMPONENT
497 The ``component'' corresponding to @code{STANDARD_INCLUDE_DIR}.
498 See @code{INCLUDE_DEFAULTS}, below, for the description of components.
499 If you do not define this macro, no component is used.
502 @defmac INCLUDE_DEFAULTS
503 Define this macro if you wish to override the entire default search path
504 for include files. For a native compiler, the default search path
505 usually consists of @code{GCC_INCLUDE_DIR}, @code{LOCAL_INCLUDE_DIR},
506 @code{SYSTEM_INCLUDE_DIR}, @code{GPLUSPLUS_INCLUDE_DIR}, and
507 @code{STANDARD_INCLUDE_DIR}. In addition, @code{GPLUSPLUS_INCLUDE_DIR}
508 and @code{GCC_INCLUDE_DIR} are defined automatically by @file{Makefile},
509 and specify private search areas for GCC@. The directory
510 @code{GPLUSPLUS_INCLUDE_DIR} is used only for C++ programs.
512 The definition should be an initializer for an array of structures.
513 Each array element should have four elements: the directory name (a
514 string constant), the component name (also a string constant), a flag
515 for C++-only directories,
516 and a flag showing that the includes in the directory don't need to be
517 wrapped in @code{extern @samp{C}} when compiling C++. Mark the end of
518 the array with a null element.
520 The component name denotes what GNU package the include file is part of,
521 if any, in all uppercase letters. For example, it might be @samp{GCC}
522 or @samp{BINUTILS}. If the package is part of a vendor-supplied
523 operating system, code the component name as @samp{0}.
525 For example, here is the definition used for VAX/VMS:
528 #define INCLUDE_DEFAULTS \
530 @{ "GNU_GXX_INCLUDE:", "G++", 1, 1@}, \
531 @{ "GNU_CC_INCLUDE:", "GCC", 0, 0@}, \
532 @{ "SYS$SYSROOT:[SYSLIB.]", 0, 0, 0@}, \
539 Here is the order of prefixes tried for exec files:
543 Any prefixes specified by the user with @option{-B}.
546 The environment variable @code{GCC_EXEC_PREFIX} or, if @code{GCC_EXEC_PREFIX}
547 is not set and the compiler has not been installed in the configure-time
548 @var{prefix}, the location in which the compiler has actually been installed.
551 The directories specified by the environment variable @code{COMPILER_PATH}.
554 The macro @code{STANDARD_EXEC_PREFIX}, if the compiler has been installed
555 in the configured-time @var{prefix}.
558 The location @file{/usr/libexec/gcc/}, but only if this is a native compiler.
561 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
564 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
568 Here is the order of prefixes tried for startfiles:
572 Any prefixes specified by the user with @option{-B}.
575 The environment variable @code{GCC_EXEC_PREFIX} or its automatically determined
576 value based on the installed toolchain location.
579 The directories specified by the environment variable @code{LIBRARY_PATH}
580 (or port-specific name; native only, cross compilers do not use this).
583 The macro @code{STANDARD_EXEC_PREFIX}, but only if the toolchain is installed
584 in the configured @var{prefix} or this is a native compiler.
587 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
590 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
594 The macro @code{MD_STARTFILE_PREFIX}, if defined, but only if this is a
595 native compiler, or we have a target system root.
598 The macro @code{MD_STARTFILE_PREFIX_1}, if defined, but only if this is a
599 native compiler, or we have a target system root.
602 The macro @code{STANDARD_STARTFILE_PREFIX}, with any sysroot modifications.
603 If this path is relative it will be prefixed by @code{GCC_EXEC_PREFIX} and
604 the machine suffix or @code{STANDARD_EXEC_PREFIX} and the machine suffix.
607 The macro @code{STANDARD_STARTFILE_PREFIX_1}, but only if this is a native
608 compiler, or we have a target system root. The default for this macro is
612 The macro @code{STANDARD_STARTFILE_PREFIX_2}, but only if this is a native
613 compiler, or we have a target system root. The default for this macro is
617 @node Run-time Target
618 @section Run-time Target Specification
619 @cindex run-time target specification
620 @cindex predefined macros
621 @cindex target specifications
623 @c prevent bad page break with this line
624 Here are run-time target specifications.
626 @defmac TARGET_CPU_CPP_BUILTINS ()
627 This function-like macro expands to a block of code that defines
628 built-in preprocessor macros and assertions for the target CPU, using
629 the functions @code{builtin_define}, @code{builtin_define_std} and
630 @code{builtin_assert}. When the front end
631 calls this macro it provides a trailing semicolon, and since it has
632 finished command line option processing your code can use those
635 @code{builtin_assert} takes a string in the form you pass to the
636 command-line option @option{-A}, such as @code{cpu=mips}, and creates
637 the assertion. @code{builtin_define} takes a string in the form
638 accepted by option @option{-D} and unconditionally defines the macro.
640 @code{builtin_define_std} takes a string representing the name of an
641 object-like macro. If it doesn't lie in the user's namespace,
642 @code{builtin_define_std} defines it unconditionally. Otherwise, it
643 defines a version with two leading underscores, and another version
644 with two leading and trailing underscores, and defines the original
645 only if an ISO standard was not requested on the command line. For
646 example, passing @code{unix} defines @code{__unix}, @code{__unix__}
647 and possibly @code{unix}; passing @code{_mips} defines @code{__mips},
648 @code{__mips__} and possibly @code{_mips}, and passing @code{_ABI64}
649 defines only @code{_ABI64}.
651 You can also test for the C dialect being compiled. The variable
652 @code{c_language} is set to one of @code{clk_c}, @code{clk_cplusplus}
653 or @code{clk_objective_c}. Note that if we are preprocessing
654 assembler, this variable will be @code{clk_c} but the function-like
655 macro @code{preprocessing_asm_p()} will return true, so you might want
656 to check for that first. If you need to check for strict ANSI, the
657 variable @code{flag_iso} can be used. The function-like macro
658 @code{preprocessing_trad_p()} can be used to check for traditional
662 @defmac TARGET_OS_CPP_BUILTINS ()
663 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
664 and is used for the target operating system instead.
667 @defmac TARGET_OBJFMT_CPP_BUILTINS ()
668 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
669 and is used for the target object format. @file{elfos.h} uses this
670 macro to define @code{__ELF__}, so you probably do not need to define
674 @deftypevar {extern int} target_flags
675 This variable is declared in @file{options.h}, which is included before
676 any target-specific headers.
679 @hook TARGET_DEFAULT_TARGET_FLAGS
680 This variable specifies the initial value of @code{target_flags}.
681 Its default setting is 0.
684 @cindex optional hardware or system features
685 @cindex features, optional, in system conventions
687 @hook TARGET_HANDLE_OPTION
688 This hook is called whenever the user specifies one of the
689 target-specific options described by the @file{.opt} definition files
690 (@pxref{Options}). It has the opportunity to do some option-specific
691 processing and should return true if the option is valid. The default
692 definition does nothing but return true.
694 @var{decoded} specifies the option and its arguments. @var{opts} and
695 @var{opts_set} are the @code{gcc_options} structures to be used for
696 storing option state, and @var{loc} is the location at which the
697 option was passed (@code{UNKNOWN_LOCATION} except for options passed
701 @hook TARGET_HANDLE_C_OPTION
702 This target hook is called whenever the user specifies one of the
703 target-specific C language family options described by the @file{.opt}
704 definition files(@pxref{Options}). It has the opportunity to do some
705 option-specific processing and should return true if the option is
706 valid. The arguments are like for @code{TARGET_HANDLE_OPTION}. The
707 default definition does nothing but return false.
709 In general, you should use @code{TARGET_HANDLE_OPTION} to handle
710 options. However, if processing an option requires routines that are
711 only available in the C (and related language) front ends, then you
712 should use @code{TARGET_HANDLE_C_OPTION} instead.
715 @hook TARGET_OBJC_CONSTRUCT_STRING_OBJECT
717 @hook TARGET_STRING_OBJECT_REF_TYPE_P
719 @hook TARGET_CHECK_STRING_OBJECT_FORMAT_ARG
721 @hook TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE
722 This target function is similar to the hook @code{TARGET_OPTION_OVERRIDE}
723 but is called when the optimize level is changed via an attribute or
724 pragma or when it is reset at the end of the code affected by the
725 attribute or pragma. It is not called at the beginning of compilation
726 when @code{TARGET_OPTION_OVERRIDE} is called so if you want to perform these
727 actions then, you should have @code{TARGET_OPTION_OVERRIDE} call
728 @code{TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE}.
731 @defmac C_COMMON_OVERRIDE_OPTIONS
732 This is similar to the @code{TARGET_OPTION_OVERRIDE} hook
733 but is only used in the C
734 language frontends (C, Objective-C, C++, Objective-C++) and so can be
735 used to alter option flag variables which only exist in those
739 @hook TARGET_OPTION_OPTIMIZATION_TABLE
740 Some machines may desire to change what optimizations are performed for
741 various optimization levels. This variable, if defined, describes
742 options to enable at particular sets of optimization levels. These
743 options are processed once
744 just after the optimization level is determined and before the remainder
745 of the command options have been parsed, so may be overridden by other
746 options passed explicitly.
748 This processing is run once at program startup and when the optimization
749 options are changed via @code{#pragma GCC optimize} or by using the
750 @code{optimize} attribute.
753 @hook TARGET_OPTION_INIT_STRUCT
755 @hook TARGET_OPTION_DEFAULT_PARAMS
757 @defmac SWITCHABLE_TARGET
758 Some targets need to switch between substantially different subtargets
759 during compilation. For example, the MIPS target has one subtarget for
760 the traditional MIPS architecture and another for MIPS16. Source code
761 can switch between these two subarchitectures using the @code{mips16}
762 and @code{nomips16} attributes.
764 Such subtargets can differ in things like the set of available
765 registers, the set of available instructions, the costs of various
766 operations, and so on. GCC caches a lot of this type of information
767 in global variables, and recomputing them for each subtarget takes a
768 significant amount of time. The compiler therefore provides a facility
769 for maintaining several versions of the global variables and quickly
770 switching between them; see @file{target-globals.h} for details.
772 Define this macro to 1 if your target needs this facility. The default
776 @node Per-Function Data
777 @section Defining data structures for per-function information.
778 @cindex per-function data
779 @cindex data structures
781 If the target needs to store information on a per-function basis, GCC
782 provides a macro and a couple of variables to allow this. Note, just
783 using statics to store the information is a bad idea, since GCC supports
784 nested functions, so you can be halfway through encoding one function
785 when another one comes along.
787 GCC defines a data structure called @code{struct function} which
788 contains all of the data specific to an individual function. This
789 structure contains a field called @code{machine} whose type is
790 @code{struct machine_function *}, which can be used by targets to point
791 to their own specific data.
793 If a target needs per-function specific data it should define the type
794 @code{struct machine_function} and also the macro @code{INIT_EXPANDERS}.
795 This macro should be used to initialize the function pointer
796 @code{init_machine_status}. This pointer is explained below.
798 One typical use of per-function, target specific data is to create an
799 RTX to hold the register containing the function's return address. This
800 RTX can then be used to implement the @code{__builtin_return_address}
801 function, for level 0.
803 Note---earlier implementations of GCC used a single data area to hold
804 all of the per-function information. Thus when processing of a nested
805 function began the old per-function data had to be pushed onto a
806 stack, and when the processing was finished, it had to be popped off the
807 stack. GCC used to provide function pointers called
808 @code{save_machine_status} and @code{restore_machine_status} to handle
809 the saving and restoring of the target specific information. Since the
810 single data area approach is no longer used, these pointers are no
813 @defmac INIT_EXPANDERS
814 Macro called to initialize any target specific information. This macro
815 is called once per function, before generation of any RTL has begun.
816 The intention of this macro is to allow the initialization of the
817 function pointer @code{init_machine_status}.
820 @deftypevar {void (*)(struct function *)} init_machine_status
821 If this function pointer is non-@code{NULL} it will be called once per
822 function, before function compilation starts, in order to allow the
823 target to perform any target specific initialization of the
824 @code{struct function} structure. It is intended that this would be
825 used to initialize the @code{machine} of that structure.
827 @code{struct machine_function} structures are expected to be freed by GC@.
828 Generally, any memory that they reference must be allocated by using
829 GC allocation, including the structure itself.
833 @section Storage Layout
834 @cindex storage layout
836 Note that the definitions of the macros in this table which are sizes or
837 alignments measured in bits do not need to be constant. They can be C
838 expressions that refer to static variables, such as the @code{target_flags}.
839 @xref{Run-time Target}.
841 @defmac BITS_BIG_ENDIAN
842 Define this macro to have the value 1 if the most significant bit in a
843 byte has the lowest number; otherwise define it to have the value zero.
844 This means that bit-field instructions count from the most significant
845 bit. If the machine has no bit-field instructions, then this must still
846 be defined, but it doesn't matter which value it is defined to. This
847 macro need not be a constant.
849 This macro does not affect the way structure fields are packed into
850 bytes or words; that is controlled by @code{BYTES_BIG_ENDIAN}.
853 @defmac BYTES_BIG_ENDIAN
854 Define this macro to have the value 1 if the most significant byte in a
855 word has the lowest number. This macro need not be a constant.
858 @defmac WORDS_BIG_ENDIAN
859 Define this macro to have the value 1 if, in a multiword object, the
860 most significant word has the lowest number. This applies to both
861 memory locations and registers; see @code{REG_WORDS_BIG_ENDIAN} if the
862 order of words in memory is not the same as the order in registers. This
863 macro need not be a constant.
866 @defmac REG_WORDS_BIG_ENDIAN
867 On some machines, the order of words in a multiword object differs between
868 registers in memory. In such a situation, define this macro to describe
869 the order of words in a register. The macro @code{WORDS_BIG_ENDIAN} controls
870 the order of words in memory.
873 @defmac FLOAT_WORDS_BIG_ENDIAN
874 Define this macro to have the value 1 if @code{DFmode}, @code{XFmode} or
875 @code{TFmode} floating point numbers are stored in memory with the word
876 containing the sign bit at the lowest address; otherwise define it to
877 have the value 0. This macro need not be a constant.
879 You need not define this macro if the ordering is the same as for
883 @defmac BITS_PER_UNIT
884 Define this macro to be the number of bits in an addressable storage
885 unit (byte). If you do not define this macro the default is 8.
888 @defmac BITS_PER_WORD
889 Number of bits in a word. If you do not define this macro, the default
890 is @code{BITS_PER_UNIT * UNITS_PER_WORD}.
893 @defmac MAX_BITS_PER_WORD
894 Maximum number of bits in a word. If this is undefined, the default is
895 @code{BITS_PER_WORD}. Otherwise, it is the constant value that is the
896 largest value that @code{BITS_PER_WORD} can have at run-time.
899 @defmac UNITS_PER_WORD
900 Number of storage units in a word; normally the size of a general-purpose
901 register, a power of two from 1 or 8.
904 @defmac MIN_UNITS_PER_WORD
905 Minimum number of units in a word. If this is undefined, the default is
906 @code{UNITS_PER_WORD}. Otherwise, it is the constant value that is the
907 smallest value that @code{UNITS_PER_WORD} can have at run-time.
911 Width of a pointer, in bits. You must specify a value no wider than the
912 width of @code{Pmode}. If it is not equal to the width of @code{Pmode},
913 you must define @code{POINTERS_EXTEND_UNSIGNED}. If you do not specify
914 a value the default is @code{BITS_PER_WORD}.
917 @defmac POINTERS_EXTEND_UNSIGNED
918 A C expression that determines how pointers should be extended from
919 @code{ptr_mode} to either @code{Pmode} or @code{word_mode}. It is
920 greater than zero if pointers should be zero-extended, zero if they
921 should be sign-extended, and negative if some other sort of conversion
922 is needed. In the last case, the extension is done by the target's
923 @code{ptr_extend} instruction.
925 You need not define this macro if the @code{ptr_mode}, @code{Pmode}
926 and @code{word_mode} are all the same width.
929 @defmac PROMOTE_MODE (@var{m}, @var{unsignedp}, @var{type})
930 A macro to update @var{m} and @var{unsignedp} when an object whose type
931 is @var{type} and which has the specified mode and signedness is to be
932 stored in a register. This macro is only called when @var{type} is a
935 On most RISC machines, which only have operations that operate on a full
936 register, define this macro to set @var{m} to @code{word_mode} if
937 @var{m} is an integer mode narrower than @code{BITS_PER_WORD}. In most
938 cases, only integer modes should be widened because wider-precision
939 floating-point operations are usually more expensive than their narrower
942 For most machines, the macro definition does not change @var{unsignedp}.
943 However, some machines, have instructions that preferentially handle
944 either signed or unsigned quantities of certain modes. For example, on
945 the DEC Alpha, 32-bit loads from memory and 32-bit add instructions
946 sign-extend the result to 64 bits. On such machines, set
947 @var{unsignedp} according to which kind of extension is more efficient.
949 Do not define this macro if it would never modify @var{m}.
952 @hook TARGET_PROMOTE_FUNCTION_MODE
953 Like @code{PROMOTE_MODE}, but it is applied to outgoing function arguments or
954 function return values. The target hook should return the new mode
955 and possibly change @code{*@var{punsignedp}} if the promotion should
956 change signedness. This function is called only for scalar @emph{or
959 @var{for_return} allows to distinguish the promotion of arguments and
960 return values. If it is @code{1}, a return value is being promoted and
961 @code{TARGET_FUNCTION_VALUE} must perform the same promotions done here.
962 If it is @code{2}, the returned mode should be that of the register in
963 which an incoming parameter is copied, or the outgoing result is computed;
964 then the hook should return the same mode as @code{promote_mode}, though
965 the signedness may be different.
967 @var{type} can be NULL when promoting function arguments of libcalls.
969 The default is to not promote arguments and return values. You can
970 also define the hook to @code{default_promote_function_mode_always_promote}
971 if you would like to apply the same rules given by @code{PROMOTE_MODE}.
974 @defmac PARM_BOUNDARY
975 Normal alignment required for function parameters on the stack, in
976 bits. All stack parameters receive at least this much alignment
977 regardless of data type. On most machines, this is the same as the
981 @defmac STACK_BOUNDARY
982 Define this macro to the minimum alignment enforced by hardware for the
983 stack pointer on this machine. The definition is a C expression for the
984 desired alignment (measured in bits). This value is used as a default
985 if @code{PREFERRED_STACK_BOUNDARY} is not defined. On most machines,
986 this should be the same as @code{PARM_BOUNDARY}.
989 @defmac PREFERRED_STACK_BOUNDARY
990 Define this macro if you wish to preserve a certain alignment for the
991 stack pointer, greater than what the hardware enforces. The definition
992 is a C expression for the desired alignment (measured in bits). This
993 macro must evaluate to a value equal to or larger than
994 @code{STACK_BOUNDARY}.
997 @defmac INCOMING_STACK_BOUNDARY
998 Define this macro if the incoming stack boundary may be different
999 from @code{PREFERRED_STACK_BOUNDARY}. This macro must evaluate
1000 to a value equal to or larger than @code{STACK_BOUNDARY}.
1003 @defmac FUNCTION_BOUNDARY
1004 Alignment required for a function entry point, in bits.
1007 @defmac BIGGEST_ALIGNMENT
1008 Biggest alignment that any data type can require on this machine, in
1009 bits. Note that this is not the biggest alignment that is supported,
1010 just the biggest alignment that, when violated, may cause a fault.
1013 @defmac MALLOC_ABI_ALIGNMENT
1014 Alignment, in bits, a C conformant malloc implementation has to
1015 provide. If not defined, the default value is @code{BITS_PER_WORD}.
1018 @defmac ATTRIBUTE_ALIGNED_VALUE
1019 Alignment used by the @code{__attribute__ ((aligned))} construct. If
1020 not defined, the default value is @code{BIGGEST_ALIGNMENT}.
1023 @defmac MINIMUM_ATOMIC_ALIGNMENT
1024 If defined, the smallest alignment, in bits, that can be given to an
1025 object that can be referenced in one operation, without disturbing any
1026 nearby object. Normally, this is @code{BITS_PER_UNIT}, but may be larger
1027 on machines that don't have byte or half-word store operations.
1030 @defmac BIGGEST_FIELD_ALIGNMENT
1031 Biggest alignment that any structure or union field can require on this
1032 machine, in bits. If defined, this overrides @code{BIGGEST_ALIGNMENT} for
1033 structure and union fields only, unless the field alignment has been set
1034 by the @code{__attribute__ ((aligned (@var{n})))} construct.
1037 @defmac ADJUST_FIELD_ALIGN (@var{field}, @var{computed})
1038 An expression for the alignment of a structure field @var{field} if the
1039 alignment computed in the usual way (including applying of
1040 @code{BIGGEST_ALIGNMENT} and @code{BIGGEST_FIELD_ALIGNMENT} to the
1041 alignment) is @var{computed}. It overrides alignment only if the
1042 field alignment has not been set by the
1043 @code{__attribute__ ((aligned (@var{n})))} construct.
1046 @defmac MAX_STACK_ALIGNMENT
1047 Biggest stack alignment guaranteed by the backend. Use this macro
1048 to specify the maximum alignment of a variable on stack.
1050 If not defined, the default value is @code{STACK_BOUNDARY}.
1052 @c FIXME: The default should be @code{PREFERRED_STACK_BOUNDARY}.
1053 @c But the fix for PR 32893 indicates that we can only guarantee
1054 @c maximum stack alignment on stack up to @code{STACK_BOUNDARY}, not
1055 @c @code{PREFERRED_STACK_BOUNDARY}, if stack alignment isn't supported.
1058 @defmac MAX_OFILE_ALIGNMENT
1059 Biggest alignment supported by the object file format of this machine.
1060 Use this macro to limit the alignment which can be specified using the
1061 @code{__attribute__ ((aligned (@var{n})))} construct. If not defined,
1062 the default value is @code{BIGGEST_ALIGNMENT}.
1064 On systems that use ELF, the default (in @file{config/elfos.h}) is
1065 the largest supported 32-bit ELF section alignment representable on
1066 a 32-bit host e.g. @samp{(((unsigned HOST_WIDEST_INT) 1 << 28) * 8)}.
1067 On 32-bit ELF the largest supported section alignment in bits is
1068 @samp{(0x80000000 * 8)}, but this is not representable on 32-bit hosts.
1071 @defmac DATA_ALIGNMENT (@var{type}, @var{basic-align})
1072 If defined, a C expression to compute the alignment for a variable in
1073 the static store. @var{type} is the data type, and @var{basic-align} is
1074 the alignment that the object would ordinarily have. The value of this
1075 macro is used instead of that alignment to align the object.
1077 If this macro is not defined, then @var{basic-align} is used.
1080 One use of this macro is to increase alignment of medium-size data to
1081 make it all fit in fewer cache lines. Another is to cause character
1082 arrays to be word-aligned so that @code{strcpy} calls that copy
1083 constants to character arrays can be done inline.
1086 @defmac CONSTANT_ALIGNMENT (@var{constant}, @var{basic-align})
1087 If defined, a C expression to compute the alignment given to a constant
1088 that is being placed in memory. @var{constant} is the constant and
1089 @var{basic-align} is the alignment that the object would ordinarily
1090 have. The value of this macro is used instead of that alignment to
1093 If this macro is not defined, then @var{basic-align} is used.
1095 The typical use of this macro is to increase alignment for string
1096 constants to be word aligned so that @code{strcpy} calls that copy
1097 constants can be done inline.
1100 @defmac LOCAL_ALIGNMENT (@var{type}, @var{basic-align})
1101 If defined, a C expression to compute the alignment for a variable in
1102 the local store. @var{type} is the data type, and @var{basic-align} is
1103 the alignment that the object would ordinarily have. The value of this
1104 macro is used instead of that alignment to align the object.
1106 If this macro is not defined, then @var{basic-align} is used.
1108 One use of this macro is to increase alignment of medium-size data to
1109 make it all fit in fewer cache lines.
1111 If the value of this macro has a type, it should be an unsigned type.
1114 @defmac STACK_SLOT_ALIGNMENT (@var{type}, @var{mode}, @var{basic-align})
1115 If defined, a C expression to compute the alignment for stack slot.
1116 @var{type} is the data type, @var{mode} is the widest mode available,
1117 and @var{basic-align} is the alignment that the slot would ordinarily
1118 have. The value of this macro is used instead of that alignment to
1121 If this macro is not defined, then @var{basic-align} is used when
1122 @var{type} is @code{NULL}. Otherwise, @code{LOCAL_ALIGNMENT} will
1125 This macro is to set alignment of stack slot to the maximum alignment
1126 of all possible modes which the slot may have.
1128 If the value of this macro has a type, it should be an unsigned type.
1131 @defmac LOCAL_DECL_ALIGNMENT (@var{decl})
1132 If defined, a C expression to compute the alignment for a local
1133 variable @var{decl}.
1135 If this macro is not defined, then
1136 @code{LOCAL_ALIGNMENT (TREE_TYPE (@var{decl}), DECL_ALIGN (@var{decl}))}
1139 One use of this macro is to increase alignment of medium-size data to
1140 make it all fit in fewer cache lines.
1142 If the value of this macro has a type, it should be an unsigned type.
1145 @defmac MINIMUM_ALIGNMENT (@var{exp}, @var{mode}, @var{align})
1146 If defined, a C expression to compute the minimum required alignment
1147 for dynamic stack realignment purposes for @var{exp} (a type or decl),
1148 @var{mode}, assuming normal alignment @var{align}.
1150 If this macro is not defined, then @var{align} will be used.
1153 @defmac EMPTY_FIELD_BOUNDARY
1154 Alignment in bits to be given to a structure bit-field that follows an
1155 empty field such as @code{int : 0;}.
1157 If @code{PCC_BITFIELD_TYPE_MATTERS} is true, it overrides this macro.
1160 @defmac STRUCTURE_SIZE_BOUNDARY
1161 Number of bits which any structure or union's size must be a multiple of.
1162 Each structure or union's size is rounded up to a multiple of this.
1164 If you do not define this macro, the default is the same as
1165 @code{BITS_PER_UNIT}.
1168 @defmac STRICT_ALIGNMENT
1169 Define this macro to be the value 1 if instructions will fail to work
1170 if given data not on the nominal alignment. If instructions will merely
1171 go slower in that case, define this macro as 0.
1174 @defmac PCC_BITFIELD_TYPE_MATTERS
1175 Define this if you wish to imitate the way many other C compilers handle
1176 alignment of bit-fields and the structures that contain them.
1178 The behavior is that the type written for a named bit-field (@code{int},
1179 @code{short}, or other integer type) imposes an alignment for the entire
1180 structure, as if the structure really did contain an ordinary field of
1181 that type. In addition, the bit-field is placed within the structure so
1182 that it would fit within such a field, not crossing a boundary for it.
1184 Thus, on most machines, a named bit-field whose type is written as
1185 @code{int} would not cross a four-byte boundary, and would force
1186 four-byte alignment for the whole structure. (The alignment used may
1187 not be four bytes; it is controlled by the other alignment parameters.)
1189 An unnamed bit-field will not affect the alignment of the containing
1192 If the macro is defined, its definition should be a C expression;
1193 a nonzero value for the expression enables this behavior.
1195 Note that if this macro is not defined, or its value is zero, some
1196 bit-fields may cross more than one alignment boundary. The compiler can
1197 support such references if there are @samp{insv}, @samp{extv}, and
1198 @samp{extzv} insns that can directly reference memory.
1200 The other known way of making bit-fields work is to define
1201 @code{STRUCTURE_SIZE_BOUNDARY} as large as @code{BIGGEST_ALIGNMENT}.
1202 Then every structure can be accessed with fullwords.
1204 Unless the machine has bit-field instructions or you define
1205 @code{STRUCTURE_SIZE_BOUNDARY} that way, you must define
1206 @code{PCC_BITFIELD_TYPE_MATTERS} to have a nonzero value.
1208 If your aim is to make GCC use the same conventions for laying out
1209 bit-fields as are used by another compiler, here is how to investigate
1210 what the other compiler does. Compile and run this program:
1229 printf ("Size of foo1 is %d\n",
1230 sizeof (struct foo1));
1231 printf ("Size of foo2 is %d\n",
1232 sizeof (struct foo2));
1237 If this prints 2 and 5, then the compiler's behavior is what you would
1238 get from @code{PCC_BITFIELD_TYPE_MATTERS}.
1241 @defmac BITFIELD_NBYTES_LIMITED
1242 Like @code{PCC_BITFIELD_TYPE_MATTERS} except that its effect is limited
1243 to aligning a bit-field within the structure.
1246 @hook TARGET_ALIGN_ANON_BITFIELD
1247 When @code{PCC_BITFIELD_TYPE_MATTERS} is true this hook will determine
1248 whether unnamed bitfields affect the alignment of the containing
1249 structure. The hook should return true if the structure should inherit
1250 the alignment requirements of an unnamed bitfield's type.
1253 @hook TARGET_NARROW_VOLATILE_BITFIELD
1254 This target hook should return @code{true} if accesses to volatile bitfields
1255 should use the narrowest mode possible. It should return @code{false} if
1256 these accesses should use the bitfield container type.
1258 The default is @code{!TARGET_STRICT_ALIGN}.
1261 @defmac MEMBER_TYPE_FORCES_BLK (@var{field}, @var{mode})
1262 Return 1 if a structure or array containing @var{field} should be accessed using
1265 If @var{field} is the only field in the structure, @var{mode} is its
1266 mode, otherwise @var{mode} is VOIDmode. @var{mode} is provided in the
1267 case where structures of one field would require the structure's mode to
1268 retain the field's mode.
1270 Normally, this is not needed.
1273 @defmac ROUND_TYPE_ALIGN (@var{type}, @var{computed}, @var{specified})
1274 Define this macro as an expression for the alignment of a type (given
1275 by @var{type} as a tree node) if the alignment computed in the usual
1276 way is @var{computed} and the alignment explicitly specified was
1279 The default is to use @var{specified} if it is larger; otherwise, use
1280 the smaller of @var{computed} and @code{BIGGEST_ALIGNMENT}
1283 @defmac MAX_FIXED_MODE_SIZE
1284 An integer expression for the size in bits of the largest integer
1285 machine mode that should actually be used. All integer machine modes of
1286 this size or smaller can be used for structures and unions with the
1287 appropriate sizes. If this macro is undefined, @code{GET_MODE_BITSIZE
1288 (DImode)} is assumed.
1291 @defmac STACK_SAVEAREA_MODE (@var{save_level})
1292 If defined, an expression of type @code{enum machine_mode} that
1293 specifies the mode of the save area operand of a
1294 @code{save_stack_@var{level}} named pattern (@pxref{Standard Names}).
1295 @var{save_level} is one of @code{SAVE_BLOCK}, @code{SAVE_FUNCTION}, or
1296 @code{SAVE_NONLOCAL} and selects which of the three named patterns is
1297 having its mode specified.
1299 You need not define this macro if it always returns @code{Pmode}. You
1300 would most commonly define this macro if the
1301 @code{save_stack_@var{level}} patterns need to support both a 32- and a
1305 @defmac STACK_SIZE_MODE
1306 If defined, an expression of type @code{enum machine_mode} that
1307 specifies the mode of the size increment operand of an
1308 @code{allocate_stack} named pattern (@pxref{Standard Names}).
1310 You need not define this macro if it always returns @code{word_mode}.
1311 You would most commonly define this macro if the @code{allocate_stack}
1312 pattern needs to support both a 32- and a 64-bit mode.
1315 @hook TARGET_LIBGCC_CMP_RETURN_MODE
1316 This target hook should return the mode to be used for the return value
1317 of compare instructions expanded to libgcc calls. If not defined
1318 @code{word_mode} is returned which is the right choice for a majority of
1322 @hook TARGET_LIBGCC_SHIFT_COUNT_MODE
1323 This target hook should return the mode to be used for the shift count operand
1324 of shift instructions expanded to libgcc calls. If not defined
1325 @code{word_mode} is returned which is the right choice for a majority of
1329 @hook TARGET_UNWIND_WORD_MODE
1330 Return machine mode to be used for @code{_Unwind_Word} type.
1331 The default is to use @code{word_mode}.
1334 @defmac ROUND_TOWARDS_ZERO
1335 If defined, this macro should be true if the prevailing rounding
1336 mode is towards zero.
1338 Defining this macro only affects the way @file{libgcc.a} emulates
1339 floating-point arithmetic.
1341 Not defining this macro is equivalent to returning zero.
1344 @defmac LARGEST_EXPONENT_IS_NORMAL (@var{size})
1345 This macro should return true if floats with @var{size}
1346 bits do not have a NaN or infinity representation, but use the largest
1347 exponent for normal numbers instead.
1349 Defining this macro only affects the way @file{libgcc.a} emulates
1350 floating-point arithmetic.
1352 The default definition of this macro returns false for all sizes.
1355 @hook TARGET_MS_BITFIELD_LAYOUT_P
1356 This target hook returns @code{true} if bit-fields in the given
1357 @var{record_type} are to be laid out following the rules of Microsoft
1358 Visual C/C++, namely: (i) a bit-field won't share the same storage
1359 unit with the previous bit-field if their underlying types have
1360 different sizes, and the bit-field will be aligned to the highest
1361 alignment of the underlying types of itself and of the previous
1362 bit-field; (ii) a zero-sized bit-field will affect the alignment of
1363 the whole enclosing structure, even if it is unnamed; except that
1364 (iii) a zero-sized bit-field will be disregarded unless it follows
1365 another bit-field of nonzero size. If this hook returns @code{true},
1366 other macros that control bit-field layout are ignored.
1368 When a bit-field is inserted into a packed record, the whole size
1369 of the underlying type is used by one or more same-size adjacent
1370 bit-fields (that is, if its long:3, 32 bits is used in the record,
1371 and any additional adjacent long bit-fields are packed into the same
1372 chunk of 32 bits. However, if the size changes, a new field of that
1373 size is allocated). In an unpacked record, this is the same as using
1374 alignment, but not equivalent when packing.
1376 If both MS bit-fields and @samp{__attribute__((packed))} are used,
1377 the latter will take precedence. If @samp{__attribute__((packed))} is
1378 used on a single field when MS bit-fields are in use, it will take
1379 precedence for that field, but the alignment of the rest of the structure
1380 may affect its placement.
1383 @hook TARGET_DECIMAL_FLOAT_SUPPORTED_P
1384 Returns true if the target supports decimal floating point.
1387 @hook TARGET_FIXED_POINT_SUPPORTED_P
1388 Returns true if the target supports fixed-point arithmetic.
1391 @hook TARGET_EXPAND_TO_RTL_HOOK
1392 This hook is called just before expansion into rtl, allowing the target
1393 to perform additional initializations or analysis before the expansion.
1394 For example, the rs6000 port uses it to allocate a scratch stack slot
1395 for use in copying SDmode values between memory and floating point
1396 registers whenever the function being expanded has any SDmode
1400 @hook TARGET_INSTANTIATE_DECLS
1401 This hook allows the backend to perform additional instantiations on rtl
1402 that are not actually in any insns yet, but will be later.
1405 @hook TARGET_MANGLE_TYPE
1406 If your target defines any fundamental types, or any types your target
1407 uses should be mangled differently from the default, define this hook
1408 to return the appropriate encoding for these types as part of a C++
1409 mangled name. The @var{type} argument is the tree structure representing
1410 the type to be mangled. The hook may be applied to trees which are
1411 not target-specific fundamental types; it should return @code{NULL}
1412 for all such types, as well as arguments it does not recognize. If the
1413 return value is not @code{NULL}, it must point to a statically-allocated
1416 Target-specific fundamental types might be new fundamental types or
1417 qualified versions of ordinary fundamental types. Encode new
1418 fundamental types as @samp{@w{u @var{n} @var{name}}}, where @var{name}
1419 is the name used for the type in source code, and @var{n} is the
1420 length of @var{name} in decimal. Encode qualified versions of
1421 ordinary types as @samp{@w{U @var{n} @var{name} @var{code}}}, where
1422 @var{name} is the name used for the type qualifier in source code,
1423 @var{n} is the length of @var{name} as above, and @var{code} is the
1424 code used to represent the unqualified version of this type. (See
1425 @code{write_builtin_type} in @file{cp/mangle.c} for the list of
1426 codes.) In both cases the spaces are for clarity; do not include any
1427 spaces in your string.
1429 This hook is applied to types prior to typedef resolution. If the mangled
1430 name for a particular type depends only on that type's main variant, you
1431 can perform typedef resolution yourself using @code{TYPE_MAIN_VARIANT}
1434 The default version of this hook always returns @code{NULL}, which is
1435 appropriate for a target that does not define any new fundamental
1440 @section Layout of Source Language Data Types
1442 These macros define the sizes and other characteristics of the standard
1443 basic data types used in programs being compiled. Unlike the macros in
1444 the previous section, these apply to specific features of C and related
1445 languages, rather than to fundamental aspects of storage layout.
1447 @defmac INT_TYPE_SIZE
1448 A C expression for the size in bits of the type @code{int} on the
1449 target machine. If you don't define this, the default is one word.
1452 @defmac SHORT_TYPE_SIZE
1453 A C expression for the size in bits of the type @code{short} on the
1454 target machine. If you don't define this, the default is half a word.
1455 (If this would be less than one storage unit, it is rounded up to one
1459 @defmac LONG_TYPE_SIZE
1460 A C expression for the size in bits of the type @code{long} on the
1461 target machine. If you don't define this, the default is one word.
1464 @defmac ADA_LONG_TYPE_SIZE
1465 On some machines, the size used for the Ada equivalent of the type
1466 @code{long} by a native Ada compiler differs from that used by C@. In
1467 that situation, define this macro to be a C expression to be used for
1468 the size of that type. If you don't define this, the default is the
1469 value of @code{LONG_TYPE_SIZE}.
1472 @defmac LONG_LONG_TYPE_SIZE
1473 A C expression for the size in bits of the type @code{long long} on the
1474 target machine. If you don't define this, the default is two
1475 words. If you want to support GNU Ada on your machine, the value of this
1476 macro must be at least 64.
1479 @defmac CHAR_TYPE_SIZE
1480 A C expression for the size in bits of the type @code{char} on the
1481 target machine. If you don't define this, the default is
1482 @code{BITS_PER_UNIT}.
1485 @defmac BOOL_TYPE_SIZE
1486 A C expression for the size in bits of the C++ type @code{bool} and
1487 C99 type @code{_Bool} on the target machine. If you don't define
1488 this, and you probably shouldn't, the default is @code{CHAR_TYPE_SIZE}.
1491 @defmac FLOAT_TYPE_SIZE
1492 A C expression for the size in bits of the type @code{float} on the
1493 target machine. If you don't define this, the default is one word.
1496 @defmac DOUBLE_TYPE_SIZE
1497 A C expression for the size in bits of the type @code{double} on the
1498 target machine. If you don't define this, the default is two
1502 @defmac LONG_DOUBLE_TYPE_SIZE
1503 A C expression for the size in bits of the type @code{long double} on
1504 the target machine. If you don't define this, the default is two
1508 @defmac SHORT_FRACT_TYPE_SIZE
1509 A C expression for the size in bits of the type @code{short _Fract} on
1510 the target machine. If you don't define this, the default is
1511 @code{BITS_PER_UNIT}.
1514 @defmac FRACT_TYPE_SIZE
1515 A C expression for the size in bits of the type @code{_Fract} on
1516 the target machine. If you don't define this, the default is
1517 @code{BITS_PER_UNIT * 2}.
1520 @defmac LONG_FRACT_TYPE_SIZE
1521 A C expression for the size in bits of the type @code{long _Fract} on
1522 the target machine. If you don't define this, the default is
1523 @code{BITS_PER_UNIT * 4}.
1526 @defmac LONG_LONG_FRACT_TYPE_SIZE
1527 A C expression for the size in bits of the type @code{long long _Fract} on
1528 the target machine. If you don't define this, the default is
1529 @code{BITS_PER_UNIT * 8}.
1532 @defmac SHORT_ACCUM_TYPE_SIZE
1533 A C expression for the size in bits of the type @code{short _Accum} on
1534 the target machine. If you don't define this, the default is
1535 @code{BITS_PER_UNIT * 2}.
1538 @defmac ACCUM_TYPE_SIZE
1539 A C expression for the size in bits of the type @code{_Accum} on
1540 the target machine. If you don't define this, the default is
1541 @code{BITS_PER_UNIT * 4}.
1544 @defmac LONG_ACCUM_TYPE_SIZE
1545 A C expression for the size in bits of the type @code{long _Accum} on
1546 the target machine. If you don't define this, the default is
1547 @code{BITS_PER_UNIT * 8}.
1550 @defmac LONG_LONG_ACCUM_TYPE_SIZE
1551 A C expression for the size in bits of the type @code{long long _Accum} on
1552 the target machine. If you don't define this, the default is
1553 @code{BITS_PER_UNIT * 16}.
1556 @defmac LIBGCC2_LONG_DOUBLE_TYPE_SIZE
1557 Define this macro if @code{LONG_DOUBLE_TYPE_SIZE} is not constant or
1558 if you want routines in @file{libgcc2.a} for a size other than
1559 @code{LONG_DOUBLE_TYPE_SIZE}. If you don't define this, the
1560 default is @code{LONG_DOUBLE_TYPE_SIZE}.
1563 @defmac LIBGCC2_HAS_DF_MODE
1564 Define this macro if neither @code{DOUBLE_TYPE_SIZE} nor
1565 @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is
1566 @code{DFmode} but you want @code{DFmode} routines in @file{libgcc2.a}
1567 anyway. If you don't define this and either @code{DOUBLE_TYPE_SIZE}
1568 or @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is 64 then the default is 1,
1572 @defmac LIBGCC2_HAS_XF_MODE
1573 Define this macro if @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is not
1574 @code{XFmode} but you want @code{XFmode} routines in @file{libgcc2.a}
1575 anyway. If you don't define this and @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE}
1576 is 80 then the default is 1, otherwise it is 0.
1579 @defmac LIBGCC2_HAS_TF_MODE
1580 Define this macro if @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is not
1581 @code{TFmode} but you want @code{TFmode} routines in @file{libgcc2.a}
1582 anyway. If you don't define this and @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE}
1583 is 128 then the default is 1, otherwise it is 0.
1586 @defmac LIBGCC2_GNU_PREFIX
1587 This macro corresponds to the @code{TARGET_LIBFUNC_GNU_PREFIX} target
1588 hook and should be defined if that hook is overriden to be true. It
1589 causes function names in libgcc to be changed to use a @code{__gnu_}
1590 prefix for their name rather than the default @code{__}. A port which
1591 uses this macro should also arrange to use @file{t-gnu-prefix} in
1592 the libgcc @file{config.host}.
1599 Define these macros to be the size in bits of the mantissa of
1600 @code{SFmode}, @code{DFmode}, @code{XFmode} and @code{TFmode} values,
1601 if the defaults in @file{libgcc2.h} are inappropriate. By default,
1602 @code{FLT_MANT_DIG} is used for @code{SF_SIZE}, @code{LDBL_MANT_DIG}
1603 for @code{XF_SIZE} and @code{TF_SIZE}, and @code{DBL_MANT_DIG} or
1604 @code{LDBL_MANT_DIG} for @code{DF_SIZE} according to whether
1605 @code{DOUBLE_TYPE_SIZE} or
1606 @code{LIBGCC2_LONG_DOUBLE_TYPE_SIZE} is 64.
1609 @defmac TARGET_FLT_EVAL_METHOD
1610 A C expression for the value for @code{FLT_EVAL_METHOD} in @file{float.h},
1611 assuming, if applicable, that the floating-point control word is in its
1612 default state. If you do not define this macro the value of
1613 @code{FLT_EVAL_METHOD} will be zero.
1616 @defmac WIDEST_HARDWARE_FP_SIZE
1617 A C expression for the size in bits of the widest floating-point format
1618 supported by the hardware. If you define this macro, you must specify a
1619 value less than or equal to the value of @code{LONG_DOUBLE_TYPE_SIZE}.
1620 If you do not define this macro, the value of @code{LONG_DOUBLE_TYPE_SIZE}
1624 @defmac DEFAULT_SIGNED_CHAR
1625 An expression whose value is 1 or 0, according to whether the type
1626 @code{char} should be signed or unsigned by default. The user can
1627 always override this default with the options @option{-fsigned-char}
1628 and @option{-funsigned-char}.
1631 @hook TARGET_DEFAULT_SHORT_ENUMS
1632 This target hook should return true if the compiler should give an
1633 @code{enum} type only as many bytes as it takes to represent the range
1634 of possible values of that type. It should return false if all
1635 @code{enum} types should be allocated like @code{int}.
1637 The default is to return false.
1641 A C expression for a string describing the name of the data type to use
1642 for size values. The typedef name @code{size_t} is defined using the
1643 contents of the string.
1645 The string can contain more than one keyword. If so, separate them with
1646 spaces, and write first any length keyword, then @code{unsigned} if
1647 appropriate, and finally @code{int}. The string must exactly match one
1648 of the data type names defined in the function
1649 @code{init_decl_processing} in the file @file{c-decl.c}. You may not
1650 omit @code{int} or change the order---that would cause the compiler to
1653 If you don't define this macro, the default is @code{"long unsigned
1657 @defmac PTRDIFF_TYPE
1658 A C expression for a string describing the name of the data type to use
1659 for the result of subtracting two pointers. The typedef name
1660 @code{ptrdiff_t} is defined using the contents of the string. See
1661 @code{SIZE_TYPE} above for more information.
1663 If you don't define this macro, the default is @code{"long int"}.
1667 A C expression for a string describing the name of the data type to use
1668 for wide characters. The typedef name @code{wchar_t} is defined using
1669 the contents of the string. See @code{SIZE_TYPE} above for more
1672 If you don't define this macro, the default is @code{"int"}.
1675 @defmac WCHAR_TYPE_SIZE
1676 A C expression for the size in bits of the data type for wide
1677 characters. This is used in @code{cpp}, which cannot make use of
1682 A C expression for a string describing the name of the data type to
1683 use for wide characters passed to @code{printf} and returned from
1684 @code{getwc}. The typedef name @code{wint_t} is defined using the
1685 contents of the string. See @code{SIZE_TYPE} above for more
1688 If you don't define this macro, the default is @code{"unsigned int"}.
1692 A C expression for a string describing the name of the data type that
1693 can represent any value of any standard or extended signed integer type.
1694 The typedef name @code{intmax_t} is defined using the contents of the
1695 string. See @code{SIZE_TYPE} above for more information.
1697 If you don't define this macro, the default is the first of
1698 @code{"int"}, @code{"long int"}, or @code{"long long int"} that has as
1699 much precision as @code{long long int}.
1702 @defmac UINTMAX_TYPE
1703 A C expression for a string describing the name of the data type that
1704 can represent any value of any standard or extended unsigned integer
1705 type. The typedef name @code{uintmax_t} is defined using the contents
1706 of the string. See @code{SIZE_TYPE} above for more information.
1708 If you don't define this macro, the default is the first of
1709 @code{"unsigned int"}, @code{"long unsigned int"}, or @code{"long long
1710 unsigned int"} that has as much precision as @code{long long unsigned
1714 @defmac SIG_ATOMIC_TYPE
1720 @defmacx UINT16_TYPE
1721 @defmacx UINT32_TYPE
1722 @defmacx UINT64_TYPE
1723 @defmacx INT_LEAST8_TYPE
1724 @defmacx INT_LEAST16_TYPE
1725 @defmacx INT_LEAST32_TYPE
1726 @defmacx INT_LEAST64_TYPE
1727 @defmacx UINT_LEAST8_TYPE
1728 @defmacx UINT_LEAST16_TYPE
1729 @defmacx UINT_LEAST32_TYPE
1730 @defmacx UINT_LEAST64_TYPE
1731 @defmacx INT_FAST8_TYPE
1732 @defmacx INT_FAST16_TYPE
1733 @defmacx INT_FAST32_TYPE
1734 @defmacx INT_FAST64_TYPE
1735 @defmacx UINT_FAST8_TYPE
1736 @defmacx UINT_FAST16_TYPE
1737 @defmacx UINT_FAST32_TYPE
1738 @defmacx UINT_FAST64_TYPE
1739 @defmacx INTPTR_TYPE
1740 @defmacx UINTPTR_TYPE
1741 C expressions for the standard types @code{sig_atomic_t},
1742 @code{int8_t}, @code{int16_t}, @code{int32_t}, @code{int64_t},
1743 @code{uint8_t}, @code{uint16_t}, @code{uint32_t}, @code{uint64_t},
1744 @code{int_least8_t}, @code{int_least16_t}, @code{int_least32_t},
1745 @code{int_least64_t}, @code{uint_least8_t}, @code{uint_least16_t},
1746 @code{uint_least32_t}, @code{uint_least64_t}, @code{int_fast8_t},
1747 @code{int_fast16_t}, @code{int_fast32_t}, @code{int_fast64_t},
1748 @code{uint_fast8_t}, @code{uint_fast16_t}, @code{uint_fast32_t},
1749 @code{uint_fast64_t}, @code{intptr_t}, and @code{uintptr_t}. See
1750 @code{SIZE_TYPE} above for more information.
1752 If any of these macros evaluates to a null pointer, the corresponding
1753 type is not supported; if GCC is configured to provide
1754 @code{<stdint.h>} in such a case, the header provided may not conform
1755 to C99, depending on the type in question. The defaults for all of
1756 these macros are null pointers.
1759 @defmac TARGET_PTRMEMFUNC_VBIT_LOCATION
1760 The C++ compiler represents a pointer-to-member-function with a struct
1767 ptrdiff_t vtable_index;
1774 The C++ compiler must use one bit to indicate whether the function that
1775 will be called through a pointer-to-member-function is virtual.
1776 Normally, we assume that the low-order bit of a function pointer must
1777 always be zero. Then, by ensuring that the vtable_index is odd, we can
1778 distinguish which variant of the union is in use. But, on some
1779 platforms function pointers can be odd, and so this doesn't work. In
1780 that case, we use the low-order bit of the @code{delta} field, and shift
1781 the remainder of the @code{delta} field to the left.
1783 GCC will automatically make the right selection about where to store
1784 this bit using the @code{FUNCTION_BOUNDARY} setting for your platform.
1785 However, some platforms such as ARM/Thumb have @code{FUNCTION_BOUNDARY}
1786 set such that functions always start at even addresses, but the lowest
1787 bit of pointers to functions indicate whether the function at that
1788 address is in ARM or Thumb mode. If this is the case of your
1789 architecture, you should define this macro to
1790 @code{ptrmemfunc_vbit_in_delta}.
1792 In general, you should not have to define this macro. On architectures
1793 in which function addresses are always even, according to
1794 @code{FUNCTION_BOUNDARY}, GCC will automatically define this macro to
1795 @code{ptrmemfunc_vbit_in_pfn}.
1798 @defmac TARGET_VTABLE_USES_DESCRIPTORS
1799 Normally, the C++ compiler uses function pointers in vtables. This
1800 macro allows the target to change to use ``function descriptors''
1801 instead. Function descriptors are found on targets for whom a
1802 function pointer is actually a small data structure. Normally the
1803 data structure consists of the actual code address plus a data
1804 pointer to which the function's data is relative.
1806 If vtables are used, the value of this macro should be the number
1807 of words that the function descriptor occupies.
1810 @defmac TARGET_VTABLE_ENTRY_ALIGN
1811 By default, the vtable entries are void pointers, the so the alignment
1812 is the same as pointer alignment. The value of this macro specifies
1813 the alignment of the vtable entry in bits. It should be defined only
1814 when special alignment is necessary. */
1817 @defmac TARGET_VTABLE_DATA_ENTRY_DISTANCE
1818 There are a few non-descriptor entries in the vtable at offsets below
1819 zero. If these entries must be padded (say, to preserve the alignment
1820 specified by @code{TARGET_VTABLE_ENTRY_ALIGN}), set this to the number
1821 of words in each data entry.
1825 @section Register Usage
1826 @cindex register usage
1828 This section explains how to describe what registers the target machine
1829 has, and how (in general) they can be used.
1831 The description of which registers a specific instruction can use is
1832 done with register classes; see @ref{Register Classes}. For information
1833 on using registers to access a stack frame, see @ref{Frame Registers}.
1834 For passing values in registers, see @ref{Register Arguments}.
1835 For returning values in registers, see @ref{Scalar Return}.
1838 * Register Basics:: Number and kinds of registers.
1839 * Allocation Order:: Order in which registers are allocated.
1840 * Values in Registers:: What kinds of values each reg can hold.
1841 * Leaf Functions:: Renumbering registers for leaf functions.
1842 * Stack Registers:: Handling a register stack such as 80387.
1845 @node Register Basics
1846 @subsection Basic Characteristics of Registers
1848 @c prevent bad page break with this line
1849 Registers have various characteristics.
1851 @defmac FIRST_PSEUDO_REGISTER
1852 Number of hardware registers known to the compiler. They receive
1853 numbers 0 through @code{FIRST_PSEUDO_REGISTER-1}; thus, the first
1854 pseudo register's number really is assigned the number
1855 @code{FIRST_PSEUDO_REGISTER}.
1858 @defmac FIXED_REGISTERS
1859 @cindex fixed register
1860 An initializer that says which registers are used for fixed purposes
1861 all throughout the compiled code and are therefore not available for
1862 general allocation. These would include the stack pointer, the frame
1863 pointer (except on machines where that can be used as a general
1864 register when no frame pointer is needed), the program counter on
1865 machines where that is considered one of the addressable registers,
1866 and any other numbered register with a standard use.
1868 This information is expressed as a sequence of numbers, separated by
1869 commas and surrounded by braces. The @var{n}th number is 1 if
1870 register @var{n} is fixed, 0 otherwise.
1872 The table initialized from this macro, and the table initialized by
1873 the following one, may be overridden at run time either automatically,
1874 by the actions of the macro @code{CONDITIONAL_REGISTER_USAGE}, or by
1875 the user with the command options @option{-ffixed-@var{reg}},
1876 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}.
1879 @defmac CALL_USED_REGISTERS
1880 @cindex call-used register
1881 @cindex call-clobbered register
1882 @cindex call-saved register
1883 Like @code{FIXED_REGISTERS} but has 1 for each register that is
1884 clobbered (in general) by function calls as well as for fixed
1885 registers. This macro therefore identifies the registers that are not
1886 available for general allocation of values that must live across
1889 If a register has 0 in @code{CALL_USED_REGISTERS}, the compiler
1890 automatically saves it on function entry and restores it on function
1891 exit, if the register is used within the function.
1894 @defmac CALL_REALLY_USED_REGISTERS
1895 @cindex call-used register
1896 @cindex call-clobbered register
1897 @cindex call-saved register
1898 Like @code{CALL_USED_REGISTERS} except this macro doesn't require
1899 that the entire set of @code{FIXED_REGISTERS} be included.
1900 (@code{CALL_USED_REGISTERS} must be a superset of @code{FIXED_REGISTERS}).
1901 This macro is optional. If not specified, it defaults to the value
1902 of @code{CALL_USED_REGISTERS}.
1905 @defmac HARD_REGNO_CALL_PART_CLOBBERED (@var{regno}, @var{mode})
1906 @cindex call-used register
1907 @cindex call-clobbered register
1908 @cindex call-saved register
1909 A C expression that is nonzero if it is not permissible to store a
1910 value of mode @var{mode} in hard register number @var{regno} across a
1911 call without some part of it being clobbered. For most machines this
1912 macro need not be defined. It is only required for machines that do not
1913 preserve the entire contents of a register across a call.
1917 @findex call_used_regs
1920 @findex reg_class_contents
1921 @hook TARGET_CONDITIONAL_REGISTER_USAGE
1922 This hook may conditionally modify five variables
1923 @code{fixed_regs}, @code{call_used_regs}, @code{global_regs},
1924 @code{reg_names}, and @code{reg_class_contents}, to take into account
1925 any dependence of these register sets on target flags. The first three
1926 of these are of type @code{char []} (interpreted as Boolean vectors).
1927 @code{global_regs} is a @code{const char *[]}, and
1928 @code{reg_class_contents} is a @code{HARD_REG_SET}. Before the macro is
1929 called, @code{fixed_regs}, @code{call_used_regs},
1930 @code{reg_class_contents}, and @code{reg_names} have been initialized
1931 from @code{FIXED_REGISTERS}, @code{CALL_USED_REGISTERS},
1932 @code{REG_CLASS_CONTENTS}, and @code{REGISTER_NAMES}, respectively.
1933 @code{global_regs} has been cleared, and any @option{-ffixed-@var{reg}},
1934 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}
1935 command options have been applied.
1937 @cindex disabling certain registers
1938 @cindex controlling register usage
1939 If the usage of an entire class of registers depends on the target
1940 flags, you may indicate this to GCC by using this macro to modify
1941 @code{fixed_regs} and @code{call_used_regs} to 1 for each of the
1942 registers in the classes which should not be used by GCC@. Also define
1943 the macro @code{REG_CLASS_FROM_LETTER} / @code{REG_CLASS_FROM_CONSTRAINT}
1944 to return @code{NO_REGS} if it
1945 is called with a letter for a class that shouldn't be used.
1947 (However, if this class is not included in @code{GENERAL_REGS} and all
1948 of the insn patterns whose constraints permit this class are
1949 controlled by target switches, then GCC will automatically avoid using
1950 these registers when the target switches are opposed to them.)
1953 @defmac INCOMING_REGNO (@var{out})
1954 Define this macro if the target machine has register windows. This C
1955 expression returns the register number as seen by the called function
1956 corresponding to the register number @var{out} as seen by the calling
1957 function. Return @var{out} if register number @var{out} is not an
1961 @defmac OUTGOING_REGNO (@var{in})
1962 Define this macro if the target machine has register windows. This C
1963 expression returns the register number as seen by the calling function
1964 corresponding to the register number @var{in} as seen by the called
1965 function. Return @var{in} if register number @var{in} is not an inbound
1969 @defmac LOCAL_REGNO (@var{regno})
1970 Define this macro if the target machine has register windows. This C
1971 expression returns true if the register is call-saved but is in the
1972 register window. Unlike most call-saved registers, such registers
1973 need not be explicitly restored on function exit or during non-local
1978 If the program counter has a register number, define this as that
1979 register number. Otherwise, do not define it.
1982 @node Allocation Order
1983 @subsection Order of Allocation of Registers
1984 @cindex order of register allocation
1985 @cindex register allocation order
1987 @c prevent bad page break with this line
1988 Registers are allocated in order.
1990 @defmac REG_ALLOC_ORDER
1991 If defined, an initializer for a vector of integers, containing the
1992 numbers of hard registers in the order in which GCC should prefer
1993 to use them (from most preferred to least).
1995 If this macro is not defined, registers are used lowest numbered first
1996 (all else being equal).
1998 One use of this macro is on machines where the highest numbered
1999 registers must always be saved and the save-multiple-registers
2000 instruction supports only sequences of consecutive registers. On such
2001 machines, define @code{REG_ALLOC_ORDER} to be an initializer that lists
2002 the highest numbered allocable register first.
2005 @defmac ADJUST_REG_ALLOC_ORDER
2006 A C statement (sans semicolon) to choose the order in which to allocate
2007 hard registers for pseudo-registers local to a basic block.
2009 Store the desired register order in the array @code{reg_alloc_order}.
2010 Element 0 should be the register to allocate first; element 1, the next
2011 register; and so on.
2013 The macro body should not assume anything about the contents of
2014 @code{reg_alloc_order} before execution of the macro.
2016 On most machines, it is not necessary to define this macro.
2019 @defmac HONOR_REG_ALLOC_ORDER
2020 Normally, IRA tries to estimate the costs for saving a register in the
2021 prologue and restoring it in the epilogue. This discourages it from
2022 using call-saved registers. If a machine wants to ensure that IRA
2023 allocates registers in the order given by REG_ALLOC_ORDER even if some
2024 call-saved registers appear earlier than call-used ones, this macro
2028 @defmac IRA_HARD_REGNO_ADD_COST_MULTIPLIER (@var{regno})
2029 In some case register allocation order is not enough for the
2030 Integrated Register Allocator (@acronym{IRA}) to generate a good code.
2031 If this macro is defined, it should return a floating point value
2032 based on @var{regno}. The cost of using @var{regno} for a pseudo will
2033 be increased by approximately the pseudo's usage frequency times the
2034 value returned by this macro. Not defining this macro is equivalent
2035 to having it always return @code{0.0}.
2037 On most machines, it is not necessary to define this macro.
2040 @node Values in Registers
2041 @subsection How Values Fit in Registers
2043 This section discusses the macros that describe which kinds of values
2044 (specifically, which machine modes) each register can hold, and how many
2045 consecutive registers are needed for a given mode.
2047 @defmac HARD_REGNO_NREGS (@var{regno}, @var{mode})
2048 A C expression for the number of consecutive hard registers, starting
2049 at register number @var{regno}, required to hold a value of mode
2050 @var{mode}. This macro must never return zero, even if a register
2051 cannot hold the requested mode - indicate that with HARD_REGNO_MODE_OK
2052 and/or CANNOT_CHANGE_MODE_CLASS instead.
2054 On a machine where all registers are exactly one word, a suitable
2055 definition of this macro is
2058 #define HARD_REGNO_NREGS(REGNO, MODE) \
2059 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) \
2064 @defmac HARD_REGNO_NREGS_HAS_PADDING (@var{regno}, @var{mode})
2065 A C expression that is nonzero if a value of mode @var{mode}, stored
2066 in memory, ends with padding that causes it to take up more space than
2067 in registers starting at register number @var{regno} (as determined by
2068 multiplying GCC's notion of the size of the register when containing
2069 this mode by the number of registers returned by
2070 @code{HARD_REGNO_NREGS}). By default this is zero.
2072 For example, if a floating-point value is stored in three 32-bit
2073 registers but takes up 128 bits in memory, then this would be
2076 This macros only needs to be defined if there are cases where
2077 @code{subreg_get_info}
2078 would otherwise wrongly determine that a @code{subreg} can be
2079 represented by an offset to the register number, when in fact such a
2080 @code{subreg} would contain some of the padding not stored in
2081 registers and so not be representable.
2084 @defmac HARD_REGNO_NREGS_WITH_PADDING (@var{regno}, @var{mode})
2085 For values of @var{regno} and @var{mode} for which
2086 @code{HARD_REGNO_NREGS_HAS_PADDING} returns nonzero, a C expression
2087 returning the greater number of registers required to hold the value
2088 including any padding. In the example above, the value would be four.
2091 @defmac REGMODE_NATURAL_SIZE (@var{mode})
2092 Define this macro if the natural size of registers that hold values
2093 of mode @var{mode} is not the word size. It is a C expression that
2094 should give the natural size in bytes for the specified mode. It is
2095 used by the register allocator to try to optimize its results. This
2096 happens for example on SPARC 64-bit where the natural size of
2097 floating-point registers is still 32-bit.
2100 @defmac HARD_REGNO_MODE_OK (@var{regno}, @var{mode})
2101 A C expression that is nonzero if it is permissible to store a value
2102 of mode @var{mode} in hard register number @var{regno} (or in several
2103 registers starting with that one). For a machine where all registers
2104 are equivalent, a suitable definition is
2107 #define HARD_REGNO_MODE_OK(REGNO, MODE) 1
2110 You need not include code to check for the numbers of fixed registers,
2111 because the allocation mechanism considers them to be always occupied.
2113 @cindex register pairs
2114 On some machines, double-precision values must be kept in even/odd
2115 register pairs. You can implement that by defining this macro to reject
2116 odd register numbers for such modes.
2118 The minimum requirement for a mode to be OK in a register is that the
2119 @samp{mov@var{mode}} instruction pattern support moves between the
2120 register and other hard register in the same class and that moving a
2121 value into the register and back out not alter it.
2123 Since the same instruction used to move @code{word_mode} will work for
2124 all narrower integer modes, it is not necessary on any machine for
2125 @code{HARD_REGNO_MODE_OK} to distinguish between these modes, provided
2126 you define patterns @samp{movhi}, etc., to take advantage of this. This
2127 is useful because of the interaction between @code{HARD_REGNO_MODE_OK}
2128 and @code{MODES_TIEABLE_P}; it is very desirable for all integer modes
2131 Many machines have special registers for floating point arithmetic.
2132 Often people assume that floating point machine modes are allowed only
2133 in floating point registers. This is not true. Any registers that
2134 can hold integers can safely @emph{hold} a floating point machine
2135 mode, whether or not floating arithmetic can be done on it in those
2136 registers. Integer move instructions can be used to move the values.
2138 On some machines, though, the converse is true: fixed-point machine
2139 modes may not go in floating registers. This is true if the floating
2140 registers normalize any value stored in them, because storing a
2141 non-floating value there would garble it. In this case,
2142 @code{HARD_REGNO_MODE_OK} should reject fixed-point machine modes in
2143 floating registers. But if the floating registers do not automatically
2144 normalize, if you can store any bit pattern in one and retrieve it
2145 unchanged without a trap, then any machine mode may go in a floating
2146 register, so you can define this macro to say so.
2148 The primary significance of special floating registers is rather that
2149 they are the registers acceptable in floating point arithmetic
2150 instructions. However, this is of no concern to
2151 @code{HARD_REGNO_MODE_OK}. You handle it by writing the proper
2152 constraints for those instructions.
2154 On some machines, the floating registers are especially slow to access,
2155 so that it is better to store a value in a stack frame than in such a
2156 register if floating point arithmetic is not being done. As long as the
2157 floating registers are not in class @code{GENERAL_REGS}, they will not
2158 be used unless some pattern's constraint asks for one.
2161 @defmac HARD_REGNO_RENAME_OK (@var{from}, @var{to})
2162 A C expression that is nonzero if it is OK to rename a hard register
2163 @var{from} to another hard register @var{to}.
2165 One common use of this macro is to prevent renaming of a register to
2166 another register that is not saved by a prologue in an interrupt
2169 The default is always nonzero.
2172 @defmac MODES_TIEABLE_P (@var{mode1}, @var{mode2})
2173 A C expression that is nonzero if a value of mode
2174 @var{mode1} is accessible in mode @var{mode2} without copying.
2176 If @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode1})} and
2177 @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode2})} are always the same for
2178 any @var{r}, then @code{MODES_TIEABLE_P (@var{mode1}, @var{mode2})}
2179 should be nonzero. If they differ for any @var{r}, you should define
2180 this macro to return zero unless some other mechanism ensures the
2181 accessibility of the value in a narrower mode.
2183 You should define this macro to return nonzero in as many cases as
2184 possible since doing so will allow GCC to perform better register
2188 @hook TARGET_HARD_REGNO_SCRATCH_OK
2189 This target hook should return @code{true} if it is OK to use a hard register
2190 @var{regno} as scratch reg in peephole2.
2192 One common use of this macro is to prevent using of a register that
2193 is not saved by a prologue in an interrupt handler.
2195 The default version of this hook always returns @code{true}.
2198 @defmac AVOID_CCMODE_COPIES
2199 Define this macro if the compiler should avoid copies to/from @code{CCmode}
2200 registers. You should only define this macro if support for copying to/from
2201 @code{CCmode} is incomplete.
2204 @node Leaf Functions
2205 @subsection Handling Leaf Functions
2207 @cindex leaf functions
2208 @cindex functions, leaf
2209 On some machines, a leaf function (i.e., one which makes no calls) can run
2210 more efficiently if it does not make its own register window. Often this
2211 means it is required to receive its arguments in the registers where they
2212 are passed by the caller, instead of the registers where they would
2215 The special treatment for leaf functions generally applies only when
2216 other conditions are met; for example, often they may use only those
2217 registers for its own variables and temporaries. We use the term ``leaf
2218 function'' to mean a function that is suitable for this special
2219 handling, so that functions with no calls are not necessarily ``leaf
2222 GCC assigns register numbers before it knows whether the function is
2223 suitable for leaf function treatment. So it needs to renumber the
2224 registers in order to output a leaf function. The following macros
2227 @defmac LEAF_REGISTERS
2228 Name of a char vector, indexed by hard register number, which
2229 contains 1 for a register that is allowable in a candidate for leaf
2232 If leaf function treatment involves renumbering the registers, then the
2233 registers marked here should be the ones before renumbering---those that
2234 GCC would ordinarily allocate. The registers which will actually be
2235 used in the assembler code, after renumbering, should not be marked with 1
2238 Define this macro only if the target machine offers a way to optimize
2239 the treatment of leaf functions.
2242 @defmac LEAF_REG_REMAP (@var{regno})
2243 A C expression whose value is the register number to which @var{regno}
2244 should be renumbered, when a function is treated as a leaf function.
2246 If @var{regno} is a register number which should not appear in a leaf
2247 function before renumbering, then the expression should yield @minus{}1, which
2248 will cause the compiler to abort.
2250 Define this macro only if the target machine offers a way to optimize the
2251 treatment of leaf functions, and registers need to be renumbered to do
2255 @findex current_function_is_leaf
2256 @findex current_function_uses_only_leaf_regs
2257 @code{TARGET_ASM_FUNCTION_PROLOGUE} and
2258 @code{TARGET_ASM_FUNCTION_EPILOGUE} must usually treat leaf functions
2259 specially. They can test the C variable @code{current_function_is_leaf}
2260 which is nonzero for leaf functions. @code{current_function_is_leaf} is
2261 set prior to local register allocation and is valid for the remaining
2262 compiler passes. They can also test the C variable
2263 @code{current_function_uses_only_leaf_regs} which is nonzero for leaf
2264 functions which only use leaf registers.
2265 @code{current_function_uses_only_leaf_regs} is valid after all passes
2266 that modify the instructions have been run and is only useful if
2267 @code{LEAF_REGISTERS} is defined.
2268 @c changed this to fix overfull. ALSO: why the "it" at the beginning
2269 @c of the next paragraph?! --mew 2feb93
2271 @node Stack Registers
2272 @subsection Registers That Form a Stack
2274 There are special features to handle computers where some of the
2275 ``registers'' form a stack. Stack registers are normally written by
2276 pushing onto the stack, and are numbered relative to the top of the
2279 Currently, GCC can only handle one group of stack-like registers, and
2280 they must be consecutively numbered. Furthermore, the existing
2281 support for stack-like registers is specific to the 80387 floating
2282 point coprocessor. If you have a new architecture that uses
2283 stack-like registers, you will need to do substantial work on
2284 @file{reg-stack.c} and write your machine description to cooperate
2285 with it, as well as defining these macros.
2288 Define this if the machine has any stack-like registers.
2291 @defmac STACK_REG_COVER_CLASS
2292 This is a cover class containing the stack registers. Define this if
2293 the machine has any stack-like registers.
2296 @defmac FIRST_STACK_REG
2297 The number of the first stack-like register. This one is the top
2301 @defmac LAST_STACK_REG
2302 The number of the last stack-like register. This one is the bottom of
2306 @node Register Classes
2307 @section Register Classes
2308 @cindex register class definitions
2309 @cindex class definitions, register
2311 On many machines, the numbered registers are not all equivalent.
2312 For example, certain registers may not be allowed for indexed addressing;
2313 certain registers may not be allowed in some instructions. These machine
2314 restrictions are described to the compiler using @dfn{register classes}.
2316 You define a number of register classes, giving each one a name and saying
2317 which of the registers belong to it. Then you can specify register classes
2318 that are allowed as operands to particular instruction patterns.
2322 In general, each register will belong to several classes. In fact, one
2323 class must be named @code{ALL_REGS} and contain all the registers. Another
2324 class must be named @code{NO_REGS} and contain no registers. Often the
2325 union of two classes will be another class; however, this is not required.
2327 @findex GENERAL_REGS
2328 One of the classes must be named @code{GENERAL_REGS}. There is nothing
2329 terribly special about the name, but the operand constraint letters
2330 @samp{r} and @samp{g} specify this class. If @code{GENERAL_REGS} is
2331 the same as @code{ALL_REGS}, just define it as a macro which expands
2334 Order the classes so that if class @var{x} is contained in class @var{y}
2335 then @var{x} has a lower class number than @var{y}.
2337 The way classes other than @code{GENERAL_REGS} are specified in operand
2338 constraints is through machine-dependent operand constraint letters.
2339 You can define such letters to correspond to various classes, then use
2340 them in operand constraints.
2342 You must define the narrowest register classes for allocatable
2343 registers, so that each class either has no subclasses, or that for
2344 some mode, the move cost between registers within the class is
2345 cheaper than moving a register in the class to or from memory
2348 You should define a class for the union of two classes whenever some
2349 instruction allows both classes. For example, if an instruction allows
2350 either a floating point (coprocessor) register or a general register for a
2351 certain operand, you should define a class @code{FLOAT_OR_GENERAL_REGS}
2352 which includes both of them. Otherwise you will get suboptimal code,
2353 or even internal compiler errors when reload cannot find a register in the
2354 class computed via @code{reg_class_subunion}.
2356 You must also specify certain redundant information about the register
2357 classes: for each class, which classes contain it and which ones are
2358 contained in it; for each pair of classes, the largest class contained
2361 When a value occupying several consecutive registers is expected in a
2362 certain class, all the registers used must belong to that class.
2363 Therefore, register classes cannot be used to enforce a requirement for
2364 a register pair to start with an even-numbered register. The way to
2365 specify this requirement is with @code{HARD_REGNO_MODE_OK}.
2367 Register classes used for input-operands of bitwise-and or shift
2368 instructions have a special requirement: each such class must have, for
2369 each fixed-point machine mode, a subclass whose registers can transfer that
2370 mode to or from memory. For example, on some machines, the operations for
2371 single-byte values (@code{QImode}) are limited to certain registers. When
2372 this is so, each register class that is used in a bitwise-and or shift
2373 instruction must have a subclass consisting of registers from which
2374 single-byte values can be loaded or stored. This is so that
2375 @code{PREFERRED_RELOAD_CLASS} can always have a possible value to return.
2377 @deftp {Data type} {enum reg_class}
2378 An enumerated type that must be defined with all the register class names
2379 as enumerated values. @code{NO_REGS} must be first. @code{ALL_REGS}
2380 must be the last register class, followed by one more enumerated value,
2381 @code{LIM_REG_CLASSES}, which is not a register class but rather
2382 tells how many classes there are.
2384 Each register class has a number, which is the value of casting
2385 the class name to type @code{int}. The number serves as an index
2386 in many of the tables described below.
2389 @defmac N_REG_CLASSES
2390 The number of distinct register classes, defined as follows:
2393 #define N_REG_CLASSES (int) LIM_REG_CLASSES
2397 @defmac REG_CLASS_NAMES
2398 An initializer containing the names of the register classes as C string
2399 constants. These names are used in writing some of the debugging dumps.
2402 @defmac REG_CLASS_CONTENTS
2403 An initializer containing the contents of the register classes, as integers
2404 which are bit masks. The @var{n}th integer specifies the contents of class
2405 @var{n}. The way the integer @var{mask} is interpreted is that
2406 register @var{r} is in the class if @code{@var{mask} & (1 << @var{r})} is 1.
2408 When the machine has more than 32 registers, an integer does not suffice.
2409 Then the integers are replaced by sub-initializers, braced groupings containing
2410 several integers. Each sub-initializer must be suitable as an initializer
2411 for the type @code{HARD_REG_SET} which is defined in @file{hard-reg-set.h}.
2412 In this situation, the first integer in each sub-initializer corresponds to
2413 registers 0 through 31, the second integer to registers 32 through 63, and
2417 @defmac REGNO_REG_CLASS (@var{regno})
2418 A C expression whose value is a register class containing hard register
2419 @var{regno}. In general there is more than one such class; choose a class
2420 which is @dfn{minimal}, meaning that no smaller class also contains the
2424 @defmac BASE_REG_CLASS
2425 A macro whose definition is the name of the class to which a valid
2426 base register must belong. A base register is one used in an address
2427 which is the register value plus a displacement.
2430 @defmac MODE_BASE_REG_CLASS (@var{mode})
2431 This is a variation of the @code{BASE_REG_CLASS} macro which allows
2432 the selection of a base register in a mode dependent manner. If
2433 @var{mode} is VOIDmode then it should return the same value as
2434 @code{BASE_REG_CLASS}.
2437 @defmac MODE_BASE_REG_REG_CLASS (@var{mode})
2438 A C expression whose value is the register class to which a valid
2439 base register must belong in order to be used in a base plus index
2440 register address. You should define this macro if base plus index
2441 addresses have different requirements than other base register uses.
2444 @defmac MODE_CODE_BASE_REG_CLASS (@var{mode}, @var{outer_code}, @var{index_code})
2445 A C expression whose value is the register class to which a valid
2446 base register must belong. @var{outer_code} and @var{index_code} define the
2447 context in which the base register occurs. @var{outer_code} is the code of
2448 the immediately enclosing expression (@code{MEM} for the top level of an
2449 address, @code{ADDRESS} for something that occurs in an
2450 @code{address_operand}). @var{index_code} is the code of the corresponding
2451 index expression if @var{outer_code} is @code{PLUS}; @code{SCRATCH} otherwise.
2454 @defmac INDEX_REG_CLASS
2455 A macro whose definition is the name of the class to which a valid
2456 index register must belong. An index register is one used in an
2457 address where its value is either multiplied by a scale factor or
2458 added to another register (as well as added to a displacement).
2461 @defmac REGNO_OK_FOR_BASE_P (@var{num})
2462 A C expression which is nonzero if register number @var{num} is
2463 suitable for use as a base register in operand addresses.
2466 @defmac REGNO_MODE_OK_FOR_BASE_P (@var{num}, @var{mode})
2467 A C expression that is just like @code{REGNO_OK_FOR_BASE_P}, except that
2468 that expression may examine the mode of the memory reference in
2469 @var{mode}. You should define this macro if the mode of the memory
2470 reference affects whether a register may be used as a base register. If
2471 you define this macro, the compiler will use it instead of
2472 @code{REGNO_OK_FOR_BASE_P}. The mode may be @code{VOIDmode} for
2473 addresses that appear outside a @code{MEM}, i.e., as an
2474 @code{address_operand}.
2477 @defmac REGNO_MODE_OK_FOR_REG_BASE_P (@var{num}, @var{mode})
2478 A C expression which is nonzero if register number @var{num} is suitable for
2479 use as a base register in base plus index operand addresses, accessing
2480 memory in mode @var{mode}. It may be either a suitable hard register or a
2481 pseudo register that has been allocated such a hard register. You should
2482 define this macro if base plus index addresses have different requirements
2483 than other base register uses.
2485 Use of this macro is deprecated; please use the more general
2486 @code{REGNO_MODE_CODE_OK_FOR_BASE_P}.
2489 @defmac REGNO_MODE_CODE_OK_FOR_BASE_P (@var{num}, @var{mode}, @var{outer_code}, @var{index_code})
2490 A C expression that is just like @code{REGNO_MODE_OK_FOR_BASE_P}, except
2491 that that expression may examine the context in which the register
2492 appears in the memory reference. @var{outer_code} is the code of the
2493 immediately enclosing expression (@code{MEM} if at the top level of the
2494 address, @code{ADDRESS} for something that occurs in an
2495 @code{address_operand}). @var{index_code} is the code of the
2496 corresponding index expression if @var{outer_code} is @code{PLUS};
2497 @code{SCRATCH} otherwise. The mode may be @code{VOIDmode} for addresses
2498 that appear outside a @code{MEM}, i.e., as an @code{address_operand}.
2501 @defmac REGNO_OK_FOR_INDEX_P (@var{num})
2502 A C expression which is nonzero if register number @var{num} is
2503 suitable for use as an index register in operand addresses. It may be
2504 either a suitable hard register or a pseudo register that has been
2505 allocated such a hard register.
2507 The difference between an index register and a base register is that
2508 the index register may be scaled. If an address involves the sum of
2509 two registers, neither one of them scaled, then either one may be
2510 labeled the ``base'' and the other the ``index''; but whichever
2511 labeling is used must fit the machine's constraints of which registers
2512 may serve in each capacity. The compiler will try both labelings,
2513 looking for one that is valid, and will reload one or both registers
2514 only if neither labeling works.
2517 @hook TARGET_PREFERRED_RENAME_CLASS
2519 @hook TARGET_PREFERRED_RELOAD_CLASS
2520 A target hook that places additional restrictions on the register class
2521 to use when it is necessary to copy value @var{x} into a register in class
2522 @var{rclass}. The value is a register class; perhaps @var{rclass}, or perhaps
2523 another, smaller class.
2525 The default version of this hook always returns value of @code{rclass} argument.
2527 Sometimes returning a more restrictive class makes better code. For
2528 example, on the 68000, when @var{x} is an integer constant that is in range
2529 for a @samp{moveq} instruction, the value of this macro is always
2530 @code{DATA_REGS} as long as @var{rclass} includes the data registers.
2531 Requiring a data register guarantees that a @samp{moveq} will be used.
2533 One case where @code{TARGET_PREFERRED_RELOAD_CLASS} must not return
2534 @var{rclass} is if @var{x} is a legitimate constant which cannot be
2535 loaded into some register class. By returning @code{NO_REGS} you can
2536 force @var{x} into a memory location. For example, rs6000 can load
2537 immediate values into general-purpose registers, but does not have an
2538 instruction for loading an immediate value into a floating-point
2539 register, so @code{TARGET_PREFERRED_RELOAD_CLASS} returns @code{NO_REGS} when
2540 @var{x} is a floating-point constant. If the constant can't be loaded
2541 into any kind of register, code generation will be better if
2542 @code{TARGET_LEGITIMATE_CONSTANT_P} makes the constant illegitimate instead
2543 of using @code{TARGET_PREFERRED_RELOAD_CLASS}.
2545 If an insn has pseudos in it after register allocation, reload will go
2546 through the alternatives and call repeatedly @code{TARGET_PREFERRED_RELOAD_CLASS}
2547 to find the best one. Returning @code{NO_REGS}, in this case, makes
2548 reload add a @code{!} in front of the constraint: the x86 back-end uses
2549 this feature to discourage usage of 387 registers when math is done in
2550 the SSE registers (and vice versa).
2553 @defmac PREFERRED_RELOAD_CLASS (@var{x}, @var{class})
2554 A C expression that places additional restrictions on the register class
2555 to use when it is necessary to copy value @var{x} into a register in class
2556 @var{class}. The value is a register class; perhaps @var{class}, or perhaps
2557 another, smaller class. On many machines, the following definition is
2561 #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS
2564 Sometimes returning a more restrictive class makes better code. For
2565 example, on the 68000, when @var{x} is an integer constant that is in range
2566 for a @samp{moveq} instruction, the value of this macro is always
2567 @code{DATA_REGS} as long as @var{class} includes the data registers.
2568 Requiring a data register guarantees that a @samp{moveq} will be used.
2570 One case where @code{PREFERRED_RELOAD_CLASS} must not return
2571 @var{class} is if @var{x} is a legitimate constant which cannot be
2572 loaded into some register class. By returning @code{NO_REGS} you can
2573 force @var{x} into a memory location. For example, rs6000 can load
2574 immediate values into general-purpose registers, but does not have an
2575 instruction for loading an immediate value into a floating-point
2576 register, so @code{PREFERRED_RELOAD_CLASS} returns @code{NO_REGS} when
2577 @var{x} is a floating-point constant. If the constant can't be loaded
2578 into any kind of register, code generation will be better if
2579 @code{TARGET_LEGITIMATE_CONSTANT_P} makes the constant illegitimate instead
2580 of using @code{TARGET_PREFERRED_RELOAD_CLASS}.
2582 If an insn has pseudos in it after register allocation, reload will go
2583 through the alternatives and call repeatedly @code{PREFERRED_RELOAD_CLASS}
2584 to find the best one. Returning @code{NO_REGS}, in this case, makes
2585 reload add a @code{!} in front of the constraint: the x86 back-end uses
2586 this feature to discourage usage of 387 registers when math is done in
2587 the SSE registers (and vice versa).
2590 @defmac PREFERRED_OUTPUT_RELOAD_CLASS (@var{x}, @var{class})
2591 Like @code{PREFERRED_RELOAD_CLASS}, but for output reloads instead of
2592 input reloads. If you don't define this macro, the default is to use
2593 @var{class}, unchanged.
2595 You can also use @code{PREFERRED_OUTPUT_RELOAD_CLASS} to discourage
2596 reload from using some alternatives, like @code{PREFERRED_RELOAD_CLASS}.
2599 @hook TARGET_PREFERRED_OUTPUT_RELOAD_CLASS
2600 Like @code{TARGET_PREFERRED_RELOAD_CLASS}, but for output reloads instead of
2603 The default version of this hook always returns value of @code{rclass}
2606 You can also use @code{TARGET_PREFERRED_OUTPUT_RELOAD_CLASS} to discourage
2607 reload from using some alternatives, like @code{TARGET_PREFERRED_RELOAD_CLASS}.
2610 @defmac LIMIT_RELOAD_CLASS (@var{mode}, @var{class})
2611 A C expression that places additional restrictions on the register class
2612 to use when it is necessary to be able to hold a value of mode
2613 @var{mode} in a reload register for which class @var{class} would
2616 Unlike @code{PREFERRED_RELOAD_CLASS}, this macro should be used when
2617 there are certain modes that simply can't go in certain reload classes.
2619 The value is a register class; perhaps @var{class}, or perhaps another,
2622 Don't define this macro unless the target machine has limitations which
2623 require the macro to do something nontrivial.
2626 @hook TARGET_SECONDARY_RELOAD
2627 Many machines have some registers that cannot be copied directly to or
2628 from memory or even from other types of registers. An example is the
2629 @samp{MQ} register, which on most machines, can only be copied to or
2630 from general registers, but not memory. Below, we shall be using the
2631 term 'intermediate register' when a move operation cannot be performed
2632 directly, but has to be done by copying the source into the intermediate
2633 register first, and then copying the intermediate register to the
2634 destination. An intermediate register always has the same mode as
2635 source and destination. Since it holds the actual value being copied,
2636 reload might apply optimizations to re-use an intermediate register
2637 and eliding the copy from the source when it can determine that the
2638 intermediate register still holds the required value.
2640 Another kind of secondary reload is required on some machines which
2641 allow copying all registers to and from memory, but require a scratch
2642 register for stores to some memory locations (e.g., those with symbolic
2643 address on the RT, and those with certain symbolic address on the SPARC
2644 when compiling PIC)@. Scratch registers need not have the same mode
2645 as the value being copied, and usually hold a different value than
2646 that being copied. Special patterns in the md file are needed to
2647 describe how the copy is performed with the help of the scratch register;
2648 these patterns also describe the number, register class(es) and mode(s)
2649 of the scratch register(s).
2651 In some cases, both an intermediate and a scratch register are required.
2653 For input reloads, this target hook is called with nonzero @var{in_p},
2654 and @var{x} is an rtx that needs to be copied to a register of class
2655 @var{reload_class} in @var{reload_mode}. For output reloads, this target
2656 hook is called with zero @var{in_p}, and a register of class @var{reload_class}
2657 needs to be copied to rtx @var{x} in @var{reload_mode}.
2659 If copying a register of @var{reload_class} from/to @var{x} requires
2660 an intermediate register, the hook @code{secondary_reload} should
2661 return the register class required for this intermediate register.
2662 If no intermediate register is required, it should return NO_REGS.
2663 If more than one intermediate register is required, describe the one
2664 that is closest in the copy chain to the reload register.
2666 If scratch registers are needed, you also have to describe how to
2667 perform the copy from/to the reload register to/from this
2668 closest intermediate register. Or if no intermediate register is
2669 required, but still a scratch register is needed, describe the
2670 copy from/to the reload register to/from the reload operand @var{x}.
2672 You do this by setting @code{sri->icode} to the instruction code of a pattern
2673 in the md file which performs the move. Operands 0 and 1 are the output
2674 and input of this copy, respectively. Operands from operand 2 onward are
2675 for scratch operands. These scratch operands must have a mode, and a
2676 single-register-class
2677 @c [later: or memory]
2680 When an intermediate register is used, the @code{secondary_reload}
2681 hook will be called again to determine how to copy the intermediate
2682 register to/from the reload operand @var{x}, so your hook must also
2683 have code to handle the register class of the intermediate operand.
2685 @c [For later: maybe we'll allow multi-alternative reload patterns -
2686 @c the port maintainer could name a mov<mode> pattern that has clobbers -
2687 @c and match the constraints of input and output to determine the required
2688 @c alternative. A restriction would be that constraints used to match
2689 @c against reloads registers would have to be written as register class
2690 @c constraints, or we need a new target macro / hook that tells us if an
2691 @c arbitrary constraint can match an unknown register of a given class.
2692 @c Such a macro / hook would also be useful in other places.]
2695 @var{x} might be a pseudo-register or a @code{subreg} of a
2696 pseudo-register, which could either be in a hard register or in memory.
2697 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2698 in memory and the hard register number if it is in a register.
2700 Scratch operands in memory (constraint @code{"=m"} / @code{"=&m"}) are
2701 currently not supported. For the time being, you will have to continue
2702 to use @code{SECONDARY_MEMORY_NEEDED} for that purpose.
2704 @code{copy_cost} also uses this target hook to find out how values are
2705 copied. If you want it to include some extra cost for the need to allocate
2706 (a) scratch register(s), set @code{sri->extra_cost} to the additional cost.
2707 Or if two dependent moves are supposed to have a lower cost than the sum
2708 of the individual moves due to expected fortuitous scheduling and/or special
2709 forwarding logic, you can set @code{sri->extra_cost} to a negative amount.
2712 @defmac SECONDARY_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2713 @defmacx SECONDARY_INPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2714 @defmacx SECONDARY_OUTPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2715 These macros are obsolete, new ports should use the target hook
2716 @code{TARGET_SECONDARY_RELOAD} instead.
2718 These are obsolete macros, replaced by the @code{TARGET_SECONDARY_RELOAD}
2719 target hook. Older ports still define these macros to indicate to the
2720 reload phase that it may
2721 need to allocate at least one register for a reload in addition to the
2722 register to contain the data. Specifically, if copying @var{x} to a
2723 register @var{class} in @var{mode} requires an intermediate register,
2724 you were supposed to define @code{SECONDARY_INPUT_RELOAD_CLASS} to return the
2725 largest register class all of whose registers can be used as
2726 intermediate registers or scratch registers.
2728 If copying a register @var{class} in @var{mode} to @var{x} requires an
2729 intermediate or scratch register, @code{SECONDARY_OUTPUT_RELOAD_CLASS}
2730 was supposed to be defined be defined to return the largest register
2731 class required. If the
2732 requirements for input and output reloads were the same, the macro
2733 @code{SECONDARY_RELOAD_CLASS} should have been used instead of defining both
2736 The values returned by these macros are often @code{GENERAL_REGS}.
2737 Return @code{NO_REGS} if no spare register is needed; i.e., if @var{x}
2738 can be directly copied to or from a register of @var{class} in
2739 @var{mode} without requiring a scratch register. Do not define this
2740 macro if it would always return @code{NO_REGS}.
2742 If a scratch register is required (either with or without an
2743 intermediate register), you were supposed to define patterns for
2744 @samp{reload_in@var{m}} or @samp{reload_out@var{m}}, as required
2745 (@pxref{Standard Names}. These patterns, which were normally
2746 implemented with a @code{define_expand}, should be similar to the
2747 @samp{mov@var{m}} patterns, except that operand 2 is the scratch
2750 These patterns need constraints for the reload register and scratch
2752 contain a single register class. If the original reload register (whose
2753 class is @var{class}) can meet the constraint given in the pattern, the
2754 value returned by these macros is used for the class of the scratch
2755 register. Otherwise, two additional reload registers are required.
2756 Their classes are obtained from the constraints in the insn pattern.
2758 @var{x} might be a pseudo-register or a @code{subreg} of a
2759 pseudo-register, which could either be in a hard register or in memory.
2760 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2761 in memory and the hard register number if it is in a register.
2763 These macros should not be used in the case where a particular class of
2764 registers can only be copied to memory and not to another class of
2765 registers. In that case, secondary reload registers are not needed and
2766 would not be helpful. Instead, a stack location must be used to perform
2767 the copy and the @code{mov@var{m}} pattern should use memory as an
2768 intermediate storage. This case often occurs between floating-point and
2772 @defmac SECONDARY_MEMORY_NEEDED (@var{class1}, @var{class2}, @var{m})
2773 Certain machines have the property that some registers cannot be copied
2774 to some other registers without using memory. Define this macro on
2775 those machines to be a C expression that is nonzero if objects of mode
2776 @var{m} in registers of @var{class1} can only be copied to registers of
2777 class @var{class2} by storing a register of @var{class1} into memory
2778 and loading that memory location into a register of @var{class2}.
2780 Do not define this macro if its value would always be zero.
2783 @defmac SECONDARY_MEMORY_NEEDED_RTX (@var{mode})
2784 Normally when @code{SECONDARY_MEMORY_NEEDED} is defined, the compiler
2785 allocates a stack slot for a memory location needed for register copies.
2786 If this macro is defined, the compiler instead uses the memory location
2787 defined by this macro.
2789 Do not define this macro if you do not define
2790 @code{SECONDARY_MEMORY_NEEDED}.
2793 @defmac SECONDARY_MEMORY_NEEDED_MODE (@var{mode})
2794 When the compiler needs a secondary memory location to copy between two
2795 registers of mode @var{mode}, it normally allocates sufficient memory to
2796 hold a quantity of @code{BITS_PER_WORD} bits and performs the store and
2797 load operations in a mode that many bits wide and whose class is the
2798 same as that of @var{mode}.
2800 This is right thing to do on most machines because it ensures that all
2801 bits of the register are copied and prevents accesses to the registers
2802 in a narrower mode, which some machines prohibit for floating-point
2805 However, this default behavior is not correct on some machines, such as
2806 the DEC Alpha, that store short integers in floating-point registers
2807 differently than in integer registers. On those machines, the default
2808 widening will not work correctly and you must define this macro to
2809 suppress that widening in some cases. See the file @file{alpha.h} for
2812 Do not define this macro if you do not define
2813 @code{SECONDARY_MEMORY_NEEDED} or if widening @var{mode} to a mode that
2814 is @code{BITS_PER_WORD} bits wide is correct for your machine.
2817 @hook TARGET_CLASS_LIKELY_SPILLED_P
2818 A target hook which returns @code{true} if pseudos that have been assigned
2819 to registers of class @var{rclass} would likely be spilled because
2820 registers of @var{rclass} are needed for spill registers.
2822 The default version of this target hook returns @code{true} if @var{rclass}
2823 has exactly one register and @code{false} otherwise. On most machines, this
2824 default should be used. Only use this target hook to some other expression
2825 if pseudos allocated by @file{local-alloc.c} end up in memory because their
2826 hard registers were needed for spill registers. If this target hook returns
2827 @code{false} for those classes, those pseudos will only be allocated by
2828 @file{global.c}, which knows how to reallocate the pseudo to another
2829 register. If there would not be another register available for reallocation,
2830 you should not change the implementation of this target hook since
2831 the only effect of such implementation would be to slow down register
2835 @hook TARGET_CLASS_MAX_NREGS
2836 A target hook returns the maximum number of consecutive registers
2837 of class @var{rclass} needed to hold a value of mode @var{mode}.
2839 This is closely related to the macro @code{HARD_REGNO_NREGS}. In fact,
2840 the value returned by @code{TARGET_CLASS_MAX_NREGS (@var{rclass},
2841 @var{mode})} target hook should be the maximum value of
2842 @code{HARD_REGNO_NREGS (@var{regno}, @var{mode})} for all @var{regno}
2843 values in the class @var{rclass}.
2845 This target hook helps control the handling of multiple-word values
2848 The default version of this target hook returns the size of @var{mode}
2852 @defmac CLASS_MAX_NREGS (@var{class}, @var{mode})
2853 A C expression for the maximum number of consecutive registers
2854 of class @var{class} needed to hold a value of mode @var{mode}.
2856 This is closely related to the macro @code{HARD_REGNO_NREGS}. In fact,
2857 the value of the macro @code{CLASS_MAX_NREGS (@var{class}, @var{mode})}
2858 should be the maximum value of @code{HARD_REGNO_NREGS (@var{regno},
2859 @var{mode})} for all @var{regno} values in the class @var{class}.
2861 This macro helps control the handling of multiple-word values
2865 @defmac CANNOT_CHANGE_MODE_CLASS (@var{from}, @var{to}, @var{class})
2866 If defined, a C expression that returns nonzero for a @var{class} for which
2867 a change from mode @var{from} to mode @var{to} is invalid.
2869 For the example, loading 32-bit integer or floating-point objects into
2870 floating-point registers on the Alpha extends them to 64 bits.
2871 Therefore loading a 64-bit object and then storing it as a 32-bit object
2872 does not store the low-order 32 bits, as would be the case for a normal
2873 register. Therefore, @file{alpha.h} defines @code{CANNOT_CHANGE_MODE_CLASS}
2877 #define CANNOT_CHANGE_MODE_CLASS(FROM, TO, CLASS) \
2878 (GET_MODE_SIZE (FROM) != GET_MODE_SIZE (TO) \
2879 ? reg_classes_intersect_p (FLOAT_REGS, (CLASS)) : 0)
2883 @node Old Constraints
2884 @section Obsolete Macros for Defining Constraints
2885 @cindex defining constraints, obsolete method
2886 @cindex constraints, defining, obsolete method
2888 Machine-specific constraints can be defined with these macros instead
2889 of the machine description constructs described in @ref{Define
2890 Constraints}. This mechanism is obsolete. New ports should not use
2891 it; old ports should convert to the new mechanism.
2893 @defmac CONSTRAINT_LEN (@var{char}, @var{str})
2894 For the constraint at the start of @var{str}, which starts with the letter
2895 @var{c}, return the length. This allows you to have register class /
2896 constant / extra constraints that are longer than a single letter;
2897 you don't need to define this macro if you can do with single-letter
2898 constraints only. The definition of this macro should use
2899 DEFAULT_CONSTRAINT_LEN for all the characters that you don't want
2900 to handle specially.
2901 There are some sanity checks in genoutput.c that check the constraint lengths
2902 for the md file, so you can also use this macro to help you while you are
2903 transitioning from a byzantine single-letter-constraint scheme: when you
2904 return a negative length for a constraint you want to re-use, genoutput
2905 will complain about every instance where it is used in the md file.
2908 @defmac REG_CLASS_FROM_LETTER (@var{char})
2909 A C expression which defines the machine-dependent operand constraint
2910 letters for register classes. If @var{char} is such a letter, the
2911 value should be the register class corresponding to it. Otherwise,
2912 the value should be @code{NO_REGS}. The register letter @samp{r},
2913 corresponding to class @code{GENERAL_REGS}, will not be passed
2914 to this macro; you do not need to handle it.
2917 @defmac REG_CLASS_FROM_CONSTRAINT (@var{char}, @var{str})
2918 Like @code{REG_CLASS_FROM_LETTER}, but you also get the constraint string
2919 passed in @var{str}, so that you can use suffixes to distinguish between
2923 @defmac CONST_OK_FOR_LETTER_P (@var{value}, @var{c})
2924 A C expression that defines the machine-dependent operand constraint
2925 letters (@samp{I}, @samp{J}, @samp{K}, @dots{} @samp{P}) that specify
2926 particular ranges of integer values. If @var{c} is one of those
2927 letters, the expression should check that @var{value}, an integer, is in
2928 the appropriate range and return 1 if so, 0 otherwise. If @var{c} is
2929 not one of those letters, the value should be 0 regardless of
2933 @defmac CONST_OK_FOR_CONSTRAINT_P (@var{value}, @var{c}, @var{str})
2934 Like @code{CONST_OK_FOR_LETTER_P}, but you also get the constraint
2935 string passed in @var{str}, so that you can use suffixes to distinguish
2936 between different variants.
2939 @defmac CONST_DOUBLE_OK_FOR_LETTER_P (@var{value}, @var{c})
2940 A C expression that defines the machine-dependent operand constraint
2941 letters that specify particular ranges of @code{const_double} values
2942 (@samp{G} or @samp{H}).
2944 If @var{c} is one of those letters, the expression should check that
2945 @var{value}, an RTX of code @code{const_double}, is in the appropriate
2946 range and return 1 if so, 0 otherwise. If @var{c} is not one of those
2947 letters, the value should be 0 regardless of @var{value}.
2949 @code{const_double} is used for all floating-point constants and for
2950 @code{DImode} fixed-point constants. A given letter can accept either
2951 or both kinds of values. It can use @code{GET_MODE} to distinguish
2952 between these kinds.
2955 @defmac CONST_DOUBLE_OK_FOR_CONSTRAINT_P (@var{value}, @var{c}, @var{str})
2956 Like @code{CONST_DOUBLE_OK_FOR_LETTER_P}, but you also get the constraint
2957 string passed in @var{str}, so that you can use suffixes to distinguish
2958 between different variants.
2961 @defmac EXTRA_CONSTRAINT (@var{value}, @var{c})
2962 A C expression that defines the optional machine-dependent constraint
2963 letters that can be used to segregate specific types of operands, usually
2964 memory references, for the target machine. Any letter that is not
2965 elsewhere defined and not matched by @code{REG_CLASS_FROM_LETTER} /
2966 @code{REG_CLASS_FROM_CONSTRAINT}
2967 may be used. Normally this macro will not be defined.
2969 If it is required for a particular target machine, it should return 1
2970 if @var{value} corresponds to the operand type represented by the
2971 constraint letter @var{c}. If @var{c} is not defined as an extra
2972 constraint, the value returned should be 0 regardless of @var{value}.
2974 For example, on the ROMP, load instructions cannot have their output
2975 in r0 if the memory reference contains a symbolic address. Constraint
2976 letter @samp{Q} is defined as representing a memory address that does
2977 @emph{not} contain a symbolic address. An alternative is specified with
2978 a @samp{Q} constraint on the input and @samp{r} on the output. The next
2979 alternative specifies @samp{m} on the input and a register class that
2980 does not include r0 on the output.
2983 @defmac EXTRA_CONSTRAINT_STR (@var{value}, @var{c}, @var{str})
2984 Like @code{EXTRA_CONSTRAINT}, but you also get the constraint string passed
2985 in @var{str}, so that you can use suffixes to distinguish between different
2989 @defmac EXTRA_MEMORY_CONSTRAINT (@var{c}, @var{str})
2990 A C expression that defines the optional machine-dependent constraint
2991 letters, amongst those accepted by @code{EXTRA_CONSTRAINT}, that should
2992 be treated like memory constraints by the reload pass.
2994 It should return 1 if the operand type represented by the constraint
2995 at the start of @var{str}, the first letter of which is the letter @var{c},
2996 comprises a subset of all memory references including
2997 all those whose address is simply a base register. This allows the reload
2998 pass to reload an operand, if it does not directly correspond to the operand
2999 type of @var{c}, by copying its address into a base register.
3001 For example, on the S/390, some instructions do not accept arbitrary
3002 memory references, but only those that do not make use of an index
3003 register. The constraint letter @samp{Q} is defined via
3004 @code{EXTRA_CONSTRAINT} as representing a memory address of this type.
3005 If the letter @samp{Q} is marked as @code{EXTRA_MEMORY_CONSTRAINT},
3006 a @samp{Q} constraint can handle any memory operand, because the
3007 reload pass knows it can be reloaded by copying the memory address
3008 into a base register if required. This is analogous to the way
3009 an @samp{o} constraint can handle any memory operand.
3012 @defmac EXTRA_ADDRESS_CONSTRAINT (@var{c}, @var{str})
3013 A C expression that defines the optional machine-dependent constraint
3014 letters, amongst those accepted by @code{EXTRA_CONSTRAINT} /
3015 @code{EXTRA_CONSTRAINT_STR}, that should
3016 be treated like address constraints by the reload pass.
3018 It should return 1 if the operand type represented by the constraint
3019 at the start of @var{str}, which starts with the letter @var{c}, comprises
3020 a subset of all memory addresses including
3021 all those that consist of just a base register. This allows the reload
3022 pass to reload an operand, if it does not directly correspond to the operand
3023 type of @var{str}, by copying it into a base register.
3025 Any constraint marked as @code{EXTRA_ADDRESS_CONSTRAINT} can only
3026 be used with the @code{address_operand} predicate. It is treated
3027 analogously to the @samp{p} constraint.
3030 @node Stack and Calling
3031 @section Stack Layout and Calling Conventions
3032 @cindex calling conventions
3034 @c prevent bad page break with this line
3035 This describes the stack layout and calling conventions.
3039 * Exception Handling::
3044 * Register Arguments::
3046 * Aggregate Return::
3051 * Stack Smashing Protection::
3055 @subsection Basic Stack Layout
3056 @cindex stack frame layout
3057 @cindex frame layout
3059 @c prevent bad page break with this line
3060 Here is the basic stack layout.
3062 @defmac STACK_GROWS_DOWNWARD
3063 Define this macro if pushing a word onto the stack moves the stack
3064 pointer to a smaller address.
3066 When we say, ``define this macro if @dots{}'', it means that the
3067 compiler checks this macro only with @code{#ifdef} so the precise
3068 definition used does not matter.
3071 @defmac STACK_PUSH_CODE
3072 This macro defines the operation used when something is pushed
3073 on the stack. In RTL, a push operation will be
3074 @code{(set (mem (STACK_PUSH_CODE (reg sp))) @dots{})}
3076 The choices are @code{PRE_DEC}, @code{POST_DEC}, @code{PRE_INC},
3077 and @code{POST_INC}. Which of these is correct depends on
3078 the stack direction and on whether the stack pointer points
3079 to the last item on the stack or whether it points to the
3080 space for the next item on the stack.
3082 The default is @code{PRE_DEC} when @code{STACK_GROWS_DOWNWARD} is
3083 defined, which is almost always right, and @code{PRE_INC} otherwise,
3084 which is often wrong.
3087 @defmac FRAME_GROWS_DOWNWARD
3088 Define this macro to nonzero value if the addresses of local variable slots
3089 are at negative offsets from the frame pointer.
3092 @defmac ARGS_GROW_DOWNWARD
3093 Define this macro if successive arguments to a function occupy decreasing
3094 addresses on the stack.
3097 @defmac STARTING_FRAME_OFFSET
3098 Offset from the frame pointer to the first local variable slot to be allocated.
3100 If @code{FRAME_GROWS_DOWNWARD}, find the next slot's offset by
3101 subtracting the first slot's length from @code{STARTING_FRAME_OFFSET}.
3102 Otherwise, it is found by adding the length of the first slot to the
3103 value @code{STARTING_FRAME_OFFSET}.
3104 @c i'm not sure if the above is still correct.. had to change it to get
3105 @c rid of an overfull. --mew 2feb93
3108 @defmac STACK_ALIGNMENT_NEEDED
3109 Define to zero to disable final alignment of the stack during reload.
3110 The nonzero default for this macro is suitable for most ports.
3112 On ports where @code{STARTING_FRAME_OFFSET} is nonzero or where there
3113 is a register save block following the local block that doesn't require
3114 alignment to @code{STACK_BOUNDARY}, it may be beneficial to disable
3115 stack alignment and do it in the backend.
3118 @defmac STACK_POINTER_OFFSET
3119 Offset from the stack pointer register to the first location at which
3120 outgoing arguments are placed. If not specified, the default value of
3121 zero is used. This is the proper value for most machines.
3123 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
3124 the first location at which outgoing arguments are placed.
3127 @defmac FIRST_PARM_OFFSET (@var{fundecl})
3128 Offset from the argument pointer register to the first argument's
3129 address. On some machines it may depend on the data type of the
3132 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
3133 the first argument's address.
3136 @defmac STACK_DYNAMIC_OFFSET (@var{fundecl})
3137 Offset from the stack pointer register to an item dynamically allocated
3138 on the stack, e.g., by @code{alloca}.
3140 The default value for this macro is @code{STACK_POINTER_OFFSET} plus the
3141 length of the outgoing arguments. The default is correct for most
3142 machines. See @file{function.c} for details.
3145 @defmac INITIAL_FRAME_ADDRESS_RTX
3146 A C expression whose value is RTL representing the address of the initial
3147 stack frame. This address is passed to @code{RETURN_ADDR_RTX} and
3148 @code{DYNAMIC_CHAIN_ADDRESS}. If you don't define this macro, a reasonable
3149 default value will be used. Define this macro in order to make frame pointer
3150 elimination work in the presence of @code{__builtin_frame_address (count)} and
3151 @code{__builtin_return_address (count)} for @code{count} not equal to zero.
3154 @defmac DYNAMIC_CHAIN_ADDRESS (@var{frameaddr})
3155 A C expression whose value is RTL representing the address in a stack
3156 frame where the pointer to the caller's frame is stored. Assume that
3157 @var{frameaddr} is an RTL expression for the address of the stack frame
3160 If you don't define this macro, the default is to return the value
3161 of @var{frameaddr}---that is, the stack frame address is also the
3162 address of the stack word that points to the previous frame.
3165 @defmac SETUP_FRAME_ADDRESSES
3166 If defined, a C expression that produces the machine-specific code to
3167 setup the stack so that arbitrary frames can be accessed. For example,
3168 on the SPARC, we must flush all of the register windows to the stack
3169 before we can access arbitrary stack frames. You will seldom need to
3173 @hook TARGET_BUILTIN_SETJMP_FRAME_VALUE
3174 This target hook should return an rtx that is used to store
3175 the address of the current frame into the built in @code{setjmp} buffer.
3176 The default value, @code{virtual_stack_vars_rtx}, is correct for most
3177 machines. One reason you may need to define this target hook is if
3178 @code{hard_frame_pointer_rtx} is the appropriate value on your machine.
3181 @defmac FRAME_ADDR_RTX (@var{frameaddr})
3182 A C expression whose value is RTL representing the value of the frame
3183 address for the current frame. @var{frameaddr} is the frame pointer
3184 of the current frame. This is used for __builtin_frame_address.
3185 You need only define this macro if the frame address is not the same
3186 as the frame pointer. Most machines do not need to define it.
3189 @defmac RETURN_ADDR_RTX (@var{count}, @var{frameaddr})
3190 A C expression whose value is RTL representing the value of the return
3191 address for the frame @var{count} steps up from the current frame, after
3192 the prologue. @var{frameaddr} is the frame pointer of the @var{count}
3193 frame, or the frame pointer of the @var{count} @minus{} 1 frame if
3194 @code{RETURN_ADDR_IN_PREVIOUS_FRAME} is defined.
3196 The value of the expression must always be the correct address when
3197 @var{count} is zero, but may be @code{NULL_RTX} if there is no way to
3198 determine the return address of other frames.
3201 @defmac RETURN_ADDR_IN_PREVIOUS_FRAME
3202 Define this if the return address of a particular stack frame is accessed
3203 from the frame pointer of the previous stack frame.
3206 @defmac INCOMING_RETURN_ADDR_RTX
3207 A C expression whose value is RTL representing the location of the
3208 incoming return address at the beginning of any function, before the
3209 prologue. This RTL is either a @code{REG}, indicating that the return
3210 value is saved in @samp{REG}, or a @code{MEM} representing a location in
3213 You only need to define this macro if you want to support call frame
3214 debugging information like that provided by DWARF 2.
3216 If this RTL is a @code{REG}, you should also define
3217 @code{DWARF_FRAME_RETURN_COLUMN} to @code{DWARF_FRAME_REGNUM (REGNO)}.
3220 @defmac DWARF_ALT_FRAME_RETURN_COLUMN
3221 A C expression whose value is an integer giving a DWARF 2 column
3222 number that may be used as an alternative return column. The column
3223 must not correspond to any gcc hard register (that is, it must not
3224 be in the range of @code{DWARF_FRAME_REGNUM}).
3226 This macro can be useful if @code{DWARF_FRAME_RETURN_COLUMN} is set to a
3227 general register, but an alternative column needs to be used for signal
3228 frames. Some targets have also used different frame return columns
3232 @defmac DWARF_ZERO_REG
3233 A C expression whose value is an integer giving a DWARF 2 register
3234 number that is considered to always have the value zero. This should
3235 only be defined if the target has an architected zero register, and
3236 someone decided it was a good idea to use that register number to
3237 terminate the stack backtrace. New ports should avoid this.
3240 @hook TARGET_DWARF_HANDLE_FRAME_UNSPEC
3241 This target hook allows the backend to emit frame-related insns that
3242 contain UNSPECs or UNSPEC_VOLATILEs. The DWARF 2 call frame debugging
3243 info engine will invoke it on insns of the form
3245 (set (reg) (unspec [@dots{}] UNSPEC_INDEX))
3249 (set (reg) (unspec_volatile [@dots{}] UNSPECV_INDEX)).
3251 to let the backend emit the call frame instructions. @var{label} is
3252 the CFI label attached to the insn, @var{pattern} is the pattern of
3253 the insn and @var{index} is @code{UNSPEC_INDEX} or @code{UNSPECV_INDEX}.
3256 @defmac INCOMING_FRAME_SP_OFFSET
3257 A C expression whose value is an integer giving the offset, in bytes,
3258 from the value of the stack pointer register to the top of the stack
3259 frame at the beginning of any function, before the prologue. The top of
3260 the frame is defined to be the value of the stack pointer in the
3261 previous frame, just before the call instruction.
3263 You only need to define this macro if you want to support call frame
3264 debugging information like that provided by DWARF 2.
3267 @defmac ARG_POINTER_CFA_OFFSET (@var{fundecl})
3268 A C expression whose value is an integer giving the offset, in bytes,
3269 from the argument pointer to the canonical frame address (cfa). The
3270 final value should coincide with that calculated by
3271 @code{INCOMING_FRAME_SP_OFFSET}. Which is unfortunately not usable
3272 during virtual register instantiation.
3274 The default value for this macro is
3275 @code{FIRST_PARM_OFFSET (fundecl) + crtl->args.pretend_args_size},
3276 which is correct for most machines; in general, the arguments are found
3277 immediately before the stack frame. Note that this is not the case on
3278 some targets that save registers into the caller's frame, such as SPARC
3279 and rs6000, and so such targets need to define this macro.
3281 You only need to define this macro if the default is incorrect, and you
3282 want to support call frame debugging information like that provided by
3286 @defmac FRAME_POINTER_CFA_OFFSET (@var{fundecl})
3287 If defined, a C expression whose value is an integer giving the offset
3288 in bytes from the frame pointer to the canonical frame address (cfa).
3289 The final value should coincide with that calculated by
3290 @code{INCOMING_FRAME_SP_OFFSET}.
3292 Normally the CFA is calculated as an offset from the argument pointer,
3293 via @code{ARG_POINTER_CFA_OFFSET}, but if the argument pointer is
3294 variable due to the ABI, this may not be possible. If this macro is
3295 defined, it implies that the virtual register instantiation should be
3296 based on the frame pointer instead of the argument pointer. Only one
3297 of @code{FRAME_POINTER_CFA_OFFSET} and @code{ARG_POINTER_CFA_OFFSET}
3301 @defmac CFA_FRAME_BASE_OFFSET (@var{fundecl})
3302 If defined, a C expression whose value is an integer giving the offset
3303 in bytes from the canonical frame address (cfa) to the frame base used
3304 in DWARF 2 debug information. The default is zero. A different value
3305 may reduce the size of debug information on some ports.
3308 @node Exception Handling
3309 @subsection Exception Handling Support
3310 @cindex exception handling
3312 @defmac EH_RETURN_DATA_REGNO (@var{N})
3313 A C expression whose value is the @var{N}th register number used for
3314 data by exception handlers, or @code{INVALID_REGNUM} if fewer than
3315 @var{N} registers are usable.
3317 The exception handling library routines communicate with the exception
3318 handlers via a set of agreed upon registers. Ideally these registers
3319 should be call-clobbered; it is possible to use call-saved registers,
3320 but may negatively impact code size. The target must support at least
3321 2 data registers, but should define 4 if there are enough free registers.
3323 You must define this macro if you want to support call frame exception
3324 handling like that provided by DWARF 2.
3327 @defmac EH_RETURN_STACKADJ_RTX
3328 A C expression whose value is RTL representing a location in which
3329 to store a stack adjustment to be applied before function return.
3330 This is used to unwind the stack to an exception handler's call frame.
3331 It will be assigned zero on code paths that return normally.
3333 Typically this is a call-clobbered hard register that is otherwise
3334 untouched by the epilogue, but could also be a stack slot.
3336 Do not define this macro if the stack pointer is saved and restored
3337 by the regular prolog and epilog code in the call frame itself; in
3338 this case, the exception handling library routines will update the
3339 stack location to be restored in place. Otherwise, you must define
3340 this macro if you want to support call frame exception handling like
3341 that provided by DWARF 2.
3344 @defmac EH_RETURN_HANDLER_RTX
3345 A C expression whose value is RTL representing a location in which
3346 to store the address of an exception handler to which we should
3347 return. It will not be assigned on code paths that return normally.
3349 Typically this is the location in the call frame at which the normal
3350 return address is stored. For targets that return by popping an
3351 address off the stack, this might be a memory address just below
3352 the @emph{target} call frame rather than inside the current call
3353 frame. If defined, @code{EH_RETURN_STACKADJ_RTX} will have already
3354 been assigned, so it may be used to calculate the location of the
3357 Some targets have more complex requirements than storing to an
3358 address calculable during initial code generation. In that case
3359 the @code{eh_return} instruction pattern should be used instead.
3361 If you want to support call frame exception handling, you must
3362 define either this macro or the @code{eh_return} instruction pattern.
3365 @defmac RETURN_ADDR_OFFSET
3366 If defined, an integer-valued C expression for which rtl will be generated
3367 to add it to the exception handler address before it is searched in the
3368 exception handling tables, and to subtract it again from the address before
3369 using it to return to the exception handler.
3372 @defmac ASM_PREFERRED_EH_DATA_FORMAT (@var{code}, @var{global})
3373 This macro chooses the encoding of pointers embedded in the exception
3374 handling sections. If at all possible, this should be defined such
3375 that the exception handling section will not require dynamic relocations,
3376 and so may be read-only.
3378 @var{code} is 0 for data, 1 for code labels, 2 for function pointers.
3379 @var{global} is true if the symbol may be affected by dynamic relocations.
3380 The macro should return a combination of the @code{DW_EH_PE_*} defines
3381 as found in @file{dwarf2.h}.
3383 If this macro is not defined, pointers will not be encoded but
3384 represented directly.
3387 @defmac ASM_MAYBE_OUTPUT_ENCODED_ADDR_RTX (@var{file}, @var{encoding}, @var{size}, @var{addr}, @var{done})
3388 This macro allows the target to emit whatever special magic is required
3389 to represent the encoding chosen by @code{ASM_PREFERRED_EH_DATA_FORMAT}.
3390 Generic code takes care of pc-relative and indirect encodings; this must
3391 be defined if the target uses text-relative or data-relative encodings.
3393 This is a C statement that branches to @var{done} if the format was
3394 handled. @var{encoding} is the format chosen, @var{size} is the number
3395 of bytes that the format occupies, @var{addr} is the @code{SYMBOL_REF}
3399 @defmac MD_FALLBACK_FRAME_STATE_FOR (@var{context}, @var{fs})
3400 This macro allows the target to add CPU and operating system specific
3401 code to the call-frame unwinder for use when there is no unwind data
3402 available. The most common reason to implement this macro is to unwind
3403 through signal frames.
3405 This macro is called from @code{uw_frame_state_for} in
3406 @file{unwind-dw2.c}, @file{unwind-dw2-xtensa.c} and
3407 @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context};
3408 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{context->ra}
3409 for the address of the code being executed and @code{context->cfa} for
3410 the stack pointer value. If the frame can be decoded, the register
3411 save addresses should be updated in @var{fs} and the macro should
3412 evaluate to @code{_URC_NO_REASON}. If the frame cannot be decoded,
3413 the macro should evaluate to @code{_URC_END_OF_STACK}.
3415 For proper signal handling in Java this macro is accompanied by
3416 @code{MAKE_THROW_FRAME}, defined in @file{libjava/include/*-signal.h} headers.
3419 @defmac MD_HANDLE_UNWABI (@var{context}, @var{fs})
3420 This macro allows the target to add operating system specific code to the
3421 call-frame unwinder to handle the IA-64 @code{.unwabi} unwinding directive,
3422 usually used for signal or interrupt frames.
3424 This macro is called from @code{uw_update_context} in @file{unwind-ia64.c}.
3425 @var{context} is an @code{_Unwind_Context};
3426 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{fs->unwabi}
3427 for the abi and context in the @code{.unwabi} directive. If the
3428 @code{.unwabi} directive can be handled, the register save addresses should
3429 be updated in @var{fs}.
3432 @defmac TARGET_USES_WEAK_UNWIND_INFO
3433 A C expression that evaluates to true if the target requires unwind
3434 info to be given comdat linkage. Define it to be @code{1} if comdat
3435 linkage is necessary. The default is @code{0}.
3438 @node Stack Checking
3439 @subsection Specifying How Stack Checking is Done
3441 GCC will check that stack references are within the boundaries of the
3442 stack, if the option @option{-fstack-check} is specified, in one of
3447 If the value of the @code{STACK_CHECK_BUILTIN} macro is nonzero, GCC
3448 will assume that you have arranged for full stack checking to be done
3449 at appropriate places in the configuration files. GCC will not do
3450 other special processing.
3453 If @code{STACK_CHECK_BUILTIN} is zero and the value of the
3454 @code{STACK_CHECK_STATIC_BUILTIN} macro is nonzero, GCC will assume
3455 that you have arranged for static stack checking (checking of the
3456 static stack frame of functions) to be done at appropriate places
3457 in the configuration files. GCC will only emit code to do dynamic
3458 stack checking (checking on dynamic stack allocations) using the third
3462 If neither of the above are true, GCC will generate code to periodically
3463 ``probe'' the stack pointer using the values of the macros defined below.
3466 If neither STACK_CHECK_BUILTIN nor STACK_CHECK_STATIC_BUILTIN is defined,
3467 GCC will change its allocation strategy for large objects if the option
3468 @option{-fstack-check} is specified: they will always be allocated
3469 dynamically if their size exceeds @code{STACK_CHECK_MAX_VAR_SIZE} bytes.
3471 @defmac STACK_CHECK_BUILTIN
3472 A nonzero value if stack checking is done by the configuration files in a
3473 machine-dependent manner. You should define this macro if stack checking
3474 is required by the ABI of your machine or if you would like to do stack
3475 checking in some more efficient way than the generic approach. The default
3476 value of this macro is zero.
3479 @defmac STACK_CHECK_STATIC_BUILTIN
3480 A nonzero value if static stack checking is done by the configuration files
3481 in a machine-dependent manner. You should define this macro if you would
3482 like to do static stack checking in some more efficient way than the generic
3483 approach. The default value of this macro is zero.
3486 @defmac STACK_CHECK_PROBE_INTERVAL_EXP
3487 An integer specifying the interval at which GCC must generate stack probe
3488 instructions, defined as 2 raised to this integer. You will normally
3489 define this macro so that the interval be no larger than the size of
3490 the ``guard pages'' at the end of a stack area. The default value
3491 of 12 (4096-byte interval) is suitable for most systems.
3494 @defmac STACK_CHECK_MOVING_SP
3495 An integer which is nonzero if GCC should move the stack pointer page by page
3496 when doing probes. This can be necessary on systems where the stack pointer
3497 contains the bottom address of the memory area accessible to the executing
3498 thread at any point in time. In this situation an alternate signal stack
3499 is required in order to be able to recover from a stack overflow. The
3500 default value of this macro is zero.
3503 @defmac STACK_CHECK_PROTECT
3504 The number of bytes of stack needed to recover from a stack overflow, for
3505 languages where such a recovery is supported. The default value of 75 words
3506 with the @code{setjmp}/@code{longjmp}-based exception handling mechanism and
3507 8192 bytes with other exception handling mechanisms should be adequate for
3511 The following macros are relevant only if neither STACK_CHECK_BUILTIN
3512 nor STACK_CHECK_STATIC_BUILTIN is defined; you can omit them altogether
3513 in the opposite case.
3515 @defmac STACK_CHECK_MAX_FRAME_SIZE
3516 The maximum size of a stack frame, in bytes. GCC will generate probe
3517 instructions in non-leaf functions to ensure at least this many bytes of
3518 stack are available. If a stack frame is larger than this size, stack
3519 checking will not be reliable and GCC will issue a warning. The
3520 default is chosen so that GCC only generates one instruction on most
3521 systems. You should normally not change the default value of this macro.
3524 @defmac STACK_CHECK_FIXED_FRAME_SIZE
3525 GCC uses this value to generate the above warning message. It
3526 represents the amount of fixed frame used by a function, not including
3527 space for any callee-saved registers, temporaries and user variables.
3528 You need only specify an upper bound for this amount and will normally
3529 use the default of four words.
3532 @defmac STACK_CHECK_MAX_VAR_SIZE
3533 The maximum size, in bytes, of an object that GCC will place in the
3534 fixed area of the stack frame when the user specifies
3535 @option{-fstack-check}.
3536 GCC computed the default from the values of the above macros and you will
3537 normally not need to override that default.
3541 @node Frame Registers
3542 @subsection Registers That Address the Stack Frame
3544 @c prevent bad page break with this line
3545 This discusses registers that address the stack frame.
3547 @defmac STACK_POINTER_REGNUM
3548 The register number of the stack pointer register, which must also be a
3549 fixed register according to @code{FIXED_REGISTERS}. On most machines,
3550 the hardware determines which register this is.
3553 @defmac FRAME_POINTER_REGNUM
3554 The register number of the frame pointer register, which is used to
3555 access automatic variables in the stack frame. On some machines, the
3556 hardware determines which register this is. On other machines, you can
3557 choose any register you wish for this purpose.
3560 @defmac HARD_FRAME_POINTER_REGNUM
3561 On some machines the offset between the frame pointer and starting
3562 offset of the automatic variables is not known until after register
3563 allocation has been done (for example, because the saved registers are
3564 between these two locations). On those machines, define
3565 @code{FRAME_POINTER_REGNUM} the number of a special, fixed register to
3566 be used internally until the offset is known, and define
3567 @code{HARD_FRAME_POINTER_REGNUM} to be the actual hard register number
3568 used for the frame pointer.
3570 You should define this macro only in the very rare circumstances when it
3571 is not possible to calculate the offset between the frame pointer and
3572 the automatic variables until after register allocation has been
3573 completed. When this macro is defined, you must also indicate in your
3574 definition of @code{ELIMINABLE_REGS} how to eliminate
3575 @code{FRAME_POINTER_REGNUM} into either @code{HARD_FRAME_POINTER_REGNUM}
3576 or @code{STACK_POINTER_REGNUM}.
3578 Do not define this macro if it would be the same as
3579 @code{FRAME_POINTER_REGNUM}.
3582 @defmac ARG_POINTER_REGNUM
3583 The register number of the arg pointer register, which is used to access
3584 the function's argument list. On some machines, this is the same as the
3585 frame pointer register. On some machines, the hardware determines which
3586 register this is. On other machines, you can choose any register you
3587 wish for this purpose. If this is not the same register as the frame
3588 pointer register, then you must mark it as a fixed register according to
3589 @code{FIXED_REGISTERS}, or arrange to be able to eliminate it
3590 (@pxref{Elimination}).
3593 @defmac HARD_FRAME_POINTER_IS_FRAME_POINTER
3594 Define this to a preprocessor constant that is nonzero if
3595 @code{hard_frame_pointer_rtx} and @code{frame_pointer_rtx} should be
3596 the same. The default definition is @samp{(HARD_FRAME_POINTER_REGNUM
3597 == FRAME_POINTER_REGNUM)}; you only need to define this macro if that
3598 definition is not suitable for use in preprocessor conditionals.
3601 @defmac HARD_FRAME_POINTER_IS_ARG_POINTER
3602 Define this to a preprocessor constant that is nonzero if
3603 @code{hard_frame_pointer_rtx} and @code{arg_pointer_rtx} should be the
3604 same. The default definition is @samp{(HARD_FRAME_POINTER_REGNUM ==
3605 ARG_POINTER_REGNUM)}; you only need to define this macro if that
3606 definition is not suitable for use in preprocessor conditionals.
3609 @defmac RETURN_ADDRESS_POINTER_REGNUM
3610 The register number of the return address pointer register, which is used to
3611 access the current function's return address from the stack. On some
3612 machines, the return address is not at a fixed offset from the frame
3613 pointer or stack pointer or argument pointer. This register can be defined
3614 to point to the return address on the stack, and then be converted by
3615 @code{ELIMINABLE_REGS} into either the frame pointer or stack pointer.
3617 Do not define this macro unless there is no other way to get the return
3618 address from the stack.
3621 @defmac STATIC_CHAIN_REGNUM
3622 @defmacx STATIC_CHAIN_INCOMING_REGNUM
3623 Register numbers used for passing a function's static chain pointer. If
3624 register windows are used, the register number as seen by the called
3625 function is @code{STATIC_CHAIN_INCOMING_REGNUM}, while the register
3626 number as seen by the calling function is @code{STATIC_CHAIN_REGNUM}. If
3627 these registers are the same, @code{STATIC_CHAIN_INCOMING_REGNUM} need
3630 The static chain register need not be a fixed register.
3632 If the static chain is passed in memory, these macros should not be
3633 defined; instead, the @code{TARGET_STATIC_CHAIN} hook should be used.
3636 @hook TARGET_STATIC_CHAIN
3637 This hook replaces the use of @code{STATIC_CHAIN_REGNUM} et al for
3638 targets that may use different static chain locations for different
3639 nested functions. This may be required if the target has function
3640 attributes that affect the calling conventions of the function and
3641 those calling conventions use different static chain locations.
3643 The default version of this hook uses @code{STATIC_CHAIN_REGNUM} et al.
3645 If the static chain is passed in memory, this hook should be used to
3646 provide rtx giving @code{mem} expressions that denote where they are stored.
3647 Often the @code{mem} expression as seen by the caller will be at an offset
3648 from the stack pointer and the @code{mem} expression as seen by the callee
3649 will be at an offset from the frame pointer.
3650 @findex stack_pointer_rtx
3651 @findex frame_pointer_rtx
3652 @findex arg_pointer_rtx
3653 The variables @code{stack_pointer_rtx}, @code{frame_pointer_rtx}, and
3654 @code{arg_pointer_rtx} will have been initialized and should be used
3655 to refer to those items.
3658 @defmac DWARF_FRAME_REGISTERS
3659 This macro specifies the maximum number of hard registers that can be
3660 saved in a call frame. This is used to size data structures used in
3661 DWARF2 exception handling.
3663 Prior to GCC 3.0, this macro was needed in order to establish a stable
3664 exception handling ABI in the face of adding new hard registers for ISA
3665 extensions. In GCC 3.0 and later, the EH ABI is insulated from changes
3666 in the number of hard registers. Nevertheless, this macro can still be
3667 used to reduce the runtime memory requirements of the exception handling
3668 routines, which can be substantial if the ISA contains a lot of
3669 registers that are not call-saved.
3671 If this macro is not defined, it defaults to
3672 @code{FIRST_PSEUDO_REGISTER}.
3675 @defmac PRE_GCC3_DWARF_FRAME_REGISTERS
3677 This macro is similar to @code{DWARF_FRAME_REGISTERS}, but is provided
3678 for backward compatibility in pre GCC 3.0 compiled code.
3680 If this macro is not defined, it defaults to
3681 @code{DWARF_FRAME_REGISTERS}.
3684 @defmac DWARF_REG_TO_UNWIND_COLUMN (@var{regno})
3686 Define this macro if the target's representation for dwarf registers
3687 is different than the internal representation for unwind column.
3688 Given a dwarf register, this macro should return the internal unwind
3689 column number to use instead.
3691 See the PowerPC's SPE target for an example.
3694 @defmac DWARF_FRAME_REGNUM (@var{regno})
3696 Define this macro if the target's representation for dwarf registers
3697 used in .eh_frame or .debug_frame is different from that used in other
3698 debug info sections. Given a GCC hard register number, this macro
3699 should return the .eh_frame register number. The default is
3700 @code{DBX_REGISTER_NUMBER (@var{regno})}.
3704 @defmac DWARF2_FRAME_REG_OUT (@var{regno}, @var{for_eh})
3706 Define this macro to map register numbers held in the call frame info
3707 that GCC has collected using @code{DWARF_FRAME_REGNUM} to those that
3708 should be output in .debug_frame (@code{@var{for_eh}} is zero) and
3709 .eh_frame (@code{@var{for_eh}} is nonzero). The default is to
3710 return @code{@var{regno}}.
3714 @defmac REG_VALUE_IN_UNWIND_CONTEXT
3716 Define this macro if the target stores register values as
3717 @code{_Unwind_Word} type in unwind context. It should be defined if
3718 target register size is larger than the size of @code{void *}. The
3719 default is to store register values as @code{void *} type.
3723 @defmac ASSUME_EXTENDED_UNWIND_CONTEXT
3725 Define this macro to be 1 if the target always uses extended unwind
3726 context with version, args_size and by_value fields. If it is undefined,
3727 it will be defined to 1 when @code{REG_VALUE_IN_UNWIND_CONTEXT} is
3728 defined and 0 otherwise.
3733 @subsection Eliminating Frame Pointer and Arg Pointer
3735 @c prevent bad page break with this line
3736 This is about eliminating the frame pointer and arg pointer.
3738 @hook TARGET_FRAME_POINTER_REQUIRED
3739 This target hook should return @code{true} if a function must have and use
3740 a frame pointer. This target hook is called in the reload pass. If its return
3741 value is @code{true} the function will have a frame pointer.
3743 This target hook can in principle examine the current function and decide
3744 according to the facts, but on most machines the constant @code{false} or the
3745 constant @code{true} suffices. Use @code{false} when the machine allows code
3746 to be generated with no frame pointer, and doing so saves some time or space.
3747 Use @code{true} when there is no possible advantage to avoiding a frame
3750 In certain cases, the compiler does not know how to produce valid code
3751 without a frame pointer. The compiler recognizes those cases and
3752 automatically gives the function a frame pointer regardless of what
3753 @code{TARGET_FRAME_POINTER_REQUIRED} returns. You don't need to worry about
3756 In a function that does not require a frame pointer, the frame pointer
3757 register can be allocated for ordinary usage, unless you mark it as a
3758 fixed register. See @code{FIXED_REGISTERS} for more information.
3760 Default return value is @code{false}.
3763 @findex get_frame_size
3764 @defmac INITIAL_FRAME_POINTER_OFFSET (@var{depth-var})
3765 A C statement to store in the variable @var{depth-var} the difference
3766 between the frame pointer and the stack pointer values immediately after
3767 the function prologue. The value would be computed from information
3768 such as the result of @code{get_frame_size ()} and the tables of
3769 registers @code{regs_ever_live} and @code{call_used_regs}.
3771 If @code{ELIMINABLE_REGS} is defined, this macro will be not be used and
3772 need not be defined. Otherwise, it must be defined even if
3773 @code{TARGET_FRAME_POINTER_REQUIRED} always returns true; in that
3774 case, you may set @var{depth-var} to anything.
3777 @defmac ELIMINABLE_REGS
3778 If defined, this macro specifies a table of register pairs used to
3779 eliminate unneeded registers that point into the stack frame. If it is not
3780 defined, the only elimination attempted by the compiler is to replace
3781 references to the frame pointer with references to the stack pointer.
3783 The definition of this macro is a list of structure initializations, each
3784 of which specifies an original and replacement register.
3786 On some machines, the position of the argument pointer is not known until
3787 the compilation is completed. In such a case, a separate hard register
3788 must be used for the argument pointer. This register can be eliminated by
3789 replacing it with either the frame pointer or the argument pointer,
3790 depending on whether or not the frame pointer has been eliminated.
3792 In this case, you might specify:
3794 #define ELIMINABLE_REGS \
3795 @{@{ARG_POINTER_REGNUM, STACK_POINTER_REGNUM@}, \
3796 @{ARG_POINTER_REGNUM, FRAME_POINTER_REGNUM@}, \
3797 @{FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM@}@}
3800 Note that the elimination of the argument pointer with the stack pointer is
3801 specified first since that is the preferred elimination.
3804 @hook TARGET_CAN_ELIMINATE
3805 This target hook should returns @code{true} if the compiler is allowed to
3806 try to replace register number @var{from_reg} with register number
3807 @var{to_reg}. This target hook need only be defined if @code{ELIMINABLE_REGS}
3808 is defined, and will usually be @code{true}, since most of the cases
3809 preventing register elimination are things that the compiler already
3812 Default return value is @code{true}.
3815 @defmac INITIAL_ELIMINATION_OFFSET (@var{from-reg}, @var{to-reg}, @var{offset-var})
3816 This macro is similar to @code{INITIAL_FRAME_POINTER_OFFSET}. It
3817 specifies the initial difference between the specified pair of
3818 registers. This macro must be defined if @code{ELIMINABLE_REGS} is
3822 @node Stack Arguments
3823 @subsection Passing Function Arguments on the Stack
3824 @cindex arguments on stack
3825 @cindex stack arguments
3827 The macros in this section control how arguments are passed
3828 on the stack. See the following section for other macros that
3829 control passing certain arguments in registers.
3831 @hook TARGET_PROMOTE_PROTOTYPES
3832 This target hook returns @code{true} if an argument declared in a
3833 prototype as an integral type smaller than @code{int} should actually be
3834 passed as an @code{int}. In addition to avoiding errors in certain
3835 cases of mismatch, it also makes for better code on certain machines.
3836 The default is to not promote prototypes.
3840 A C expression. If nonzero, push insns will be used to pass
3842 If the target machine does not have a push instruction, set it to zero.
3843 That directs GCC to use an alternate strategy: to
3844 allocate the entire argument block and then store the arguments into
3845 it. When @code{PUSH_ARGS} is nonzero, @code{PUSH_ROUNDING} must be defined too.
3848 @defmac PUSH_ARGS_REVERSED
3849 A C expression. If nonzero, function arguments will be evaluated from
3850 last to first, rather than from first to last. If this macro is not
3851 defined, it defaults to @code{PUSH_ARGS} on targets where the stack
3852 and args grow in opposite directions, and 0 otherwise.
3855 @defmac PUSH_ROUNDING (@var{npushed})
3856 A C expression that is the number of bytes actually pushed onto the
3857 stack when an instruction attempts to push @var{npushed} bytes.
3859 On some machines, the definition
3862 #define PUSH_ROUNDING(BYTES) (BYTES)
3866 will suffice. But on other machines, instructions that appear
3867 to push one byte actually push two bytes in an attempt to maintain
3868 alignment. Then the definition should be
3871 #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1)
3874 If the value of this macro has a type, it should be an unsigned type.
3877 @findex current_function_outgoing_args_size
3878 @defmac ACCUMULATE_OUTGOING_ARGS
3879 A C expression. If nonzero, the maximum amount of space required for outgoing arguments
3880 will be computed and placed into the variable
3881 @code{current_function_outgoing_args_size}. No space will be pushed
3882 onto the stack for each call; instead, the function prologue should
3883 increase the stack frame size by this amount.
3885 Setting both @code{PUSH_ARGS} and @code{ACCUMULATE_OUTGOING_ARGS}
3889 @defmac REG_PARM_STACK_SPACE (@var{fndecl})
3890 Define this macro if functions should assume that stack space has been
3891 allocated for arguments even when their values are passed in
3894 The value of this macro is the size, in bytes, of the area reserved for
3895 arguments passed in registers for the function represented by @var{fndecl},
3896 which can be zero if GCC is calling a library function.
3897 The argument @var{fndecl} can be the FUNCTION_DECL, or the type itself
3900 This space can be allocated by the caller, or be a part of the
3901 machine-dependent stack frame: @code{OUTGOING_REG_PARM_STACK_SPACE} says
3904 @c above is overfull. not sure what to do. --mew 5feb93 did
3905 @c something, not sure if it looks good. --mew 10feb93
3907 @defmac OUTGOING_REG_PARM_STACK_SPACE (@var{fntype})
3908 Define this to a nonzero value if it is the responsibility of the
3909 caller to allocate the area reserved for arguments passed in registers
3910 when calling a function of @var{fntype}. @var{fntype} may be NULL
3911 if the function called is a library function.
3913 If @code{ACCUMULATE_OUTGOING_ARGS} is defined, this macro controls
3914 whether the space for these arguments counts in the value of
3915 @code{current_function_outgoing_args_size}.
3918 @defmac STACK_PARMS_IN_REG_PARM_AREA
3919 Define this macro if @code{REG_PARM_STACK_SPACE} is defined, but the
3920 stack parameters don't skip the area specified by it.
3921 @c i changed this, makes more sens and it should have taken care of the
3922 @c overfull.. not as specific, tho. --mew 5feb93
3924 Normally, when a parameter is not passed in registers, it is placed on the
3925 stack beyond the @code{REG_PARM_STACK_SPACE} area. Defining this macro
3926 suppresses this behavior and causes the parameter to be passed on the
3927 stack in its natural location.
3930 @hook TARGET_RETURN_POPS_ARGS
3931 This target hook returns the number of bytes of its own arguments that
3932 a function pops on returning, or 0 if the function pops no arguments
3933 and the caller must therefore pop them all after the function returns.
3935 @var{fundecl} is a C variable whose value is a tree node that describes
3936 the function in question. Normally it is a node of type
3937 @code{FUNCTION_DECL} that describes the declaration of the function.
3938 From this you can obtain the @code{DECL_ATTRIBUTES} of the function.
3940 @var{funtype} is a C variable whose value is a tree node that
3941 describes the function in question. Normally it is a node of type
3942 @code{FUNCTION_TYPE} that describes the data type of the function.
3943 From this it is possible to obtain the data types of the value and
3944 arguments (if known).
3946 When a call to a library function is being considered, @var{fundecl}
3947 will contain an identifier node for the library function. Thus, if
3948 you need to distinguish among various library functions, you can do so
3949 by their names. Note that ``library function'' in this context means
3950 a function used to perform arithmetic, whose name is known specially
3951 in the compiler and was not mentioned in the C code being compiled.
3953 @var{size} is the number of bytes of arguments passed on the
3954 stack. If a variable number of bytes is passed, it is zero, and
3955 argument popping will always be the responsibility of the calling function.
3957 On the VAX, all functions always pop their arguments, so the definition
3958 of this macro is @var{size}. On the 68000, using the standard
3959 calling convention, no functions pop their arguments, so the value of
3960 the macro is always 0 in this case. But an alternative calling
3961 convention is available in which functions that take a fixed number of
3962 arguments pop them but other functions (such as @code{printf}) pop
3963 nothing (the caller pops all). When this convention is in use,
3964 @var{funtype} is examined to determine whether a function takes a fixed
3965 number of arguments.
3968 @defmac CALL_POPS_ARGS (@var{cum})
3969 A C expression that should indicate the number of bytes a call sequence
3970 pops off the stack. It is added to the value of @code{RETURN_POPS_ARGS}
3971 when compiling a function call.
3973 @var{cum} is the variable in which all arguments to the called function
3974 have been accumulated.
3976 On certain architectures, such as the SH5, a call trampoline is used
3977 that pops certain registers off the stack, depending on the arguments
3978 that have been passed to the function. Since this is a property of the
3979 call site, not of the called function, @code{RETURN_POPS_ARGS} is not
3983 @node Register Arguments
3984 @subsection Passing Arguments in Registers
3985 @cindex arguments in registers
3986 @cindex registers arguments
3988 This section describes the macros which let you control how various
3989 types of arguments are passed in registers or how they are arranged in
3992 @hook TARGET_FUNCTION_ARG
3993 Return an RTX indicating whether a function argument is passed in a
3994 register and if so, which register.
3996 The arguments are @var{ca}, which summarizes all the previous
3997 arguments; @var{mode}, the machine mode of the argument; @var{type},
3998 the data type of the argument as a tree node or 0 if that is not known
3999 (which happens for C support library functions); and @var{named},
4000 which is @code{true} for an ordinary argument and @code{false} for
4001 nameless arguments that correspond to @samp{@dots{}} in the called
4002 function's prototype. @var{type} can be an incomplete type if a
4003 syntax error has previously occurred.
4005 The return value is usually either a @code{reg} RTX for the hard
4006 register in which to pass the argument, or zero to pass the argument
4009 The value of the expression can also be a @code{parallel} RTX@. This is
4010 used when an argument is passed in multiple locations. The mode of the
4011 @code{parallel} should be the mode of the entire argument. The
4012 @code{parallel} holds any number of @code{expr_list} pairs; each one
4013 describes where part of the argument is passed. In each
4014 @code{expr_list} the first operand must be a @code{reg} RTX for the hard
4015 register in which to pass this part of the argument, and the mode of the
4016 register RTX indicates how large this part of the argument is. The
4017 second operand of the @code{expr_list} is a @code{const_int} which gives
4018 the offset in bytes into the entire argument of where this part starts.
4019 As a special exception the first @code{expr_list} in the @code{parallel}
4020 RTX may have a first operand of zero. This indicates that the entire
4021 argument is also stored on the stack.
4023 The last time this hook is called, it is called with @code{MODE ==
4024 VOIDmode}, and its result is passed to the @code{call} or @code{call_value}
4025 pattern as operands 2 and 3 respectively.
4027 @cindex @file{stdarg.h} and register arguments
4028 The usual way to make the ISO library @file{stdarg.h} work on a
4029 machine where some arguments are usually passed in registers, is to
4030 cause nameless arguments to be passed on the stack instead. This is
4031 done by making @code{TARGET_FUNCTION_ARG} return 0 whenever
4032 @var{named} is @code{false}.
4034 @cindex @code{TARGET_MUST_PASS_IN_STACK}, and @code{TARGET_FUNCTION_ARG}
4035 @cindex @code{REG_PARM_STACK_SPACE}, and @code{TARGET_FUNCTION_ARG}
4036 You may use the hook @code{targetm.calls.must_pass_in_stack}
4037 in the definition of this macro to determine if this argument is of a
4038 type that must be passed in the stack. If @code{REG_PARM_STACK_SPACE}
4039 is not defined and @code{TARGET_FUNCTION_ARG} returns nonzero for such an
4040 argument, the compiler will abort. If @code{REG_PARM_STACK_SPACE} is
4041 defined, the argument will be computed in the stack and then loaded into
4045 @hook TARGET_MUST_PASS_IN_STACK
4046 This target hook should return @code{true} if we should not pass @var{type}
4047 solely in registers. The file @file{expr.h} defines a
4048 definition that is usually appropriate, refer to @file{expr.h} for additional
4052 @hook TARGET_FUNCTION_INCOMING_ARG
4053 Define this hook if the target machine has ``register windows'', so
4054 that the register in which a function sees an arguments is not
4055 necessarily the same as the one in which the caller passed the
4058 For such machines, @code{TARGET_FUNCTION_ARG} computes the register in
4059 which the caller passes the value, and
4060 @code{TARGET_FUNCTION_INCOMING_ARG} should be defined in a similar
4061 fashion to tell the function being called where the arguments will
4064 If @code{TARGET_FUNCTION_INCOMING_ARG} is not defined,
4065 @code{TARGET_FUNCTION_ARG} serves both purposes.
4068 @hook TARGET_ARG_PARTIAL_BYTES
4069 This target hook returns the number of bytes at the beginning of an
4070 argument that must be put in registers. The value must be zero for
4071 arguments that are passed entirely in registers or that are entirely
4072 pushed on the stack.
4074 On some machines, certain arguments must be passed partially in
4075 registers and partially in memory. On these machines, typically the
4076 first few words of arguments are passed in registers, and the rest
4077 on the stack. If a multi-word argument (a @code{double} or a
4078 structure) crosses that boundary, its first few words must be passed
4079 in registers and the rest must be pushed. This macro tells the
4080 compiler when this occurs, and how many bytes should go in registers.
4082 @code{TARGET_FUNCTION_ARG} for these arguments should return the first
4083 register to be used by the caller for this argument; likewise
4084 @code{TARGET_FUNCTION_INCOMING_ARG}, for the called function.
4087 @hook TARGET_PASS_BY_REFERENCE
4088 This target hook should return @code{true} if an argument at the
4089 position indicated by @var{cum} should be passed by reference. This
4090 predicate is queried after target independent reasons for being
4091 passed by reference, such as @code{TREE_ADDRESSABLE (type)}.
4093 If the hook returns true, a copy of that argument is made in memory and a
4094 pointer to the argument is passed instead of the argument itself.
4095 The pointer is passed in whatever way is appropriate for passing a pointer
4099 @hook TARGET_CALLEE_COPIES
4100 The function argument described by the parameters to this hook is
4101 known to be passed by reference. The hook should return true if the
4102 function argument should be copied by the callee instead of copied
4105 For any argument for which the hook returns true, if it can be
4106 determined that the argument is not modified, then a copy need
4109 The default version of this hook always returns false.
4112 @defmac CUMULATIVE_ARGS
4113 A C type for declaring a variable that is used as the first argument
4114 of @code{TARGET_FUNCTION_ARG} and other related values. For some
4115 target machines, the type @code{int} suffices and can hold the number
4116 of bytes of argument so far.
4118 There is no need to record in @code{CUMULATIVE_ARGS} anything about the
4119 arguments that have been passed on the stack. The compiler has other
4120 variables to keep track of that. For target machines on which all
4121 arguments are passed on the stack, there is no need to store anything in
4122 @code{CUMULATIVE_ARGS}; however, the data structure must exist and
4123 should not be empty, so use @code{int}.
4126 @defmac OVERRIDE_ABI_FORMAT (@var{fndecl})
4127 If defined, this macro is called before generating any code for a
4128 function, but after the @var{cfun} descriptor for the function has been
4129 created. The back end may use this macro to update @var{cfun} to
4130 reflect an ABI other than that which would normally be used by default.
4131 If the compiler is generating code for a compiler-generated function,
4132 @var{fndecl} may be @code{NULL}.
4135 @defmac INIT_CUMULATIVE_ARGS (@var{cum}, @var{fntype}, @var{libname}, @var{fndecl}, @var{n_named_args})
4136 A C statement (sans semicolon) for initializing the variable
4137 @var{cum} for the state at the beginning of the argument list. The
4138 variable has type @code{CUMULATIVE_ARGS}. The value of @var{fntype}
4139 is the tree node for the data type of the function which will receive
4140 the args, or 0 if the args are to a compiler support library function.
4141 For direct calls that are not libcalls, @var{fndecl} contain the
4142 declaration node of the function. @var{fndecl} is also set when
4143 @code{INIT_CUMULATIVE_ARGS} is used to find arguments for the function
4144 being compiled. @var{n_named_args} is set to the number of named
4145 arguments, including a structure return address if it is passed as a
4146 parameter, when making a call. When processing incoming arguments,
4147 @var{n_named_args} is set to @minus{}1.
4149 When processing a call to a compiler support library function,
4150 @var{libname} identifies which one. It is a @code{symbol_ref} rtx which
4151 contains the name of the function, as a string. @var{libname} is 0 when
4152 an ordinary C function call is being processed. Thus, each time this
4153 macro is called, either @var{libname} or @var{fntype} is nonzero, but
4154 never both of them at once.
4157 @defmac INIT_CUMULATIVE_LIBCALL_ARGS (@var{cum}, @var{mode}, @var{libname})
4158 Like @code{INIT_CUMULATIVE_ARGS} but only used for outgoing libcalls,
4159 it gets a @code{MODE} argument instead of @var{fntype}, that would be
4160 @code{NULL}. @var{indirect} would always be zero, too. If this macro
4161 is not defined, @code{INIT_CUMULATIVE_ARGS (cum, NULL_RTX, libname,
4162 0)} is used instead.
4165 @defmac INIT_CUMULATIVE_INCOMING_ARGS (@var{cum}, @var{fntype}, @var{libname})
4166 Like @code{INIT_CUMULATIVE_ARGS} but overrides it for the purposes of
4167 finding the arguments for the function being compiled. If this macro is
4168 undefined, @code{INIT_CUMULATIVE_ARGS} is used instead.
4170 The value passed for @var{libname} is always 0, since library routines
4171 with special calling conventions are never compiled with GCC@. The
4172 argument @var{libname} exists for symmetry with
4173 @code{INIT_CUMULATIVE_ARGS}.
4174 @c could use "this macro" in place of @code{INIT_CUMULATIVE_ARGS}, maybe.
4175 @c --mew 5feb93 i switched the order of the sentences. --mew 10feb93
4178 @hook TARGET_FUNCTION_ARG_ADVANCE
4179 This hook updates the summarizer variable pointed to by @var{ca} to
4180 advance past an argument in the argument list. The values @var{mode},
4181 @var{type} and @var{named} describe that argument. Once this is done,
4182 the variable @var{cum} is suitable for analyzing the @emph{following}
4183 argument with @code{TARGET_FUNCTION_ARG}, etc.
4185 This hook need not do anything if the argument in question was passed
4186 on the stack. The compiler knows how to track the amount of stack space
4187 used for arguments without any special help.
4190 @defmac FUNCTION_ARG_OFFSET (@var{mode}, @var{type})
4191 If defined, a C expression that is the number of bytes to add to the
4192 offset of the argument passed in memory. This is needed for the SPU,
4193 which passes @code{char} and @code{short} arguments in the preferred
4194 slot that is in the middle of the quad word instead of starting at the
4198 @defmac FUNCTION_ARG_PADDING (@var{mode}, @var{type})
4199 If defined, a C expression which determines whether, and in which direction,
4200 to pad out an argument with extra space. The value should be of type
4201 @code{enum direction}: either @code{upward} to pad above the argument,
4202 @code{downward} to pad below, or @code{none} to inhibit padding.
4204 The @emph{amount} of padding is not controlled by this macro, but by the
4205 target hook @code{TARGET_FUNCTION_ARG_ROUND_BOUNDARY}. It is
4206 always just enough to reach the next multiple of that boundary.
4208 This macro has a default definition which is right for most systems.
4209 For little-endian machines, the default is to pad upward. For
4210 big-endian machines, the default is to pad downward for an argument of
4211 constant size shorter than an @code{int}, and upward otherwise.
4214 @defmac PAD_VARARGS_DOWN
4215 If defined, a C expression which determines whether the default
4216 implementation of va_arg will attempt to pad down before reading the
4217 next argument, if that argument is smaller than its aligned space as
4218 controlled by @code{PARM_BOUNDARY}. If this macro is not defined, all such
4219 arguments are padded down if @code{BYTES_BIG_ENDIAN} is true.
4222 @defmac BLOCK_REG_PADDING (@var{mode}, @var{type}, @var{first})
4223 Specify padding for the last element of a block move between registers and
4224 memory. @var{first} is nonzero if this is the only element. Defining this
4225 macro allows better control of register function parameters on big-endian
4226 machines, without using @code{PARALLEL} rtl. In particular,
4227 @code{MUST_PASS_IN_STACK} need not test padding and mode of types in
4228 registers, as there is no longer a "wrong" part of a register; For example,
4229 a three byte aggregate may be passed in the high part of a register if so
4233 @hook TARGET_FUNCTION_ARG_BOUNDARY
4234 This hook returns the alignment boundary, in bits, of an argument
4235 with the specified mode and type. The default hook returns
4236 @code{PARM_BOUNDARY} for all arguments.
4239 @hook TARGET_FUNCTION_ARG_ROUND_BOUNDARY
4241 @defmac FUNCTION_ARG_REGNO_P (@var{regno})
4242 A C expression that is nonzero if @var{regno} is the number of a hard
4243 register in which function arguments are sometimes passed. This does
4244 @emph{not} include implicit arguments such as the static chain and
4245 the structure-value address. On many machines, no registers can be
4246 used for this purpose since all function arguments are pushed on the
4250 @hook TARGET_SPLIT_COMPLEX_ARG
4251 This hook should return true if parameter of type @var{type} are passed
4252 as two scalar parameters. By default, GCC will attempt to pack complex
4253 arguments into the target's word size. Some ABIs require complex arguments
4254 to be split and treated as their individual components. For example, on
4255 AIX64, complex floats should be passed in a pair of floating point
4256 registers, even though a complex float would fit in one 64-bit floating
4259 The default value of this hook is @code{NULL}, which is treated as always
4263 @hook TARGET_BUILD_BUILTIN_VA_LIST
4264 This hook returns a type node for @code{va_list} for the target.
4265 The default version of the hook returns @code{void*}.
4268 @hook TARGET_ENUM_VA_LIST_P
4269 This target hook is used in function @code{c_common_nodes_and_builtins}
4270 to iterate through the target specific builtin types for va_list. The
4271 variable @var{idx} is used as iterator. @var{pname} has to be a pointer
4272 to a @code{const char *} and @var{ptree} a pointer to a @code{tree} typed
4274 The arguments @var{pname} and @var{ptree} are used to store the result of
4275 this macro and are set to the name of the va_list builtin type and its
4277 If the return value of this macro is zero, then there is no more element.
4278 Otherwise the @var{IDX} should be increased for the next call of this
4279 macro to iterate through all types.
4282 @hook TARGET_FN_ABI_VA_LIST
4283 This hook returns the va_list type of the calling convention specified by
4285 The default version of this hook returns @code{va_list_type_node}.
4288 @hook TARGET_CANONICAL_VA_LIST_TYPE
4289 This hook returns the va_list type of the calling convention specified by the
4290 type of @var{type}. If @var{type} is not a valid va_list type, it returns
4294 @hook TARGET_GIMPLIFY_VA_ARG_EXPR
4295 This hook performs target-specific gimplification of
4296 @code{VA_ARG_EXPR}. The first two parameters correspond to the
4297 arguments to @code{va_arg}; the latter two are as in
4298 @code{gimplify.c:gimplify_expr}.
4301 @hook TARGET_VALID_POINTER_MODE
4302 Define this to return nonzero if the port can handle pointers
4303 with machine mode @var{mode}. The default version of this
4304 hook returns true for both @code{ptr_mode} and @code{Pmode}.
4307 @hook TARGET_REF_MAY_ALIAS_ERRNO
4309 @hook TARGET_SCALAR_MODE_SUPPORTED_P
4310 Define this to return nonzero if the port is prepared to handle
4311 insns involving scalar mode @var{mode}. For a scalar mode to be
4312 considered supported, all the basic arithmetic and comparisons
4315 The default version of this hook returns true for any mode
4316 required to handle the basic C types (as defined by the port).
4317 Included here are the double-word arithmetic supported by the
4318 code in @file{optabs.c}.
4321 @hook TARGET_VECTOR_MODE_SUPPORTED_P
4322 Define this to return nonzero if the port is prepared to handle
4323 insns involving vector mode @var{mode}. At the very least, it
4324 must have move patterns for this mode.
4327 @hook TARGET_ARRAY_MODE_SUPPORTED_P
4329 @hook TARGET_SMALL_REGISTER_CLASSES_FOR_MODE_P
4330 Define this to return nonzero for machine modes for which the port has
4331 small register classes. If this target hook returns nonzero for a given
4332 @var{mode}, the compiler will try to minimize the lifetime of registers
4333 in @var{mode}. The hook may be called with @code{VOIDmode} as argument.
4334 In this case, the hook is expected to return nonzero if it returns nonzero
4337 On some machines, it is risky to let hard registers live across arbitrary
4338 insns. Typically, these machines have instructions that require values
4339 to be in specific registers (like an accumulator), and reload will fail
4340 if the required hard register is used for another purpose across such an
4343 Passes before reload do not know which hard registers will be used
4344 in an instruction, but the machine modes of the registers set or used in
4345 the instruction are already known. And for some machines, register
4346 classes are small for, say, integer registers but not for floating point
4347 registers. For example, the AMD x86-64 architecture requires specific
4348 registers for the legacy x86 integer instructions, but there are many
4349 SSE registers for floating point operations. On such targets, a good
4350 strategy may be to return nonzero from this hook for @code{INTEGRAL_MODE_P}
4351 machine modes but zero for the SSE register classes.
4353 The default version of this hook returns false for any mode. It is always
4354 safe to redefine this hook to return with a nonzero value. But if you
4355 unnecessarily define it, you will reduce the amount of optimizations
4356 that can be performed in some cases. If you do not define this hook
4357 to return a nonzero value when it is required, the compiler will run out
4358 of spill registers and print a fatal error message.
4361 @hook TARGET_FLAGS_REGNUM
4364 @subsection How Scalar Function Values Are Returned
4365 @cindex return values in registers
4366 @cindex values, returned by functions
4367 @cindex scalars, returned as values
4369 This section discusses the macros that control returning scalars as
4370 values---values that can fit in registers.
4372 @hook TARGET_FUNCTION_VALUE
4374 Define this to return an RTX representing the place where a function
4375 returns or receives a value of data type @var{ret_type}, a tree node
4376 representing a data type. @var{fn_decl_or_type} is a tree node
4377 representing @code{FUNCTION_DECL} or @code{FUNCTION_TYPE} of a
4378 function being called. If @var{outgoing} is false, the hook should
4379 compute the register in which the caller will see the return value.
4380 Otherwise, the hook should return an RTX representing the place where
4381 a function returns a value.
4383 On many machines, only @code{TYPE_MODE (@var{ret_type})} is relevant.
4384 (Actually, on most machines, scalar values are returned in the same
4385 place regardless of mode.) The value of the expression is usually a
4386 @code{reg} RTX for the hard register where the return value is stored.
4387 The value can also be a @code{parallel} RTX, if the return value is in
4388 multiple places. See @code{TARGET_FUNCTION_ARG} for an explanation of the
4389 @code{parallel} form. Note that the callee will populate every
4390 location specified in the @code{parallel}, but if the first element of
4391 the @code{parallel} contains the whole return value, callers will use
4392 that element as the canonical location and ignore the others. The m68k
4393 port uses this type of @code{parallel} to return pointers in both
4394 @samp{%a0} (the canonical location) and @samp{%d0}.
4396 If @code{TARGET_PROMOTE_FUNCTION_RETURN} returns true, you must apply
4397 the same promotion rules specified in @code{PROMOTE_MODE} if
4398 @var{valtype} is a scalar type.
4400 If the precise function being called is known, @var{func} is a tree
4401 node (@code{FUNCTION_DECL}) for it; otherwise, @var{func} is a null
4402 pointer. This makes it possible to use a different value-returning
4403 convention for specific functions when all their calls are
4406 Some target machines have ``register windows'' so that the register in
4407 which a function returns its value is not the same as the one in which
4408 the caller sees the value. For such machines, you should return
4409 different RTX depending on @var{outgoing}.
4411 @code{TARGET_FUNCTION_VALUE} is not used for return values with
4412 aggregate data types, because these are returned in another way. See
4413 @code{TARGET_STRUCT_VALUE_RTX} and related macros, below.
4416 @defmac FUNCTION_VALUE (@var{valtype}, @var{func})
4417 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE} for
4418 a new target instead.
4421 @defmac LIBCALL_VALUE (@var{mode})
4422 A C expression to create an RTX representing the place where a library
4423 function returns a value of mode @var{mode}.
4425 Note that ``library function'' in this context means a compiler
4426 support routine, used to perform arithmetic, whose name is known
4427 specially by the compiler and was not mentioned in the C code being
4431 @hook TARGET_LIBCALL_VALUE
4432 Define this hook if the back-end needs to know the name of the libcall
4433 function in order to determine where the result should be returned.
4435 The mode of the result is given by @var{mode} and the name of the called
4436 library function is given by @var{fun}. The hook should return an RTX
4437 representing the place where the library function result will be returned.
4439 If this hook is not defined, then LIBCALL_VALUE will be used.
4442 @defmac FUNCTION_VALUE_REGNO_P (@var{regno})
4443 A C expression that is nonzero if @var{regno} is the number of a hard
4444 register in which the values of called function may come back.
4446 A register whose use for returning values is limited to serving as the
4447 second of a pair (for a value of type @code{double}, say) need not be
4448 recognized by this macro. So for most machines, this definition
4452 #define FUNCTION_VALUE_REGNO_P(N) ((N) == 0)
4455 If the machine has register windows, so that the caller and the called
4456 function use different registers for the return value, this macro
4457 should recognize only the caller's register numbers.
4459 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE_REGNO_P}
4460 for a new target instead.
4463 @hook TARGET_FUNCTION_VALUE_REGNO_P
4464 A target hook that return @code{true} if @var{regno} is the number of a hard
4465 register in which the values of called function may come back.
4467 A register whose use for returning values is limited to serving as the
4468 second of a pair (for a value of type @code{double}, say) need not be
4469 recognized by this target hook.
4471 If the machine has register windows, so that the caller and the called
4472 function use different registers for the return value, this target hook
4473 should recognize only the caller's register numbers.
4475 If this hook is not defined, then FUNCTION_VALUE_REGNO_P will be used.
4478 @defmac APPLY_RESULT_SIZE
4479 Define this macro if @samp{untyped_call} and @samp{untyped_return}
4480 need more space than is implied by @code{FUNCTION_VALUE_REGNO_P} for
4481 saving and restoring an arbitrary return value.
4484 @hook TARGET_RETURN_IN_MSB
4485 This hook should return true if values of type @var{type} are returned
4486 at the most significant end of a register (in other words, if they are
4487 padded at the least significant end). You can assume that @var{type}
4488 is returned in a register; the caller is required to check this.
4490 Note that the register provided by @code{TARGET_FUNCTION_VALUE} must
4491 be able to hold the complete return value. For example, if a 1-, 2-
4492 or 3-byte structure is returned at the most significant end of a
4493 4-byte register, @code{TARGET_FUNCTION_VALUE} should provide an
4497 @node Aggregate Return
4498 @subsection How Large Values Are Returned
4499 @cindex aggregates as return values
4500 @cindex large return values
4501 @cindex returning aggregate values
4502 @cindex structure value address
4504 When a function value's mode is @code{BLKmode} (and in some other
4505 cases), the value is not returned according to
4506 @code{TARGET_FUNCTION_VALUE} (@pxref{Scalar Return}). Instead, the
4507 caller passes the address of a block of memory in which the value
4508 should be stored. This address is called the @dfn{structure value
4511 This section describes how to control returning structure values in
4514 @hook TARGET_RETURN_IN_MEMORY
4515 This target hook should return a nonzero value to say to return the
4516 function value in memory, just as large structures are always returned.
4517 Here @var{type} will be the data type of the value, and @var{fntype}
4518 will be the type of the function doing the returning, or @code{NULL} for
4521 Note that values of mode @code{BLKmode} must be explicitly handled
4522 by this function. Also, the option @option{-fpcc-struct-return}
4523 takes effect regardless of this macro. On most systems, it is
4524 possible to leave the hook undefined; this causes a default
4525 definition to be used, whose value is the constant 1 for @code{BLKmode}
4526 values, and 0 otherwise.
4528 Do not use this hook to indicate that structures and unions should always
4529 be returned in memory. You should instead use @code{DEFAULT_PCC_STRUCT_RETURN}
4533 @defmac DEFAULT_PCC_STRUCT_RETURN
4534 Define this macro to be 1 if all structure and union return values must be
4535 in memory. Since this results in slower code, this should be defined
4536 only if needed for compatibility with other compilers or with an ABI@.
4537 If you define this macro to be 0, then the conventions used for structure
4538 and union return values are decided by the @code{TARGET_RETURN_IN_MEMORY}
4541 If not defined, this defaults to the value 1.
4544 @hook TARGET_STRUCT_VALUE_RTX
4545 This target hook should return the location of the structure value
4546 address (normally a @code{mem} or @code{reg}), or 0 if the address is
4547 passed as an ``invisible'' first argument. Note that @var{fndecl} may
4548 be @code{NULL}, for libcalls. You do not need to define this target
4549 hook if the address is always passed as an ``invisible'' first
4552 On some architectures the place where the structure value address
4553 is found by the called function is not the same place that the
4554 caller put it. This can be due to register windows, or it could
4555 be because the function prologue moves it to a different place.
4556 @var{incoming} is @code{1} or @code{2} when the location is needed in
4557 the context of the called function, and @code{0} in the context of
4560 If @var{incoming} is nonzero and the address is to be found on the
4561 stack, return a @code{mem} which refers to the frame pointer. If
4562 @var{incoming} is @code{2}, the result is being used to fetch the
4563 structure value address at the beginning of a function. If you need
4564 to emit adjusting code, you should do it at this point.
4567 @defmac PCC_STATIC_STRUCT_RETURN
4568 Define this macro if the usual system convention on the target machine
4569 for returning structures and unions is for the called function to return
4570 the address of a static variable containing the value.
4572 Do not define this if the usual system convention is for the caller to
4573 pass an address to the subroutine.
4575 This macro has effect in @option{-fpcc-struct-return} mode, but it does
4576 nothing when you use @option{-freg-struct-return} mode.
4579 @hook TARGET_GET_RAW_RESULT_MODE
4581 @hook TARGET_GET_RAW_ARG_MODE
4584 @subsection Caller-Saves Register Allocation
4586 If you enable it, GCC can save registers around function calls. This
4587 makes it possible to use call-clobbered registers to hold variables that
4588 must live across calls.
4590 @defmac CALLER_SAVE_PROFITABLE (@var{refs}, @var{calls})
4591 A C expression to determine whether it is worthwhile to consider placing
4592 a pseudo-register in a call-clobbered hard register and saving and
4593 restoring it around each function call. The expression should be 1 when
4594 this is worth doing, and 0 otherwise.
4596 If you don't define this macro, a default is used which is good on most
4597 machines: @code{4 * @var{calls} < @var{refs}}.
4600 @defmac HARD_REGNO_CALLER_SAVE_MODE (@var{regno}, @var{nregs})
4601 A C expression specifying which mode is required for saving @var{nregs}
4602 of a pseudo-register in call-clobbered hard register @var{regno}. If
4603 @var{regno} is unsuitable for caller save, @code{VOIDmode} should be
4604 returned. For most machines this macro need not be defined since GCC
4605 will select the smallest suitable mode.
4608 @node Function Entry
4609 @subsection Function Entry and Exit
4610 @cindex function entry and exit
4614 This section describes the macros that output function entry
4615 (@dfn{prologue}) and exit (@dfn{epilogue}) code.
4617 @hook TARGET_ASM_FUNCTION_PROLOGUE
4618 If defined, a function that outputs the assembler code for entry to a
4619 function. The prologue is responsible for setting up the stack frame,
4620 initializing the frame pointer register, saving registers that must be
4621 saved, and allocating @var{size} additional bytes of storage for the
4622 local variables. @var{size} is an integer. @var{file} is a stdio
4623 stream to which the assembler code should be output.
4625 The label for the beginning of the function need not be output by this
4626 macro. That has already been done when the macro is run.
4628 @findex regs_ever_live
4629 To determine which registers to save, the macro can refer to the array
4630 @code{regs_ever_live}: element @var{r} is nonzero if hard register
4631 @var{r} is used anywhere within the function. This implies the function
4632 prologue should save register @var{r}, provided it is not one of the
4633 call-used registers. (@code{TARGET_ASM_FUNCTION_EPILOGUE} must likewise use
4634 @code{regs_ever_live}.)
4636 On machines that have ``register windows'', the function entry code does
4637 not save on the stack the registers that are in the windows, even if
4638 they are supposed to be preserved by function calls; instead it takes
4639 appropriate steps to ``push'' the register stack, if any non-call-used
4640 registers are used in the function.
4642 @findex frame_pointer_needed
4643 On machines where functions may or may not have frame-pointers, the
4644 function entry code must vary accordingly; it must set up the frame
4645 pointer if one is wanted, and not otherwise. To determine whether a
4646 frame pointer is in wanted, the macro can refer to the variable
4647 @code{frame_pointer_needed}. The variable's value will be 1 at run
4648 time in a function that needs a frame pointer. @xref{Elimination}.
4650 The function entry code is responsible for allocating any stack space
4651 required for the function. This stack space consists of the regions
4652 listed below. In most cases, these regions are allocated in the
4653 order listed, with the last listed region closest to the top of the
4654 stack (the lowest address if @code{STACK_GROWS_DOWNWARD} is defined, and
4655 the highest address if it is not defined). You can use a different order
4656 for a machine if doing so is more convenient or required for
4657 compatibility reasons. Except in cases where required by standard
4658 or by a debugger, there is no reason why the stack layout used by GCC
4659 need agree with that used by other compilers for a machine.
4662 @hook TARGET_ASM_FUNCTION_END_PROLOGUE
4663 If defined, a function that outputs assembler code at the end of a
4664 prologue. This should be used when the function prologue is being
4665 emitted as RTL, and you have some extra assembler that needs to be
4666 emitted. @xref{prologue instruction pattern}.
4669 @hook TARGET_ASM_FUNCTION_BEGIN_EPILOGUE
4670 If defined, a function that outputs assembler code at the start of an
4671 epilogue. This should be used when the function epilogue is being
4672 emitted as RTL, and you have some extra assembler that needs to be
4673 emitted. @xref{epilogue instruction pattern}.
4676 @hook TARGET_ASM_FUNCTION_EPILOGUE
4677 If defined, a function that outputs the assembler code for exit from a
4678 function. The epilogue is responsible for restoring the saved
4679 registers and stack pointer to their values when the function was
4680 called, and returning control to the caller. This macro takes the
4681 same arguments as the macro @code{TARGET_ASM_FUNCTION_PROLOGUE}, and the
4682 registers to restore are determined from @code{regs_ever_live} and
4683 @code{CALL_USED_REGISTERS} in the same way.
4685 On some machines, there is a single instruction that does all the work
4686 of returning from the function. On these machines, give that
4687 instruction the name @samp{return} and do not define the macro
4688 @code{TARGET_ASM_FUNCTION_EPILOGUE} at all.
4690 Do not define a pattern named @samp{return} if you want the
4691 @code{TARGET_ASM_FUNCTION_EPILOGUE} to be used. If you want the target
4692 switches to control whether return instructions or epilogues are used,
4693 define a @samp{return} pattern with a validity condition that tests the
4694 target switches appropriately. If the @samp{return} pattern's validity
4695 condition is false, epilogues will be used.
4697 On machines where functions may or may not have frame-pointers, the
4698 function exit code must vary accordingly. Sometimes the code for these
4699 two cases is completely different. To determine whether a frame pointer
4700 is wanted, the macro can refer to the variable
4701 @code{frame_pointer_needed}. The variable's value will be 1 when compiling
4702 a function that needs a frame pointer.
4704 Normally, @code{TARGET_ASM_FUNCTION_PROLOGUE} and
4705 @code{TARGET_ASM_FUNCTION_EPILOGUE} must treat leaf functions specially.
4706 The C variable @code{current_function_is_leaf} is nonzero for such a
4707 function. @xref{Leaf Functions}.
4709 On some machines, some functions pop their arguments on exit while
4710 others leave that for the caller to do. For example, the 68020 when
4711 given @option{-mrtd} pops arguments in functions that take a fixed
4712 number of arguments.
4714 @findex current_function_pops_args
4715 Your definition of the macro @code{RETURN_POPS_ARGS} decides which
4716 functions pop their own arguments. @code{TARGET_ASM_FUNCTION_EPILOGUE}
4717 needs to know what was decided. The number of bytes of the current
4718 function's arguments that this function should pop is available in
4719 @code{crtl->args.pops_args}. @xref{Scalar Return}.
4724 @findex current_function_pretend_args_size
4725 A region of @code{current_function_pretend_args_size} bytes of
4726 uninitialized space just underneath the first argument arriving on the
4727 stack. (This may not be at the very start of the allocated stack region
4728 if the calling sequence has pushed anything else since pushing the stack
4729 arguments. But usually, on such machines, nothing else has been pushed
4730 yet, because the function prologue itself does all the pushing.) This
4731 region is used on machines where an argument may be passed partly in
4732 registers and partly in memory, and, in some cases to support the
4733 features in @code{<stdarg.h>}.
4736 An area of memory used to save certain registers used by the function.
4737 The size of this area, which may also include space for such things as
4738 the return address and pointers to previous stack frames, is
4739 machine-specific and usually depends on which registers have been used
4740 in the function. Machines with register windows often do not require
4744 A region of at least @var{size} bytes, possibly rounded up to an allocation
4745 boundary, to contain the local variables of the function. On some machines,
4746 this region and the save area may occur in the opposite order, with the
4747 save area closer to the top of the stack.
4750 @cindex @code{ACCUMULATE_OUTGOING_ARGS} and stack frames
4751 Optionally, when @code{ACCUMULATE_OUTGOING_ARGS} is defined, a region of
4752 @code{current_function_outgoing_args_size} bytes to be used for outgoing
4753 argument lists of the function. @xref{Stack Arguments}.
4756 @defmac EXIT_IGNORE_STACK
4757 Define this macro as a C expression that is nonzero if the return
4758 instruction or the function epilogue ignores the value of the stack
4759 pointer; in other words, if it is safe to delete an instruction to
4760 adjust the stack pointer before a return from the function. The
4763 Note that this macro's value is relevant only for functions for which
4764 frame pointers are maintained. It is never safe to delete a final
4765 stack adjustment in a function that has no frame pointer, and the
4766 compiler knows this regardless of @code{EXIT_IGNORE_STACK}.
4769 @defmac EPILOGUE_USES (@var{regno})
4770 Define this macro as a C expression that is nonzero for registers that are
4771 used by the epilogue or the @samp{return} pattern. The stack and frame
4772 pointer registers are already assumed to be used as needed.
4775 @defmac EH_USES (@var{regno})
4776 Define this macro as a C expression that is nonzero for registers that are
4777 used by the exception handling mechanism, and so should be considered live
4778 on entry to an exception edge.
4781 @defmac DELAY_SLOTS_FOR_EPILOGUE
4782 Define this macro if the function epilogue contains delay slots to which
4783 instructions from the rest of the function can be ``moved''. The
4784 definition should be a C expression whose value is an integer
4785 representing the number of delay slots there.
4788 @defmac ELIGIBLE_FOR_EPILOGUE_DELAY (@var{insn}, @var{n})
4789 A C expression that returns 1 if @var{insn} can be placed in delay
4790 slot number @var{n} of the epilogue.
4792 The argument @var{n} is an integer which identifies the delay slot now
4793 being considered (since different slots may have different rules of
4794 eligibility). It is never negative and is always less than the number
4795 of epilogue delay slots (what @code{DELAY_SLOTS_FOR_EPILOGUE} returns).
4796 If you reject a particular insn for a given delay slot, in principle, it
4797 may be reconsidered for a subsequent delay slot. Also, other insns may
4798 (at least in principle) be considered for the so far unfilled delay
4801 @findex current_function_epilogue_delay_list
4802 @findex final_scan_insn
4803 The insns accepted to fill the epilogue delay slots are put in an RTL
4804 list made with @code{insn_list} objects, stored in the variable
4805 @code{current_function_epilogue_delay_list}. The insn for the first
4806 delay slot comes first in the list. Your definition of the macro
4807 @code{TARGET_ASM_FUNCTION_EPILOGUE} should fill the delay slots by
4808 outputting the insns in this list, usually by calling
4809 @code{final_scan_insn}.
4811 You need not define this macro if you did not define
4812 @code{DELAY_SLOTS_FOR_EPILOGUE}.
4815 @hook TARGET_ASM_OUTPUT_MI_THUNK
4816 A function that outputs the assembler code for a thunk
4817 function, used to implement C++ virtual function calls with multiple
4818 inheritance. The thunk acts as a wrapper around a virtual function,
4819 adjusting the implicit object parameter before handing control off to
4822 First, emit code to add the integer @var{delta} to the location that
4823 contains the incoming first argument. Assume that this argument
4824 contains a pointer, and is the one used to pass the @code{this} pointer
4825 in C++. This is the incoming argument @emph{before} the function prologue,
4826 e.g.@: @samp{%o0} on a sparc. The addition must preserve the values of
4827 all other incoming arguments.
4829 Then, if @var{vcall_offset} is nonzero, an additional adjustment should be
4830 made after adding @code{delta}. In particular, if @var{p} is the
4831 adjusted pointer, the following adjustment should be made:
4834 p += (*((ptrdiff_t **)p))[vcall_offset/sizeof(ptrdiff_t)]
4837 After the additions, emit code to jump to @var{function}, which is a
4838 @code{FUNCTION_DECL}. This is a direct pure jump, not a call, and does
4839 not touch the return address. Hence returning from @var{FUNCTION} will
4840 return to whoever called the current @samp{thunk}.
4842 The effect must be as if @var{function} had been called directly with
4843 the adjusted first argument. This macro is responsible for emitting all
4844 of the code for a thunk function; @code{TARGET_ASM_FUNCTION_PROLOGUE}
4845 and @code{TARGET_ASM_FUNCTION_EPILOGUE} are not invoked.
4847 The @var{thunk_fndecl} is redundant. (@var{delta} and @var{function}
4848 have already been extracted from it.) It might possibly be useful on
4849 some targets, but probably not.
4851 If you do not define this macro, the target-independent code in the C++
4852 front end will generate a less efficient heavyweight thunk that calls
4853 @var{function} instead of jumping to it. The generic approach does
4854 not support varargs.
4857 @hook TARGET_ASM_CAN_OUTPUT_MI_THUNK
4858 A function that returns true if TARGET_ASM_OUTPUT_MI_THUNK would be able
4859 to output the assembler code for the thunk function specified by the
4860 arguments it is passed, and false otherwise. In the latter case, the
4861 generic approach will be used by the C++ front end, with the limitations
4866 @subsection Generating Code for Profiling
4867 @cindex profiling, code generation
4869 These macros will help you generate code for profiling.
4871 @defmac FUNCTION_PROFILER (@var{file}, @var{labelno})
4872 A C statement or compound statement to output to @var{file} some
4873 assembler code to call the profiling subroutine @code{mcount}.
4876 The details of how @code{mcount} expects to be called are determined by
4877 your operating system environment, not by GCC@. To figure them out,
4878 compile a small program for profiling using the system's installed C
4879 compiler and look at the assembler code that results.
4881 Older implementations of @code{mcount} expect the address of a counter
4882 variable to be loaded into some register. The name of this variable is
4883 @samp{LP} followed by the number @var{labelno}, so you would generate
4884 the name using @samp{LP%d} in a @code{fprintf}.
4887 @defmac PROFILE_HOOK
4888 A C statement or compound statement to output to @var{file} some assembly
4889 code to call the profiling subroutine @code{mcount} even the target does
4890 not support profiling.
4893 @defmac NO_PROFILE_COUNTERS
4894 Define this macro to be an expression with a nonzero value if the
4895 @code{mcount} subroutine on your system does not need a counter variable
4896 allocated for each function. This is true for almost all modern
4897 implementations. If you define this macro, you must not use the
4898 @var{labelno} argument to @code{FUNCTION_PROFILER}.
4901 @defmac PROFILE_BEFORE_PROLOGUE
4902 Define this macro if the code for function profiling should come before
4903 the function prologue. Normally, the profiling code comes after.
4907 @subsection Permitting tail calls
4910 @hook TARGET_FUNCTION_OK_FOR_SIBCALL
4911 True if it is ok to do sibling call optimization for the specified
4912 call expression @var{exp}. @var{decl} will be the called function,
4913 or @code{NULL} if this is an indirect call.
4915 It is not uncommon for limitations of calling conventions to prevent
4916 tail calls to functions outside the current unit of translation, or
4917 during PIC compilation. The hook is used to enforce these restrictions,
4918 as the @code{sibcall} md pattern can not fail, or fall over to a
4919 ``normal'' call. The criteria for successful sibling call optimization
4920 may vary greatly between different architectures.
4923 @hook TARGET_EXTRA_LIVE_ON_ENTRY
4924 Add any hard registers to @var{regs} that are live on entry to the
4925 function. This hook only needs to be defined to provide registers that
4926 cannot be found by examination of FUNCTION_ARG_REGNO_P, the callee saved
4927 registers, STATIC_CHAIN_INCOMING_REGNUM, STATIC_CHAIN_REGNUM,
4928 TARGET_STRUCT_VALUE_RTX, FRAME_POINTER_REGNUM, EH_USES,
4929 FRAME_POINTER_REGNUM, ARG_POINTER_REGNUM, and the PIC_OFFSET_TABLE_REGNUM.
4932 @node Stack Smashing Protection
4933 @subsection Stack smashing protection
4934 @cindex stack smashing protection
4936 @hook TARGET_STACK_PROTECT_GUARD
4937 This hook returns a @code{DECL} node for the external variable to use
4938 for the stack protection guard. This variable is initialized by the
4939 runtime to some random value and is used to initialize the guard value
4940 that is placed at the top of the local stack frame. The type of this
4941 variable must be @code{ptr_type_node}.
4943 The default version of this hook creates a variable called
4944 @samp{__stack_chk_guard}, which is normally defined in @file{libgcc2.c}.
4947 @hook TARGET_STACK_PROTECT_FAIL
4948 This hook returns a tree expression that alerts the runtime that the
4949 stack protect guard variable has been modified. This expression should
4950 involve a call to a @code{noreturn} function.
4952 The default version of this hook invokes a function called
4953 @samp{__stack_chk_fail}, taking no arguments. This function is
4954 normally defined in @file{libgcc2.c}.
4957 @hook TARGET_SUPPORTS_SPLIT_STACK
4960 @section Implementing the Varargs Macros
4961 @cindex varargs implementation
4963 GCC comes with an implementation of @code{<varargs.h>} and
4964 @code{<stdarg.h>} that work without change on machines that pass arguments
4965 on the stack. Other machines require their own implementations of
4966 varargs, and the two machine independent header files must have
4967 conditionals to include it.
4969 ISO @code{<stdarg.h>} differs from traditional @code{<varargs.h>} mainly in
4970 the calling convention for @code{va_start}. The traditional
4971 implementation takes just one argument, which is the variable in which
4972 to store the argument pointer. The ISO implementation of
4973 @code{va_start} takes an additional second argument. The user is
4974 supposed to write the last named argument of the function here.
4976 However, @code{va_start} should not use this argument. The way to find
4977 the end of the named arguments is with the built-in functions described
4980 @defmac __builtin_saveregs ()
4981 Use this built-in function to save the argument registers in memory so
4982 that the varargs mechanism can access them. Both ISO and traditional
4983 versions of @code{va_start} must use @code{__builtin_saveregs}, unless
4984 you use @code{TARGET_SETUP_INCOMING_VARARGS} (see below) instead.
4986 On some machines, @code{__builtin_saveregs} is open-coded under the
4987 control of the target hook @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. On
4988 other machines, it calls a routine written in assembler language,
4989 found in @file{libgcc2.c}.
4991 Code generated for the call to @code{__builtin_saveregs} appears at the
4992 beginning of the function, as opposed to where the call to
4993 @code{__builtin_saveregs} is written, regardless of what the code is.
4994 This is because the registers must be saved before the function starts
4995 to use them for its own purposes.
4996 @c i rewrote the first sentence above to fix an overfull hbox. --mew
5000 @defmac __builtin_next_arg (@var{lastarg})
5001 This builtin returns the address of the first anonymous stack
5002 argument, as type @code{void *}. If @code{ARGS_GROW_DOWNWARD}, it
5003 returns the address of the location above the first anonymous stack
5004 argument. Use it in @code{va_start} to initialize the pointer for
5005 fetching arguments from the stack. Also use it in @code{va_start} to
5006 verify that the second parameter @var{lastarg} is the last named argument
5007 of the current function.
5010 @defmac __builtin_classify_type (@var{object})
5011 Since each machine has its own conventions for which data types are
5012 passed in which kind of register, your implementation of @code{va_arg}
5013 has to embody these conventions. The easiest way to categorize the
5014 specified data type is to use @code{__builtin_classify_type} together
5015 with @code{sizeof} and @code{__alignof__}.
5017 @code{__builtin_classify_type} ignores the value of @var{object},
5018 considering only its data type. It returns an integer describing what
5019 kind of type that is---integer, floating, pointer, structure, and so on.
5021 The file @file{typeclass.h} defines an enumeration that you can use to
5022 interpret the values of @code{__builtin_classify_type}.
5025 These machine description macros help implement varargs:
5027 @hook TARGET_EXPAND_BUILTIN_SAVEREGS
5028 If defined, this hook produces the machine-specific code for a call to
5029 @code{__builtin_saveregs}. This code will be moved to the very
5030 beginning of the function, before any parameter access are made. The
5031 return value of this function should be an RTX that contains the value
5032 to use as the return of @code{__builtin_saveregs}.
5035 @hook TARGET_SETUP_INCOMING_VARARGS
5036 This target hook offers an alternative to using
5037 @code{__builtin_saveregs} and defining the hook
5038 @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. Use it to store the anonymous
5039 register arguments into the stack so that all the arguments appear to
5040 have been passed consecutively on the stack. Once this is done, you can
5041 use the standard implementation of varargs that works for machines that
5042 pass all their arguments on the stack.
5044 The argument @var{args_so_far} points to the @code{CUMULATIVE_ARGS} data
5045 structure, containing the values that are obtained after processing the
5046 named arguments. The arguments @var{mode} and @var{type} describe the
5047 last named argument---its machine mode and its data type as a tree node.
5049 The target hook should do two things: first, push onto the stack all the
5050 argument registers @emph{not} used for the named arguments, and second,
5051 store the size of the data thus pushed into the @code{int}-valued
5052 variable pointed to by @var{pretend_args_size}. The value that you
5053 store here will serve as additional offset for setting up the stack
5056 Because you must generate code to push the anonymous arguments at
5057 compile time without knowing their data types,
5058 @code{TARGET_SETUP_INCOMING_VARARGS} is only useful on machines that
5059 have just a single category of argument register and use it uniformly
5062 If the argument @var{second_time} is nonzero, it means that the
5063 arguments of the function are being analyzed for the second time. This
5064 happens for an inline function, which is not actually compiled until the
5065 end of the source file. The hook @code{TARGET_SETUP_INCOMING_VARARGS} should
5066 not generate any instructions in this case.
5069 @hook TARGET_STRICT_ARGUMENT_NAMING
5070 Define this hook to return @code{true} if the location where a function
5071 argument is passed depends on whether or not it is a named argument.
5073 This hook controls how the @var{named} argument to @code{TARGET_FUNCTION_ARG}
5074 is set for varargs and stdarg functions. If this hook returns
5075 @code{true}, the @var{named} argument is always true for named
5076 arguments, and false for unnamed arguments. If it returns @code{false},
5077 but @code{TARGET_PRETEND_OUTGOING_VARARGS_NAMED} returns @code{true},
5078 then all arguments are treated as named. Otherwise, all named arguments
5079 except the last are treated as named.
5081 You need not define this hook if it always returns @code{false}.
5084 @hook TARGET_PRETEND_OUTGOING_VARARGS_NAMED
5085 If you need to conditionally change ABIs so that one works with
5086 @code{TARGET_SETUP_INCOMING_VARARGS}, but the other works like neither
5087 @code{TARGET_SETUP_INCOMING_VARARGS} nor @code{TARGET_STRICT_ARGUMENT_NAMING} was
5088 defined, then define this hook to return @code{true} if
5089 @code{TARGET_SETUP_INCOMING_VARARGS} is used, @code{false} otherwise.
5090 Otherwise, you should not define this hook.
5094 @section Trampolines for Nested Functions
5095 @cindex trampolines for nested functions
5096 @cindex nested functions, trampolines for
5098 A @dfn{trampoline} is a small piece of code that is created at run time
5099 when the address of a nested function is taken. It normally resides on
5100 the stack, in the stack frame of the containing function. These macros
5101 tell GCC how to generate code to allocate and initialize a
5104 The instructions in the trampoline must do two things: load a constant
5105 address into the static chain register, and jump to the real address of
5106 the nested function. On CISC machines such as the m68k, this requires
5107 two instructions, a move immediate and a jump. Then the two addresses
5108 exist in the trampoline as word-long immediate operands. On RISC
5109 machines, it is often necessary to load each address into a register in
5110 two parts. Then pieces of each address form separate immediate
5113 The code generated to initialize the trampoline must store the variable
5114 parts---the static chain value and the function address---into the
5115 immediate operands of the instructions. On a CISC machine, this is
5116 simply a matter of copying each address to a memory reference at the
5117 proper offset from the start of the trampoline. On a RISC machine, it
5118 may be necessary to take out pieces of the address and store them
5121 @hook TARGET_ASM_TRAMPOLINE_TEMPLATE
5122 This hook is called by @code{assemble_trampoline_template} to output,
5123 on the stream @var{f}, assembler code for a block of data that contains
5124 the constant parts of a trampoline. This code should not include a
5125 label---the label is taken care of automatically.
5127 If you do not define this hook, it means no template is needed
5128 for the target. Do not define this hook on systems where the block move
5129 code to copy the trampoline into place would be larger than the code
5130 to generate it on the spot.
5133 @defmac TRAMPOLINE_SECTION
5134 Return the section into which the trampoline template is to be placed
5135 (@pxref{Sections}). The default value is @code{readonly_data_section}.
5138 @defmac TRAMPOLINE_SIZE
5139 A C expression for the size in bytes of the trampoline, as an integer.
5142 @defmac TRAMPOLINE_ALIGNMENT
5143 Alignment required for trampolines, in bits.
5145 If you don't define this macro, the value of @code{FUNCTION_ALIGNMENT}
5146 is used for aligning trampolines.
5149 @hook TARGET_TRAMPOLINE_INIT
5150 This hook is called to initialize a trampoline.
5151 @var{m_tramp} is an RTX for the memory block for the trampoline; @var{fndecl}
5152 is the @code{FUNCTION_DECL} for the nested function; @var{static_chain} is an
5153 RTX for the static chain value that should be passed to the function
5156 If the target defines @code{TARGET_ASM_TRAMPOLINE_TEMPLATE}, then the
5157 first thing this hook should do is emit a block move into @var{m_tramp}
5158 from the memory block returned by @code{assemble_trampoline_template}.
5159 Note that the block move need only cover the constant parts of the
5160 trampoline. If the target isolates the variable parts of the trampoline
5161 to the end, not all @code{TRAMPOLINE_SIZE} bytes need be copied.
5163 If the target requires any other actions, such as flushing caches or
5164 enabling stack execution, these actions should be performed after
5165 initializing the trampoline proper.
5168 @hook TARGET_TRAMPOLINE_ADJUST_ADDRESS
5169 This hook should perform any machine-specific adjustment in
5170 the address of the trampoline. Its argument contains the address of the
5171 memory block that was passed to @code{TARGET_TRAMPOLINE_INIT}. In case
5172 the address to be used for a function call should be different from the
5173 address at which the template was stored, the different address should
5174 be returned; otherwise @var{addr} should be returned unchanged.
5175 If this hook is not defined, @var{addr} will be used for function calls.
5178 Implementing trampolines is difficult on many machines because they have
5179 separate instruction and data caches. Writing into a stack location
5180 fails to clear the memory in the instruction cache, so when the program
5181 jumps to that location, it executes the old contents.
5183 Here are two possible solutions. One is to clear the relevant parts of
5184 the instruction cache whenever a trampoline is set up. The other is to
5185 make all trampolines identical, by having them jump to a standard
5186 subroutine. The former technique makes trampoline execution faster; the
5187 latter makes initialization faster.
5189 To clear the instruction cache when a trampoline is initialized, define
5190 the following macro.
5192 @defmac CLEAR_INSN_CACHE (@var{beg}, @var{end})
5193 If defined, expands to a C expression clearing the @emph{instruction
5194 cache} in the specified interval. The definition of this macro would
5195 typically be a series of @code{asm} statements. Both @var{beg} and
5196 @var{end} are both pointer expressions.
5199 To use a standard subroutine, define the following macro. In addition,
5200 you must make sure that the instructions in a trampoline fill an entire
5201 cache line with identical instructions, or else ensure that the
5202 beginning of the trampoline code is always aligned at the same point in
5203 its cache line. Look in @file{m68k.h} as a guide.
5205 @defmac TRANSFER_FROM_TRAMPOLINE
5206 Define this macro if trampolines need a special subroutine to do their
5207 work. The macro should expand to a series of @code{asm} statements
5208 which will be compiled with GCC@. They go in a library function named
5209 @code{__transfer_from_trampoline}.
5211 If you need to avoid executing the ordinary prologue code of a compiled
5212 C function when you jump to the subroutine, you can do so by placing a
5213 special label of your own in the assembler code. Use one @code{asm}
5214 statement to generate an assembler label, and another to make the label
5215 global. Then trampolines can use that label to jump directly to your
5216 special assembler code.
5220 @section Implicit Calls to Library Routines
5221 @cindex library subroutine names
5222 @cindex @file{libgcc.a}
5224 @c prevent bad page break with this line
5225 Here is an explanation of implicit calls to library routines.
5227 @defmac DECLARE_LIBRARY_RENAMES
5228 This macro, if defined, should expand to a piece of C code that will get
5229 expanded when compiling functions for libgcc.a. It can be used to
5230 provide alternate names for GCC's internal library functions if there
5231 are ABI-mandated names that the compiler should provide.
5234 @findex set_optab_libfunc
5235 @findex init_one_libfunc
5236 @hook TARGET_INIT_LIBFUNCS
5237 This hook should declare additional library routines or rename
5238 existing ones, using the functions @code{set_optab_libfunc} and
5239 @code{init_one_libfunc} defined in @file{optabs.c}.
5240 @code{init_optabs} calls this macro after initializing all the normal
5243 The default is to do nothing. Most ports don't need to define this hook.
5246 @hook TARGET_LIBFUNC_GNU_PREFIX
5248 @defmac FLOAT_LIB_COMPARE_RETURNS_BOOL (@var{mode}, @var{comparison})
5249 This macro should return @code{true} if the library routine that
5250 implements the floating point comparison operator @var{comparison} in
5251 mode @var{mode} will return a boolean, and @var{false} if it will
5254 GCC's own floating point libraries return tristates from the
5255 comparison operators, so the default returns false always. Most ports
5256 don't need to define this macro.
5259 @defmac TARGET_LIB_INT_CMP_BIASED
5260 This macro should evaluate to @code{true} if the integer comparison
5261 functions (like @code{__cmpdi2}) return 0 to indicate that the first
5262 operand is smaller than the second, 1 to indicate that they are equal,
5263 and 2 to indicate that the first operand is greater than the second.
5264 If this macro evaluates to @code{false} the comparison functions return
5265 @minus{}1, 0, and 1 instead of 0, 1, and 2. If the target uses the routines
5266 in @file{libgcc.a}, you do not need to define this macro.
5269 @cindex @code{EDOM}, implicit usage
5272 The value of @code{EDOM} on the target machine, as a C integer constant
5273 expression. If you don't define this macro, GCC does not attempt to
5274 deposit the value of @code{EDOM} into @code{errno} directly. Look in
5275 @file{/usr/include/errno.h} to find the value of @code{EDOM} on your
5278 If you do not define @code{TARGET_EDOM}, then compiled code reports
5279 domain errors by calling the library function and letting it report the
5280 error. If mathematical functions on your system use @code{matherr} when
5281 there is an error, then you should leave @code{TARGET_EDOM} undefined so
5282 that @code{matherr} is used normally.
5285 @cindex @code{errno}, implicit usage
5286 @defmac GEN_ERRNO_RTX
5287 Define this macro as a C expression to create an rtl expression that
5288 refers to the global ``variable'' @code{errno}. (On certain systems,
5289 @code{errno} may not actually be a variable.) If you don't define this
5290 macro, a reasonable default is used.
5293 @cindex C99 math functions, implicit usage
5294 @defmac TARGET_C99_FUNCTIONS
5295 When this macro is nonzero, GCC will implicitly optimize @code{sin} calls into
5296 @code{sinf} and similarly for other functions defined by C99 standard. The
5297 default is zero because a number of existing systems lack support for these
5298 functions in their runtime so this macro needs to be redefined to one on
5299 systems that do support the C99 runtime.
5302 @cindex sincos math function, implicit usage
5303 @defmac TARGET_HAS_SINCOS
5304 When this macro is nonzero, GCC will implicitly optimize calls to @code{sin}
5305 and @code{cos} with the same argument to a call to @code{sincos}. The
5306 default is zero. The target has to provide the following functions:
5308 void sincos(double x, double *sin, double *cos);
5309 void sincosf(float x, float *sin, float *cos);
5310 void sincosl(long double x, long double *sin, long double *cos);
5314 @defmac NEXT_OBJC_RUNTIME
5315 Define this macro to generate code for Objective-C message sending using
5316 the calling convention of the NeXT system. This calling convention
5317 involves passing the object, the selector and the method arguments all
5318 at once to the method-lookup library function.
5320 The default calling convention passes just the object and the selector
5321 to the lookup function, which returns a pointer to the method.
5324 @node Addressing Modes
5325 @section Addressing Modes
5326 @cindex addressing modes
5328 @c prevent bad page break with this line
5329 This is about addressing modes.
5331 @defmac HAVE_PRE_INCREMENT
5332 @defmacx HAVE_PRE_DECREMENT
5333 @defmacx HAVE_POST_INCREMENT
5334 @defmacx HAVE_POST_DECREMENT
5335 A C expression that is nonzero if the machine supports pre-increment,
5336 pre-decrement, post-increment, or post-decrement addressing respectively.
5339 @defmac HAVE_PRE_MODIFY_DISP
5340 @defmacx HAVE_POST_MODIFY_DISP
5341 A C expression that is nonzero if the machine supports pre- or
5342 post-address side-effect generation involving constants other than
5343 the size of the memory operand.
5346 @defmac HAVE_PRE_MODIFY_REG
5347 @defmacx HAVE_POST_MODIFY_REG
5348 A C expression that is nonzero if the machine supports pre- or
5349 post-address side-effect generation involving a register displacement.
5352 @defmac CONSTANT_ADDRESS_P (@var{x})
5353 A C expression that is 1 if the RTX @var{x} is a constant which
5354 is a valid address. On most machines the default definition of
5355 @code{(CONSTANT_P (@var{x}) && GET_CODE (@var{x}) != CONST_DOUBLE)}
5356 is acceptable, but a few machines are more restrictive as to which
5357 constant addresses are supported.
5360 @defmac CONSTANT_P (@var{x})
5361 @code{CONSTANT_P}, which is defined by target-independent code,
5362 accepts integer-values expressions whose values are not explicitly
5363 known, such as @code{symbol_ref}, @code{label_ref}, and @code{high}
5364 expressions and @code{const} arithmetic expressions, in addition to
5365 @code{const_int} and @code{const_double} expressions.
5368 @defmac MAX_REGS_PER_ADDRESS
5369 A number, the maximum number of registers that can appear in a valid
5370 memory address. Note that it is up to you to specify a value equal to
5371 the maximum number that @code{TARGET_LEGITIMATE_ADDRESS_P} would ever
5375 @hook TARGET_LEGITIMATE_ADDRESS_P
5376 A function that returns whether @var{x} (an RTX) is a legitimate memory
5377 address on the target machine for a memory operand of mode @var{mode}.
5379 Legitimate addresses are defined in two variants: a strict variant and a
5380 non-strict one. The @var{strict} parameter chooses which variant is
5381 desired by the caller.
5383 The strict variant is used in the reload pass. It must be defined so
5384 that any pseudo-register that has not been allocated a hard register is
5385 considered a memory reference. This is because in contexts where some
5386 kind of register is required, a pseudo-register with no hard register
5387 must be rejected. For non-hard registers, the strict variant should look
5388 up the @code{reg_renumber} array; it should then proceed using the hard
5389 register number in the array, or treat the pseudo as a memory reference
5390 if the array holds @code{-1}.
5392 The non-strict variant is used in other passes. It must be defined to
5393 accept all pseudo-registers in every context where some kind of
5394 register is required.
5396 Normally, constant addresses which are the sum of a @code{symbol_ref}
5397 and an integer are stored inside a @code{const} RTX to mark them as
5398 constant. Therefore, there is no need to recognize such sums
5399 specifically as legitimate addresses. Normally you would simply
5400 recognize any @code{const} as legitimate.
5402 Usually @code{PRINT_OPERAND_ADDRESS} is not prepared to handle constant
5403 sums that are not marked with @code{const}. It assumes that a naked
5404 @code{plus} indicates indexing. If so, then you @emph{must} reject such
5405 naked constant sums as illegitimate addresses, so that none of them will
5406 be given to @code{PRINT_OPERAND_ADDRESS}.
5408 @cindex @code{TARGET_ENCODE_SECTION_INFO} and address validation
5409 On some machines, whether a symbolic address is legitimate depends on
5410 the section that the address refers to. On these machines, define the
5411 target hook @code{TARGET_ENCODE_SECTION_INFO} to store the information
5412 into the @code{symbol_ref}, and then check for it here. When you see a
5413 @code{const}, you will have to look inside it to find the
5414 @code{symbol_ref} in order to determine the section. @xref{Assembler
5417 @cindex @code{GO_IF_LEGITIMATE_ADDRESS}
5418 Some ports are still using a deprecated legacy substitute for
5419 this hook, the @code{GO_IF_LEGITIMATE_ADDRESS} macro. This macro
5423 #define GO_IF_LEGITIMATE_ADDRESS (@var{mode}, @var{x}, @var{label})
5427 and should @code{goto @var{label}} if the address @var{x} is a valid
5428 address on the target machine for a memory operand of mode @var{mode}.
5430 @findex REG_OK_STRICT
5431 Compiler source files that want to use the strict variant of this
5432 macro define the macro @code{REG_OK_STRICT}. You should use an
5433 @code{#ifdef REG_OK_STRICT} conditional to define the strict variant in
5434 that case and the non-strict variant otherwise.
5436 Using the hook is usually simpler because it limits the number of
5437 files that are recompiled when changes are made.
5440 @defmac TARGET_MEM_CONSTRAINT
5441 A single character to be used instead of the default @code{'m'}
5442 character for general memory addresses. This defines the constraint
5443 letter which matches the memory addresses accepted by
5444 @code{TARGET_LEGITIMATE_ADDRESS_P}. Define this macro if you want to
5445 support new address formats in your back end without changing the
5446 semantics of the @code{'m'} constraint. This is necessary in order to
5447 preserve functionality of inline assembly constructs using the
5448 @code{'m'} constraint.
5451 @defmac FIND_BASE_TERM (@var{x})
5452 A C expression to determine the base term of address @var{x},
5453 or to provide a simplified version of @var{x} from which @file{alias.c}
5454 can easily find the base term. This macro is used in only two places:
5455 @code{find_base_value} and @code{find_base_term} in @file{alias.c}.
5457 It is always safe for this macro to not be defined. It exists so
5458 that alias analysis can understand machine-dependent addresses.
5460 The typical use of this macro is to handle addresses containing
5461 a label_ref or symbol_ref within an UNSPEC@.
5464 @hook TARGET_LEGITIMIZE_ADDRESS
5465 This hook is given an invalid memory address @var{x} for an
5466 operand of mode @var{mode} and should try to return a valid memory
5469 @findex break_out_memory_refs
5470 @var{x} will always be the result of a call to @code{break_out_memory_refs},
5471 and @var{oldx} will be the operand that was given to that function to produce
5474 The code of the hook should not alter the substructure of
5475 @var{x}. If it transforms @var{x} into a more legitimate form, it
5476 should return the new @var{x}.
5478 It is not necessary for this hook to come up with a legitimate address.
5479 The compiler has standard ways of doing so in all cases. In fact, it
5480 is safe to omit this hook or make it return @var{x} if it cannot find
5481 a valid way to legitimize the address. But often a machine-dependent
5482 strategy can generate better code.
5485 @defmac LEGITIMIZE_RELOAD_ADDRESS (@var{x}, @var{mode}, @var{opnum}, @var{type}, @var{ind_levels}, @var{win})
5486 A C compound statement that attempts to replace @var{x}, which is an address
5487 that needs reloading, with a valid memory address for an operand of mode
5488 @var{mode}. @var{win} will be a C statement label elsewhere in the code.
5489 It is not necessary to define this macro, but it might be useful for
5490 performance reasons.
5492 For example, on the i386, it is sometimes possible to use a single
5493 reload register instead of two by reloading a sum of two pseudo
5494 registers into a register. On the other hand, for number of RISC
5495 processors offsets are limited so that often an intermediate address
5496 needs to be generated in order to address a stack slot. By defining
5497 @code{LEGITIMIZE_RELOAD_ADDRESS} appropriately, the intermediate addresses
5498 generated for adjacent some stack slots can be made identical, and thus
5501 @emph{Note}: This macro should be used with caution. It is necessary
5502 to know something of how reload works in order to effectively use this,
5503 and it is quite easy to produce macros that build in too much knowledge
5504 of reload internals.
5506 @emph{Note}: This macro must be able to reload an address created by a
5507 previous invocation of this macro. If it fails to handle such addresses
5508 then the compiler may generate incorrect code or abort.
5511 The macro definition should use @code{push_reload} to indicate parts that
5512 need reloading; @var{opnum}, @var{type} and @var{ind_levels} are usually
5513 suitable to be passed unaltered to @code{push_reload}.
5515 The code generated by this macro must not alter the substructure of
5516 @var{x}. If it transforms @var{x} into a more legitimate form, it
5517 should assign @var{x} (which will always be a C variable) a new value.
5518 This also applies to parts that you change indirectly by calling
5521 @findex strict_memory_address_p
5522 The macro definition may use @code{strict_memory_address_p} to test if
5523 the address has become legitimate.
5526 If you want to change only a part of @var{x}, one standard way of doing
5527 this is to use @code{copy_rtx}. Note, however, that it unshares only a
5528 single level of rtl. Thus, if the part to be changed is not at the
5529 top level, you'll need to replace first the top level.
5530 It is not necessary for this macro to come up with a legitimate
5531 address; but often a machine-dependent strategy can generate better code.
5534 @hook TARGET_MODE_DEPENDENT_ADDRESS_P
5535 This hook returns @code{true} if memory address @var{addr} can have
5536 different meanings depending on the machine mode of the memory
5537 reference it is used for or if the address is valid for some modes
5540 Autoincrement and autodecrement addresses typically have mode-dependent
5541 effects because the amount of the increment or decrement is the size
5542 of the operand being addressed. Some machines have other mode-dependent
5543 addresses. Many RISC machines have no mode-dependent addresses.
5545 You may assume that @var{addr} is a valid address for the machine.
5547 The default version of this hook returns @code{false}.
5550 @defmac GO_IF_MODE_DEPENDENT_ADDRESS (@var{addr}, @var{label})
5551 A C statement or compound statement with a conditional @code{goto
5552 @var{label};} executed if memory address @var{x} (an RTX) can have
5553 different meanings depending on the machine mode of the memory
5554 reference it is used for or if the address is valid for some modes
5557 Autoincrement and autodecrement addresses typically have mode-dependent
5558 effects because the amount of the increment or decrement is the size
5559 of the operand being addressed. Some machines have other mode-dependent
5560 addresses. Many RISC machines have no mode-dependent addresses.
5562 You may assume that @var{addr} is a valid address for the machine.
5564 These are obsolete macros, replaced by the
5565 @code{TARGET_MODE_DEPENDENT_ADDRESS_P} target hook.
5568 @hook TARGET_LEGITIMATE_CONSTANT_P
5569 This hook returns true if @var{x} is a legitimate constant for a
5570 @var{mode}-mode immediate operand on the target machine. You can assume that
5571 @var{x} satisfies @code{CONSTANT_P}, so you need not check this.
5573 The default definition returns true.
5576 @hook TARGET_DELEGITIMIZE_ADDRESS
5577 This hook is used to undo the possibly obfuscating effects of the
5578 @code{LEGITIMIZE_ADDRESS} and @code{LEGITIMIZE_RELOAD_ADDRESS} target
5579 macros. Some backend implementations of these macros wrap symbol
5580 references inside an @code{UNSPEC} rtx to represent PIC or similar
5581 addressing modes. This target hook allows GCC's optimizers to understand
5582 the semantics of these opaque @code{UNSPEC}s by converting them back
5583 into their original form.
5586 @hook TARGET_CANNOT_FORCE_CONST_MEM
5587 This hook should return true if @var{x} is of a form that cannot (or
5588 should not) be spilled to the constant pool. @var{mode} is the mode
5591 The default version of this hook returns false.
5593 The primary reason to define this hook is to prevent reload from
5594 deciding that a non-legitimate constant would be better reloaded
5595 from the constant pool instead of spilling and reloading a register
5596 holding the constant. This restriction is often true of addresses
5597 of TLS symbols for various targets.
5600 @hook TARGET_USE_BLOCKS_FOR_CONSTANT_P
5601 This hook should return true if pool entries for constant @var{x} can
5602 be placed in an @code{object_block} structure. @var{mode} is the mode
5605 The default version returns false for all constants.
5608 @hook TARGET_BUILTIN_RECIPROCAL
5609 This hook should return the DECL of a function that implements reciprocal of
5610 the builtin function with builtin function code @var{fn}, or
5611 @code{NULL_TREE} if such a function is not available. @var{md_fn} is true
5612 when @var{fn} is a code of a machine-dependent builtin function. When
5613 @var{sqrt} is true, additional optimizations that apply only to the reciprocal
5614 of a square root function are performed, and only reciprocals of @code{sqrt}
5618 @hook TARGET_VECTORIZE_BUILTIN_MASK_FOR_LOAD
5619 This hook should return the DECL of a function @var{f} that given an
5620 address @var{addr} as an argument returns a mask @var{m} that can be
5621 used to extract from two vectors the relevant data that resides in
5622 @var{addr} in case @var{addr} is not properly aligned.
5624 The autovectorizer, when vectorizing a load operation from an address
5625 @var{addr} that may be unaligned, will generate two vector loads from
5626 the two aligned addresses around @var{addr}. It then generates a
5627 @code{REALIGN_LOAD} operation to extract the relevant data from the
5628 two loaded vectors. The first two arguments to @code{REALIGN_LOAD},
5629 @var{v1} and @var{v2}, are the two vectors, each of size @var{VS}, and
5630 the third argument, @var{OFF}, defines how the data will be extracted
5631 from these two vectors: if @var{OFF} is 0, then the returned vector is
5632 @var{v2}; otherwise, the returned vector is composed from the last
5633 @var{VS}-@var{OFF} elements of @var{v1} concatenated to the first
5634 @var{OFF} elements of @var{v2}.
5636 If this hook is defined, the autovectorizer will generate a call
5637 to @var{f} (using the DECL tree that this hook returns) and will
5638 use the return value of @var{f} as the argument @var{OFF} to
5639 @code{REALIGN_LOAD}. Therefore, the mask @var{m} returned by @var{f}
5640 should comply with the semantics expected by @code{REALIGN_LOAD}
5642 If this hook is not defined, then @var{addr} will be used as
5643 the argument @var{OFF} to @code{REALIGN_LOAD}, in which case the low
5644 log2(@var{VS}) @minus{} 1 bits of @var{addr} will be considered.
5647 @hook TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_EVEN
5648 This hook should return the DECL of a function @var{f} that implements
5649 widening multiplication of the even elements of two input vectors of type @var{x}.
5651 If this hook is defined, the autovectorizer will use it along with the
5652 @code{TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_ODD} target hook when vectorizing
5653 widening multiplication in cases that the order of the results does not have to be
5654 preserved (e.g.@: used only by a reduction computation). Otherwise, the
5655 @code{widen_mult_hi/lo} idioms will be used.
5658 @hook TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_ODD
5659 This hook should return the DECL of a function @var{f} that implements
5660 widening multiplication of the odd elements of two input vectors of type @var{x}.
5662 If this hook is defined, the autovectorizer will use it along with the
5663 @code{TARGET_VECTORIZE_BUILTIN_MUL_WIDEN_EVEN} target hook when vectorizing
5664 widening multiplication in cases that the order of the results does not have to be
5665 preserved (e.g.@: used only by a reduction computation). Otherwise, the
5666 @code{widen_mult_hi/lo} idioms will be used.
5669 @hook TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST
5670 Returns cost of different scalar or vector statements for vectorization cost model.
5671 For vector memory operations the cost may depend on type (@var{vectype}) and
5672 misalignment value (@var{misalign}).
5675 @hook TARGET_VECTORIZE_VECTOR_ALIGNMENT_REACHABLE
5676 Return true if vector alignment is reachable (by peeling N iterations) for the given type.
5679 @hook TARGET_VECTORIZE_BUILTIN_VEC_PERM
5680 Target builtin that implements vector permute.
5683 @hook TARGET_VECTORIZE_BUILTIN_VEC_PERM_OK
5684 Return true if a vector created for @code{builtin_vec_perm} is valid.
5687 @hook TARGET_VECTORIZE_BUILTIN_CONVERSION
5688 This hook should return the DECL of a function that implements conversion of the
5689 input vector of type @var{src_type} to type @var{dest_type}.
5690 The value of @var{code} is one of the enumerators in @code{enum tree_code} and
5691 specifies how the conversion is to be applied
5692 (truncation, rounding, etc.).
5694 If this hook is defined, the autovectorizer will use the
5695 @code{TARGET_VECTORIZE_BUILTIN_CONVERSION} target hook when vectorizing
5696 conversion. Otherwise, it will return @code{NULL_TREE}.
5699 @hook TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION
5700 This hook should return the decl of a function that implements the
5701 vectorized variant of the builtin function with builtin function code
5702 @var{code} or @code{NULL_TREE} if such a function is not available.
5703 The value of @var{fndecl} is the builtin function declaration. The
5704 return type of the vectorized function shall be of vector type
5705 @var{vec_type_out} and the argument types should be @var{vec_type_in}.
5708 @hook TARGET_VECTORIZE_SUPPORT_VECTOR_MISALIGNMENT
5709 This hook should return true if the target supports misaligned vector
5710 store/load of a specific factor denoted in the @var{misalignment}
5711 parameter. The vector store/load should be of machine mode @var{mode} and
5712 the elements in the vectors should be of type @var{type}. @var{is_packed}
5713 parameter is true if the memory access is defined in a packed struct.
5716 @hook TARGET_VECTORIZE_PREFERRED_SIMD_MODE
5717 This hook should return the preferred mode for vectorizing scalar
5718 mode @var{mode}. The default is
5719 equal to @code{word_mode}, because the vectorizer can do some
5720 transformations even in absence of specialized @acronym{SIMD} hardware.
5723 @hook TARGET_VECTORIZE_AUTOVECTORIZE_VECTOR_SIZES
5724 This hook should return a mask of sizes that should be iterated over
5725 after trying to autovectorize using the vector size derived from the
5726 mode returned by @code{TARGET_VECTORIZE_PREFERRED_SIMD_MODE}.
5727 The default is zero which means to not iterate over other vector sizes.
5730 @node Anchored Addresses
5731 @section Anchored Addresses
5732 @cindex anchored addresses
5733 @cindex @option{-fsection-anchors}
5735 GCC usually addresses every static object as a separate entity.
5736 For example, if we have:
5740 int foo (void) @{ return a + b + c; @}
5743 the code for @code{foo} will usually calculate three separate symbolic
5744 addresses: those of @code{a}, @code{b} and @code{c}. On some targets,
5745 it would be better to calculate just one symbolic address and access
5746 the three variables relative to it. The equivalent pseudocode would
5752 register int *xr = &x;
5753 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
5757 (which isn't valid C). We refer to shared addresses like @code{x} as
5758 ``section anchors''. Their use is controlled by @option{-fsection-anchors}.
5760 The hooks below describe the target properties that GCC needs to know
5761 in order to make effective use of section anchors. It won't use
5762 section anchors at all unless either @code{TARGET_MIN_ANCHOR_OFFSET}
5763 or @code{TARGET_MAX_ANCHOR_OFFSET} is set to a nonzero value.
5765 @hook TARGET_MIN_ANCHOR_OFFSET
5766 The minimum offset that should be applied to a section anchor.
5767 On most targets, it should be the smallest offset that can be
5768 applied to a base register while still giving a legitimate address
5769 for every mode. The default value is 0.
5772 @hook TARGET_MAX_ANCHOR_OFFSET
5773 Like @code{TARGET_MIN_ANCHOR_OFFSET}, but the maximum (inclusive)
5774 offset that should be applied to section anchors. The default
5778 @hook TARGET_ASM_OUTPUT_ANCHOR
5779 Write the assembly code to define section anchor @var{x}, which is a
5780 @code{SYMBOL_REF} for which @samp{SYMBOL_REF_ANCHOR_P (@var{x})} is true.
5781 The hook is called with the assembly output position set to the beginning
5782 of @code{SYMBOL_REF_BLOCK (@var{x})}.
5784 If @code{ASM_OUTPUT_DEF} is available, the hook's default definition uses
5785 it to define the symbol as @samp{. + SYMBOL_REF_BLOCK_OFFSET (@var{x})}.
5786 If @code{ASM_OUTPUT_DEF} is not available, the hook's default definition
5787 is @code{NULL}, which disables the use of section anchors altogether.
5790 @hook TARGET_USE_ANCHORS_FOR_SYMBOL_P
5791 Return true if GCC should attempt to use anchors to access @code{SYMBOL_REF}
5792 @var{x}. You can assume @samp{SYMBOL_REF_HAS_BLOCK_INFO_P (@var{x})} and
5793 @samp{!SYMBOL_REF_ANCHOR_P (@var{x})}.
5795 The default version is correct for most targets, but you might need to
5796 intercept this hook to handle things like target-specific attributes
5797 or target-specific sections.
5800 @node Condition Code
5801 @section Condition Code Status
5802 @cindex condition code status
5804 The macros in this section can be split in two families, according to the
5805 two ways of representing condition codes in GCC.
5807 The first representation is the so called @code{(cc0)} representation
5808 (@pxref{Jump Patterns}), where all instructions can have an implicit
5809 clobber of the condition codes. The second is the condition code
5810 register representation, which provides better schedulability for
5811 architectures that do have a condition code register, but on which
5812 most instructions do not affect it. The latter category includes
5815 The implicit clobbering poses a strong restriction on the placement of
5816 the definition and use of the condition code, which need to be in adjacent
5817 insns for machines using @code{(cc0)}. This can prevent important
5818 optimizations on some machines. For example, on the IBM RS/6000, there
5819 is a delay for taken branches unless the condition code register is set
5820 three instructions earlier than the conditional branch. The instruction
5821 scheduler cannot perform this optimization if it is not permitted to
5822 separate the definition and use of the condition code register.
5824 For this reason, it is possible and suggested to use a register to
5825 represent the condition code for new ports. If there is a specific
5826 condition code register in the machine, use a hard register. If the
5827 condition code or comparison result can be placed in any general register,
5828 or if there are multiple condition registers, use a pseudo register.
5829 Registers used to store the condition code value will usually have a mode
5830 that is in class @code{MODE_CC}.
5832 Alternatively, you can use @code{BImode} if the comparison operator is
5833 specified already in the compare instruction. In this case, you are not
5834 interested in most macros in this section.
5837 * CC0 Condition Codes:: Old style representation of condition codes.
5838 * MODE_CC Condition Codes:: Modern representation of condition codes.
5839 * Cond Exec Macros:: Macros to control conditional execution.
5842 @node CC0 Condition Codes
5843 @subsection Representation of condition codes using @code{(cc0)}
5847 The file @file{conditions.h} defines a variable @code{cc_status} to
5848 describe how the condition code was computed (in case the interpretation of
5849 the condition code depends on the instruction that it was set by). This
5850 variable contains the RTL expressions on which the condition code is
5851 currently based, and several standard flags.
5853 Sometimes additional machine-specific flags must be defined in the machine
5854 description header file. It can also add additional machine-specific
5855 information by defining @code{CC_STATUS_MDEP}.
5857 @defmac CC_STATUS_MDEP
5858 C code for a data type which is used for declaring the @code{mdep}
5859 component of @code{cc_status}. It defaults to @code{int}.
5861 This macro is not used on machines that do not use @code{cc0}.
5864 @defmac CC_STATUS_MDEP_INIT
5865 A C expression to initialize the @code{mdep} field to ``empty''.
5866 The default definition does nothing, since most machines don't use
5867 the field anyway. If you want to use the field, you should probably
5868 define this macro to initialize it.
5870 This macro is not used on machines that do not use @code{cc0}.
5873 @defmac NOTICE_UPDATE_CC (@var{exp}, @var{insn})
5874 A C compound statement to set the components of @code{cc_status}
5875 appropriately for an insn @var{insn} whose body is @var{exp}. It is
5876 this macro's responsibility to recognize insns that set the condition
5877 code as a byproduct of other activity as well as those that explicitly
5880 This macro is not used on machines that do not use @code{cc0}.
5882 If there are insns that do not set the condition code but do alter
5883 other machine registers, this macro must check to see whether they
5884 invalidate the expressions that the condition code is recorded as
5885 reflecting. For example, on the 68000, insns that store in address
5886 registers do not set the condition code, which means that usually
5887 @code{NOTICE_UPDATE_CC} can leave @code{cc_status} unaltered for such
5888 insns. But suppose that the previous insn set the condition code
5889 based on location @samp{a4@@(102)} and the current insn stores a new
5890 value in @samp{a4}. Although the condition code is not changed by
5891 this, it will no longer be true that it reflects the contents of
5892 @samp{a4@@(102)}. Therefore, @code{NOTICE_UPDATE_CC} must alter
5893 @code{cc_status} in this case to say that nothing is known about the
5894 condition code value.
5896 The definition of @code{NOTICE_UPDATE_CC} must be prepared to deal
5897 with the results of peephole optimization: insns whose patterns are
5898 @code{parallel} RTXs containing various @code{reg}, @code{mem} or
5899 constants which are just the operands. The RTL structure of these
5900 insns is not sufficient to indicate what the insns actually do. What
5901 @code{NOTICE_UPDATE_CC} should do when it sees one is just to run
5902 @code{CC_STATUS_INIT}.
5904 A possible definition of @code{NOTICE_UPDATE_CC} is to call a function
5905 that looks at an attribute (@pxref{Insn Attributes}) named, for example,
5906 @samp{cc}. This avoids having detailed information about patterns in
5907 two places, the @file{md} file and in @code{NOTICE_UPDATE_CC}.
5910 @node MODE_CC Condition Codes
5911 @subsection Representation of condition codes using registers
5915 @defmac SELECT_CC_MODE (@var{op}, @var{x}, @var{y})
5916 On many machines, the condition code may be produced by other instructions
5917 than compares, for example the branch can use directly the condition
5918 code set by a subtract instruction. However, on some machines
5919 when the condition code is set this way some bits (such as the overflow
5920 bit) are not set in the same way as a test instruction, so that a different
5921 branch instruction must be used for some conditional branches. When
5922 this happens, use the machine mode of the condition code register to
5923 record different formats of the condition code register. Modes can
5924 also be used to record which compare instruction (e.g. a signed or an
5925 unsigned comparison) produced the condition codes.
5927 If other modes than @code{CCmode} are required, add them to
5928 @file{@var{machine}-modes.def} and define @code{SELECT_CC_MODE} to choose
5929 a mode given an operand of a compare. This is needed because the modes
5930 have to be chosen not only during RTL generation but also, for example,
5931 by instruction combination. The result of @code{SELECT_CC_MODE} should
5932 be consistent with the mode used in the patterns; for example to support
5933 the case of the add on the SPARC discussed above, we have the pattern
5937 [(set (reg:CC_NOOV 0)
5939 (plus:SI (match_operand:SI 0 "register_operand" "%r")
5940 (match_operand:SI 1 "arith_operand" "rI"))
5947 together with a @code{SELECT_CC_MODE} that returns @code{CC_NOOVmode}
5948 for comparisons whose argument is a @code{plus}:
5951 #define SELECT_CC_MODE(OP,X,Y) \
5952 (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \
5953 ? ((OP == EQ || OP == NE) ? CCFPmode : CCFPEmode) \
5954 : ((GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \
5955 || GET_CODE (X) == NEG) \
5956 ? CC_NOOVmode : CCmode))
5959 Another reason to use modes is to retain information on which operands
5960 were used by the comparison; see @code{REVERSIBLE_CC_MODE} later in
5963 You should define this macro if and only if you define extra CC modes
5964 in @file{@var{machine}-modes.def}.
5967 @defmac CANONICALIZE_COMPARISON (@var{code}, @var{op0}, @var{op1})
5968 On some machines not all possible comparisons are defined, but you can
5969 convert an invalid comparison into a valid one. For example, the Alpha
5970 does not have a @code{GT} comparison, but you can use an @code{LT}
5971 comparison instead and swap the order of the operands.
5973 On such machines, define this macro to be a C statement to do any
5974 required conversions. @var{code} is the initial comparison code
5975 and @var{op0} and @var{op1} are the left and right operands of the
5976 comparison, respectively. You should modify @var{code}, @var{op0}, and
5977 @var{op1} as required.
5979 GCC will not assume that the comparison resulting from this macro is
5980 valid but will see if the resulting insn matches a pattern in the
5983 You need not define this macro if it would never change the comparison
5987 @defmac REVERSIBLE_CC_MODE (@var{mode})
5988 A C expression whose value is one if it is always safe to reverse a
5989 comparison whose mode is @var{mode}. If @code{SELECT_CC_MODE}
5990 can ever return @var{mode} for a floating-point inequality comparison,
5991 then @code{REVERSIBLE_CC_MODE (@var{mode})} must be zero.
5993 You need not define this macro if it would always returns zero or if the
5994 floating-point format is anything other than @code{IEEE_FLOAT_FORMAT}.
5995 For example, here is the definition used on the SPARC, where floating-point
5996 inequality comparisons are always given @code{CCFPEmode}:
5999 #define REVERSIBLE_CC_MODE(MODE) ((MODE) != CCFPEmode)
6003 @defmac REVERSE_CONDITION (@var{code}, @var{mode})
6004 A C expression whose value is reversed condition code of the @var{code} for
6005 comparison done in CC_MODE @var{mode}. The macro is used only in case
6006 @code{REVERSIBLE_CC_MODE (@var{mode})} is nonzero. Define this macro in case
6007 machine has some non-standard way how to reverse certain conditionals. For
6008 instance in case all floating point conditions are non-trapping, compiler may
6009 freely convert unordered compares to ordered one. Then definition may look
6013 #define REVERSE_CONDITION(CODE, MODE) \
6014 ((MODE) != CCFPmode ? reverse_condition (CODE) \
6015 : reverse_condition_maybe_unordered (CODE))
6019 @hook TARGET_FIXED_CONDITION_CODE_REGS
6020 On targets which do not use @code{(cc0)}, and which use a hard
6021 register rather than a pseudo-register to hold condition codes, the
6022 regular CSE passes are often not able to identify cases in which the
6023 hard register is set to a common value. Use this hook to enable a
6024 small pass which optimizes such cases. This hook should return true
6025 to enable this pass, and it should set the integers to which its
6026 arguments point to the hard register numbers used for condition codes.
6027 When there is only one such register, as is true on most systems, the
6028 integer pointed to by @var{p2} should be set to
6029 @code{INVALID_REGNUM}.
6031 The default version of this hook returns false.
6034 @hook TARGET_CC_MODES_COMPATIBLE
6035 On targets which use multiple condition code modes in class
6036 @code{MODE_CC}, it is sometimes the case that a comparison can be
6037 validly done in more than one mode. On such a system, define this
6038 target hook to take two mode arguments and to return a mode in which
6039 both comparisons may be validly done. If there is no such mode,
6040 return @code{VOIDmode}.
6042 The default version of this hook checks whether the modes are the
6043 same. If they are, it returns that mode. If they are different, it
6044 returns @code{VOIDmode}.
6047 @node Cond Exec Macros
6048 @subsection Macros to control conditional execution
6049 @findex conditional execution
6052 There is one macro that may need to be defined for targets
6053 supporting conditional execution, independent of how they
6054 represent conditional branches.
6056 @defmac REVERSE_CONDEXEC_PREDICATES_P (@var{op1}, @var{op2})
6057 A C expression that returns true if the conditional execution predicate
6058 @var{op1}, a comparison operation, is the inverse of @var{op2} and vice
6059 versa. Define this to return 0 if the target has conditional execution
6060 predicates that cannot be reversed safely. There is no need to validate
6061 that the arguments of op1 and op2 are the same, this is done separately.
6062 If no expansion is specified, this macro is defined as follows:
6065 #define REVERSE_CONDEXEC_PREDICATES_P (x, y) \
6066 (GET_CODE ((x)) == reversed_comparison_code ((y), NULL))
6071 @section Describing Relative Costs of Operations
6072 @cindex costs of instructions
6073 @cindex relative costs
6074 @cindex speed of instructions
6076 These macros let you describe the relative speed of various operations
6077 on the target machine.
6079 @defmac REGISTER_MOVE_COST (@var{mode}, @var{from}, @var{to})
6080 A C expression for the cost of moving data of mode @var{mode} from a
6081 register in class @var{from} to one in class @var{to}. The classes are
6082 expressed using the enumeration values such as @code{GENERAL_REGS}. A
6083 value of 2 is the default; other values are interpreted relative to
6086 It is not required that the cost always equal 2 when @var{from} is the
6087 same as @var{to}; on some machines it is expensive to move between
6088 registers if they are not general registers.
6090 If reload sees an insn consisting of a single @code{set} between two
6091 hard registers, and if @code{REGISTER_MOVE_COST} applied to their
6092 classes returns a value of 2, reload does not check to ensure that the
6093 constraints of the insn are met. Setting a cost of other than 2 will
6094 allow reload to verify that the constraints are met. You should do this
6095 if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
6097 These macros are obsolete, new ports should use the target hook
6098 @code{TARGET_REGISTER_MOVE_COST} instead.
6101 @hook TARGET_REGISTER_MOVE_COST
6102 This target hook should return the cost of moving data of mode @var{mode}
6103 from a register in class @var{from} to one in class @var{to}. The classes
6104 are expressed using the enumeration values such as @code{GENERAL_REGS}.
6105 A value of 2 is the default; other values are interpreted relative to
6108 It is not required that the cost always equal 2 when @var{from} is the
6109 same as @var{to}; on some machines it is expensive to move between
6110 registers if they are not general registers.
6112 If reload sees an insn consisting of a single @code{set} between two
6113 hard registers, and if @code{TARGET_REGISTER_MOVE_COST} applied to their
6114 classes returns a value of 2, reload does not check to ensure that the
6115 constraints of the insn are met. Setting a cost of other than 2 will
6116 allow reload to verify that the constraints are met. You should do this
6117 if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
6119 The default version of this function returns 2.
6122 @defmac MEMORY_MOVE_COST (@var{mode}, @var{class}, @var{in})
6123 A C expression for the cost of moving data of mode @var{mode} between a
6124 register of class @var{class} and memory; @var{in} is zero if the value
6125 is to be written to memory, nonzero if it is to be read in. This cost
6126 is relative to those in @code{REGISTER_MOVE_COST}. If moving between
6127 registers and memory is more expensive than between two registers, you
6128 should define this macro to express the relative cost.
6130 If you do not define this macro, GCC uses a default cost of 4 plus
6131 the cost of copying via a secondary reload register, if one is
6132 needed. If your machine requires a secondary reload register to copy
6133 between memory and a register of @var{class} but the reload mechanism is
6134 more complex than copying via an intermediate, define this macro to
6135 reflect the actual cost of the move.
6137 GCC defines the function @code{memory_move_secondary_cost} if
6138 secondary reloads are needed. It computes the costs due to copying via
6139 a secondary register. If your machine copies from memory using a
6140 secondary register in the conventional way but the default base value of
6141 4 is not correct for your machine, define this macro to add some other
6142 value to the result of that function. The arguments to that function
6143 are the same as to this macro.
6145 These macros are obsolete, new ports should use the target hook
6146 @code{TARGET_MEMORY_MOVE_COST} instead.
6149 @hook TARGET_MEMORY_MOVE_COST
6150 This target hook should return the cost of moving data of mode @var{mode}
6151 between a register of class @var{rclass} and memory; @var{in} is @code{false}
6152 if the value is to be written to memory, @code{true} if it is to be read in.
6153 This cost is relative to those in @code{TARGET_REGISTER_MOVE_COST}.
6154 If moving between registers and memory is more expensive than between two
6155 registers, you should add this target hook to express the relative cost.
6157 If you do not add this target hook, GCC uses a default cost of 4 plus
6158 the cost of copying via a secondary reload register, if one is
6159 needed. If your machine requires a secondary reload register to copy
6160 between memory and a register of @var{rclass} but the reload mechanism is
6161 more complex than copying via an intermediate, use this target hook to
6162 reflect the actual cost of the move.
6164 GCC defines the function @code{memory_move_secondary_cost} if
6165 secondary reloads are needed. It computes the costs due to copying via
6166 a secondary register. If your machine copies from memory using a
6167 secondary register in the conventional way but the default base value of
6168 4 is not correct for your machine, use this target hook to add some other
6169 value to the result of that function. The arguments to that function
6170 are the same as to this target hook.
6173 @defmac BRANCH_COST (@var{speed_p}, @var{predictable_p})
6174 A C expression for the cost of a branch instruction. A value of 1 is
6175 the default; other values are interpreted relative to that. Parameter
6176 @var{speed_p} is true when the branch in question should be optimized
6177 for speed. When it is false, @code{BRANCH_COST} should return a value
6178 optimal for code size rather than performance. @var{predictable_p} is
6179 true for well-predicted branches. On many architectures the
6180 @code{BRANCH_COST} can be reduced then.
6183 Here are additional macros which do not specify precise relative costs,
6184 but only that certain actions are more expensive than GCC would
6187 @defmac SLOW_BYTE_ACCESS
6188 Define this macro as a C expression which is nonzero if accessing less
6189 than a word of memory (i.e.@: a @code{char} or a @code{short}) is no
6190 faster than accessing a word of memory, i.e., if such access
6191 require more than one instruction or if there is no difference in cost
6192 between byte and (aligned) word loads.
6194 When this macro is not defined, the compiler will access a field by
6195 finding the smallest containing object; when it is defined, a fullword
6196 load will be used if alignment permits. Unless bytes accesses are
6197 faster than word accesses, using word accesses is preferable since it
6198 may eliminate subsequent memory access if subsequent accesses occur to
6199 other fields in the same word of the structure, but to different bytes.
6202 @defmac SLOW_UNALIGNED_ACCESS (@var{mode}, @var{alignment})
6203 Define this macro to be the value 1 if memory accesses described by the
6204 @var{mode} and @var{alignment} parameters have a cost many times greater
6205 than aligned accesses, for example if they are emulated in a trap
6208 When this macro is nonzero, the compiler will act as if
6209 @code{STRICT_ALIGNMENT} were nonzero when generating code for block
6210 moves. This can cause significantly more instructions to be produced.
6211 Therefore, do not set this macro nonzero if unaligned accesses only add a
6212 cycle or two to the time for a memory access.
6214 If the value of this macro is always zero, it need not be defined. If
6215 this macro is defined, it should produce a nonzero value when
6216 @code{STRICT_ALIGNMENT} is nonzero.
6219 @defmac MOVE_RATIO (@var{speed})
6220 The threshold of number of scalar memory-to-memory move insns, @emph{below}
6221 which a sequence of insns should be generated instead of a
6222 string move insn or a library call. Increasing the value will always
6223 make code faster, but eventually incurs high cost in increased code size.
6225 Note that on machines where the corresponding move insn is a
6226 @code{define_expand} that emits a sequence of insns, this macro counts
6227 the number of such sequences.
6229 The parameter @var{speed} is true if the code is currently being
6230 optimized for speed rather than size.
6232 If you don't define this, a reasonable default is used.
6235 @defmac MOVE_BY_PIECES_P (@var{size}, @var{alignment})
6236 A C expression used to determine whether @code{move_by_pieces} will be used to
6237 copy a chunk of memory, or whether some other block move mechanism
6238 will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6239 than @code{MOVE_RATIO}.
6242 @defmac MOVE_MAX_PIECES
6243 A C expression used by @code{move_by_pieces} to determine the largest unit
6244 a load or store used to copy memory is. Defaults to @code{MOVE_MAX}.
6247 @defmac CLEAR_RATIO (@var{speed})
6248 The threshold of number of scalar move insns, @emph{below} which a sequence
6249 of insns should be generated to clear memory instead of a string clear insn
6250 or a library call. Increasing the value will always make code faster, but
6251 eventually incurs high cost in increased code size.
6253 The parameter @var{speed} is true if the code is currently being
6254 optimized for speed rather than size.
6256 If you don't define this, a reasonable default is used.
6259 @defmac CLEAR_BY_PIECES_P (@var{size}, @var{alignment})
6260 A C expression used to determine whether @code{clear_by_pieces} will be used
6261 to clear a chunk of memory, or whether some other block clear mechanism
6262 will be used. Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6263 than @code{CLEAR_RATIO}.
6266 @defmac SET_RATIO (@var{speed})
6267 The threshold of number of scalar move insns, @emph{below} which a sequence
6268 of insns should be generated to set memory to a constant value, instead of
6269 a block set insn or a library call.
6270 Increasing the value will always make code faster, but
6271 eventually incurs high cost in increased code size.
6273 The parameter @var{speed} is true if the code is currently being
6274 optimized for speed rather than size.
6276 If you don't define this, it defaults to the value of @code{MOVE_RATIO}.
6279 @defmac SET_BY_PIECES_P (@var{size}, @var{alignment})
6280 A C expression used to determine whether @code{store_by_pieces} will be
6281 used to set a chunk of memory to a constant value, or whether some
6282 other mechanism will be used. Used by @code{__builtin_memset} when
6283 storing values other than constant zero.
6284 Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6285 than @code{SET_RATIO}.
6288 @defmac STORE_BY_PIECES_P (@var{size}, @var{alignment})
6289 A C expression used to determine whether @code{store_by_pieces} will be
6290 used to set a chunk of memory to a constant string value, or whether some
6291 other mechanism will be used. Used by @code{__builtin_strcpy} when
6292 called with a constant source string.
6293 Defaults to 1 if @code{move_by_pieces_ninsns} returns less
6294 than @code{MOVE_RATIO}.
6297 @defmac USE_LOAD_POST_INCREMENT (@var{mode})
6298 A C expression used to determine whether a load postincrement is a good
6299 thing to use for a given mode. Defaults to the value of
6300 @code{HAVE_POST_INCREMENT}.
6303 @defmac USE_LOAD_POST_DECREMENT (@var{mode})
6304 A C expression used to determine whether a load postdecrement is a good
6305 thing to use for a given mode. Defaults to the value of
6306 @code{HAVE_POST_DECREMENT}.
6309 @defmac USE_LOAD_PRE_INCREMENT (@var{mode})
6310 A C expression used to determine whether a load preincrement is a good
6311 thing to use for a given mode. Defaults to the value of
6312 @code{HAVE_PRE_INCREMENT}.
6315 @defmac USE_LOAD_PRE_DECREMENT (@var{mode})
6316 A C expression used to determine whether a load predecrement is a good
6317 thing to use for a given mode. Defaults to the value of
6318 @code{HAVE_PRE_DECREMENT}.
6321 @defmac USE_STORE_POST_INCREMENT (@var{mode})
6322 A C expression used to determine whether a store postincrement is a good
6323 thing to use for a given mode. Defaults to the value of
6324 @code{HAVE_POST_INCREMENT}.
6327 @defmac USE_STORE_POST_DECREMENT (@var{mode})
6328 A C expression used to determine whether a store postdecrement is a good
6329 thing to use for a given mode. Defaults to the value of
6330 @code{HAVE_POST_DECREMENT}.
6333 @defmac USE_STORE_PRE_INCREMENT (@var{mode})
6334 This macro is used to determine whether a store preincrement is a good
6335 thing to use for a given mode. Defaults to the value of
6336 @code{HAVE_PRE_INCREMENT}.
6339 @defmac USE_STORE_PRE_DECREMENT (@var{mode})
6340 This macro is used to determine whether a store predecrement is a good
6341 thing to use for a given mode. Defaults to the value of
6342 @code{HAVE_PRE_DECREMENT}.
6345 @defmac NO_FUNCTION_CSE
6346 Define this macro if it is as good or better to call a constant
6347 function address than to call an address kept in a register.
6350 @defmac RANGE_TEST_NON_SHORT_CIRCUIT
6351 Define this macro if a non-short-circuit operation produced by
6352 @samp{fold_range_test ()} is optimal. This macro defaults to true if
6353 @code{BRANCH_COST} is greater than or equal to the value 2.
6356 @hook TARGET_RTX_COSTS
6357 This target hook describes the relative costs of RTL expressions.
6359 The cost may depend on the precise form of the expression, which is
6360 available for examination in @var{x}, and the fact that @var{x} appears
6361 as operand @var{opno} of an expression with rtx code @var{outer_code}.
6362 That is, the hook can assume that there is some rtx @var{y} such
6363 that @samp{GET_CODE (@var{y}) == @var{outer_code}} and such that
6364 either (a) @samp{XEXP (@var{y}, @var{opno}) == @var{x}} or
6365 (b) @samp{XVEC (@var{y}, @var{opno})} contains @var{x}.
6367 @var{code} is @var{x}'s expression code---redundant, since it can be
6368 obtained with @code{GET_CODE (@var{x})}.
6370 In implementing this hook, you can use the construct
6371 @code{COSTS_N_INSNS (@var{n})} to specify a cost equal to @var{n} fast
6374 On entry to the hook, @code{*@var{total}} contains a default estimate
6375 for the cost of the expression. The hook should modify this value as
6376 necessary. Traditionally, the default costs are @code{COSTS_N_INSNS (5)}
6377 for multiplications, @code{COSTS_N_INSNS (7)} for division and modulus
6378 operations, and @code{COSTS_N_INSNS (1)} for all other operations.
6380 When optimizing for code size, i.e.@: when @code{speed} is
6381 false, this target hook should be used to estimate the relative
6382 size cost of an expression, again relative to @code{COSTS_N_INSNS}.
6384 The hook returns true when all subexpressions of @var{x} have been
6385 processed, and false when @code{rtx_cost} should recurse.
6388 @hook TARGET_ADDRESS_COST
6389 This hook computes the cost of an addressing mode that contains
6390 @var{address}. If not defined, the cost is computed from
6391 the @var{address} expression and the @code{TARGET_RTX_COST} hook.
6393 For most CISC machines, the default cost is a good approximation of the
6394 true cost of the addressing mode. However, on RISC machines, all
6395 instructions normally have the same length and execution time. Hence
6396 all addresses will have equal costs.
6398 In cases where more than one form of an address is known, the form with
6399 the lowest cost will be used. If multiple forms have the same, lowest,
6400 cost, the one that is the most complex will be used.
6402 For example, suppose an address that is equal to the sum of a register
6403 and a constant is used twice in the same basic block. When this macro
6404 is not defined, the address will be computed in a register and memory
6405 references will be indirect through that register. On machines where
6406 the cost of the addressing mode containing the sum is no higher than
6407 that of a simple indirect reference, this will produce an additional
6408 instruction and possibly require an additional register. Proper
6409 specification of this macro eliminates this overhead for such machines.
6411 This hook is never called with an invalid address.
6413 On machines where an address involving more than one register is as
6414 cheap as an address computation involving only one register, defining
6415 @code{TARGET_ADDRESS_COST} to reflect this can cause two registers to
6416 be live over a region of code where only one would have been if
6417 @code{TARGET_ADDRESS_COST} were not defined in that manner. This effect
6418 should be considered in the definition of this macro. Equivalent costs
6419 should probably only be given to addresses with different numbers of
6420 registers on machines with lots of registers.
6424 @section Adjusting the Instruction Scheduler
6426 The instruction scheduler may need a fair amount of machine-specific
6427 adjustment in order to produce good code. GCC provides several target
6428 hooks for this purpose. It is usually enough to define just a few of
6429 them: try the first ones in this list first.
6431 @hook TARGET_SCHED_ISSUE_RATE
6432 This hook returns the maximum number of instructions that can ever
6433 issue at the same time on the target machine. The default is one.
6434 Although the insn scheduler can define itself the possibility of issue
6435 an insn on the same cycle, the value can serve as an additional
6436 constraint to issue insns on the same simulated processor cycle (see
6437 hooks @samp{TARGET_SCHED_REORDER} and @samp{TARGET_SCHED_REORDER2}).
6438 This value must be constant over the entire compilation. If you need
6439 it to vary depending on what the instructions are, you must use
6440 @samp{TARGET_SCHED_VARIABLE_ISSUE}.
6443 @hook TARGET_SCHED_VARIABLE_ISSUE
6444 This hook is executed by the scheduler after it has scheduled an insn
6445 from the ready list. It should return the number of insns which can
6446 still be issued in the current cycle. The default is
6447 @samp{@w{@var{more} - 1}} for insns other than @code{CLOBBER} and
6448 @code{USE}, which normally are not counted against the issue rate.
6449 You should define this hook if some insns take more machine resources
6450 than others, so that fewer insns can follow them in the same cycle.
6451 @var{file} is either a null pointer, or a stdio stream to write any
6452 debug output to. @var{verbose} is the verbose level provided by
6453 @option{-fsched-verbose-@var{n}}. @var{insn} is the instruction that
6457 @hook TARGET_SCHED_ADJUST_COST
6458 This function corrects the value of @var{cost} based on the
6459 relationship between @var{insn} and @var{dep_insn} through the
6460 dependence @var{link}. It should return the new value. The default
6461 is to make no adjustment to @var{cost}. This can be used for example
6462 to specify to the scheduler using the traditional pipeline description
6463 that an output- or anti-dependence does not incur the same cost as a
6464 data-dependence. If the scheduler using the automaton based pipeline
6465 description, the cost of anti-dependence is zero and the cost of
6466 output-dependence is maximum of one and the difference of latency
6467 times of the first and the second insns. If these values are not
6468 acceptable, you could use the hook to modify them too. See also
6469 @pxref{Processor pipeline description}.
6472 @hook TARGET_SCHED_ADJUST_PRIORITY
6473 This hook adjusts the integer scheduling priority @var{priority} of
6474 @var{insn}. It should return the new priority. Increase the priority to
6475 execute @var{insn} earlier, reduce the priority to execute @var{insn}
6476 later. Do not define this hook if you do not need to adjust the
6477 scheduling priorities of insns.
6480 @hook TARGET_SCHED_REORDER
6481 This hook is executed by the scheduler after it has scheduled the ready
6482 list, to allow the machine description to reorder it (for example to
6483 combine two small instructions together on @samp{VLIW} machines).
6484 @var{file} is either a null pointer, or a stdio stream to write any
6485 debug output to. @var{verbose} is the verbose level provided by
6486 @option{-fsched-verbose-@var{n}}. @var{ready} is a pointer to the ready
6487 list of instructions that are ready to be scheduled. @var{n_readyp} is
6488 a pointer to the number of elements in the ready list. The scheduler
6489 reads the ready list in reverse order, starting with
6490 @var{ready}[@var{*n_readyp} @minus{} 1] and going to @var{ready}[0]. @var{clock}
6491 is the timer tick of the scheduler. You may modify the ready list and
6492 the number of ready insns. The return value is the number of insns that
6493 can issue this cycle; normally this is just @code{issue_rate}. See also
6494 @samp{TARGET_SCHED_REORDER2}.
6497 @hook TARGET_SCHED_REORDER2
6498 Like @samp{TARGET_SCHED_REORDER}, but called at a different time. That
6499 function is called whenever the scheduler starts a new cycle. This one
6500 is called once per iteration over a cycle, immediately after
6501 @samp{TARGET_SCHED_VARIABLE_ISSUE}; it can reorder the ready list and
6502 return the number of insns to be scheduled in the same cycle. Defining
6503 this hook can be useful if there are frequent situations where
6504 scheduling one insn causes other insns to become ready in the same
6505 cycle. These other insns can then be taken into account properly.
6508 @hook TARGET_SCHED_DEPENDENCIES_EVALUATION_HOOK
6509 This hook is called after evaluation forward dependencies of insns in
6510 chain given by two parameter values (@var{head} and @var{tail}
6511 correspondingly) but before insns scheduling of the insn chain. For
6512 example, it can be used for better insn classification if it requires
6513 analysis of dependencies. This hook can use backward and forward
6514 dependencies of the insn scheduler because they are already
6518 @hook TARGET_SCHED_INIT
6519 This hook is executed by the scheduler at the beginning of each block of
6520 instructions that are to be scheduled. @var{file} is either a null
6521 pointer, or a stdio stream to write any debug output to. @var{verbose}
6522 is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6523 @var{max_ready} is the maximum number of insns in the current scheduling
6524 region that can be live at the same time. This can be used to allocate
6525 scratch space if it is needed, e.g.@: by @samp{TARGET_SCHED_REORDER}.
6528 @hook TARGET_SCHED_FINISH
6529 This hook is executed by the scheduler at the end of each block of
6530 instructions that are to be scheduled. It can be used to perform
6531 cleanup of any actions done by the other scheduling hooks. @var{file}
6532 is either a null pointer, or a stdio stream to write any debug output
6533 to. @var{verbose} is the verbose level provided by
6534 @option{-fsched-verbose-@var{n}}.
6537 @hook TARGET_SCHED_INIT_GLOBAL
6538 This hook is executed by the scheduler after function level initializations.
6539 @var{file} is either a null pointer, or a stdio stream to write any debug output to.
6540 @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6541 @var{old_max_uid} is the maximum insn uid when scheduling begins.
6544 @hook TARGET_SCHED_FINISH_GLOBAL
6545 This is the cleanup hook corresponding to @code{TARGET_SCHED_INIT_GLOBAL}.
6546 @var{file} is either a null pointer, or a stdio stream to write any debug output to.
6547 @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6550 @hook TARGET_SCHED_DFA_PRE_CYCLE_INSN
6551 The hook returns an RTL insn. The automaton state used in the
6552 pipeline hazard recognizer is changed as if the insn were scheduled
6553 when the new simulated processor cycle starts. Usage of the hook may
6554 simplify the automaton pipeline description for some @acronym{VLIW}
6555 processors. If the hook is defined, it is used only for the automaton
6556 based pipeline description. The default is not to change the state
6557 when the new simulated processor cycle starts.
6560 @hook TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN
6561 The hook can be used to initialize data used by the previous hook.
6564 @hook TARGET_SCHED_DFA_POST_CYCLE_INSN
6565 The hook is analogous to @samp{TARGET_SCHED_DFA_PRE_CYCLE_INSN} but used
6566 to changed the state as if the insn were scheduled when the new
6567 simulated processor cycle finishes.
6570 @hook TARGET_SCHED_INIT_DFA_POST_CYCLE_INSN
6571 The hook is analogous to @samp{TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN} but
6572 used to initialize data used by the previous hook.
6575 @hook TARGET_SCHED_DFA_PRE_ADVANCE_CYCLE
6576 The hook to notify target that the current simulated cycle is about to finish.
6577 The hook is analogous to @samp{TARGET_SCHED_DFA_PRE_CYCLE_INSN} but used
6578 to change the state in more complicated situations - e.g., when advancing
6579 state on a single insn is not enough.
6582 @hook TARGET_SCHED_DFA_POST_ADVANCE_CYCLE
6583 The hook to notify target that new simulated cycle has just started.
6584 The hook is analogous to @samp{TARGET_SCHED_DFA_POST_CYCLE_INSN} but used
6585 to change the state in more complicated situations - e.g., when advancing
6586 state on a single insn is not enough.
6589 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD
6590 This hook controls better choosing an insn from the ready insn queue
6591 for the @acronym{DFA}-based insn scheduler. Usually the scheduler
6592 chooses the first insn from the queue. If the hook returns a positive
6593 value, an additional scheduler code tries all permutations of
6594 @samp{TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD ()}
6595 subsequent ready insns to choose an insn whose issue will result in
6596 maximal number of issued insns on the same cycle. For the
6597 @acronym{VLIW} processor, the code could actually solve the problem of
6598 packing simple insns into the @acronym{VLIW} insn. Of course, if the
6599 rules of @acronym{VLIW} packing are described in the automaton.
6601 This code also could be used for superscalar @acronym{RISC}
6602 processors. Let us consider a superscalar @acronym{RISC} processor
6603 with 3 pipelines. Some insns can be executed in pipelines @var{A} or
6604 @var{B}, some insns can be executed only in pipelines @var{B} or
6605 @var{C}, and one insn can be executed in pipeline @var{B}. The
6606 processor may issue the 1st insn into @var{A} and the 2nd one into
6607 @var{B}. In this case, the 3rd insn will wait for freeing @var{B}
6608 until the next cycle. If the scheduler issues the 3rd insn the first,
6609 the processor could issue all 3 insns per cycle.
6611 Actually this code demonstrates advantages of the automaton based
6612 pipeline hazard recognizer. We try quickly and easy many insn
6613 schedules to choose the best one.
6615 The default is no multipass scheduling.
6618 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD
6620 This hook controls what insns from the ready insn queue will be
6621 considered for the multipass insn scheduling. If the hook returns
6622 zero for @var{insn}, the insn will be not chosen to
6625 The default is that any ready insns can be chosen to be issued.
6628 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_BEGIN
6629 This hook prepares the target backend for a new round of multipass
6633 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_ISSUE
6634 This hook is called when multipass scheduling evaluates instruction INSN.
6637 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_BACKTRACK
6638 This is called when multipass scheduling backtracks from evaluation of
6642 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_END
6643 This hook notifies the target about the result of the concluded current
6644 round of multipass scheduling.
6647 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_INIT
6648 This hook initializes target-specific data used in multipass scheduling.
6651 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_FINI
6652 This hook finalizes target-specific data used in multipass scheduling.
6655 @hook TARGET_SCHED_DFA_NEW_CYCLE
6656 This hook is called by the insn scheduler before issuing @var{insn}
6657 on cycle @var{clock}. If the hook returns nonzero,
6658 @var{insn} is not issued on this processor cycle. Instead,
6659 the processor cycle is advanced. If *@var{sort_p}
6660 is zero, the insn ready queue is not sorted on the new cycle
6661 start as usually. @var{dump} and @var{verbose} specify the file and
6662 verbosity level to use for debugging output.
6663 @var{last_clock} and @var{clock} are, respectively, the
6664 processor cycle on which the previous insn has been issued,
6665 and the current processor cycle.
6668 @hook TARGET_SCHED_IS_COSTLY_DEPENDENCE
6669 This hook is used to define which dependences are considered costly by
6670 the target, so costly that it is not advisable to schedule the insns that
6671 are involved in the dependence too close to one another. The parameters
6672 to this hook are as follows: The first parameter @var{_dep} is the dependence
6673 being evaluated. The second parameter @var{cost} is the cost of the
6674 dependence as estimated by the scheduler, and the third
6675 parameter @var{distance} is the distance in cycles between the two insns.
6676 The hook returns @code{true} if considering the distance between the two
6677 insns the dependence between them is considered costly by the target,
6678 and @code{false} otherwise.
6680 Defining this hook can be useful in multiple-issue out-of-order machines,
6681 where (a) it's practically hopeless to predict the actual data/resource
6682 delays, however: (b) there's a better chance to predict the actual grouping
6683 that will be formed, and (c) correctly emulating the grouping can be very
6684 important. In such targets one may want to allow issuing dependent insns
6685 closer to one another---i.e., closer than the dependence distance; however,
6686 not in cases of ``costly dependences'', which this hooks allows to define.
6689 @hook TARGET_SCHED_H_I_D_EXTENDED
6690 This hook is called by the insn scheduler after emitting a new instruction to
6691 the instruction stream. The hook notifies a target backend to extend its
6692 per instruction data structures.
6695 @hook TARGET_SCHED_ALLOC_SCHED_CONTEXT
6696 Return a pointer to a store large enough to hold target scheduling context.
6699 @hook TARGET_SCHED_INIT_SCHED_CONTEXT
6700 Initialize store pointed to by @var{tc} to hold target scheduling context.
6701 It @var{clean_p} is true then initialize @var{tc} as if scheduler is at the
6702 beginning of the block. Otherwise, copy the current context into @var{tc}.
6705 @hook TARGET_SCHED_SET_SCHED_CONTEXT
6706 Copy target scheduling context pointed to by @var{tc} to the current context.
6709 @hook TARGET_SCHED_CLEAR_SCHED_CONTEXT
6710 Deallocate internal data in target scheduling context pointed to by @var{tc}.
6713 @hook TARGET_SCHED_FREE_SCHED_CONTEXT
6714 Deallocate a store for target scheduling context pointed to by @var{tc}.
6717 @hook TARGET_SCHED_SPECULATE_INSN
6718 This hook is called by the insn scheduler when @var{insn} has only
6719 speculative dependencies and therefore can be scheduled speculatively.
6720 The hook is used to check if the pattern of @var{insn} has a speculative
6721 version and, in case of successful check, to generate that speculative
6722 pattern. The hook should return 1, if the instruction has a speculative form,
6723 or @minus{}1, if it doesn't. @var{request} describes the type of requested
6724 speculation. If the return value equals 1 then @var{new_pat} is assigned
6725 the generated speculative pattern.
6728 @hook TARGET_SCHED_NEEDS_BLOCK_P
6729 This hook is called by the insn scheduler during generation of recovery code
6730 for @var{insn}. It should return @code{true}, if the corresponding check
6731 instruction should branch to recovery code, or @code{false} otherwise.
6734 @hook TARGET_SCHED_GEN_SPEC_CHECK
6735 This hook is called by the insn scheduler to generate a pattern for recovery
6736 check instruction. If @var{mutate_p} is zero, then @var{insn} is a
6737 speculative instruction for which the check should be generated.
6738 @var{label} is either a label of a basic block, where recovery code should
6739 be emitted, or a null pointer, when requested check doesn't branch to
6740 recovery code (a simple check). If @var{mutate_p} is nonzero, then
6741 a pattern for a branchy check corresponding to a simple check denoted by
6742 @var{insn} should be generated. In this case @var{label} can't be null.
6745 @hook TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD_SPEC
6746 This hook is used as a workaround for
6747 @samp{TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD} not being
6748 called on the first instruction of the ready list. The hook is used to
6749 discard speculative instructions that stand first in the ready list from
6750 being scheduled on the current cycle. If the hook returns @code{false},
6751 @var{insn} will not be chosen to be issued.
6752 For non-speculative instructions,
6753 the hook should always return @code{true}. For example, in the ia64 backend
6754 the hook is used to cancel data speculative insns when the ALAT table
6758 @hook TARGET_SCHED_SET_SCHED_FLAGS
6759 This hook is used by the insn scheduler to find out what features should be
6761 The structure *@var{spec_info} should be filled in by the target.
6762 The structure describes speculation types that can be used in the scheduler.
6765 @hook TARGET_SCHED_SMS_RES_MII
6766 This hook is called by the swing modulo scheduler to calculate a
6767 resource-based lower bound which is based on the resources available in
6768 the machine and the resources required by each instruction. The target
6769 backend can use @var{g} to calculate such bound. A very simple lower
6770 bound will be used in case this hook is not implemented: the total number
6771 of instructions divided by the issue rate.
6774 @hook TARGET_SCHED_DISPATCH
6775 This hook is called by Haifa Scheduler. It returns true if dispatch scheduling
6776 is supported in hardware and the condition specified in the parameter is true.
6779 @hook TARGET_SCHED_DISPATCH_DO
6780 This hook is called by Haifa Scheduler. It performs the operation specified
6781 in its second parameter.
6784 @hook TARGET_SCHED_EXPOSED_PIPELINE
6787 @section Dividing the Output into Sections (Texts, Data, @dots{})
6788 @c the above section title is WAY too long. maybe cut the part between
6789 @c the (...)? --mew 10feb93
6791 An object file is divided into sections containing different types of
6792 data. In the most common case, there are three sections: the @dfn{text
6793 section}, which holds instructions and read-only data; the @dfn{data
6794 section}, which holds initialized writable data; and the @dfn{bss
6795 section}, which holds uninitialized data. Some systems have other kinds
6798 @file{varasm.c} provides several well-known sections, such as
6799 @code{text_section}, @code{data_section} and @code{bss_section}.
6800 The normal way of controlling a @code{@var{foo}_section} variable
6801 is to define the associated @code{@var{FOO}_SECTION_ASM_OP} macro,
6802 as described below. The macros are only read once, when @file{varasm.c}
6803 initializes itself, so their values must be run-time constants.
6804 They may however depend on command-line flags.
6806 @emph{Note:} Some run-time files, such @file{crtstuff.c}, also make
6807 use of the @code{@var{FOO}_SECTION_ASM_OP} macros, and expect them
6808 to be string literals.
6810 Some assemblers require a different string to be written every time a
6811 section is selected. If your assembler falls into this category, you
6812 should define the @code{TARGET_ASM_INIT_SECTIONS} hook and use
6813 @code{get_unnamed_section} to set up the sections.
6815 You must always create a @code{text_section}, either by defining
6816 @code{TEXT_SECTION_ASM_OP} or by initializing @code{text_section}
6817 in @code{TARGET_ASM_INIT_SECTIONS}. The same is true of
6818 @code{data_section} and @code{DATA_SECTION_ASM_OP}. If you do not
6819 create a distinct @code{readonly_data_section}, the default is to
6820 reuse @code{text_section}.
6822 All the other @file{varasm.c} sections are optional, and are null
6823 if the target does not provide them.
6825 @defmac TEXT_SECTION_ASM_OP
6826 A C expression whose value is a string, including spacing, containing the
6827 assembler operation that should precede instructions and read-only data.
6828 Normally @code{"\t.text"} is right.
6831 @defmac HOT_TEXT_SECTION_NAME
6832 If defined, a C string constant for the name of the section containing most
6833 frequently executed functions of the program. If not defined, GCC will provide
6834 a default definition if the target supports named sections.
6837 @defmac UNLIKELY_EXECUTED_TEXT_SECTION_NAME
6838 If defined, a C string constant for the name of the section containing unlikely
6839 executed functions in the program.
6842 @defmac DATA_SECTION_ASM_OP
6843 A C expression whose value is a string, including spacing, containing the
6844 assembler operation to identify the following data as writable initialized
6845 data. Normally @code{"\t.data"} is right.
6848 @defmac SDATA_SECTION_ASM_OP
6849 If defined, a C expression whose value is a string, including spacing,
6850 containing the assembler operation to identify the following data as
6851 initialized, writable small data.
6854 @defmac READONLY_DATA_SECTION_ASM_OP
6855 A C expression whose value is a string, including spacing, containing the
6856 assembler operation to identify the following data as read-only initialized
6860 @defmac BSS_SECTION_ASM_OP
6861 If defined, a C expression whose value is a string, including spacing,
6862 containing the assembler operation to identify the following data as
6863 uninitialized global data. If not defined, and
6864 @code{ASM_OUTPUT_ALIGNED_BSS} not defined,
6865 uninitialized global data will be output in the data section if
6866 @option{-fno-common} is passed, otherwise @code{ASM_OUTPUT_COMMON} will be
6870 @defmac SBSS_SECTION_ASM_OP
6871 If defined, a C expression whose value is a string, including spacing,
6872 containing the assembler operation to identify the following data as
6873 uninitialized, writable small data.
6876 @defmac TLS_COMMON_ASM_OP
6877 If defined, a C expression whose value is a string containing the
6878 assembler operation to identify the following data as thread-local
6879 common data. The default is @code{".tls_common"}.
6882 @defmac TLS_SECTION_ASM_FLAG
6883 If defined, a C expression whose value is a character constant
6884 containing the flag used to mark a section as a TLS section. The
6885 default is @code{'T'}.
6888 @defmac INIT_SECTION_ASM_OP
6889 If defined, a C expression whose value is a string, including spacing,
6890 containing the assembler operation to identify the following data as
6891 initialization code. If not defined, GCC will assume such a section does
6892 not exist. This section has no corresponding @code{init_section}
6893 variable; it is used entirely in runtime code.
6896 @defmac FINI_SECTION_ASM_OP
6897 If defined, a C expression whose value is a string, including spacing,
6898 containing the assembler operation to identify the following data as
6899 finalization code. If not defined, GCC will assume such a section does
6900 not exist. This section has no corresponding @code{fini_section}
6901 variable; it is used entirely in runtime code.
6904 @defmac INIT_ARRAY_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 part of the @code{.init_array} (or equivalent) section. If not
6908 defined, GCC will assume such a section does not exist. Do not define
6909 both this macro and @code{INIT_SECTION_ASM_OP}.
6912 @defmac FINI_ARRAY_SECTION_ASM_OP
6913 If defined, a C expression whose value is a string, including spacing,
6914 containing the assembler operation to identify the following data as
6915 part of the @code{.fini_array} (or equivalent) section. If not
6916 defined, GCC will assume such a section does not exist. Do not define
6917 both this macro and @code{FINI_SECTION_ASM_OP}.
6920 @defmac CRT_CALL_STATIC_FUNCTION (@var{section_op}, @var{function})
6921 If defined, an ASM statement that switches to a different section
6922 via @var{section_op}, calls @var{function}, and switches back to
6923 the text section. This is used in @file{crtstuff.c} if
6924 @code{INIT_SECTION_ASM_OP} or @code{FINI_SECTION_ASM_OP} to calls
6925 to initialization and finalization functions from the init and fini
6926 sections. By default, this macro uses a simple function call. Some
6927 ports need hand-crafted assembly code to avoid dependencies on
6928 registers initialized in the function prologue or to ensure that
6929 constant pools don't end up too far way in the text section.
6932 @defmac TARGET_LIBGCC_SDATA_SECTION
6933 If defined, a string which names the section into which small
6934 variables defined in crtstuff and libgcc should go. This is useful
6935 when the target has options for optimizing access to small data, and
6936 you want the crtstuff and libgcc routines to be conservative in what
6937 they expect of your application yet liberal in what your application
6938 expects. For example, for targets with a @code{.sdata} section (like
6939 MIPS), you could compile crtstuff with @code{-G 0} so that it doesn't
6940 require small data support from your application, but use this macro
6941 to put small data into @code{.sdata} so that your application can
6942 access these variables whether it uses small data or not.
6945 @defmac FORCE_CODE_SECTION_ALIGN
6946 If defined, an ASM statement that aligns a code section to some
6947 arbitrary boundary. This is used to force all fragments of the
6948 @code{.init} and @code{.fini} sections to have to same alignment
6949 and thus prevent the linker from having to add any padding.
6952 @defmac JUMP_TABLES_IN_TEXT_SECTION
6953 Define this macro to be an expression with a nonzero value if jump
6954 tables (for @code{tablejump} insns) should be output in the text
6955 section, along with the assembler instructions. Otherwise, the
6956 readonly data section is used.
6958 This macro is irrelevant if there is no separate readonly data section.
6961 @hook TARGET_ASM_INIT_SECTIONS
6962 Define this hook if you need to do something special to set up the
6963 @file{varasm.c} sections, or if your target has some special sections
6964 of its own that you need to create.
6966 GCC calls this hook after processing the command line, but before writing
6967 any assembly code, and before calling any of the section-returning hooks
6971 @hook TARGET_ASM_RELOC_RW_MASK
6972 Return a mask describing how relocations should be treated when
6973 selecting sections. Bit 1 should be set if global relocations
6974 should be placed in a read-write section; bit 0 should be set if
6975 local relocations should be placed in a read-write section.
6977 The default version of this function returns 3 when @option{-fpic}
6978 is in effect, and 0 otherwise. The hook is typically redefined
6979 when the target cannot support (some kinds of) dynamic relocations
6980 in read-only sections even in executables.
6983 @hook TARGET_ASM_SELECT_SECTION
6984 Return the section into which @var{exp} should be placed. You can
6985 assume that @var{exp} is either a @code{VAR_DECL} node or a constant of
6986 some sort. @var{reloc} indicates whether the initial value of @var{exp}
6987 requires link-time relocations. Bit 0 is set when variable contains
6988 local relocations only, while bit 1 is set for global relocations.
6989 @var{align} is the constant alignment in bits.
6991 The default version of this function takes care of putting read-only
6992 variables in @code{readonly_data_section}.
6994 See also @var{USE_SELECT_SECTION_FOR_FUNCTIONS}.
6997 @defmac USE_SELECT_SECTION_FOR_FUNCTIONS
6998 Define this macro if you wish TARGET_ASM_SELECT_SECTION to be called
6999 for @code{FUNCTION_DECL}s as well as for variables and constants.
7001 In the case of a @code{FUNCTION_DECL}, @var{reloc} will be zero if the
7002 function has been determined to be likely to be called, and nonzero if
7003 it is unlikely to be called.
7006 @hook TARGET_ASM_UNIQUE_SECTION
7007 Build up a unique section name, expressed as a @code{STRING_CST} node,
7008 and assign it to @samp{DECL_SECTION_NAME (@var{decl})}.
7009 As with @code{TARGET_ASM_SELECT_SECTION}, @var{reloc} indicates whether
7010 the initial value of @var{exp} requires link-time relocations.
7012 The default version of this function appends the symbol name to the
7013 ELF section name that would normally be used for the symbol. For
7014 example, the function @code{foo} would be placed in @code{.text.foo}.
7015 Whatever the actual target object format, this is often good enough.
7018 @hook TARGET_ASM_FUNCTION_RODATA_SECTION
7019 Return the readonly data section associated with
7020 @samp{DECL_SECTION_NAME (@var{decl})}.
7021 The default version of this function selects @code{.gnu.linkonce.r.name} if
7022 the function's section is @code{.gnu.linkonce.t.name}, @code{.rodata.name}
7023 if function is in @code{.text.name}, and the normal readonly-data section
7027 @hook TARGET_ASM_MERGEABLE_RODATA_PREFIX
7029 @hook TARGET_ASM_SELECT_RTX_SECTION
7030 Return the section into which a constant @var{x}, of mode @var{mode},
7031 should be placed. You can assume that @var{x} is some kind of
7032 constant in RTL@. The argument @var{mode} is redundant except in the
7033 case of a @code{const_int} rtx. @var{align} is the constant alignment
7036 The default version of this function takes care of putting symbolic
7037 constants in @code{flag_pic} mode in @code{data_section} and everything
7038 else in @code{readonly_data_section}.
7041 @hook TARGET_MANGLE_DECL_ASSEMBLER_NAME
7042 Define this hook if you need to postprocess the assembler name generated
7043 by target-independent code. The @var{id} provided to this hook will be
7044 the computed name (e.g., the macro @code{DECL_NAME} of the @var{decl} in C,
7045 or the mangled name of the @var{decl} in C++). The return value of the
7046 hook is an @code{IDENTIFIER_NODE} for the appropriate mangled name on
7047 your target system. The default implementation of this hook just
7048 returns the @var{id} provided.
7051 @hook TARGET_ENCODE_SECTION_INFO
7052 Define this hook if references to a symbol or a constant must be
7053 treated differently depending on something about the variable or
7054 function named by the symbol (such as what section it is in).
7056 The hook is executed immediately after rtl has been created for
7057 @var{decl}, which may be a variable or function declaration or
7058 an entry in the constant pool. In either case, @var{rtl} is the
7059 rtl in question. Do @emph{not} use @code{DECL_RTL (@var{decl})}
7060 in this hook; that field may not have been initialized yet.
7062 In the case of a constant, it is safe to assume that the rtl is
7063 a @code{mem} whose address is a @code{symbol_ref}. Most decls
7064 will also have this form, but that is not guaranteed. Global
7065 register variables, for instance, will have a @code{reg} for their
7066 rtl. (Normally the right thing to do with such unusual rtl is
7069 The @var{new_decl_p} argument will be true if this is the first time
7070 that @code{TARGET_ENCODE_SECTION_INFO} has been invoked on this decl. It will
7071 be false for subsequent invocations, which will happen for duplicate
7072 declarations. Whether or not anything must be done for the duplicate
7073 declaration depends on whether the hook examines @code{DECL_ATTRIBUTES}.
7074 @var{new_decl_p} is always true when the hook is called for a constant.
7076 @cindex @code{SYMBOL_REF_FLAG}, in @code{TARGET_ENCODE_SECTION_INFO}
7077 The usual thing for this hook to do is to record flags in the
7078 @code{symbol_ref}, using @code{SYMBOL_REF_FLAG} or @code{SYMBOL_REF_FLAGS}.
7079 Historically, the name string was modified if it was necessary to
7080 encode more than one bit of information, but this practice is now
7081 discouraged; use @code{SYMBOL_REF_FLAGS}.
7083 The default definition of this hook, @code{default_encode_section_info}
7084 in @file{varasm.c}, sets a number of commonly-useful bits in
7085 @code{SYMBOL_REF_FLAGS}. Check whether the default does what you need
7086 before overriding it.
7089 @hook TARGET_STRIP_NAME_ENCODING
7090 Decode @var{name} and return the real name part, sans
7091 the characters that @code{TARGET_ENCODE_SECTION_INFO}
7095 @hook TARGET_IN_SMALL_DATA_P
7096 Returns true if @var{exp} should be placed into a ``small data'' section.
7097 The default version of this hook always returns false.
7100 @hook TARGET_HAVE_SRODATA_SECTION
7101 Contains the value true if the target places read-only
7102 ``small data'' into a separate section. The default value is false.
7105 @hook TARGET_PROFILE_BEFORE_PROLOGUE
7107 @hook TARGET_BINDS_LOCAL_P
7108 Returns true if @var{exp} names an object for which name resolution
7109 rules must resolve to the current ``module'' (dynamic shared library
7110 or executable image).
7112 The default version of this hook implements the name resolution rules
7113 for ELF, which has a looser model of global name binding than other
7114 currently supported object file formats.
7117 @hook TARGET_HAVE_TLS
7118 Contains the value true if the target supports thread-local storage.
7119 The default value is false.
7124 @section Position Independent Code
7125 @cindex position independent code
7128 This section describes macros that help implement generation of position
7129 independent code. Simply defining these macros is not enough to
7130 generate valid PIC; you must also add support to the hook
7131 @code{TARGET_LEGITIMATE_ADDRESS_P} and to the macro
7132 @code{PRINT_OPERAND_ADDRESS}, as well as @code{LEGITIMIZE_ADDRESS}. You
7133 must modify the definition of @samp{movsi} to do something appropriate
7134 when the source operand contains a symbolic address. You may also
7135 need to alter the handling of switch statements so that they use
7137 @c i rearranged the order of the macros above to try to force one of
7138 @c them to the next line, to eliminate an overfull hbox. --mew 10feb93
7140 @defmac PIC_OFFSET_TABLE_REGNUM
7141 The register number of the register used to address a table of static
7142 data addresses in memory. In some cases this register is defined by a
7143 processor's ``application binary interface'' (ABI)@. When this macro
7144 is defined, RTL is generated for this register once, as with the stack
7145 pointer and frame pointer registers. If this macro is not defined, it
7146 is up to the machine-dependent files to allocate such a register (if
7147 necessary). Note that this register must be fixed when in use (e.g.@:
7148 when @code{flag_pic} is true).
7151 @defmac PIC_OFFSET_TABLE_REG_CALL_CLOBBERED
7152 A C expression that is nonzero if the register defined by
7153 @code{PIC_OFFSET_TABLE_REGNUM} is clobbered by calls. If not defined,
7154 the default is zero. Do not define
7155 this macro if @code{PIC_OFFSET_TABLE_REGNUM} is not defined.
7158 @defmac LEGITIMATE_PIC_OPERAND_P (@var{x})
7159 A C expression that is nonzero if @var{x} is a legitimate immediate
7160 operand on the target machine when generating position independent code.
7161 You can assume that @var{x} satisfies @code{CONSTANT_P}, so you need not
7162 check this. You can also assume @var{flag_pic} is true, so you need not
7163 check it either. You need not define this macro if all constants
7164 (including @code{SYMBOL_REF}) can be immediate operands when generating
7165 position independent code.
7168 @node Assembler Format
7169 @section Defining the Output Assembler Language
7171 This section describes macros whose principal purpose is to describe how
7172 to write instructions in assembler language---rather than what the
7176 * File Framework:: Structural information for the assembler file.
7177 * Data Output:: Output of constants (numbers, strings, addresses).
7178 * Uninitialized Data:: Output of uninitialized variables.
7179 * Label Output:: Output and generation of labels.
7180 * Initialization:: General principles of initialization
7181 and termination routines.
7182 * Macros for Initialization::
7183 Specific macros that control the handling of
7184 initialization and termination routines.
7185 * Instruction Output:: Output of actual instructions.
7186 * Dispatch Tables:: Output of jump tables.
7187 * Exception Region Output:: Output of exception region code.
7188 * Alignment Output:: Pseudo ops for alignment and skipping data.
7191 @node File Framework
7192 @subsection The Overall Framework of an Assembler File
7193 @cindex assembler format
7194 @cindex output of assembler code
7196 @c prevent bad page break with this line
7197 This describes the overall framework of an assembly file.
7199 @findex default_file_start
7200 @hook TARGET_ASM_FILE_START
7201 Output to @code{asm_out_file} any text which the assembler expects to
7202 find at the beginning of a file. The default behavior is controlled
7203 by two flags, documented below. Unless your target's assembler is
7204 quite unusual, if you override the default, you should call
7205 @code{default_file_start} at some point in your target hook. This
7206 lets other target files rely on these variables.
7209 @hook TARGET_ASM_FILE_START_APP_OFF
7210 If this flag is true, the text of the macro @code{ASM_APP_OFF} will be
7211 printed as the very first line in the assembly file, unless
7212 @option{-fverbose-asm} is in effect. (If that macro has been defined
7213 to the empty string, this variable has no effect.) With the normal
7214 definition of @code{ASM_APP_OFF}, the effect is to notify the GNU
7215 assembler that it need not bother stripping comments or extra
7216 whitespace from its input. This allows it to work a bit faster.
7218 The default is false. You should not set it to true unless you have
7219 verified that your port does not generate any extra whitespace or
7220 comments that will cause GAS to issue errors in NO_APP mode.
7223 @hook TARGET_ASM_FILE_START_FILE_DIRECTIVE
7224 If this flag is true, @code{output_file_directive} will be called
7225 for the primary source file, immediately after printing
7226 @code{ASM_APP_OFF} (if that is enabled). Most ELF assemblers expect
7227 this to be done. The default is false.
7230 @hook TARGET_ASM_FILE_END
7231 Output to @code{asm_out_file} any text which the assembler expects
7232 to find at the end of a file. The default is to output nothing.
7235 @deftypefun void file_end_indicate_exec_stack ()
7236 Some systems use a common convention, the @samp{.note.GNU-stack}
7237 special section, to indicate whether or not an object file relies on
7238 the stack being executable. If your system uses this convention, you
7239 should define @code{TARGET_ASM_FILE_END} to this function. If you
7240 need to do other things in that hook, have your hook function call
7244 @hook TARGET_ASM_LTO_START
7245 Output to @code{asm_out_file} any text which the assembler expects
7246 to find at the start of an LTO section. The default is to output
7250 @hook TARGET_ASM_LTO_END
7251 Output to @code{asm_out_file} any text which the assembler expects
7252 to find at the end of an LTO section. The default is to output
7256 @hook TARGET_ASM_CODE_END
7257 Output to @code{asm_out_file} any text which is needed before emitting
7258 unwind info and debug info at the end of a file. Some targets emit
7259 here PIC setup thunks that cannot be emitted at the end of file,
7260 because they couldn't have unwind info then. The default is to output
7264 @defmac ASM_COMMENT_START
7265 A C string constant describing how to begin a comment in the target
7266 assembler language. The compiler assumes that the comment will end at
7267 the end of the line.
7271 A C string constant for text to be output before each @code{asm}
7272 statement or group of consecutive ones. Normally this is
7273 @code{"#APP"}, which is a comment that has no effect on most
7274 assemblers but tells the GNU assembler that it must check the lines
7275 that follow for all valid assembler constructs.
7279 A C string constant for text to be output after each @code{asm}
7280 statement or group of consecutive ones. Normally this is
7281 @code{"#NO_APP"}, which tells the GNU assembler to resume making the
7282 time-saving assumptions that are valid for ordinary compiler output.
7285 @defmac ASM_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name})
7286 A C statement to output COFF information or DWARF debugging information
7287 which indicates that filename @var{name} is the current source file to
7288 the stdio stream @var{stream}.
7290 This macro need not be defined if the standard form of output
7291 for the file format in use is appropriate.
7294 @hook TARGET_ASM_OUTPUT_SOURCE_FILENAME
7296 @defmac OUTPUT_QUOTED_STRING (@var{stream}, @var{string})
7297 A C statement to output the string @var{string} to the stdio stream
7298 @var{stream}. If you do not call the function @code{output_quoted_string}
7299 in your config files, GCC will only call it to output filenames to
7300 the assembler source. So you can use it to canonicalize the format
7301 of the filename using this macro.
7304 @defmac ASM_OUTPUT_IDENT (@var{stream}, @var{string})
7305 A C statement to output something to the assembler file to handle a
7306 @samp{#ident} directive containing the text @var{string}. If this
7307 macro is not defined, nothing is output for a @samp{#ident} directive.
7310 @hook TARGET_ASM_NAMED_SECTION
7311 Output assembly directives to switch to section @var{name}. The section
7312 should have attributes as specified by @var{flags}, which is a bit mask
7313 of the @code{SECTION_*} flags defined in @file{output.h}. If @var{decl}
7314 is non-NULL, it is the @code{VAR_DECL} or @code{FUNCTION_DECL} with which
7315 this section is associated.
7318 @hook TARGET_ASM_FUNCTION_SECTION
7319 Return preferred text (sub)section for function @var{decl}.
7320 Main purpose of this function is to separate cold, normal and hot
7321 functions. @var{startup} is true when function is known to be used only
7322 at startup (from static constructors or it is @code{main()}).
7323 @var{exit} is true when function is known to be used only at exit
7324 (from static destructors).
7325 Return NULL if function should go to default text section.
7328 @hook TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS
7330 @hook TARGET_HAVE_NAMED_SECTIONS
7331 This flag is true if the target supports @code{TARGET_ASM_NAMED_SECTION}.
7332 It must not be modified by command-line option processing.
7335 @anchor{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}
7336 @hook TARGET_HAVE_SWITCHABLE_BSS_SECTIONS
7337 This flag is true if we can create zeroed data by switching to a BSS
7338 section and then using @code{ASM_OUTPUT_SKIP} to allocate the space.
7339 This is true on most ELF targets.
7342 @hook TARGET_SECTION_TYPE_FLAGS
7343 Choose a set of section attributes for use by @code{TARGET_ASM_NAMED_SECTION}
7344 based on a variable or function decl, a section name, and whether or not the
7345 declaration's initializer may contain runtime relocations. @var{decl} may be
7346 null, in which case read-write data should be assumed.
7348 The default version of this function handles choosing code vs data,
7349 read-only vs read-write data, and @code{flag_pic}. You should only
7350 need to override this if your target has special flags that might be
7351 set via @code{__attribute__}.
7354 @hook TARGET_ASM_RECORD_GCC_SWITCHES
7355 Provides the target with the ability to record the gcc command line
7356 switches that have been passed to the compiler, and options that are
7357 enabled. The @var{type} argument specifies what is being recorded.
7358 It can take the following values:
7361 @item SWITCH_TYPE_PASSED
7362 @var{text} is a command line switch that has been set by the user.
7364 @item SWITCH_TYPE_ENABLED
7365 @var{text} is an option which has been enabled. This might be as a
7366 direct result of a command line switch, or because it is enabled by
7367 default or because it has been enabled as a side effect of a different
7368 command line switch. For example, the @option{-O2} switch enables
7369 various different individual optimization passes.
7371 @item SWITCH_TYPE_DESCRIPTIVE
7372 @var{text} is either NULL or some descriptive text which should be
7373 ignored. If @var{text} is NULL then it is being used to warn the
7374 target hook that either recording is starting or ending. The first
7375 time @var{type} is SWITCH_TYPE_DESCRIPTIVE and @var{text} is NULL, the
7376 warning is for start up and the second time the warning is for
7377 wind down. This feature is to allow the target hook to make any
7378 necessary preparations before it starts to record switches and to
7379 perform any necessary tidying up after it has finished recording
7382 @item SWITCH_TYPE_LINE_START
7383 This option can be ignored by this target hook.
7385 @item SWITCH_TYPE_LINE_END
7386 This option can be ignored by this target hook.
7389 The hook's return value must be zero. Other return values may be
7390 supported in the future.
7392 By default this hook is set to NULL, but an example implementation is
7393 provided for ELF based targets. Called @var{elf_record_gcc_switches},
7394 it records the switches as ASCII text inside a new, string mergeable
7395 section in the assembler output file. The name of the new section is
7396 provided by the @code{TARGET_ASM_RECORD_GCC_SWITCHES_SECTION} target
7400 @hook TARGET_ASM_RECORD_GCC_SWITCHES_SECTION
7401 This is the name of the section that will be created by the example
7402 ELF implementation of the @code{TARGET_ASM_RECORD_GCC_SWITCHES} target
7408 @subsection Output of Data
7411 @hook TARGET_ASM_BYTE_OP
7412 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_HI_OP
7413 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_SI_OP
7414 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_DI_OP
7415 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_TI_OP
7416 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_HI_OP
7417 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_SI_OP
7418 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_DI_OP
7419 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_TI_OP
7420 These hooks specify assembly directives for creating certain kinds
7421 of integer object. The @code{TARGET_ASM_BYTE_OP} directive creates a
7422 byte-sized object, the @code{TARGET_ASM_ALIGNED_HI_OP} one creates an
7423 aligned two-byte object, and so on. Any of the hooks may be
7424 @code{NULL}, indicating that no suitable directive is available.
7426 The compiler will print these strings at the start of a new line,
7427 followed immediately by the object's initial value. In most cases,
7428 the string should contain a tab, a pseudo-op, and then another tab.
7431 @hook TARGET_ASM_INTEGER
7432 The @code{assemble_integer} function uses this hook to output an
7433 integer object. @var{x} is the object's value, @var{size} is its size
7434 in bytes and @var{aligned_p} indicates whether it is aligned. The
7435 function should return @code{true} if it was able to output the
7436 object. If it returns false, @code{assemble_integer} will try to
7437 split the object into smaller parts.
7439 The default implementation of this hook will use the
7440 @code{TARGET_ASM_BYTE_OP} family of strings, returning @code{false}
7441 when the relevant string is @code{NULL}.
7444 @hook TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA
7445 A target hook to recognize @var{rtx} patterns that @code{output_addr_const}
7446 can't deal with, and output assembly code to @var{file} corresponding to
7447 the pattern @var{x}. This may be used to allow machine-dependent
7448 @code{UNSPEC}s to appear within constants.
7450 If target hook fails to recognize a pattern, it must return @code{false},
7451 so that a standard error message is printed. If it prints an error message
7452 itself, by calling, for example, @code{output_operand_lossage}, it may just
7456 @defmac OUTPUT_ADDR_CONST_EXTRA (@var{stream}, @var{x}, @var{fail})
7457 A C statement to recognize @var{rtx} patterns that
7458 @code{output_addr_const} can't deal with, and output assembly code to
7459 @var{stream} corresponding to the pattern @var{x}. This may be used to
7460 allow machine-dependent @code{UNSPEC}s to appear within constants.
7462 If @code{OUTPUT_ADDR_CONST_EXTRA} fails to recognize a pattern, it must
7463 @code{goto fail}, so that a standard error message is printed. If it
7464 prints an error message itself, by calling, for example,
7465 @code{output_operand_lossage}, it may just complete normally.
7468 @defmac ASM_OUTPUT_ASCII (@var{stream}, @var{ptr}, @var{len})
7469 A C statement to output to the stdio stream @var{stream} an assembler
7470 instruction to assemble a string constant containing the @var{len}
7471 bytes at @var{ptr}. @var{ptr} will be a C expression of type
7472 @code{char *} and @var{len} a C expression of type @code{int}.
7474 If the assembler has a @code{.ascii} pseudo-op as found in the
7475 Berkeley Unix assembler, do not define the macro
7476 @code{ASM_OUTPUT_ASCII}.
7479 @defmac ASM_OUTPUT_FDESC (@var{stream}, @var{decl}, @var{n})
7480 A C statement to output word @var{n} of a function descriptor for
7481 @var{decl}. This must be defined if @code{TARGET_VTABLE_USES_DESCRIPTORS}
7482 is defined, and is otherwise unused.
7485 @defmac CONSTANT_POOL_BEFORE_FUNCTION
7486 You may define this macro as a C expression. You should define the
7487 expression to have a nonzero value if GCC should output the constant
7488 pool for a function before the code for the function, or a zero value if
7489 GCC should output the constant pool after the function. If you do
7490 not define this macro, the usual case, GCC will output the constant
7491 pool before the function.
7494 @defmac ASM_OUTPUT_POOL_PROLOGUE (@var{file}, @var{funname}, @var{fundecl}, @var{size})
7495 A C statement to output assembler commands to define the start of the
7496 constant pool for a function. @var{funname} is a string giving
7497 the name of the function. Should the return type of the function
7498 be required, it can be obtained via @var{fundecl}. @var{size}
7499 is the size, in bytes, of the constant pool that will be written
7500 immediately after this call.
7502 If no constant-pool prefix is required, the usual case, this macro need
7506 @defmac ASM_OUTPUT_SPECIAL_POOL_ENTRY (@var{file}, @var{x}, @var{mode}, @var{align}, @var{labelno}, @var{jumpto})
7507 A C statement (with or without semicolon) to output a constant in the
7508 constant pool, if it needs special treatment. (This macro need not do
7509 anything for RTL expressions that can be output normally.)
7511 The argument @var{file} is the standard I/O stream to output the
7512 assembler code on. @var{x} is the RTL expression for the constant to
7513 output, and @var{mode} is the machine mode (in case @var{x} is a
7514 @samp{const_int}). @var{align} is the required alignment for the value
7515 @var{x}; you should output an assembler directive to force this much
7518 The argument @var{labelno} is a number to use in an internal label for
7519 the address of this pool entry. The definition of this macro is
7520 responsible for outputting the label definition at the proper place.
7521 Here is how to do this:
7524 @code{(*targetm.asm_out.internal_label)} (@var{file}, "LC", @var{labelno});
7527 When you output a pool entry specially, you should end with a
7528 @code{goto} to the label @var{jumpto}. This will prevent the same pool
7529 entry from being output a second time in the usual manner.
7531 You need not define this macro if it would do nothing.
7534 @defmac ASM_OUTPUT_POOL_EPILOGUE (@var{file} @var{funname} @var{fundecl} @var{size})
7535 A C statement to output assembler commands to at the end of the constant
7536 pool for a function. @var{funname} is a string giving the name of the
7537 function. Should the return type of the function be required, you can
7538 obtain it via @var{fundecl}. @var{size} is the size, in bytes, of the
7539 constant pool that GCC wrote immediately before this call.
7541 If no constant-pool epilogue is required, the usual case, you need not
7545 @defmac IS_ASM_LOGICAL_LINE_SEPARATOR (@var{C}, @var{STR})
7546 Define this macro as a C expression which is nonzero if @var{C} is
7547 used as a logical line separator by the assembler. @var{STR} points
7548 to the position in the string where @var{C} was found; this can be used if
7549 a line separator uses multiple characters.
7551 If you do not define this macro, the default is that only
7552 the character @samp{;} is treated as a logical line separator.
7555 @hook TARGET_ASM_OPEN_PAREN
7556 These target hooks are C string constants, describing the syntax in the
7557 assembler for grouping arithmetic expressions. If not overridden, they
7558 default to normal parentheses, which is correct for most assemblers.
7561 These macros are provided by @file{real.h} for writing the definitions
7562 of @code{ASM_OUTPUT_DOUBLE} and the like:
7564 @defmac REAL_VALUE_TO_TARGET_SINGLE (@var{x}, @var{l})
7565 @defmacx REAL_VALUE_TO_TARGET_DOUBLE (@var{x}, @var{l})
7566 @defmacx REAL_VALUE_TO_TARGET_LONG_DOUBLE (@var{x}, @var{l})
7567 @defmacx REAL_VALUE_TO_TARGET_DECIMAL32 (@var{x}, @var{l})
7568 @defmacx REAL_VALUE_TO_TARGET_DECIMAL64 (@var{x}, @var{l})
7569 @defmacx REAL_VALUE_TO_TARGET_DECIMAL128 (@var{x}, @var{l})
7570 These translate @var{x}, of type @code{REAL_VALUE_TYPE}, to the
7571 target's floating point representation, and store its bit pattern in
7572 the variable @var{l}. For @code{REAL_VALUE_TO_TARGET_SINGLE} and
7573 @code{REAL_VALUE_TO_TARGET_DECIMAL32}, this variable should be a
7574 simple @code{long int}. For the others, it should be an array of
7575 @code{long int}. The number of elements in this array is determined
7576 by the size of the desired target floating point data type: 32 bits of
7577 it go in each @code{long int} array element. Each array element holds
7578 32 bits of the result, even if @code{long int} is wider than 32 bits
7579 on the host machine.
7581 The array element values are designed so that you can print them out
7582 using @code{fprintf} in the order they should appear in the target
7586 @node Uninitialized Data
7587 @subsection Output of Uninitialized Variables
7589 Each of the macros in this section is used to do the whole job of
7590 outputting a single uninitialized variable.
7592 @defmac ASM_OUTPUT_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
7593 A C statement (sans semicolon) to output to the stdio stream
7594 @var{stream} the assembler definition of a common-label named
7595 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7596 is the size rounded up to whatever alignment the caller wants. It is
7597 possible that @var{size} may be zero, for instance if a struct with no
7598 other member than a zero-length array is defined. In this case, the
7599 backend must output a symbol definition that allocates at least one
7600 byte, both so that the address of the resulting object does not compare
7601 equal to any other, and because some object formats cannot even express
7602 the concept of a zero-sized common symbol, as that is how they represent
7603 an ordinary undefined external.
7605 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7606 output the name itself; before and after that, output the additional
7607 assembler syntax for defining the name, and a newline.
7609 This macro controls how the assembler definitions of uninitialized
7610 common global variables are output.
7613 @defmac ASM_OUTPUT_ALIGNED_COMMON (@var{stream}, @var{name}, @var{size}, @var{alignment})
7614 Like @code{ASM_OUTPUT_COMMON} except takes the required alignment as a
7615 separate, explicit argument. If you define this macro, it is used in
7616 place of @code{ASM_OUTPUT_COMMON}, and gives you more flexibility in
7617 handling the required alignment of the variable. The alignment is specified
7618 as the number of bits.
7621 @defmac ASM_OUTPUT_ALIGNED_DECL_COMMON (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7622 Like @code{ASM_OUTPUT_ALIGNED_COMMON} except that @var{decl} of the
7623 variable to be output, if there is one, or @code{NULL_TREE} if there
7624 is no corresponding variable. If you define this macro, GCC will use it
7625 in place of both @code{ASM_OUTPUT_COMMON} and
7626 @code{ASM_OUTPUT_ALIGNED_COMMON}. Define this macro when you need to see
7627 the variable's decl in order to chose what to output.
7630 @defmac ASM_OUTPUT_ALIGNED_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7631 A C statement (sans semicolon) to output to the stdio stream
7632 @var{stream} the assembler definition of uninitialized global @var{decl} named
7633 @var{name} whose size is @var{size} bytes. The variable @var{alignment}
7634 is the alignment specified as the number of bits.
7636 Try to use function @code{asm_output_aligned_bss} defined in file
7637 @file{varasm.c} when defining this macro. If unable, use the expression
7638 @code{assemble_name (@var{stream}, @var{name})} to output the name itself;
7639 before and after that, output the additional assembler syntax for defining
7640 the name, and a newline.
7642 There are two ways of handling global BSS@. One is to define this macro.
7643 The other is to have @code{TARGET_ASM_SELECT_SECTION} return a
7644 switchable BSS section (@pxref{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}).
7645 You do not need to do both.
7647 Some languages do not have @code{common} data, and require a
7648 non-common form of global BSS in order to handle uninitialized globals
7649 efficiently. C++ is one example of this. However, if the target does
7650 not support global BSS, the front end may choose to make globals
7651 common in order to save space in the object file.
7654 @defmac ASM_OUTPUT_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded})
7655 A C statement (sans semicolon) to output to the stdio stream
7656 @var{stream} the assembler definition of a local-common-label named
7657 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7658 is the size rounded up to whatever alignment the caller wants.
7660 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7661 output the name itself; before and after that, output the additional
7662 assembler syntax for defining the name, and a newline.
7664 This macro controls how the assembler definitions of uninitialized
7665 static variables are output.
7668 @defmac ASM_OUTPUT_ALIGNED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{alignment})
7669 Like @code{ASM_OUTPUT_LOCAL} except takes the required alignment as a
7670 separate, explicit argument. If you define this macro, it is used in
7671 place of @code{ASM_OUTPUT_LOCAL}, and gives you more flexibility in
7672 handling the required alignment of the variable. The alignment is specified
7673 as the number of bits.
7676 @defmac ASM_OUTPUT_ALIGNED_DECL_LOCAL (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7677 Like @code{ASM_OUTPUT_ALIGNED_DECL} except that @var{decl} of the
7678 variable to be output, if there is one, or @code{NULL_TREE} if there
7679 is no corresponding variable. If you define this macro, GCC will use it
7680 in place of both @code{ASM_OUTPUT_DECL} and
7681 @code{ASM_OUTPUT_ALIGNED_DECL}. Define this macro when you need to see
7682 the variable's decl in order to chose what to output.
7686 @subsection Output and Generation of Labels
7688 @c prevent bad page break with this line
7689 This is about outputting labels.
7691 @findex assemble_name
7692 @defmac ASM_OUTPUT_LABEL (@var{stream}, @var{name})
7693 A C statement (sans semicolon) to output to the stdio stream
7694 @var{stream} the assembler definition of a label named @var{name}.
7695 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7696 output the name itself; before and after that, output the additional
7697 assembler syntax for defining the name, and a newline. A default
7698 definition of this macro is provided which is correct for most systems.
7701 @defmac ASM_OUTPUT_FUNCTION_LABEL (@var{stream}, @var{name}, @var{decl})
7702 A C statement (sans semicolon) to output to the stdio stream
7703 @var{stream} the assembler definition of a label named @var{name} of
7705 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7706 output the name itself; before and after that, output the additional
7707 assembler syntax for defining the name, and a newline. A default
7708 definition of this macro is provided which is correct for most systems.
7710 If this macro is not defined, then the function name is defined in the
7711 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
7714 @findex assemble_name_raw
7715 @defmac ASM_OUTPUT_INTERNAL_LABEL (@var{stream}, @var{name})
7716 Identical to @code{ASM_OUTPUT_LABEL}, except that @var{name} is known
7717 to refer to a compiler-generated label. The default definition uses
7718 @code{assemble_name_raw}, which is like @code{assemble_name} except
7719 that it is more efficient.
7723 A C string containing the appropriate assembler directive to specify the
7724 size of a symbol, without any arguments. On systems that use ELF, the
7725 default (in @file{config/elfos.h}) is @samp{"\t.size\t"}; on other
7726 systems, the default is not to define this macro.
7728 Define this macro only if it is correct to use the default definitions
7729 of @code{ASM_OUTPUT_SIZE_DIRECTIVE} and @code{ASM_OUTPUT_MEASURED_SIZE}
7730 for your system. If you need your own custom definitions of those
7731 macros, or if you do not need explicit symbol sizes at all, do not
7735 @defmac ASM_OUTPUT_SIZE_DIRECTIVE (@var{stream}, @var{name}, @var{size})
7736 A C statement (sans semicolon) to output to the stdio stream
7737 @var{stream} a directive telling the assembler that the size of the
7738 symbol @var{name} is @var{size}. @var{size} is a @code{HOST_WIDE_INT}.
7739 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
7743 @defmac ASM_OUTPUT_MEASURED_SIZE (@var{stream}, @var{name})
7744 A C statement (sans semicolon) to output to the stdio stream
7745 @var{stream} a directive telling the assembler to calculate the size of
7746 the symbol @var{name} by subtracting its address from the current
7749 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
7750 provided. The default assumes that the assembler recognizes a special
7751 @samp{.} symbol as referring to the current address, and can calculate
7752 the difference between this and another symbol. If your assembler does
7753 not recognize @samp{.} or cannot do calculations with it, you will need
7754 to redefine @code{ASM_OUTPUT_MEASURED_SIZE} to use some other technique.
7758 A C string containing the appropriate assembler directive to specify the
7759 type of a symbol, without any arguments. On systems that use ELF, the
7760 default (in @file{config/elfos.h}) is @samp{"\t.type\t"}; on other
7761 systems, the default is not to define this macro.
7763 Define this macro only if it is correct to use the default definition of
7764 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
7765 custom definition of this macro, or if you do not need explicit symbol
7766 types at all, do not define this macro.
7769 @defmac TYPE_OPERAND_FMT
7770 A C string which specifies (using @code{printf} syntax) the format of
7771 the second operand to @code{TYPE_ASM_OP}. On systems that use ELF, the
7772 default (in @file{config/elfos.h}) is @samp{"@@%s"}; on other systems,
7773 the default is not to define this macro.
7775 Define this macro only if it is correct to use the default definition of
7776 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
7777 custom definition of this macro, or if you do not need explicit symbol
7778 types at all, do not define this macro.
7781 @defmac ASM_OUTPUT_TYPE_DIRECTIVE (@var{stream}, @var{type})
7782 A C statement (sans semicolon) to output to the stdio stream
7783 @var{stream} a directive telling the assembler that the type of the
7784 symbol @var{name} is @var{type}. @var{type} is a C string; currently,
7785 that string is always either @samp{"function"} or @samp{"object"}, but
7786 you should not count on this.
7788 If you define @code{TYPE_ASM_OP} and @code{TYPE_OPERAND_FMT}, a default
7789 definition of this macro is provided.
7792 @defmac ASM_DECLARE_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl})
7793 A C statement (sans semicolon) to output to the stdio stream
7794 @var{stream} any text necessary for declaring the name @var{name} of a
7795 function which is being defined. This macro is responsible for
7796 outputting the label definition (perhaps using
7797 @code{ASM_OUTPUT_FUNCTION_LABEL}). The argument @var{decl} is the
7798 @code{FUNCTION_DECL} tree node representing the function.
7800 If this macro is not defined, then the function name is defined in the
7801 usual manner as a label (by means of @code{ASM_OUTPUT_FUNCTION_LABEL}).
7803 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition
7807 @defmac ASM_DECLARE_FUNCTION_SIZE (@var{stream}, @var{name}, @var{decl})
7808 A C statement (sans semicolon) to output to the stdio stream
7809 @var{stream} any text necessary for declaring the size of a function
7810 which is being defined. The argument @var{name} is the name of the
7811 function. The argument @var{decl} is the @code{FUNCTION_DECL} tree node
7812 representing the function.
7814 If this macro is not defined, then the function size is not defined.
7816 You may wish to use @code{ASM_OUTPUT_MEASURED_SIZE} in the definition
7820 @defmac ASM_DECLARE_OBJECT_NAME (@var{stream}, @var{name}, @var{decl})
7821 A C statement (sans semicolon) to output to the stdio stream
7822 @var{stream} any text necessary for declaring the name @var{name} of an
7823 initialized variable which is being defined. This macro must output the
7824 label definition (perhaps using @code{ASM_OUTPUT_LABEL}). The argument
7825 @var{decl} is the @code{VAR_DECL} tree node representing the variable.
7827 If this macro is not defined, then the variable name is defined in the
7828 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
7830 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} and/or
7831 @code{ASM_OUTPUT_SIZE_DIRECTIVE} in the definition of this macro.
7834 @hook TARGET_ASM_DECLARE_CONSTANT_NAME
7835 A target hook to output to the stdio stream @var{file} any text necessary
7836 for declaring the name @var{name} of a constant which is being defined. This
7837 target hook is responsible for outputting the label definition (perhaps using
7838 @code{assemble_label}). The argument @var{exp} is the value of the constant,
7839 and @var{size} is the size of the constant in bytes. The @var{name}
7840 will be an internal label.
7842 The default version of this target hook, define the @var{name} in the
7843 usual manner as a label (by means of @code{assemble_label}).
7845 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in this target hook.
7848 @defmac ASM_DECLARE_REGISTER_GLOBAL (@var{stream}, @var{decl}, @var{regno}, @var{name})
7849 A C statement (sans semicolon) to output to the stdio stream
7850 @var{stream} any text necessary for claiming a register @var{regno}
7851 for a global variable @var{decl} with name @var{name}.
7853 If you don't define this macro, that is equivalent to defining it to do
7857 @defmac ASM_FINISH_DECLARE_OBJECT (@var{stream}, @var{decl}, @var{toplevel}, @var{atend})
7858 A C statement (sans semicolon) to finish up declaring a variable name
7859 once the compiler has processed its initializer fully and thus has had a
7860 chance to determine the size of an array when controlled by an
7861 initializer. This is used on systems where it's necessary to declare
7862 something about the size of the object.
7864 If you don't define this macro, that is equivalent to defining it to do
7867 You may wish to use @code{ASM_OUTPUT_SIZE_DIRECTIVE} and/or
7868 @code{ASM_OUTPUT_MEASURED_SIZE} in the definition of this macro.
7871 @hook TARGET_ASM_GLOBALIZE_LABEL
7872 This target hook is a function to output to the stdio stream
7873 @var{stream} some commands that will make the label @var{name} global;
7874 that is, available for reference from other files.
7876 The default implementation relies on a proper definition of
7877 @code{GLOBAL_ASM_OP}.
7880 @hook TARGET_ASM_GLOBALIZE_DECL_NAME
7881 This target hook is a function to output to the stdio stream
7882 @var{stream} some commands that will make the name associated with @var{decl}
7883 global; that is, available for reference from other files.
7885 The default implementation uses the TARGET_ASM_GLOBALIZE_LABEL target hook.
7888 @defmac ASM_WEAKEN_LABEL (@var{stream}, @var{name})
7889 A C statement (sans semicolon) to output to the stdio stream
7890 @var{stream} some commands that will make the label @var{name} weak;
7891 that is, available for reference from other files but only used if
7892 no other definition is available. Use the expression
7893 @code{assemble_name (@var{stream}, @var{name})} to output the name
7894 itself; before and after that, output the additional assembler syntax
7895 for making that name weak, and a newline.
7897 If you don't define this macro or @code{ASM_WEAKEN_DECL}, GCC will not
7898 support weak symbols and you should not define the @code{SUPPORTS_WEAK}
7902 @defmac ASM_WEAKEN_DECL (@var{stream}, @var{decl}, @var{name}, @var{value})
7903 Combines (and replaces) the function of @code{ASM_WEAKEN_LABEL} and
7904 @code{ASM_OUTPUT_WEAK_ALIAS}, allowing access to the associated function
7905 or variable decl. If @var{value} is not @code{NULL}, this C statement
7906 should output to the stdio stream @var{stream} assembler code which
7907 defines (equates) the weak symbol @var{name} to have the value
7908 @var{value}. If @var{value} is @code{NULL}, it should output commands
7909 to make @var{name} weak.
7912 @defmac ASM_OUTPUT_WEAKREF (@var{stream}, @var{decl}, @var{name}, @var{value})
7913 Outputs a directive that enables @var{name} to be used to refer to
7914 symbol @var{value} with weak-symbol semantics. @code{decl} is the
7915 declaration of @code{name}.
7918 @defmac SUPPORTS_WEAK
7919 A preprocessor constant expression which evaluates to true if the target
7920 supports weak symbols.
7922 If you don't define this macro, @file{defaults.h} provides a default
7923 definition. If either @code{ASM_WEAKEN_LABEL} or @code{ASM_WEAKEN_DECL}
7924 is defined, the default definition is @samp{1}; otherwise, it is @samp{0}.
7927 @defmac TARGET_SUPPORTS_WEAK
7928 A C expression which evaluates to true if the target supports weak symbols.
7930 If you don't define this macro, @file{defaults.h} provides a default
7931 definition. The default definition is @samp{(SUPPORTS_WEAK)}. Define
7932 this macro if you want to control weak symbol support with a compiler
7933 flag such as @option{-melf}.
7936 @defmac MAKE_DECL_ONE_ONLY (@var{decl})
7937 A C statement (sans semicolon) to mark @var{decl} to be emitted as a
7938 public symbol such that extra copies in multiple translation units will
7939 be discarded by the linker. Define this macro if your object file
7940 format provides support for this concept, such as the @samp{COMDAT}
7941 section flags in the Microsoft Windows PE/COFF format, and this support
7942 requires changes to @var{decl}, such as putting it in a separate section.
7945 @defmac SUPPORTS_ONE_ONLY
7946 A C expression which evaluates to true if the target supports one-only
7949 If you don't define this macro, @file{varasm.c} provides a default
7950 definition. If @code{MAKE_DECL_ONE_ONLY} is defined, the default
7951 definition is @samp{1}; otherwise, it is @samp{0}. Define this macro if
7952 you want to control one-only symbol support with a compiler flag, or if
7953 setting the @code{DECL_ONE_ONLY} flag is enough to mark a declaration to
7954 be emitted as one-only.
7957 @hook TARGET_ASM_ASSEMBLE_VISIBILITY
7958 This target hook is a function to output to @var{asm_out_file} some
7959 commands that will make the symbol(s) associated with @var{decl} have
7960 hidden, protected or internal visibility as specified by @var{visibility}.
7963 @defmac TARGET_WEAK_NOT_IN_ARCHIVE_TOC
7964 A C expression that evaluates to true if the target's linker expects
7965 that weak symbols do not appear in a static archive's table of contents.
7966 The default is @code{0}.
7968 Leaving weak symbols out of an archive's table of contents means that,
7969 if a symbol will only have a definition in one translation unit and
7970 will have undefined references from other translation units, that
7971 symbol should not be weak. Defining this macro to be nonzero will
7972 thus have the effect that certain symbols that would normally be weak
7973 (explicit template instantiations, and vtables for polymorphic classes
7974 with noninline key methods) will instead be nonweak.
7976 The C++ ABI requires this macro to be zero. Define this macro for
7977 targets where full C++ ABI compliance is impossible and where linker
7978 restrictions require weak symbols to be left out of a static archive's
7982 @defmac ASM_OUTPUT_EXTERNAL (@var{stream}, @var{decl}, @var{name})
7983 A C statement (sans semicolon) to output to the stdio stream
7984 @var{stream} any text necessary for declaring the name of an external
7985 symbol named @var{name} which is referenced in this compilation but
7986 not defined. The value of @var{decl} is the tree node for the
7989 This macro need not be defined if it does not need to output anything.
7990 The GNU assembler and most Unix assemblers don't require anything.
7993 @hook TARGET_ASM_EXTERNAL_LIBCALL
7994 This target hook is a function to output to @var{asm_out_file} an assembler
7995 pseudo-op to declare a library function name external. The name of the
7996 library function is given by @var{symref}, which is a @code{symbol_ref}.
7999 @hook TARGET_ASM_MARK_DECL_PRESERVED
8000 This target hook is a function to output to @var{asm_out_file} an assembler
8001 directive to annotate @var{symbol} as used. The Darwin target uses the
8002 .no_dead_code_strip directive.
8005 @defmac ASM_OUTPUT_LABELREF (@var{stream}, @var{name})
8006 A C statement (sans semicolon) to output to the stdio stream
8007 @var{stream} a reference in assembler syntax to a label named
8008 @var{name}. This should add @samp{_} to the front of the name, if that
8009 is customary on your operating system, as it is in most Berkeley Unix
8010 systems. This macro is used in @code{assemble_name}.
8013 @hook TARGET_MANGLE_ASSEMBLER_NAME
8015 @defmac ASM_OUTPUT_SYMBOL_REF (@var{stream}, @var{sym})
8016 A C statement (sans semicolon) to output a reference to
8017 @code{SYMBOL_REF} @var{sym}. If not defined, @code{assemble_name}
8018 will be used to output the name of the symbol. This macro may be used
8019 to modify the way a symbol is referenced depending on information
8020 encoded by @code{TARGET_ENCODE_SECTION_INFO}.
8023 @defmac ASM_OUTPUT_LABEL_REF (@var{stream}, @var{buf})
8024 A C statement (sans semicolon) to output a reference to @var{buf}, the
8025 result of @code{ASM_GENERATE_INTERNAL_LABEL}. If not defined,
8026 @code{assemble_name} will be used to output the name of the symbol.
8027 This macro is not used by @code{output_asm_label}, or the @code{%l}
8028 specifier that calls it; the intention is that this macro should be set
8029 when it is necessary to output a label differently when its address is
8033 @hook TARGET_ASM_INTERNAL_LABEL
8034 A function to output to the stdio stream @var{stream} a label whose
8035 name is made from the string @var{prefix} and the number @var{labelno}.
8037 It is absolutely essential that these labels be distinct from the labels
8038 used for user-level functions and variables. Otherwise, certain programs
8039 will have name conflicts with internal labels.
8041 It is desirable to exclude internal labels from the symbol table of the
8042 object file. Most assemblers have a naming convention for labels that
8043 should be excluded; on many systems, the letter @samp{L} at the
8044 beginning of a label has this effect. You should find out what
8045 convention your system uses, and follow it.
8047 The default version of this function utilizes @code{ASM_GENERATE_INTERNAL_LABEL}.
8050 @defmac ASM_OUTPUT_DEBUG_LABEL (@var{stream}, @var{prefix}, @var{num})
8051 A C statement to output to the stdio stream @var{stream} a debug info
8052 label whose name is made from the string @var{prefix} and the number
8053 @var{num}. This is useful for VLIW targets, where debug info labels
8054 may need to be treated differently than branch target labels. On some
8055 systems, branch target labels must be at the beginning of instruction
8056 bundles, but debug info labels can occur in the middle of instruction
8059 If this macro is not defined, then @code{(*targetm.asm_out.internal_label)} will be
8063 @defmac ASM_GENERATE_INTERNAL_LABEL (@var{string}, @var{prefix}, @var{num})
8064 A C statement to store into the string @var{string} a label whose name
8065 is made from the string @var{prefix} and the number @var{num}.
8067 This string, when output subsequently by @code{assemble_name}, should
8068 produce the output that @code{(*targetm.asm_out.internal_label)} would produce
8069 with the same @var{prefix} and @var{num}.
8071 If the string begins with @samp{*}, then @code{assemble_name} will
8072 output the rest of the string unchanged. It is often convenient for
8073 @code{ASM_GENERATE_INTERNAL_LABEL} to use @samp{*} in this way. If the
8074 string doesn't start with @samp{*}, then @code{ASM_OUTPUT_LABELREF} gets
8075 to output the string, and may change it. (Of course,
8076 @code{ASM_OUTPUT_LABELREF} is also part of your machine description, so
8077 you should know what it does on your machine.)
8080 @defmac ASM_FORMAT_PRIVATE_NAME (@var{outvar}, @var{name}, @var{number})
8081 A C expression to assign to @var{outvar} (which is a variable of type
8082 @code{char *}) a newly allocated string made from the string
8083 @var{name} and the number @var{number}, with some suitable punctuation
8084 added. Use @code{alloca} to get space for the string.
8086 The string will be used as an argument to @code{ASM_OUTPUT_LABELREF} to
8087 produce an assembler label for an internal static variable whose name is
8088 @var{name}. Therefore, the string must be such as to result in valid
8089 assembler code. The argument @var{number} is different each time this
8090 macro is executed; it prevents conflicts between similarly-named
8091 internal static variables in different scopes.
8093 Ideally this string should not be a valid C identifier, to prevent any
8094 conflict with the user's own symbols. Most assemblers allow periods
8095 or percent signs in assembler symbols; putting at least one of these
8096 between the name and the number will suffice.
8098 If this macro is not defined, a default definition will be provided
8099 which is correct for most systems.
8102 @defmac ASM_OUTPUT_DEF (@var{stream}, @var{name}, @var{value})
8103 A C statement to output to the stdio stream @var{stream} assembler code
8104 which defines (equates) the symbol @var{name} to have the value @var{value}.
8107 If @code{SET_ASM_OP} is defined, a default definition is provided which is
8108 correct for most systems.
8111 @defmac ASM_OUTPUT_DEF_FROM_DECLS (@var{stream}, @var{decl_of_name}, @var{decl_of_value})
8112 A C statement to output to the stdio stream @var{stream} assembler code
8113 which defines (equates) the symbol whose tree node is @var{decl_of_name}
8114 to have the value of the tree node @var{decl_of_value}. This macro will
8115 be used in preference to @samp{ASM_OUTPUT_DEF} if it is defined and if
8116 the tree nodes are available.
8119 If @code{SET_ASM_OP} is defined, a default definition is provided which is
8120 correct for most systems.
8123 @defmac TARGET_DEFERRED_OUTPUT_DEFS (@var{decl_of_name}, @var{decl_of_value})
8124 A C statement that evaluates to true if the assembler code which defines
8125 (equates) the symbol whose tree node is @var{decl_of_name} to have the value
8126 of the tree node @var{decl_of_value} should be emitted near the end of the
8127 current compilation unit. The default is to not defer output of defines.
8128 This macro affects defines output by @samp{ASM_OUTPUT_DEF} and
8129 @samp{ASM_OUTPUT_DEF_FROM_DECLS}.
8132 @defmac ASM_OUTPUT_WEAK_ALIAS (@var{stream}, @var{name}, @var{value})
8133 A C statement to output to the stdio stream @var{stream} assembler code
8134 which defines (equates) the weak symbol @var{name} to have the value
8135 @var{value}. If @var{value} is @code{NULL}, it defines @var{name} as
8136 an undefined weak symbol.
8138 Define this macro if the target only supports weak aliases; define
8139 @code{ASM_OUTPUT_DEF} instead if possible.
8142 @defmac OBJC_GEN_METHOD_LABEL (@var{buf}, @var{is_inst}, @var{class_name}, @var{cat_name}, @var{sel_name})
8143 Define this macro to override the default assembler names used for
8144 Objective-C methods.
8146 The default name is a unique method number followed by the name of the
8147 class (e.g.@: @samp{_1_Foo}). For methods in categories, the name of
8148 the category is also included in the assembler name (e.g.@:
8151 These names are safe on most systems, but make debugging difficult since
8152 the method's selector is not present in the name. Therefore, particular
8153 systems define other ways of computing names.
8155 @var{buf} is an expression of type @code{char *} which gives you a
8156 buffer in which to store the name; its length is as long as
8157 @var{class_name}, @var{cat_name} and @var{sel_name} put together, plus
8158 50 characters extra.
8160 The argument @var{is_inst} specifies whether the method is an instance
8161 method or a class method; @var{class_name} is the name of the class;
8162 @var{cat_name} is the name of the category (or @code{NULL} if the method is not
8163 in a category); and @var{sel_name} is the name of the selector.
8165 On systems where the assembler can handle quoted names, you can use this
8166 macro to provide more human-readable names.
8169 @defmac ASM_DECLARE_CLASS_REFERENCE (@var{stream}, @var{name})
8170 A C statement (sans semicolon) to output to the stdio stream
8171 @var{stream} commands to declare that the label @var{name} is an
8172 Objective-C class reference. This is only needed for targets whose
8173 linkers have special support for NeXT-style runtimes.
8176 @defmac ASM_DECLARE_UNRESOLVED_REFERENCE (@var{stream}, @var{name})
8177 A C statement (sans semicolon) to output to the stdio stream
8178 @var{stream} commands to declare that the label @var{name} is an
8179 unresolved Objective-C class reference. This is only needed for targets
8180 whose linkers have special support for NeXT-style runtimes.
8183 @node Initialization
8184 @subsection How Initialization Functions Are Handled
8185 @cindex initialization routines
8186 @cindex termination routines
8187 @cindex constructors, output of
8188 @cindex destructors, output of
8190 The compiled code for certain languages includes @dfn{constructors}
8191 (also called @dfn{initialization routines})---functions to initialize
8192 data in the program when the program is started. These functions need
8193 to be called before the program is ``started''---that is to say, before
8194 @code{main} is called.
8196 Compiling some languages generates @dfn{destructors} (also called
8197 @dfn{termination routines}) that should be called when the program
8200 To make the initialization and termination functions work, the compiler
8201 must output something in the assembler code to cause those functions to
8202 be called at the appropriate time. When you port the compiler to a new
8203 system, you need to specify how to do this.
8205 There are two major ways that GCC currently supports the execution of
8206 initialization and termination functions. Each way has two variants.
8207 Much of the structure is common to all four variations.
8209 @findex __CTOR_LIST__
8210 @findex __DTOR_LIST__
8211 The linker must build two lists of these functions---a list of
8212 initialization functions, called @code{__CTOR_LIST__}, and a list of
8213 termination functions, called @code{__DTOR_LIST__}.
8215 Each list always begins with an ignored function pointer (which may hold
8216 0, @minus{}1, or a count of the function pointers after it, depending on
8217 the environment). This is followed by a series of zero or more function
8218 pointers to constructors (or destructors), followed by a function
8219 pointer containing zero.
8221 Depending on the operating system and its executable file format, either
8222 @file{crtstuff.c} or @file{libgcc2.c} traverses these lists at startup
8223 time and exit time. Constructors are called in reverse order of the
8224 list; destructors in forward order.
8226 The best way to handle static constructors works only for object file
8227 formats which provide arbitrarily-named sections. A section is set
8228 aside for a list of constructors, and another for a list of destructors.
8229 Traditionally these are called @samp{.ctors} and @samp{.dtors}. Each
8230 object file that defines an initialization function also puts a word in
8231 the constructor section to point to that function. The linker
8232 accumulates all these words into one contiguous @samp{.ctors} section.
8233 Termination functions are handled similarly.
8235 This method will be chosen as the default by @file{target-def.h} if
8236 @code{TARGET_ASM_NAMED_SECTION} is defined. A target that does not
8237 support arbitrary sections, but does support special designated
8238 constructor and destructor sections may define @code{CTORS_SECTION_ASM_OP}
8239 and @code{DTORS_SECTION_ASM_OP} to achieve the same effect.
8241 When arbitrary sections are available, there are two variants, depending
8242 upon how the code in @file{crtstuff.c} is called. On systems that
8243 support a @dfn{.init} section which is executed at program startup,
8244 parts of @file{crtstuff.c} are compiled into that section. The
8245 program is linked by the @command{gcc} driver like this:
8248 ld -o @var{output_file} crti.o crtbegin.o @dots{} -lgcc crtend.o crtn.o
8251 The prologue of a function (@code{__init}) appears in the @code{.init}
8252 section of @file{crti.o}; the epilogue appears in @file{crtn.o}. Likewise
8253 for the function @code{__fini} in the @dfn{.fini} section. Normally these
8254 files are provided by the operating system or by the GNU C library, but
8255 are provided by GCC for a few targets.
8257 The objects @file{crtbegin.o} and @file{crtend.o} are (for most targets)
8258 compiled from @file{crtstuff.c}. They contain, among other things, code
8259 fragments within the @code{.init} and @code{.fini} sections that branch
8260 to routines in the @code{.text} section. The linker will pull all parts
8261 of a section together, which results in a complete @code{__init} function
8262 that invokes the routines we need at startup.
8264 To use this variant, you must define the @code{INIT_SECTION_ASM_OP}
8267 If no init section is available, when GCC compiles any function called
8268 @code{main} (or more accurately, any function designated as a program
8269 entry point by the language front end calling @code{expand_main_function}),
8270 it inserts a procedure call to @code{__main} as the first executable code
8271 after the function prologue. The @code{__main} function is defined
8272 in @file{libgcc2.c} and runs the global constructors.
8274 In file formats that don't support arbitrary sections, there are again
8275 two variants. In the simplest variant, the GNU linker (GNU @code{ld})
8276 and an `a.out' format must be used. In this case,
8277 @code{TARGET_ASM_CONSTRUCTOR} is defined to produce a @code{.stabs}
8278 entry of type @samp{N_SETT}, referencing the name @code{__CTOR_LIST__},
8279 and with the address of the void function containing the initialization
8280 code as its value. The GNU linker recognizes this as a request to add
8281 the value to a @dfn{set}; the values are accumulated, and are eventually
8282 placed in the executable as a vector in the format described above, with
8283 a leading (ignored) count and a trailing zero element.
8284 @code{TARGET_ASM_DESTRUCTOR} is handled similarly. Since no init
8285 section is available, the absence of @code{INIT_SECTION_ASM_OP} causes
8286 the compilation of @code{main} to call @code{__main} as above, starting
8287 the initialization process.
8289 The last variant uses neither arbitrary sections nor the GNU linker.
8290 This is preferable when you want to do dynamic linking and when using
8291 file formats which the GNU linker does not support, such as `ECOFF'@. In
8292 this case, @code{TARGET_HAVE_CTORS_DTORS} is false, initialization and
8293 termination functions are recognized simply by their names. This requires
8294 an extra program in the linkage step, called @command{collect2}. This program
8295 pretends to be the linker, for use with GCC; it does its job by running
8296 the ordinary linker, but also arranges to include the vectors of
8297 initialization and termination functions. These functions are called
8298 via @code{__main} as described above. In order to use this method,
8299 @code{use_collect2} must be defined in the target in @file{config.gcc}.
8302 The following section describes the specific macros that control and
8303 customize the handling of initialization and termination functions.
8306 @node Macros for Initialization
8307 @subsection Macros Controlling Initialization Routines
8309 Here are the macros that control how the compiler handles initialization
8310 and termination functions:
8312 @defmac INIT_SECTION_ASM_OP
8313 If defined, a C string constant, including spacing, for the assembler
8314 operation to identify the following data as initialization code. If not
8315 defined, GCC will assume such a section does not exist. When you are
8316 using special sections for initialization and termination functions, this
8317 macro also controls how @file{crtstuff.c} and @file{libgcc2.c} arrange to
8318 run the initialization functions.
8321 @defmac HAS_INIT_SECTION
8322 If defined, @code{main} will not call @code{__main} as described above.
8323 This macro should be defined for systems that control start-up code
8324 on a symbol-by-symbol basis, such as OSF/1, and should not
8325 be defined explicitly for systems that support @code{INIT_SECTION_ASM_OP}.
8328 @defmac LD_INIT_SWITCH
8329 If defined, a C string constant for a switch that tells the linker that
8330 the following symbol is an initialization routine.
8333 @defmac LD_FINI_SWITCH
8334 If defined, a C string constant for a switch that tells the linker that
8335 the following symbol is a finalization routine.
8338 @defmac COLLECT_SHARED_INIT_FUNC (@var{stream}, @var{func})
8339 If defined, a C statement that will write a function that can be
8340 automatically called when a shared library is loaded. The function
8341 should call @var{func}, which takes no arguments. If not defined, and
8342 the object format requires an explicit initialization function, then a
8343 function called @code{_GLOBAL__DI} will be generated.
8345 This function and the following one are used by collect2 when linking a
8346 shared library that needs constructors or destructors, or has DWARF2
8347 exception tables embedded in the code.
8350 @defmac COLLECT_SHARED_FINI_FUNC (@var{stream}, @var{func})
8351 If defined, a C statement that will write a function that can be
8352 automatically called when a shared library is unloaded. The function
8353 should call @var{func}, which takes no arguments. If not defined, and
8354 the object format requires an explicit finalization function, then a
8355 function called @code{_GLOBAL__DD} will be generated.
8358 @defmac INVOKE__main
8359 If defined, @code{main} will call @code{__main} despite the presence of
8360 @code{INIT_SECTION_ASM_OP}. This macro should be defined for systems
8361 where the init section is not actually run automatically, but is still
8362 useful for collecting the lists of constructors and destructors.
8365 @defmac SUPPORTS_INIT_PRIORITY
8366 If nonzero, the C++ @code{init_priority} attribute is supported and the
8367 compiler should emit instructions to control the order of initialization
8368 of objects. If zero, the compiler will issue an error message upon
8369 encountering an @code{init_priority} attribute.
8372 @hook TARGET_HAVE_CTORS_DTORS
8373 This value is true if the target supports some ``native'' method of
8374 collecting constructors and destructors to be run at startup and exit.
8375 It is false if we must use @command{collect2}.
8378 @hook TARGET_ASM_CONSTRUCTOR
8379 If defined, a function that outputs assembler code to arrange to call
8380 the function referenced by @var{symbol} at initialization time.
8382 Assume that @var{symbol} is a @code{SYMBOL_REF} for a function taking
8383 no arguments and with no return value. If the target supports initialization
8384 priorities, @var{priority} is a value between 0 and @code{MAX_INIT_PRIORITY};
8385 otherwise it must be @code{DEFAULT_INIT_PRIORITY}.
8387 If this macro is not defined by the target, a suitable default will
8388 be chosen if (1) the target supports arbitrary section names, (2) the
8389 target defines @code{CTORS_SECTION_ASM_OP}, or (3) @code{USE_COLLECT2}
8393 @hook TARGET_ASM_DESTRUCTOR
8394 This is like @code{TARGET_ASM_CONSTRUCTOR} but used for termination
8395 functions rather than initialization functions.
8398 If @code{TARGET_HAVE_CTORS_DTORS} is true, the initialization routine
8399 generated for the generated object file will have static linkage.
8401 If your system uses @command{collect2} as the means of processing
8402 constructors, then that program normally uses @command{nm} to scan
8403 an object file for constructor functions to be called.
8405 On certain kinds of systems, you can define this macro to make
8406 @command{collect2} work faster (and, in some cases, make it work at all):
8408 @defmac OBJECT_FORMAT_COFF
8409 Define this macro if the system uses COFF (Common Object File Format)
8410 object files, so that @command{collect2} can assume this format and scan
8411 object files directly for dynamic constructor/destructor functions.
8413 This macro is effective only in a native compiler; @command{collect2} as
8414 part of a cross compiler always uses @command{nm} for the target machine.
8417 @defmac REAL_NM_FILE_NAME
8418 Define this macro as a C string constant containing the file name to use
8419 to execute @command{nm}. The default is to search the path normally for
8424 @command{collect2} calls @command{nm} to scan object files for static
8425 constructors and destructors and LTO info. By default, @option{-n} is
8426 passed. Define @code{NM_FLAGS} to a C string constant if other options
8427 are needed to get the same output format as GNU @command{nm -n}
8431 If your system supports shared libraries and has a program to list the
8432 dynamic dependencies of a given library or executable, you can define
8433 these macros to enable support for running initialization and
8434 termination functions in shared libraries:
8437 Define this macro to a C string constant containing the name of the program
8438 which lists dynamic dependencies, like @command{ldd} under SunOS 4.
8441 @defmac PARSE_LDD_OUTPUT (@var{ptr})
8442 Define this macro to be C code that extracts filenames from the output
8443 of the program denoted by @code{LDD_SUFFIX}. @var{ptr} is a variable
8444 of type @code{char *} that points to the beginning of a line of output
8445 from @code{LDD_SUFFIX}. If the line lists a dynamic dependency, the
8446 code must advance @var{ptr} to the beginning of the filename on that
8447 line. Otherwise, it must set @var{ptr} to @code{NULL}.
8450 @defmac SHLIB_SUFFIX
8451 Define this macro to a C string constant containing the default shared
8452 library extension of the target (e.g., @samp{".so"}). @command{collect2}
8453 strips version information after this suffix when generating global
8454 constructor and destructor names. This define is only needed on targets
8455 that use @command{collect2} to process constructors and destructors.
8458 @node Instruction Output
8459 @subsection Output of Assembler Instructions
8461 @c prevent bad page break with this line
8462 This describes assembler instruction output.
8464 @defmac REGISTER_NAMES
8465 A C initializer containing the assembler's names for the machine
8466 registers, each one as a C string constant. This is what translates
8467 register numbers in the compiler into assembler language.
8470 @defmac ADDITIONAL_REGISTER_NAMES
8471 If defined, a C initializer for an array of structures containing a name
8472 and a register number. This macro defines additional names for hard
8473 registers, thus allowing the @code{asm} option in declarations to refer
8474 to registers using alternate names.
8477 @defmac OVERLAPPING_REGISTER_NAMES
8478 If defined, a C initializer for an array of structures containing a
8479 name, a register number and a count of the number of consecutive
8480 machine registers the name overlaps. This macro defines additional
8481 names for hard registers, thus allowing the @code{asm} option in
8482 declarations to refer to registers using alternate names. Unlike
8483 @code{ADDITIONAL_REGISTER_NAMES}, this macro should be used when the
8484 register name implies multiple underlying registers.
8486 This macro should be used when it is important that a clobber in an
8487 @code{asm} statement clobbers all the underlying values implied by the
8488 register name. For example, on ARM, clobbering the double-precision
8489 VFP register ``d0'' implies clobbering both single-precision registers
8493 @defmac ASM_OUTPUT_OPCODE (@var{stream}, @var{ptr})
8494 Define this macro if you are using an unusual assembler that
8495 requires different names for the machine instructions.
8497 The definition is a C statement or statements which output an
8498 assembler instruction opcode to the stdio stream @var{stream}. The
8499 macro-operand @var{ptr} is a variable of type @code{char *} which
8500 points to the opcode name in its ``internal'' form---the form that is
8501 written in the machine description. The definition should output the
8502 opcode name to @var{stream}, performing any translation you desire, and
8503 increment the variable @var{ptr} to point at the end of the opcode
8504 so that it will not be output twice.
8506 In fact, your macro definition may process less than the entire opcode
8507 name, or more than the opcode name; but if you want to process text
8508 that includes @samp{%}-sequences to substitute operands, you must take
8509 care of the substitution yourself. Just be sure to increment
8510 @var{ptr} over whatever text should not be output normally.
8512 @findex recog_data.operand
8513 If you need to look at the operand values, they can be found as the
8514 elements of @code{recog_data.operand}.
8516 If the macro definition does nothing, the instruction is output
8520 @defmac FINAL_PRESCAN_INSN (@var{insn}, @var{opvec}, @var{noperands})
8521 If defined, a C statement to be executed just prior to the output of
8522 assembler code for @var{insn}, to modify the extracted operands so
8523 they will be output differently.
8525 Here the argument @var{opvec} is the vector containing the operands
8526 extracted from @var{insn}, and @var{noperands} is the number of
8527 elements of the vector which contain meaningful data for this insn.
8528 The contents of this vector are what will be used to convert the insn
8529 template into assembler code, so you can change the assembler output
8530 by changing the contents of the vector.
8532 This macro is useful when various assembler syntaxes share a single
8533 file of instruction patterns; by defining this macro differently, you
8534 can cause a large class of instructions to be output differently (such
8535 as with rearranged operands). Naturally, variations in assembler
8536 syntax affecting individual insn patterns ought to be handled by
8537 writing conditional output routines in those patterns.
8539 If this macro is not defined, it is equivalent to a null statement.
8542 @hook TARGET_ASM_FINAL_POSTSCAN_INSN
8543 If defined, this target hook is a function which is executed just after the
8544 output of assembler code for @var{insn}, to change the mode of the assembler
8547 Here the argument @var{opvec} is the vector containing the operands
8548 extracted from @var{insn}, and @var{noperands} is the number of
8549 elements of the vector which contain meaningful data for this insn.
8550 The contents of this vector are what was used to convert the insn
8551 template into assembler code, so you can change the assembler mode
8552 by checking the contents of the vector.
8555 @defmac PRINT_OPERAND (@var{stream}, @var{x}, @var{code})
8556 A C compound statement to output to stdio stream @var{stream} the
8557 assembler syntax for an instruction operand @var{x}. @var{x} is an
8560 @var{code} is a value that can be used to specify one of several ways
8561 of printing the operand. It is used when identical operands must be
8562 printed differently depending on the context. @var{code} comes from
8563 the @samp{%} specification that was used to request printing of the
8564 operand. If the specification was just @samp{%@var{digit}} then
8565 @var{code} is 0; if the specification was @samp{%@var{ltr}
8566 @var{digit}} then @var{code} is the ASCII code for @var{ltr}.
8569 If @var{x} is a register, this macro should print the register's name.
8570 The names can be found in an array @code{reg_names} whose type is
8571 @code{char *[]}. @code{reg_names} is initialized from
8572 @code{REGISTER_NAMES}.
8574 When the machine description has a specification @samp{%@var{punct}}
8575 (a @samp{%} followed by a punctuation character), this macro is called
8576 with a null pointer for @var{x} and the punctuation character for
8580 @defmac PRINT_OPERAND_PUNCT_VALID_P (@var{code})
8581 A C expression which evaluates to true if @var{code} is a valid
8582 punctuation character for use in the @code{PRINT_OPERAND} macro. If
8583 @code{PRINT_OPERAND_PUNCT_VALID_P} is not defined, it means that no
8584 punctuation characters (except for the standard one, @samp{%}) are used
8588 @defmac PRINT_OPERAND_ADDRESS (@var{stream}, @var{x})
8589 A C compound statement to output to stdio stream @var{stream} the
8590 assembler syntax for an instruction operand that is a memory reference
8591 whose address is @var{x}. @var{x} is an RTL expression.
8593 @cindex @code{TARGET_ENCODE_SECTION_INFO} usage
8594 On some machines, the syntax for a symbolic address depends on the
8595 section that the address refers to. On these machines, define the hook
8596 @code{TARGET_ENCODE_SECTION_INFO} to store the information into the
8597 @code{symbol_ref}, and then check for it here. @xref{Assembler
8601 @findex dbr_sequence_length
8602 @defmac DBR_OUTPUT_SEQEND (@var{file})
8603 A C statement, to be executed after all slot-filler instructions have
8604 been output. If necessary, call @code{dbr_sequence_length} to
8605 determine the number of slots filled in a sequence (zero if not
8606 currently outputting a sequence), to decide how many no-ops to output,
8609 Don't define this macro if it has nothing to do, but it is helpful in
8610 reading assembly output if the extent of the delay sequence is made
8611 explicit (e.g.@: with white space).
8614 @findex final_sequence
8615 Note that output routines for instructions with delay slots must be
8616 prepared to deal with not being output as part of a sequence
8617 (i.e.@: when the scheduling pass is not run, or when no slot fillers could be
8618 found.) The variable @code{final_sequence} is null when not
8619 processing a sequence, otherwise it contains the @code{sequence} rtx
8623 @defmac REGISTER_PREFIX
8624 @defmacx LOCAL_LABEL_PREFIX
8625 @defmacx USER_LABEL_PREFIX
8626 @defmacx IMMEDIATE_PREFIX
8627 If defined, C string expressions to be used for the @samp{%R}, @samp{%L},
8628 @samp{%U}, and @samp{%I} options of @code{asm_fprintf} (see
8629 @file{final.c}). These are useful when a single @file{md} file must
8630 support multiple assembler formats. In that case, the various @file{tm.h}
8631 files can define these macros differently.
8634 @defmac ASM_FPRINTF_EXTENSIONS (@var{file}, @var{argptr}, @var{format})
8635 If defined this macro should expand to a series of @code{case}
8636 statements which will be parsed inside the @code{switch} statement of
8637 the @code{asm_fprintf} function. This allows targets to define extra
8638 printf formats which may useful when generating their assembler
8639 statements. Note that uppercase letters are reserved for future
8640 generic extensions to asm_fprintf, and so are not available to target
8641 specific code. The output file is given by the parameter @var{file}.
8642 The varargs input pointer is @var{argptr} and the rest of the format
8643 string, starting the character after the one that is being switched
8644 upon, is pointed to by @var{format}.
8647 @defmac ASSEMBLER_DIALECT
8648 If your target supports multiple dialects of assembler language (such as
8649 different opcodes), define this macro as a C expression that gives the
8650 numeric index of the assembler language dialect to use, with zero as the
8653 If this macro is defined, you may use constructs of the form
8655 @samp{@{option0|option1|option2@dots{}@}}
8658 in the output templates of patterns (@pxref{Output Template}) or in the
8659 first argument of @code{asm_fprintf}. This construct outputs
8660 @samp{option0}, @samp{option1}, @samp{option2}, etc., if the value of
8661 @code{ASSEMBLER_DIALECT} is zero, one, two, etc. Any special characters
8662 within these strings retain their usual meaning. If there are fewer
8663 alternatives within the braces than the value of
8664 @code{ASSEMBLER_DIALECT}, the construct outputs nothing.
8666 If you do not define this macro, the characters @samp{@{}, @samp{|} and
8667 @samp{@}} do not have any special meaning when used in templates or
8668 operands to @code{asm_fprintf}.
8670 Define the macros @code{REGISTER_PREFIX}, @code{LOCAL_LABEL_PREFIX},
8671 @code{USER_LABEL_PREFIX} and @code{IMMEDIATE_PREFIX} if you can express
8672 the variations in assembler language syntax with that mechanism. Define
8673 @code{ASSEMBLER_DIALECT} and use the @samp{@{option0|option1@}} syntax
8674 if the syntax variant are larger and involve such things as different
8675 opcodes or operand order.
8678 @defmac ASM_OUTPUT_REG_PUSH (@var{stream}, @var{regno})
8679 A C expression to output to @var{stream} some assembler code
8680 which will push hard register number @var{regno} onto the stack.
8681 The code need not be optimal, since this macro is used only when
8685 @defmac ASM_OUTPUT_REG_POP (@var{stream}, @var{regno})
8686 A C expression to output to @var{stream} some assembler code
8687 which will pop hard register number @var{regno} off of the stack.
8688 The code need not be optimal, since this macro is used only when
8692 @node Dispatch Tables
8693 @subsection Output of Dispatch Tables
8695 @c prevent bad page break with this line
8696 This concerns dispatch tables.
8698 @cindex dispatch table
8699 @defmac ASM_OUTPUT_ADDR_DIFF_ELT (@var{stream}, @var{body}, @var{value}, @var{rel})
8700 A C statement to output to the stdio stream @var{stream} an assembler
8701 pseudo-instruction to generate a difference between two labels.
8702 @var{value} and @var{rel} are the numbers of two internal labels. The
8703 definitions of these labels are output using
8704 @code{(*targetm.asm_out.internal_label)}, and they must be printed in the same
8705 way here. For example,
8708 fprintf (@var{stream}, "\t.word L%d-L%d\n",
8709 @var{value}, @var{rel})
8712 You must provide this macro on machines where the addresses in a
8713 dispatch table are relative to the table's own address. If defined, GCC
8714 will also use this macro on all machines when producing PIC@.
8715 @var{body} is the body of the @code{ADDR_DIFF_VEC}; it is provided so that the
8716 mode and flags can be read.
8719 @defmac ASM_OUTPUT_ADDR_VEC_ELT (@var{stream}, @var{value})
8720 This macro should be provided on machines where the addresses
8721 in a dispatch table are absolute.
8723 The definition should be a C statement to output to the stdio stream
8724 @var{stream} an assembler pseudo-instruction to generate a reference to
8725 a label. @var{value} is the number of an internal label whose
8726 definition is output using @code{(*targetm.asm_out.internal_label)}.
8730 fprintf (@var{stream}, "\t.word L%d\n", @var{value})
8734 @defmac ASM_OUTPUT_CASE_LABEL (@var{stream}, @var{prefix}, @var{num}, @var{table})
8735 Define this if the label before a jump-table needs to be output
8736 specially. The first three arguments are the same as for
8737 @code{(*targetm.asm_out.internal_label)}; the fourth argument is the
8738 jump-table which follows (a @code{jump_insn} containing an
8739 @code{addr_vec} or @code{addr_diff_vec}).
8741 This feature is used on system V to output a @code{swbeg} statement
8744 If this macro is not defined, these labels are output with
8745 @code{(*targetm.asm_out.internal_label)}.
8748 @defmac ASM_OUTPUT_CASE_END (@var{stream}, @var{num}, @var{table})
8749 Define this if something special must be output at the end of a
8750 jump-table. The definition should be a C statement to be executed
8751 after the assembler code for the table is written. It should write
8752 the appropriate code to stdio stream @var{stream}. The argument
8753 @var{table} is the jump-table insn, and @var{num} is the label-number
8754 of the preceding label.
8756 If this macro is not defined, nothing special is output at the end of
8760 @hook TARGET_ASM_EMIT_UNWIND_LABEL
8761 This target hook emits a label at the beginning of each FDE@. It
8762 should be defined on targets where FDEs need special labels, and it
8763 should write the appropriate label, for the FDE associated with the
8764 function declaration @var{decl}, to the stdio stream @var{stream}.
8765 The third argument, @var{for_eh}, is a boolean: true if this is for an
8766 exception table. The fourth argument, @var{empty}, is a boolean:
8767 true if this is a placeholder label for an omitted FDE@.
8769 The default is that FDEs are not given nonlocal labels.
8772 @hook TARGET_ASM_EMIT_EXCEPT_TABLE_LABEL
8773 This target hook emits a label at the beginning of the exception table.
8774 It should be defined on targets where it is desirable for the table
8775 to be broken up according to function.
8777 The default is that no label is emitted.
8780 @hook TARGET_ASM_EMIT_EXCEPT_PERSONALITY
8782 @hook TARGET_ASM_UNWIND_EMIT
8783 This target hook emits assembly directives required to unwind the
8784 given instruction. This is only used when @code{TARGET_EXCEPT_UNWIND_INFO}
8785 returns @code{UI_TARGET}.
8788 @hook TARGET_ASM_UNWIND_EMIT_BEFORE_INSN
8790 @node Exception Region Output
8791 @subsection Assembler Commands for Exception Regions
8793 @c prevent bad page break with this line
8795 This describes commands marking the start and the end of an exception
8798 @defmac EH_FRAME_SECTION_NAME
8799 If defined, a C string constant for the name of the section containing
8800 exception handling frame unwind information. If not defined, GCC will
8801 provide a default definition if the target supports named sections.
8802 @file{crtstuff.c} uses this macro to switch to the appropriate section.
8804 You should define this symbol if your target supports DWARF 2 frame
8805 unwind information and the default definition does not work.
8808 @defmac EH_FRAME_IN_DATA_SECTION
8809 If defined, DWARF 2 frame unwind information will be placed in the
8810 data section even though the target supports named sections. This
8811 might be necessary, for instance, if the system linker does garbage
8812 collection and sections cannot be marked as not to be collected.
8814 Do not define this macro unless @code{TARGET_ASM_NAMED_SECTION} is
8818 @defmac EH_TABLES_CAN_BE_READ_ONLY
8819 Define this macro to 1 if your target is such that no frame unwind
8820 information encoding used with non-PIC code will ever require a
8821 runtime relocation, but the linker may not support merging read-only
8822 and read-write sections into a single read-write section.
8825 @defmac MASK_RETURN_ADDR
8826 An rtx used to mask the return address found via @code{RETURN_ADDR_RTX}, so
8827 that it does not contain any extraneous set bits in it.
8830 @defmac DWARF2_UNWIND_INFO
8831 Define this macro to 0 if your target supports DWARF 2 frame unwind
8832 information, but it does not yet work with exception handling.
8833 Otherwise, if your target supports this information (if it defines
8834 @code{INCOMING_RETURN_ADDR_RTX} and either @code{UNALIGNED_INT_ASM_OP}
8835 or @code{OBJECT_FORMAT_ELF}), GCC will provide a default definition of 1.
8838 @hook TARGET_EXCEPT_UNWIND_INFO
8839 This hook defines the mechanism that will be used for exception handling
8840 by the target. If the target has ABI specified unwind tables, the hook
8841 should return @code{UI_TARGET}. If the target is to use the
8842 @code{setjmp}/@code{longjmp}-based exception handling scheme, the hook
8843 should return @code{UI_SJLJ}. If the target supports DWARF 2 frame unwind
8844 information, the hook should return @code{UI_DWARF2}.
8846 A target may, if exceptions are disabled, choose to return @code{UI_NONE}.
8847 This may end up simplifying other parts of target-specific code. The
8848 default implementation of this hook never returns @code{UI_NONE}.
8850 Note that the value returned by this hook should be constant. It should
8851 not depend on anything except the command-line switches described by
8852 @var{opts}. In particular, the
8853 setting @code{UI_SJLJ} must be fixed at compiler start-up as C pre-processor
8854 macros and builtin functions related to exception handling are set up
8855 depending on this setting.
8857 The default implementation of the hook first honors the
8858 @option{--enable-sjlj-exceptions} configure option, then
8859 @code{DWARF2_UNWIND_INFO}, and finally defaults to @code{UI_SJLJ}. If
8860 @code{DWARF2_UNWIND_INFO} depends on command-line options, the target
8861 must define this hook so that @var{opts} is used correctly.
8864 @hook TARGET_UNWIND_TABLES_DEFAULT
8865 This variable should be set to @code{true} if the target ABI requires unwinding
8866 tables even when exceptions are not used. It must not be modified by
8867 command-line option processing.
8870 @defmac DONT_USE_BUILTIN_SETJMP
8871 Define this macro to 1 if the @code{setjmp}/@code{longjmp}-based scheme
8872 should use the @code{setjmp}/@code{longjmp} functions from the C library
8873 instead of the @code{__builtin_setjmp}/@code{__builtin_longjmp} machinery.
8876 @defmac DWARF_CIE_DATA_ALIGNMENT
8877 This macro need only be defined if the target might save registers in the
8878 function prologue at an offset to the stack pointer that is not aligned to
8879 @code{UNITS_PER_WORD}. The definition should be the negative minimum
8880 alignment if @code{STACK_GROWS_DOWNWARD} is defined, and the positive
8881 minimum alignment otherwise. @xref{SDB and DWARF}. Only applicable if
8882 the target supports DWARF 2 frame unwind information.
8885 @hook TARGET_TERMINATE_DW2_EH_FRAME_INFO
8886 Contains the value true if the target should add a zero word onto the
8887 end of a Dwarf-2 frame info section when used for exception handling.
8888 Default value is false if @code{EH_FRAME_SECTION_NAME} is defined, and
8892 @hook TARGET_DWARF_REGISTER_SPAN
8893 Given a register, this hook should return a parallel of registers to
8894 represent where to find the register pieces. Define this hook if the
8895 register and its mode are represented in Dwarf in non-contiguous
8896 locations, or if the register should be represented in more than one
8897 register in Dwarf. Otherwise, this hook should return @code{NULL_RTX}.
8898 If not defined, the default is to return @code{NULL_RTX}.
8901 @hook TARGET_INIT_DWARF_REG_SIZES_EXTRA
8902 If some registers are represented in Dwarf-2 unwind information in
8903 multiple pieces, define this hook to fill in information about the
8904 sizes of those pieces in the table used by the unwinder at runtime.
8905 It will be called by @code{expand_builtin_init_dwarf_reg_sizes} after
8906 filling in a single size corresponding to each hard register;
8907 @var{address} is the address of the table.
8910 @hook TARGET_ASM_TTYPE
8911 This hook is used to output a reference from a frame unwinding table to
8912 the type_info object identified by @var{sym}. It should return @code{true}
8913 if the reference was output. Returning @code{false} will cause the
8914 reference to be output using the normal Dwarf2 routines.
8917 @hook TARGET_ARM_EABI_UNWINDER
8918 This flag should be set to @code{true} on targets that use an ARM EABI
8919 based unwinding library, and @code{false} on other targets. This effects
8920 the format of unwinding tables, and how the unwinder in entered after
8921 running a cleanup. The default is @code{false}.
8924 @node Alignment Output
8925 @subsection Assembler Commands for Alignment
8927 @c prevent bad page break with this line
8928 This describes commands for alignment.
8930 @defmac JUMP_ALIGN (@var{label})
8931 The alignment (log base 2) to put in front of @var{label}, which is
8932 a common destination of jumps and has no fallthru incoming edge.
8934 This macro need not be defined if you don't want any special alignment
8935 to be done at such a time. Most machine descriptions do not currently
8938 Unless it's necessary to inspect the @var{label} parameter, it is better
8939 to set the variable @var{align_jumps} in the target's
8940 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
8941 selection in @var{align_jumps} in a @code{JUMP_ALIGN} implementation.
8944 @hook TARGET_ASM_JUMP_ALIGN_MAX_SKIP
8945 The maximum number of bytes to skip before @var{label} when applying
8946 @code{JUMP_ALIGN}. This works only if
8947 @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
8950 @defmac LABEL_ALIGN_AFTER_BARRIER (@var{label})
8951 The alignment (log base 2) to put in front of @var{label}, which follows
8954 This macro need not be defined if you don't want any special alignment
8955 to be done at such a time. Most machine descriptions do not currently
8959 @hook TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP
8960 The maximum number of bytes to skip before @var{label} when applying
8961 @code{LABEL_ALIGN_AFTER_BARRIER}. This works only if
8962 @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
8965 @defmac LOOP_ALIGN (@var{label})
8966 The alignment (log base 2) to put in front of @var{label}, which follows
8967 a @code{NOTE_INSN_LOOP_BEG} note.
8969 This macro need not be defined if you don't want any special alignment
8970 to be done at such a time. Most machine descriptions do not currently
8973 Unless it's necessary to inspect the @var{label} parameter, it is better
8974 to set the variable @code{align_loops} in the target's
8975 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
8976 selection in @code{align_loops} in a @code{LOOP_ALIGN} implementation.
8979 @hook TARGET_ASM_LOOP_ALIGN_MAX_SKIP
8980 The maximum number of bytes to skip when applying @code{LOOP_ALIGN} to
8981 @var{label}. This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is
8985 @defmac LABEL_ALIGN (@var{label})
8986 The alignment (log base 2) to put in front of @var{label}.
8987 If @code{LABEL_ALIGN_AFTER_BARRIER} / @code{LOOP_ALIGN} specify a different alignment,
8988 the maximum of the specified values is used.
8990 Unless it's necessary to inspect the @var{label} parameter, it is better
8991 to set the variable @code{align_labels} in the target's
8992 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
8993 selection in @code{align_labels} in a @code{LABEL_ALIGN} implementation.
8996 @hook TARGET_ASM_LABEL_ALIGN_MAX_SKIP
8997 The maximum number of bytes to skip when applying @code{LABEL_ALIGN}
8998 to @var{label}. This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN}
9002 @defmac ASM_OUTPUT_SKIP (@var{stream}, @var{nbytes})
9003 A C statement to output to the stdio stream @var{stream} an assembler
9004 instruction to advance the location counter by @var{nbytes} bytes.
9005 Those bytes should be zero when loaded. @var{nbytes} will be a C
9006 expression of type @code{unsigned HOST_WIDE_INT}.
9009 @defmac ASM_NO_SKIP_IN_TEXT
9010 Define this macro if @code{ASM_OUTPUT_SKIP} should not be used in the
9011 text section because it fails to put zeros in the bytes that are skipped.
9012 This is true on many Unix systems, where the pseudo--op to skip bytes
9013 produces no-op instructions rather than zeros when used in the text
9017 @defmac ASM_OUTPUT_ALIGN (@var{stream}, @var{power})
9018 A C statement to output to the stdio stream @var{stream} an assembler
9019 command to advance the location counter to a multiple of 2 to the
9020 @var{power} bytes. @var{power} will be a C expression of type @code{int}.
9023 @defmac ASM_OUTPUT_ALIGN_WITH_NOP (@var{stream}, @var{power})
9024 Like @code{ASM_OUTPUT_ALIGN}, except that the ``nop'' instruction is used
9025 for padding, if necessary.
9028 @defmac ASM_OUTPUT_MAX_SKIP_ALIGN (@var{stream}, @var{power}, @var{max_skip})
9029 A C statement to output to the stdio stream @var{stream} an assembler
9030 command to advance the location counter to a multiple of 2 to the
9031 @var{power} bytes, but only if @var{max_skip} or fewer bytes are needed to
9032 satisfy the alignment request. @var{power} and @var{max_skip} will be
9033 a C expression of type @code{int}.
9037 @node Debugging Info
9038 @section Controlling Debugging Information Format
9040 @c prevent bad page break with this line
9041 This describes how to specify debugging information.
9044 * All Debuggers:: Macros that affect all debugging formats uniformly.
9045 * DBX Options:: Macros enabling specific options in DBX format.
9046 * DBX Hooks:: Hook macros for varying DBX format.
9047 * File Names and DBX:: Macros controlling output of file names in DBX format.
9048 * SDB and DWARF:: Macros for SDB (COFF) and DWARF formats.
9049 * VMS Debug:: Macros for VMS debug format.
9053 @subsection Macros Affecting All Debugging Formats
9055 @c prevent bad page break with this line
9056 These macros affect all debugging formats.
9058 @defmac DBX_REGISTER_NUMBER (@var{regno})
9059 A C expression that returns the DBX register number for the compiler
9060 register number @var{regno}. In the default macro provided, the value
9061 of this expression will be @var{regno} itself. But sometimes there are
9062 some registers that the compiler knows about and DBX does not, or vice
9063 versa. In such cases, some register may need to have one number in the
9064 compiler and another for DBX@.
9066 If two registers have consecutive numbers inside GCC, and they can be
9067 used as a pair to hold a multiword value, then they @emph{must} have
9068 consecutive numbers after renumbering with @code{DBX_REGISTER_NUMBER}.
9069 Otherwise, debuggers will be unable to access such a pair, because they
9070 expect register pairs to be consecutive in their own numbering scheme.
9072 If you find yourself defining @code{DBX_REGISTER_NUMBER} in way that
9073 does not preserve register pairs, then what you must do instead is
9074 redefine the actual register numbering scheme.
9077 @defmac DEBUGGER_AUTO_OFFSET (@var{x})
9078 A C expression that returns the integer offset value for an automatic
9079 variable having address @var{x} (an RTL expression). The default
9080 computation assumes that @var{x} is based on the frame-pointer and
9081 gives the offset from the frame-pointer. This is required for targets
9082 that produce debugging output for DBX or COFF-style debugging output
9083 for SDB and allow the frame-pointer to be eliminated when the
9084 @option{-g} options is used.
9087 @defmac DEBUGGER_ARG_OFFSET (@var{offset}, @var{x})
9088 A C expression that returns the integer offset value for an argument
9089 having address @var{x} (an RTL expression). The nominal offset is
9093 @defmac PREFERRED_DEBUGGING_TYPE
9094 A C expression that returns the type of debugging output GCC should
9095 produce when the user specifies just @option{-g}. Define
9096 this if you have arranged for GCC to support more than one format of
9097 debugging output. Currently, the allowable values are @code{DBX_DEBUG},
9098 @code{SDB_DEBUG}, @code{DWARF_DEBUG}, @code{DWARF2_DEBUG},
9099 @code{XCOFF_DEBUG}, @code{VMS_DEBUG}, and @code{VMS_AND_DWARF2_DEBUG}.
9101 When the user specifies @option{-ggdb}, GCC normally also uses the
9102 value of this macro to select the debugging output format, but with two
9103 exceptions. If @code{DWARF2_DEBUGGING_INFO} is defined, GCC uses the
9104 value @code{DWARF2_DEBUG}. Otherwise, if @code{DBX_DEBUGGING_INFO} is
9105 defined, GCC uses @code{DBX_DEBUG}.
9107 The value of this macro only affects the default debugging output; the
9108 user can always get a specific type of output by using @option{-gstabs},
9109 @option{-gcoff}, @option{-gdwarf-2}, @option{-gxcoff}, or @option{-gvms}.
9113 @subsection Specific Options for DBX Output
9115 @c prevent bad page break with this line
9116 These are specific options for DBX output.
9118 @defmac DBX_DEBUGGING_INFO
9119 Define this macro if GCC should produce debugging output for DBX
9120 in response to the @option{-g} option.
9123 @defmac XCOFF_DEBUGGING_INFO
9124 Define this macro if GCC should produce XCOFF format debugging output
9125 in response to the @option{-g} option. This is a variant of DBX format.
9128 @defmac DEFAULT_GDB_EXTENSIONS
9129 Define this macro to control whether GCC should by default generate
9130 GDB's extended version of DBX debugging information (assuming DBX-format
9131 debugging information is enabled at all). If you don't define the
9132 macro, the default is 1: always generate the extended information
9133 if there is any occasion to.
9136 @defmac DEBUG_SYMS_TEXT
9137 Define this macro if all @code{.stabs} commands should be output while
9138 in the text section.
9141 @defmac ASM_STABS_OP
9142 A C string constant, including spacing, naming the assembler pseudo op to
9143 use instead of @code{"\t.stabs\t"} to define an ordinary debugging symbol.
9144 If you don't define this macro, @code{"\t.stabs\t"} is used. This macro
9145 applies only to DBX debugging information format.
9148 @defmac ASM_STABD_OP
9149 A C string constant, including spacing, naming the assembler pseudo op to
9150 use instead of @code{"\t.stabd\t"} to define a debugging symbol whose
9151 value is the current location. If you don't define this macro,
9152 @code{"\t.stabd\t"} is used. This macro applies only to DBX debugging
9156 @defmac ASM_STABN_OP
9157 A C string constant, including spacing, naming the assembler pseudo op to
9158 use instead of @code{"\t.stabn\t"} to define a debugging symbol with no
9159 name. If you don't define this macro, @code{"\t.stabn\t"} is used. This
9160 macro applies only to DBX debugging information format.
9163 @defmac DBX_NO_XREFS
9164 Define this macro if DBX on your system does not support the construct
9165 @samp{xs@var{tagname}}. On some systems, this construct is used to
9166 describe a forward reference to a structure named @var{tagname}.
9167 On other systems, this construct is not supported at all.
9170 @defmac DBX_CONTIN_LENGTH
9171 A symbol name in DBX-format debugging information is normally
9172 continued (split into two separate @code{.stabs} directives) when it
9173 exceeds a certain length (by default, 80 characters). On some
9174 operating systems, DBX requires this splitting; on others, splitting
9175 must not be done. You can inhibit splitting by defining this macro
9176 with the value zero. You can override the default splitting-length by
9177 defining this macro as an expression for the length you desire.
9180 @defmac DBX_CONTIN_CHAR
9181 Normally continuation is indicated by adding a @samp{\} character to
9182 the end of a @code{.stabs} string when a continuation follows. To use
9183 a different character instead, define this macro as a character
9184 constant for the character you want to use. Do not define this macro
9185 if backslash is correct for your system.
9188 @defmac DBX_STATIC_STAB_DATA_SECTION
9189 Define this macro if it is necessary to go to the data section before
9190 outputting the @samp{.stabs} pseudo-op for a non-global static
9194 @defmac DBX_TYPE_DECL_STABS_CODE
9195 The value to use in the ``code'' field of the @code{.stabs} directive
9196 for a typedef. The default is @code{N_LSYM}.
9199 @defmac DBX_STATIC_CONST_VAR_CODE
9200 The value to use in the ``code'' field of the @code{.stabs} directive
9201 for a static variable located in the text section. DBX format does not
9202 provide any ``right'' way to do this. The default is @code{N_FUN}.
9205 @defmac DBX_REGPARM_STABS_CODE
9206 The value to use in the ``code'' field of the @code{.stabs} directive
9207 for a parameter passed in registers. DBX format does not provide any
9208 ``right'' way to do this. The default is @code{N_RSYM}.
9211 @defmac DBX_REGPARM_STABS_LETTER
9212 The letter to use in DBX symbol data to identify a symbol as a parameter
9213 passed in registers. DBX format does not customarily provide any way to
9214 do this. The default is @code{'P'}.
9217 @defmac DBX_FUNCTION_FIRST
9218 Define this macro if the DBX information for a function and its
9219 arguments should precede the assembler code for the function. Normally,
9220 in DBX format, the debugging information entirely follows the assembler
9224 @defmac DBX_BLOCKS_FUNCTION_RELATIVE
9225 Define this macro, with value 1, if the value of a symbol describing
9226 the scope of a block (@code{N_LBRAC} or @code{N_RBRAC}) should be
9227 relative to the start of the enclosing function. Normally, GCC uses
9228 an absolute address.
9231 @defmac DBX_LINES_FUNCTION_RELATIVE
9232 Define this macro, with value 1, if the value of a symbol indicating
9233 the current line number (@code{N_SLINE}) should be relative to the
9234 start of the enclosing function. Normally, GCC uses an absolute address.
9237 @defmac DBX_USE_BINCL
9238 Define this macro if GCC should generate @code{N_BINCL} and
9239 @code{N_EINCL} stabs for included header files, as on Sun systems. This
9240 macro also directs GCC to output a type number as a pair of a file
9241 number and a type number within the file. Normally, GCC does not
9242 generate @code{N_BINCL} or @code{N_EINCL} stabs, and it outputs a single
9243 number for a type number.
9247 @subsection Open-Ended Hooks for DBX Format
9249 @c prevent bad page break with this line
9250 These are hooks for DBX format.
9252 @defmac DBX_OUTPUT_LBRAC (@var{stream}, @var{name})
9253 Define this macro to say how to output to @var{stream} the debugging
9254 information for the start of a scope level for variable names. The
9255 argument @var{name} is the name of an assembler symbol (for use with
9256 @code{assemble_name}) whose value is the address where the scope begins.
9259 @defmac DBX_OUTPUT_RBRAC (@var{stream}, @var{name})
9260 Like @code{DBX_OUTPUT_LBRAC}, but for the end of a scope level.
9263 @defmac DBX_OUTPUT_NFUN (@var{stream}, @var{lscope_label}, @var{decl})
9264 Define this macro if the target machine requires special handling to
9265 output an @code{N_FUN} entry for the function @var{decl}.
9268 @defmac DBX_OUTPUT_SOURCE_LINE (@var{stream}, @var{line}, @var{counter})
9269 A C statement to output DBX debugging information before code for line
9270 number @var{line} of the current source file to the stdio stream
9271 @var{stream}. @var{counter} is the number of time the macro was
9272 invoked, including the current invocation; it is intended to generate
9273 unique labels in the assembly output.
9275 This macro should not be defined if the default output is correct, or
9276 if it can be made correct by defining @code{DBX_LINES_FUNCTION_RELATIVE}.
9279 @defmac NO_DBX_FUNCTION_END
9280 Some stabs encapsulation formats (in particular ECOFF), cannot handle the
9281 @code{.stabs "",N_FUN,,0,0,Lscope-function-1} gdb dbx extension construct.
9282 On those machines, define this macro to turn this feature off without
9283 disturbing the rest of the gdb extensions.
9286 @defmac NO_DBX_BNSYM_ENSYM
9287 Some assemblers cannot handle the @code{.stabd BNSYM/ENSYM,0,0} gdb dbx
9288 extension construct. On those machines, define this macro to turn this
9289 feature off without disturbing the rest of the gdb extensions.
9292 @node File Names and DBX
9293 @subsection File Names in DBX Format
9295 @c prevent bad page break with this line
9296 This describes file names in DBX format.
9298 @defmac DBX_OUTPUT_MAIN_SOURCE_FILENAME (@var{stream}, @var{name})
9299 A C statement to output DBX debugging information to the stdio stream
9300 @var{stream}, which indicates that file @var{name} is the main source
9301 file---the file specified as the input file for compilation.
9302 This macro is called only once, at the beginning of compilation.
9304 This macro need not be defined if the standard form of output
9305 for DBX debugging information is appropriate.
9307 It may be necessary to refer to a label equal to the beginning of the
9308 text section. You can use @samp{assemble_name (stream, ltext_label_name)}
9309 to do so. If you do this, you must also set the variable
9310 @var{used_ltext_label_name} to @code{true}.
9313 @defmac NO_DBX_MAIN_SOURCE_DIRECTORY
9314 Define this macro, with value 1, if GCC should not emit an indication
9315 of the current directory for compilation and current source language at
9316 the beginning of the file.
9319 @defmac NO_DBX_GCC_MARKER
9320 Define this macro, with value 1, if GCC should not emit an indication
9321 that this object file was compiled by GCC@. The default is to emit
9322 an @code{N_OPT} stab at the beginning of every source file, with
9323 @samp{gcc2_compiled.} for the string and value 0.
9326 @defmac DBX_OUTPUT_MAIN_SOURCE_FILE_END (@var{stream}, @var{name})
9327 A C statement to output DBX debugging information at the end of
9328 compilation of the main source file @var{name}. Output should be
9329 written to the stdio stream @var{stream}.
9331 If you don't define this macro, nothing special is output at the end
9332 of compilation, which is correct for most machines.
9335 @defmac DBX_OUTPUT_NULL_N_SO_AT_MAIN_SOURCE_FILE_END
9336 Define this macro @emph{instead of} defining
9337 @code{DBX_OUTPUT_MAIN_SOURCE_FILE_END}, if what needs to be output at
9338 the end of compilation is an @code{N_SO} stab with an empty string,
9339 whose value is the highest absolute text address in the file.
9344 @subsection Macros for SDB and DWARF Output
9346 @c prevent bad page break with this line
9347 Here are macros for SDB and DWARF output.
9349 @defmac SDB_DEBUGGING_INFO
9350 Define this macro if GCC should produce COFF-style debugging output
9351 for SDB in response to the @option{-g} option.
9354 @defmac DWARF2_DEBUGGING_INFO
9355 Define this macro if GCC should produce dwarf version 2 format
9356 debugging output in response to the @option{-g} option.
9358 @hook TARGET_DWARF_CALLING_CONVENTION
9359 Define this to enable the dwarf attribute @code{DW_AT_calling_convention} to
9360 be emitted for each function. Instead of an integer return the enum
9361 value for the @code{DW_CC_} tag.
9364 To support optional call frame debugging information, you must also
9365 define @code{INCOMING_RETURN_ADDR_RTX} and either set
9366 @code{RTX_FRAME_RELATED_P} on the prologue insns if you use RTL for the
9367 prologue, or call @code{dwarf2out_def_cfa} and @code{dwarf2out_reg_save}
9368 as appropriate from @code{TARGET_ASM_FUNCTION_PROLOGUE} if you don't.
9371 @defmac DWARF2_FRAME_INFO
9372 Define this macro to a nonzero value if GCC should always output
9373 Dwarf 2 frame information. If @code{TARGET_EXCEPT_UNWIND_INFO}
9374 (@pxref{Exception Region Output}) returns @code{UI_DWARF2}, and
9375 exceptions are enabled, GCC will output this information not matter
9376 how you define @code{DWARF2_FRAME_INFO}.
9379 @hook TARGET_DEBUG_UNWIND_INFO
9380 This hook defines the mechanism that will be used for describing frame
9381 unwind information to the debugger. Normally the hook will return
9382 @code{UI_DWARF2} if DWARF 2 debug information is enabled, and
9383 return @code{UI_NONE} otherwise.
9385 A target may return @code{UI_DWARF2} even when DWARF 2 debug information
9386 is disabled in order to always output DWARF 2 frame information.
9388 A target may return @code{UI_TARGET} if it has ABI specified unwind tables.
9389 This will suppress generation of the normal debug frame unwind information.
9392 @defmac DWARF2_ASM_LINE_DEBUG_INFO
9393 Define this macro to be a nonzero value if the assembler can generate Dwarf 2
9394 line debug info sections. This will result in much more compact line number
9395 tables, and hence is desirable if it works.
9398 @hook TARGET_WANT_DEBUG_PUB_SECTIONS
9400 @hook TARGET_DELAY_SCHED2
9402 @hook TARGET_DELAY_VARTRACK
9404 @defmac ASM_OUTPUT_DWARF_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
9405 A C statement to issue assembly directives that create a difference
9406 @var{lab1} minus @var{lab2}, using an integer of the given @var{size}.
9409 @defmac ASM_OUTPUT_DWARF_VMS_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
9410 A C statement to issue assembly directives that create a difference
9411 between the two given labels in system defined units, e.g. instruction
9412 slots on IA64 VMS, using an integer of the given size.
9415 @defmac ASM_OUTPUT_DWARF_OFFSET (@var{stream}, @var{size}, @var{label}, @var{section})
9416 A C statement to issue assembly directives that create a
9417 section-relative reference to the given @var{label}, using an integer of the
9418 given @var{size}. The label is known to be defined in the given @var{section}.
9421 @defmac ASM_OUTPUT_DWARF_PCREL (@var{stream}, @var{size}, @var{label})
9422 A C statement to issue assembly directives that create a self-relative
9423 reference to the given @var{label}, using an integer of the given @var{size}.
9426 @defmac ASM_OUTPUT_DWARF_TABLE_REF (@var{label})
9427 A C statement to issue assembly directives that create a reference to
9428 the DWARF table identifier @var{label} from the current section. This
9429 is used on some systems to avoid garbage collecting a DWARF table which
9430 is referenced by a function.
9433 @hook TARGET_ASM_OUTPUT_DWARF_DTPREL
9434 If defined, this target hook is a function which outputs a DTP-relative
9435 reference to the given TLS symbol of the specified size.
9438 @defmac PUT_SDB_@dots{}
9439 Define these macros to override the assembler syntax for the special
9440 SDB assembler directives. See @file{sdbout.c} for a list of these
9441 macros and their arguments. If the standard syntax is used, you need
9442 not define them yourself.
9446 Some assemblers do not support a semicolon as a delimiter, even between
9447 SDB assembler directives. In that case, define this macro to be the
9448 delimiter to use (usually @samp{\n}). It is not necessary to define
9449 a new set of @code{PUT_SDB_@var{op}} macros if this is the only change
9453 @defmac SDB_ALLOW_UNKNOWN_REFERENCES
9454 Define this macro to allow references to unknown structure,
9455 union, or enumeration tags to be emitted. Standard COFF does not
9456 allow handling of unknown references, MIPS ECOFF has support for
9460 @defmac SDB_ALLOW_FORWARD_REFERENCES
9461 Define this macro to allow references to structure, union, or
9462 enumeration tags that have not yet been seen to be handled. Some
9463 assemblers choke if forward tags are used, while some require it.
9466 @defmac SDB_OUTPUT_SOURCE_LINE (@var{stream}, @var{line})
9467 A C statement to output SDB debugging information before code for line
9468 number @var{line} of the current source file to the stdio stream
9469 @var{stream}. The default is to emit an @code{.ln} directive.
9474 @subsection Macros for VMS Debug Format
9476 @c prevent bad page break with this line
9477 Here are macros for VMS debug format.
9479 @defmac VMS_DEBUGGING_INFO
9480 Define this macro if GCC should produce debugging output for VMS
9481 in response to the @option{-g} option. The default behavior for VMS
9482 is to generate minimal debug info for a traceback in the absence of
9483 @option{-g} unless explicitly overridden with @option{-g0}. This
9484 behavior is controlled by @code{TARGET_OPTION_OPTIMIZATION} and
9485 @code{TARGET_OPTION_OVERRIDE}.
9488 @node Floating Point
9489 @section Cross Compilation and Floating Point
9490 @cindex cross compilation and floating point
9491 @cindex floating point and cross compilation
9493 While all modern machines use twos-complement representation for integers,
9494 there are a variety of representations for floating point numbers. This
9495 means that in a cross-compiler the representation of floating point numbers
9496 in the compiled program may be different from that used in the machine
9497 doing the compilation.
9499 Because different representation systems may offer different amounts of
9500 range and precision, all floating point constants must be represented in
9501 the target machine's format. Therefore, the cross compiler cannot
9502 safely use the host machine's floating point arithmetic; it must emulate
9503 the target's arithmetic. To ensure consistency, GCC always uses
9504 emulation to work with floating point values, even when the host and
9505 target floating point formats are identical.
9507 The following macros are provided by @file{real.h} for the compiler to
9508 use. All parts of the compiler which generate or optimize
9509 floating-point calculations must use these macros. They may evaluate
9510 their operands more than once, so operands must not have side effects.
9512 @defmac REAL_VALUE_TYPE
9513 The C data type to be used to hold a floating point value in the target
9514 machine's format. Typically this is a @code{struct} containing an
9515 array of @code{HOST_WIDE_INT}, but all code should treat it as an opaque
9519 @deftypefn Macro int REAL_VALUES_EQUAL (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
9520 Compares for equality the two values, @var{x} and @var{y}. If the target
9521 floating point format supports negative zeroes and/or NaNs,
9522 @samp{REAL_VALUES_EQUAL (-0.0, 0.0)} is true, and
9523 @samp{REAL_VALUES_EQUAL (NaN, NaN)} is false.
9526 @deftypefn Macro int REAL_VALUES_LESS (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
9527 Tests whether @var{x} is less than @var{y}.
9530 @deftypefn Macro HOST_WIDE_INT REAL_VALUE_FIX (REAL_VALUE_TYPE @var{x})
9531 Truncates @var{x} to a signed integer, rounding toward zero.
9534 @deftypefn Macro {unsigned HOST_WIDE_INT} REAL_VALUE_UNSIGNED_FIX (REAL_VALUE_TYPE @var{x})
9535 Truncates @var{x} to an unsigned integer, rounding toward zero. If
9536 @var{x} is negative, returns zero.
9539 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ATOF (const char *@var{string}, enum machine_mode @var{mode})
9540 Converts @var{string} into a floating point number in the target machine's
9541 representation for mode @var{mode}. This routine can handle both
9542 decimal and hexadecimal floating point constants, using the syntax
9543 defined by the C language for both.
9546 @deftypefn Macro int REAL_VALUE_NEGATIVE (REAL_VALUE_TYPE @var{x})
9547 Returns 1 if @var{x} is negative (including negative zero), 0 otherwise.
9550 @deftypefn Macro int REAL_VALUE_ISINF (REAL_VALUE_TYPE @var{x})
9551 Determines whether @var{x} represents infinity (positive or negative).
9554 @deftypefn Macro int REAL_VALUE_ISNAN (REAL_VALUE_TYPE @var{x})
9555 Determines whether @var{x} represents a ``NaN'' (not-a-number).
9558 @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})
9559 Calculates an arithmetic operation on the two floating point values
9560 @var{x} and @var{y}, storing the result in @var{output} (which must be a
9563 The operation to be performed is specified by @var{code}. Only the
9564 following codes are supported: @code{PLUS_EXPR}, @code{MINUS_EXPR},
9565 @code{MULT_EXPR}, @code{RDIV_EXPR}, @code{MAX_EXPR}, @code{MIN_EXPR}.
9567 If @code{REAL_ARITHMETIC} is asked to evaluate division by zero and the
9568 target's floating point format cannot represent infinity, it will call
9569 @code{abort}. Callers should check for this situation first, using
9570 @code{MODE_HAS_INFINITIES}. @xref{Storage Layout}.
9573 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_NEGATE (REAL_VALUE_TYPE @var{x})
9574 Returns the negative of the floating point value @var{x}.
9577 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ABS (REAL_VALUE_TYPE @var{x})
9578 Returns the absolute value of @var{x}.
9581 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_TRUNCATE (REAL_VALUE_TYPE @var{mode}, enum machine_mode @var{x})
9582 Truncates the floating point value @var{x} to fit in @var{mode}. The
9583 return value is still a full-size @code{REAL_VALUE_TYPE}, but it has an
9584 appropriate bit pattern to be output as a floating constant whose
9585 precision accords with mode @var{mode}.
9588 @deftypefn Macro void REAL_VALUE_TO_INT (HOST_WIDE_INT @var{low}, HOST_WIDE_INT @var{high}, REAL_VALUE_TYPE @var{x})
9589 Converts a floating point value @var{x} into a double-precision integer
9590 which is then stored into @var{low} and @var{high}. If the value is not
9591 integral, it is truncated.
9594 @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})
9595 Converts a double-precision integer found in @var{low} and @var{high},
9596 into a floating point value which is then stored into @var{x}. The
9597 value is truncated to fit in mode @var{mode}.
9600 @node Mode Switching
9601 @section Mode Switching Instructions
9602 @cindex mode switching
9603 The following macros control mode switching optimizations:
9605 @defmac OPTIMIZE_MODE_SWITCHING (@var{entity})
9606 Define this macro if the port needs extra instructions inserted for mode
9607 switching in an optimizing compilation.
9609 For an example, the SH4 can perform both single and double precision
9610 floating point operations, but to perform a single precision operation,
9611 the FPSCR PR bit has to be cleared, while for a double precision
9612 operation, this bit has to be set. Changing the PR bit requires a general
9613 purpose register as a scratch register, hence these FPSCR sets have to
9614 be inserted before reload, i.e.@: you can't put this into instruction emitting
9615 or @code{TARGET_MACHINE_DEPENDENT_REORG}.
9617 You can have multiple entities that are mode-switched, and select at run time
9618 which entities actually need it. @code{OPTIMIZE_MODE_SWITCHING} should
9619 return nonzero for any @var{entity} that needs mode-switching.
9620 If you define this macro, you also have to define
9621 @code{NUM_MODES_FOR_MODE_SWITCHING}, @code{MODE_NEEDED},
9622 @code{MODE_PRIORITY_TO_MODE} and @code{EMIT_MODE_SET}.
9623 @code{MODE_AFTER}, @code{MODE_ENTRY}, and @code{MODE_EXIT}
9627 @defmac NUM_MODES_FOR_MODE_SWITCHING
9628 If you define @code{OPTIMIZE_MODE_SWITCHING}, you have to define this as
9629 initializer for an array of integers. Each initializer element
9630 N refers to an entity that needs mode switching, and specifies the number
9631 of different modes that might need to be set for this entity.
9632 The position of the initializer in the initializer---starting counting at
9633 zero---determines the integer that is used to refer to the mode-switched
9635 In macros that take mode arguments / yield a mode result, modes are
9636 represented as numbers 0 @dots{} N @minus{} 1. N is used to specify that no mode
9637 switch is needed / supplied.
9640 @defmac MODE_NEEDED (@var{entity}, @var{insn})
9641 @var{entity} is an integer specifying a mode-switched entity. If
9642 @code{OPTIMIZE_MODE_SWITCHING} is defined, you must define this macro to
9643 return an integer value not larger than the corresponding element in
9644 @code{NUM_MODES_FOR_MODE_SWITCHING}, to denote the mode that @var{entity} must
9645 be switched into prior to the execution of @var{insn}.
9648 @defmac MODE_AFTER (@var{mode}, @var{insn})
9649 If this macro is defined, it is evaluated for every @var{insn} during
9650 mode switching. It determines the mode that an insn results in (if
9651 different from the incoming mode).
9654 @defmac MODE_ENTRY (@var{entity})
9655 If this macro is defined, it is evaluated for every @var{entity} that needs
9656 mode switching. It should evaluate to an integer, which is a mode that
9657 @var{entity} is assumed to be switched to at function entry. If @code{MODE_ENTRY}
9658 is defined then @code{MODE_EXIT} must be defined.
9661 @defmac MODE_EXIT (@var{entity})
9662 If this macro is defined, it is evaluated for every @var{entity} that needs
9663 mode switching. It should evaluate to an integer, which is a mode that
9664 @var{entity} is assumed to be switched to at function exit. If @code{MODE_EXIT}
9665 is defined then @code{MODE_ENTRY} must be defined.
9668 @defmac MODE_PRIORITY_TO_MODE (@var{entity}, @var{n})
9669 This macro specifies the order in which modes for @var{entity} are processed.
9670 0 is the highest priority, @code{NUM_MODES_FOR_MODE_SWITCHING[@var{entity}] - 1} the
9671 lowest. The value of the macro should be an integer designating a mode
9672 for @var{entity}. For any fixed @var{entity}, @code{mode_priority_to_mode}
9673 (@var{entity}, @var{n}) shall be a bijection in 0 @dots{}
9674 @code{num_modes_for_mode_switching[@var{entity}] - 1}.
9677 @defmac EMIT_MODE_SET (@var{entity}, @var{mode}, @var{hard_regs_live})
9678 Generate one or more insns to set @var{entity} to @var{mode}.
9679 @var{hard_reg_live} is the set of hard registers live at the point where
9680 the insn(s) are to be inserted.
9683 @node Target Attributes
9684 @section Defining target-specific uses of @code{__attribute__}
9685 @cindex target attributes
9686 @cindex machine attributes
9687 @cindex attributes, target-specific
9689 Target-specific attributes may be defined for functions, data and types.
9690 These are described using the following target hooks; they also need to
9691 be documented in @file{extend.texi}.
9693 @hook TARGET_ATTRIBUTE_TABLE
9694 If defined, this target hook points to an array of @samp{struct
9695 attribute_spec} (defined in @file{tree.h}) specifying the machine
9696 specific attributes for this target and some of the restrictions on the
9697 entities to which these attributes are applied and the arguments they
9701 @hook TARGET_ATTRIBUTE_TAKES_IDENTIFIER_P
9702 If defined, this target hook is a function which returns true if the
9703 machine-specific attribute named @var{name} expects an identifier
9704 given as its first argument to be passed on as a plain identifier, not
9705 subjected to name lookup. If this is not defined, the default is
9706 false for all machine-specific attributes.
9709 @hook TARGET_COMP_TYPE_ATTRIBUTES
9710 If defined, this target hook is a function which returns zero if the attributes on
9711 @var{type1} and @var{type2} are incompatible, one if they are compatible,
9712 and two if they are nearly compatible (which causes a warning to be
9713 generated). If this is not defined, machine-specific attributes are
9714 supposed always to be compatible.
9717 @hook TARGET_SET_DEFAULT_TYPE_ATTRIBUTES
9718 If defined, this target hook is a function which assigns default attributes to
9719 the newly defined @var{type}.
9722 @hook TARGET_MERGE_TYPE_ATTRIBUTES
9723 Define this target hook if the merging of type attributes needs special
9724 handling. If defined, the result is a list of the combined
9725 @code{TYPE_ATTRIBUTES} of @var{type1} and @var{type2}. It is assumed
9726 that @code{comptypes} has already been called and returned 1. This
9727 function may call @code{merge_attributes} to handle machine-independent
9731 @hook TARGET_MERGE_DECL_ATTRIBUTES
9732 Define this target hook if the merging of decl attributes needs special
9733 handling. If defined, the result is a list of the combined
9734 @code{DECL_ATTRIBUTES} of @var{olddecl} and @var{newdecl}.
9735 @var{newdecl} is a duplicate declaration of @var{olddecl}. Examples of
9736 when this is needed are when one attribute overrides another, or when an
9737 attribute is nullified by a subsequent definition. This function may
9738 call @code{merge_attributes} to handle machine-independent merging.
9740 @findex TARGET_DLLIMPORT_DECL_ATTRIBUTES
9741 If the only target-specific handling you require is @samp{dllimport}
9742 for Microsoft Windows targets, you should define the macro
9743 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES} to @code{1}. The compiler
9744 will then define a function called
9745 @code{merge_dllimport_decl_attributes} which can then be defined as
9746 the expansion of @code{TARGET_MERGE_DECL_ATTRIBUTES}. You can also
9747 add @code{handle_dll_attribute} in the attribute table for your port
9748 to perform initial processing of the @samp{dllimport} and
9749 @samp{dllexport} attributes. This is done in @file{i386/cygwin.h} and
9750 @file{i386/i386.c}, for example.
9753 @hook TARGET_VALID_DLLIMPORT_ATTRIBUTE_P
9755 @defmac TARGET_DECLSPEC
9756 Define this macro to a nonzero value if you want to treat
9757 @code{__declspec(X)} as equivalent to @code{__attribute((X))}. By
9758 default, this behavior is enabled only for targets that define
9759 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES}. The current implementation
9760 of @code{__declspec} is via a built-in macro, but you should not rely
9761 on this implementation detail.
9764 @hook TARGET_INSERT_ATTRIBUTES
9765 Define this target hook if you want to be able to add attributes to a decl
9766 when it is being created. This is normally useful for back ends which
9767 wish to implement a pragma by using the attributes which correspond to
9768 the pragma's effect. The @var{node} argument is the decl which is being
9769 created. The @var{attr_ptr} argument is a pointer to the attribute list
9770 for this decl. The list itself should not be modified, since it may be
9771 shared with other decls, but attributes may be chained on the head of
9772 the list and @code{*@var{attr_ptr}} modified to point to the new
9773 attributes, or a copy of the list may be made if further changes are
9777 @hook TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P
9779 This target hook returns @code{true} if it is ok to inline @var{fndecl}
9780 into the current function, despite its having target-specific
9781 attributes, @code{false} otherwise. By default, if a function has a
9782 target specific attribute attached to it, it will not be inlined.
9785 @hook TARGET_OPTION_VALID_ATTRIBUTE_P
9786 This hook is called to parse the @code{attribute(option("..."))}, and
9787 it allows the function to set different target machine compile time
9788 options for the current function that might be different than the
9789 options specified on the command line. The hook should return
9790 @code{true} if the options are valid.
9792 The hook should set the @var{DECL_FUNCTION_SPECIFIC_TARGET} field in
9793 the function declaration to hold a pointer to a target specific
9794 @var{struct cl_target_option} structure.
9797 @hook TARGET_OPTION_SAVE
9798 This hook is called to save any additional target specific information
9799 in the @var{struct cl_target_option} structure for function specific
9801 @xref{Option file format}.
9804 @hook TARGET_OPTION_RESTORE
9805 This hook is called to restore any additional target specific
9806 information in the @var{struct cl_target_option} structure for
9807 function specific options.
9810 @hook TARGET_OPTION_PRINT
9811 This hook is called to print any additional target specific
9812 information in the @var{struct cl_target_option} structure for
9813 function specific options.
9816 @hook TARGET_OPTION_PRAGMA_PARSE
9817 This target hook parses the options for @code{#pragma GCC option} to
9818 set the machine specific options for functions that occur later in the
9819 input stream. The options should be the same as handled by the
9820 @code{TARGET_OPTION_VALID_ATTRIBUTE_P} hook.
9823 @hook TARGET_OPTION_OVERRIDE
9824 Sometimes certain combinations of command options do not make sense on
9825 a particular target machine. You can override the hook
9826 @code{TARGET_OPTION_OVERRIDE} to take account of this. This hooks is called
9827 once just after all the command options have been parsed.
9829 Don't use this hook to turn on various extra optimizations for
9830 @option{-O}. That is what @code{TARGET_OPTION_OPTIMIZATION} is for.
9832 If you need to do something whenever the optimization level is
9833 changed via the optimize attribute or pragma, see
9834 @code{TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE}
9837 @hook TARGET_CAN_INLINE_P
9838 This target hook returns @code{false} if the @var{caller} function
9839 cannot inline @var{callee}, based on target specific information. By
9840 default, inlining is not allowed if the callee function has function
9841 specific target options and the caller does not use the same options.
9845 @section Emulating TLS
9846 @cindex Emulated TLS
9848 For targets whose psABI does not provide Thread Local Storage via
9849 specific relocations and instruction sequences, an emulation layer is
9850 used. A set of target hooks allows this emulation layer to be
9851 configured for the requirements of a particular target. For instance
9852 the psABI may in fact specify TLS support in terms of an emulation
9855 The emulation layer works by creating a control object for every TLS
9856 object. To access the TLS object, a lookup function is provided
9857 which, when given the address of the control object, will return the
9858 address of the current thread's instance of the TLS object.
9860 @hook TARGET_EMUTLS_GET_ADDRESS
9861 Contains the name of the helper function that uses a TLS control
9862 object to locate a TLS instance. The default causes libgcc's
9863 emulated TLS helper function to be used.
9866 @hook TARGET_EMUTLS_REGISTER_COMMON
9867 Contains the name of the helper function that should be used at
9868 program startup to register TLS objects that are implicitly
9869 initialized to zero. If this is @code{NULL}, all TLS objects will
9870 have explicit initializers. The default causes libgcc's emulated TLS
9871 registration function to be used.
9874 @hook TARGET_EMUTLS_VAR_SECTION
9875 Contains the name of the section in which TLS control variables should
9876 be placed. The default of @code{NULL} allows these to be placed in
9880 @hook TARGET_EMUTLS_TMPL_SECTION
9881 Contains the name of the section in which TLS initializers should be
9882 placed. The default of @code{NULL} allows these to be placed in any
9886 @hook TARGET_EMUTLS_VAR_PREFIX
9887 Contains the prefix to be prepended to TLS control variable names.
9888 The default of @code{NULL} uses a target-specific prefix.
9891 @hook TARGET_EMUTLS_TMPL_PREFIX
9892 Contains the prefix to be prepended to TLS initializer objects. The
9893 default of @code{NULL} uses a target-specific prefix.
9896 @hook TARGET_EMUTLS_VAR_FIELDS
9897 Specifies a function that generates the FIELD_DECLs for a TLS control
9898 object type. @var{type} is the RECORD_TYPE the fields are for and
9899 @var{name} should be filled with the structure tag, if the default of
9900 @code{__emutls_object} is unsuitable. The default creates a type suitable
9901 for libgcc's emulated TLS function.
9904 @hook TARGET_EMUTLS_VAR_INIT
9905 Specifies a function that generates the CONSTRUCTOR to initialize a
9906 TLS control object. @var{var} is the TLS control object, @var{decl}
9907 is the TLS object and @var{tmpl_addr} is the address of the
9908 initializer. The default initializes libgcc's emulated TLS control object.
9911 @hook TARGET_EMUTLS_VAR_ALIGN_FIXED
9912 Specifies whether the alignment of TLS control variable objects is
9913 fixed and should not be increased as some backends may do to optimize
9914 single objects. The default is false.
9917 @hook TARGET_EMUTLS_DEBUG_FORM_TLS_ADDRESS
9918 Specifies whether a DWARF @code{DW_OP_form_tls_address} location descriptor
9919 may be used to describe emulated TLS control objects.
9922 @node MIPS Coprocessors
9923 @section Defining coprocessor specifics for MIPS targets.
9924 @cindex MIPS coprocessor-definition macros
9926 The MIPS specification allows MIPS implementations to have as many as 4
9927 coprocessors, each with as many as 32 private registers. GCC supports
9928 accessing these registers and transferring values between the registers
9929 and memory using asm-ized variables. For example:
9932 register unsigned int cp0count asm ("c0r1");
9938 (``c0r1'' is the default name of register 1 in coprocessor 0; alternate
9939 names may be added as described below, or the default names may be
9940 overridden entirely in @code{SUBTARGET_CONDITIONAL_REGISTER_USAGE}.)
9942 Coprocessor registers are assumed to be epilogue-used; sets to them will
9943 be preserved even if it does not appear that the register is used again
9944 later in the function.
9946 Another note: according to the MIPS spec, coprocessor 1 (if present) is
9947 the FPU@. One accesses COP1 registers through standard mips
9948 floating-point support; they are not included in this mechanism.
9950 There is one macro used in defining the MIPS coprocessor interface which
9951 you may want to override in subtargets; it is described below.
9953 @defmac ALL_COP_ADDITIONAL_REGISTER_NAMES
9954 A comma-separated list (with leading comma) of pairs describing the
9955 alternate names of coprocessor registers. The format of each entry should be
9957 @{ @var{alternatename}, @var{register_number}@}
9963 @section Parameters for Precompiled Header Validity Checking
9964 @cindex parameters, precompiled headers
9966 @hook TARGET_GET_PCH_VALIDITY
9967 This hook returns a pointer to the data needed by
9968 @code{TARGET_PCH_VALID_P} and sets
9969 @samp{*@var{sz}} to the size of the data in bytes.
9972 @hook TARGET_PCH_VALID_P
9973 This hook checks whether the options used to create a PCH file are
9974 compatible with the current settings. It returns @code{NULL}
9975 if so and a suitable error message if not. Error messages will
9976 be presented to the user and must be localized using @samp{_(@var{msg})}.
9978 @var{data} is the data that was returned by @code{TARGET_GET_PCH_VALIDITY}
9979 when the PCH file was created and @var{sz} is the size of that data in bytes.
9980 It's safe to assume that the data was created by the same version of the
9981 compiler, so no format checking is needed.
9983 The default definition of @code{default_pch_valid_p} should be
9984 suitable for most targets.
9987 @hook TARGET_CHECK_PCH_TARGET_FLAGS
9988 If this hook is nonnull, the default implementation of
9989 @code{TARGET_PCH_VALID_P} will use it to check for compatible values
9990 of @code{target_flags}. @var{pch_flags} specifies the value that
9991 @code{target_flags} had when the PCH file was created. The return
9992 value is the same as for @code{TARGET_PCH_VALID_P}.
9996 @section C++ ABI parameters
9997 @cindex parameters, c++ abi
9999 @hook TARGET_CXX_GUARD_TYPE
10000 Define this hook to override the integer type used for guard variables.
10001 These are used to implement one-time construction of static objects. The
10002 default is long_long_integer_type_node.
10005 @hook TARGET_CXX_GUARD_MASK_BIT
10006 This hook determines how guard variables are used. It should return
10007 @code{false} (the default) if the first byte should be used. A return value of
10008 @code{true} indicates that only the least significant bit should be used.
10011 @hook TARGET_CXX_GET_COOKIE_SIZE
10012 This hook returns the size of the cookie to use when allocating an array
10013 whose elements have the indicated @var{type}. Assumes that it is already
10014 known that a cookie is needed. The default is
10015 @code{max(sizeof (size_t), alignof(type))}, as defined in section 2.7 of the
10016 IA64/Generic C++ ABI@.
10019 @hook TARGET_CXX_COOKIE_HAS_SIZE
10020 This hook should return @code{true} if the element size should be stored in
10021 array cookies. The default is to return @code{false}.
10024 @hook TARGET_CXX_IMPORT_EXPORT_CLASS
10025 If defined by a backend this hook allows the decision made to export
10026 class @var{type} to be overruled. Upon entry @var{import_export}
10027 will contain 1 if the class is going to be exported, @minus{}1 if it is going
10028 to be imported and 0 otherwise. This function should return the
10029 modified value and perform any other actions necessary to support the
10030 backend's targeted operating system.
10033 @hook TARGET_CXX_CDTOR_RETURNS_THIS
10034 This hook should return @code{true} if constructors and destructors return
10035 the address of the object created/destroyed. The default is to return
10039 @hook TARGET_CXX_KEY_METHOD_MAY_BE_INLINE
10040 This hook returns true if the key method for a class (i.e., the method
10041 which, if defined in the current translation unit, causes the virtual
10042 table to be emitted) may be an inline function. Under the standard
10043 Itanium C++ ABI the key method may be an inline function so long as
10044 the function is not declared inline in the class definition. Under
10045 some variants of the ABI, an inline function can never be the key
10046 method. The default is to return @code{true}.
10049 @hook TARGET_CXX_DETERMINE_CLASS_DATA_VISIBILITY
10051 @hook TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT
10052 This hook returns true (the default) if virtual tables and other
10053 similar implicit class data objects are always COMDAT if they have
10054 external linkage. If this hook returns false, then class data for
10055 classes whose virtual table will be emitted in only one translation
10056 unit will not be COMDAT.
10059 @hook TARGET_CXX_LIBRARY_RTTI_COMDAT
10060 This hook returns true (the default) if the RTTI information for
10061 the basic types which is defined in the C++ runtime should always
10062 be COMDAT, false if it should not be COMDAT.
10065 @hook TARGET_CXX_USE_AEABI_ATEXIT
10066 This hook returns true if @code{__aeabi_atexit} (as defined by the ARM EABI)
10067 should be used to register static destructors when @option{-fuse-cxa-atexit}
10068 is in effect. The default is to return false to use @code{__cxa_atexit}.
10071 @hook TARGET_CXX_USE_ATEXIT_FOR_CXA_ATEXIT
10072 This hook returns true if the target @code{atexit} function can be used
10073 in the same manner as @code{__cxa_atexit} to register C++ static
10074 destructors. This requires that @code{atexit}-registered functions in
10075 shared libraries are run in the correct order when the libraries are
10076 unloaded. The default is to return false.
10079 @hook TARGET_CXX_ADJUST_CLASS_AT_DEFINITION
10081 @node Named Address Spaces
10082 @section Adding support for named address spaces
10083 @cindex named address spaces
10085 The draft technical report of the ISO/IEC JTC1 S22 WG14 N1275
10086 standards committee, @cite{Programming Languages - C - Extensions to
10087 support embedded processors}, specifies a syntax for embedded
10088 processors to specify alternate address spaces. You can configure a
10089 GCC port to support section 5.1 of the draft report to add support for
10090 address spaces other than the default address space. These address
10091 spaces are new keywords that are similar to the @code{volatile} and
10092 @code{const} type attributes.
10094 Pointers to named address spaces can have a different size than
10095 pointers to the generic address space.
10097 For example, the SPU port uses the @code{__ea} address space to refer
10098 to memory in the host processor, rather than memory local to the SPU
10099 processor. Access to memory in the @code{__ea} address space involves
10100 issuing DMA operations to move data between the host processor and the
10101 local processor memory address space. Pointers in the @code{__ea}
10102 address space are either 32 bits or 64 bits based on the
10103 @option{-mea32} or @option{-mea64} switches (native SPU pointers are
10106 Internally, address spaces are represented as a small integer in the
10107 range 0 to 15 with address space 0 being reserved for the generic
10110 To register a named address space qualifier keyword with the C front end,
10111 the target may call the @code{c_register_addr_space} routine. For example,
10112 the SPU port uses the following to declare @code{__ea} as the keyword for
10113 named address space #1:
10115 #define ADDR_SPACE_EA 1
10116 c_register_addr_space ("__ea", ADDR_SPACE_EA);
10119 @hook TARGET_ADDR_SPACE_POINTER_MODE
10120 Define this to return the machine mode to use for pointers to
10121 @var{address_space} if the target supports named address spaces.
10122 The default version of this hook returns @code{ptr_mode} for the
10123 generic address space only.
10126 @hook TARGET_ADDR_SPACE_ADDRESS_MODE
10127 Define this to return the machine mode to use for addresses in
10128 @var{address_space} if the target supports named address spaces.
10129 The default version of this hook returns @code{Pmode} for the
10130 generic address space only.
10133 @hook TARGET_ADDR_SPACE_VALID_POINTER_MODE
10134 Define this to return nonzero if the port can handle pointers
10135 with machine mode @var{mode} to address space @var{as}. This target
10136 hook is the same as the @code{TARGET_VALID_POINTER_MODE} target hook,
10137 except that it includes explicit named address space support. The default
10138 version of this hook returns true for the modes returned by either the
10139 @code{TARGET_ADDR_SPACE_POINTER_MODE} or @code{TARGET_ADDR_SPACE_ADDRESS_MODE}
10140 target hooks for the given address space.
10143 @hook TARGET_ADDR_SPACE_LEGITIMATE_ADDRESS_P
10144 Define this to return true if @var{exp} is a valid address for mode
10145 @var{mode} in the named address space @var{as}. The @var{strict}
10146 parameter says whether strict addressing is in effect after reload has
10147 finished. This target hook is the same as the
10148 @code{TARGET_LEGITIMATE_ADDRESS_P} target hook, except that it includes
10149 explicit named address space support.
10152 @hook TARGET_ADDR_SPACE_LEGITIMIZE_ADDRESS
10153 Define this to modify an invalid address @var{x} to be a valid address
10154 with mode @var{mode} in the named address space @var{as}. This target
10155 hook is the same as the @code{TARGET_LEGITIMIZE_ADDRESS} target hook,
10156 except that it includes explicit named address space support.
10159 @hook TARGET_ADDR_SPACE_SUBSET_P
10160 Define this to return whether the @var{subset} named address space is
10161 contained within the @var{superset} named address space. Pointers to
10162 a named address space that is a subset of another named address space
10163 will be converted automatically without a cast if used together in
10164 arithmetic operations. Pointers to a superset address space can be
10165 converted to pointers to a subset address space via explicit casts.
10168 @hook TARGET_ADDR_SPACE_CONVERT
10169 Define this to convert the pointer expression represented by the RTL
10170 @var{op} with type @var{from_type} that points to a named address
10171 space to a new pointer expression with type @var{to_type} that points
10172 to a different named address space. When this hook it called, it is
10173 guaranteed that one of the two address spaces is a subset of the other,
10174 as determined by the @code{TARGET_ADDR_SPACE_SUBSET_P} target hook.
10178 @section Miscellaneous Parameters
10179 @cindex parameters, miscellaneous
10181 @c prevent bad page break with this line
10182 Here are several miscellaneous parameters.
10184 @defmac HAS_LONG_COND_BRANCH
10185 Define this boolean macro to indicate whether or not your architecture
10186 has conditional branches that can span all of memory. It is used in
10187 conjunction with an optimization that partitions hot and cold basic
10188 blocks into separate sections of the executable. If this macro is
10189 set to false, gcc will convert any conditional branches that attempt
10190 to cross between sections into unconditional branches or indirect jumps.
10193 @defmac HAS_LONG_UNCOND_BRANCH
10194 Define this boolean macro to indicate whether or not your architecture
10195 has unconditional branches that can span all of memory. It is used in
10196 conjunction with an optimization that partitions hot and cold basic
10197 blocks into separate sections of the executable. If this macro is
10198 set to false, gcc will convert any unconditional branches that attempt
10199 to cross between sections into indirect jumps.
10202 @defmac CASE_VECTOR_MODE
10203 An alias for a machine mode name. This is the machine mode that
10204 elements of a jump-table should have.
10207 @defmac CASE_VECTOR_SHORTEN_MODE (@var{min_offset}, @var{max_offset}, @var{body})
10208 Optional: return the preferred mode for an @code{addr_diff_vec}
10209 when the minimum and maximum offset are known. If you define this,
10210 it enables extra code in branch shortening to deal with @code{addr_diff_vec}.
10211 To make this work, you also have to define @code{INSN_ALIGN} and
10212 make the alignment for @code{addr_diff_vec} explicit.
10213 The @var{body} argument is provided so that the offset_unsigned and scale
10214 flags can be updated.
10217 @defmac CASE_VECTOR_PC_RELATIVE
10218 Define this macro to be a C expression to indicate when jump-tables
10219 should contain relative addresses. You need not define this macro if
10220 jump-tables never contain relative addresses, or jump-tables should
10221 contain relative addresses only when @option{-fPIC} or @option{-fPIC}
10225 @hook TARGET_CASE_VALUES_THRESHOLD
10226 This function return the smallest number of different values for which it
10227 is best to use a jump-table instead of a tree of conditional branches.
10228 The default is four for machines with a @code{casesi} instruction and
10229 five otherwise. This is best for most machines.
10232 @defmac CASE_USE_BIT_TESTS
10233 Define this macro to be a C expression to indicate whether C switch
10234 statements may be implemented by a sequence of bit tests. This is
10235 advantageous on processors that can efficiently implement left shift
10236 of 1 by the number of bits held in a register, but inappropriate on
10237 targets that would require a loop. By default, this macro returns
10238 @code{true} if the target defines an @code{ashlsi3} pattern, and
10239 @code{false} otherwise.
10242 @defmac WORD_REGISTER_OPERATIONS
10243 Define this macro if operations between registers with integral mode
10244 smaller than a word are always performed on the entire register.
10245 Most RISC machines have this property and most CISC machines do not.
10248 @defmac LOAD_EXTEND_OP (@var{mem_mode})
10249 Define this macro to be a C expression indicating when insns that read
10250 memory in @var{mem_mode}, an integral mode narrower than a word, set the
10251 bits outside of @var{mem_mode} to be either the sign-extension or the
10252 zero-extension of the data read. Return @code{SIGN_EXTEND} for values
10253 of @var{mem_mode} for which the
10254 insn sign-extends, @code{ZERO_EXTEND} for which it zero-extends, and
10255 @code{UNKNOWN} for other modes.
10257 This macro is not called with @var{mem_mode} non-integral or with a width
10258 greater than or equal to @code{BITS_PER_WORD}, so you may return any
10259 value in this case. Do not define this macro if it would always return
10260 @code{UNKNOWN}. On machines where this macro is defined, you will normally
10261 define it as the constant @code{SIGN_EXTEND} or @code{ZERO_EXTEND}.
10263 You may return a non-@code{UNKNOWN} value even if for some hard registers
10264 the sign extension is not performed, if for the @code{REGNO_REG_CLASS}
10265 of these hard registers @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero
10266 when the @var{from} mode is @var{mem_mode} and the @var{to} mode is any
10267 integral mode larger than this but not larger than @code{word_mode}.
10269 You must return @code{UNKNOWN} if for some hard registers that allow this
10270 mode, @code{CANNOT_CHANGE_MODE_CLASS} says that they cannot change to
10271 @code{word_mode}, but that they can change to another integral mode that
10272 is larger then @var{mem_mode} but still smaller than @code{word_mode}.
10275 @defmac SHORT_IMMEDIATES_SIGN_EXTEND
10276 Define this macro if loading short immediate values into registers sign
10280 @defmac FIXUNS_TRUNC_LIKE_FIX_TRUNC
10281 Define this macro if the same instructions that convert a floating
10282 point number to a signed fixed point number also convert validly to an
10286 @hook TARGET_MIN_DIVISIONS_FOR_RECIP_MUL
10287 When @option{-ffast-math} is in effect, GCC tries to optimize
10288 divisions by the same divisor, by turning them into multiplications by
10289 the reciprocal. This target hook specifies the minimum number of divisions
10290 that should be there for GCC to perform the optimization for a variable
10291 of mode @var{mode}. The default implementation returns 3 if the machine
10292 has an instruction for the division, and 2 if it does not.
10296 The maximum number of bytes that a single instruction can move quickly
10297 between memory and registers or between two memory locations.
10300 @defmac MAX_MOVE_MAX
10301 The maximum number of bytes that a single instruction can move quickly
10302 between memory and registers or between two memory locations. If this
10303 is undefined, the default is @code{MOVE_MAX}. Otherwise, it is the
10304 constant value that is the largest value that @code{MOVE_MAX} can have
10308 @defmac SHIFT_COUNT_TRUNCATED
10309 A C expression that is nonzero if on this machine the number of bits
10310 actually used for the count of a shift operation is equal to the number
10311 of bits needed to represent the size of the object being shifted. When
10312 this macro is nonzero, the compiler will assume that it is safe to omit
10313 a sign-extend, zero-extend, and certain bitwise `and' instructions that
10314 truncates the count of a shift operation. On machines that have
10315 instructions that act on bit-fields at variable positions, which may
10316 include `bit test' instructions, a nonzero @code{SHIFT_COUNT_TRUNCATED}
10317 also enables deletion of truncations of the values that serve as
10318 arguments to bit-field instructions.
10320 If both types of instructions truncate the count (for shifts) and
10321 position (for bit-field operations), or if no variable-position bit-field
10322 instructions exist, you should define this macro.
10324 However, on some machines, such as the 80386 and the 680x0, truncation
10325 only applies to shift operations and not the (real or pretended)
10326 bit-field operations. Define @code{SHIFT_COUNT_TRUNCATED} to be zero on
10327 such machines. Instead, add patterns to the @file{md} file that include
10328 the implied truncation of the shift instructions.
10330 You need not define this macro if it would always have the value of zero.
10333 @anchor{TARGET_SHIFT_TRUNCATION_MASK}
10334 @hook TARGET_SHIFT_TRUNCATION_MASK
10335 This function describes how the standard shift patterns for @var{mode}
10336 deal with shifts by negative amounts or by more than the width of the mode.
10337 @xref{shift patterns}.
10339 On many machines, the shift patterns will apply a mask @var{m} to the
10340 shift count, meaning that a fixed-width shift of @var{x} by @var{y} is
10341 equivalent to an arbitrary-width shift of @var{x} by @var{y & m}. If
10342 this is true for mode @var{mode}, the function should return @var{m},
10343 otherwise it should return 0. A return value of 0 indicates that no
10344 particular behavior is guaranteed.
10346 Note that, unlike @code{SHIFT_COUNT_TRUNCATED}, this function does
10347 @emph{not} apply to general shift rtxes; it applies only to instructions
10348 that are generated by the named shift patterns.
10350 The default implementation of this function returns
10351 @code{GET_MODE_BITSIZE (@var{mode}) - 1} if @code{SHIFT_COUNT_TRUNCATED}
10352 and 0 otherwise. This definition is always safe, but if
10353 @code{SHIFT_COUNT_TRUNCATED} is false, and some shift patterns
10354 nevertheless truncate the shift count, you may get better code
10358 @defmac TRULY_NOOP_TRUNCATION (@var{outprec}, @var{inprec})
10359 A C expression which is nonzero if on this machine it is safe to
10360 ``convert'' an integer of @var{inprec} bits to one of @var{outprec}
10361 bits (where @var{outprec} is smaller than @var{inprec}) by merely
10362 operating on it as if it had only @var{outprec} bits.
10364 On many machines, this expression can be 1.
10366 @c rearranged this, removed the phrase "it is reported that". this was
10367 @c to fix an overfull hbox. --mew 10feb93
10368 When @code{TRULY_NOOP_TRUNCATION} returns 1 for a pair of sizes for
10369 modes for which @code{MODES_TIEABLE_P} is 0, suboptimal code can result.
10370 If this is the case, making @code{TRULY_NOOP_TRUNCATION} return 0 in
10371 such cases may improve things.
10374 @hook TARGET_MODE_REP_EXTENDED
10375 The representation of an integral mode can be such that the values
10376 are always extended to a wider integral mode. Return
10377 @code{SIGN_EXTEND} if values of @var{mode} are represented in
10378 sign-extended form to @var{rep_mode}. Return @code{UNKNOWN}
10379 otherwise. (Currently, none of the targets use zero-extended
10380 representation this way so unlike @code{LOAD_EXTEND_OP},
10381 @code{TARGET_MODE_REP_EXTENDED} is expected to return either
10382 @code{SIGN_EXTEND} or @code{UNKNOWN}. Also no target extends
10383 @var{mode} to @var{rep_mode} so that @var{rep_mode} is not the next
10384 widest integral mode and currently we take advantage of this fact.)
10386 Similarly to @code{LOAD_EXTEND_OP} you may return a non-@code{UNKNOWN}
10387 value even if the extension is not performed on certain hard registers
10388 as long as for the @code{REGNO_REG_CLASS} of these hard registers
10389 @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero.
10391 Note that @code{TARGET_MODE_REP_EXTENDED} and @code{LOAD_EXTEND_OP}
10392 describe two related properties. If you define
10393 @code{TARGET_MODE_REP_EXTENDED (mode, word_mode)} you probably also want
10394 to define @code{LOAD_EXTEND_OP (mode)} to return the same type of
10397 In order to enforce the representation of @code{mode},
10398 @code{TRULY_NOOP_TRUNCATION} should return false when truncating to
10402 @defmac STORE_FLAG_VALUE
10403 A C expression describing the value returned by a comparison operator
10404 with an integral mode and stored by a store-flag instruction
10405 (@samp{cstore@var{mode}4}) when the condition is true. This description must
10406 apply to @emph{all} the @samp{cstore@var{mode}4} patterns and all the
10407 comparison operators whose results have a @code{MODE_INT} mode.
10409 A value of 1 or @minus{}1 means that the instruction implementing the
10410 comparison operator returns exactly 1 or @minus{}1 when the comparison is true
10411 and 0 when the comparison is false. Otherwise, the value indicates
10412 which bits of the result are guaranteed to be 1 when the comparison is
10413 true. This value is interpreted in the mode of the comparison
10414 operation, which is given by the mode of the first operand in the
10415 @samp{cstore@var{mode}4} pattern. Either the low bit or the sign bit of
10416 @code{STORE_FLAG_VALUE} be on. Presently, only those bits are used by
10419 If @code{STORE_FLAG_VALUE} is neither 1 or @minus{}1, the compiler will
10420 generate code that depends only on the specified bits. It can also
10421 replace comparison operators with equivalent operations if they cause
10422 the required bits to be set, even if the remaining bits are undefined.
10423 For example, on a machine whose comparison operators return an
10424 @code{SImode} value and where @code{STORE_FLAG_VALUE} is defined as
10425 @samp{0x80000000}, saying that just the sign bit is relevant, the
10429 (ne:SI (and:SI @var{x} (const_int @var{power-of-2})) (const_int 0))
10433 can be converted to
10436 (ashift:SI @var{x} (const_int @var{n}))
10440 where @var{n} is the appropriate shift count to move the bit being
10441 tested into the sign bit.
10443 There is no way to describe a machine that always sets the low-order bit
10444 for a true value, but does not guarantee the value of any other bits,
10445 but we do not know of any machine that has such an instruction. If you
10446 are trying to port GCC to such a machine, include an instruction to
10447 perform a logical-and of the result with 1 in the pattern for the
10448 comparison operators and let us know at @email{gcc@@gcc.gnu.org}.
10450 Often, a machine will have multiple instructions that obtain a value
10451 from a comparison (or the condition codes). Here are rules to guide the
10452 choice of value for @code{STORE_FLAG_VALUE}, and hence the instructions
10457 Use the shortest sequence that yields a valid definition for
10458 @code{STORE_FLAG_VALUE}. It is more efficient for the compiler to
10459 ``normalize'' the value (convert it to, e.g., 1 or 0) than for the
10460 comparison operators to do so because there may be opportunities to
10461 combine the normalization with other operations.
10464 For equal-length sequences, use a value of 1 or @minus{}1, with @minus{}1 being
10465 slightly preferred on machines with expensive jumps and 1 preferred on
10469 As a second choice, choose a value of @samp{0x80000001} if instructions
10470 exist that set both the sign and low-order bits but do not define the
10474 Otherwise, use a value of @samp{0x80000000}.
10477 Many machines can produce both the value chosen for
10478 @code{STORE_FLAG_VALUE} and its negation in the same number of
10479 instructions. On those machines, you should also define a pattern for
10480 those cases, e.g., one matching
10483 (set @var{A} (neg:@var{m} (ne:@var{m} @var{B} @var{C})))
10486 Some machines can also perform @code{and} or @code{plus} operations on
10487 condition code values with less instructions than the corresponding
10488 @samp{cstore@var{mode}4} insn followed by @code{and} or @code{plus}. On those
10489 machines, define the appropriate patterns. Use the names @code{incscc}
10490 and @code{decscc}, respectively, for the patterns which perform
10491 @code{plus} or @code{minus} operations on condition code values. See
10492 @file{rs6000.md} for some examples. The GNU Superoptimizer can be used to
10493 find such instruction sequences on other machines.
10495 If this macro is not defined, the default value, 1, is used. You need
10496 not define @code{STORE_FLAG_VALUE} if the machine has no store-flag
10497 instructions, or if the value generated by these instructions is 1.
10500 @defmac FLOAT_STORE_FLAG_VALUE (@var{mode})
10501 A C expression that gives a nonzero @code{REAL_VALUE_TYPE} value that is
10502 returned when comparison operators with floating-point results are true.
10503 Define this macro on machines that have comparison operations that return
10504 floating-point values. If there are no such operations, do not define
10508 @defmac VECTOR_STORE_FLAG_VALUE (@var{mode})
10509 A C expression that gives a rtx representing the nonzero true element
10510 for vector comparisons. The returned rtx should be valid for the inner
10511 mode of @var{mode} which is guaranteed to be a vector mode. Define
10512 this macro on machines that have vector comparison operations that
10513 return a vector result. If there are no such operations, do not define
10514 this macro. Typically, this macro is defined as @code{const1_rtx} or
10515 @code{constm1_rtx}. This macro may return @code{NULL_RTX} to prevent
10516 the compiler optimizing such vector comparison operations for the
10520 @defmac CLZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
10521 @defmacx CTZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
10522 A C expression that indicates whether the architecture defines a value
10523 for @code{clz} or @code{ctz} with a zero operand.
10524 A result of @code{0} indicates the value is undefined.
10525 If the value is defined for only the RTL expression, the macro should
10526 evaluate to @code{1}; if the value applies also to the corresponding optab
10527 entry (which is normally the case if it expands directly into
10528 the corresponding RTL), then the macro should evaluate to @code{2}.
10529 In the cases where the value is defined, @var{value} should be set to
10532 If this macro is not defined, the value of @code{clz} or
10533 @code{ctz} at zero is assumed to be undefined.
10535 This macro must be defined if the target's expansion for @code{ffs}
10536 relies on a particular value to get correct results. Otherwise it
10537 is not necessary, though it may be used to optimize some corner cases, and
10538 to provide a default expansion for the @code{ffs} optab.
10540 Note that regardless of this macro the ``definedness'' of @code{clz}
10541 and @code{ctz} at zero do @emph{not} extend to the builtin functions
10542 visible to the user. Thus one may be free to adjust the value at will
10543 to match the target expansion of these operations without fear of
10548 An alias for the machine mode for pointers. On most machines, define
10549 this to be the integer mode corresponding to the width of a hardware
10550 pointer; @code{SImode} on 32-bit machine or @code{DImode} on 64-bit machines.
10551 On some machines you must define this to be one of the partial integer
10552 modes, such as @code{PSImode}.
10554 The width of @code{Pmode} must be at least as large as the value of
10555 @code{POINTER_SIZE}. If it is not equal, you must define the macro
10556 @code{POINTERS_EXTEND_UNSIGNED} to specify how pointers are extended
10560 @defmac FUNCTION_MODE
10561 An alias for the machine mode used for memory references to functions
10562 being called, in @code{call} RTL expressions. On most CISC machines,
10563 where an instruction can begin at any byte address, this should be
10564 @code{QImode}. On most RISC machines, where all instructions have fixed
10565 size and alignment, this should be a mode with the same size and alignment
10566 as the machine instruction words - typically @code{SImode} or @code{HImode}.
10569 @defmac STDC_0_IN_SYSTEM_HEADERS
10570 In normal operation, the preprocessor expands @code{__STDC__} to the
10571 constant 1, to signify that GCC conforms to ISO Standard C@. On some
10572 hosts, like Solaris, the system compiler uses a different convention,
10573 where @code{__STDC__} is normally 0, but is 1 if the user specifies
10574 strict conformance to the C Standard.
10576 Defining @code{STDC_0_IN_SYSTEM_HEADERS} makes GNU CPP follows the host
10577 convention when processing system header files, but when processing user
10578 files @code{__STDC__} will always expand to 1.
10581 @defmac NO_IMPLICIT_EXTERN_C
10582 Define this macro if the system header files support C++ as well as C@.
10583 This macro inhibits the usual method of using system header files in
10584 C++, which is to pretend that the file's contents are enclosed in
10585 @samp{extern "C" @{@dots{}@}}.
10590 @defmac REGISTER_TARGET_PRAGMAS ()
10591 Define this macro if you want to implement any target-specific pragmas.
10592 If defined, it is a C expression which makes a series of calls to
10593 @code{c_register_pragma} or @code{c_register_pragma_with_expansion}
10594 for each pragma. The macro may also do any
10595 setup required for the pragmas.
10597 The primary reason to define this macro is to provide compatibility with
10598 other compilers for the same target. In general, we discourage
10599 definition of target-specific pragmas for GCC@.
10601 If the pragma can be implemented by attributes then you should consider
10602 defining the target hook @samp{TARGET_INSERT_ATTRIBUTES} as well.
10604 Preprocessor macros that appear on pragma lines are not expanded. All
10605 @samp{#pragma} directives that do not match any registered pragma are
10606 silently ignored, unless the user specifies @option{-Wunknown-pragmas}.
10609 @deftypefun void c_register_pragma (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
10610 @deftypefunx void c_register_pragma_with_expansion (const char *@var{space}, const char *@var{name}, void (*@var{callback}) (struct cpp_reader *))
10612 Each call to @code{c_register_pragma} or
10613 @code{c_register_pragma_with_expansion} establishes one pragma. The
10614 @var{callback} routine will be called when the preprocessor encounters a
10618 #pragma [@var{space}] @var{name} @dots{}
10621 @var{space} is the case-sensitive namespace of the pragma, or
10622 @code{NULL} to put the pragma in the global namespace. The callback
10623 routine receives @var{pfile} as its first argument, which can be passed
10624 on to cpplib's functions if necessary. You can lex tokens after the
10625 @var{name} by calling @code{pragma_lex}. Tokens that are not read by the
10626 callback will be silently ignored. The end of the line is indicated by
10627 a token of type @code{CPP_EOF}. Macro expansion occurs on the
10628 arguments of pragmas registered with
10629 @code{c_register_pragma_with_expansion} but not on the arguments of
10630 pragmas registered with @code{c_register_pragma}.
10632 Note that the use of @code{pragma_lex} is specific to the C and C++
10633 compilers. It will not work in the Java or Fortran compilers, or any
10634 other language compilers for that matter. Thus if @code{pragma_lex} is going
10635 to be called from target-specific code, it must only be done so when
10636 building the C and C++ compilers. This can be done by defining the
10637 variables @code{c_target_objs} and @code{cxx_target_objs} in the
10638 target entry in the @file{config.gcc} file. These variables should name
10639 the target-specific, language-specific object file which contains the
10640 code that uses @code{pragma_lex}. Note it will also be necessary to add a
10641 rule to the makefile fragment pointed to by @code{tmake_file} that shows
10642 how to build this object file.
10645 @defmac HANDLE_PRAGMA_PACK_WITH_EXPANSION
10646 Define this macro if macros should be expanded in the
10647 arguments of @samp{#pragma pack}.
10650 @hook TARGET_HANDLE_PRAGMA_EXTERN_PREFIX
10652 @defmac TARGET_DEFAULT_PACK_STRUCT
10653 If your target requires a structure packing default other than 0 (meaning
10654 the machine default), define this macro to the necessary value (in bytes).
10655 This must be a value that would also be valid to use with
10656 @samp{#pragma pack()} (that is, a small power of two).
10659 @defmac DOLLARS_IN_IDENTIFIERS
10660 Define this macro to control use of the character @samp{$} in
10661 identifier names for the C family of languages. 0 means @samp{$} is
10662 not allowed by default; 1 means it is allowed. 1 is the default;
10663 there is no need to define this macro in that case.
10666 @defmac NO_DOLLAR_IN_LABEL
10667 Define this macro if the assembler does not accept the character
10668 @samp{$} in label names. By default constructors and destructors in
10669 G++ have @samp{$} in the identifiers. If this macro is defined,
10670 @samp{.} is used instead.
10673 @defmac NO_DOT_IN_LABEL
10674 Define this macro if the assembler does not accept the character
10675 @samp{.} in label names. By default constructors and destructors in G++
10676 have names that use @samp{.}. If this macro is defined, these names
10677 are rewritten to avoid @samp{.}.
10680 @defmac INSN_SETS_ARE_DELAYED (@var{insn})
10681 Define this macro as a C expression that is nonzero if it is safe for the
10682 delay slot scheduler to place instructions in the delay slot of @var{insn},
10683 even if they appear to use a resource set or clobbered in @var{insn}.
10684 @var{insn} is always a @code{jump_insn} or an @code{insn}; GCC knows that
10685 every @code{call_insn} has this behavior. On machines where some @code{insn}
10686 or @code{jump_insn} is really a function call and hence has this behavior,
10687 you should define this macro.
10689 You need not define this macro if it would always return zero.
10692 @defmac INSN_REFERENCES_ARE_DELAYED (@var{insn})
10693 Define this macro as a C expression that is nonzero if it is safe for the
10694 delay slot scheduler to place instructions in the delay slot of @var{insn},
10695 even if they appear to set or clobber a resource referenced in @var{insn}.
10696 @var{insn} is always a @code{jump_insn} or an @code{insn}. On machines where
10697 some @code{insn} or @code{jump_insn} is really a function call and its operands
10698 are registers whose use is actually in the subroutine it calls, you should
10699 define this macro. Doing so allows the delay slot scheduler to move
10700 instructions which copy arguments into the argument registers into the delay
10701 slot of @var{insn}.
10703 You need not define this macro if it would always return zero.
10706 @defmac MULTIPLE_SYMBOL_SPACES
10707 Define this macro as a C expression that is nonzero if, in some cases,
10708 global symbols from one translation unit may not be bound to undefined
10709 symbols in another translation unit without user intervention. For
10710 instance, under Microsoft Windows symbols must be explicitly imported
10711 from shared libraries (DLLs).
10713 You need not define this macro if it would always evaluate to zero.
10716 @hook TARGET_MD_ASM_CLOBBERS
10717 This target hook should add to @var{clobbers} @code{STRING_CST} trees for
10718 any hard regs the port wishes to automatically clobber for an asm.
10719 It should return the result of the last @code{tree_cons} used to add a
10720 clobber. The @var{outputs}, @var{inputs} and @var{clobber} lists are the
10721 corresponding parameters to the asm and may be inspected to avoid
10722 clobbering a register that is an input or output of the asm. You can use
10723 @code{tree_overlaps_hard_reg_set}, declared in @file{tree.h}, to test
10724 for overlap with regards to asm-declared registers.
10727 @defmac MATH_LIBRARY
10728 Define this macro as a C string constant for the linker argument to link
10729 in the system math library, minus the initial @samp{"-l"}, or
10730 @samp{""} if the target does not have a
10731 separate math library.
10733 You need only define this macro if the default of @samp{"m"} is wrong.
10736 @defmac LIBRARY_PATH_ENV
10737 Define this macro as a C string constant for the environment variable that
10738 specifies where the linker should look for libraries.
10740 You need only define this macro if the default of @samp{"LIBRARY_PATH"}
10744 @defmac TARGET_POSIX_IO
10745 Define this macro if the target supports the following POSIX@ file
10746 functions, access, mkdir and file locking with fcntl / F_SETLKW@.
10747 Defining @code{TARGET_POSIX_IO} will enable the test coverage code
10748 to use file locking when exiting a program, which avoids race conditions
10749 if the program has forked. It will also create directories at run-time
10750 for cross-profiling.
10753 @defmac MAX_CONDITIONAL_EXECUTE
10755 A C expression for the maximum number of instructions to execute via
10756 conditional execution instructions instead of a branch. A value of
10757 @code{BRANCH_COST}+1 is the default if the machine does not use cc0, and
10758 1 if it does use cc0.
10761 @defmac IFCVT_MODIFY_TESTS (@var{ce_info}, @var{true_expr}, @var{false_expr})
10762 Used if the target needs to perform machine-dependent modifications on the
10763 conditionals used for turning basic blocks into conditionally executed code.
10764 @var{ce_info} points to a data structure, @code{struct ce_if_block}, which
10765 contains information about the currently processed blocks. @var{true_expr}
10766 and @var{false_expr} are the tests that are used for converting the
10767 then-block and the else-block, respectively. Set either @var{true_expr} or
10768 @var{false_expr} to a null pointer if the tests cannot be converted.
10771 @defmac IFCVT_MODIFY_MULTIPLE_TESTS (@var{ce_info}, @var{bb}, @var{true_expr}, @var{false_expr})
10772 Like @code{IFCVT_MODIFY_TESTS}, but used when converting more complicated
10773 if-statements into conditions combined by @code{and} and @code{or} operations.
10774 @var{bb} contains the basic block that contains the test that is currently
10775 being processed and about to be turned into a condition.
10778 @defmac IFCVT_MODIFY_INSN (@var{ce_info}, @var{pattern}, @var{insn})
10779 A C expression to modify the @var{PATTERN} of an @var{INSN} that is to
10780 be converted to conditional execution format. @var{ce_info} points to
10781 a data structure, @code{struct ce_if_block}, which contains information
10782 about the currently processed blocks.
10785 @defmac IFCVT_MODIFY_FINAL (@var{ce_info})
10786 A C expression to perform any final machine dependent modifications in
10787 converting code to conditional execution. The involved basic blocks
10788 can be found in the @code{struct ce_if_block} structure that is pointed
10789 to by @var{ce_info}.
10792 @defmac IFCVT_MODIFY_CANCEL (@var{ce_info})
10793 A C expression to cancel any machine dependent modifications in
10794 converting code to conditional execution. The involved basic blocks
10795 can be found in the @code{struct ce_if_block} structure that is pointed
10796 to by @var{ce_info}.
10799 @defmac IFCVT_INIT_EXTRA_FIELDS (@var{ce_info})
10800 A C expression to initialize any extra fields in a @code{struct ce_if_block}
10801 structure, which are defined by the @code{IFCVT_EXTRA_FIELDS} macro.
10804 @defmac IFCVT_EXTRA_FIELDS
10805 If defined, it should expand to a set of field declarations that will be
10806 added to the @code{struct ce_if_block} structure. These should be initialized
10807 by the @code{IFCVT_INIT_EXTRA_FIELDS} macro.
10810 @hook TARGET_MACHINE_DEPENDENT_REORG
10811 If non-null, this hook performs a target-specific pass over the
10812 instruction stream. The compiler will run it at all optimization levels,
10813 just before the point at which it normally does delayed-branch scheduling.
10815 The exact purpose of the hook varies from target to target. Some use
10816 it to do transformations that are necessary for correctness, such as
10817 laying out in-function constant pools or avoiding hardware hazards.
10818 Others use it as an opportunity to do some machine-dependent optimizations.
10820 You need not implement the hook if it has nothing to do. The default
10821 definition is null.
10824 @hook TARGET_INIT_BUILTINS
10825 Define this hook if you have any machine-specific built-in functions
10826 that need to be defined. It should be a function that performs the
10829 Machine specific built-in functions can be useful to expand special machine
10830 instructions that would otherwise not normally be generated because
10831 they have no equivalent in the source language (for example, SIMD vector
10832 instructions or prefetch instructions).
10834 To create a built-in function, call the function
10835 @code{lang_hooks.builtin_function}
10836 which is defined by the language front end. You can use any type nodes set
10837 up by @code{build_common_tree_nodes};
10838 only language front ends that use those two functions will call
10839 @samp{TARGET_INIT_BUILTINS}.
10842 @hook TARGET_BUILTIN_DECL
10843 Define this hook if you have any machine-specific built-in functions
10844 that need to be defined. It should be a function that returns the
10845 builtin function declaration for the builtin function code @var{code}.
10846 If there is no such builtin and it cannot be initialized at this time
10847 if @var{initialize_p} is true the function should return @code{NULL_TREE}.
10848 If @var{code} is out of range the function should return
10849 @code{error_mark_node}.
10852 @hook TARGET_EXPAND_BUILTIN
10854 Expand a call to a machine specific built-in function that was set up by
10855 @samp{TARGET_INIT_BUILTINS}. @var{exp} is the expression for the
10856 function call; the result should go to @var{target} if that is
10857 convenient, and have mode @var{mode} if that is convenient.
10858 @var{subtarget} may be used as the target for computing one of
10859 @var{exp}'s operands. @var{ignore} is nonzero if the value is to be
10860 ignored. This function should return the result of the call to the
10864 @hook TARGET_RESOLVE_OVERLOADED_BUILTIN
10865 Select a replacement for a machine specific built-in function that
10866 was set up by @samp{TARGET_INIT_BUILTINS}. This is done
10867 @emph{before} regular type checking, and so allows the target to
10868 implement a crude form of function overloading. @var{fndecl} is the
10869 declaration of the built-in function. @var{arglist} is the list of
10870 arguments passed to the built-in function. The result is a
10871 complete expression that implements the operation, usually
10872 another @code{CALL_EXPR}.
10873 @var{arglist} really has type @samp{VEC(tree,gc)*}
10876 @hook TARGET_FOLD_BUILTIN
10877 Fold a call to a machine specific built-in function that was set up by
10878 @samp{TARGET_INIT_BUILTINS}. @var{fndecl} is the declaration of the
10879 built-in function. @var{n_args} is the number of arguments passed to
10880 the function; the arguments themselves are pointed to by @var{argp}.
10881 The result is another tree containing a simplified expression for the
10882 call's result. If @var{ignore} is true the value will be ignored.
10885 @hook TARGET_INVALID_WITHIN_DOLOOP
10887 Take an instruction in @var{insn} and return NULL if it is valid within a
10888 low-overhead loop, otherwise return a string explaining why doloop
10889 could not be applied.
10891 Many targets use special registers for low-overhead looping. For any
10892 instruction that clobbers these this function should return a string indicating
10893 the reason why the doloop could not be applied.
10894 By default, the RTL loop optimizer does not use a present doloop pattern for
10895 loops containing function calls or branch on table instructions.
10898 @defmac MD_CAN_REDIRECT_BRANCH (@var{branch1}, @var{branch2})
10900 Take a branch insn in @var{branch1} and another in @var{branch2}.
10901 Return true if redirecting @var{branch1} to the destination of
10902 @var{branch2} is possible.
10904 On some targets, branches may have a limited range. Optimizing the
10905 filling of delay slots can result in branches being redirected, and this
10906 may in turn cause a branch offset to overflow.
10909 @hook TARGET_COMMUTATIVE_P
10910 This target hook returns @code{true} if @var{x} is considered to be commutative.
10911 Usually, this is just COMMUTATIVE_P (@var{x}), but the HP PA doesn't consider
10912 PLUS to be commutative inside a MEM@. @var{outer_code} is the rtx code
10913 of the enclosing rtl, if known, otherwise it is UNKNOWN.
10916 @hook TARGET_ALLOCATE_INITIAL_VALUE
10918 When the initial value of a hard register has been copied in a pseudo
10919 register, it is often not necessary to actually allocate another register
10920 to this pseudo register, because the original hard register or a stack slot
10921 it has been saved into can be used. @code{TARGET_ALLOCATE_INITIAL_VALUE}
10922 is called at the start of register allocation once for each hard register
10923 that had its initial value copied by using
10924 @code{get_func_hard_reg_initial_val} or @code{get_hard_reg_initial_val}.
10925 Possible values are @code{NULL_RTX}, if you don't want
10926 to do any special allocation, a @code{REG} rtx---that would typically be
10927 the hard register itself, if it is known not to be clobbered---or a
10929 If you are returning a @code{MEM}, this is only a hint for the allocator;
10930 it might decide to use another register anyways.
10931 You may use @code{current_function_leaf_function} in the hook, functions
10932 that use @code{REG_N_SETS}, to determine if the hard
10933 register in question will not be clobbered.
10934 The default value of this hook is @code{NULL}, which disables any special
10938 @hook TARGET_UNSPEC_MAY_TRAP_P
10939 This target hook returns nonzero if @var{x}, an @code{unspec} or
10940 @code{unspec_volatile} operation, might cause a trap. Targets can use
10941 this hook to enhance precision of analysis for @code{unspec} and
10942 @code{unspec_volatile} operations. You may call @code{may_trap_p_1}
10943 to analyze inner elements of @var{x} in which case @var{flags} should be
10947 @hook TARGET_SET_CURRENT_FUNCTION
10948 The compiler invokes this hook whenever it changes its current function
10949 context (@code{cfun}). You can define this function if
10950 the back end needs to perform any initialization or reset actions on a
10951 per-function basis. For example, it may be used to implement function
10952 attributes that affect register usage or code generation patterns.
10953 The argument @var{decl} is the declaration for the new function context,
10954 and may be null to indicate that the compiler has left a function context
10955 and is returning to processing at the top level.
10956 The default hook function does nothing.
10958 GCC sets @code{cfun} to a dummy function context during initialization of
10959 some parts of the back end. The hook function is not invoked in this
10960 situation; you need not worry about the hook being invoked recursively,
10961 or when the back end is in a partially-initialized state.
10962 @code{cfun} might be @code{NULL} to indicate processing at top level,
10963 outside of any function scope.
10966 @defmac TARGET_OBJECT_SUFFIX
10967 Define this macro to be a C string representing the suffix for object
10968 files on your target machine. If you do not define this macro, GCC will
10969 use @samp{.o} as the suffix for object files.
10972 @defmac TARGET_EXECUTABLE_SUFFIX
10973 Define this macro to be a C string representing the suffix to be
10974 automatically added to executable files on your target machine. If you
10975 do not define this macro, GCC will use the null string as the suffix for
10979 @defmac COLLECT_EXPORT_LIST
10980 If defined, @code{collect2} will scan the individual object files
10981 specified on its command line and create an export list for the linker.
10982 Define this macro for systems like AIX, where the linker discards
10983 object files that are not referenced from @code{main} and uses export
10987 @defmac MODIFY_JNI_METHOD_CALL (@var{mdecl})
10988 Define this macro to a C expression representing a variant of the
10989 method call @var{mdecl}, if Java Native Interface (JNI) methods
10990 must be invoked differently from other methods on your target.
10991 For example, on 32-bit Microsoft Windows, JNI methods must be invoked using
10992 the @code{stdcall} calling convention and this macro is then
10993 defined as this expression:
10996 build_type_attribute_variant (@var{mdecl},
10998 (get_identifier ("stdcall"),
11003 @hook TARGET_CANNOT_MODIFY_JUMPS_P
11004 This target hook returns @code{true} past the point in which new jump
11005 instructions could be created. On machines that require a register for
11006 every jump such as the SHmedia ISA of SH5, this point would typically be
11007 reload, so this target hook should be defined to a function such as:
11011 cannot_modify_jumps_past_reload_p ()
11013 return (reload_completed || reload_in_progress);
11018 @hook TARGET_BRANCH_TARGET_REGISTER_CLASS
11019 This target hook returns a register class for which branch target register
11020 optimizations should be applied. All registers in this class should be
11021 usable interchangeably. After reload, registers in this class will be
11022 re-allocated and loads will be hoisted out of loops and be subjected
11023 to inter-block scheduling.
11026 @hook TARGET_BRANCH_TARGET_REGISTER_CALLEE_SAVED
11027 Branch target register optimization will by default exclude callee-saved
11029 that are not already live during the current function; if this target hook
11030 returns true, they will be included. The target code must than make sure
11031 that all target registers in the class returned by
11032 @samp{TARGET_BRANCH_TARGET_REGISTER_CLASS} that might need saving are
11033 saved. @var{after_prologue_epilogue_gen} indicates if prologues and
11034 epilogues have already been generated. Note, even if you only return
11035 true when @var{after_prologue_epilogue_gen} is false, you still are likely
11036 to have to make special provisions in @code{INITIAL_ELIMINATION_OFFSET}
11037 to reserve space for caller-saved target registers.
11040 @hook TARGET_HAVE_CONDITIONAL_EXECUTION
11041 This target hook returns true if the target supports conditional execution.
11042 This target hook is required only when the target has several different
11043 modes and they have different conditional execution capability, such as ARM.
11046 @hook TARGET_LOOP_UNROLL_ADJUST
11047 This target hook returns a new value for the number of times @var{loop}
11048 should be unrolled. The parameter @var{nunroll} is the number of times
11049 the loop is to be unrolled. The parameter @var{loop} is a pointer to
11050 the loop, which is going to be checked for unrolling. This target hook
11051 is required only when the target has special constraints like maximum
11052 number of memory accesses.
11055 @defmac POWI_MAX_MULTS
11056 If defined, this macro is interpreted as a signed integer C expression
11057 that specifies the maximum number of floating point multiplications
11058 that should be emitted when expanding exponentiation by an integer
11059 constant inline. When this value is defined, exponentiation requiring
11060 more than this number of multiplications is implemented by calling the
11061 system library's @code{pow}, @code{powf} or @code{powl} routines.
11062 The default value places no upper bound on the multiplication count.
11065 @deftypefn Macro void TARGET_EXTRA_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
11066 This target hook should register any extra include files for the
11067 target. The parameter @var{stdinc} indicates if normal include files
11068 are present. The parameter @var{sysroot} is the system root directory.
11069 The parameter @var{iprefix} is the prefix for the gcc directory.
11072 @deftypefn Macro void TARGET_EXTRA_PRE_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
11073 This target hook should register any extra include files for the
11074 target before any standard headers. The parameter @var{stdinc}
11075 indicates if normal include files are present. The parameter
11076 @var{sysroot} is the system root directory. The parameter
11077 @var{iprefix} is the prefix for the gcc directory.
11080 @deftypefn Macro void TARGET_OPTF (char *@var{path})
11081 This target hook should register special include paths for the target.
11082 The parameter @var{path} is the include to register. On Darwin
11083 systems, this is used for Framework includes, which have semantics
11084 that are different from @option{-I}.
11087 @defmac bool TARGET_USE_LOCAL_THUNK_ALIAS_P (tree @var{fndecl})
11088 This target macro returns @code{true} if it is safe to use a local alias
11089 for a virtual function @var{fndecl} when constructing thunks,
11090 @code{false} otherwise. By default, the macro returns @code{true} for all
11091 functions, if a target supports aliases (i.e.@: defines
11092 @code{ASM_OUTPUT_DEF}), @code{false} otherwise,
11095 @defmac TARGET_FORMAT_TYPES
11096 If defined, this macro is the name of a global variable containing
11097 target-specific format checking information for the @option{-Wformat}
11098 option. The default is to have no target-specific format checks.
11101 @defmac TARGET_N_FORMAT_TYPES
11102 If defined, this macro is the number of entries in
11103 @code{TARGET_FORMAT_TYPES}.
11106 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES
11107 If defined, this macro is the name of a global variable containing
11108 target-specific format overrides for the @option{-Wformat} option. The
11109 default is to have no target-specific format overrides. If defined,
11110 @code{TARGET_FORMAT_TYPES} must be defined, too.
11113 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES_COUNT
11114 If defined, this macro specifies the number of entries in
11115 @code{TARGET_OVERRIDES_FORMAT_ATTRIBUTES}.
11118 @defmac TARGET_OVERRIDES_FORMAT_INIT
11119 If defined, this macro specifies the optional initialization
11120 routine for target specific customizations of the system printf
11121 and scanf formatter settings.
11124 @hook TARGET_RELAXED_ORDERING
11125 If set to @code{true}, means that the target's memory model does not
11126 guarantee that loads which do not depend on one another will access
11127 main memory in the order of the instruction stream; if ordering is
11128 important, an explicit memory barrier must be used. This is true of
11129 many recent processors which implement a policy of ``relaxed,''
11130 ``weak,'' or ``release'' memory consistency, such as Alpha, PowerPC,
11131 and ia64. The default is @code{false}.
11134 @hook TARGET_INVALID_ARG_FOR_UNPROTOTYPED_FN
11135 If defined, this macro returns the diagnostic message when it is
11136 illegal to pass argument @var{val} to function @var{funcdecl}
11137 with prototype @var{typelist}.
11140 @hook TARGET_INVALID_CONVERSION
11141 If defined, this macro returns the diagnostic message when it is
11142 invalid to convert from @var{fromtype} to @var{totype}, or @code{NULL}
11143 if validity should be determined by the front end.
11146 @hook TARGET_INVALID_UNARY_OP
11147 If defined, this macro returns the diagnostic message when it is
11148 invalid to apply operation @var{op} (where unary plus is denoted by
11149 @code{CONVERT_EXPR}) to an operand of type @var{type}, or @code{NULL}
11150 if validity should be determined by the front end.
11153 @hook TARGET_INVALID_BINARY_OP
11154 If defined, this macro returns the diagnostic message when it is
11155 invalid to apply operation @var{op} to operands of types @var{type1}
11156 and @var{type2}, or @code{NULL} if validity should be determined by
11160 @hook TARGET_INVALID_PARAMETER_TYPE
11161 If defined, this macro returns the diagnostic message when it is
11162 invalid for functions to include parameters of type @var{type},
11163 or @code{NULL} if validity should be determined by
11164 the front end. This is currently used only by the C and C++ front ends.
11167 @hook TARGET_INVALID_RETURN_TYPE
11168 If defined, this macro returns the diagnostic message when it is
11169 invalid for functions to have return type @var{type},
11170 or @code{NULL} if validity should be determined by
11171 the front end. This is currently used only by the C and C++ front ends.
11174 @hook TARGET_PROMOTED_TYPE
11175 If defined, this target hook returns the type to which values of
11176 @var{type} should be promoted when they appear in expressions,
11177 analogous to the integer promotions, or @code{NULL_TREE} to use the
11178 front end's normal promotion rules. This hook is useful when there are
11179 target-specific types with special promotion rules.
11180 This is currently used only by the C and C++ front ends.
11183 @hook TARGET_CONVERT_TO_TYPE
11184 If defined, this hook returns the result of converting @var{expr} to
11185 @var{type}. It should return the converted expression,
11186 or @code{NULL_TREE} to apply the front end's normal conversion rules.
11187 This hook is useful when there are target-specific types with special
11189 This is currently used only by the C and C++ front ends.
11192 @defmac TARGET_USE_JCR_SECTION
11193 This macro determines whether to use the JCR section to register Java
11194 classes. By default, TARGET_USE_JCR_SECTION is defined to 1 if both
11195 SUPPORTS_WEAK and TARGET_HAVE_NAMED_SECTIONS are true, else 0.
11199 This macro determines the size of the objective C jump buffer for the
11200 NeXT runtime. By default, OBJC_JBLEN is defined to an innocuous value.
11203 @defmac LIBGCC2_UNWIND_ATTRIBUTE
11204 Define this macro if any target-specific attributes need to be attached
11205 to the functions in @file{libgcc} that provide low-level support for
11206 call stack unwinding. It is used in declarations in @file{unwind-generic.h}
11207 and the associated definitions of those functions.
11210 @hook TARGET_UPDATE_STACK_BOUNDARY
11211 Define this macro to update the current function stack boundary if
11215 @hook TARGET_GET_DRAP_RTX
11216 This hook should return an rtx for Dynamic Realign Argument Pointer (DRAP) if a
11217 different argument pointer register is needed to access the function's
11218 argument list due to stack realignment. Return @code{NULL} if no DRAP
11222 @hook TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS
11223 When optimization is disabled, this hook indicates whether or not
11224 arguments should be allocated to stack slots. Normally, GCC allocates
11225 stacks slots for arguments when not optimizing in order to make
11226 debugging easier. However, when a function is declared with
11227 @code{__attribute__((naked))}, there is no stack frame, and the compiler
11228 cannot safely move arguments from the registers in which they are passed
11229 to the stack. Therefore, this hook should return true in general, but
11230 false for naked functions. The default implementation always returns true.
11233 @hook TARGET_CONST_ANCHOR
11234 On some architectures it can take multiple instructions to synthesize
11235 a constant. If there is another constant already in a register that
11236 is close enough in value then it is preferable that the new constant
11237 is computed from this register using immediate addition or
11238 subtraction. We accomplish this through CSE. Besides the value of
11239 the constant we also add a lower and an upper constant anchor to the
11240 available expressions. These are then queried when encountering new
11241 constants. The anchors are computed by rounding the constant up and
11242 down to a multiple of the value of @code{TARGET_CONST_ANCHOR}.
11243 @code{TARGET_CONST_ANCHOR} should be the maximum positive value
11244 accepted by immediate-add plus one. We currently assume that the
11245 value of @code{TARGET_CONST_ANCHOR} is a power of 2. For example, on
11246 MIPS, where add-immediate takes a 16-bit signed value,
11247 @code{TARGET_CONST_ANCHOR} is set to @samp{0x8000}. The default value
11248 is zero, which disables this optimization. @end deftypevr