1 @c Copyright (C) 1988-2014 Free Software Foundation, Inc.
2 @c This is part of the GCC manual.
3 @c For copying conditions, see the file gcc.texi.
6 @chapter Target Description Macros and Functions
7 @cindex machine description macros
8 @cindex target description macros
9 @cindex macros, target description
10 @cindex @file{tm.h} macros
12 In addition to the file @file{@var{machine}.md}, a machine description
13 includes a C header file conventionally given the name
14 @file{@var{machine}.h} and a C source file named @file{@var{machine}.c}.
15 The header file defines numerous macros that convey the information
16 about the target machine that does not fit into the scheme of the
17 @file{.md} file. The file @file{tm.h} should be a link to
18 @file{@var{machine}.h}. The header file @file{config.h} includes
19 @file{tm.h} and most compiler source files include @file{config.h}. The
20 source file defines a variable @code{targetm}, which is a structure
21 containing pointers to functions and data relating to the target
22 machine. @file{@var{machine}.c} should also contain their definitions,
23 if they are not defined elsewhere in GCC, and other functions called
24 through the macros defined in the @file{.h} file.
27 * Target Structure:: The @code{targetm} variable.
28 * Driver:: Controlling how the driver runs the compilation passes.
29 * Run-time Target:: Defining @samp{-m} options like @option{-m68000} and @option{-m68020}.
30 * Per-Function Data:: Defining data structures for per-function information.
31 * Storage Layout:: Defining sizes and alignments of data.
32 * Type Layout:: Defining sizes and properties of basic user data types.
33 * Registers:: Naming and describing the hardware registers.
34 * Register Classes:: Defining the classes of hardware registers.
35 * Stack and Calling:: Defining which way the stack grows and by how much.
36 * Varargs:: Defining the varargs macros.
37 * Trampolines:: Code set up at run time to enter a nested function.
38 * Library Calls:: Controlling how library routines are implicitly called.
39 * Addressing Modes:: Defining addressing modes valid for memory operands.
40 * Anchored Addresses:: Defining how @option{-fsection-anchors} should work.
41 * Condition Code:: Defining how insns update the condition code.
42 * Costs:: Defining relative costs of different operations.
43 * Scheduling:: Adjusting the behavior of the instruction scheduler.
44 * Sections:: Dividing storage into text, data, and other sections.
45 * PIC:: Macros for position independent code.
46 * Assembler Format:: Defining how to write insns and pseudo-ops to output.
47 * Debugging Info:: Defining the format of debugging output.
48 * Floating Point:: Handling floating point for cross-compilers.
49 * Mode Switching:: Insertion of mode-switching instructions.
50 * Target Attributes:: Defining target-specific uses of @code{__attribute__}.
51 * Emulated TLS:: Emulated TLS support.
52 * MIPS Coprocessors:: MIPS coprocessor support and how to customize it.
53 * PCH Target:: Validity checking for precompiled headers.
54 * C++ ABI:: Controlling C++ ABI changes.
55 * Named Address Spaces:: Adding support for named address spaces
56 * Misc:: Everything else.
59 @node Target Structure
60 @section The Global @code{targetm} Variable
62 @cindex target functions
64 @deftypevar {struct gcc_target} targetm
65 The target @file{.c} file must define the global @code{targetm} variable
66 which contains pointers to functions and data relating to the target
67 machine. The variable is declared in @file{target.h};
68 @file{target-def.h} defines the macro @code{TARGET_INITIALIZER} which is
69 used to initialize the variable, and macros for the default initializers
70 for elements of the structure. The @file{.c} file should override those
71 macros for which the default definition is inappropriate. For example:
74 #include "target-def.h"
76 /* @r{Initialize the GCC target structure.} */
78 #undef TARGET_COMP_TYPE_ATTRIBUTES
79 #define TARGET_COMP_TYPE_ATTRIBUTES @var{machine}_comp_type_attributes
81 struct gcc_target targetm = TARGET_INITIALIZER;
85 Where a macro should be defined in the @file{.c} file in this manner to
86 form part of the @code{targetm} structure, it is documented below as a
87 ``Target Hook'' with a prototype. Many macros will change in future
88 from being defined in the @file{.h} file to being part of the
89 @code{targetm} structure.
91 Similarly, there is a @code{targetcm} variable for hooks that are
92 specific to front ends for C-family languages, documented as ``C
93 Target Hook''. This is declared in @file{c-family/c-target.h}, the
94 initializer @code{TARGETCM_INITIALIZER} in
95 @file{c-family/c-target-def.h}. If targets initialize @code{targetcm}
96 themselves, they should set @code{target_has_targetcm=yes} in
97 @file{config.gcc}; otherwise a default definition is used.
99 Similarly, there is a @code{targetm_common} variable for hooks that
100 are shared between the compiler driver and the compilers proper,
101 documented as ``Common Target Hook''. This is declared in
102 @file{common/common-target.h}, the initializer
103 @code{TARGETM_COMMON_INITIALIZER} in
104 @file{common/common-target-def.h}. If targets initialize
105 @code{targetm_common} themselves, they should set
106 @code{target_has_targetm_common=yes} in @file{config.gcc}; otherwise a
107 default definition is used.
110 @section Controlling the Compilation Driver, @file{gcc}
112 @cindex controlling the compilation driver
114 @c prevent bad page break with this line
115 You can control the compilation driver.
117 @defmac DRIVER_SELF_SPECS
118 A list of specs for the driver itself. It should be a suitable
119 initializer for an array of strings, with no surrounding braces.
121 The driver applies these specs to its own command line between loading
122 default @file{specs} files (but not command-line specified ones) and
123 choosing the multilib directory or running any subcommands. It
124 applies them in the order given, so each spec can depend on the
125 options added by earlier ones. It is also possible to remove options
126 using @samp{%<@var{option}} in the usual way.
128 This macro can be useful when a port has several interdependent target
129 options. It provides a way of standardizing the command line so
130 that the other specs are easier to write.
132 Do not define this macro if it does not need to do anything.
135 @defmac OPTION_DEFAULT_SPECS
136 A list of specs used to support configure-time default options (i.e.@:
137 @option{--with} options) in the driver. It should be a suitable initializer
138 for an array of structures, each containing two strings, without the
139 outermost pair of surrounding braces.
141 The first item in the pair is the name of the default. This must match
142 the code in @file{config.gcc} for the target. The second item is a spec
143 to apply if a default with this name was specified. The string
144 @samp{%(VALUE)} in the spec will be replaced by the value of the default
145 everywhere it occurs.
147 The driver will apply these specs to its own command line between loading
148 default @file{specs} files and processing @code{DRIVER_SELF_SPECS}, using
149 the same mechanism as @code{DRIVER_SELF_SPECS}.
151 Do not define this macro if it does not need to do anything.
155 A C string constant that tells the GCC driver program options to
156 pass to CPP@. It can also specify how to translate options you
157 give to GCC into options for GCC to pass to the CPP@.
159 Do not define this macro if it does not need to do anything.
162 @defmac CPLUSPLUS_CPP_SPEC
163 This macro is just like @code{CPP_SPEC}, but is used for C++, rather
164 than C@. If you do not define this macro, then the value of
165 @code{CPP_SPEC} (if any) will be used instead.
169 A C string constant that tells the GCC driver program options to
170 pass to @code{cc1}, @code{cc1plus}, @code{f771}, and the other language
172 It can also specify how to translate options you give to GCC into options
173 for GCC to pass to front ends.
175 Do not define this macro if it does not need to do anything.
179 A C string constant that tells the GCC driver program options to
180 pass to @code{cc1plus}. It can also specify how to translate options you
181 give to GCC into options for GCC to pass to the @code{cc1plus}.
183 Do not define this macro if it does not need to do anything.
184 Note that everything defined in CC1_SPEC is already passed to
185 @code{cc1plus} so there is no need to duplicate the contents of
186 CC1_SPEC in CC1PLUS_SPEC@.
190 A C string constant that tells the GCC driver program options to
191 pass to the assembler. It can also specify how to translate options
192 you give to GCC into options for GCC to pass to the assembler.
193 See the file @file{sun3.h} for an example of this.
195 Do not define this macro if it does not need to do anything.
198 @defmac ASM_FINAL_SPEC
199 A C string constant that tells the GCC driver program how to
200 run any programs which cleanup after the normal assembler.
201 Normally, this is not needed. See the file @file{mips.h} for
204 Do not define this macro if it does not need to do anything.
207 @defmac AS_NEEDS_DASH_FOR_PIPED_INPUT
208 Define this macro, with no value, if the driver should give the assembler
209 an argument consisting of a single dash, @option{-}, to instruct it to
210 read from its standard input (which will be a pipe connected to the
211 output of the compiler proper). This argument is given after any
212 @option{-o} option specifying the name of the output file.
214 If you do not define this macro, the assembler is assumed to read its
215 standard input if given no non-option arguments. If your assembler
216 cannot read standard input at all, use a @samp{%@{pipe:%e@}} construct;
217 see @file{mips.h} for instance.
221 A C string constant that tells the GCC driver program options to
222 pass to the linker. It can also specify how to translate options you
223 give to GCC into options for GCC to pass to the linker.
225 Do not define this macro if it does not need to do anything.
229 Another C string constant used much like @code{LINK_SPEC}. The difference
230 between the two is that @code{LIB_SPEC} is used at the end of the
231 command given to the linker.
233 If this macro is not defined, a default is provided that
234 loads the standard C library from the usual place. See @file{gcc.c}.
238 Another C string constant that tells the GCC driver program
239 how and when to place a reference to @file{libgcc.a} into the
240 linker command line. This constant is placed both before and after
241 the value of @code{LIB_SPEC}.
243 If this macro is not defined, the GCC driver provides a default that
244 passes the string @option{-lgcc} to the linker.
247 @defmac REAL_LIBGCC_SPEC
248 By default, if @code{ENABLE_SHARED_LIBGCC} is defined, the
249 @code{LIBGCC_SPEC} is not directly used by the driver program but is
250 instead modified to refer to different versions of @file{libgcc.a}
251 depending on the values of the command line flags @option{-static},
252 @option{-shared}, @option{-static-libgcc}, and @option{-shared-libgcc}. On
253 targets where these modifications are inappropriate, define
254 @code{REAL_LIBGCC_SPEC} instead. @code{REAL_LIBGCC_SPEC} tells the
255 driver how to place a reference to @file{libgcc} on the link command
256 line, but, unlike @code{LIBGCC_SPEC}, it is used unmodified.
259 @defmac USE_LD_AS_NEEDED
260 A macro that controls the modifications to @code{LIBGCC_SPEC}
261 mentioned in @code{REAL_LIBGCC_SPEC}. If nonzero, a spec will be
262 generated that uses @option{--as-needed} or equivalent options and the
263 shared @file{libgcc} in place of the
264 static exception handler library, when linking without any of
265 @code{-static}, @code{-static-libgcc}, or @code{-shared-libgcc}.
269 If defined, this C string constant is added to @code{LINK_SPEC}.
270 When @code{USE_LD_AS_NEEDED} is zero or undefined, it also affects
271 the modifications to @code{LIBGCC_SPEC} mentioned in
272 @code{REAL_LIBGCC_SPEC}.
275 @defmac STARTFILE_SPEC
276 Another C string constant used much like @code{LINK_SPEC}. The
277 difference between the two is that @code{STARTFILE_SPEC} is used at
278 the very beginning of the command given to the linker.
280 If this macro is not defined, a default is provided that loads the
281 standard C startup file from the usual place. See @file{gcc.c}.
285 Another C string constant used much like @code{LINK_SPEC}. The
286 difference between the two is that @code{ENDFILE_SPEC} is used at
287 the very end of the command given to the linker.
289 Do not define this macro if it does not need to do anything.
292 @defmac THREAD_MODEL_SPEC
293 GCC @code{-v} will print the thread model GCC was configured to use.
294 However, this doesn't work on platforms that are multilibbed on thread
295 models, such as AIX 4.3. On such platforms, define
296 @code{THREAD_MODEL_SPEC} such that it evaluates to a string without
297 blanks that names one of the recognized thread models. @code{%*}, the
298 default value of this macro, will expand to the value of
299 @code{thread_file} set in @file{config.gcc}.
302 @defmac SYSROOT_SUFFIX_SPEC
303 Define this macro to add a suffix to the target sysroot when GCC is
304 configured with a sysroot. This will cause GCC to search for usr/lib,
305 et al, within sysroot+suffix.
308 @defmac SYSROOT_HEADERS_SUFFIX_SPEC
309 Define this macro to add a headers_suffix to the target sysroot when
310 GCC is configured with a sysroot. This will cause GCC to pass the
311 updated sysroot+headers_suffix to CPP, causing it to search for
312 usr/include, et al, within sysroot+headers_suffix.
316 Define this macro to provide additional specifications to put in the
317 @file{specs} file that can be used in various specifications like
320 The definition should be an initializer for an array of structures,
321 containing a string constant, that defines the specification name, and a
322 string constant that provides the specification.
324 Do not define this macro if it does not need to do anything.
326 @code{EXTRA_SPECS} is useful when an architecture contains several
327 related targets, which have various @code{@dots{}_SPECS} which are similar
328 to each other, and the maintainer would like one central place to keep
331 For example, the PowerPC System V.4 targets use @code{EXTRA_SPECS} to
332 define either @code{_CALL_SYSV} when the System V calling sequence is
333 used or @code{_CALL_AIX} when the older AIX-based calling sequence is
336 The @file{config/rs6000/rs6000.h} target file defines:
339 #define EXTRA_SPECS \
340 @{ "cpp_sysv_default", CPP_SYSV_DEFAULT @},
342 #define CPP_SYS_DEFAULT ""
345 The @file{config/rs6000/sysv.h} target file defines:
349 "%@{posix: -D_POSIX_SOURCE @} \
350 %@{mcall-sysv: -D_CALL_SYSV @} \
351 %@{!mcall-sysv: %(cpp_sysv_default) @} \
352 %@{msoft-float: -D_SOFT_FLOAT@} %@{mcpu=403: -D_SOFT_FLOAT@}"
354 #undef CPP_SYSV_DEFAULT
355 #define CPP_SYSV_DEFAULT "-D_CALL_SYSV"
358 while the @file{config/rs6000/eabiaix.h} target file defines
359 @code{CPP_SYSV_DEFAULT} as:
362 #undef CPP_SYSV_DEFAULT
363 #define CPP_SYSV_DEFAULT "-D_CALL_AIX"
367 @defmac LINK_LIBGCC_SPECIAL_1
368 Define this macro if the driver program should find the library
369 @file{libgcc.a}. If you do not define this macro, the driver program will pass
370 the argument @option{-lgcc} to tell the linker to do the search.
373 @defmac LINK_GCC_C_SEQUENCE_SPEC
374 The sequence in which libgcc and libc are specified to the linker.
375 By default this is @code{%G %L %G}.
378 @defmac LINK_COMMAND_SPEC
379 A C string constant giving the complete command line need to execute the
380 linker. When you do this, you will need to update your port each time a
381 change is made to the link command line within @file{gcc.c}. Therefore,
382 define this macro only if you need to completely redefine the command
383 line for invoking the linker and there is no other way to accomplish
384 the effect you need. Overriding this macro may be avoidable by overriding
385 @code{LINK_GCC_C_SEQUENCE_SPEC} instead.
388 @deftypevr {Common Target Hook} bool TARGET_ALWAYS_STRIP_DOTDOT
389 True if @file{..} components should always be removed from directory names computed relative to GCC's internal directories, false (default) if such components should be preserved and directory names containing them passed to other tools such as the linker.
392 @defmac MULTILIB_DEFAULTS
393 Define this macro as a C expression for the initializer of an array of
394 string to tell the driver program which options are defaults for this
395 target and thus do not need to be handled specially when using
396 @code{MULTILIB_OPTIONS}.
398 Do not define this macro if @code{MULTILIB_OPTIONS} is not defined in
399 the target makefile fragment or if none of the options listed in
400 @code{MULTILIB_OPTIONS} are set by default.
401 @xref{Target Fragment}.
404 @defmac RELATIVE_PREFIX_NOT_LINKDIR
405 Define this macro to tell @command{gcc} that it should only translate
406 a @option{-B} prefix into a @option{-L} linker option if the prefix
407 indicates an absolute file name.
410 @defmac MD_EXEC_PREFIX
411 If defined, this macro is an additional prefix to try after
412 @code{STANDARD_EXEC_PREFIX}. @code{MD_EXEC_PREFIX} is not searched
413 when the compiler is built as a cross
414 compiler. If you define @code{MD_EXEC_PREFIX}, then be sure to add it
415 to the list of directories used to find the assembler in @file{configure.in}.
418 @defmac STANDARD_STARTFILE_PREFIX
419 Define this macro as a C string constant if you wish to override the
420 standard choice of @code{libdir} as the default prefix to
421 try when searching for startup files such as @file{crt0.o}.
422 @code{STANDARD_STARTFILE_PREFIX} is not searched when the compiler
423 is built as a cross compiler.
426 @defmac STANDARD_STARTFILE_PREFIX_1
427 Define this macro as a C string constant if you wish to override the
428 standard choice of @code{/lib} as a prefix to try after the default prefix
429 when searching for startup files such as @file{crt0.o}.
430 @code{STANDARD_STARTFILE_PREFIX_1} is not searched when the compiler
431 is built as a cross compiler.
434 @defmac STANDARD_STARTFILE_PREFIX_2
435 Define this macro as a C string constant if you wish to override the
436 standard choice of @code{/lib} as yet another prefix to try after the
437 default prefix when searching for startup files such as @file{crt0.o}.
438 @code{STANDARD_STARTFILE_PREFIX_2} is not searched when the compiler
439 is built as a cross compiler.
442 @defmac MD_STARTFILE_PREFIX
443 If defined, this macro supplies an additional prefix to try after the
444 standard prefixes. @code{MD_EXEC_PREFIX} is not searched when the
445 compiler is built as a cross compiler.
448 @defmac MD_STARTFILE_PREFIX_1
449 If defined, this macro supplies yet another prefix to try after the
450 standard prefixes. It is not searched when the compiler is built as a
454 @defmac INIT_ENVIRONMENT
455 Define this macro as a C string constant if you wish to set environment
456 variables for programs called by the driver, such as the assembler and
457 loader. The driver passes the value of this macro to @code{putenv} to
458 initialize the necessary environment variables.
461 @defmac LOCAL_INCLUDE_DIR
462 Define this macro as a C string constant if you wish to override the
463 standard choice of @file{/usr/local/include} as the default prefix to
464 try when searching for local header files. @code{LOCAL_INCLUDE_DIR}
465 comes before @code{NATIVE_SYSTEM_HEADER_DIR} (set in
466 @file{config.gcc}, normally @file{/usr/include}) in the search order.
468 Cross compilers do not search either @file{/usr/local/include} or its
472 @defmac NATIVE_SYSTEM_HEADER_COMPONENT
473 The ``component'' corresponding to @code{NATIVE_SYSTEM_HEADER_DIR}.
474 See @code{INCLUDE_DEFAULTS}, below, for the description of components.
475 If you do not define this macro, no component is used.
478 @defmac INCLUDE_DEFAULTS
479 Define this macro if you wish to override the entire default search path
480 for include files. For a native compiler, the default search path
481 usually consists of @code{GCC_INCLUDE_DIR}, @code{LOCAL_INCLUDE_DIR},
482 @code{GPLUSPLUS_INCLUDE_DIR}, and
483 @code{NATIVE_SYSTEM_HEADER_DIR}. In addition, @code{GPLUSPLUS_INCLUDE_DIR}
484 and @code{GCC_INCLUDE_DIR} are defined automatically by @file{Makefile},
485 and specify private search areas for GCC@. The directory
486 @code{GPLUSPLUS_INCLUDE_DIR} is used only for C++ programs.
488 The definition should be an initializer for an array of structures.
489 Each array element should have four elements: the directory name (a
490 string constant), the component name (also a string constant), a flag
491 for C++-only directories,
492 and a flag showing that the includes in the directory don't need to be
493 wrapped in @code{extern @samp{C}} when compiling C++. Mark the end of
494 the array with a null element.
496 The component name denotes what GNU package the include file is part of,
497 if any, in all uppercase letters. For example, it might be @samp{GCC}
498 or @samp{BINUTILS}. If the package is part of a vendor-supplied
499 operating system, code the component name as @samp{0}.
501 For example, here is the definition used for VAX/VMS:
504 #define INCLUDE_DEFAULTS \
506 @{ "GNU_GXX_INCLUDE:", "G++", 1, 1@}, \
507 @{ "GNU_CC_INCLUDE:", "GCC", 0, 0@}, \
508 @{ "SYS$SYSROOT:[SYSLIB.]", 0, 0, 0@}, \
515 Here is the order of prefixes tried for exec files:
519 Any prefixes specified by the user with @option{-B}.
522 The environment variable @code{GCC_EXEC_PREFIX} or, if @code{GCC_EXEC_PREFIX}
523 is not set and the compiler has not been installed in the configure-time
524 @var{prefix}, the location in which the compiler has actually been installed.
527 The directories specified by the environment variable @code{COMPILER_PATH}.
530 The macro @code{STANDARD_EXEC_PREFIX}, if the compiler has been installed
531 in the configured-time @var{prefix}.
534 The location @file{/usr/libexec/gcc/}, but only if this is a native compiler.
537 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
540 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
544 Here is the order of prefixes tried for startfiles:
548 Any prefixes specified by the user with @option{-B}.
551 The environment variable @code{GCC_EXEC_PREFIX} or its automatically determined
552 value based on the installed toolchain location.
555 The directories specified by the environment variable @code{LIBRARY_PATH}
556 (or port-specific name; native only, cross compilers do not use this).
559 The macro @code{STANDARD_EXEC_PREFIX}, but only if the toolchain is installed
560 in the configured @var{prefix} or this is a native compiler.
563 The location @file{/usr/lib/gcc/}, but only if this is a native compiler.
566 The macro @code{MD_EXEC_PREFIX}, if defined, but only if this is a native
570 The macro @code{MD_STARTFILE_PREFIX}, if defined, but only if this is a
571 native compiler, or we have a target system root.
574 The macro @code{MD_STARTFILE_PREFIX_1}, if defined, but only if this is a
575 native compiler, or we have a target system root.
578 The macro @code{STANDARD_STARTFILE_PREFIX}, with any sysroot modifications.
579 If this path is relative it will be prefixed by @code{GCC_EXEC_PREFIX} and
580 the machine suffix or @code{STANDARD_EXEC_PREFIX} and the machine suffix.
583 The macro @code{STANDARD_STARTFILE_PREFIX_1}, but only if this is a native
584 compiler, or we have a target system root. The default for this macro is
588 The macro @code{STANDARD_STARTFILE_PREFIX_2}, but only if this is a native
589 compiler, or we have a target system root. The default for this macro is
593 @node Run-time Target
594 @section Run-time Target Specification
595 @cindex run-time target specification
596 @cindex predefined macros
597 @cindex target specifications
599 @c prevent bad page break with this line
600 Here are run-time target specifications.
602 @defmac TARGET_CPU_CPP_BUILTINS ()
603 This function-like macro expands to a block of code that defines
604 built-in preprocessor macros and assertions for the target CPU, using
605 the functions @code{builtin_define}, @code{builtin_define_std} and
606 @code{builtin_assert}. When the front end
607 calls this macro it provides a trailing semicolon, and since it has
608 finished command line option processing your code can use those
611 @code{builtin_assert} takes a string in the form you pass to the
612 command-line option @option{-A}, such as @code{cpu=mips}, and creates
613 the assertion. @code{builtin_define} takes a string in the form
614 accepted by option @option{-D} and unconditionally defines the macro.
616 @code{builtin_define_std} takes a string representing the name of an
617 object-like macro. If it doesn't lie in the user's namespace,
618 @code{builtin_define_std} defines it unconditionally. Otherwise, it
619 defines a version with two leading underscores, and another version
620 with two leading and trailing underscores, and defines the original
621 only if an ISO standard was not requested on the command line. For
622 example, passing @code{unix} defines @code{__unix}, @code{__unix__}
623 and possibly @code{unix}; passing @code{_mips} defines @code{__mips},
624 @code{__mips__} and possibly @code{_mips}, and passing @code{_ABI64}
625 defines only @code{_ABI64}.
627 You can also test for the C dialect being compiled. The variable
628 @code{c_language} is set to one of @code{clk_c}, @code{clk_cplusplus}
629 or @code{clk_objective_c}. Note that if we are preprocessing
630 assembler, this variable will be @code{clk_c} but the function-like
631 macro @code{preprocessing_asm_p()} will return true, so you might want
632 to check for that first. If you need to check for strict ANSI, the
633 variable @code{flag_iso} can be used. The function-like macro
634 @code{preprocessing_trad_p()} can be used to check for traditional
638 @defmac TARGET_OS_CPP_BUILTINS ()
639 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
640 and is used for the target operating system instead.
643 @defmac TARGET_OBJFMT_CPP_BUILTINS ()
644 Similarly to @code{TARGET_CPU_CPP_BUILTINS} but this macro is optional
645 and is used for the target object format. @file{elfos.h} uses this
646 macro to define @code{__ELF__}, so you probably do not need to define
650 @deftypevar {extern int} target_flags
651 This variable is declared in @file{options.h}, which is included before
652 any target-specific headers.
655 @deftypevr {Common Target Hook} int TARGET_DEFAULT_TARGET_FLAGS
656 This variable specifies the initial value of @code{target_flags}.
657 Its default setting is 0.
660 @cindex optional hardware or system features
661 @cindex features, optional, in system conventions
663 @deftypefn {Common Target Hook} bool TARGET_HANDLE_OPTION (struct gcc_options *@var{opts}, struct gcc_options *@var{opts_set}, const struct cl_decoded_option *@var{decoded}, location_t @var{loc})
664 This hook is called whenever the user specifies one of the
665 target-specific options described by the @file{.opt} definition files
666 (@pxref{Options}). It has the opportunity to do some option-specific
667 processing and should return true if the option is valid. The default
668 definition does nothing but return true.
670 @var{decoded} specifies the option and its arguments. @var{opts} and
671 @var{opts_set} are the @code{gcc_options} structures to be used for
672 storing option state, and @var{loc} is the location at which the
673 option was passed (@code{UNKNOWN_LOCATION} except for options passed
677 @deftypefn {C Target Hook} bool TARGET_HANDLE_C_OPTION (size_t @var{code}, const char *@var{arg}, int @var{value})
678 This target hook is called whenever the user specifies one of the
679 target-specific C language family options described by the @file{.opt}
680 definition files(@pxref{Options}). It has the opportunity to do some
681 option-specific processing and should return true if the option is
682 valid. The arguments are like for @code{TARGET_HANDLE_OPTION}. The
683 default definition does nothing but return false.
685 In general, you should use @code{TARGET_HANDLE_OPTION} to handle
686 options. However, if processing an option requires routines that are
687 only available in the C (and related language) front ends, then you
688 should use @code{TARGET_HANDLE_C_OPTION} instead.
691 @deftypefn {C Target Hook} tree TARGET_OBJC_CONSTRUCT_STRING_OBJECT (tree @var{string})
692 Targets may provide a string object type that can be used within and between C, C++ and their respective Objective-C dialects. A string object might, for example, embed encoding and length information. These objects are considered opaque to the compiler and handled as references. An ideal implementation makes the composition of the string object match that of the Objective-C @code{NSString} (@code{NXString} for GNUStep), allowing efficient interworking between C-only and Objective-C code. If a target implements string objects then this hook should return a reference to such an object constructed from the normal `C' string representation provided in @var{string}. At present, the hook is used by Objective-C only, to obtain a common-format string object when the target provides one.
695 @deftypefn {C Target Hook} void TARGET_OBJC_DECLARE_UNRESOLVED_CLASS_REFERENCE (const char *@var{classname})
696 Declare that Objective C class @var{classname} is referenced by the current TU.
699 @deftypefn {C Target Hook} void TARGET_OBJC_DECLARE_CLASS_DEFINITION (const char *@var{classname})
700 Declare that Objective C class @var{classname} is defined by the current TU.
703 @deftypefn {C Target Hook} bool TARGET_STRING_OBJECT_REF_TYPE_P (const_tree @var{stringref})
704 If a target implements string objects then this hook should return @code{true} if @var{stringref} is a valid reference to such an object.
707 @deftypefn {C Target Hook} void TARGET_CHECK_STRING_OBJECT_FORMAT_ARG (tree @var{format_arg}, tree @var{args_list})
708 If a target implements string objects then this hook should should provide a facility to check the function arguments in @var{args_list} against the format specifiers in @var{format_arg} where the type of @var{format_arg} is one recognized as a valid string reference type.
711 @deftypefn {Target Hook} void TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE (void)
712 This target function is similar to the hook @code{TARGET_OPTION_OVERRIDE}
713 but is called when the optimize level is changed via an attribute or
714 pragma or when it is reset at the end of the code affected by the
715 attribute or pragma. It is not called at the beginning of compilation
716 when @code{TARGET_OPTION_OVERRIDE} is called so if you want to perform these
717 actions then, you should have @code{TARGET_OPTION_OVERRIDE} call
718 @code{TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE}.
721 @defmac C_COMMON_OVERRIDE_OPTIONS
722 This is similar to the @code{TARGET_OPTION_OVERRIDE} hook
723 but is only used in the C
724 language frontends (C, Objective-C, C++, Objective-C++) and so can be
725 used to alter option flag variables which only exist in those
729 @deftypevr {Common Target Hook} {const struct default_options *} TARGET_OPTION_OPTIMIZATION_TABLE
730 Some machines may desire to change what optimizations are performed for
731 various optimization levels. This variable, if defined, describes
732 options to enable at particular sets of optimization levels. These
733 options are processed once
734 just after the optimization level is determined and before the remainder
735 of the command options have been parsed, so may be overridden by other
736 options passed explicitly.
738 This processing is run once at program startup and when the optimization
739 options are changed via @code{#pragma GCC optimize} or by using the
740 @code{optimize} attribute.
743 @deftypefn {Common Target Hook} void TARGET_OPTION_INIT_STRUCT (struct gcc_options *@var{opts})
744 Set target-dependent initial values of fields in @var{opts}.
747 @deftypefn {Common Target Hook} void TARGET_OPTION_DEFAULT_PARAMS (void)
748 Set target-dependent default values for @option{--param} settings, using calls to @code{set_default_param_value}.
751 @defmac SWITCHABLE_TARGET
752 Some targets need to switch between substantially different subtargets
753 during compilation. For example, the MIPS target has one subtarget for
754 the traditional MIPS architecture and another for MIPS16. Source code
755 can switch between these two subarchitectures using the @code{mips16}
756 and @code{nomips16} attributes.
758 Such subtargets can differ in things like the set of available
759 registers, the set of available instructions, the costs of various
760 operations, and so on. GCC caches a lot of this type of information
761 in global variables, and recomputing them for each subtarget takes a
762 significant amount of time. The compiler therefore provides a facility
763 for maintaining several versions of the global variables and quickly
764 switching between them; see @file{target-globals.h} for details.
766 Define this macro to 1 if your target needs this facility. The default
770 @deftypefn {Target Hook} bool TARGET_FLOAT_EXCEPTIONS_ROUNDING_SUPPORTED_P (void)
771 Returns true if the target supports IEEE 754 floating-point exceptions and rounding modes, false otherwise. This is intended to relate to the @code{float} and @code{double} types, but not necessarily @code{long double}. By default, returns true if the @code{adddf3} instruction pattern is available and false otherwise, on the assumption that hardware floating point supports exceptions and rounding modes but software floating point does not.
774 @node Per-Function Data
775 @section Defining data structures for per-function information.
776 @cindex per-function data
777 @cindex data structures
779 If the target needs to store information on a per-function basis, GCC
780 provides a macro and a couple of variables to allow this. Note, just
781 using statics to store the information is a bad idea, since GCC supports
782 nested functions, so you can be halfway through encoding one function
783 when another one comes along.
785 GCC defines a data structure called @code{struct function} which
786 contains all of the data specific to an individual function. This
787 structure contains a field called @code{machine} whose type is
788 @code{struct machine_function *}, which can be used by targets to point
789 to their own specific data.
791 If a target needs per-function specific data it should define the type
792 @code{struct machine_function} and also the macro @code{INIT_EXPANDERS}.
793 This macro should be used to initialize the function pointer
794 @code{init_machine_status}. This pointer is explained below.
796 One typical use of per-function, target specific data is to create an
797 RTX to hold the register containing the function's return address. This
798 RTX can then be used to implement the @code{__builtin_return_address}
799 function, for level 0.
801 Note---earlier implementations of GCC used a single data area to hold
802 all of the per-function information. Thus when processing of a nested
803 function began the old per-function data had to be pushed onto a
804 stack, and when the processing was finished, it had to be popped off the
805 stack. GCC used to provide function pointers called
806 @code{save_machine_status} and @code{restore_machine_status} to handle
807 the saving and restoring of the target specific information. Since the
808 single data area approach is no longer used, these pointers are no
811 @defmac INIT_EXPANDERS
812 Macro called to initialize any target specific information. This macro
813 is called once per function, before generation of any RTL has begun.
814 The intention of this macro is to allow the initialization of the
815 function pointer @code{init_machine_status}.
818 @deftypevar {void (*)(struct function *)} init_machine_status
819 If this function pointer is non-@code{NULL} it will be called once per
820 function, before function compilation starts, in order to allow the
821 target to perform any target specific initialization of the
822 @code{struct function} structure. It is intended that this would be
823 used to initialize the @code{machine} of that structure.
825 @code{struct machine_function} structures are expected to be freed by GC@.
826 Generally, any memory that they reference must be allocated by using
827 GC allocation, including the structure itself.
831 @section Storage Layout
832 @cindex storage layout
834 Note that the definitions of the macros in this table which are sizes or
835 alignments measured in bits do not need to be constant. They can be C
836 expressions that refer to static variables, such as the @code{target_flags}.
837 @xref{Run-time Target}.
839 @defmac BITS_BIG_ENDIAN
840 Define this macro to have the value 1 if the most significant bit in a
841 byte has the lowest number; otherwise define it to have the value zero.
842 This means that bit-field instructions count from the most significant
843 bit. If the machine has no bit-field instructions, then this must still
844 be defined, but it doesn't matter which value it is defined to. This
845 macro need not be a constant.
847 This macro does not affect the way structure fields are packed into
848 bytes or words; that is controlled by @code{BYTES_BIG_ENDIAN}.
851 @defmac BYTES_BIG_ENDIAN
852 Define this macro to have the value 1 if the most significant byte in a
853 word has the lowest number. This macro need not be a constant.
856 @defmac WORDS_BIG_ENDIAN
857 Define this macro to have the value 1 if, in a multiword object, the
858 most significant word has the lowest number. This applies to both
859 memory locations and registers; see @code{REG_WORDS_BIG_ENDIAN} if the
860 order of words in memory is not the same as the order in registers. This
861 macro need not be a constant.
864 @defmac REG_WORDS_BIG_ENDIAN
865 On some machines, the order of words in a multiword object differs between
866 registers in memory. In such a situation, define this macro to describe
867 the order of words in a register. The macro @code{WORDS_BIG_ENDIAN} controls
868 the order of words in memory.
871 @defmac FLOAT_WORDS_BIG_ENDIAN
872 Define this macro to have the value 1 if @code{DFmode}, @code{XFmode} or
873 @code{TFmode} floating point numbers are stored in memory with the word
874 containing the sign bit at the lowest address; otherwise define it to
875 have the value 0. This macro need not be a constant.
877 You need not define this macro if the ordering is the same as for
881 @defmac BITS_PER_WORD
882 Number of bits in a word. If you do not define this macro, the default
883 is @code{BITS_PER_UNIT * UNITS_PER_WORD}.
886 @defmac MAX_BITS_PER_WORD
887 Maximum number of bits in a word. If this is undefined, the default is
888 @code{BITS_PER_WORD}. Otherwise, it is the constant value that is the
889 largest value that @code{BITS_PER_WORD} can have at run-time.
892 @defmac UNITS_PER_WORD
893 Number of storage units in a word; normally the size of a general-purpose
894 register, a power of two from 1 or 8.
897 @defmac MIN_UNITS_PER_WORD
898 Minimum number of units in a word. If this is undefined, the default is
899 @code{UNITS_PER_WORD}. Otherwise, it is the constant value that is the
900 smallest value that @code{UNITS_PER_WORD} can have at run-time.
904 Width of a pointer, in bits. You must specify a value no wider than the
905 width of @code{Pmode}. If it is not equal to the width of @code{Pmode},
906 you must define @code{POINTERS_EXTEND_UNSIGNED}. If you do not specify
907 a value the default is @code{BITS_PER_WORD}.
910 @defmac POINTERS_EXTEND_UNSIGNED
911 A C expression that determines how pointers should be extended from
912 @code{ptr_mode} to either @code{Pmode} or @code{word_mode}. It is
913 greater than zero if pointers should be zero-extended, zero if they
914 should be sign-extended, and negative if some other sort of conversion
915 is needed. In the last case, the extension is done by the target's
916 @code{ptr_extend} instruction.
918 You need not define this macro if the @code{ptr_mode}, @code{Pmode}
919 and @code{word_mode} are all the same width.
922 @defmac PROMOTE_MODE (@var{m}, @var{unsignedp}, @var{type})
923 A macro to update @var{m} and @var{unsignedp} when an object whose type
924 is @var{type} and which has the specified mode and signedness is to be
925 stored in a register. This macro is only called when @var{type} is a
928 On most RISC machines, which only have operations that operate on a full
929 register, define this macro to set @var{m} to @code{word_mode} if
930 @var{m} is an integer mode narrower than @code{BITS_PER_WORD}. In most
931 cases, only integer modes should be widened because wider-precision
932 floating-point operations are usually more expensive than their narrower
935 For most machines, the macro definition does not change @var{unsignedp}.
936 However, some machines, have instructions that preferentially handle
937 either signed or unsigned quantities of certain modes. For example, on
938 the DEC Alpha, 32-bit loads from memory and 32-bit add instructions
939 sign-extend the result to 64 bits. On such machines, set
940 @var{unsignedp} according to which kind of extension is more efficient.
942 Do not define this macro if it would never modify @var{m}.
945 @deftypefn {Target Hook} machine_mode TARGET_PROMOTE_FUNCTION_MODE (const_tree @var{type}, machine_mode @var{mode}, int *@var{punsignedp}, const_tree @var{funtype}, int @var{for_return})
946 Like @code{PROMOTE_MODE}, but it is applied to outgoing function arguments or
947 function return values. The target hook should return the new mode
948 and possibly change @code{*@var{punsignedp}} if the promotion should
949 change signedness. This function is called only for scalar @emph{or
952 @var{for_return} allows to distinguish the promotion of arguments and
953 return values. If it is @code{1}, a return value is being promoted and
954 @code{TARGET_FUNCTION_VALUE} must perform the same promotions done here.
955 If it is @code{2}, the returned mode should be that of the register in
956 which an incoming parameter is copied, or the outgoing result is computed;
957 then the hook should return the same mode as @code{promote_mode}, though
958 the signedness may be different.
960 @var{type} can be NULL when promoting function arguments of libcalls.
962 The default is to not promote arguments and return values. You can
963 also define the hook to @code{default_promote_function_mode_always_promote}
964 if you would like to apply the same rules given by @code{PROMOTE_MODE}.
967 @defmac PARM_BOUNDARY
968 Normal alignment required for function parameters on the stack, in
969 bits. All stack parameters receive at least this much alignment
970 regardless of data type. On most machines, this is the same as the
974 @defmac STACK_BOUNDARY
975 Define this macro to the minimum alignment enforced by hardware for the
976 stack pointer on this machine. The definition is a C expression for the
977 desired alignment (measured in bits). This value is used as a default
978 if @code{PREFERRED_STACK_BOUNDARY} is not defined. On most machines,
979 this should be the same as @code{PARM_BOUNDARY}.
982 @defmac PREFERRED_STACK_BOUNDARY
983 Define this macro if you wish to preserve a certain alignment for the
984 stack pointer, greater than what the hardware enforces. The definition
985 is a C expression for the desired alignment (measured in bits). This
986 macro must evaluate to a value equal to or larger than
987 @code{STACK_BOUNDARY}.
990 @defmac INCOMING_STACK_BOUNDARY
991 Define this macro if the incoming stack boundary may be different
992 from @code{PREFERRED_STACK_BOUNDARY}. This macro must evaluate
993 to a value equal to or larger than @code{STACK_BOUNDARY}.
996 @defmac FUNCTION_BOUNDARY
997 Alignment required for a function entry point, in bits.
1000 @defmac BIGGEST_ALIGNMENT
1001 Biggest alignment that any data type can require on this machine, in
1002 bits. Note that this is not the biggest alignment that is supported,
1003 just the biggest alignment that, when violated, may cause a fault.
1006 @defmac MALLOC_ABI_ALIGNMENT
1007 Alignment, in bits, a C conformant malloc implementation has to
1008 provide. If not defined, the default value is @code{BITS_PER_WORD}.
1011 @defmac ATTRIBUTE_ALIGNED_VALUE
1012 Alignment used by the @code{__attribute__ ((aligned))} construct. If
1013 not defined, the default value is @code{BIGGEST_ALIGNMENT}.
1016 @defmac MINIMUM_ATOMIC_ALIGNMENT
1017 If defined, the smallest alignment, in bits, that can be given to an
1018 object that can be referenced in one operation, without disturbing any
1019 nearby object. Normally, this is @code{BITS_PER_UNIT}, but may be larger
1020 on machines that don't have byte or half-word store operations.
1023 @defmac BIGGEST_FIELD_ALIGNMENT
1024 Biggest alignment that any structure or union field can require on this
1025 machine, in bits. If defined, this overrides @code{BIGGEST_ALIGNMENT} for
1026 structure and union fields only, unless the field alignment has been set
1027 by the @code{__attribute__ ((aligned (@var{n})))} construct.
1030 @defmac ADJUST_FIELD_ALIGN (@var{field}, @var{computed})
1031 An expression for the alignment of a structure field @var{field} if the
1032 alignment computed in the usual way (including applying of
1033 @code{BIGGEST_ALIGNMENT} and @code{BIGGEST_FIELD_ALIGNMENT} to the
1034 alignment) is @var{computed}. It overrides alignment only if the
1035 field alignment has not been set by the
1036 @code{__attribute__ ((aligned (@var{n})))} construct.
1039 @defmac MAX_STACK_ALIGNMENT
1040 Biggest stack alignment guaranteed by the backend. Use this macro
1041 to specify the maximum alignment of a variable on stack.
1043 If not defined, the default value is @code{STACK_BOUNDARY}.
1045 @c FIXME: The default should be @code{PREFERRED_STACK_BOUNDARY}.
1046 @c But the fix for PR 32893 indicates that we can only guarantee
1047 @c maximum stack alignment on stack up to @code{STACK_BOUNDARY}, not
1048 @c @code{PREFERRED_STACK_BOUNDARY}, if stack alignment isn't supported.
1051 @defmac MAX_OFILE_ALIGNMENT
1052 Biggest alignment supported by the object file format of this machine.
1053 Use this macro to limit the alignment which can be specified using the
1054 @code{__attribute__ ((aligned (@var{n})))} construct. If not defined,
1055 the default value is @code{BIGGEST_ALIGNMENT}.
1057 On systems that use ELF, the default (in @file{config/elfos.h}) is
1058 the largest supported 32-bit ELF section alignment representable on
1059 a 32-bit host e.g. @samp{(((uint64_t) 1 << 28) * 8)}.
1060 On 32-bit ELF the largest supported section alignment in bits is
1061 @samp{(0x80000000 * 8)}, but this is not representable on 32-bit hosts.
1064 @defmac DATA_ALIGNMENT (@var{type}, @var{basic-align})
1065 If defined, a C expression to compute the alignment for a variable in
1066 the static store. @var{type} is the data type, and @var{basic-align} is
1067 the alignment that the object would ordinarily have. The value of this
1068 macro is used instead of that alignment to align the object.
1070 If this macro is not defined, then @var{basic-align} is used.
1073 One use of this macro is to increase alignment of medium-size data to
1074 make it all fit in fewer cache lines. Another is to cause character
1075 arrays to be word-aligned so that @code{strcpy} calls that copy
1076 constants to character arrays can be done inline.
1079 @defmac DATA_ABI_ALIGNMENT (@var{type}, @var{basic-align})
1080 Similar to @code{DATA_ALIGNMENT}, but for the cases where the ABI mandates
1081 some alignment increase, instead of optimization only purposes. E.g.@
1082 AMD x86-64 psABI says that variables with array type larger than 15 bytes
1083 must be aligned to 16 byte boundaries.
1085 If this macro is not defined, then @var{basic-align} is used.
1088 @defmac CONSTANT_ALIGNMENT (@var{constant}, @var{basic-align})
1089 If defined, a C expression to compute the alignment given to a constant
1090 that is being placed in memory. @var{constant} is the constant and
1091 @var{basic-align} is the alignment that the object would ordinarily
1092 have. The value of this macro is used instead of that alignment to
1095 If this macro is not defined, then @var{basic-align} is used.
1097 The typical use of this macro is to increase alignment for string
1098 constants to be word aligned so that @code{strcpy} calls that copy
1099 constants can be done inline.
1102 @defmac LOCAL_ALIGNMENT (@var{type}, @var{basic-align})
1103 If defined, a C expression to compute the alignment for a variable in
1104 the local store. @var{type} is the data type, and @var{basic-align} is
1105 the alignment that the object would ordinarily have. The value of this
1106 macro is used instead of that alignment to align the object.
1108 If this macro is not defined, then @var{basic-align} is used.
1110 One use of this macro is to increase alignment of medium-size data to
1111 make it all fit in fewer cache lines.
1113 If the value of this macro has a type, it should be an unsigned type.
1116 @deftypefn {Target Hook} HOST_WIDE_INT TARGET_VECTOR_ALIGNMENT (const_tree @var{type})
1117 This hook can be used to define the alignment for a vector of type
1118 @var{type}, in order to comply with a platform ABI. The default is to
1119 require natural alignment for vector types. The alignment returned by
1120 this hook must be a power-of-two multiple of the default alignment of
1121 the vector element type.
1124 @defmac STACK_SLOT_ALIGNMENT (@var{type}, @var{mode}, @var{basic-align})
1125 If defined, a C expression to compute the alignment for stack slot.
1126 @var{type} is the data type, @var{mode} is the widest mode available,
1127 and @var{basic-align} is the alignment that the slot would ordinarily
1128 have. The value of this macro is used instead of that alignment to
1131 If this macro is not defined, then @var{basic-align} is used when
1132 @var{type} is @code{NULL}. Otherwise, @code{LOCAL_ALIGNMENT} will
1135 This macro is to set alignment of stack slot to the maximum alignment
1136 of all possible modes which the slot may have.
1138 If the value of this macro has a type, it should be an unsigned type.
1141 @defmac LOCAL_DECL_ALIGNMENT (@var{decl})
1142 If defined, a C expression to compute the alignment for a local
1143 variable @var{decl}.
1145 If this macro is not defined, then
1146 @code{LOCAL_ALIGNMENT (TREE_TYPE (@var{decl}), DECL_ALIGN (@var{decl}))}
1149 One use of this macro is to increase alignment of medium-size data to
1150 make it all fit in fewer cache lines.
1152 If the value of this macro has a type, it should be an unsigned type.
1155 @defmac MINIMUM_ALIGNMENT (@var{exp}, @var{mode}, @var{align})
1156 If defined, a C expression to compute the minimum required alignment
1157 for dynamic stack realignment purposes for @var{exp} (a type or decl),
1158 @var{mode}, assuming normal alignment @var{align}.
1160 If this macro is not defined, then @var{align} will be used.
1163 @defmac EMPTY_FIELD_BOUNDARY
1164 Alignment in bits to be given to a structure bit-field that follows an
1165 empty field such as @code{int : 0;}.
1167 If @code{PCC_BITFIELD_TYPE_MATTERS} is true, it overrides this macro.
1170 @defmac STRUCTURE_SIZE_BOUNDARY
1171 Number of bits which any structure or union's size must be a multiple of.
1172 Each structure or union's size is rounded up to a multiple of this.
1174 If you do not define this macro, the default is the same as
1175 @code{BITS_PER_UNIT}.
1178 @defmac STRICT_ALIGNMENT
1179 Define this macro to be the value 1 if instructions will fail to work
1180 if given data not on the nominal alignment. If instructions will merely
1181 go slower in that case, define this macro as 0.
1184 @defmac PCC_BITFIELD_TYPE_MATTERS
1185 Define this if you wish to imitate the way many other C compilers handle
1186 alignment of bit-fields and the structures that contain them.
1188 The behavior is that the type written for a named bit-field (@code{int},
1189 @code{short}, or other integer type) imposes an alignment for the entire
1190 structure, as if the structure really did contain an ordinary field of
1191 that type. In addition, the bit-field is placed within the structure so
1192 that it would fit within such a field, not crossing a boundary for it.
1194 Thus, on most machines, a named bit-field whose type is written as
1195 @code{int} would not cross a four-byte boundary, and would force
1196 four-byte alignment for the whole structure. (The alignment used may
1197 not be four bytes; it is controlled by the other alignment parameters.)
1199 An unnamed bit-field will not affect the alignment of the containing
1202 If the macro is defined, its definition should be a C expression;
1203 a nonzero value for the expression enables this behavior.
1205 Note that if this macro is not defined, or its value is zero, some
1206 bit-fields may cross more than one alignment boundary. The compiler can
1207 support such references if there are @samp{insv}, @samp{extv}, and
1208 @samp{extzv} insns that can directly reference memory.
1210 The other known way of making bit-fields work is to define
1211 @code{STRUCTURE_SIZE_BOUNDARY} as large as @code{BIGGEST_ALIGNMENT}.
1212 Then every structure can be accessed with fullwords.
1214 Unless the machine has bit-field instructions or you define
1215 @code{STRUCTURE_SIZE_BOUNDARY} that way, you must define
1216 @code{PCC_BITFIELD_TYPE_MATTERS} to have a nonzero value.
1218 If your aim is to make GCC use the same conventions for laying out
1219 bit-fields as are used by another compiler, here is how to investigate
1220 what the other compiler does. Compile and run this program:
1239 printf ("Size of foo1 is %d\n",
1240 sizeof (struct foo1));
1241 printf ("Size of foo2 is %d\n",
1242 sizeof (struct foo2));
1247 If this prints 2 and 5, then the compiler's behavior is what you would
1248 get from @code{PCC_BITFIELD_TYPE_MATTERS}.
1251 @defmac BITFIELD_NBYTES_LIMITED
1252 Like @code{PCC_BITFIELD_TYPE_MATTERS} except that its effect is limited
1253 to aligning a bit-field within the structure.
1256 @deftypefn {Target Hook} bool TARGET_ALIGN_ANON_BITFIELD (void)
1257 When @code{PCC_BITFIELD_TYPE_MATTERS} is true this hook will determine
1258 whether unnamed bitfields affect the alignment of the containing
1259 structure. The hook should return true if the structure should inherit
1260 the alignment requirements of an unnamed bitfield's type.
1263 @deftypefn {Target Hook} bool TARGET_NARROW_VOLATILE_BITFIELD (void)
1264 This target hook should return @code{true} if accesses to volatile bitfields
1265 should use the narrowest mode possible. It should return @code{false} if
1266 these accesses should use the bitfield container type.
1268 The default is @code{false}.
1271 @deftypefn {Target Hook} bool TARGET_MEMBER_TYPE_FORCES_BLK (const_tree @var{field}, machine_mode @var{mode})
1272 Return true if a structure, union or array containing @var{field} should
1273 be accessed using @code{BLKMODE}.
1275 If @var{field} is the only field in the structure, @var{mode} is its
1276 mode, otherwise @var{mode} is VOIDmode. @var{mode} is provided in the
1277 case where structures of one field would require the structure's mode to
1278 retain the field's mode.
1280 Normally, this is not needed.
1283 @defmac ROUND_TYPE_ALIGN (@var{type}, @var{computed}, @var{specified})
1284 Define this macro as an expression for the alignment of a type (given
1285 by @var{type} as a tree node) if the alignment computed in the usual
1286 way is @var{computed} and the alignment explicitly specified was
1289 The default is to use @var{specified} if it is larger; otherwise, use
1290 the smaller of @var{computed} and @code{BIGGEST_ALIGNMENT}
1293 @defmac MAX_FIXED_MODE_SIZE
1294 An integer expression for the size in bits of the largest integer
1295 machine mode that should actually be used. All integer machine modes of
1296 this size or smaller can be used for structures and unions with the
1297 appropriate sizes. If this macro is undefined, @code{GET_MODE_BITSIZE
1298 (DImode)} is assumed.
1301 @defmac STACK_SAVEAREA_MODE (@var{save_level})
1302 If defined, an expression of type @code{machine_mode} that
1303 specifies the mode of the save area operand of a
1304 @code{save_stack_@var{level}} named pattern (@pxref{Standard Names}).
1305 @var{save_level} is one of @code{SAVE_BLOCK}, @code{SAVE_FUNCTION}, or
1306 @code{SAVE_NONLOCAL} and selects which of the three named patterns is
1307 having its mode specified.
1309 You need not define this macro if it always returns @code{Pmode}. You
1310 would most commonly define this macro if the
1311 @code{save_stack_@var{level}} patterns need to support both a 32- and a
1315 @defmac STACK_SIZE_MODE
1316 If defined, an expression of type @code{machine_mode} that
1317 specifies the mode of the size increment operand of an
1318 @code{allocate_stack} named pattern (@pxref{Standard Names}).
1320 You need not define this macro if it always returns @code{word_mode}.
1321 You would most commonly define this macro if the @code{allocate_stack}
1322 pattern needs to support both a 32- and a 64-bit mode.
1325 @deftypefn {Target Hook} machine_mode TARGET_LIBGCC_CMP_RETURN_MODE (void)
1326 This target hook should return the mode to be used for the return value
1327 of compare instructions expanded to libgcc calls. If not defined
1328 @code{word_mode} is returned which is the right choice for a majority of
1332 @deftypefn {Target Hook} machine_mode TARGET_LIBGCC_SHIFT_COUNT_MODE (void)
1333 This target hook should return the mode to be used for the shift count operand
1334 of shift instructions expanded to libgcc calls. If not defined
1335 @code{word_mode} is returned which is the right choice for a majority of
1339 @deftypefn {Target Hook} machine_mode TARGET_UNWIND_WORD_MODE (void)
1340 Return machine mode to be used for @code{_Unwind_Word} type.
1341 The default is to use @code{word_mode}.
1344 @deftypefn {Target Hook} bool TARGET_MS_BITFIELD_LAYOUT_P (const_tree @var{record_type})
1345 This target hook returns @code{true} if bit-fields in the given
1346 @var{record_type} are to be laid out following the rules of Microsoft
1347 Visual C/C++, namely: (i) a bit-field won't share the same storage
1348 unit with the previous bit-field if their underlying types have
1349 different sizes, and the bit-field will be aligned to the highest
1350 alignment of the underlying types of itself and of the previous
1351 bit-field; (ii) a zero-sized bit-field will affect the alignment of
1352 the whole enclosing structure, even if it is unnamed; except that
1353 (iii) a zero-sized bit-field will be disregarded unless it follows
1354 another bit-field of nonzero size. If this hook returns @code{true},
1355 other macros that control bit-field layout are ignored.
1357 When a bit-field is inserted into a packed record, the whole size
1358 of the underlying type is used by one or more same-size adjacent
1359 bit-fields (that is, if its long:3, 32 bits is used in the record,
1360 and any additional adjacent long bit-fields are packed into the same
1361 chunk of 32 bits. However, if the size changes, a new field of that
1362 size is allocated). In an unpacked record, this is the same as using
1363 alignment, but not equivalent when packing.
1365 If both MS bit-fields and @samp{__attribute__((packed))} are used,
1366 the latter will take precedence. If @samp{__attribute__((packed))} is
1367 used on a single field when MS bit-fields are in use, it will take
1368 precedence for that field, but the alignment of the rest of the structure
1369 may affect its placement.
1372 @deftypefn {Target Hook} bool TARGET_DECIMAL_FLOAT_SUPPORTED_P (void)
1373 Returns true if the target supports decimal floating point.
1376 @deftypefn {Target Hook} bool TARGET_FIXED_POINT_SUPPORTED_P (void)
1377 Returns true if the target supports fixed-point arithmetic.
1380 @deftypefn {Target Hook} void TARGET_EXPAND_TO_RTL_HOOK (void)
1381 This hook is called just before expansion into rtl, allowing the target
1382 to perform additional initializations or analysis before the expansion.
1383 For example, the rs6000 port uses it to allocate a scratch stack slot
1384 for use in copying SDmode values between memory and floating point
1385 registers whenever the function being expanded has any SDmode
1389 @deftypefn {Target Hook} void TARGET_INSTANTIATE_DECLS (void)
1390 This hook allows the backend to perform additional instantiations on rtl
1391 that are not actually in any insns yet, but will be later.
1394 @deftypefn {Target Hook} {const char *} TARGET_MANGLE_TYPE (const_tree @var{type})
1395 If your target defines any fundamental types, or any types your target
1396 uses should be mangled differently from the default, define this hook
1397 to return the appropriate encoding for these types as part of a C++
1398 mangled name. The @var{type} argument is the tree structure representing
1399 the type to be mangled. The hook may be applied to trees which are
1400 not target-specific fundamental types; it should return @code{NULL}
1401 for all such types, as well as arguments it does not recognize. If the
1402 return value is not @code{NULL}, it must point to a statically-allocated
1405 Target-specific fundamental types might be new fundamental types or
1406 qualified versions of ordinary fundamental types. Encode new
1407 fundamental types as @samp{@w{u @var{n} @var{name}}}, where @var{name}
1408 is the name used for the type in source code, and @var{n} is the
1409 length of @var{name} in decimal. Encode qualified versions of
1410 ordinary types as @samp{@w{U @var{n} @var{name} @var{code}}}, where
1411 @var{name} is the name used for the type qualifier in source code,
1412 @var{n} is the length of @var{name} as above, and @var{code} is the
1413 code used to represent the unqualified version of this type. (See
1414 @code{write_builtin_type} in @file{cp/mangle.c} for the list of
1415 codes.) In both cases the spaces are for clarity; do not include any
1416 spaces in your string.
1418 This hook is applied to types prior to typedef resolution. If the mangled
1419 name for a particular type depends only on that type's main variant, you
1420 can perform typedef resolution yourself using @code{TYPE_MAIN_VARIANT}
1423 The default version of this hook always returns @code{NULL}, which is
1424 appropriate for a target that does not define any new fundamental
1429 @section Layout of Source Language Data Types
1431 These macros define the sizes and other characteristics of the standard
1432 basic data types used in programs being compiled. Unlike the macros in
1433 the previous section, these apply to specific features of C and related
1434 languages, rather than to fundamental aspects of storage layout.
1436 @defmac INT_TYPE_SIZE
1437 A C expression for the size in bits of the type @code{int} on the
1438 target machine. If you don't define this, the default is one word.
1441 @defmac SHORT_TYPE_SIZE
1442 A C expression for the size in bits of the type @code{short} on the
1443 target machine. If you don't define this, the default is half a word.
1444 (If this would be less than one storage unit, it is rounded up to one
1448 @defmac LONG_TYPE_SIZE
1449 A C expression for the size in bits of the type @code{long} on the
1450 target machine. If you don't define this, the default is one word.
1453 @defmac ADA_LONG_TYPE_SIZE
1454 On some machines, the size used for the Ada equivalent of the type
1455 @code{long} by a native Ada compiler differs from that used by C@. In
1456 that situation, define this macro to be a C expression to be used for
1457 the size of that type. If you don't define this, the default is the
1458 value of @code{LONG_TYPE_SIZE}.
1461 @defmac LONG_LONG_TYPE_SIZE
1462 A C expression for the size in bits of the type @code{long long} on the
1463 target machine. If you don't define this, the default is two
1464 words. If you want to support GNU Ada on your machine, the value of this
1465 macro must be at least 64.
1468 @defmac CHAR_TYPE_SIZE
1469 A C expression for the size in bits of the type @code{char} on the
1470 target machine. If you don't define this, the default is
1471 @code{BITS_PER_UNIT}.
1474 @defmac BOOL_TYPE_SIZE
1475 A C expression for the size in bits of the C++ type @code{bool} and
1476 C99 type @code{_Bool} on the target machine. If you don't define
1477 this, and you probably shouldn't, the default is @code{CHAR_TYPE_SIZE}.
1480 @defmac FLOAT_TYPE_SIZE
1481 A C expression for the size in bits of the type @code{float} on the
1482 target machine. If you don't define this, the default is one word.
1485 @defmac DOUBLE_TYPE_SIZE
1486 A C expression for the size in bits of the type @code{double} on the
1487 target machine. If you don't define this, the default is two
1491 @defmac LONG_DOUBLE_TYPE_SIZE
1492 A C expression for the size in bits of the type @code{long double} on
1493 the target machine. If you don't define this, the default is two
1497 @defmac SHORT_FRACT_TYPE_SIZE
1498 A C expression for the size in bits of the type @code{short _Fract} on
1499 the target machine. If you don't define this, the default is
1500 @code{BITS_PER_UNIT}.
1503 @defmac FRACT_TYPE_SIZE
1504 A C expression for the size in bits of the type @code{_Fract} on
1505 the target machine. If you don't define this, the default is
1506 @code{BITS_PER_UNIT * 2}.
1509 @defmac LONG_FRACT_TYPE_SIZE
1510 A C expression for the size in bits of the type @code{long _Fract} on
1511 the target machine. If you don't define this, the default is
1512 @code{BITS_PER_UNIT * 4}.
1515 @defmac LONG_LONG_FRACT_TYPE_SIZE
1516 A C expression for the size in bits of the type @code{long long _Fract} on
1517 the target machine. If you don't define this, the default is
1518 @code{BITS_PER_UNIT * 8}.
1521 @defmac SHORT_ACCUM_TYPE_SIZE
1522 A C expression for the size in bits of the type @code{short _Accum} on
1523 the target machine. If you don't define this, the default is
1524 @code{BITS_PER_UNIT * 2}.
1527 @defmac ACCUM_TYPE_SIZE
1528 A C expression for the size in bits of the type @code{_Accum} on
1529 the target machine. If you don't define this, the default is
1530 @code{BITS_PER_UNIT * 4}.
1533 @defmac LONG_ACCUM_TYPE_SIZE
1534 A C expression for the size in bits of the type @code{long _Accum} on
1535 the target machine. If you don't define this, the default is
1536 @code{BITS_PER_UNIT * 8}.
1539 @defmac LONG_LONG_ACCUM_TYPE_SIZE
1540 A C expression for the size in bits of the type @code{long long _Accum} on
1541 the target machine. If you don't define this, the default is
1542 @code{BITS_PER_UNIT * 16}.
1545 @defmac LIBGCC2_GNU_PREFIX
1546 This macro corresponds to the @code{TARGET_LIBFUNC_GNU_PREFIX} target
1547 hook and should be defined if that hook is overriden to be true. It
1548 causes function names in libgcc to be changed to use a @code{__gnu_}
1549 prefix for their name rather than the default @code{__}. A port which
1550 uses this macro should also arrange to use @file{t-gnu-prefix} in
1551 the libgcc @file{config.host}.
1554 @defmac TARGET_FLT_EVAL_METHOD
1555 A C expression for the value for @code{FLT_EVAL_METHOD} in @file{float.h},
1556 assuming, if applicable, that the floating-point control word is in its
1557 default state. If you do not define this macro the value of
1558 @code{FLT_EVAL_METHOD} will be zero.
1561 @defmac WIDEST_HARDWARE_FP_SIZE
1562 A C expression for the size in bits of the widest floating-point format
1563 supported by the hardware. If you define this macro, you must specify a
1564 value less than or equal to the value of @code{LONG_DOUBLE_TYPE_SIZE}.
1565 If you do not define this macro, the value of @code{LONG_DOUBLE_TYPE_SIZE}
1569 @defmac DEFAULT_SIGNED_CHAR
1570 An expression whose value is 1 or 0, according to whether the type
1571 @code{char} should be signed or unsigned by default. The user can
1572 always override this default with the options @option{-fsigned-char}
1573 and @option{-funsigned-char}.
1576 @deftypefn {Target Hook} bool TARGET_DEFAULT_SHORT_ENUMS (void)
1577 This target hook should return true if the compiler should give an
1578 @code{enum} type only as many bytes as it takes to represent the range
1579 of possible values of that type. It should return false if all
1580 @code{enum} types should be allocated like @code{int}.
1582 The default is to return false.
1586 A C expression for a string describing the name of the data type to use
1587 for size values. The typedef name @code{size_t} is defined using the
1588 contents of the string.
1590 The string can contain more than one keyword. If so, separate them with
1591 spaces, and write first any length keyword, then @code{unsigned} if
1592 appropriate, and finally @code{int}. The string must exactly match one
1593 of the data type names defined in the function
1594 @code{c_common_nodes_and_builtins} in the file @file{c-family/c-common.c}.
1595 You may not omit @code{int} or change the order---that would cause the
1596 compiler to crash on startup.
1598 If you don't define this macro, the default is @code{"long unsigned
1603 GCC defines internal types (@code{sizetype}, @code{ssizetype},
1604 @code{bitsizetype} and @code{sbitsizetype}) for expressions
1605 dealing with size. This macro is a C expression for a string describing
1606 the name of the data type from which the precision of @code{sizetype}
1609 The string has the same restrictions as @code{SIZE_TYPE} string.
1611 If you don't define this macro, the default is @code{SIZE_TYPE}.
1614 @defmac PTRDIFF_TYPE
1615 A C expression for a string describing the name of the data type to use
1616 for the result of subtracting two pointers. The typedef name
1617 @code{ptrdiff_t} is defined using the contents of the string. See
1618 @code{SIZE_TYPE} above for more information.
1620 If you don't define this macro, the default is @code{"long int"}.
1624 A C expression for a string describing the name of the data type to use
1625 for wide characters. The typedef name @code{wchar_t} is defined using
1626 the contents of the string. See @code{SIZE_TYPE} above for more
1629 If you don't define this macro, the default is @code{"int"}.
1632 @defmac WCHAR_TYPE_SIZE
1633 A C expression for the size in bits of the data type for wide
1634 characters. This is used in @code{cpp}, which cannot make use of
1639 A C expression for a string describing the name of the data type to
1640 use for wide characters passed to @code{printf} and returned from
1641 @code{getwc}. The typedef name @code{wint_t} is defined using the
1642 contents of the string. See @code{SIZE_TYPE} above for more
1645 If you don't define this macro, the default is @code{"unsigned int"}.
1649 A C expression for a string describing the name of the data type that
1650 can represent any value of any standard or extended signed integer type.
1651 The typedef name @code{intmax_t} is defined using the contents of the
1652 string. See @code{SIZE_TYPE} above for more information.
1654 If you don't define this macro, the default is the first of
1655 @code{"int"}, @code{"long int"}, or @code{"long long int"} that has as
1656 much precision as @code{long long int}.
1659 @defmac UINTMAX_TYPE
1660 A C expression for a string describing the name of the data type that
1661 can represent any value of any standard or extended unsigned integer
1662 type. The typedef name @code{uintmax_t} is defined using the contents
1663 of the string. See @code{SIZE_TYPE} above for more information.
1665 If you don't define this macro, the default is the first of
1666 @code{"unsigned int"}, @code{"long unsigned int"}, or @code{"long long
1667 unsigned int"} that has as much precision as @code{long long unsigned
1671 @defmac SIG_ATOMIC_TYPE
1677 @defmacx UINT16_TYPE
1678 @defmacx UINT32_TYPE
1679 @defmacx UINT64_TYPE
1680 @defmacx INT_LEAST8_TYPE
1681 @defmacx INT_LEAST16_TYPE
1682 @defmacx INT_LEAST32_TYPE
1683 @defmacx INT_LEAST64_TYPE
1684 @defmacx UINT_LEAST8_TYPE
1685 @defmacx UINT_LEAST16_TYPE
1686 @defmacx UINT_LEAST32_TYPE
1687 @defmacx UINT_LEAST64_TYPE
1688 @defmacx INT_FAST8_TYPE
1689 @defmacx INT_FAST16_TYPE
1690 @defmacx INT_FAST32_TYPE
1691 @defmacx INT_FAST64_TYPE
1692 @defmacx UINT_FAST8_TYPE
1693 @defmacx UINT_FAST16_TYPE
1694 @defmacx UINT_FAST32_TYPE
1695 @defmacx UINT_FAST64_TYPE
1696 @defmacx INTPTR_TYPE
1697 @defmacx UINTPTR_TYPE
1698 C expressions for the standard types @code{sig_atomic_t},
1699 @code{int8_t}, @code{int16_t}, @code{int32_t}, @code{int64_t},
1700 @code{uint8_t}, @code{uint16_t}, @code{uint32_t}, @code{uint64_t},
1701 @code{int_least8_t}, @code{int_least16_t}, @code{int_least32_t},
1702 @code{int_least64_t}, @code{uint_least8_t}, @code{uint_least16_t},
1703 @code{uint_least32_t}, @code{uint_least64_t}, @code{int_fast8_t},
1704 @code{int_fast16_t}, @code{int_fast32_t}, @code{int_fast64_t},
1705 @code{uint_fast8_t}, @code{uint_fast16_t}, @code{uint_fast32_t},
1706 @code{uint_fast64_t}, @code{intptr_t}, and @code{uintptr_t}. See
1707 @code{SIZE_TYPE} above for more information.
1709 If any of these macros evaluates to a null pointer, the corresponding
1710 type is not supported; if GCC is configured to provide
1711 @code{<stdint.h>} in such a case, the header provided may not conform
1712 to C99, depending on the type in question. The defaults for all of
1713 these macros are null pointers.
1716 @defmac TARGET_PTRMEMFUNC_VBIT_LOCATION
1717 The C++ compiler represents a pointer-to-member-function with a struct
1724 ptrdiff_t vtable_index;
1731 The C++ compiler must use one bit to indicate whether the function that
1732 will be called through a pointer-to-member-function is virtual.
1733 Normally, we assume that the low-order bit of a function pointer must
1734 always be zero. Then, by ensuring that the vtable_index is odd, we can
1735 distinguish which variant of the union is in use. But, on some
1736 platforms function pointers can be odd, and so this doesn't work. In
1737 that case, we use the low-order bit of the @code{delta} field, and shift
1738 the remainder of the @code{delta} field to the left.
1740 GCC will automatically make the right selection about where to store
1741 this bit using the @code{FUNCTION_BOUNDARY} setting for your platform.
1742 However, some platforms such as ARM/Thumb have @code{FUNCTION_BOUNDARY}
1743 set such that functions always start at even addresses, but the lowest
1744 bit of pointers to functions indicate whether the function at that
1745 address is in ARM or Thumb mode. If this is the case of your
1746 architecture, you should define this macro to
1747 @code{ptrmemfunc_vbit_in_delta}.
1749 In general, you should not have to define this macro. On architectures
1750 in which function addresses are always even, according to
1751 @code{FUNCTION_BOUNDARY}, GCC will automatically define this macro to
1752 @code{ptrmemfunc_vbit_in_pfn}.
1755 @defmac TARGET_VTABLE_USES_DESCRIPTORS
1756 Normally, the C++ compiler uses function pointers in vtables. This
1757 macro allows the target to change to use ``function descriptors''
1758 instead. Function descriptors are found on targets for whom a
1759 function pointer is actually a small data structure. Normally the
1760 data structure consists of the actual code address plus a data
1761 pointer to which the function's data is relative.
1763 If vtables are used, the value of this macro should be the number
1764 of words that the function descriptor occupies.
1767 @defmac TARGET_VTABLE_ENTRY_ALIGN
1768 By default, the vtable entries are void pointers, the so the alignment
1769 is the same as pointer alignment. The value of this macro specifies
1770 the alignment of the vtable entry in bits. It should be defined only
1771 when special alignment is necessary. */
1774 @defmac TARGET_VTABLE_DATA_ENTRY_DISTANCE
1775 There are a few non-descriptor entries in the vtable at offsets below
1776 zero. If these entries must be padded (say, to preserve the alignment
1777 specified by @code{TARGET_VTABLE_ENTRY_ALIGN}), set this to the number
1778 of words in each data entry.
1782 @section Register Usage
1783 @cindex register usage
1785 This section explains how to describe what registers the target machine
1786 has, and how (in general) they can be used.
1788 The description of which registers a specific instruction can use is
1789 done with register classes; see @ref{Register Classes}. For information
1790 on using registers to access a stack frame, see @ref{Frame Registers}.
1791 For passing values in registers, see @ref{Register Arguments}.
1792 For returning values in registers, see @ref{Scalar Return}.
1795 * Register Basics:: Number and kinds of registers.
1796 * Allocation Order:: Order in which registers are allocated.
1797 * Values in Registers:: What kinds of values each reg can hold.
1798 * Leaf Functions:: Renumbering registers for leaf functions.
1799 * Stack Registers:: Handling a register stack such as 80387.
1802 @node Register Basics
1803 @subsection Basic Characteristics of Registers
1805 @c prevent bad page break with this line
1806 Registers have various characteristics.
1808 @defmac FIRST_PSEUDO_REGISTER
1809 Number of hardware registers known to the compiler. They receive
1810 numbers 0 through @code{FIRST_PSEUDO_REGISTER-1}; thus, the first
1811 pseudo register's number really is assigned the number
1812 @code{FIRST_PSEUDO_REGISTER}.
1815 @defmac FIXED_REGISTERS
1816 @cindex fixed register
1817 An initializer that says which registers are used for fixed purposes
1818 all throughout the compiled code and are therefore not available for
1819 general allocation. These would include the stack pointer, the frame
1820 pointer (except on machines where that can be used as a general
1821 register when no frame pointer is needed), the program counter on
1822 machines where that is considered one of the addressable registers,
1823 and any other numbered register with a standard use.
1825 This information is expressed as a sequence of numbers, separated by
1826 commas and surrounded by braces. The @var{n}th number is 1 if
1827 register @var{n} is fixed, 0 otherwise.
1829 The table initialized from this macro, and the table initialized by
1830 the following one, may be overridden at run time either automatically,
1831 by the actions of the macro @code{CONDITIONAL_REGISTER_USAGE}, or by
1832 the user with the command options @option{-ffixed-@var{reg}},
1833 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}.
1836 @defmac CALL_USED_REGISTERS
1837 @cindex call-used register
1838 @cindex call-clobbered register
1839 @cindex call-saved register
1840 Like @code{FIXED_REGISTERS} but has 1 for each register that is
1841 clobbered (in general) by function calls as well as for fixed
1842 registers. This macro therefore identifies the registers that are not
1843 available for general allocation of values that must live across
1846 If a register has 0 in @code{CALL_USED_REGISTERS}, the compiler
1847 automatically saves it on function entry and restores it on function
1848 exit, if the register is used within the function.
1851 @defmac CALL_REALLY_USED_REGISTERS
1852 @cindex call-used register
1853 @cindex call-clobbered register
1854 @cindex call-saved register
1855 Like @code{CALL_USED_REGISTERS} except this macro doesn't require
1856 that the entire set of @code{FIXED_REGISTERS} be included.
1857 (@code{CALL_USED_REGISTERS} must be a superset of @code{FIXED_REGISTERS}).
1858 This macro is optional. If not specified, it defaults to the value
1859 of @code{CALL_USED_REGISTERS}.
1862 @defmac HARD_REGNO_CALL_PART_CLOBBERED (@var{regno}, @var{mode})
1863 @cindex call-used register
1864 @cindex call-clobbered register
1865 @cindex call-saved register
1866 A C expression that is nonzero if it is not permissible to store a
1867 value of mode @var{mode} in hard register number @var{regno} across a
1868 call without some part of it being clobbered. For most machines this
1869 macro need not be defined. It is only required for machines that do not
1870 preserve the entire contents of a register across a call.
1874 @findex call_used_regs
1877 @findex reg_class_contents
1878 @deftypefn {Target Hook} void TARGET_CONDITIONAL_REGISTER_USAGE (void)
1879 This hook may conditionally modify five variables
1880 @code{fixed_regs}, @code{call_used_regs}, @code{global_regs},
1881 @code{reg_names}, and @code{reg_class_contents}, to take into account
1882 any dependence of these register sets on target flags. The first three
1883 of these are of type @code{char []} (interpreted as Boolean vectors).
1884 @code{global_regs} is a @code{const char *[]}, and
1885 @code{reg_class_contents} is a @code{HARD_REG_SET}. Before the macro is
1886 called, @code{fixed_regs}, @code{call_used_regs},
1887 @code{reg_class_contents}, and @code{reg_names} have been initialized
1888 from @code{FIXED_REGISTERS}, @code{CALL_USED_REGISTERS},
1889 @code{REG_CLASS_CONTENTS}, and @code{REGISTER_NAMES}, respectively.
1890 @code{global_regs} has been cleared, and any @option{-ffixed-@var{reg}},
1891 @option{-fcall-used-@var{reg}} and @option{-fcall-saved-@var{reg}}
1892 command options have been applied.
1894 @cindex disabling certain registers
1895 @cindex controlling register usage
1896 If the usage of an entire class of registers depends on the target
1897 flags, you may indicate this to GCC by using this macro to modify
1898 @code{fixed_regs} and @code{call_used_regs} to 1 for each of the
1899 registers in the classes which should not be used by GCC@. Also make
1900 @code{define_register_constraint}s return @code{NO_REGS} for constraints
1901 that shouldn't be used.
1903 (However, if this class is not included in @code{GENERAL_REGS} and all
1904 of the insn patterns whose constraints permit this class are
1905 controlled by target switches, then GCC will automatically avoid using
1906 these registers when the target switches are opposed to them.)
1909 @defmac INCOMING_REGNO (@var{out})
1910 Define this macro if the target machine has register windows. This C
1911 expression returns the register number as seen by the called function
1912 corresponding to the register number @var{out} as seen by the calling
1913 function. Return @var{out} if register number @var{out} is not an
1917 @defmac OUTGOING_REGNO (@var{in})
1918 Define this macro if the target machine has register windows. This C
1919 expression returns the register number as seen by the calling function
1920 corresponding to the register number @var{in} as seen by the called
1921 function. Return @var{in} if register number @var{in} is not an inbound
1925 @defmac LOCAL_REGNO (@var{regno})
1926 Define this macro if the target machine has register windows. This C
1927 expression returns true if the register is call-saved but is in the
1928 register window. Unlike most call-saved registers, such registers
1929 need not be explicitly restored on function exit or during non-local
1934 If the program counter has a register number, define this as that
1935 register number. Otherwise, do not define it.
1938 @node Allocation Order
1939 @subsection Order of Allocation of Registers
1940 @cindex order of register allocation
1941 @cindex register allocation order
1943 @c prevent bad page break with this line
1944 Registers are allocated in order.
1946 @defmac REG_ALLOC_ORDER
1947 If defined, an initializer for a vector of integers, containing the
1948 numbers of hard registers in the order in which GCC should prefer
1949 to use them (from most preferred to least).
1951 If this macro is not defined, registers are used lowest numbered first
1952 (all else being equal).
1954 One use of this macro is on machines where the highest numbered
1955 registers must always be saved and the save-multiple-registers
1956 instruction supports only sequences of consecutive registers. On such
1957 machines, define @code{REG_ALLOC_ORDER} to be an initializer that lists
1958 the highest numbered allocable register first.
1961 @defmac ADJUST_REG_ALLOC_ORDER
1962 A C statement (sans semicolon) to choose the order in which to allocate
1963 hard registers for pseudo-registers local to a basic block.
1965 Store the desired register order in the array @code{reg_alloc_order}.
1966 Element 0 should be the register to allocate first; element 1, the next
1967 register; and so on.
1969 The macro body should not assume anything about the contents of
1970 @code{reg_alloc_order} before execution of the macro.
1972 On most machines, it is not necessary to define this macro.
1975 @defmac HONOR_REG_ALLOC_ORDER
1976 Normally, IRA tries to estimate the costs for saving a register in the
1977 prologue and restoring it in the epilogue. This discourages it from
1978 using call-saved registers. If a machine wants to ensure that IRA
1979 allocates registers in the order given by REG_ALLOC_ORDER even if some
1980 call-saved registers appear earlier than call-used ones, then define this
1981 macro as a C expression to nonzero. Default is 0.
1984 @defmac IRA_HARD_REGNO_ADD_COST_MULTIPLIER (@var{regno})
1985 In some case register allocation order is not enough for the
1986 Integrated Register Allocator (@acronym{IRA}) to generate a good code.
1987 If this macro is defined, it should return a floating point value
1988 based on @var{regno}. The cost of using @var{regno} for a pseudo will
1989 be increased by approximately the pseudo's usage frequency times the
1990 value returned by this macro. Not defining this macro is equivalent
1991 to having it always return @code{0.0}.
1993 On most machines, it is not necessary to define this macro.
1996 @node Values in Registers
1997 @subsection How Values Fit in Registers
1999 This section discusses the macros that describe which kinds of values
2000 (specifically, which machine modes) each register can hold, and how many
2001 consecutive registers are needed for a given mode.
2003 @defmac HARD_REGNO_NREGS (@var{regno}, @var{mode})
2004 A C expression for the number of consecutive hard registers, starting
2005 at register number @var{regno}, required to hold a value of mode
2006 @var{mode}. This macro must never return zero, even if a register
2007 cannot hold the requested mode - indicate that with HARD_REGNO_MODE_OK
2008 and/or CANNOT_CHANGE_MODE_CLASS instead.
2010 On a machine where all registers are exactly one word, a suitable
2011 definition of this macro is
2014 #define HARD_REGNO_NREGS(REGNO, MODE) \
2015 ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) \
2020 @defmac HARD_REGNO_NREGS_HAS_PADDING (@var{regno}, @var{mode})
2021 A C expression that is nonzero if a value of mode @var{mode}, stored
2022 in memory, ends with padding that causes it to take up more space than
2023 in registers starting at register number @var{regno} (as determined by
2024 multiplying GCC's notion of the size of the register when containing
2025 this mode by the number of registers returned by
2026 @code{HARD_REGNO_NREGS}). By default this is zero.
2028 For example, if a floating-point value is stored in three 32-bit
2029 registers but takes up 128 bits in memory, then this would be
2032 This macros only needs to be defined if there are cases where
2033 @code{subreg_get_info}
2034 would otherwise wrongly determine that a @code{subreg} can be
2035 represented by an offset to the register number, when in fact such a
2036 @code{subreg} would contain some of the padding not stored in
2037 registers and so not be representable.
2040 @defmac HARD_REGNO_NREGS_WITH_PADDING (@var{regno}, @var{mode})
2041 For values of @var{regno} and @var{mode} for which
2042 @code{HARD_REGNO_NREGS_HAS_PADDING} returns nonzero, a C expression
2043 returning the greater number of registers required to hold the value
2044 including any padding. In the example above, the value would be four.
2047 @defmac REGMODE_NATURAL_SIZE (@var{mode})
2048 Define this macro if the natural size of registers that hold values
2049 of mode @var{mode} is not the word size. It is a C expression that
2050 should give the natural size in bytes for the specified mode. It is
2051 used by the register allocator to try to optimize its results. This
2052 happens for example on SPARC 64-bit where the natural size of
2053 floating-point registers is still 32-bit.
2056 @defmac HARD_REGNO_MODE_OK (@var{regno}, @var{mode})
2057 A C expression that is nonzero if it is permissible to store a value
2058 of mode @var{mode} in hard register number @var{regno} (or in several
2059 registers starting with that one). For a machine where all registers
2060 are equivalent, a suitable definition is
2063 #define HARD_REGNO_MODE_OK(REGNO, MODE) 1
2066 You need not include code to check for the numbers of fixed registers,
2067 because the allocation mechanism considers them to be always occupied.
2069 @cindex register pairs
2070 On some machines, double-precision values must be kept in even/odd
2071 register pairs. You can implement that by defining this macro to reject
2072 odd register numbers for such modes.
2074 The minimum requirement for a mode to be OK in a register is that the
2075 @samp{mov@var{mode}} instruction pattern support moves between the
2076 register and other hard register in the same class and that moving a
2077 value into the register and back out not alter it.
2079 Since the same instruction used to move @code{word_mode} will work for
2080 all narrower integer modes, it is not necessary on any machine for
2081 @code{HARD_REGNO_MODE_OK} to distinguish between these modes, provided
2082 you define patterns @samp{movhi}, etc., to take advantage of this. This
2083 is useful because of the interaction between @code{HARD_REGNO_MODE_OK}
2084 and @code{MODES_TIEABLE_P}; it is very desirable for all integer modes
2087 Many machines have special registers for floating point arithmetic.
2088 Often people assume that floating point machine modes are allowed only
2089 in floating point registers. This is not true. Any registers that
2090 can hold integers can safely @emph{hold} a floating point machine
2091 mode, whether or not floating arithmetic can be done on it in those
2092 registers. Integer move instructions can be used to move the values.
2094 On some machines, though, the converse is true: fixed-point machine
2095 modes may not go in floating registers. This is true if the floating
2096 registers normalize any value stored in them, because storing a
2097 non-floating value there would garble it. In this case,
2098 @code{HARD_REGNO_MODE_OK} should reject fixed-point machine modes in
2099 floating registers. But if the floating registers do not automatically
2100 normalize, if you can store any bit pattern in one and retrieve it
2101 unchanged without a trap, then any machine mode may go in a floating
2102 register, so you can define this macro to say so.
2104 The primary significance of special floating registers is rather that
2105 they are the registers acceptable in floating point arithmetic
2106 instructions. However, this is of no concern to
2107 @code{HARD_REGNO_MODE_OK}. You handle it by writing the proper
2108 constraints for those instructions.
2110 On some machines, the floating registers are especially slow to access,
2111 so that it is better to store a value in a stack frame than in such a
2112 register if floating point arithmetic is not being done. As long as the
2113 floating registers are not in class @code{GENERAL_REGS}, they will not
2114 be used unless some pattern's constraint asks for one.
2117 @defmac HARD_REGNO_RENAME_OK (@var{from}, @var{to})
2118 A C expression that is nonzero if it is OK to rename a hard register
2119 @var{from} to another hard register @var{to}.
2121 One common use of this macro is to prevent renaming of a register to
2122 another register that is not saved by a prologue in an interrupt
2125 The default is always nonzero.
2128 @defmac MODES_TIEABLE_P (@var{mode1}, @var{mode2})
2129 A C expression that is nonzero if a value of mode
2130 @var{mode1} is accessible in mode @var{mode2} without copying.
2132 If @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode1})} and
2133 @code{HARD_REGNO_MODE_OK (@var{r}, @var{mode2})} are always the same for
2134 any @var{r}, then @code{MODES_TIEABLE_P (@var{mode1}, @var{mode2})}
2135 should be nonzero. If they differ for any @var{r}, you should define
2136 this macro to return zero unless some other mechanism ensures the
2137 accessibility of the value in a narrower mode.
2139 You should define this macro to return nonzero in as many cases as
2140 possible since doing so will allow GCC to perform better register
2144 @deftypefn {Target Hook} bool TARGET_HARD_REGNO_SCRATCH_OK (unsigned int @var{regno})
2145 This target hook should return @code{true} if it is OK to use a hard register
2146 @var{regno} as scratch reg in peephole2.
2148 One common use of this macro is to prevent using of a register that
2149 is not saved by a prologue in an interrupt handler.
2151 The default version of this hook always returns @code{true}.
2154 @defmac AVOID_CCMODE_COPIES
2155 Define this macro if the compiler should avoid copies to/from @code{CCmode}
2156 registers. You should only define this macro if support for copying to/from
2157 @code{CCmode} is incomplete.
2160 @node Leaf Functions
2161 @subsection Handling Leaf Functions
2163 @cindex leaf functions
2164 @cindex functions, leaf
2165 On some machines, a leaf function (i.e., one which makes no calls) can run
2166 more efficiently if it does not make its own register window. Often this
2167 means it is required to receive its arguments in the registers where they
2168 are passed by the caller, instead of the registers where they would
2171 The special treatment for leaf functions generally applies only when
2172 other conditions are met; for example, often they may use only those
2173 registers for its own variables and temporaries. We use the term ``leaf
2174 function'' to mean a function that is suitable for this special
2175 handling, so that functions with no calls are not necessarily ``leaf
2178 GCC assigns register numbers before it knows whether the function is
2179 suitable for leaf function treatment. So it needs to renumber the
2180 registers in order to output a leaf function. The following macros
2183 @defmac LEAF_REGISTERS
2184 Name of a char vector, indexed by hard register number, which
2185 contains 1 for a register that is allowable in a candidate for leaf
2188 If leaf function treatment involves renumbering the registers, then the
2189 registers marked here should be the ones before renumbering---those that
2190 GCC would ordinarily allocate. The registers which will actually be
2191 used in the assembler code, after renumbering, should not be marked with 1
2194 Define this macro only if the target machine offers a way to optimize
2195 the treatment of leaf functions.
2198 @defmac LEAF_REG_REMAP (@var{regno})
2199 A C expression whose value is the register number to which @var{regno}
2200 should be renumbered, when a function is treated as a leaf function.
2202 If @var{regno} is a register number which should not appear in a leaf
2203 function before renumbering, then the expression should yield @minus{}1, which
2204 will cause the compiler to abort.
2206 Define this macro only if the target machine offers a way to optimize the
2207 treatment of leaf functions, and registers need to be renumbered to do
2211 @findex current_function_is_leaf
2212 @findex current_function_uses_only_leaf_regs
2213 @code{TARGET_ASM_FUNCTION_PROLOGUE} and
2214 @code{TARGET_ASM_FUNCTION_EPILOGUE} must usually treat leaf functions
2215 specially. They can test the C variable @code{current_function_is_leaf}
2216 which is nonzero for leaf functions. @code{current_function_is_leaf} is
2217 set prior to local register allocation and is valid for the remaining
2218 compiler passes. They can also test the C variable
2219 @code{current_function_uses_only_leaf_regs} which is nonzero for leaf
2220 functions which only use leaf registers.
2221 @code{current_function_uses_only_leaf_regs} is valid after all passes
2222 that modify the instructions have been run and is only useful if
2223 @code{LEAF_REGISTERS} is defined.
2224 @c changed this to fix overfull. ALSO: why the "it" at the beginning
2225 @c of the next paragraph?! --mew 2feb93
2227 @node Stack Registers
2228 @subsection Registers That Form a Stack
2230 There are special features to handle computers where some of the
2231 ``registers'' form a stack. Stack registers are normally written by
2232 pushing onto the stack, and are numbered relative to the top of the
2235 Currently, GCC can only handle one group of stack-like registers, and
2236 they must be consecutively numbered. Furthermore, the existing
2237 support for stack-like registers is specific to the 80387 floating
2238 point coprocessor. If you have a new architecture that uses
2239 stack-like registers, you will need to do substantial work on
2240 @file{reg-stack.c} and write your machine description to cooperate
2241 with it, as well as defining these macros.
2244 Define this if the machine has any stack-like registers.
2247 @defmac STACK_REG_COVER_CLASS
2248 This is a cover class containing the stack registers. Define this if
2249 the machine has any stack-like registers.
2252 @defmac FIRST_STACK_REG
2253 The number of the first stack-like register. This one is the top
2257 @defmac LAST_STACK_REG
2258 The number of the last stack-like register. This one is the bottom of
2262 @node Register Classes
2263 @section Register Classes
2264 @cindex register class definitions
2265 @cindex class definitions, register
2267 On many machines, the numbered registers are not all equivalent.
2268 For example, certain registers may not be allowed for indexed addressing;
2269 certain registers may not be allowed in some instructions. These machine
2270 restrictions are described to the compiler using @dfn{register classes}.
2272 You define a number of register classes, giving each one a name and saying
2273 which of the registers belong to it. Then you can specify register classes
2274 that are allowed as operands to particular instruction patterns.
2278 In general, each register will belong to several classes. In fact, one
2279 class must be named @code{ALL_REGS} and contain all the registers. Another
2280 class must be named @code{NO_REGS} and contain no registers. Often the
2281 union of two classes will be another class; however, this is not required.
2283 @findex GENERAL_REGS
2284 One of the classes must be named @code{GENERAL_REGS}. There is nothing
2285 terribly special about the name, but the operand constraint letters
2286 @samp{r} and @samp{g} specify this class. If @code{GENERAL_REGS} is
2287 the same as @code{ALL_REGS}, just define it as a macro which expands
2290 Order the classes so that if class @var{x} is contained in class @var{y}
2291 then @var{x} has a lower class number than @var{y}.
2293 The way classes other than @code{GENERAL_REGS} are specified in operand
2294 constraints is through machine-dependent operand constraint letters.
2295 You can define such letters to correspond to various classes, then use
2296 them in operand constraints.
2298 You must define the narrowest register classes for allocatable
2299 registers, so that each class either has no subclasses, or that for
2300 some mode, the move cost between registers within the class is
2301 cheaper than moving a register in the class to or from memory
2304 You should define a class for the union of two classes whenever some
2305 instruction allows both classes. For example, if an instruction allows
2306 either a floating point (coprocessor) register or a general register for a
2307 certain operand, you should define a class @code{FLOAT_OR_GENERAL_REGS}
2308 which includes both of them. Otherwise you will get suboptimal code,
2309 or even internal compiler errors when reload cannot find a register in the
2310 class computed via @code{reg_class_subunion}.
2312 You must also specify certain redundant information about the register
2313 classes: for each class, which classes contain it and which ones are
2314 contained in it; for each pair of classes, the largest class contained
2317 When a value occupying several consecutive registers is expected in a
2318 certain class, all the registers used must belong to that class.
2319 Therefore, register classes cannot be used to enforce a requirement for
2320 a register pair to start with an even-numbered register. The way to
2321 specify this requirement is with @code{HARD_REGNO_MODE_OK}.
2323 Register classes used for input-operands of bitwise-and or shift
2324 instructions have a special requirement: each such class must have, for
2325 each fixed-point machine mode, a subclass whose registers can transfer that
2326 mode to or from memory. For example, on some machines, the operations for
2327 single-byte values (@code{QImode}) are limited to certain registers. When
2328 this is so, each register class that is used in a bitwise-and or shift
2329 instruction must have a subclass consisting of registers from which
2330 single-byte values can be loaded or stored. This is so that
2331 @code{PREFERRED_RELOAD_CLASS} can always have a possible value to return.
2333 @deftp {Data type} {enum reg_class}
2334 An enumerated type that must be defined with all the register class names
2335 as enumerated values. @code{NO_REGS} must be first. @code{ALL_REGS}
2336 must be the last register class, followed by one more enumerated value,
2337 @code{LIM_REG_CLASSES}, which is not a register class but rather
2338 tells how many classes there are.
2340 Each register class has a number, which is the value of casting
2341 the class name to type @code{int}. The number serves as an index
2342 in many of the tables described below.
2345 @defmac N_REG_CLASSES
2346 The number of distinct register classes, defined as follows:
2349 #define N_REG_CLASSES (int) LIM_REG_CLASSES
2353 @defmac REG_CLASS_NAMES
2354 An initializer containing the names of the register classes as C string
2355 constants. These names are used in writing some of the debugging dumps.
2358 @defmac REG_CLASS_CONTENTS
2359 An initializer containing the contents of the register classes, as integers
2360 which are bit masks. The @var{n}th integer specifies the contents of class
2361 @var{n}. The way the integer @var{mask} is interpreted is that
2362 register @var{r} is in the class if @code{@var{mask} & (1 << @var{r})} is 1.
2364 When the machine has more than 32 registers, an integer does not suffice.
2365 Then the integers are replaced by sub-initializers, braced groupings containing
2366 several integers. Each sub-initializer must be suitable as an initializer
2367 for the type @code{HARD_REG_SET} which is defined in @file{hard-reg-set.h}.
2368 In this situation, the first integer in each sub-initializer corresponds to
2369 registers 0 through 31, the second integer to registers 32 through 63, and
2373 @defmac REGNO_REG_CLASS (@var{regno})
2374 A C expression whose value is a register class containing hard register
2375 @var{regno}. In general there is more than one such class; choose a class
2376 which is @dfn{minimal}, meaning that no smaller class also contains the
2380 @defmac BASE_REG_CLASS
2381 A macro whose definition is the name of the class to which a valid
2382 base register must belong. A base register is one used in an address
2383 which is the register value plus a displacement.
2386 @defmac MODE_BASE_REG_CLASS (@var{mode})
2387 This is a variation of the @code{BASE_REG_CLASS} macro which allows
2388 the selection of a base register in a mode dependent manner. If
2389 @var{mode} is VOIDmode then it should return the same value as
2390 @code{BASE_REG_CLASS}.
2393 @defmac MODE_BASE_REG_REG_CLASS (@var{mode})
2394 A C expression whose value is the register class to which a valid
2395 base register must belong in order to be used in a base plus index
2396 register address. You should define this macro if base plus index
2397 addresses have different requirements than other base register uses.
2400 @defmac MODE_CODE_BASE_REG_CLASS (@var{mode}, @var{address_space}, @var{outer_code}, @var{index_code})
2401 A C expression whose value is the register class to which a valid
2402 base register for a memory reference in mode @var{mode} to address
2403 space @var{address_space} must belong. @var{outer_code} and @var{index_code}
2404 define the context in which the base register occurs. @var{outer_code} is
2405 the code of the immediately enclosing expression (@code{MEM} for the top level
2406 of an address, @code{ADDRESS} for something that occurs in an
2407 @code{address_operand}). @var{index_code} is the code of the corresponding
2408 index expression if @var{outer_code} is @code{PLUS}; @code{SCRATCH} otherwise.
2411 @defmac INDEX_REG_CLASS
2412 A macro whose definition is the name of the class to which a valid
2413 index register must belong. An index register is one used in an
2414 address where its value is either multiplied by a scale factor or
2415 added to another register (as well as added to a displacement).
2418 @defmac REGNO_OK_FOR_BASE_P (@var{num})
2419 A C expression which is nonzero if register number @var{num} is
2420 suitable for use as a base register in operand addresses.
2423 @defmac REGNO_MODE_OK_FOR_BASE_P (@var{num}, @var{mode})
2424 A C expression that is just like @code{REGNO_OK_FOR_BASE_P}, except that
2425 that expression may examine the mode of the memory reference in
2426 @var{mode}. You should define this macro if the mode of the memory
2427 reference affects whether a register may be used as a base register. If
2428 you define this macro, the compiler will use it instead of
2429 @code{REGNO_OK_FOR_BASE_P}. The mode may be @code{VOIDmode} for
2430 addresses that appear outside a @code{MEM}, i.e., as an
2431 @code{address_operand}.
2434 @defmac REGNO_MODE_OK_FOR_REG_BASE_P (@var{num}, @var{mode})
2435 A C expression which is nonzero if register number @var{num} is suitable for
2436 use as a base register in base plus index operand addresses, accessing
2437 memory in mode @var{mode}. It may be either a suitable hard register or a
2438 pseudo register that has been allocated such a hard register. You should
2439 define this macro if base plus index addresses have different requirements
2440 than other base register uses.
2442 Use of this macro is deprecated; please use the more general
2443 @code{REGNO_MODE_CODE_OK_FOR_BASE_P}.
2446 @defmac REGNO_MODE_CODE_OK_FOR_BASE_P (@var{num}, @var{mode}, @var{address_space}, @var{outer_code}, @var{index_code})
2447 A C expression which is nonzero if register number @var{num} is
2448 suitable for use as a base register in operand addresses, accessing
2449 memory in mode @var{mode} in address space @var{address_space}.
2450 This is similar to @code{REGNO_MODE_OK_FOR_BASE_P}, except
2451 that that expression may examine the context in which the register
2452 appears in the memory reference. @var{outer_code} is the code of the
2453 immediately enclosing expression (@code{MEM} if at the top level of the
2454 address, @code{ADDRESS} for something that occurs in an
2455 @code{address_operand}). @var{index_code} is the code of the
2456 corresponding index expression if @var{outer_code} is @code{PLUS};
2457 @code{SCRATCH} otherwise. The mode may be @code{VOIDmode} for addresses
2458 that appear outside a @code{MEM}, i.e., as an @code{address_operand}.
2461 @defmac REGNO_OK_FOR_INDEX_P (@var{num})
2462 A C expression which is nonzero if register number @var{num} is
2463 suitable for use as an index register in operand addresses. It may be
2464 either a suitable hard register or a pseudo register that has been
2465 allocated such a hard register.
2467 The difference between an index register and a base register is that
2468 the index register may be scaled. If an address involves the sum of
2469 two registers, neither one of them scaled, then either one may be
2470 labeled the ``base'' and the other the ``index''; but whichever
2471 labeling is used must fit the machine's constraints of which registers
2472 may serve in each capacity. The compiler will try both labelings,
2473 looking for one that is valid, and will reload one or both registers
2474 only if neither labeling works.
2477 @deftypefn {Target Hook} reg_class_t TARGET_PREFERRED_RENAME_CLASS (reg_class_t @var{rclass})
2478 A target hook that places additional preference on the register class to use when it is necessary to rename a register in class @var{rclass} to another class, or perhaps @var{NO_REGS}, if no preferred register class is found or hook @code{preferred_rename_class} is not implemented. Sometimes returning a more restrictive class makes better code. For example, on ARM, thumb-2 instructions using @code{LO_REGS} may be smaller than instructions using @code{GENERIC_REGS}. By returning @code{LO_REGS} from @code{preferred_rename_class}, code size can be reduced.
2481 @deftypefn {Target Hook} reg_class_t TARGET_PREFERRED_RELOAD_CLASS (rtx @var{x}, reg_class_t @var{rclass})
2482 A target hook that places additional restrictions on the register class
2483 to use when it is necessary to copy value @var{x} into a register in class
2484 @var{rclass}. The value is a register class; perhaps @var{rclass}, or perhaps
2485 another, smaller class.
2487 The default version of this hook always returns value of @code{rclass} argument.
2489 Sometimes returning a more restrictive class makes better code. For
2490 example, on the 68000, when @var{x} is an integer constant that is in range
2491 for a @samp{moveq} instruction, the value of this macro is always
2492 @code{DATA_REGS} as long as @var{rclass} includes the data registers.
2493 Requiring a data register guarantees that a @samp{moveq} will be used.
2495 One case where @code{TARGET_PREFERRED_RELOAD_CLASS} must not return
2496 @var{rclass} is if @var{x} is a legitimate constant which cannot be
2497 loaded into some register class. By returning @code{NO_REGS} you can
2498 force @var{x} into a memory location. For example, rs6000 can load
2499 immediate values into general-purpose registers, but does not have an
2500 instruction for loading an immediate value into a floating-point
2501 register, so @code{TARGET_PREFERRED_RELOAD_CLASS} returns @code{NO_REGS} when
2502 @var{x} is a floating-point constant. If the constant can't be loaded
2503 into any kind of register, code generation will be better if
2504 @code{TARGET_LEGITIMATE_CONSTANT_P} makes the constant illegitimate instead
2505 of using @code{TARGET_PREFERRED_RELOAD_CLASS}.
2507 If an insn has pseudos in it after register allocation, reload will go
2508 through the alternatives and call repeatedly @code{TARGET_PREFERRED_RELOAD_CLASS}
2509 to find the best one. Returning @code{NO_REGS}, in this case, makes
2510 reload add a @code{!} in front of the constraint: the x86 back-end uses
2511 this feature to discourage usage of 387 registers when math is done in
2512 the SSE registers (and vice versa).
2515 @defmac PREFERRED_RELOAD_CLASS (@var{x}, @var{class})
2516 A C expression that places additional restrictions on the register class
2517 to use when it is necessary to copy value @var{x} into a register in class
2518 @var{class}. The value is a register class; perhaps @var{class}, or perhaps
2519 another, smaller class. On many machines, the following definition is
2523 #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS
2526 Sometimes returning a more restrictive class makes better code. For
2527 example, on the 68000, when @var{x} is an integer constant that is in range
2528 for a @samp{moveq} instruction, the value of this macro is always
2529 @code{DATA_REGS} as long as @var{class} includes the data registers.
2530 Requiring a data register guarantees that a @samp{moveq} will be used.
2532 One case where @code{PREFERRED_RELOAD_CLASS} must not return
2533 @var{class} is if @var{x} is a legitimate constant which cannot be
2534 loaded into some register class. By returning @code{NO_REGS} you can
2535 force @var{x} into a memory location. For example, rs6000 can load
2536 immediate values into general-purpose registers, but does not have an
2537 instruction for loading an immediate value into a floating-point
2538 register, so @code{PREFERRED_RELOAD_CLASS} returns @code{NO_REGS} when
2539 @var{x} is a floating-point constant. If the constant can't be loaded
2540 into any kind of register, code generation will be better if
2541 @code{TARGET_LEGITIMATE_CONSTANT_P} makes the constant illegitimate instead
2542 of using @code{TARGET_PREFERRED_RELOAD_CLASS}.
2544 If an insn has pseudos in it after register allocation, reload will go
2545 through the alternatives and call repeatedly @code{PREFERRED_RELOAD_CLASS}
2546 to find the best one. Returning @code{NO_REGS}, in this case, makes
2547 reload add a @code{!} in front of the constraint: the x86 back-end uses
2548 this feature to discourage usage of 387 registers when math is done in
2549 the SSE registers (and vice versa).
2552 @deftypefn {Target Hook} reg_class_t TARGET_PREFERRED_OUTPUT_RELOAD_CLASS (rtx @var{x}, reg_class_t @var{rclass})
2553 Like @code{TARGET_PREFERRED_RELOAD_CLASS}, but for output reloads instead of
2556 The default version of this hook always returns value of @code{rclass}
2559 You can also use @code{TARGET_PREFERRED_OUTPUT_RELOAD_CLASS} to discourage
2560 reload from using some alternatives, like @code{TARGET_PREFERRED_RELOAD_CLASS}.
2563 @defmac LIMIT_RELOAD_CLASS (@var{mode}, @var{class})
2564 A C expression that places additional restrictions on the register class
2565 to use when it is necessary to be able to hold a value of mode
2566 @var{mode} in a reload register for which class @var{class} would
2569 Unlike @code{PREFERRED_RELOAD_CLASS}, this macro should be used when
2570 there are certain modes that simply can't go in certain reload classes.
2572 The value is a register class; perhaps @var{class}, or perhaps another,
2575 Don't define this macro unless the target machine has limitations which
2576 require the macro to do something nontrivial.
2579 @deftypefn {Target Hook} reg_class_t TARGET_SECONDARY_RELOAD (bool @var{in_p}, rtx @var{x}, reg_class_t @var{reload_class}, machine_mode @var{reload_mode}, secondary_reload_info *@var{sri})
2580 Many machines have some registers that cannot be copied directly to or
2581 from memory or even from other types of registers. An example is the
2582 @samp{MQ} register, which on most machines, can only be copied to or
2583 from general registers, but not memory. Below, we shall be using the
2584 term 'intermediate register' when a move operation cannot be performed
2585 directly, but has to be done by copying the source into the intermediate
2586 register first, and then copying the intermediate register to the
2587 destination. An intermediate register always has the same mode as
2588 source and destination. Since it holds the actual value being copied,
2589 reload might apply optimizations to re-use an intermediate register
2590 and eliding the copy from the source when it can determine that the
2591 intermediate register still holds the required value.
2593 Another kind of secondary reload is required on some machines which
2594 allow copying all registers to and from memory, but require a scratch
2595 register for stores to some memory locations (e.g., those with symbolic
2596 address on the RT, and those with certain symbolic address on the SPARC
2597 when compiling PIC)@. Scratch registers need not have the same mode
2598 as the value being copied, and usually hold a different value than
2599 that being copied. Special patterns in the md file are needed to
2600 describe how the copy is performed with the help of the scratch register;
2601 these patterns also describe the number, register class(es) and mode(s)
2602 of the scratch register(s).
2604 In some cases, both an intermediate and a scratch register are required.
2606 For input reloads, this target hook is called with nonzero @var{in_p},
2607 and @var{x} is an rtx that needs to be copied to a register of class
2608 @var{reload_class} in @var{reload_mode}. For output reloads, this target
2609 hook is called with zero @var{in_p}, and a register of class @var{reload_class}
2610 needs to be copied to rtx @var{x} in @var{reload_mode}.
2612 If copying a register of @var{reload_class} from/to @var{x} requires
2613 an intermediate register, the hook @code{secondary_reload} should
2614 return the register class required for this intermediate register.
2615 If no intermediate register is required, it should return NO_REGS.
2616 If more than one intermediate register is required, describe the one
2617 that is closest in the copy chain to the reload register.
2619 If scratch registers are needed, you also have to describe how to
2620 perform the copy from/to the reload register to/from this
2621 closest intermediate register. Or if no intermediate register is
2622 required, but still a scratch register is needed, describe the
2623 copy from/to the reload register to/from the reload operand @var{x}.
2625 You do this by setting @code{sri->icode} to the instruction code of a pattern
2626 in the md file which performs the move. Operands 0 and 1 are the output
2627 and input of this copy, respectively. Operands from operand 2 onward are
2628 for scratch operands. These scratch operands must have a mode, and a
2629 single-register-class
2630 @c [later: or memory]
2633 When an intermediate register is used, the @code{secondary_reload}
2634 hook will be called again to determine how to copy the intermediate
2635 register to/from the reload operand @var{x}, so your hook must also
2636 have code to handle the register class of the intermediate operand.
2638 @c [For later: maybe we'll allow multi-alternative reload patterns -
2639 @c the port maintainer could name a mov<mode> pattern that has clobbers -
2640 @c and match the constraints of input and output to determine the required
2641 @c alternative. A restriction would be that constraints used to match
2642 @c against reloads registers would have to be written as register class
2643 @c constraints, or we need a new target macro / hook that tells us if an
2644 @c arbitrary constraint can match an unknown register of a given class.
2645 @c Such a macro / hook would also be useful in other places.]
2648 @var{x} might be a pseudo-register or a @code{subreg} of a
2649 pseudo-register, which could either be in a hard register or in memory.
2650 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2651 in memory and the hard register number if it is in a register.
2653 Scratch operands in memory (constraint @code{"=m"} / @code{"=&m"}) are
2654 currently not supported. For the time being, you will have to continue
2655 to use @code{SECONDARY_MEMORY_NEEDED} for that purpose.
2657 @code{copy_cost} also uses this target hook to find out how values are
2658 copied. If you want it to include some extra cost for the need to allocate
2659 (a) scratch register(s), set @code{sri->extra_cost} to the additional cost.
2660 Or if two dependent moves are supposed to have a lower cost than the sum
2661 of the individual moves due to expected fortuitous scheduling and/or special
2662 forwarding logic, you can set @code{sri->extra_cost} to a negative amount.
2665 @defmac SECONDARY_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2666 @defmacx SECONDARY_INPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2667 @defmacx SECONDARY_OUTPUT_RELOAD_CLASS (@var{class}, @var{mode}, @var{x})
2668 These macros are obsolete, new ports should use the target hook
2669 @code{TARGET_SECONDARY_RELOAD} instead.
2671 These are obsolete macros, replaced by the @code{TARGET_SECONDARY_RELOAD}
2672 target hook. Older ports still define these macros to indicate to the
2673 reload phase that it may
2674 need to allocate at least one register for a reload in addition to the
2675 register to contain the data. Specifically, if copying @var{x} to a
2676 register @var{class} in @var{mode} requires an intermediate register,
2677 you were supposed to define @code{SECONDARY_INPUT_RELOAD_CLASS} to return the
2678 largest register class all of whose registers can be used as
2679 intermediate registers or scratch registers.
2681 If copying a register @var{class} in @var{mode} to @var{x} requires an
2682 intermediate or scratch register, @code{SECONDARY_OUTPUT_RELOAD_CLASS}
2683 was supposed to be defined be defined to return the largest register
2684 class required. If the
2685 requirements for input and output reloads were the same, the macro
2686 @code{SECONDARY_RELOAD_CLASS} should have been used instead of defining both
2689 The values returned by these macros are often @code{GENERAL_REGS}.
2690 Return @code{NO_REGS} if no spare register is needed; i.e., if @var{x}
2691 can be directly copied to or from a register of @var{class} in
2692 @var{mode} without requiring a scratch register. Do not define this
2693 macro if it would always return @code{NO_REGS}.
2695 If a scratch register is required (either with or without an
2696 intermediate register), you were supposed to define patterns for
2697 @samp{reload_in@var{m}} or @samp{reload_out@var{m}}, as required
2698 (@pxref{Standard Names}. These patterns, which were normally
2699 implemented with a @code{define_expand}, should be similar to the
2700 @samp{mov@var{m}} patterns, except that operand 2 is the scratch
2703 These patterns need constraints for the reload register and scratch
2705 contain a single register class. If the original reload register (whose
2706 class is @var{class}) can meet the constraint given in the pattern, the
2707 value returned by these macros is used for the class of the scratch
2708 register. Otherwise, two additional reload registers are required.
2709 Their classes are obtained from the constraints in the insn pattern.
2711 @var{x} might be a pseudo-register or a @code{subreg} of a
2712 pseudo-register, which could either be in a hard register or in memory.
2713 Use @code{true_regnum} to find out; it will return @minus{}1 if the pseudo is
2714 in memory and the hard register number if it is in a register.
2716 These macros should not be used in the case where a particular class of
2717 registers can only be copied to memory and not to another class of
2718 registers. In that case, secondary reload registers are not needed and
2719 would not be helpful. Instead, a stack location must be used to perform
2720 the copy and the @code{mov@var{m}} pattern should use memory as an
2721 intermediate storage. This case often occurs between floating-point and
2725 @defmac SECONDARY_MEMORY_NEEDED (@var{class1}, @var{class2}, @var{m})
2726 Certain machines have the property that some registers cannot be copied
2727 to some other registers without using memory. Define this macro on
2728 those machines to be a C expression that is nonzero if objects of mode
2729 @var{m} in registers of @var{class1} can only be copied to registers of
2730 class @var{class2} by storing a register of @var{class1} into memory
2731 and loading that memory location into a register of @var{class2}.
2733 Do not define this macro if its value would always be zero.
2736 @defmac SECONDARY_MEMORY_NEEDED_RTX (@var{mode})
2737 Normally when @code{SECONDARY_MEMORY_NEEDED} is defined, the compiler
2738 allocates a stack slot for a memory location needed for register copies.
2739 If this macro is defined, the compiler instead uses the memory location
2740 defined by this macro.
2742 Do not define this macro if you do not define
2743 @code{SECONDARY_MEMORY_NEEDED}.
2746 @defmac SECONDARY_MEMORY_NEEDED_MODE (@var{mode})
2747 When the compiler needs a secondary memory location to copy between two
2748 registers of mode @var{mode}, it normally allocates sufficient memory to
2749 hold a quantity of @code{BITS_PER_WORD} bits and performs the store and
2750 load operations in a mode that many bits wide and whose class is the
2751 same as that of @var{mode}.
2753 This is right thing to do on most machines because it ensures that all
2754 bits of the register are copied and prevents accesses to the registers
2755 in a narrower mode, which some machines prohibit for floating-point
2758 However, this default behavior is not correct on some machines, such as
2759 the DEC Alpha, that store short integers in floating-point registers
2760 differently than in integer registers. On those machines, the default
2761 widening will not work correctly and you must define this macro to
2762 suppress that widening in some cases. See the file @file{alpha.h} for
2765 Do not define this macro if you do not define
2766 @code{SECONDARY_MEMORY_NEEDED} or if widening @var{mode} to a mode that
2767 is @code{BITS_PER_WORD} bits wide is correct for your machine.
2770 @deftypefn {Target Hook} bool TARGET_CLASS_LIKELY_SPILLED_P (reg_class_t @var{rclass})
2771 A target hook which returns @code{true} if pseudos that have been assigned
2772 to registers of class @var{rclass} would likely be spilled because
2773 registers of @var{rclass} are needed for spill registers.
2775 The default version of this target hook returns @code{true} if @var{rclass}
2776 has exactly one register and @code{false} otherwise. On most machines, this
2777 default should be used. For generally register-starved machines, such as
2778 i386, or machines with right register constraints, such as SH, this hook
2779 can be used to avoid excessive spilling.
2781 This hook is also used by some of the global intra-procedural code
2782 transformations to throtle code motion, to avoid increasing register
2786 @deftypefn {Target Hook} {unsigned char} TARGET_CLASS_MAX_NREGS (reg_class_t @var{rclass}, machine_mode @var{mode})
2787 A target hook returns the maximum number of consecutive registers
2788 of class @var{rclass} needed to hold a value of mode @var{mode}.
2790 This is closely related to the macro @code{HARD_REGNO_NREGS}. In fact,
2791 the value returned by @code{TARGET_CLASS_MAX_NREGS (@var{rclass},
2792 @var{mode})} target hook should be the maximum value of
2793 @code{HARD_REGNO_NREGS (@var{regno}, @var{mode})} for all @var{regno}
2794 values in the class @var{rclass}.
2796 This target hook helps control the handling of multiple-word values
2799 The default version of this target hook returns the size of @var{mode}
2803 @defmac CLASS_MAX_NREGS (@var{class}, @var{mode})
2804 A C expression for the maximum number of consecutive registers
2805 of class @var{class} needed to hold a value of mode @var{mode}.
2807 This is closely related to the macro @code{HARD_REGNO_NREGS}. In fact,
2808 the value of the macro @code{CLASS_MAX_NREGS (@var{class}, @var{mode})}
2809 should be the maximum value of @code{HARD_REGNO_NREGS (@var{regno},
2810 @var{mode})} for all @var{regno} values in the class @var{class}.
2812 This macro helps control the handling of multiple-word values
2816 @defmac CANNOT_CHANGE_MODE_CLASS (@var{from}, @var{to}, @var{class})
2817 If defined, a C expression that returns nonzero for a @var{class} for which
2818 a change from mode @var{from} to mode @var{to} is invalid.
2820 For the example, loading 32-bit integer or floating-point objects into
2821 floating-point registers on the Alpha extends them to 64 bits.
2822 Therefore loading a 64-bit object and then storing it as a 32-bit object
2823 does not store the low-order 32 bits, as would be the case for a normal
2824 register. Therefore, @file{alpha.h} defines @code{CANNOT_CHANGE_MODE_CLASS}
2828 #define CANNOT_CHANGE_MODE_CLASS(FROM, TO, CLASS) \
2829 (GET_MODE_SIZE (FROM) != GET_MODE_SIZE (TO) \
2830 ? reg_classes_intersect_p (FLOAT_REGS, (CLASS)) : 0)
2834 @deftypefn {Target Hook} bool TARGET_LRA_P (void)
2835 A target hook which returns true if we use LRA instead of reload pass. It means that LRA was ported to the target. The default version of this target hook returns always false.
2838 @deftypefn {Target Hook} int TARGET_REGISTER_PRIORITY (int)
2839 A target hook which returns the register priority number to which the register @var{hard_regno} belongs to. The bigger the number, the more preferable the hard register usage (when all other conditions are the same). This hook can be used to prefer some hard register over others in LRA. For example, some x86-64 register usage needs additional prefix which makes instructions longer. The hook can return lower priority number for such registers make them less favorable and as result making the generated code smaller. The default version of this target hook returns always zero.
2842 @deftypefn {Target Hook} bool TARGET_REGISTER_USAGE_LEVELING_P (void)
2843 A target hook which returns true if we need register usage leveling. That means if a few hard registers are equally good for the assignment, we choose the least used hard register. The register usage leveling may be profitable for some targets. Don't use the usage leveling for targets with conditional execution or targets with big register files as it hurts if-conversion and cross-jumping optimizations. The default version of this target hook returns always false.
2846 @deftypefn {Target Hook} bool TARGET_DIFFERENT_ADDR_DISPLACEMENT_P (void)
2847 A target hook which returns true if an address with the same structure can have different maximal legitimate displacement. For example, the displacement can depend on memory mode or on operand combinations in the insn. The default version of this target hook returns always false.
2850 @deftypefn {Target Hook} reg_class_t TARGET_SPILL_CLASS (reg_class_t, @var{machine_mode})
2851 This hook defines a class of registers which could be used for spilling pseudos of the given mode and class, or @code{NO_REGS} if only memory should be used. Not defining this hook is equivalent to returning @code{NO_REGS} for all inputs.
2854 @deftypefn {Target Hook} machine_mode TARGET_CSTORE_MODE (enum insn_code @var{icode})
2855 This hook defines the machine mode to use for the boolean result of conditional store patterns. The ICODE argument is the instruction code for the cstore being performed. Not definiting this hook is the same as accepting the mode encoded into operand 0 of the cstore expander patterns.
2858 @node Stack and Calling
2859 @section Stack Layout and Calling Conventions
2860 @cindex calling conventions
2862 @c prevent bad page break with this line
2863 This describes the stack layout and calling conventions.
2867 * Exception Handling::
2872 * Register Arguments::
2874 * Aggregate Return::
2879 * Stack Smashing Protection::
2880 * Miscellaneous Register Hooks::
2884 @subsection Basic Stack Layout
2885 @cindex stack frame layout
2886 @cindex frame layout
2888 @c prevent bad page break with this line
2889 Here is the basic stack layout.
2891 @defmac STACK_GROWS_DOWNWARD
2892 Define this macro if pushing a word onto the stack moves the stack
2893 pointer to a smaller address.
2895 When we say, ``define this macro if @dots{}'', it means that the
2896 compiler checks this macro only with @code{#ifdef} so the precise
2897 definition used does not matter.
2900 @defmac STACK_PUSH_CODE
2901 This macro defines the operation used when something is pushed
2902 on the stack. In RTL, a push operation will be
2903 @code{(set (mem (STACK_PUSH_CODE (reg sp))) @dots{})}
2905 The choices are @code{PRE_DEC}, @code{POST_DEC}, @code{PRE_INC},
2906 and @code{POST_INC}. Which of these is correct depends on
2907 the stack direction and on whether the stack pointer points
2908 to the last item on the stack or whether it points to the
2909 space for the next item on the stack.
2911 The default is @code{PRE_DEC} when @code{STACK_GROWS_DOWNWARD} is
2912 defined, which is almost always right, and @code{PRE_INC} otherwise,
2913 which is often wrong.
2916 @defmac FRAME_GROWS_DOWNWARD
2917 Define this macro to nonzero value if the addresses of local variable slots
2918 are at negative offsets from the frame pointer.
2921 @defmac ARGS_GROW_DOWNWARD
2922 Define this macro if successive arguments to a function occupy decreasing
2923 addresses on the stack.
2926 @defmac STARTING_FRAME_OFFSET
2927 Offset from the frame pointer to the first local variable slot to be allocated.
2929 If @code{FRAME_GROWS_DOWNWARD}, find the next slot's offset by
2930 subtracting the first slot's length from @code{STARTING_FRAME_OFFSET}.
2931 Otherwise, it is found by adding the length of the first slot to the
2932 value @code{STARTING_FRAME_OFFSET}.
2933 @c i'm not sure if the above is still correct.. had to change it to get
2934 @c rid of an overfull. --mew 2feb93
2937 @defmac STACK_ALIGNMENT_NEEDED
2938 Define to zero to disable final alignment of the stack during reload.
2939 The nonzero default for this macro is suitable for most ports.
2941 On ports where @code{STARTING_FRAME_OFFSET} is nonzero or where there
2942 is a register save block following the local block that doesn't require
2943 alignment to @code{STACK_BOUNDARY}, it may be beneficial to disable
2944 stack alignment and do it in the backend.
2947 @defmac STACK_POINTER_OFFSET
2948 Offset from the stack pointer register to the first location at which
2949 outgoing arguments are placed. If not specified, the default value of
2950 zero is used. This is the proper value for most machines.
2952 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
2953 the first location at which outgoing arguments are placed.
2956 @defmac FIRST_PARM_OFFSET (@var{fundecl})
2957 Offset from the argument pointer register to the first argument's
2958 address. On some machines it may depend on the data type of the
2961 If @code{ARGS_GROW_DOWNWARD}, this is the offset to the location above
2962 the first argument's address.
2965 @defmac STACK_DYNAMIC_OFFSET (@var{fundecl})
2966 Offset from the stack pointer register to an item dynamically allocated
2967 on the stack, e.g., by @code{alloca}.
2969 The default value for this macro is @code{STACK_POINTER_OFFSET} plus the
2970 length of the outgoing arguments. The default is correct for most
2971 machines. See @file{function.c} for details.
2974 @defmac INITIAL_FRAME_ADDRESS_RTX
2975 A C expression whose value is RTL representing the address of the initial
2976 stack frame. This address is passed to @code{RETURN_ADDR_RTX} and
2977 @code{DYNAMIC_CHAIN_ADDRESS}. If you don't define this macro, a reasonable
2978 default value will be used. Define this macro in order to make frame pointer
2979 elimination work in the presence of @code{__builtin_frame_address (count)} and
2980 @code{__builtin_return_address (count)} for @code{count} not equal to zero.
2983 @defmac DYNAMIC_CHAIN_ADDRESS (@var{frameaddr})
2984 A C expression whose value is RTL representing the address in a stack
2985 frame where the pointer to the caller's frame is stored. Assume that
2986 @var{frameaddr} is an RTL expression for the address of the stack frame
2989 If you don't define this macro, the default is to return the value
2990 of @var{frameaddr}---that is, the stack frame address is also the
2991 address of the stack word that points to the previous frame.
2994 @defmac SETUP_FRAME_ADDRESSES
2995 If defined, a C expression that produces the machine-specific code to
2996 setup the stack so that arbitrary frames can be accessed. For example,
2997 on the SPARC, we must flush all of the register windows to the stack
2998 before we can access arbitrary stack frames. You will seldom need to
3002 @deftypefn {Target Hook} rtx TARGET_BUILTIN_SETJMP_FRAME_VALUE (void)
3003 This target hook should return an rtx that is used to store
3004 the address of the current frame into the built in @code{setjmp} buffer.
3005 The default value, @code{virtual_stack_vars_rtx}, is correct for most
3006 machines. One reason you may need to define this target hook is if
3007 @code{hard_frame_pointer_rtx} is the appropriate value on your machine.
3010 @defmac FRAME_ADDR_RTX (@var{frameaddr})
3011 A C expression whose value is RTL representing the value of the frame
3012 address for the current frame. @var{frameaddr} is the frame pointer
3013 of the current frame. This is used for __builtin_frame_address.
3014 You need only define this macro if the frame address is not the same
3015 as the frame pointer. Most machines do not need to define it.
3018 @defmac RETURN_ADDR_RTX (@var{count}, @var{frameaddr})
3019 A C expression whose value is RTL representing the value of the return
3020 address for the frame @var{count} steps up from the current frame, after
3021 the prologue. @var{frameaddr} is the frame pointer of the @var{count}
3022 frame, or the frame pointer of the @var{count} @minus{} 1 frame if
3023 @code{RETURN_ADDR_IN_PREVIOUS_FRAME} is defined.
3025 The value of the expression must always be the correct address when
3026 @var{count} is zero, but may be @code{NULL_RTX} if there is no way to
3027 determine the return address of other frames.
3030 @defmac RETURN_ADDR_IN_PREVIOUS_FRAME
3031 Define this if the return address of a particular stack frame is accessed
3032 from the frame pointer of the previous stack frame.
3035 @defmac INCOMING_RETURN_ADDR_RTX
3036 A C expression whose value is RTL representing the location of the
3037 incoming return address at the beginning of any function, before the
3038 prologue. This RTL is either a @code{REG}, indicating that the return
3039 value is saved in @samp{REG}, or a @code{MEM} representing a location in
3042 You only need to define this macro if you want to support call frame
3043 debugging information like that provided by DWARF 2.
3045 If this RTL is a @code{REG}, you should also define
3046 @code{DWARF_FRAME_RETURN_COLUMN} to @code{DWARF_FRAME_REGNUM (REGNO)}.
3049 @defmac DWARF_ALT_FRAME_RETURN_COLUMN
3050 A C expression whose value is an integer giving a DWARF 2 column
3051 number that may be used as an alternative return column. The column
3052 must not correspond to any gcc hard register (that is, it must not
3053 be in the range of @code{DWARF_FRAME_REGNUM}).
3055 This macro can be useful if @code{DWARF_FRAME_RETURN_COLUMN} is set to a
3056 general register, but an alternative column needs to be used for signal
3057 frames. Some targets have also used different frame return columns
3061 @defmac DWARF_ZERO_REG
3062 A C expression whose value is an integer giving a DWARF 2 register
3063 number that is considered to always have the value zero. This should
3064 only be defined if the target has an architected zero register, and
3065 someone decided it was a good idea to use that register number to
3066 terminate the stack backtrace. New ports should avoid this.
3069 @deftypefn {Target Hook} void TARGET_DWARF_HANDLE_FRAME_UNSPEC (const char *@var{label}, rtx @var{pattern}, int @var{index})
3070 This target hook allows the backend to emit frame-related insns that
3071 contain UNSPECs or UNSPEC_VOLATILEs. The DWARF 2 call frame debugging
3072 info engine will invoke it on insns of the form
3074 (set (reg) (unspec [@dots{}] UNSPEC_INDEX))
3078 (set (reg) (unspec_volatile [@dots{}] UNSPECV_INDEX)).
3080 to let the backend emit the call frame instructions. @var{label} is
3081 the CFI label attached to the insn, @var{pattern} is the pattern of
3082 the insn and @var{index} is @code{UNSPEC_INDEX} or @code{UNSPECV_INDEX}.
3085 @defmac INCOMING_FRAME_SP_OFFSET
3086 A C expression whose value is an integer giving the offset, in bytes,
3087 from the value of the stack pointer register to the top of the stack
3088 frame at the beginning of any function, before the prologue. The top of
3089 the frame is defined to be the value of the stack pointer in the
3090 previous frame, just before the call instruction.
3092 You only need to define this macro if you want to support call frame
3093 debugging information like that provided by DWARF 2.
3096 @defmac ARG_POINTER_CFA_OFFSET (@var{fundecl})
3097 A C expression whose value is an integer giving the offset, in bytes,
3098 from the argument pointer to the canonical frame address (cfa). The
3099 final value should coincide with that calculated by
3100 @code{INCOMING_FRAME_SP_OFFSET}. Which is unfortunately not usable
3101 during virtual register instantiation.
3103 The default value for this macro is
3104 @code{FIRST_PARM_OFFSET (fundecl) + crtl->args.pretend_args_size},
3105 which is correct for most machines; in general, the arguments are found
3106 immediately before the stack frame. Note that this is not the case on
3107 some targets that save registers into the caller's frame, such as SPARC
3108 and rs6000, and so such targets need to define this macro.
3110 You only need to define this macro if the default is incorrect, and you
3111 want to support call frame debugging information like that provided by
3115 @defmac FRAME_POINTER_CFA_OFFSET (@var{fundecl})
3116 If defined, a C expression whose value is an integer giving the offset
3117 in bytes from the frame pointer to the canonical frame address (cfa).
3118 The final value should coincide with that calculated by
3119 @code{INCOMING_FRAME_SP_OFFSET}.
3121 Normally the CFA is calculated as an offset from the argument pointer,
3122 via @code{ARG_POINTER_CFA_OFFSET}, but if the argument pointer is
3123 variable due to the ABI, this may not be possible. If this macro is
3124 defined, it implies that the virtual register instantiation should be
3125 based on the frame pointer instead of the argument pointer. Only one
3126 of @code{FRAME_POINTER_CFA_OFFSET} and @code{ARG_POINTER_CFA_OFFSET}
3130 @defmac CFA_FRAME_BASE_OFFSET (@var{fundecl})
3131 If defined, a C expression whose value is an integer giving the offset
3132 in bytes from the canonical frame address (cfa) to the frame base used
3133 in DWARF 2 debug information. The default is zero. A different value
3134 may reduce the size of debug information on some ports.
3137 @node Exception Handling
3138 @subsection Exception Handling Support
3139 @cindex exception handling
3141 @defmac EH_RETURN_DATA_REGNO (@var{N})
3142 A C expression whose value is the @var{N}th register number used for
3143 data by exception handlers, or @code{INVALID_REGNUM} if fewer than
3144 @var{N} registers are usable.
3146 The exception handling library routines communicate with the exception
3147 handlers via a set of agreed upon registers. Ideally these registers
3148 should be call-clobbered; it is possible to use call-saved registers,
3149 but may negatively impact code size. The target must support at least
3150 2 data registers, but should define 4 if there are enough free registers.
3152 You must define this macro if you want to support call frame exception
3153 handling like that provided by DWARF 2.
3156 @defmac EH_RETURN_STACKADJ_RTX
3157 A C expression whose value is RTL representing a location in which
3158 to store a stack adjustment to be applied before function return.
3159 This is used to unwind the stack to an exception handler's call frame.
3160 It will be assigned zero on code paths that return normally.
3162 Typically this is a call-clobbered hard register that is otherwise
3163 untouched by the epilogue, but could also be a stack slot.
3165 Do not define this macro if the stack pointer is saved and restored
3166 by the regular prolog and epilog code in the call frame itself; in
3167 this case, the exception handling library routines will update the
3168 stack location to be restored in place. Otherwise, you must define
3169 this macro if you want to support call frame exception handling like
3170 that provided by DWARF 2.
3173 @defmac EH_RETURN_HANDLER_RTX
3174 A C expression whose value is RTL representing a location in which
3175 to store the address of an exception handler to which we should
3176 return. It will not be assigned on code paths that return normally.
3178 Typically this is the location in the call frame at which the normal
3179 return address is stored. For targets that return by popping an
3180 address off the stack, this might be a memory address just below
3181 the @emph{target} call frame rather than inside the current call
3182 frame. If defined, @code{EH_RETURN_STACKADJ_RTX} will have already
3183 been assigned, so it may be used to calculate the location of the
3186 Some targets have more complex requirements than storing to an
3187 address calculable during initial code generation. In that case
3188 the @code{eh_return} instruction pattern should be used instead.
3190 If you want to support call frame exception handling, you must
3191 define either this macro or the @code{eh_return} instruction pattern.
3194 @defmac RETURN_ADDR_OFFSET
3195 If defined, an integer-valued C expression for which rtl will be generated
3196 to add it to the exception handler address before it is searched in the
3197 exception handling tables, and to subtract it again from the address before
3198 using it to return to the exception handler.
3201 @defmac ASM_PREFERRED_EH_DATA_FORMAT (@var{code}, @var{global})
3202 This macro chooses the encoding of pointers embedded in the exception
3203 handling sections. If at all possible, this should be defined such
3204 that the exception handling section will not require dynamic relocations,
3205 and so may be read-only.
3207 @var{code} is 0 for data, 1 for code labels, 2 for function pointers.
3208 @var{global} is true if the symbol may be affected by dynamic relocations.
3209 The macro should return a combination of the @code{DW_EH_PE_*} defines
3210 as found in @file{dwarf2.h}.
3212 If this macro is not defined, pointers will not be encoded but
3213 represented directly.
3216 @defmac ASM_MAYBE_OUTPUT_ENCODED_ADDR_RTX (@var{file}, @var{encoding}, @var{size}, @var{addr}, @var{done})
3217 This macro allows the target to emit whatever special magic is required
3218 to represent the encoding chosen by @code{ASM_PREFERRED_EH_DATA_FORMAT}.
3219 Generic code takes care of pc-relative and indirect encodings; this must
3220 be defined if the target uses text-relative or data-relative encodings.
3222 This is a C statement that branches to @var{done} if the format was
3223 handled. @var{encoding} is the format chosen, @var{size} is the number
3224 of bytes that the format occupies, @var{addr} is the @code{SYMBOL_REF}
3228 @defmac MD_FALLBACK_FRAME_STATE_FOR (@var{context}, @var{fs})
3229 This macro allows the target to add CPU and operating system specific
3230 code to the call-frame unwinder for use when there is no unwind data
3231 available. The most common reason to implement this macro is to unwind
3232 through signal frames.
3234 This macro is called from @code{uw_frame_state_for} in
3235 @file{unwind-dw2.c}, @file{unwind-dw2-xtensa.c} and
3236 @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context};
3237 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{context->ra}
3238 for the address of the code being executed and @code{context->cfa} for
3239 the stack pointer value. If the frame can be decoded, the register
3240 save addresses should be updated in @var{fs} and the macro should
3241 evaluate to @code{_URC_NO_REASON}. If the frame cannot be decoded,
3242 the macro should evaluate to @code{_URC_END_OF_STACK}.
3244 For proper signal handling in Java this macro is accompanied by
3245 @code{MAKE_THROW_FRAME}, defined in @file{libjava/include/*-signal.h} headers.
3248 @defmac MD_HANDLE_UNWABI (@var{context}, @var{fs})
3249 This macro allows the target to add operating system specific code to the
3250 call-frame unwinder to handle the IA-64 @code{.unwabi} unwinding directive,
3251 usually used for signal or interrupt frames.
3253 This macro is called from @code{uw_update_context} in libgcc's
3254 @file{unwind-ia64.c}. @var{context} is an @code{_Unwind_Context};
3255 @var{fs} is an @code{_Unwind_FrameState}. Examine @code{fs->unwabi}
3256 for the abi and context in the @code{.unwabi} directive. If the
3257 @code{.unwabi} directive can be handled, the register save addresses should
3258 be updated in @var{fs}.
3261 @defmac TARGET_USES_WEAK_UNWIND_INFO
3262 A C expression that evaluates to true if the target requires unwind
3263 info to be given comdat linkage. Define it to be @code{1} if comdat
3264 linkage is necessary. The default is @code{0}.
3267 @node Stack Checking
3268 @subsection Specifying How Stack Checking is Done
3270 GCC will check that stack references are within the boundaries of the
3271 stack, if the option @option{-fstack-check} is specified, in one of
3276 If the value of the @code{STACK_CHECK_BUILTIN} macro is nonzero, GCC
3277 will assume that you have arranged for full stack checking to be done
3278 at appropriate places in the configuration files. GCC will not do
3279 other special processing.
3282 If @code{STACK_CHECK_BUILTIN} is zero and the value of the
3283 @code{STACK_CHECK_STATIC_BUILTIN} macro is nonzero, GCC will assume
3284 that you have arranged for static stack checking (checking of the
3285 static stack frame of functions) to be done at appropriate places
3286 in the configuration files. GCC will only emit code to do dynamic
3287 stack checking (checking on dynamic stack allocations) using the third
3291 If neither of the above are true, GCC will generate code to periodically
3292 ``probe'' the stack pointer using the values of the macros defined below.
3295 If neither STACK_CHECK_BUILTIN nor STACK_CHECK_STATIC_BUILTIN is defined,
3296 GCC will change its allocation strategy for large objects if the option
3297 @option{-fstack-check} is specified: they will always be allocated
3298 dynamically if their size exceeds @code{STACK_CHECK_MAX_VAR_SIZE} bytes.
3300 @defmac STACK_CHECK_BUILTIN
3301 A nonzero value if stack checking is done by the configuration files in a
3302 machine-dependent manner. You should define this macro if stack checking
3303 is required by the ABI of your machine or if you would like to do stack
3304 checking in some more efficient way than the generic approach. The default
3305 value of this macro is zero.
3308 @defmac STACK_CHECK_STATIC_BUILTIN
3309 A nonzero value if static stack checking is done by the configuration files
3310 in a machine-dependent manner. You should define this macro if you would
3311 like to do static stack checking in some more efficient way than the generic
3312 approach. The default value of this macro is zero.
3315 @defmac STACK_CHECK_PROBE_INTERVAL_EXP
3316 An integer specifying the interval at which GCC must generate stack probe
3317 instructions, defined as 2 raised to this integer. You will normally
3318 define this macro so that the interval be no larger than the size of
3319 the ``guard pages'' at the end of a stack area. The default value
3320 of 12 (4096-byte interval) is suitable for most systems.
3323 @defmac STACK_CHECK_MOVING_SP
3324 An integer which is nonzero if GCC should move the stack pointer page by page
3325 when doing probes. This can be necessary on systems where the stack pointer
3326 contains the bottom address of the memory area accessible to the executing
3327 thread at any point in time. In this situation an alternate signal stack
3328 is required in order to be able to recover from a stack overflow. The
3329 default value of this macro is zero.
3332 @defmac STACK_CHECK_PROTECT
3333 The number of bytes of stack needed to recover from a stack overflow, for
3334 languages where such a recovery is supported. The default value of 75 words
3335 with the @code{setjmp}/@code{longjmp}-based exception handling mechanism and
3336 8192 bytes with other exception handling mechanisms should be adequate for
3340 The following macros are relevant only if neither STACK_CHECK_BUILTIN
3341 nor STACK_CHECK_STATIC_BUILTIN is defined; you can omit them altogether
3342 in the opposite case.
3344 @defmac STACK_CHECK_MAX_FRAME_SIZE
3345 The maximum size of a stack frame, in bytes. GCC will generate probe
3346 instructions in non-leaf functions to ensure at least this many bytes of
3347 stack are available. If a stack frame is larger than this size, stack
3348 checking will not be reliable and GCC will issue a warning. The
3349 default is chosen so that GCC only generates one instruction on most
3350 systems. You should normally not change the default value of this macro.
3353 @defmac STACK_CHECK_FIXED_FRAME_SIZE
3354 GCC uses this value to generate the above warning message. It
3355 represents the amount of fixed frame used by a function, not including
3356 space for any callee-saved registers, temporaries and user variables.
3357 You need only specify an upper bound for this amount and will normally
3358 use the default of four words.
3361 @defmac STACK_CHECK_MAX_VAR_SIZE
3362 The maximum size, in bytes, of an object that GCC will place in the
3363 fixed area of the stack frame when the user specifies
3364 @option{-fstack-check}.
3365 GCC computed the default from the values of the above macros and you will
3366 normally not need to override that default.
3370 @node Frame Registers
3371 @subsection Registers That Address the Stack Frame
3373 @c prevent bad page break with this line
3374 This discusses registers that address the stack frame.
3376 @defmac STACK_POINTER_REGNUM
3377 The register number of the stack pointer register, which must also be a
3378 fixed register according to @code{FIXED_REGISTERS}. On most machines,
3379 the hardware determines which register this is.
3382 @defmac FRAME_POINTER_REGNUM
3383 The register number of the frame pointer register, which is used to
3384 access automatic variables in the stack frame. On some machines, the
3385 hardware determines which register this is. On other machines, you can
3386 choose any register you wish for this purpose.
3389 @defmac HARD_FRAME_POINTER_REGNUM
3390 On some machines the offset between the frame pointer and starting
3391 offset of the automatic variables is not known until after register
3392 allocation has been done (for example, because the saved registers are
3393 between these two locations). On those machines, define
3394 @code{FRAME_POINTER_REGNUM} the number of a special, fixed register to
3395 be used internally until the offset is known, and define
3396 @code{HARD_FRAME_POINTER_REGNUM} to be the actual hard register number
3397 used for the frame pointer.
3399 You should define this macro only in the very rare circumstances when it
3400 is not possible to calculate the offset between the frame pointer and
3401 the automatic variables until after register allocation has been
3402 completed. When this macro is defined, you must also indicate in your
3403 definition of @code{ELIMINABLE_REGS} how to eliminate
3404 @code{FRAME_POINTER_REGNUM} into either @code{HARD_FRAME_POINTER_REGNUM}
3405 or @code{STACK_POINTER_REGNUM}.
3407 Do not define this macro if it would be the same as
3408 @code{FRAME_POINTER_REGNUM}.
3411 @defmac ARG_POINTER_REGNUM
3412 The register number of the arg pointer register, which is used to access
3413 the function's argument list. On some machines, this is the same as the
3414 frame pointer register. On some machines, the hardware determines which
3415 register this is. On other machines, you can choose any register you
3416 wish for this purpose. If this is not the same register as the frame
3417 pointer register, then you must mark it as a fixed register according to
3418 @code{FIXED_REGISTERS}, or arrange to be able to eliminate it
3419 (@pxref{Elimination}).
3422 @defmac HARD_FRAME_POINTER_IS_FRAME_POINTER
3423 Define this to a preprocessor constant that is nonzero if
3424 @code{hard_frame_pointer_rtx} and @code{frame_pointer_rtx} should be
3425 the same. The default definition is @samp{(HARD_FRAME_POINTER_REGNUM
3426 == FRAME_POINTER_REGNUM)}; you only need to define this macro if that
3427 definition is not suitable for use in preprocessor conditionals.
3430 @defmac HARD_FRAME_POINTER_IS_ARG_POINTER
3431 Define this to a preprocessor constant that is nonzero if
3432 @code{hard_frame_pointer_rtx} and @code{arg_pointer_rtx} should be the
3433 same. The default definition is @samp{(HARD_FRAME_POINTER_REGNUM ==
3434 ARG_POINTER_REGNUM)}; you only need to define this macro if that
3435 definition is not suitable for use in preprocessor conditionals.
3438 @defmac RETURN_ADDRESS_POINTER_REGNUM
3439 The register number of the return address pointer register, which is used to
3440 access the current function's return address from the stack. On some
3441 machines, the return address is not at a fixed offset from the frame
3442 pointer or stack pointer or argument pointer. This register can be defined
3443 to point to the return address on the stack, and then be converted by
3444 @code{ELIMINABLE_REGS} into either the frame pointer or stack pointer.
3446 Do not define this macro unless there is no other way to get the return
3447 address from the stack.
3450 @defmac STATIC_CHAIN_REGNUM
3451 @defmacx STATIC_CHAIN_INCOMING_REGNUM
3452 Register numbers used for passing a function's static chain pointer. If
3453 register windows are used, the register number as seen by the called
3454 function is @code{STATIC_CHAIN_INCOMING_REGNUM}, while the register
3455 number as seen by the calling function is @code{STATIC_CHAIN_REGNUM}. If
3456 these registers are the same, @code{STATIC_CHAIN_INCOMING_REGNUM} need
3459 The static chain register need not be a fixed register.
3461 If the static chain is passed in memory, these macros should not be
3462 defined; instead, the @code{TARGET_STATIC_CHAIN} hook should be used.
3465 @deftypefn {Target Hook} rtx TARGET_STATIC_CHAIN (const_tree @var{fndecl}, bool @var{incoming_p})
3466 This hook replaces the use of @code{STATIC_CHAIN_REGNUM} et al for
3467 targets that may use different static chain locations for different
3468 nested functions. This may be required if the target has function
3469 attributes that affect the calling conventions of the function and
3470 those calling conventions use different static chain locations.
3472 The default version of this hook uses @code{STATIC_CHAIN_REGNUM} et al.
3474 If the static chain is passed in memory, this hook should be used to
3475 provide rtx giving @code{mem} expressions that denote where they are stored.
3476 Often the @code{mem} expression as seen by the caller will be at an offset
3477 from the stack pointer and the @code{mem} expression as seen by the callee
3478 will be at an offset from the frame pointer.
3479 @findex stack_pointer_rtx
3480 @findex frame_pointer_rtx
3481 @findex arg_pointer_rtx
3482 The variables @code{stack_pointer_rtx}, @code{frame_pointer_rtx}, and
3483 @code{arg_pointer_rtx} will have been initialized and should be used
3484 to refer to those items.
3487 @defmac DWARF_FRAME_REGISTERS
3488 This macro specifies the maximum number of hard registers that can be
3489 saved in a call frame. This is used to size data structures used in
3490 DWARF2 exception handling.
3492 Prior to GCC 3.0, this macro was needed in order to establish a stable
3493 exception handling ABI in the face of adding new hard registers for ISA
3494 extensions. In GCC 3.0 and later, the EH ABI is insulated from changes
3495 in the number of hard registers. Nevertheless, this macro can still be
3496 used to reduce the runtime memory requirements of the exception handling
3497 routines, which can be substantial if the ISA contains a lot of
3498 registers that are not call-saved.
3500 If this macro is not defined, it defaults to
3501 @code{FIRST_PSEUDO_REGISTER}.
3504 @defmac PRE_GCC3_DWARF_FRAME_REGISTERS
3506 This macro is similar to @code{DWARF_FRAME_REGISTERS}, but is provided
3507 for backward compatibility in pre GCC 3.0 compiled code.
3509 If this macro is not defined, it defaults to
3510 @code{DWARF_FRAME_REGISTERS}.
3513 @defmac DWARF_REG_TO_UNWIND_COLUMN (@var{regno})
3515 Define this macro if the target's representation for dwarf registers
3516 is different than the internal representation for unwind column.
3517 Given a dwarf register, this macro should return the internal unwind
3518 column number to use instead.
3520 See the PowerPC's SPE target for an example.
3523 @defmac DWARF_FRAME_REGNUM (@var{regno})
3525 Define this macro if the target's representation for dwarf registers
3526 used in .eh_frame or .debug_frame is different from that used in other
3527 debug info sections. Given a GCC hard register number, this macro
3528 should return the .eh_frame register number. The default is
3529 @code{DBX_REGISTER_NUMBER (@var{regno})}.
3533 @defmac DWARF2_FRAME_REG_OUT (@var{regno}, @var{for_eh})
3535 Define this macro to map register numbers held in the call frame info
3536 that GCC has collected using @code{DWARF_FRAME_REGNUM} to those that
3537 should be output in .debug_frame (@code{@var{for_eh}} is zero) and
3538 .eh_frame (@code{@var{for_eh}} is nonzero). The default is to
3539 return @code{@var{regno}}.
3543 @defmac REG_VALUE_IN_UNWIND_CONTEXT
3545 Define this macro if the target stores register values as
3546 @code{_Unwind_Word} type in unwind context. It should be defined if
3547 target register size is larger than the size of @code{void *}. The
3548 default is to store register values as @code{void *} type.
3552 @defmac ASSUME_EXTENDED_UNWIND_CONTEXT
3554 Define this macro to be 1 if the target always uses extended unwind
3555 context with version, args_size and by_value fields. If it is undefined,
3556 it will be defined to 1 when @code{REG_VALUE_IN_UNWIND_CONTEXT} is
3557 defined and 0 otherwise.
3562 @subsection Eliminating Frame Pointer and Arg Pointer
3564 @c prevent bad page break with this line
3565 This is about eliminating the frame pointer and arg pointer.
3567 @deftypefn {Target Hook} bool TARGET_FRAME_POINTER_REQUIRED (void)
3568 This target hook should return @code{true} if a function must have and use
3569 a frame pointer. This target hook is called in the reload pass. If its return
3570 value is @code{true} the function will have a frame pointer.
3572 This target hook can in principle examine the current function and decide
3573 according to the facts, but on most machines the constant @code{false} or the
3574 constant @code{true} suffices. Use @code{false} when the machine allows code
3575 to be generated with no frame pointer, and doing so saves some time or space.
3576 Use @code{true} when there is no possible advantage to avoiding a frame
3579 In certain cases, the compiler does not know how to produce valid code
3580 without a frame pointer. The compiler recognizes those cases and
3581 automatically gives the function a frame pointer regardless of what
3582 @code{TARGET_FRAME_POINTER_REQUIRED} returns. You don't need to worry about
3585 In a function that does not require a frame pointer, the frame pointer
3586 register can be allocated for ordinary usage, unless you mark it as a
3587 fixed register. See @code{FIXED_REGISTERS} for more information.
3589 Default return value is @code{false}.
3592 @findex get_frame_size
3593 @defmac INITIAL_FRAME_POINTER_OFFSET (@var{depth-var})
3594 A C statement to store in the variable @var{depth-var} the difference
3595 between the frame pointer and the stack pointer values immediately after
3596 the function prologue. The value would be computed from information
3597 such as the result of @code{get_frame_size ()} and the tables of
3598 registers @code{regs_ever_live} and @code{call_used_regs}.
3600 If @code{ELIMINABLE_REGS} is defined, this macro will be not be used and
3601 need not be defined. Otherwise, it must be defined even if
3602 @code{TARGET_FRAME_POINTER_REQUIRED} always returns true; in that
3603 case, you may set @var{depth-var} to anything.
3606 @defmac ELIMINABLE_REGS
3607 If defined, this macro specifies a table of register pairs used to
3608 eliminate unneeded registers that point into the stack frame. If it is not
3609 defined, the only elimination attempted by the compiler is to replace
3610 references to the frame pointer with references to the stack pointer.
3612 The definition of this macro is a list of structure initializations, each
3613 of which specifies an original and replacement register.
3615 On some machines, the position of the argument pointer is not known until
3616 the compilation is completed. In such a case, a separate hard register
3617 must be used for the argument pointer. This register can be eliminated by
3618 replacing it with either the frame pointer or the argument pointer,
3619 depending on whether or not the frame pointer has been eliminated.
3621 In this case, you might specify:
3623 #define ELIMINABLE_REGS \
3624 @{@{ARG_POINTER_REGNUM, STACK_POINTER_REGNUM@}, \
3625 @{ARG_POINTER_REGNUM, FRAME_POINTER_REGNUM@}, \
3626 @{FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM@}@}
3629 Note that the elimination of the argument pointer with the stack pointer is
3630 specified first since that is the preferred elimination.
3633 @deftypefn {Target Hook} bool TARGET_CAN_ELIMINATE (const int @var{from_reg}, const int @var{to_reg})
3634 This target hook should returns @code{true} if the compiler is allowed to
3635 try to replace register number @var{from_reg} with register number
3636 @var{to_reg}. This target hook need only be defined if @code{ELIMINABLE_REGS}
3637 is defined, and will usually be @code{true}, since most of the cases
3638 preventing register elimination are things that the compiler already
3641 Default return value is @code{true}.
3644 @defmac INITIAL_ELIMINATION_OFFSET (@var{from-reg}, @var{to-reg}, @var{offset-var})
3645 This macro is similar to @code{INITIAL_FRAME_POINTER_OFFSET}. It
3646 specifies the initial difference between the specified pair of
3647 registers. This macro must be defined if @code{ELIMINABLE_REGS} is
3651 @node Stack Arguments
3652 @subsection Passing Function Arguments on the Stack
3653 @cindex arguments on stack
3654 @cindex stack arguments
3656 The macros in this section control how arguments are passed
3657 on the stack. See the following section for other macros that
3658 control passing certain arguments in registers.
3660 @deftypefn {Target Hook} bool TARGET_PROMOTE_PROTOTYPES (const_tree @var{fntype})
3661 This target hook returns @code{true} if an argument declared in a
3662 prototype as an integral type smaller than @code{int} should actually be
3663 passed as an @code{int}. In addition to avoiding errors in certain
3664 cases of mismatch, it also makes for better code on certain machines.
3665 The default is to not promote prototypes.
3669 A C expression. If nonzero, push insns will be used to pass
3671 If the target machine does not have a push instruction, set it to zero.
3672 That directs GCC to use an alternate strategy: to
3673 allocate the entire argument block and then store the arguments into
3674 it. When @code{PUSH_ARGS} is nonzero, @code{PUSH_ROUNDING} must be defined too.
3677 @defmac PUSH_ARGS_REVERSED
3678 A C expression. If nonzero, function arguments will be evaluated from
3679 last to first, rather than from first to last. If this macro is not
3680 defined, it defaults to @code{PUSH_ARGS} on targets where the stack
3681 and args grow in opposite directions, and 0 otherwise.
3684 @defmac PUSH_ROUNDING (@var{npushed})
3685 A C expression that is the number of bytes actually pushed onto the
3686 stack when an instruction attempts to push @var{npushed} bytes.
3688 On some machines, the definition
3691 #define PUSH_ROUNDING(BYTES) (BYTES)
3695 will suffice. But on other machines, instructions that appear
3696 to push one byte actually push two bytes in an attempt to maintain
3697 alignment. Then the definition should be
3700 #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1)
3703 If the value of this macro has a type, it should be an unsigned type.
3706 @findex outgoing_args_size
3707 @findex crtl->outgoing_args_size
3708 @defmac ACCUMULATE_OUTGOING_ARGS
3709 A C expression. If nonzero, the maximum amount of space required for outgoing arguments
3710 will be computed and placed into
3711 @code{crtl->outgoing_args_size}. No space will be pushed
3712 onto the stack for each call; instead, the function prologue should
3713 increase the stack frame size by this amount.
3715 Setting both @code{PUSH_ARGS} and @code{ACCUMULATE_OUTGOING_ARGS}
3719 @defmac REG_PARM_STACK_SPACE (@var{fndecl})
3720 Define this macro if functions should assume that stack space has been
3721 allocated for arguments even when their values are passed in
3724 The value of this macro is the size, in bytes, of the area reserved for
3725 arguments passed in registers for the function represented by @var{fndecl},
3726 which can be zero if GCC is calling a library function.
3727 The argument @var{fndecl} can be the FUNCTION_DECL, or the type itself
3730 This space can be allocated by the caller, or be a part of the
3731 machine-dependent stack frame: @code{OUTGOING_REG_PARM_STACK_SPACE} says
3734 @c above is overfull. not sure what to do. --mew 5feb93 did
3735 @c something, not sure if it looks good. --mew 10feb93
3737 @defmac INCOMING_REG_PARM_STACK_SPACE (@var{fndecl})
3738 Like @code{REG_PARM_STACK_SPACE}, but for incoming register arguments.
3739 Define this macro if space guaranteed when compiling a function body
3740 is different to space required when making a call, a situation that
3741 can arise with K&R style function definitions.
3744 @defmac OUTGOING_REG_PARM_STACK_SPACE (@var{fntype})
3745 Define this to a nonzero value if it is the responsibility of the
3746 caller to allocate the area reserved for arguments passed in registers
3747 when calling a function of @var{fntype}. @var{fntype} may be NULL
3748 if the function called is a library function.
3750 If @code{ACCUMULATE_OUTGOING_ARGS} is defined, this macro controls
3751 whether the space for these arguments counts in the value of
3752 @code{crtl->outgoing_args_size}.
3755 @defmac STACK_PARMS_IN_REG_PARM_AREA
3756 Define this macro if @code{REG_PARM_STACK_SPACE} is defined, but the
3757 stack parameters don't skip the area specified by it.
3758 @c i changed this, makes more sens and it should have taken care of the
3759 @c overfull.. not as specific, tho. --mew 5feb93
3761 Normally, when a parameter is not passed in registers, it is placed on the
3762 stack beyond the @code{REG_PARM_STACK_SPACE} area. Defining this macro
3763 suppresses this behavior and causes the parameter to be passed on the
3764 stack in its natural location.
3767 @deftypefn {Target Hook} int TARGET_RETURN_POPS_ARGS (tree @var{fundecl}, tree @var{funtype}, int @var{size})
3768 This target hook returns the number of bytes of its own arguments that
3769 a function pops on returning, or 0 if the function pops no arguments
3770 and the caller must therefore pop them all after the function returns.
3772 @var{fundecl} is a C variable whose value is a tree node that describes
3773 the function in question. Normally it is a node of type
3774 @code{FUNCTION_DECL} that describes the declaration of the function.
3775 From this you can obtain the @code{DECL_ATTRIBUTES} of the function.
3777 @var{funtype} is a C variable whose value is a tree node that
3778 describes the function in question. Normally it is a node of type
3779 @code{FUNCTION_TYPE} that describes the data type of the function.
3780 From this it is possible to obtain the data types of the value and
3781 arguments (if known).
3783 When a call to a library function is being considered, @var{fundecl}
3784 will contain an identifier node for the library function. Thus, if
3785 you need to distinguish among various library functions, you can do so
3786 by their names. Note that ``library function'' in this context means
3787 a function used to perform arithmetic, whose name is known specially
3788 in the compiler and was not mentioned in the C code being compiled.
3790 @var{size} is the number of bytes of arguments passed on the
3791 stack. If a variable number of bytes is passed, it is zero, and
3792 argument popping will always be the responsibility of the calling function.
3794 On the VAX, all functions always pop their arguments, so the definition
3795 of this macro is @var{size}. On the 68000, using the standard
3796 calling convention, no functions pop their arguments, so the value of
3797 the macro is always 0 in this case. But an alternative calling
3798 convention is available in which functions that take a fixed number of
3799 arguments pop them but other functions (such as @code{printf}) pop
3800 nothing (the caller pops all). When this convention is in use,
3801 @var{funtype} is examined to determine whether a function takes a fixed
3802 number of arguments.
3805 @defmac CALL_POPS_ARGS (@var{cum})
3806 A C expression that should indicate the number of bytes a call sequence
3807 pops off the stack. It is added to the value of @code{RETURN_POPS_ARGS}
3808 when compiling a function call.
3810 @var{cum} is the variable in which all arguments to the called function
3811 have been accumulated.
3813 On certain architectures, such as the SH5, a call trampoline is used
3814 that pops certain registers off the stack, depending on the arguments
3815 that have been passed to the function. Since this is a property of the
3816 call site, not of the called function, @code{RETURN_POPS_ARGS} is not
3820 @node Register Arguments
3821 @subsection Passing Arguments in Registers
3822 @cindex arguments in registers
3823 @cindex registers arguments
3825 This section describes the macros which let you control how various
3826 types of arguments are passed in registers or how they are arranged in
3829 @deftypefn {Target Hook} rtx TARGET_FUNCTION_ARG (cumulative_args_t @var{ca}, machine_mode @var{mode}, const_tree @var{type}, bool @var{named})
3830 Return an RTX indicating whether a function argument is passed in a
3831 register and if so, which register.
3833 The arguments are @var{ca}, which summarizes all the previous
3834 arguments; @var{mode}, the machine mode of the argument; @var{type},
3835 the data type of the argument as a tree node or 0 if that is not known
3836 (which happens for C support library functions); and @var{named},
3837 which is @code{true} for an ordinary argument and @code{false} for
3838 nameless arguments that correspond to @samp{@dots{}} in the called
3839 function's prototype. @var{type} can be an incomplete type if a
3840 syntax error has previously occurred.
3842 The return value is usually either a @code{reg} RTX for the hard
3843 register in which to pass the argument, or zero to pass the argument
3846 The return value can be a @code{const_int} which means argument is
3847 passed in a target specific slot with specified number. Target hooks
3848 should be used to store or load argument in such case. See
3849 @code{TARGET_STORE_BOUNDS_FOR_ARG} and @code{TARGET_LOAD_BOUNDS_FOR_ARG}
3850 for more information.
3852 The value of the expression can also be a @code{parallel} RTX@. This is
3853 used when an argument is passed in multiple locations. The mode of the
3854 @code{parallel} should be the mode of the entire argument. The
3855 @code{parallel} holds any number of @code{expr_list} pairs; each one
3856 describes where part of the argument is passed. In each
3857 @code{expr_list} the first operand must be a @code{reg} RTX for the hard
3858 register in which to pass this part of the argument, and the mode of the
3859 register RTX indicates how large this part of the argument is. The
3860 second operand of the @code{expr_list} is a @code{const_int} which gives
3861 the offset in bytes into the entire argument of where this part starts.
3862 As a special exception the first @code{expr_list} in the @code{parallel}
3863 RTX may have a first operand of zero. This indicates that the entire
3864 argument is also stored on the stack.
3866 The last time this hook is called, it is called with @code{MODE ==
3867 VOIDmode}, and its result is passed to the @code{call} or @code{call_value}
3868 pattern as operands 2 and 3 respectively.
3870 @cindex @file{stdarg.h} and register arguments
3871 The usual way to make the ISO library @file{stdarg.h} work on a
3872 machine where some arguments are usually passed in registers, is to
3873 cause nameless arguments to be passed on the stack instead. This is
3874 done by making @code{TARGET_FUNCTION_ARG} return 0 whenever
3875 @var{named} is @code{false}.
3877 @cindex @code{TARGET_MUST_PASS_IN_STACK}, and @code{TARGET_FUNCTION_ARG}
3878 @cindex @code{REG_PARM_STACK_SPACE}, and @code{TARGET_FUNCTION_ARG}
3879 You may use the hook @code{targetm.calls.must_pass_in_stack}
3880 in the definition of this macro to determine if this argument is of a
3881 type that must be passed in the stack. If @code{REG_PARM_STACK_SPACE}
3882 is not defined and @code{TARGET_FUNCTION_ARG} returns nonzero for such an
3883 argument, the compiler will abort. If @code{REG_PARM_STACK_SPACE} is
3884 defined, the argument will be computed in the stack and then loaded into
3888 @deftypefn {Target Hook} bool TARGET_MUST_PASS_IN_STACK (machine_mode @var{mode}, const_tree @var{type})
3889 This target hook should return @code{true} if we should not pass @var{type}
3890 solely in registers. The file @file{expr.h} defines a
3891 definition that is usually appropriate, refer to @file{expr.h} for additional
3895 @deftypefn {Target Hook} rtx TARGET_FUNCTION_INCOMING_ARG (cumulative_args_t @var{ca}, machine_mode @var{mode}, const_tree @var{type}, bool @var{named})
3896 Define this hook if the target machine has ``register windows'', so
3897 that the register in which a function sees an arguments is not
3898 necessarily the same as the one in which the caller passed the
3901 For such machines, @code{TARGET_FUNCTION_ARG} computes the register in
3902 which the caller passes the value, and
3903 @code{TARGET_FUNCTION_INCOMING_ARG} should be defined in a similar
3904 fashion to tell the function being called where the arguments will
3907 If @code{TARGET_FUNCTION_INCOMING_ARG} is not defined,
3908 @code{TARGET_FUNCTION_ARG} serves both purposes.
3911 @deftypefn {Target Hook} bool TARGET_USE_PSEUDO_PIC_REG (void)
3912 This hook should return 1 in case pseudo register should be created
3913 for pic_offset_table_rtx during function expand.
3916 @deftypefn {Target Hook} void TARGET_INIT_PIC_REG (void)
3917 Perform a target dependent initialization of pic_offset_table_rtx.
3918 This hook is called at the start of register allocation.
3921 @deftypefn {Target Hook} int TARGET_ARG_PARTIAL_BYTES (cumulative_args_t @var{cum}, machine_mode @var{mode}, tree @var{type}, bool @var{named})
3922 This target hook returns the number of bytes at the beginning of an
3923 argument that must be put in registers. The value must be zero for
3924 arguments that are passed entirely in registers or that are entirely
3925 pushed on the stack.
3927 On some machines, certain arguments must be passed partially in
3928 registers and partially in memory. On these machines, typically the
3929 first few words of arguments are passed in registers, and the rest
3930 on the stack. If a multi-word argument (a @code{double} or a
3931 structure) crosses that boundary, its first few words must be passed
3932 in registers and the rest must be pushed. This macro tells the
3933 compiler when this occurs, and how many bytes should go in registers.
3935 @code{TARGET_FUNCTION_ARG} for these arguments should return the first
3936 register to be used by the caller for this argument; likewise
3937 @code{TARGET_FUNCTION_INCOMING_ARG}, for the called function.
3940 @deftypefn {Target Hook} bool TARGET_PASS_BY_REFERENCE (cumulative_args_t @var{cum}, machine_mode @var{mode}, const_tree @var{type}, bool @var{named})
3941 This target hook should return @code{true} if an argument at the
3942 position indicated by @var{cum} should be passed by reference. This
3943 predicate is queried after target independent reasons for being
3944 passed by reference, such as @code{TREE_ADDRESSABLE (type)}.
3946 If the hook returns true, a copy of that argument is made in memory and a
3947 pointer to the argument is passed instead of the argument itself.
3948 The pointer is passed in whatever way is appropriate for passing a pointer
3952 @deftypefn {Target Hook} bool TARGET_CALLEE_COPIES (cumulative_args_t @var{cum}, machine_mode @var{mode}, const_tree @var{type}, bool @var{named})
3953 The function argument described by the parameters to this hook is
3954 known to be passed by reference. The hook should return true if the
3955 function argument should be copied by the callee instead of copied
3958 For any argument for which the hook returns true, if it can be
3959 determined that the argument is not modified, then a copy need
3962 The default version of this hook always returns false.
3965 @defmac CUMULATIVE_ARGS
3966 A C type for declaring a variable that is used as the first argument
3967 of @code{TARGET_FUNCTION_ARG} and other related values. For some
3968 target machines, the type @code{int} suffices and can hold the number
3969 of bytes of argument so far.
3971 There is no need to record in @code{CUMULATIVE_ARGS} anything about the
3972 arguments that have been passed on the stack. The compiler has other
3973 variables to keep track of that. For target machines on which all
3974 arguments are passed on the stack, there is no need to store anything in
3975 @code{CUMULATIVE_ARGS}; however, the data structure must exist and
3976 should not be empty, so use @code{int}.
3979 @defmac OVERRIDE_ABI_FORMAT (@var{fndecl})
3980 If defined, this macro is called before generating any code for a
3981 function, but after the @var{cfun} descriptor for the function has been
3982 created. The back end may use this macro to update @var{cfun} to
3983 reflect an ABI other than that which would normally be used by default.
3984 If the compiler is generating code for a compiler-generated function,
3985 @var{fndecl} may be @code{NULL}.
3988 @defmac INIT_CUMULATIVE_ARGS (@var{cum}, @var{fntype}, @var{libname}, @var{fndecl}, @var{n_named_args})
3989 A C statement (sans semicolon) for initializing the variable
3990 @var{cum} for the state at the beginning of the argument list. The
3991 variable has type @code{CUMULATIVE_ARGS}. The value of @var{fntype}
3992 is the tree node for the data type of the function which will receive
3993 the args, or 0 if the args are to a compiler support library function.
3994 For direct calls that are not libcalls, @var{fndecl} contain the
3995 declaration node of the function. @var{fndecl} is also set when
3996 @code{INIT_CUMULATIVE_ARGS} is used to find arguments for the function
3997 being compiled. @var{n_named_args} is set to the number of named
3998 arguments, including a structure return address if it is passed as a
3999 parameter, when making a call. When processing incoming arguments,
4000 @var{n_named_args} is set to @minus{}1.
4002 When processing a call to a compiler support library function,
4003 @var{libname} identifies which one. It is a @code{symbol_ref} rtx which
4004 contains the name of the function, as a string. @var{libname} is 0 when
4005 an ordinary C function call is being processed. Thus, each time this
4006 macro is called, either @var{libname} or @var{fntype} is nonzero, but
4007 never both of them at once.
4010 @defmac INIT_CUMULATIVE_LIBCALL_ARGS (@var{cum}, @var{mode}, @var{libname})
4011 Like @code{INIT_CUMULATIVE_ARGS} but only used for outgoing libcalls,
4012 it gets a @code{MODE} argument instead of @var{fntype}, that would be
4013 @code{NULL}. @var{indirect} would always be zero, too. If this macro
4014 is not defined, @code{INIT_CUMULATIVE_ARGS (cum, NULL_RTX, libname,
4015 0)} is used instead.
4018 @defmac INIT_CUMULATIVE_INCOMING_ARGS (@var{cum}, @var{fntype}, @var{libname})
4019 Like @code{INIT_CUMULATIVE_ARGS} but overrides it for the purposes of
4020 finding the arguments for the function being compiled. If this macro is
4021 undefined, @code{INIT_CUMULATIVE_ARGS} is used instead.
4023 The value passed for @var{libname} is always 0, since library routines
4024 with special calling conventions are never compiled with GCC@. The
4025 argument @var{libname} exists for symmetry with
4026 @code{INIT_CUMULATIVE_ARGS}.
4027 @c could use "this macro" in place of @code{INIT_CUMULATIVE_ARGS}, maybe.
4028 @c --mew 5feb93 i switched the order of the sentences. --mew 10feb93
4031 @deftypefn {Target Hook} void TARGET_FUNCTION_ARG_ADVANCE (cumulative_args_t @var{ca}, machine_mode @var{mode}, const_tree @var{type}, bool @var{named})
4032 This hook updates the summarizer variable pointed to by @var{ca} to
4033 advance past an argument in the argument list. The values @var{mode},
4034 @var{type} and @var{named} describe that argument. Once this is done,
4035 the variable @var{cum} is suitable for analyzing the @emph{following}
4036 argument with @code{TARGET_FUNCTION_ARG}, etc.
4038 This hook need not do anything if the argument in question was passed
4039 on the stack. The compiler knows how to track the amount of stack space
4040 used for arguments without any special help.
4043 @defmac FUNCTION_ARG_OFFSET (@var{mode}, @var{type})
4044 If defined, a C expression that is the number of bytes to add to the
4045 offset of the argument passed in memory. This is needed for the SPU,
4046 which passes @code{char} and @code{short} arguments in the preferred
4047 slot that is in the middle of the quad word instead of starting at the
4051 @defmac FUNCTION_ARG_PADDING (@var{mode}, @var{type})
4052 If defined, a C expression which determines whether, and in which direction,
4053 to pad out an argument with extra space. The value should be of type
4054 @code{enum direction}: either @code{upward} to pad above the argument,
4055 @code{downward} to pad below, or @code{none} to inhibit padding.
4057 The @emph{amount} of padding is not controlled by this macro, but by the
4058 target hook @code{TARGET_FUNCTION_ARG_ROUND_BOUNDARY}. It is
4059 always just enough to reach the next multiple of that boundary.
4061 This macro has a default definition which is right for most systems.
4062 For little-endian machines, the default is to pad upward. For
4063 big-endian machines, the default is to pad downward for an argument of
4064 constant size shorter than an @code{int}, and upward otherwise.
4067 @defmac PAD_VARARGS_DOWN
4068 If defined, a C expression which determines whether the default
4069 implementation of va_arg will attempt to pad down before reading the
4070 next argument, if that argument is smaller than its aligned space as
4071 controlled by @code{PARM_BOUNDARY}. If this macro is not defined, all such
4072 arguments are padded down if @code{BYTES_BIG_ENDIAN} is true.
4075 @defmac BLOCK_REG_PADDING (@var{mode}, @var{type}, @var{first})
4076 Specify padding for the last element of a block move between registers and
4077 memory. @var{first} is nonzero if this is the only element. Defining this
4078 macro allows better control of register function parameters on big-endian
4079 machines, without using @code{PARALLEL} rtl. In particular,
4080 @code{MUST_PASS_IN_STACK} need not test padding and mode of types in
4081 registers, as there is no longer a "wrong" part of a register; For example,
4082 a three byte aggregate may be passed in the high part of a register if so
4086 @deftypefn {Target Hook} {unsigned int} TARGET_FUNCTION_ARG_BOUNDARY (machine_mode @var{mode}, const_tree @var{type})
4087 This hook returns the alignment boundary, in bits, of an argument
4088 with the specified mode and type. The default hook returns
4089 @code{PARM_BOUNDARY} for all arguments.
4092 @deftypefn {Target Hook} {unsigned int} TARGET_FUNCTION_ARG_ROUND_BOUNDARY (machine_mode @var{mode}, const_tree @var{type})
4093 Normally, the size of an argument is rounded up to @code{PARM_BOUNDARY},
4094 which is the default value for this hook. You can define this hook to
4095 return a different value if an argument size must be rounded to a larger
4099 @defmac FUNCTION_ARG_REGNO_P (@var{regno})
4100 A C expression that is nonzero if @var{regno} is the number of a hard
4101 register in which function arguments are sometimes passed. This does
4102 @emph{not} include implicit arguments such as the static chain and
4103 the structure-value address. On many machines, no registers can be
4104 used for this purpose since all function arguments are pushed on the
4108 @deftypefn {Target Hook} bool TARGET_SPLIT_COMPLEX_ARG (const_tree @var{type})
4109 This hook should return true if parameter of type @var{type} are passed
4110 as two scalar parameters. By default, GCC will attempt to pack complex
4111 arguments into the target's word size. Some ABIs require complex arguments
4112 to be split and treated as their individual components. For example, on
4113 AIX64, complex floats should be passed in a pair of floating point
4114 registers, even though a complex float would fit in one 64-bit floating
4117 The default value of this hook is @code{NULL}, which is treated as always
4121 @deftypefn {Target Hook} tree TARGET_BUILD_BUILTIN_VA_LIST (void)
4122 This hook returns a type node for @code{va_list} for the target.
4123 The default version of the hook returns @code{void*}.
4126 @deftypefn {Target Hook} int TARGET_ENUM_VA_LIST_P (int @var{idx}, const char **@var{pname}, tree *@var{ptree})
4127 This target hook is used in function @code{c_common_nodes_and_builtins}
4128 to iterate through the target specific builtin types for va_list. The
4129 variable @var{idx} is used as iterator. @var{pname} has to be a pointer
4130 to a @code{const char *} and @var{ptree} a pointer to a @code{tree} typed
4132 The arguments @var{pname} and @var{ptree} are used to store the result of
4133 this macro and are set to the name of the va_list builtin type and its
4135 If the return value of this macro is zero, then there is no more element.
4136 Otherwise the @var{IDX} should be increased for the next call of this
4137 macro to iterate through all types.
4140 @deftypefn {Target Hook} tree TARGET_FN_ABI_VA_LIST (tree @var{fndecl})
4141 This hook returns the va_list type of the calling convention specified by
4143 The default version of this hook returns @code{va_list_type_node}.
4146 @deftypefn {Target Hook} tree TARGET_CANONICAL_VA_LIST_TYPE (tree @var{type})
4147 This hook returns the va_list type of the calling convention specified by the
4148 type of @var{type}. If @var{type} is not a valid va_list type, it returns
4152 @deftypefn {Target Hook} tree TARGET_GIMPLIFY_VA_ARG_EXPR (tree @var{valist}, tree @var{type}, gimple_seq *@var{pre_p}, gimple_seq *@var{post_p})
4153 This hook performs target-specific gimplification of
4154 @code{VA_ARG_EXPR}. The first two parameters correspond to the
4155 arguments to @code{va_arg}; the latter two are as in
4156 @code{gimplify.c:gimplify_expr}.
4159 @deftypefn {Target Hook} bool TARGET_VALID_POINTER_MODE (machine_mode @var{mode})
4160 Define this to return nonzero if the port can handle pointers
4161 with machine mode @var{mode}. The default version of this
4162 hook returns true for both @code{ptr_mode} and @code{Pmode}.
4165 @deftypefn {Target Hook} bool TARGET_REF_MAY_ALIAS_ERRNO (struct ao_ref *@var{ref})
4166 Define this to return nonzero if the memory reference @var{ref} may alias with the system C library errno location. The default version of this hook assumes the system C library errno location is either a declaration of type int or accessed by dereferencing a pointer to int.
4169 @deftypefn {Target Hook} bool TARGET_SCALAR_MODE_SUPPORTED_P (machine_mode @var{mode})
4170 Define this to return nonzero if the port is prepared to handle
4171 insns involving scalar mode @var{mode}. For a scalar mode to be
4172 considered supported, all the basic arithmetic and comparisons
4175 The default version of this hook returns true for any mode
4176 required to handle the basic C types (as defined by the port).
4177 Included here are the double-word arithmetic supported by the
4178 code in @file{optabs.c}.
4181 @deftypefn {Target Hook} bool TARGET_VECTOR_MODE_SUPPORTED_P (machine_mode @var{mode})
4182 Define this to return nonzero if the port is prepared to handle
4183 insns involving vector mode @var{mode}. At the very least, it
4184 must have move patterns for this mode.
4187 @deftypefn {Target Hook} bool TARGET_ARRAY_MODE_SUPPORTED_P (machine_mode @var{mode}, unsigned HOST_WIDE_INT @var{nelems})
4188 Return true if GCC should try to use a scalar mode to store an array
4189 of @var{nelems} elements, given that each element has mode @var{mode}.
4190 Returning true here overrides the usual @code{MAX_FIXED_MODE} limit
4191 and allows GCC to use any defined integer mode.
4193 One use of this hook is to support vector load and store operations
4194 that operate on several homogeneous vectors. For example, ARM NEON
4195 has operations like:
4198 int8x8x3_t vld3_s8 (const int8_t *)
4201 where the return type is defined as:
4204 typedef struct int8x8x3_t
4210 If this hook allows @code{val} to have a scalar mode, then
4211 @code{int8x8x3_t} can have the same mode. GCC can then store
4212 @code{int8x8x3_t}s in registers rather than forcing them onto the stack.
4215 @deftypefn {Target Hook} bool TARGET_LIBGCC_FLOATING_MODE_SUPPORTED_P (machine_mode @var{mode})
4216 Define this to return nonzero if libgcc provides support for the
4217 floating-point mode @var{mode}, which is known to pass
4218 @code{TARGET_SCALAR_MODE_SUPPORTED_P}. The default version of this
4219 hook returns true for all of @code{SFmode}, @code{DFmode},
4220 @code{XFmode} and @code{TFmode}, if such modes exist.
4223 @deftypefn {Target Hook} bool TARGET_SMALL_REGISTER_CLASSES_FOR_MODE_P (machine_mode @var{mode})
4224 Define this to return nonzero for machine modes for which the port has
4225 small register classes. If this target hook returns nonzero for a given
4226 @var{mode}, the compiler will try to minimize the lifetime of registers
4227 in @var{mode}. The hook may be called with @code{VOIDmode} as argument.
4228 In this case, the hook is expected to return nonzero if it returns nonzero
4231 On some machines, it is risky to let hard registers live across arbitrary
4232 insns. Typically, these machines have instructions that require values
4233 to be in specific registers (like an accumulator), and reload will fail
4234 if the required hard register is used for another purpose across such an
4237 Passes before reload do not know which hard registers will be used
4238 in an instruction, but the machine modes of the registers set or used in
4239 the instruction are already known. And for some machines, register
4240 classes are small for, say, integer registers but not for floating point
4241 registers. For example, the AMD x86-64 architecture requires specific
4242 registers for the legacy x86 integer instructions, but there are many
4243 SSE registers for floating point operations. On such targets, a good
4244 strategy may be to return nonzero from this hook for @code{INTEGRAL_MODE_P}
4245 machine modes but zero for the SSE register classes.
4247 The default version of this hook returns false for any mode. It is always
4248 safe to redefine this hook to return with a nonzero value. But if you
4249 unnecessarily define it, you will reduce the amount of optimizations
4250 that can be performed in some cases. If you do not define this hook
4251 to return a nonzero value when it is required, the compiler will run out
4252 of spill registers and print a fatal error message.
4255 @deftypevr {Target Hook} {unsigned int} TARGET_FLAGS_REGNUM
4256 If the target has a dedicated flags register, and it needs to use the post-reload comparison elimination pass, then this value should be set appropriately.
4260 @subsection How Scalar Function Values Are Returned
4261 @cindex return values in registers
4262 @cindex values, returned by functions
4263 @cindex scalars, returned as values
4265 This section discusses the macros that control returning scalars as
4266 values---values that can fit in registers.
4268 @deftypefn {Target Hook} rtx TARGET_FUNCTION_VALUE (const_tree @var{ret_type}, const_tree @var{fn_decl_or_type}, bool @var{outgoing})
4270 Define this to return an RTX representing the place where a function
4271 returns or receives a value of data type @var{ret_type}, a tree node
4272 representing a data type. @var{fn_decl_or_type} is a tree node
4273 representing @code{FUNCTION_DECL} or @code{FUNCTION_TYPE} of a
4274 function being called. If @var{outgoing} is false, the hook should
4275 compute the register in which the caller will see the return value.
4276 Otherwise, the hook should return an RTX representing the place where
4277 a function returns a value.
4279 On many machines, only @code{TYPE_MODE (@var{ret_type})} is relevant.
4280 (Actually, on most machines, scalar values are returned in the same
4281 place regardless of mode.) The value of the expression is usually a
4282 @code{reg} RTX for the hard register where the return value is stored.
4283 The value can also be a @code{parallel} RTX, if the return value is in
4284 multiple places. See @code{TARGET_FUNCTION_ARG} for an explanation of the
4285 @code{parallel} form. Note that the callee will populate every
4286 location specified in the @code{parallel}, but if the first element of
4287 the @code{parallel} contains the whole return value, callers will use
4288 that element as the canonical location and ignore the others. The m68k
4289 port uses this type of @code{parallel} to return pointers in both
4290 @samp{%a0} (the canonical location) and @samp{%d0}.
4292 If @code{TARGET_PROMOTE_FUNCTION_RETURN} returns true, you must apply
4293 the same promotion rules specified in @code{PROMOTE_MODE} if
4294 @var{valtype} is a scalar type.
4296 If the precise function being called is known, @var{func} is a tree
4297 node (@code{FUNCTION_DECL}) for it; otherwise, @var{func} is a null
4298 pointer. This makes it possible to use a different value-returning
4299 convention for specific functions when all their calls are
4302 Some target machines have ``register windows'' so that the register in
4303 which a function returns its value is not the same as the one in which
4304 the caller sees the value. For such machines, you should return
4305 different RTX depending on @var{outgoing}.
4307 @code{TARGET_FUNCTION_VALUE} is not used for return values with
4308 aggregate data types, because these are returned in another way. See
4309 @code{TARGET_STRUCT_VALUE_RTX} and related macros, below.
4312 @defmac FUNCTION_VALUE (@var{valtype}, @var{func})
4313 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE} for
4314 a new target instead.
4317 @defmac LIBCALL_VALUE (@var{mode})
4318 A C expression to create an RTX representing the place where a library
4319 function returns a value of mode @var{mode}.
4321 Note that ``library function'' in this context means a compiler
4322 support routine, used to perform arithmetic, whose name is known
4323 specially by the compiler and was not mentioned in the C code being
4327 @deftypefn {Target Hook} rtx TARGET_LIBCALL_VALUE (machine_mode @var{mode}, const_rtx @var{fun})
4328 Define this hook if the back-end needs to know the name of the libcall
4329 function in order to determine where the result should be returned.
4331 The mode of the result is given by @var{mode} and the name of the called
4332 library function is given by @var{fun}. The hook should return an RTX
4333 representing the place where the library function result will be returned.
4335 If this hook is not defined, then LIBCALL_VALUE will be used.
4338 @defmac FUNCTION_VALUE_REGNO_P (@var{regno})
4339 A C expression that is nonzero if @var{regno} is the number of a hard
4340 register in which the values of called function may come back.
4342 A register whose use for returning values is limited to serving as the
4343 second of a pair (for a value of type @code{double}, say) need not be
4344 recognized by this macro. So for most machines, this definition
4348 #define FUNCTION_VALUE_REGNO_P(N) ((N) == 0)
4351 If the machine has register windows, so that the caller and the called
4352 function use different registers for the return value, this macro
4353 should recognize only the caller's register numbers.
4355 This macro has been deprecated. Use @code{TARGET_FUNCTION_VALUE_REGNO_P}
4356 for a new target instead.
4359 @deftypefn {Target Hook} bool TARGET_FUNCTION_VALUE_REGNO_P (const unsigned int @var{regno})
4360 A target hook that return @code{true} if @var{regno} is the number of a hard
4361 register in which the values of called function may come back.
4363 A register whose use for returning values is limited to serving as the
4364 second of a pair (for a value of type @code{double}, say) need not be
4365 recognized by this target hook.
4367 If the machine has register windows, so that the caller and the called
4368 function use different registers for the return value, this target hook
4369 should recognize only the caller's register numbers.
4371 If this hook is not defined, then FUNCTION_VALUE_REGNO_P will be used.
4374 @defmac APPLY_RESULT_SIZE
4375 Define this macro if @samp{untyped_call} and @samp{untyped_return}
4376 need more space than is implied by @code{FUNCTION_VALUE_REGNO_P} for
4377 saving and restoring an arbitrary return value.
4380 @deftypevr {Target Hook} bool TARGET_OMIT_STRUCT_RETURN_REG
4381 Normally, when a function returns a structure by memory, the address
4382 is passed as an invisible pointer argument, but the compiler also
4383 arranges to return the address from the function like it would a normal
4384 pointer return value. Define this to true if that behaviour is
4385 undesirable on your target.
4388 @deftypefn {Target Hook} bool TARGET_RETURN_IN_MSB (const_tree @var{type})
4389 This hook should return true if values of type @var{type} are returned
4390 at the most significant end of a register (in other words, if they are
4391 padded at the least significant end). You can assume that @var{type}
4392 is returned in a register; the caller is required to check this.
4394 Note that the register provided by @code{TARGET_FUNCTION_VALUE} must
4395 be able to hold the complete return value. For example, if a 1-, 2-
4396 or 3-byte structure is returned at the most significant end of a
4397 4-byte register, @code{TARGET_FUNCTION_VALUE} should provide an
4401 @node Aggregate Return
4402 @subsection How Large Values Are Returned
4403 @cindex aggregates as return values
4404 @cindex large return values
4405 @cindex returning aggregate values
4406 @cindex structure value address
4408 When a function value's mode is @code{BLKmode} (and in some other
4409 cases), the value is not returned according to
4410 @code{TARGET_FUNCTION_VALUE} (@pxref{Scalar Return}). Instead, the
4411 caller passes the address of a block of memory in which the value
4412 should be stored. This address is called the @dfn{structure value
4415 This section describes how to control returning structure values in
4418 @deftypefn {Target Hook} bool TARGET_RETURN_IN_MEMORY (const_tree @var{type}, const_tree @var{fntype})
4419 This target hook should return a nonzero value to say to return the
4420 function value in memory, just as large structures are always returned.
4421 Here @var{type} will be the data type of the value, and @var{fntype}
4422 will be the type of the function doing the returning, or @code{NULL} for
4425 Note that values of mode @code{BLKmode} must be explicitly handled
4426 by this function. Also, the option @option{-fpcc-struct-return}
4427 takes effect regardless of this macro. On most systems, it is
4428 possible to leave the hook undefined; this causes a default
4429 definition to be used, whose value is the constant 1 for @code{BLKmode}
4430 values, and 0 otherwise.
4432 Do not use this hook to indicate that structures and unions should always
4433 be returned in memory. You should instead use @code{DEFAULT_PCC_STRUCT_RETURN}
4437 @defmac DEFAULT_PCC_STRUCT_RETURN
4438 Define this macro to be 1 if all structure and union return values must be
4439 in memory. Since this results in slower code, this should be defined
4440 only if needed for compatibility with other compilers or with an ABI@.
4441 If you define this macro to be 0, then the conventions used for structure
4442 and union return values are decided by the @code{TARGET_RETURN_IN_MEMORY}
4445 If not defined, this defaults to the value 1.
4448 @deftypefn {Target Hook} rtx TARGET_STRUCT_VALUE_RTX (tree @var{fndecl}, int @var{incoming})
4449 This target hook should return the location of the structure value
4450 address (normally a @code{mem} or @code{reg}), or 0 if the address is
4451 passed as an ``invisible'' first argument. Note that @var{fndecl} may
4452 be @code{NULL}, for libcalls. You do not need to define this target
4453 hook if the address is always passed as an ``invisible'' first
4456 On some architectures the place where the structure value address
4457 is found by the called function is not the same place that the
4458 caller put it. This can be due to register windows, or it could
4459 be because the function prologue moves it to a different place.
4460 @var{incoming} is @code{1} or @code{2} when the location is needed in
4461 the context of the called function, and @code{0} in the context of
4464 If @var{incoming} is nonzero and the address is to be found on the
4465 stack, return a @code{mem} which refers to the frame pointer. If
4466 @var{incoming} is @code{2}, the result is being used to fetch the
4467 structure value address at the beginning of a function. If you need
4468 to emit adjusting code, you should do it at this point.
4471 @defmac PCC_STATIC_STRUCT_RETURN
4472 Define this macro if the usual system convention on the target machine
4473 for returning structures and unions is for the called function to return
4474 the address of a static variable containing the value.
4476 Do not define this if the usual system convention is for the caller to
4477 pass an address to the subroutine.
4479 This macro has effect in @option{-fpcc-struct-return} mode, but it does
4480 nothing when you use @option{-freg-struct-return} mode.
4483 @deftypefn {Target Hook} machine_mode TARGET_GET_RAW_RESULT_MODE (int @var{regno})
4484 This target hook returns the mode to be used when accessing raw return registers in @code{__builtin_return}. Define this macro if the value in @var{reg_raw_mode} is not correct.
4487 @deftypefn {Target Hook} machine_mode TARGET_GET_RAW_ARG_MODE (int @var{regno})
4488 This target hook returns the mode to be used when accessing raw argument registers in @code{__builtin_apply_args}. Define this macro if the value in @var{reg_raw_mode} is not correct.
4492 @subsection Caller-Saves Register Allocation
4494 If you enable it, GCC can save registers around function calls. This
4495 makes it possible to use call-clobbered registers to hold variables that
4496 must live across calls.
4498 @defmac HARD_REGNO_CALLER_SAVE_MODE (@var{regno}, @var{nregs})
4499 A C expression specifying which mode is required for saving @var{nregs}
4500 of a pseudo-register in call-clobbered hard register @var{regno}. If
4501 @var{regno} is unsuitable for caller save, @code{VOIDmode} should be
4502 returned. For most machines this macro need not be defined since GCC
4503 will select the smallest suitable mode.
4506 @node Function Entry
4507 @subsection Function Entry and Exit
4508 @cindex function entry and exit
4512 This section describes the macros that output function entry
4513 (@dfn{prologue}) and exit (@dfn{epilogue}) code.
4515 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_PROLOGUE (FILE *@var{file}, HOST_WIDE_INT @var{size})
4516 If defined, a function that outputs the assembler code for entry to a
4517 function. The prologue is responsible for setting up the stack frame,
4518 initializing the frame pointer register, saving registers that must be
4519 saved, and allocating @var{size} additional bytes of storage for the
4520 local variables. @var{size} is an integer. @var{file} is a stdio
4521 stream to which the assembler code should be output.
4523 The label for the beginning of the function need not be output by this
4524 macro. That has already been done when the macro is run.
4526 @findex regs_ever_live
4527 To determine which registers to save, the macro can refer to the array
4528 @code{regs_ever_live}: element @var{r} is nonzero if hard register
4529 @var{r} is used anywhere within the function. This implies the function
4530 prologue should save register @var{r}, provided it is not one of the
4531 call-used registers. (@code{TARGET_ASM_FUNCTION_EPILOGUE} must likewise use
4532 @code{regs_ever_live}.)
4534 On machines that have ``register windows'', the function entry code does
4535 not save on the stack the registers that are in the windows, even if
4536 they are supposed to be preserved by function calls; instead it takes
4537 appropriate steps to ``push'' the register stack, if any non-call-used
4538 registers are used in the function.
4540 @findex frame_pointer_needed
4541 On machines where functions may or may not have frame-pointers, the
4542 function entry code must vary accordingly; it must set up the frame
4543 pointer if one is wanted, and not otherwise. To determine whether a
4544 frame pointer is in wanted, the macro can refer to the variable
4545 @code{frame_pointer_needed}. The variable's value will be 1 at run
4546 time in a function that needs a frame pointer. @xref{Elimination}.
4548 The function entry code is responsible for allocating any stack space
4549 required for the function. This stack space consists of the regions
4550 listed below. In most cases, these regions are allocated in the
4551 order listed, with the last listed region closest to the top of the
4552 stack (the lowest address if @code{STACK_GROWS_DOWNWARD} is defined, and
4553 the highest address if it is not defined). You can use a different order
4554 for a machine if doing so is more convenient or required for
4555 compatibility reasons. Except in cases where required by standard
4556 or by a debugger, there is no reason why the stack layout used by GCC
4557 need agree with that used by other compilers for a machine.
4560 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_END_PROLOGUE (FILE *@var{file})
4561 If defined, a function that outputs assembler code at the end of a
4562 prologue. This should be used when the function prologue is being
4563 emitted as RTL, and you have some extra assembler that needs to be
4564 emitted. @xref{prologue instruction pattern}.
4567 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_BEGIN_EPILOGUE (FILE *@var{file})
4568 If defined, a function that outputs assembler code at the start of an
4569 epilogue. This should be used when the function epilogue is being
4570 emitted as RTL, and you have some extra assembler that needs to be
4571 emitted. @xref{epilogue instruction pattern}.
4574 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_EPILOGUE (FILE *@var{file}, HOST_WIDE_INT @var{size})
4575 If defined, a function that outputs the assembler code for exit from a
4576 function. The epilogue is responsible for restoring the saved
4577 registers and stack pointer to their values when the function was
4578 called, and returning control to the caller. This macro takes the
4579 same arguments as the macro @code{TARGET_ASM_FUNCTION_PROLOGUE}, and the
4580 registers to restore are determined from @code{regs_ever_live} and
4581 @code{CALL_USED_REGISTERS} in the same way.
4583 On some machines, there is a single instruction that does all the work
4584 of returning from the function. On these machines, give that
4585 instruction the name @samp{return} and do not define the macro
4586 @code{TARGET_ASM_FUNCTION_EPILOGUE} at all.
4588 Do not define a pattern named @samp{return} if you want the
4589 @code{TARGET_ASM_FUNCTION_EPILOGUE} to be used. If you want the target
4590 switches to control whether return instructions or epilogues are used,
4591 define a @samp{return} pattern with a validity condition that tests the
4592 target switches appropriately. If the @samp{return} pattern's validity
4593 condition is false, epilogues will be used.
4595 On machines where functions may or may not have frame-pointers, the
4596 function exit code must vary accordingly. Sometimes the code for these
4597 two cases is completely different. To determine whether a frame pointer
4598 is wanted, the macro can refer to the variable
4599 @code{frame_pointer_needed}. The variable's value will be 1 when compiling
4600 a function that needs a frame pointer.
4602 Normally, @code{TARGET_ASM_FUNCTION_PROLOGUE} and
4603 @code{TARGET_ASM_FUNCTION_EPILOGUE} must treat leaf functions specially.
4604 The C variable @code{current_function_is_leaf} is nonzero for such a
4605 function. @xref{Leaf Functions}.
4607 On some machines, some functions pop their arguments on exit while
4608 others leave that for the caller to do. For example, the 68020 when
4609 given @option{-mrtd} pops arguments in functions that take a fixed
4610 number of arguments.
4613 @findex crtl->args.pops_args
4614 Your definition of the macro @code{RETURN_POPS_ARGS} decides which
4615 functions pop their own arguments. @code{TARGET_ASM_FUNCTION_EPILOGUE}
4616 needs to know what was decided. The number of bytes of the current
4617 function's arguments that this function should pop is available in
4618 @code{crtl->args.pops_args}. @xref{Scalar Return}.
4623 @findex pretend_args_size
4624 @findex crtl->args.pretend_args_size
4625 A region of @code{crtl->args.pretend_args_size} bytes of
4626 uninitialized space just underneath the first argument arriving on the
4627 stack. (This may not be at the very start of the allocated stack region
4628 if the calling sequence has pushed anything else since pushing the stack
4629 arguments. But usually, on such machines, nothing else has been pushed
4630 yet, because the function prologue itself does all the pushing.) This
4631 region is used on machines where an argument may be passed partly in
4632 registers and partly in memory, and, in some cases to support the
4633 features in @code{<stdarg.h>}.
4636 An area of memory used to save certain registers used by the function.
4637 The size of this area, which may also include space for such things as
4638 the return address and pointers to previous stack frames, is
4639 machine-specific and usually depends on which registers have been used
4640 in the function. Machines with register windows often do not require
4644 A region of at least @var{size} bytes, possibly rounded up to an allocation
4645 boundary, to contain the local variables of the function. On some machines,
4646 this region and the save area may occur in the opposite order, with the
4647 save area closer to the top of the stack.
4650 @cindex @code{ACCUMULATE_OUTGOING_ARGS} and stack frames
4651 Optionally, when @code{ACCUMULATE_OUTGOING_ARGS} is defined, a region of
4652 @code{crtl->outgoing_args_size} bytes to be used for outgoing
4653 argument lists of the function. @xref{Stack Arguments}.
4656 @defmac EXIT_IGNORE_STACK
4657 Define this macro as a C expression that is nonzero if the return
4658 instruction or the function epilogue ignores the value of the stack
4659 pointer; in other words, if it is safe to delete an instruction to
4660 adjust the stack pointer before a return from the function. The
4663 Note that this macro's value is relevant only for functions for which
4664 frame pointers are maintained. It is never safe to delete a final
4665 stack adjustment in a function that has no frame pointer, and the
4666 compiler knows this regardless of @code{EXIT_IGNORE_STACK}.
4669 @defmac EPILOGUE_USES (@var{regno})
4670 Define this macro as a C expression that is nonzero for registers that are
4671 used by the epilogue or the @samp{return} pattern. The stack and frame
4672 pointer registers are already assumed to be used as needed.
4675 @defmac EH_USES (@var{regno})
4676 Define this macro as a C expression that is nonzero for registers that are
4677 used by the exception handling mechanism, and so should be considered live
4678 on entry to an exception edge.
4681 @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_MI_THUNK (FILE *@var{file}, tree @var{thunk_fndecl}, HOST_WIDE_INT @var{delta}, HOST_WIDE_INT @var{vcall_offset}, tree @var{function})
4682 A function that outputs the assembler code for a thunk
4683 function, used to implement C++ virtual function calls with multiple
4684 inheritance. The thunk acts as a wrapper around a virtual function,
4685 adjusting the implicit object parameter before handing control off to
4688 First, emit code to add the integer @var{delta} to the location that
4689 contains the incoming first argument. Assume that this argument
4690 contains a pointer, and is the one used to pass the @code{this} pointer
4691 in C++. This is the incoming argument @emph{before} the function prologue,
4692 e.g.@: @samp{%o0} on a sparc. The addition must preserve the values of
4693 all other incoming arguments.
4695 Then, if @var{vcall_offset} is nonzero, an additional adjustment should be
4696 made after adding @code{delta}. In particular, if @var{p} is the
4697 adjusted pointer, the following adjustment should be made:
4700 p += (*((ptrdiff_t **)p))[vcall_offset/sizeof(ptrdiff_t)]
4703 After the additions, emit code to jump to @var{function}, which is a
4704 @code{FUNCTION_DECL}. This is a direct pure jump, not a call, and does
4705 not touch the return address. Hence returning from @var{FUNCTION} will
4706 return to whoever called the current @samp{thunk}.
4708 The effect must be as if @var{function} had been called directly with
4709 the adjusted first argument. This macro is responsible for emitting all
4710 of the code for a thunk function; @code{TARGET_ASM_FUNCTION_PROLOGUE}
4711 and @code{TARGET_ASM_FUNCTION_EPILOGUE} are not invoked.
4713 The @var{thunk_fndecl} is redundant. (@var{delta} and @var{function}
4714 have already been extracted from it.) It might possibly be useful on
4715 some targets, but probably not.
4717 If you do not define this macro, the target-independent code in the C++
4718 front end will generate a less efficient heavyweight thunk that calls
4719 @var{function} instead of jumping to it. The generic approach does
4720 not support varargs.
4723 @deftypefn {Target Hook} bool TARGET_ASM_CAN_OUTPUT_MI_THUNK (const_tree @var{thunk_fndecl}, HOST_WIDE_INT @var{delta}, HOST_WIDE_INT @var{vcall_offset}, const_tree @var{function})
4724 A function that returns true if TARGET_ASM_OUTPUT_MI_THUNK would be able
4725 to output the assembler code for the thunk function specified by the
4726 arguments it is passed, and false otherwise. In the latter case, the
4727 generic approach will be used by the C++ front end, with the limitations
4732 @subsection Generating Code for Profiling
4733 @cindex profiling, code generation
4735 These macros will help you generate code for profiling.
4737 @defmac FUNCTION_PROFILER (@var{file}, @var{labelno})
4738 A C statement or compound statement to output to @var{file} some
4739 assembler code to call the profiling subroutine @code{mcount}.
4742 The details of how @code{mcount} expects to be called are determined by
4743 your operating system environment, not by GCC@. To figure them out,
4744 compile a small program for profiling using the system's installed C
4745 compiler and look at the assembler code that results.
4747 Older implementations of @code{mcount} expect the address of a counter
4748 variable to be loaded into some register. The name of this variable is
4749 @samp{LP} followed by the number @var{labelno}, so you would generate
4750 the name using @samp{LP%d} in a @code{fprintf}.
4753 @defmac PROFILE_HOOK
4754 A C statement or compound statement to output to @var{file} some assembly
4755 code to call the profiling subroutine @code{mcount} even the target does
4756 not support profiling.
4759 @defmac NO_PROFILE_COUNTERS
4760 Define this macro to be an expression with a nonzero value if the
4761 @code{mcount} subroutine on your system does not need a counter variable
4762 allocated for each function. This is true for almost all modern
4763 implementations. If you define this macro, you must not use the
4764 @var{labelno} argument to @code{FUNCTION_PROFILER}.
4767 @defmac PROFILE_BEFORE_PROLOGUE
4768 Define this macro if the code for function profiling should come before
4769 the function prologue. Normally, the profiling code comes after.
4772 @deftypefn {Target Hook} bool TARGET_KEEP_LEAF_WHEN_PROFILED (void)
4773 This target hook returns true if the target wants the leaf flag for the current function to stay true even if it calls mcount. This might make sense for targets using the leaf flag only to determine whether a stack frame needs to be generated or not and for which the call to mcount is generated before the function prologue.
4777 @subsection Permitting tail calls
4780 @deftypefn {Target Hook} bool TARGET_FUNCTION_OK_FOR_SIBCALL (tree @var{decl}, tree @var{exp})
4781 True if it is OK to do sibling call optimization for the specified
4782 call expression @var{exp}. @var{decl} will be the called function,
4783 or @code{NULL} if this is an indirect call.
4785 It is not uncommon for limitations of calling conventions to prevent
4786 tail calls to functions outside the current unit of translation, or
4787 during PIC compilation. The hook is used to enforce these restrictions,
4788 as the @code{sibcall} md pattern can not fail, or fall over to a
4789 ``normal'' call. The criteria for successful sibling call optimization
4790 may vary greatly between different architectures.
4793 @deftypefn {Target Hook} void TARGET_EXTRA_LIVE_ON_ENTRY (bitmap @var{regs})
4794 Add any hard registers to @var{regs} that are live on entry to the
4795 function. This hook only needs to be defined to provide registers that
4796 cannot be found by examination of FUNCTION_ARG_REGNO_P, the callee saved
4797 registers, STATIC_CHAIN_INCOMING_REGNUM, STATIC_CHAIN_REGNUM,
4798 TARGET_STRUCT_VALUE_RTX, FRAME_POINTER_REGNUM, EH_USES,
4799 FRAME_POINTER_REGNUM, ARG_POINTER_REGNUM, and the PIC_OFFSET_TABLE_REGNUM.
4802 @deftypefn {Target Hook} void TARGET_SET_UP_BY_PROLOGUE (struct hard_reg_set_container *@var{})
4803 This hook should add additional registers that are computed by the prologue to the hard regset for shrink-wrapping optimization purposes.
4806 @deftypefn {Target Hook} bool TARGET_WARN_FUNC_RETURN (tree)
4807 True if a function's return statements should be checked for matching the function's return type. This includes checking for falling off the end of a non-void function. Return false if no such check should be made.
4810 @node Stack Smashing Protection
4811 @subsection Stack smashing protection
4812 @cindex stack smashing protection
4814 @deftypefn {Target Hook} tree TARGET_STACK_PROTECT_GUARD (void)
4815 This hook returns a @code{DECL} node for the external variable to use
4816 for the stack protection guard. This variable is initialized by the
4817 runtime to some random value and is used to initialize the guard value
4818 that is placed at the top of the local stack frame. The type of this
4819 variable must be @code{ptr_type_node}.
4821 The default version of this hook creates a variable called
4822 @samp{__stack_chk_guard}, which is normally defined in @file{libgcc2.c}.
4825 @deftypefn {Target Hook} tree TARGET_STACK_PROTECT_FAIL (void)
4826 This hook returns a @code{CALL_EXPR} that alerts the runtime that the
4827 stack protect guard variable has been modified. This expression should
4828 involve a call to a @code{noreturn} function.
4830 The default version of this hook invokes a function called
4831 @samp{__stack_chk_fail}, taking no arguments. This function is
4832 normally defined in @file{libgcc2.c}.
4835 @deftypefn {Common Target Hook} bool TARGET_SUPPORTS_SPLIT_STACK (bool @var{report}, struct gcc_options *@var{opts})
4836 Whether this target supports splitting the stack when the options described in @var{opts} have been passed. This is called after options have been parsed, so the target may reject splitting the stack in some configurations. The default version of this hook returns false. If @var{report} is true, this function may issue a warning or error; if @var{report} is false, it must simply return a value
4839 @node Miscellaneous Register Hooks
4840 @subsection Miscellaneous register hooks
4841 @cindex miscellaneous register hooks
4843 @deftypevr {Target Hook} bool TARGET_CALL_FUSAGE_CONTAINS_NON_CALLEE_CLOBBERS
4844 Set to true if each call that binds to a local definition explicitly
4845 clobbers or sets all non-fixed registers modified by performing the call.
4846 That is, by the call pattern itself, or by code that might be inserted by the
4847 linker (e.g. stubs, veneers, branch islands), but not including those
4848 modifiable by the callee. The affected registers may be mentioned explicitly
4849 in the call pattern, or included as clobbers in CALL_INSN_FUNCTION_USAGE.
4850 The default version of this hook is set to false. The purpose of this hook
4851 is to enable the fuse-caller-save optimization.
4855 @section Implementing the Varargs Macros
4856 @cindex varargs implementation
4858 GCC comes with an implementation of @code{<varargs.h>} and
4859 @code{<stdarg.h>} that work without change on machines that pass arguments
4860 on the stack. Other machines require their own implementations of
4861 varargs, and the two machine independent header files must have
4862 conditionals to include it.
4864 ISO @code{<stdarg.h>} differs from traditional @code{<varargs.h>} mainly in
4865 the calling convention for @code{va_start}. The traditional
4866 implementation takes just one argument, which is the variable in which
4867 to store the argument pointer. The ISO implementation of
4868 @code{va_start} takes an additional second argument. The user is
4869 supposed to write the last named argument of the function here.
4871 However, @code{va_start} should not use this argument. The way to find
4872 the end of the named arguments is with the built-in functions described
4875 @defmac __builtin_saveregs ()
4876 Use this built-in function to save the argument registers in memory so
4877 that the varargs mechanism can access them. Both ISO and traditional
4878 versions of @code{va_start} must use @code{__builtin_saveregs}, unless
4879 you use @code{TARGET_SETUP_INCOMING_VARARGS} (see below) instead.
4881 On some machines, @code{__builtin_saveregs} is open-coded under the
4882 control of the target hook @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. On
4883 other machines, it calls a routine written in assembler language,
4884 found in @file{libgcc2.c}.
4886 Code generated for the call to @code{__builtin_saveregs} appears at the
4887 beginning of the function, as opposed to where the call to
4888 @code{__builtin_saveregs} is written, regardless of what the code is.
4889 This is because the registers must be saved before the function starts
4890 to use them for its own purposes.
4891 @c i rewrote the first sentence above to fix an overfull hbox. --mew
4895 @defmac __builtin_next_arg (@var{lastarg})
4896 This builtin returns the address of the first anonymous stack
4897 argument, as type @code{void *}. If @code{ARGS_GROW_DOWNWARD}, it
4898 returns the address of the location above the first anonymous stack
4899 argument. Use it in @code{va_start} to initialize the pointer for
4900 fetching arguments from the stack. Also use it in @code{va_start} to
4901 verify that the second parameter @var{lastarg} is the last named argument
4902 of the current function.
4905 @defmac __builtin_classify_type (@var{object})
4906 Since each machine has its own conventions for which data types are
4907 passed in which kind of register, your implementation of @code{va_arg}
4908 has to embody these conventions. The easiest way to categorize the
4909 specified data type is to use @code{__builtin_classify_type} together
4910 with @code{sizeof} and @code{__alignof__}.
4912 @code{__builtin_classify_type} ignores the value of @var{object},
4913 considering only its data type. It returns an integer describing what
4914 kind of type that is---integer, floating, pointer, structure, and so on.
4916 The file @file{typeclass.h} defines an enumeration that you can use to
4917 interpret the values of @code{__builtin_classify_type}.
4920 These machine description macros help implement varargs:
4922 @deftypefn {Target Hook} rtx TARGET_EXPAND_BUILTIN_SAVEREGS (void)
4923 If defined, this hook produces the machine-specific code for a call to
4924 @code{__builtin_saveregs}. This code will be moved to the very
4925 beginning of the function, before any parameter access are made. The
4926 return value of this function should be an RTX that contains the value
4927 to use as the return of @code{__builtin_saveregs}.
4930 @deftypefn {Target Hook} void TARGET_SETUP_INCOMING_VARARGS (cumulative_args_t @var{args_so_far}, machine_mode @var{mode}, tree @var{type}, int *@var{pretend_args_size}, int @var{second_time})
4931 This target hook offers an alternative to using
4932 @code{__builtin_saveregs} and defining the hook
4933 @code{TARGET_EXPAND_BUILTIN_SAVEREGS}. Use it to store the anonymous
4934 register arguments into the stack so that all the arguments appear to
4935 have been passed consecutively on the stack. Once this is done, you can
4936 use the standard implementation of varargs that works for machines that
4937 pass all their arguments on the stack.
4939 The argument @var{args_so_far} points to the @code{CUMULATIVE_ARGS} data
4940 structure, containing the values that are obtained after processing the
4941 named arguments. The arguments @var{mode} and @var{type} describe the
4942 last named argument---its machine mode and its data type as a tree node.
4944 The target hook should do two things: first, push onto the stack all the
4945 argument registers @emph{not} used for the named arguments, and second,
4946 store the size of the data thus pushed into the @code{int}-valued
4947 variable pointed to by @var{pretend_args_size}. The value that you
4948 store here will serve as additional offset for setting up the stack
4951 Because you must generate code to push the anonymous arguments at
4952 compile time without knowing their data types,
4953 @code{TARGET_SETUP_INCOMING_VARARGS} is only useful on machines that
4954 have just a single category of argument register and use it uniformly
4957 If the argument @var{second_time} is nonzero, it means that the
4958 arguments of the function are being analyzed for the second time. This
4959 happens for an inline function, which is not actually compiled until the
4960 end of the source file. The hook @code{TARGET_SETUP_INCOMING_VARARGS} should
4961 not generate any instructions in this case.
4964 @deftypefn {Target Hook} bool TARGET_STRICT_ARGUMENT_NAMING (cumulative_args_t @var{ca})
4965 Define this hook to return @code{true} if the location where a function
4966 argument is passed depends on whether or not it is a named argument.
4968 This hook controls how the @var{named} argument to @code{TARGET_FUNCTION_ARG}
4969 is set for varargs and stdarg functions. If this hook returns
4970 @code{true}, the @var{named} argument is always true for named
4971 arguments, and false for unnamed arguments. If it returns @code{false},
4972 but @code{TARGET_PRETEND_OUTGOING_VARARGS_NAMED} returns @code{true},
4973 then all arguments are treated as named. Otherwise, all named arguments
4974 except the last are treated as named.
4976 You need not define this hook if it always returns @code{false}.
4979 @deftypefn {Target Hook} void TARGET_CALL_ARGS (rtx, @var{tree})
4980 While generating RTL for a function call, this target hook is invoked once
4981 for each argument passed to the function, either a register returned by
4982 @code{TARGET_FUNCTION_ARG} or a memory location. It is called just
4983 before the point where argument registers are stored. The type of the
4984 function to be called is also passed as the second argument; it is
4985 @code{NULL_TREE} for libcalls. The @code{TARGET_END_CALL_ARGS} hook is
4986 invoked just after the code to copy the return reg has been emitted.
4987 This functionality can be used to perform special setup of call argument
4988 registers if a target needs it.
4989 For functions without arguments, the hook is called once with @code{pc_rtx}
4990 passed instead of an argument register.
4991 Most ports do not need to implement anything for this hook.
4994 @deftypefn {Target Hook} void TARGET_END_CALL_ARGS (void)
4995 This target hook is invoked while generating RTL for a function call,
4996 just after the point where the return reg is copied into a pseudo. It
4997 signals that all the call argument and return registers for the just
4998 emitted call are now no longer in use.
4999 Most ports do not need to implement anything for this hook.
5002 @deftypefn {Target Hook} bool TARGET_PRETEND_OUTGOING_VARARGS_NAMED (cumulative_args_t @var{ca})
5003 If you need to conditionally change ABIs so that one works with
5004 @code{TARGET_SETUP_INCOMING_VARARGS}, but the other works like neither
5005 @code{TARGET_SETUP_INCOMING_VARARGS} nor @code{TARGET_STRICT_ARGUMENT_NAMING} was
5006 defined, then define this hook to return @code{true} if
5007 @code{TARGET_SETUP_INCOMING_VARARGS} is used, @code{false} otherwise.
5008 Otherwise, you should not define this hook.
5011 @deftypefn {Target Hook} rtx TARGET_LOAD_BOUNDS_FOR_ARG (rtx @var{slot}, rtx @var{arg}, rtx @var{slot_no})
5012 This hook is used by expand pass to emit insn to load bounds of
5013 @var{arg} passed in @var{slot}. Expand pass uses this hook in case
5014 bounds of @var{arg} are not passed in register. If @var{slot} is a
5015 memory, then bounds are loaded as for regular pointer loaded from
5016 memory. If @var{slot} is not a memory then @var{slot_no} is an integer
5017 constant holding number of the target dependent special slot which
5018 should be used to obtain bounds. Hook returns RTX holding loaded bounds.
5021 @deftypefn {Target Hook} void TARGET_STORE_BOUNDS_FOR_ARG (rtx @var{arg}, rtx @var{slot}, rtx @var{bounds}, rtx @var{slot_no})
5022 This hook is used by expand pass to emit insns to store @var{bounds} of
5023 @var{arg} passed in @var{slot}. Expand pass uses this hook in case
5024 @var{bounds} of @var{arg} are not passed in register. If @var{slot} is a
5025 memory, then @var{bounds} are stored as for regular pointer stored in
5026 memory. If @var{slot} is not a memory then @var{slot_no} is an integer
5027 constant holding number of the target dependent special slot which
5028 should be used to store @var{bounds}.
5031 @deftypefn {Target Hook} rtx TARGET_LOAD_RETURNED_BOUNDS (rtx @var{slot})
5032 This hook is used by expand pass to emit insn to load bounds
5033 returned by function call in @var{slot}. Hook returns RTX holding
5037 @deftypefn {Target Hook} void TARGET_STORE_RETURNED_BOUNDS (rtx @var{slot}, rtx @var{bounds})
5038 This hook is used by expand pass to emit insn to store @var{bounds}
5039 returned by function call into @var{slot}.
5042 @deftypefn {Target Hook} rtx TARGET_CHKP_FUNCTION_VALUE_BOUNDS (const_tree @var{ret_type}, const_tree @var{fn_decl_or_type}, bool @var{outgoing})
5043 Define this to return an RTX representing the place where a function
5044 returns bounds for returned pointers. Arguments meaning is similar to
5045 @code{TARGET_FUNCTION_VALUE}.
5048 @deftypefn {Target Hook} void TARGET_SETUP_INCOMING_VARARG_BOUNDS (cumulative_args_t @var{args_so_far}, enum machine_mode @var{mode}, tree @var{type}, int *@var{pretend_args_size}, int @var{second_time})
5049 Use it to store bounds for anonymous register arguments stored
5050 into the stack. Arguments meaning is similar to
5051 @code{TARGET_SETUP_INCOMING_VARARGS}.
5055 @section Trampolines for Nested Functions
5056 @cindex trampolines for nested functions
5057 @cindex nested functions, trampolines for
5059 A @dfn{trampoline} is a small piece of code that is created at run time
5060 when the address of a nested function is taken. It normally resides on
5061 the stack, in the stack frame of the containing function. These macros
5062 tell GCC how to generate code to allocate and initialize a
5065 The instructions in the trampoline must do two things: load a constant
5066 address into the static chain register, and jump to the real address of
5067 the nested function. On CISC machines such as the m68k, this requires
5068 two instructions, a move immediate and a jump. Then the two addresses
5069 exist in the trampoline as word-long immediate operands. On RISC
5070 machines, it is often necessary to load each address into a register in
5071 two parts. Then pieces of each address form separate immediate
5074 The code generated to initialize the trampoline must store the variable
5075 parts---the static chain value and the function address---into the
5076 immediate operands of the instructions. On a CISC machine, this is
5077 simply a matter of copying each address to a memory reference at the
5078 proper offset from the start of the trampoline. On a RISC machine, it
5079 may be necessary to take out pieces of the address and store them
5082 @deftypefn {Target Hook} void TARGET_ASM_TRAMPOLINE_TEMPLATE (FILE *@var{f})
5083 This hook is called by @code{assemble_trampoline_template} to output,
5084 on the stream @var{f}, assembler code for a block of data that contains
5085 the constant parts of a trampoline. This code should not include a
5086 label---the label is taken care of automatically.
5088 If you do not define this hook, it means no template is needed
5089 for the target. Do not define this hook on systems where the block move
5090 code to copy the trampoline into place would be larger than the code
5091 to generate it on the spot.
5094 @defmac TRAMPOLINE_SECTION
5095 Return the section into which the trampoline template is to be placed
5096 (@pxref{Sections}). The default value is @code{readonly_data_section}.
5099 @defmac TRAMPOLINE_SIZE
5100 A C expression for the size in bytes of the trampoline, as an integer.
5103 @defmac TRAMPOLINE_ALIGNMENT
5104 Alignment required for trampolines, in bits.
5106 If you don't define this macro, the value of @code{FUNCTION_ALIGNMENT}
5107 is used for aligning trampolines.
5110 @deftypefn {Target Hook} void TARGET_TRAMPOLINE_INIT (rtx @var{m_tramp}, tree @var{fndecl}, rtx @var{static_chain})
5111 This hook is called to initialize a trampoline.
5112 @var{m_tramp} is an RTX for the memory block for the trampoline; @var{fndecl}
5113 is the @code{FUNCTION_DECL} for the nested function; @var{static_chain} is an
5114 RTX for the static chain value that should be passed to the function
5117 If the target defines @code{TARGET_ASM_TRAMPOLINE_TEMPLATE}, then the
5118 first thing this hook should do is emit a block move into @var{m_tramp}
5119 from the memory block returned by @code{assemble_trampoline_template}.
5120 Note that the block move need only cover the constant parts of the
5121 trampoline. If the target isolates the variable parts of the trampoline
5122 to the end, not all @code{TRAMPOLINE_SIZE} bytes need be copied.
5124 If the target requires any other actions, such as flushing caches or
5125 enabling stack execution, these actions should be performed after
5126 initializing the trampoline proper.
5129 @deftypefn {Target Hook} rtx TARGET_TRAMPOLINE_ADJUST_ADDRESS (rtx @var{addr})
5130 This hook should perform any machine-specific adjustment in
5131 the address of the trampoline. Its argument contains the address of the
5132 memory block that was passed to @code{TARGET_TRAMPOLINE_INIT}. In case
5133 the address to be used for a function call should be different from the
5134 address at which the template was stored, the different address should
5135 be returned; otherwise @var{addr} should be returned unchanged.
5136 If this hook is not defined, @var{addr} will be used for function calls.
5139 Implementing trampolines is difficult on many machines because they have
5140 separate instruction and data caches. Writing into a stack location
5141 fails to clear the memory in the instruction cache, so when the program
5142 jumps to that location, it executes the old contents.
5144 Here are two possible solutions. One is to clear the relevant parts of
5145 the instruction cache whenever a trampoline is set up. The other is to
5146 make all trampolines identical, by having them jump to a standard
5147 subroutine. The former technique makes trampoline execution faster; the
5148 latter makes initialization faster.
5150 To clear the instruction cache when a trampoline is initialized, define
5151 the following macro.
5153 @defmac CLEAR_INSN_CACHE (@var{beg}, @var{end})
5154 If defined, expands to a C expression clearing the @emph{instruction
5155 cache} in the specified interval. The definition of this macro would
5156 typically be a series of @code{asm} statements. Both @var{beg} and
5157 @var{end} are both pointer expressions.
5160 To use a standard subroutine, define the following macro. In addition,
5161 you must make sure that the instructions in a trampoline fill an entire
5162 cache line with identical instructions, or else ensure that the
5163 beginning of the trampoline code is always aligned at the same point in
5164 its cache line. Look in @file{m68k.h} as a guide.
5166 @defmac TRANSFER_FROM_TRAMPOLINE
5167 Define this macro if trampolines need a special subroutine to do their
5168 work. The macro should expand to a series of @code{asm} statements
5169 which will be compiled with GCC@. They go in a library function named
5170 @code{__transfer_from_trampoline}.
5172 If you need to avoid executing the ordinary prologue code of a compiled
5173 C function when you jump to the subroutine, you can do so by placing a
5174 special label of your own in the assembler code. Use one @code{asm}
5175 statement to generate an assembler label, and another to make the label
5176 global. Then trampolines can use that label to jump directly to your
5177 special assembler code.
5181 @section Implicit Calls to Library Routines
5182 @cindex library subroutine names
5183 @cindex @file{libgcc.a}
5185 @c prevent bad page break with this line
5186 Here is an explanation of implicit calls to library routines.
5188 @defmac DECLARE_LIBRARY_RENAMES
5189 This macro, if defined, should expand to a piece of C code that will get
5190 expanded when compiling functions for libgcc.a. It can be used to
5191 provide alternate names for GCC's internal library functions if there
5192 are ABI-mandated names that the compiler should provide.
5195 @findex set_optab_libfunc
5196 @findex init_one_libfunc
5197 @deftypefn {Target Hook} void TARGET_INIT_LIBFUNCS (void)
5198 This hook should declare additional library routines or rename
5199 existing ones, using the functions @code{set_optab_libfunc} and
5200 @code{init_one_libfunc} defined in @file{optabs.c}.
5201 @code{init_optabs} calls this macro after initializing all the normal
5204 The default is to do nothing. Most ports don't need to define this hook.
5207 @deftypevr {Target Hook} bool TARGET_LIBFUNC_GNU_PREFIX
5208 If false (the default), internal library routines start with two
5209 underscores. If set to true, these routines start with @code{__gnu_}
5210 instead. E.g., @code{__muldi3} changes to @code{__gnu_muldi3}. This
5211 currently only affects functions defined in @file{libgcc2.c}. If this
5212 is set to true, the @file{tm.h} file must also
5213 @code{#define LIBGCC2_GNU_PREFIX}.
5216 @defmac FLOAT_LIB_COMPARE_RETURNS_BOOL (@var{mode}, @var{comparison})
5217 This macro should return @code{true} if the library routine that
5218 implements the floating point comparison operator @var{comparison} in
5219 mode @var{mode} will return a boolean, and @var{false} if it will
5222 GCC's own floating point libraries return tristates from the
5223 comparison operators, so the default returns false always. Most ports
5224 don't need to define this macro.
5227 @defmac TARGET_LIB_INT_CMP_BIASED
5228 This macro should evaluate to @code{true} if the integer comparison
5229 functions (like @code{__cmpdi2}) return 0 to indicate that the first
5230 operand is smaller than the second, 1 to indicate that they are equal,
5231 and 2 to indicate that the first operand is greater than the second.
5232 If this macro evaluates to @code{false} the comparison functions return
5233 @minus{}1, 0, and 1 instead of 0, 1, and 2. If the target uses the routines
5234 in @file{libgcc.a}, you do not need to define this macro.
5237 @defmac TARGET_HAS_NO_HW_DIVIDE
5238 This macro should be defined if the target has no hardware divide
5239 instructions. If this macro is defined, GCC will use an algorithm which
5240 make use of simple logical and arithmetic operations for 64-bit
5241 division. If the macro is not defined, GCC will use an algorithm which
5242 make use of a 64-bit by 32-bit divide primitive.
5245 @cindex @code{EDOM}, implicit usage
5248 The value of @code{EDOM} on the target machine, as a C integer constant
5249 expression. If you don't define this macro, GCC does not attempt to
5250 deposit the value of @code{EDOM} into @code{errno} directly. Look in
5251 @file{/usr/include/errno.h} to find the value of @code{EDOM} on your
5254 If you do not define @code{TARGET_EDOM}, then compiled code reports
5255 domain errors by calling the library function and letting it report the
5256 error. If mathematical functions on your system use @code{matherr} when
5257 there is an error, then you should leave @code{TARGET_EDOM} undefined so
5258 that @code{matherr} is used normally.
5261 @cindex @code{errno}, implicit usage
5262 @defmac GEN_ERRNO_RTX
5263 Define this macro as a C expression to create an rtl expression that
5264 refers to the global ``variable'' @code{errno}. (On certain systems,
5265 @code{errno} may not actually be a variable.) If you don't define this
5266 macro, a reasonable default is used.
5269 @deftypefn {Target Hook} bool TARGET_LIBC_HAS_FUNCTION (enum function_class @var{fn_class})
5270 This hook determines whether a function from a class of functions
5271 @var{fn_class} is present at the runtime.
5274 @defmac NEXT_OBJC_RUNTIME
5275 Set this macro to 1 to use the "NeXT" Objective-C message sending conventions
5276 by default. This calling convention involves passing the object, the selector
5277 and the method arguments all at once to the method-lookup library function.
5278 This is the usual setting when targeting Darwin/Mac OS X systems, which have
5279 the NeXT runtime installed.
5281 If the macro is set to 0, the "GNU" Objective-C message sending convention
5282 will be used by default. This convention passes just the object and the
5283 selector to the method-lookup function, which returns a pointer to the method.
5285 In either case, it remains possible to select code-generation for the alternate
5286 scheme, by means of compiler command line switches.
5289 @node Addressing Modes
5290 @section Addressing Modes
5291 @cindex addressing modes
5293 @c prevent bad page break with this line
5294 This is about addressing modes.
5296 @defmac HAVE_PRE_INCREMENT
5297 @defmacx HAVE_PRE_DECREMENT
5298 @defmacx HAVE_POST_INCREMENT
5299 @defmacx HAVE_POST_DECREMENT
5300 A C expression that is nonzero if the machine supports pre-increment,
5301 pre-decrement, post-increment, or post-decrement addressing respectively.
5304 @defmac HAVE_PRE_MODIFY_DISP
5305 @defmacx HAVE_POST_MODIFY_DISP
5306 A C expression that is nonzero if the machine supports pre- or
5307 post-address side-effect generation involving constants other than
5308 the size of the memory operand.
5311 @defmac HAVE_PRE_MODIFY_REG
5312 @defmacx HAVE_POST_MODIFY_REG
5313 A C expression that is nonzero if the machine supports pre- or
5314 post-address side-effect generation involving a register displacement.
5317 @defmac CONSTANT_ADDRESS_P (@var{x})
5318 A C expression that is 1 if the RTX @var{x} is a constant which
5319 is a valid address. On most machines the default definition of
5320 @code{(CONSTANT_P (@var{x}) && GET_CODE (@var{x}) != CONST_DOUBLE)}
5321 is acceptable, but a few machines are more restrictive as to which
5322 constant addresses are supported.
5325 @defmac CONSTANT_P (@var{x})
5326 @code{CONSTANT_P}, which is defined by target-independent code,
5327 accepts integer-values expressions whose values are not explicitly
5328 known, such as @code{symbol_ref}, @code{label_ref}, and @code{high}
5329 expressions and @code{const} arithmetic expressions, in addition to
5330 @code{const_int} and @code{const_double} expressions.
5333 @defmac MAX_REGS_PER_ADDRESS
5334 A number, the maximum number of registers that can appear in a valid
5335 memory address. Note that it is up to you to specify a value equal to
5336 the maximum number that @code{TARGET_LEGITIMATE_ADDRESS_P} would ever
5340 @deftypefn {Target Hook} bool TARGET_LEGITIMATE_ADDRESS_P (machine_mode @var{mode}, rtx @var{x}, bool @var{strict})
5341 A function that returns whether @var{x} (an RTX) is a legitimate memory
5342 address on the target machine for a memory operand of mode @var{mode}.
5344 Legitimate addresses are defined in two variants: a strict variant and a
5345 non-strict one. The @var{strict} parameter chooses which variant is
5346 desired by the caller.
5348 The strict variant is used in the reload pass. It must be defined so
5349 that any pseudo-register that has not been allocated a hard register is
5350 considered a memory reference. This is because in contexts where some
5351 kind of register is required, a pseudo-register with no hard register
5352 must be rejected. For non-hard registers, the strict variant should look
5353 up the @code{reg_renumber} array; it should then proceed using the hard
5354 register number in the array, or treat the pseudo as a memory reference
5355 if the array holds @code{-1}.
5357 The non-strict variant is used in other passes. It must be defined to
5358 accept all pseudo-registers in every context where some kind of
5359 register is required.
5361 Normally, constant addresses which are the sum of a @code{symbol_ref}
5362 and an integer are stored inside a @code{const} RTX to mark them as
5363 constant. Therefore, there is no need to recognize such sums
5364 specifically as legitimate addresses. Normally you would simply
5365 recognize any @code{const} as legitimate.
5367 Usually @code{PRINT_OPERAND_ADDRESS} is not prepared to handle constant
5368 sums that are not marked with @code{const}. It assumes that a naked
5369 @code{plus} indicates indexing. If so, then you @emph{must} reject such
5370 naked constant sums as illegitimate addresses, so that none of them will
5371 be given to @code{PRINT_OPERAND_ADDRESS}.
5373 @cindex @code{TARGET_ENCODE_SECTION_INFO} and address validation
5374 On some machines, whether a symbolic address is legitimate depends on
5375 the section that the address refers to. On these machines, define the
5376 target hook @code{TARGET_ENCODE_SECTION_INFO} to store the information
5377 into the @code{symbol_ref}, and then check for it here. When you see a
5378 @code{const}, you will have to look inside it to find the
5379 @code{symbol_ref} in order to determine the section. @xref{Assembler
5382 @cindex @code{GO_IF_LEGITIMATE_ADDRESS}
5383 Some ports are still using a deprecated legacy substitute for
5384 this hook, the @code{GO_IF_LEGITIMATE_ADDRESS} macro. This macro
5388 #define GO_IF_LEGITIMATE_ADDRESS (@var{mode}, @var{x}, @var{label})
5392 and should @code{goto @var{label}} if the address @var{x} is a valid
5393 address on the target machine for a memory operand of mode @var{mode}.
5395 @findex REG_OK_STRICT
5396 Compiler source files that want to use the strict variant of this
5397 macro define the macro @code{REG_OK_STRICT}. You should use an
5398 @code{#ifdef REG_OK_STRICT} conditional to define the strict variant in
5399 that case and the non-strict variant otherwise.
5401 Using the hook is usually simpler because it limits the number of
5402 files that are recompiled when changes are made.
5405 @defmac TARGET_MEM_CONSTRAINT
5406 A single character to be used instead of the default @code{'m'}
5407 character for general memory addresses. This defines the constraint
5408 letter which matches the memory addresses accepted by
5409 @code{TARGET_LEGITIMATE_ADDRESS_P}. Define this macro if you want to
5410 support new address formats in your back end without changing the
5411 semantics of the @code{'m'} constraint. This is necessary in order to
5412 preserve functionality of inline assembly constructs using the
5413 @code{'m'} constraint.
5416 @defmac FIND_BASE_TERM (@var{x})
5417 A C expression to determine the base term of address @var{x},
5418 or to provide a simplified version of @var{x} from which @file{alias.c}
5419 can easily find the base term. This macro is used in only two places:
5420 @code{find_base_value} and @code{find_base_term} in @file{alias.c}.
5422 It is always safe for this macro to not be defined. It exists so
5423 that alias analysis can understand machine-dependent addresses.
5425 The typical use of this macro is to handle addresses containing
5426 a label_ref or symbol_ref within an UNSPEC@.
5429 @deftypefn {Target Hook} rtx TARGET_LEGITIMIZE_ADDRESS (rtx @var{x}, rtx @var{oldx}, machine_mode @var{mode})
5430 This hook is given an invalid memory address @var{x} for an
5431 operand of mode @var{mode} and should try to return a valid memory
5434 @findex break_out_memory_refs
5435 @var{x} will always be the result of a call to @code{break_out_memory_refs},
5436 and @var{oldx} will be the operand that was given to that function to produce
5439 The code of the hook should not alter the substructure of
5440 @var{x}. If it transforms @var{x} into a more legitimate form, it
5441 should return the new @var{x}.
5443 It is not necessary for this hook to come up with a legitimate address,
5444 with the exception of native TLS addresses (@pxref{Emulated TLS}).
5445 The compiler has standard ways of doing so in all cases. In fact, if
5446 the target supports only emulated TLS, it
5447 is safe to omit this hook or make it return @var{x} if it cannot find
5448 a valid way to legitimize the address. But often a machine-dependent
5449 strategy can generate better code.
5452 @defmac LEGITIMIZE_RELOAD_ADDRESS (@var{x}, @var{mode}, @var{opnum}, @var{type}, @var{ind_levels}, @var{win})
5453 A C compound statement that attempts to replace @var{x}, which is an address
5454 that needs reloading, with a valid memory address for an operand of mode
5455 @var{mode}. @var{win} will be a C statement label elsewhere in the code.
5456 It is not necessary to define this macro, but it might be useful for
5457 performance reasons.
5459 For example, on the i386, it is sometimes possible to use a single
5460 reload register instead of two by reloading a sum of two pseudo
5461 registers into a register. On the other hand, for number of RISC
5462 processors offsets are limited so that often an intermediate address
5463 needs to be generated in order to address a stack slot. By defining
5464 @code{LEGITIMIZE_RELOAD_ADDRESS} appropriately, the intermediate addresses
5465 generated for adjacent some stack slots can be made identical, and thus
5468 @emph{Note}: This macro should be used with caution. It is necessary
5469 to know something of how reload works in order to effectively use this,
5470 and it is quite easy to produce macros that build in too much knowledge
5471 of reload internals.
5473 @emph{Note}: This macro must be able to reload an address created by a
5474 previous invocation of this macro. If it fails to handle such addresses
5475 then the compiler may generate incorrect code or abort.
5478 The macro definition should use @code{push_reload} to indicate parts that
5479 need reloading; @var{opnum}, @var{type} and @var{ind_levels} are usually
5480 suitable to be passed unaltered to @code{push_reload}.
5482 The code generated by this macro must not alter the substructure of
5483 @var{x}. If it transforms @var{x} into a more legitimate form, it
5484 should assign @var{x} (which will always be a C variable) a new value.
5485 This also applies to parts that you change indirectly by calling
5488 @findex strict_memory_address_p
5489 The macro definition may use @code{strict_memory_address_p} to test if
5490 the address has become legitimate.
5493 If you want to change only a part of @var{x}, one standard way of doing
5494 this is to use @code{copy_rtx}. Note, however, that it unshares only a
5495 single level of rtl. Thus, if the part to be changed is not at the
5496 top level, you'll need to replace first the top level.
5497 It is not necessary for this macro to come up with a legitimate
5498 address; but often a machine-dependent strategy can generate better code.
5501 @deftypefn {Target Hook} bool TARGET_MODE_DEPENDENT_ADDRESS_P (const_rtx @var{addr}, addr_space_t @var{addrspace})
5502 This hook returns @code{true} if memory address @var{addr} in address
5503 space @var{addrspace} can have
5504 different meanings depending on the machine mode of the memory
5505 reference it is used for or if the address is valid for some modes
5508 Autoincrement and autodecrement addresses typically have mode-dependent
5509 effects because the amount of the increment or decrement is the size
5510 of the operand being addressed. Some machines have other mode-dependent
5511 addresses. Many RISC machines have no mode-dependent addresses.
5513 You may assume that @var{addr} is a valid address for the machine.
5515 The default version of this hook returns @code{false}.
5518 @deftypefn {Target Hook} bool TARGET_LEGITIMATE_CONSTANT_P (machine_mode @var{mode}, rtx @var{x})
5519 This hook returns true if @var{x} is a legitimate constant for a
5520 @var{mode}-mode immediate operand on the target machine. You can assume that
5521 @var{x} satisfies @code{CONSTANT_P}, so you need not check this.
5523 The default definition returns true.
5526 @deftypefn {Target Hook} rtx TARGET_DELEGITIMIZE_ADDRESS (rtx @var{x})
5527 This hook is used to undo the possibly obfuscating effects of the
5528 @code{LEGITIMIZE_ADDRESS} and @code{LEGITIMIZE_RELOAD_ADDRESS} target
5529 macros. Some backend implementations of these macros wrap symbol
5530 references inside an @code{UNSPEC} rtx to represent PIC or similar
5531 addressing modes. This target hook allows GCC's optimizers to understand
5532 the semantics of these opaque @code{UNSPEC}s by converting them back
5533 into their original form.
5536 @deftypefn {Target Hook} bool TARGET_CONST_NOT_OK_FOR_DEBUG_P (rtx @var{x})
5537 This hook should return true if @var{x} should not be emitted into
5541 @deftypefn {Target Hook} bool TARGET_CANNOT_FORCE_CONST_MEM (machine_mode @var{mode}, rtx @var{x})
5542 This hook should return true if @var{x} is of a form that cannot (or
5543 should not) be spilled to the constant pool. @var{mode} is the mode
5546 The default version of this hook returns false.
5548 The primary reason to define this hook is to prevent reload from
5549 deciding that a non-legitimate constant would be better reloaded
5550 from the constant pool instead of spilling and reloading a register
5551 holding the constant. This restriction is often true of addresses
5552 of TLS symbols for various targets.
5555 @deftypefn {Target Hook} bool TARGET_USE_BLOCKS_FOR_CONSTANT_P (machine_mode @var{mode}, const_rtx @var{x})
5556 This hook should return true if pool entries for constant @var{x} can
5557 be placed in an @code{object_block} structure. @var{mode} is the mode
5560 The default version returns false for all constants.
5563 @deftypefn {Target Hook} bool TARGET_USE_BLOCKS_FOR_DECL_P (const_tree @var{decl})
5564 This hook should return true if pool entries for @var{decl} should
5565 be placed in an @code{object_block} structure.
5567 The default version returns true for all decls.
5570 @deftypefn {Target Hook} tree TARGET_BUILTIN_RECIPROCAL (unsigned @var{fn}, bool @var{md_fn}, bool @var{sqrt})
5571 This hook should return the DECL of a function that implements reciprocal of
5572 the builtin function with builtin function code @var{fn}, or
5573 @code{NULL_TREE} if such a function is not available. @var{md_fn} is true
5574 when @var{fn} is a code of a machine-dependent builtin function. When
5575 @var{sqrt} is true, additional optimizations that apply only to the reciprocal
5576 of a square root function are performed, and only reciprocals of @code{sqrt}
5580 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_MASK_FOR_LOAD (void)
5581 This hook should return the DECL of a function @var{f} that given an
5582 address @var{addr} as an argument returns a mask @var{m} that can be
5583 used to extract from two vectors the relevant data that resides in
5584 @var{addr} in case @var{addr} is not properly aligned.
5586 The autovectorizer, when vectorizing a load operation from an address
5587 @var{addr} that may be unaligned, will generate two vector loads from
5588 the two aligned addresses around @var{addr}. It then generates a
5589 @code{REALIGN_LOAD} operation to extract the relevant data from the
5590 two loaded vectors. The first two arguments to @code{REALIGN_LOAD},
5591 @var{v1} and @var{v2}, are the two vectors, each of size @var{VS}, and
5592 the third argument, @var{OFF}, defines how the data will be extracted
5593 from these two vectors: if @var{OFF} is 0, then the returned vector is
5594 @var{v2}; otherwise, the returned vector is composed from the last
5595 @var{VS}-@var{OFF} elements of @var{v1} concatenated to the first
5596 @var{OFF} elements of @var{v2}.
5598 If this hook is defined, the autovectorizer will generate a call
5599 to @var{f} (using the DECL tree that this hook returns) and will
5600 use the return value of @var{f} as the argument @var{OFF} to
5601 @code{REALIGN_LOAD}. Therefore, the mask @var{m} returned by @var{f}
5602 should comply with the semantics expected by @code{REALIGN_LOAD}
5604 If this hook is not defined, then @var{addr} will be used as
5605 the argument @var{OFF} to @code{REALIGN_LOAD}, in which case the low
5606 log2(@var{VS}) @minus{} 1 bits of @var{addr} will be considered.
5609 @deftypefn {Target Hook} int TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST (enum vect_cost_for_stmt @var{type_of_cost}, tree @var{vectype}, int @var{misalign})
5610 Returns cost of different scalar or vector statements for vectorization cost model.
5611 For vector memory operations the cost may depend on type (@var{vectype}) and
5612 misalignment value (@var{misalign}).
5615 @deftypefn {Target Hook} bool TARGET_VECTORIZE_VECTOR_ALIGNMENT_REACHABLE (const_tree @var{type}, bool @var{is_packed})
5616 Return true if vector alignment is reachable (by peeling N iterations) for the given type.
5619 @deftypefn {Target Hook} bool TARGET_VECTORIZE_VEC_PERM_CONST_OK (machine_mode, const unsigned char *@var{sel})
5620 Return true if a vector created for @code{vec_perm_const} is valid.
5623 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_CONVERSION (unsigned @var{code}, tree @var{dest_type}, tree @var{src_type})
5624 This hook should return the DECL of a function that implements conversion of the
5625 input vector of type @var{src_type} to type @var{dest_type}.
5626 The value of @var{code} is one of the enumerators in @code{enum tree_code} and
5627 specifies how the conversion is to be applied
5628 (truncation, rounding, etc.).
5630 If this hook is defined, the autovectorizer will use the
5631 @code{TARGET_VECTORIZE_BUILTIN_CONVERSION} target hook when vectorizing
5632 conversion. Otherwise, it will return @code{NULL_TREE}.
5635 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION (tree @var{fndecl}, tree @var{vec_type_out}, tree @var{vec_type_in})
5636 This hook should return the decl of a function that implements the
5637 vectorized variant of the builtin function with builtin function code
5638 @var{code} or @code{NULL_TREE} if such a function is not available.
5639 The value of @var{fndecl} is the builtin function declaration. The
5640 return type of the vectorized function shall be of vector type
5641 @var{vec_type_out} and the argument types should be @var{vec_type_in}.
5644 @deftypefn {Target Hook} bool TARGET_VECTORIZE_SUPPORT_VECTOR_MISALIGNMENT (machine_mode @var{mode}, const_tree @var{type}, int @var{misalignment}, bool @var{is_packed})
5645 This hook should return true if the target supports misaligned vector
5646 store/load of a specific factor denoted in the @var{misalignment}
5647 parameter. The vector store/load should be of machine mode @var{mode} and
5648 the elements in the vectors should be of type @var{type}. @var{is_packed}
5649 parameter is true if the memory access is defined in a packed struct.
5652 @deftypefn {Target Hook} machine_mode TARGET_VECTORIZE_PREFERRED_SIMD_MODE (machine_mode @var{mode})
5653 This hook should return the preferred mode for vectorizing scalar
5654 mode @var{mode}. The default is
5655 equal to @code{word_mode}, because the vectorizer can do some
5656 transformations even in absence of specialized @acronym{SIMD} hardware.
5659 @deftypefn {Target Hook} {unsigned int} TARGET_VECTORIZE_AUTOVECTORIZE_VECTOR_SIZES (void)
5660 This hook should return a mask of sizes that should be iterated over
5661 after trying to autovectorize using the vector size derived from the
5662 mode returned by @code{TARGET_VECTORIZE_PREFERRED_SIMD_MODE}.
5663 The default is zero which means to not iterate over other vector sizes.
5666 @deftypefn {Target Hook} {void *} TARGET_VECTORIZE_INIT_COST (struct loop *@var{loop_info})
5667 This hook should initialize target-specific data structures in preparation for modeling the costs of vectorizing a loop or basic block. The default allocates three unsigned integers for accumulating costs for the prologue, body, and epilogue of the loop or basic block. If @var{loop_info} is non-NULL, it identifies the loop being vectorized; otherwise a single block is being vectorized.
5670 @deftypefn {Target Hook} unsigned TARGET_VECTORIZE_ADD_STMT_COST (void *@var{data}, int @var{count}, enum vect_cost_for_stmt @var{kind}, struct _stmt_vec_info *@var{stmt_info}, int @var{misalign}, enum vect_cost_model_location @var{where})
5671 This hook should update the target-specific @var{data} in response to adding @var{count} copies of the given @var{kind} of statement to a loop or basic block. The default adds the builtin vectorizer cost for the copies of the statement to the accumulator specified by @var{where}, (the prologue, body, or epilogue) and returns the amount added. The return value should be viewed as a tentative cost that may later be revised.
5674 @deftypefn {Target Hook} void TARGET_VECTORIZE_FINISH_COST (void *@var{data}, unsigned *@var{prologue_cost}, unsigned *@var{body_cost}, unsigned *@var{epilogue_cost})
5675 This hook should complete calculations of the cost of vectorizing a loop or basic block based on @var{data}, and return the prologue, body, and epilogue costs as unsigned integers. The default returns the value of the three accumulators.
5678 @deftypefn {Target Hook} void TARGET_VECTORIZE_DESTROY_COST_DATA (void *@var{data})
5679 This hook should release @var{data} and any related data structures allocated by TARGET_VECTORIZE_INIT_COST. The default releases the accumulator.
5682 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_TM_LOAD (tree)
5683 This hook should return the built-in decl needed to load a vector of the given type within a transaction.
5686 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_TM_STORE (tree)
5687 This hook should return the built-in decl needed to store a vector of the given type within a transaction.
5690 @deftypefn {Target Hook} tree TARGET_VECTORIZE_BUILTIN_GATHER (const_tree @var{mem_vectype}, const_tree @var{index_type}, int @var{scale})
5691 Target builtin that implements vector gather operation. @var{mem_vectype}
5692 is the vector type of the load and @var{index_type} is scalar type of
5693 the index, scaled by @var{scale}.
5694 The default is @code{NULL_TREE} which means to not vectorize gather
5698 @deftypefn {Target Hook} int TARGET_SIMD_CLONE_COMPUTE_VECSIZE_AND_SIMDLEN (struct cgraph_node *@var{}, struct cgraph_simd_clone *@var{}, @var{tree}, @var{int})
5699 This hook should set @var{vecsize_mangle}, @var{vecsize_int}, @var{vecsize_float}
5700 fields in @var{simd_clone} structure pointed by @var{clone_info} argument and also
5701 @var{simdlen} field if it was previously 0.
5702 The hook should return 0 if SIMD clones shouldn't be emitted,
5703 or number of @var{vecsize_mangle} variants that should be emitted.
5706 @deftypefn {Target Hook} void TARGET_SIMD_CLONE_ADJUST (struct cgraph_node *@var{})
5707 This hook should add implicit @code{attribute(target("..."))} attribute
5708 to SIMD clone @var{node} if needed.
5711 @deftypefn {Target Hook} int TARGET_SIMD_CLONE_USABLE (struct cgraph_node *@var{})
5712 This hook should return -1 if SIMD clone @var{node} shouldn't be used
5713 in vectorized loops in current function, or non-negative number if it is
5714 usable. In that case, the smaller the number is, the more desirable it is
5718 @node Anchored Addresses
5719 @section Anchored Addresses
5720 @cindex anchored addresses
5721 @cindex @option{-fsection-anchors}
5723 GCC usually addresses every static object as a separate entity.
5724 For example, if we have:
5728 int foo (void) @{ return a + b + c; @}
5731 the code for @code{foo} will usually calculate three separate symbolic
5732 addresses: those of @code{a}, @code{b} and @code{c}. On some targets,
5733 it would be better to calculate just one symbolic address and access
5734 the three variables relative to it. The equivalent pseudocode would
5740 register int *xr = &x;
5741 return xr[&a - &x] + xr[&b - &x] + xr[&c - &x];
5745 (which isn't valid C). We refer to shared addresses like @code{x} as
5746 ``section anchors''. Their use is controlled by @option{-fsection-anchors}.
5748 The hooks below describe the target properties that GCC needs to know
5749 in order to make effective use of section anchors. It won't use
5750 section anchors at all unless either @code{TARGET_MIN_ANCHOR_OFFSET}
5751 or @code{TARGET_MAX_ANCHOR_OFFSET} is set to a nonzero value.
5753 @deftypevr {Target Hook} HOST_WIDE_INT TARGET_MIN_ANCHOR_OFFSET
5754 The minimum offset that should be applied to a section anchor.
5755 On most targets, it should be the smallest offset that can be
5756 applied to a base register while still giving a legitimate address
5757 for every mode. The default value is 0.
5760 @deftypevr {Target Hook} HOST_WIDE_INT TARGET_MAX_ANCHOR_OFFSET
5761 Like @code{TARGET_MIN_ANCHOR_OFFSET}, but the maximum (inclusive)
5762 offset that should be applied to section anchors. The default
5766 @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_ANCHOR (rtx @var{x})
5767 Write the assembly code to define section anchor @var{x}, which is a
5768 @code{SYMBOL_REF} for which @samp{SYMBOL_REF_ANCHOR_P (@var{x})} is true.
5769 The hook is called with the assembly output position set to the beginning
5770 of @code{SYMBOL_REF_BLOCK (@var{x})}.
5772 If @code{ASM_OUTPUT_DEF} is available, the hook's default definition uses
5773 it to define the symbol as @samp{. + SYMBOL_REF_BLOCK_OFFSET (@var{x})}.
5774 If @code{ASM_OUTPUT_DEF} is not available, the hook's default definition
5775 is @code{NULL}, which disables the use of section anchors altogether.
5778 @deftypefn {Target Hook} bool TARGET_USE_ANCHORS_FOR_SYMBOL_P (const_rtx @var{x})
5779 Return true if GCC should attempt to use anchors to access @code{SYMBOL_REF}
5780 @var{x}. You can assume @samp{SYMBOL_REF_HAS_BLOCK_INFO_P (@var{x})} and
5781 @samp{!SYMBOL_REF_ANCHOR_P (@var{x})}.
5783 The default version is correct for most targets, but you might need to
5784 intercept this hook to handle things like target-specific attributes
5785 or target-specific sections.
5788 @node Condition Code
5789 @section Condition Code Status
5790 @cindex condition code status
5792 The macros in this section can be split in two families, according to the
5793 two ways of representing condition codes in GCC.
5795 The first representation is the so called @code{(cc0)} representation
5796 (@pxref{Jump Patterns}), where all instructions can have an implicit
5797 clobber of the condition codes. The second is the condition code
5798 register representation, which provides better schedulability for
5799 architectures that do have a condition code register, but on which
5800 most instructions do not affect it. The latter category includes
5803 The implicit clobbering poses a strong restriction on the placement of
5804 the definition and use of the condition code. In the past the definition
5805 and use were always adjacent. However, recent changes to support trapping
5806 arithmatic may result in the definition and user being in different blocks.
5807 Thus, there may be a @code{NOTE_INSN_BASIC_BLOCK} between them. Additionally,
5808 the definition may be the source of exception handling edges.
5810 These restrictions can prevent important
5811 optimizations on some machines. For example, on the IBM RS/6000, there
5812 is a delay for taken branches unless the condition code register is set
5813 three instructions earlier than the conditional branch. The instruction
5814 scheduler cannot perform this optimization if it is not permitted to
5815 separate the definition and use of the condition code register.
5817 For this reason, it is possible and suggested to use a register to
5818 represent the condition code for new ports. If there is a specific
5819 condition code register in the machine, use a hard register. If the
5820 condition code or comparison result can be placed in any general register,
5821 or if there are multiple condition registers, use a pseudo register.
5822 Registers used to store the condition code value will usually have a mode
5823 that is in class @code{MODE_CC}.
5825 Alternatively, you can use @code{BImode} if the comparison operator is
5826 specified already in the compare instruction. In this case, you are not
5827 interested in most macros in this section.
5830 * CC0 Condition Codes:: Old style representation of condition codes.
5831 * MODE_CC Condition Codes:: Modern representation of condition codes.
5834 @node CC0 Condition Codes
5835 @subsection Representation of condition codes using @code{(cc0)}
5839 The file @file{conditions.h} defines a variable @code{cc_status} to
5840 describe how the condition code was computed (in case the interpretation of
5841 the condition code depends on the instruction that it was set by). This
5842 variable contains the RTL expressions on which the condition code is
5843 currently based, and several standard flags.
5845 Sometimes additional machine-specific flags must be defined in the machine
5846 description header file. It can also add additional machine-specific
5847 information by defining @code{CC_STATUS_MDEP}.
5849 @defmac CC_STATUS_MDEP
5850 C code for a data type which is used for declaring the @code{mdep}
5851 component of @code{cc_status}. It defaults to @code{int}.
5853 This macro is not used on machines that do not use @code{cc0}.
5856 @defmac CC_STATUS_MDEP_INIT
5857 A C expression to initialize the @code{mdep} field to ``empty''.
5858 The default definition does nothing, since most machines don't use
5859 the field anyway. If you want to use the field, you should probably
5860 define this macro to initialize it.
5862 This macro is not used on machines that do not use @code{cc0}.
5865 @defmac NOTICE_UPDATE_CC (@var{exp}, @var{insn})
5866 A C compound statement to set the components of @code{cc_status}
5867 appropriately for an insn @var{insn} whose body is @var{exp}. It is
5868 this macro's responsibility to recognize insns that set the condition
5869 code as a byproduct of other activity as well as those that explicitly
5872 This macro is not used on machines that do not use @code{cc0}.
5874 If there are insns that do not set the condition code but do alter
5875 other machine registers, this macro must check to see whether they
5876 invalidate the expressions that the condition code is recorded as
5877 reflecting. For example, on the 68000, insns that store in address
5878 registers do not set the condition code, which means that usually
5879 @code{NOTICE_UPDATE_CC} can leave @code{cc_status} unaltered for such
5880 insns. But suppose that the previous insn set the condition code
5881 based on location @samp{a4@@(102)} and the current insn stores a new
5882 value in @samp{a4}. Although the condition code is not changed by
5883 this, it will no longer be true that it reflects the contents of
5884 @samp{a4@@(102)}. Therefore, @code{NOTICE_UPDATE_CC} must alter
5885 @code{cc_status} in this case to say that nothing is known about the
5886 condition code value.
5888 The definition of @code{NOTICE_UPDATE_CC} must be prepared to deal
5889 with the results of peephole optimization: insns whose patterns are
5890 @code{parallel} RTXs containing various @code{reg}, @code{mem} or
5891 constants which are just the operands. The RTL structure of these
5892 insns is not sufficient to indicate what the insns actually do. What
5893 @code{NOTICE_UPDATE_CC} should do when it sees one is just to run
5894 @code{CC_STATUS_INIT}.
5896 A possible definition of @code{NOTICE_UPDATE_CC} is to call a function
5897 that looks at an attribute (@pxref{Insn Attributes}) named, for example,
5898 @samp{cc}. This avoids having detailed information about patterns in
5899 two places, the @file{md} file and in @code{NOTICE_UPDATE_CC}.
5902 @node MODE_CC Condition Codes
5903 @subsection Representation of condition codes using registers
5907 @defmac SELECT_CC_MODE (@var{op}, @var{x}, @var{y})
5908 On many machines, the condition code may be produced by other instructions
5909 than compares, for example the branch can use directly the condition
5910 code set by a subtract instruction. However, on some machines
5911 when the condition code is set this way some bits (such as the overflow
5912 bit) are not set in the same way as a test instruction, so that a different
5913 branch instruction must be used for some conditional branches. When
5914 this happens, use the machine mode of the condition code register to
5915 record different formats of the condition code register. Modes can
5916 also be used to record which compare instruction (e.g. a signed or an
5917 unsigned comparison) produced the condition codes.
5919 If other modes than @code{CCmode} are required, add them to
5920 @file{@var{machine}-modes.def} and define @code{SELECT_CC_MODE} to choose
5921 a mode given an operand of a compare. This is needed because the modes
5922 have to be chosen not only during RTL generation but also, for example,
5923 by instruction combination. The result of @code{SELECT_CC_MODE} should
5924 be consistent with the mode used in the patterns; for example to support
5925 the case of the add on the SPARC discussed above, we have the pattern
5929 [(set (reg:CC_NOOV 0)
5931 (plus:SI (match_operand:SI 0 "register_operand" "%r")
5932 (match_operand:SI 1 "arith_operand" "rI"))
5939 together with a @code{SELECT_CC_MODE} that returns @code{CC_NOOVmode}
5940 for comparisons whose argument is a @code{plus}:
5943 #define SELECT_CC_MODE(OP,X,Y) \
5944 (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \
5945 ? ((OP == LT || OP == LE || OP == GT || OP == GE) \
5946 ? CCFPEmode : CCFPmode) \
5947 : ((GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \
5948 || GET_CODE (X) == NEG || GET_CODE (x) == ASHIFT) \
5949 ? CC_NOOVmode : CCmode))
5952 Another reason to use modes is to retain information on which operands
5953 were used by the comparison; see @code{REVERSIBLE_CC_MODE} later in
5956 You should define this macro if and only if you define extra CC modes
5957 in @file{@var{machine}-modes.def}.
5960 @deftypefn {Target Hook} void TARGET_CANONICALIZE_COMPARISON (int *@var{code}, rtx *@var{op0}, rtx *@var{op1}, bool @var{op0_preserve_value})
5961 On some machines not all possible comparisons are defined, but you can
5962 convert an invalid comparison into a valid one. For example, the Alpha
5963 does not have a @code{GT} comparison, but you can use an @code{LT}
5964 comparison instead and swap the order of the operands.
5966 On such machines, implement this hook to do any required conversions.
5967 @var{code} is the initial comparison code and @var{op0} and @var{op1}
5968 are the left and right operands of the comparison, respectively. If
5969 @var{op0_preserve_value} is @code{true} the implementation is not
5970 allowed to change the value of @var{op0} since the value might be used
5971 in RTXs which aren't comparisons. E.g. the implementation is not
5972 allowed to swap operands in that case.
5974 GCC will not assume that the comparison resulting from this macro is
5975 valid but will see if the resulting insn matches a pattern in the
5978 You need not to implement this hook if it would never change the
5979 comparison code or operands.
5982 @defmac REVERSIBLE_CC_MODE (@var{mode})
5983 A C expression whose value is one if it is always safe to reverse a
5984 comparison whose mode is @var{mode}. If @code{SELECT_CC_MODE}
5985 can ever return @var{mode} for a floating-point inequality comparison,
5986 then @code{REVERSIBLE_CC_MODE (@var{mode})} must be zero.
5988 You need not define this macro if it would always returns zero or if the
5989 floating-point format is anything other than @code{IEEE_FLOAT_FORMAT}.
5990 For example, here is the definition used on the SPARC, where floating-point
5991 inequality comparisons are given either @code{CCFPEmode} or @code{CCFPmode}:
5994 #define REVERSIBLE_CC_MODE(MODE) \
5995 ((MODE) != CCFPEmode && (MODE) != CCFPmode)
5999 @defmac REVERSE_CONDITION (@var{code}, @var{mode})
6000 A C expression whose value is reversed condition code of the @var{code} for
6001 comparison done in CC_MODE @var{mode}. The macro is used only in case
6002 @code{REVERSIBLE_CC_MODE (@var{mode})} is nonzero. Define this macro in case
6003 machine has some non-standard way how to reverse certain conditionals. For
6004 instance in case all floating point conditions are non-trapping, compiler may
6005 freely convert unordered compares to ordered ones. Then definition may look
6009 #define REVERSE_CONDITION(CODE, MODE) \
6010 ((MODE) != CCFPmode ? reverse_condition (CODE) \
6011 : reverse_condition_maybe_unordered (CODE))
6015 @deftypefn {Target Hook} bool TARGET_FIXED_CONDITION_CODE_REGS (unsigned int *@var{p1}, unsigned int *@var{p2})
6016 On targets which do not use @code{(cc0)}, and which use a hard
6017 register rather than a pseudo-register to hold condition codes, the
6018 regular CSE passes are often not able to identify cases in which the
6019 hard register is set to a common value. Use this hook to enable a
6020 small pass which optimizes such cases. This hook should return true
6021 to enable this pass, and it should set the integers to which its
6022 arguments point to the hard register numbers used for condition codes.
6023 When there is only one such register, as is true on most systems, the
6024 integer pointed to by @var{p2} should be set to
6025 @code{INVALID_REGNUM}.
6027 The default version of this hook returns false.
6030 @deftypefn {Target Hook} machine_mode TARGET_CC_MODES_COMPATIBLE (machine_mode @var{m1}, machine_mode @var{m2})
6031 On targets which use multiple condition code modes in class
6032 @code{MODE_CC}, it is sometimes the case that a comparison can be
6033 validly done in more than one mode. On such a system, define this
6034 target hook to take two mode arguments and to return a mode in which
6035 both comparisons may be validly done. If there is no such mode,
6036 return @code{VOIDmode}.
6038 The default version of this hook checks whether the modes are the
6039 same. If they are, it returns that mode. If they are different, it
6040 returns @code{VOIDmode}.
6044 @section Describing Relative Costs of Operations
6045 @cindex costs of instructions
6046 @cindex relative costs
6047 @cindex speed of instructions
6049 These macros let you describe the relative speed of various operations
6050 on the target machine.
6052 @defmac REGISTER_MOVE_COST (@var{mode}, @var{from}, @var{to})
6053 A C expression for the cost of moving data of mode @var{mode} from a
6054 register in class @var{from} to one in class @var{to}. The classes are
6055 expressed using the enumeration values such as @code{GENERAL_REGS}. A
6056 value of 2 is the default; other values are interpreted relative to
6059 It is not required that the cost always equal 2 when @var{from} is the
6060 same as @var{to}; on some machines it is expensive to move between
6061 registers if they are not general registers.
6063 If reload sees an insn consisting of a single @code{set} between two
6064 hard registers, and if @code{REGISTER_MOVE_COST} applied to their
6065 classes returns a value of 2, reload does not check to ensure that the
6066 constraints of the insn are met. Setting a cost of other than 2 will
6067 allow reload to verify that the constraints are met. You should do this
6068 if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
6070 These macros are obsolete, new ports should use the target hook
6071 @code{TARGET_REGISTER_MOVE_COST} instead.
6074 @deftypefn {Target Hook} int TARGET_REGISTER_MOVE_COST (machine_mode @var{mode}, reg_class_t @var{from}, reg_class_t @var{to})
6075 This target hook should return the cost of moving data of mode @var{mode}
6076 from a register in class @var{from} to one in class @var{to}. The classes
6077 are expressed using the enumeration values such as @code{GENERAL_REGS}.
6078 A value of 2 is the default; other values are interpreted relative to
6081 It is not required that the cost always equal 2 when @var{from} is the
6082 same as @var{to}; on some machines it is expensive to move between
6083 registers if they are not general registers.
6085 If reload sees an insn consisting of a single @code{set} between two
6086 hard registers, and if @code{TARGET_REGISTER_MOVE_COST} applied to their
6087 classes returns a value of 2, reload does not check to ensure that the
6088 constraints of the insn are met. Setting a cost of other than 2 will
6089 allow reload to verify that the constraints are met. You should do this
6090 if the @samp{mov@var{m}} pattern's constraints do not allow such copying.
6092 The default version of this function returns 2.
6095 @defmac MEMORY_MOVE_COST (@var{mode}, @var{class}, @var{in})
6096 A C expression for the cost of moving data of mode @var{mode} between a
6097 register of class @var{class} and memory; @var{in} is zero if the value
6098 is to be written to memory, nonzero if it is to be read in. This cost
6099 is relative to those in @code{REGISTER_MOVE_COST}. If moving between
6100 registers and memory is more expensive than between two registers, you
6101 should define this macro to express the relative cost.
6103 If you do not define this macro, GCC uses a default cost of 4 plus
6104 the cost of copying via a secondary reload register, if one is
6105 needed. If your machine requires a secondary reload register to copy
6106 between memory and a register of @var{class} but the reload mechanism is
6107 more complex than copying via an intermediate, define this macro to
6108 reflect the actual cost of the move.
6110 GCC defines the function @code{memory_move_secondary_cost} if
6111 secondary reloads are needed. It computes the costs due to copying via
6112 a secondary register. If your machine copies from memory using a
6113 secondary register in the conventional way but the default base value of
6114 4 is not correct for your machine, define this macro to add some other
6115 value to the result of that function. The arguments to that function
6116 are the same as to this macro.
6118 These macros are obsolete, new ports should use the target hook
6119 @code{TARGET_MEMORY_MOVE_COST} instead.
6122 @deftypefn {Target Hook} int TARGET_MEMORY_MOVE_COST (machine_mode @var{mode}, reg_class_t @var{rclass}, bool @var{in})
6123 This target hook should return the cost of moving data of mode @var{mode}
6124 between a register of class @var{rclass} and memory; @var{in} is @code{false}
6125 if the value is to be written to memory, @code{true} if it is to be read in.
6126 This cost is relative to those in @code{TARGET_REGISTER_MOVE_COST}.
6127 If moving between registers and memory is more expensive than between two
6128 registers, you should add this target hook to express the relative cost.
6130 If you do not add this target hook, 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{rclass} but the reload mechanism is
6134 more complex than copying via an intermediate, use this target hook 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, use this target hook to add some other
6142 value to the result of that function. The arguments to that function
6143 are the same as to this target hook.
6146 @defmac BRANCH_COST (@var{speed_p}, @var{predictable_p})
6147 A C expression for the cost of a branch instruction. A value of 1 is
6148 the default; other values are interpreted relative to that. Parameter
6149 @var{speed_p} is true when the branch in question should be optimized
6150 for speed. When it is false, @code{BRANCH_COST} should return a value
6151 optimal for code size rather than performance. @var{predictable_p} is
6152 true for well-predicted branches. On many architectures the
6153 @code{BRANCH_COST} can be reduced then.
6156 Here are additional macros which do not specify precise relative costs,
6157 but only that certain actions are more expensive than GCC would
6160 @defmac SLOW_BYTE_ACCESS
6161 Define this macro as a C expression which is nonzero if accessing less
6162 than a word of memory (i.e.@: a @code{char} or a @code{short}) is no
6163 faster than accessing a word of memory, i.e., if such access
6164 require more than one instruction or if there is no difference in cost
6165 between byte and (aligned) word loads.
6167 When this macro is not defined, the compiler will access a field by
6168 finding the smallest containing object; when it is defined, a fullword
6169 load will be used if alignment permits. Unless bytes accesses are
6170 faster than word accesses, using word accesses is preferable since it
6171 may eliminate subsequent memory access if subsequent accesses occur to
6172 other fields in the same word of the structure, but to different bytes.
6175 @defmac SLOW_UNALIGNED_ACCESS (@var{mode}, @var{alignment})
6176 Define this macro to be the value 1 if memory accesses described by the
6177 @var{mode} and @var{alignment} parameters have a cost many times greater
6178 than aligned accesses, for example if they are emulated in a trap
6181 When this macro is nonzero, the compiler will act as if
6182 @code{STRICT_ALIGNMENT} were nonzero when generating code for block
6183 moves. This can cause significantly more instructions to be produced.
6184 Therefore, do not set this macro nonzero if unaligned accesses only add a
6185 cycle or two to the time for a memory access.
6187 If the value of this macro is always zero, it need not be defined. If
6188 this macro is defined, it should produce a nonzero value when
6189 @code{STRICT_ALIGNMENT} is nonzero.
6192 @defmac MOVE_RATIO (@var{speed})
6193 The threshold of number of scalar memory-to-memory move insns, @emph{below}
6194 which a sequence of insns should be generated instead of a
6195 string move insn or a library call. Increasing the value will always
6196 make code faster, but eventually incurs high cost in increased code size.
6198 Note that on machines where the corresponding move insn is a
6199 @code{define_expand} that emits a sequence of insns, this macro counts
6200 the number of such sequences.
6202 The parameter @var{speed} is true if the code is currently being
6203 optimized for speed rather than size.
6205 If you don't define this, a reasonable default is used.
6208 @deftypefn {Target Hook} bool TARGET_USE_BY_PIECES_INFRASTRUCTURE_P (unsigned int @var{size}, unsigned int @var{alignment}, enum by_pieces_operation @var{op}, bool @var{speed_p})
6209 GCC will attempt several strategies when asked to copy between
6210 two areas of memory, or to set, clear or store to memory, for example
6211 when copying a @code{struct}. The @code{by_pieces} infrastructure
6212 implements such memory operations as a sequence of load, store or move
6213 insns. Alternate strategies are to expand the
6214 @code{movmem} or @code{setmem} optabs, to emit a library call, or to emit
6215 unit-by-unit, loop-based operations.
6217 This target hook should return true if, for a memory operation with a
6218 given @var{size} and @var{alignment}, using the @code{by_pieces}
6219 infrastructure is expected to result in better code generation.
6220 Both @var{size} and @var{alignment} are measured in terms of storage
6223 The parameter @var{op} is one of: @code{CLEAR_BY_PIECES},
6224 @code{MOVE_BY_PIECES}, @code{SET_BY_PIECES}, @code{STORE_BY_PIECES}.
6225 These describe the type of memory operation under consideration.
6227 The parameter @var{speed_p} is true if the code is currently being
6228 optimized for speed rather than size.
6230 Returning true for higher values of @var{size} can improve code generation
6231 for speed if the target does not provide an implementation of the
6232 @code{movmem} or @code{setmem} standard names, if the @code{movmem} or
6233 @code{setmem} implementation would be more expensive than a sequence of
6234 insns, or if the overhead of a library call would dominate that of
6235 the body of the memory operation.
6237 Returning true for higher values of @code{size} may also cause an increase
6238 in code size, for example where the number of insns emitted to perform a
6239 move would be greater than that of a library call.
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 SET_RATIO (@var{speed})
6260 The threshold of number of scalar move insns, @emph{below} which a sequence
6261 of insns should be generated to set memory to a constant value, instead of
6262 a block set insn or a library call.
6263 Increasing the value will always make code faster, but
6264 eventually incurs high cost in increased code size.
6266 The parameter @var{speed} is true if the code is currently being
6267 optimized for speed rather than size.
6269 If you don't define this, it defaults to the value of @code{MOVE_RATIO}.
6272 @defmac USE_LOAD_POST_INCREMENT (@var{mode})
6273 A C expression used to determine whether a load postincrement is a good
6274 thing to use for a given mode. Defaults to the value of
6275 @code{HAVE_POST_INCREMENT}.
6278 @defmac USE_LOAD_POST_DECREMENT (@var{mode})
6279 A C expression used to determine whether a load postdecrement is a good
6280 thing to use for a given mode. Defaults to the value of
6281 @code{HAVE_POST_DECREMENT}.
6284 @defmac USE_LOAD_PRE_INCREMENT (@var{mode})
6285 A C expression used to determine whether a load preincrement is a good
6286 thing to use for a given mode. Defaults to the value of
6287 @code{HAVE_PRE_INCREMENT}.
6290 @defmac USE_LOAD_PRE_DECREMENT (@var{mode})
6291 A C expression used to determine whether a load predecrement is a good
6292 thing to use for a given mode. Defaults to the value of
6293 @code{HAVE_PRE_DECREMENT}.
6296 @defmac USE_STORE_POST_INCREMENT (@var{mode})
6297 A C expression used to determine whether a store postincrement is a good
6298 thing to use for a given mode. Defaults to the value of
6299 @code{HAVE_POST_INCREMENT}.
6302 @defmac USE_STORE_POST_DECREMENT (@var{mode})
6303 A C expression used to determine whether a store postdecrement is a good
6304 thing to use for a given mode. Defaults to the value of
6305 @code{HAVE_POST_DECREMENT}.
6308 @defmac USE_STORE_PRE_INCREMENT (@var{mode})
6309 This macro is used to determine whether a store preincrement is a good
6310 thing to use for a given mode. Defaults to the value of
6311 @code{HAVE_PRE_INCREMENT}.
6314 @defmac USE_STORE_PRE_DECREMENT (@var{mode})
6315 This macro is used to determine whether a store predecrement is a good
6316 thing to use for a given mode. Defaults to the value of
6317 @code{HAVE_PRE_DECREMENT}.
6320 @defmac NO_FUNCTION_CSE
6321 Define this macro if it is as good or better to call a constant
6322 function address than to call an address kept in a register.
6325 @defmac LOGICAL_OP_NON_SHORT_CIRCUIT
6326 Define this macro if a non-short-circuit operation produced by
6327 @samp{fold_range_test ()} is optimal. This macro defaults to true if
6328 @code{BRANCH_COST} is greater than or equal to the value 2.
6331 @deftypefn {Target Hook} bool TARGET_RTX_COSTS (rtx @var{x}, int @var{code}, int @var{outer_code}, int @var{opno}, int *@var{total}, bool @var{speed})
6332 This target hook describes the relative costs of RTL expressions.
6334 The cost may depend on the precise form of the expression, which is
6335 available for examination in @var{x}, and the fact that @var{x} appears
6336 as operand @var{opno} of an expression with rtx code @var{outer_code}.
6337 That is, the hook can assume that there is some rtx @var{y} such
6338 that @samp{GET_CODE (@var{y}) == @var{outer_code}} and such that
6339 either (a) @samp{XEXP (@var{y}, @var{opno}) == @var{x}} or
6340 (b) @samp{XVEC (@var{y}, @var{opno})} contains @var{x}.
6342 @var{code} is @var{x}'s expression code---redundant, since it can be
6343 obtained with @code{GET_CODE (@var{x})}.
6345 In implementing this hook, you can use the construct
6346 @code{COSTS_N_INSNS (@var{n})} to specify a cost equal to @var{n} fast
6349 On entry to the hook, @code{*@var{total}} contains a default estimate
6350 for the cost of the expression. The hook should modify this value as
6351 necessary. Traditionally, the default costs are @code{COSTS_N_INSNS (5)}
6352 for multiplications, @code{COSTS_N_INSNS (7)} for division and modulus
6353 operations, and @code{COSTS_N_INSNS (1)} for all other operations.
6355 When optimizing for code size, i.e.@: when @code{speed} is
6356 false, this target hook should be used to estimate the relative
6357 size cost of an expression, again relative to @code{COSTS_N_INSNS}.
6359 The hook returns true when all subexpressions of @var{x} have been
6360 processed, and false when @code{rtx_cost} should recurse.
6363 @deftypefn {Target Hook} int TARGET_ADDRESS_COST (rtx @var{address}, machine_mode @var{mode}, addr_space_t @var{as}, bool @var{speed})
6364 This hook computes the cost of an addressing mode that contains
6365 @var{address}. If not defined, the cost is computed from
6366 the @var{address} expression and the @code{TARGET_RTX_COST} hook.
6368 For most CISC machines, the default cost is a good approximation of the
6369 true cost of the addressing mode. However, on RISC machines, all
6370 instructions normally have the same length and execution time. Hence
6371 all addresses will have equal costs.
6373 In cases where more than one form of an address is known, the form with
6374 the lowest cost will be used. If multiple forms have the same, lowest,
6375 cost, the one that is the most complex will be used.
6377 For example, suppose an address that is equal to the sum of a register
6378 and a constant is used twice in the same basic block. When this macro
6379 is not defined, the address will be computed in a register and memory
6380 references will be indirect through that register. On machines where
6381 the cost of the addressing mode containing the sum is no higher than
6382 that of a simple indirect reference, this will produce an additional
6383 instruction and possibly require an additional register. Proper
6384 specification of this macro eliminates this overhead for such machines.
6386 This hook is never called with an invalid address.
6388 On machines where an address involving more than one register is as
6389 cheap as an address computation involving only one register, defining
6390 @code{TARGET_ADDRESS_COST} to reflect this can cause two registers to
6391 be live over a region of code where only one would have been if
6392 @code{TARGET_ADDRESS_COST} were not defined in that manner. This effect
6393 should be considered in the definition of this macro. Equivalent costs
6394 should probably only be given to addresses with different numbers of
6395 registers on machines with lots of registers.
6399 @section Adjusting the Instruction Scheduler
6401 The instruction scheduler may need a fair amount of machine-specific
6402 adjustment in order to produce good code. GCC provides several target
6403 hooks for this purpose. It is usually enough to define just a few of
6404 them: try the first ones in this list first.
6406 @deftypefn {Target Hook} int TARGET_SCHED_ISSUE_RATE (void)
6407 This hook returns the maximum number of instructions that can ever
6408 issue at the same time on the target machine. The default is one.
6409 Although the insn scheduler can define itself the possibility of issue
6410 an insn on the same cycle, the value can serve as an additional
6411 constraint to issue insns on the same simulated processor cycle (see
6412 hooks @samp{TARGET_SCHED_REORDER} and @samp{TARGET_SCHED_REORDER2}).
6413 This value must be constant over the entire compilation. If you need
6414 it to vary depending on what the instructions are, you must use
6415 @samp{TARGET_SCHED_VARIABLE_ISSUE}.
6418 @deftypefn {Target Hook} int TARGET_SCHED_VARIABLE_ISSUE (FILE *@var{file}, int @var{verbose}, rtx_insn *@var{insn}, int @var{more})
6419 This hook is executed by the scheduler after it has scheduled an insn
6420 from the ready list. It should return the number of insns which can
6421 still be issued in the current cycle. The default is
6422 @samp{@w{@var{more} - 1}} for insns other than @code{CLOBBER} and
6423 @code{USE}, which normally are not counted against the issue rate.
6424 You should define this hook if some insns take more machine resources
6425 than others, so that fewer insns can follow them in the same cycle.
6426 @var{file} is either a null pointer, or a stdio stream to write any
6427 debug output to. @var{verbose} is the verbose level provided by
6428 @option{-fsched-verbose-@var{n}}. @var{insn} is the instruction that
6432 @deftypefn {Target Hook} int TARGET_SCHED_ADJUST_COST (rtx_insn *@var{insn}, rtx @var{link}, rtx_insn *@var{dep_insn}, int @var{cost})
6433 This function corrects the value of @var{cost} based on the
6434 relationship between @var{insn} and @var{dep_insn} through the
6435 dependence @var{link}. It should return the new value. The default
6436 is to make no adjustment to @var{cost}. This can be used for example
6437 to specify to the scheduler using the traditional pipeline description
6438 that an output- or anti-dependence does not incur the same cost as a
6439 data-dependence. If the scheduler using the automaton based pipeline
6440 description, the cost of anti-dependence is zero and the cost of
6441 output-dependence is maximum of one and the difference of latency
6442 times of the first and the second insns. If these values are not
6443 acceptable, you could use the hook to modify them too. See also
6444 @pxref{Processor pipeline description}.
6447 @deftypefn {Target Hook} int TARGET_SCHED_ADJUST_PRIORITY (rtx_insn *@var{insn}, int @var{priority})
6448 This hook adjusts the integer scheduling priority @var{priority} of
6449 @var{insn}. It should return the new priority. Increase the priority to
6450 execute @var{insn} earlier, reduce the priority to execute @var{insn}
6451 later. Do not define this hook if you do not need to adjust the
6452 scheduling priorities of insns.
6455 @deftypefn {Target Hook} int TARGET_SCHED_REORDER (FILE *@var{file}, int @var{verbose}, rtx_insn **@var{ready}, int *@var{n_readyp}, int @var{clock})
6456 This hook is executed by the scheduler after it has scheduled the ready
6457 list, to allow the machine description to reorder it (for example to
6458 combine two small instructions together on @samp{VLIW} machines).
6459 @var{file} is either a null pointer, or a stdio stream to write any
6460 debug output to. @var{verbose} is the verbose level provided by
6461 @option{-fsched-verbose-@var{n}}. @var{ready} is a pointer to the ready
6462 list of instructions that are ready to be scheduled. @var{n_readyp} is
6463 a pointer to the number of elements in the ready list. The scheduler
6464 reads the ready list in reverse order, starting with
6465 @var{ready}[@var{*n_readyp} @minus{} 1] and going to @var{ready}[0]. @var{clock}
6466 is the timer tick of the scheduler. You may modify the ready list and
6467 the number of ready insns. The return value is the number of insns that
6468 can issue this cycle; normally this is just @code{issue_rate}. See also
6469 @samp{TARGET_SCHED_REORDER2}.
6472 @deftypefn {Target Hook} int TARGET_SCHED_REORDER2 (FILE *@var{file}, int @var{verbose}, rtx_insn **@var{ready}, int *@var{n_readyp}, int @var{clock})
6473 Like @samp{TARGET_SCHED_REORDER}, but called at a different time. That
6474 function is called whenever the scheduler starts a new cycle. This one
6475 is called once per iteration over a cycle, immediately after
6476 @samp{TARGET_SCHED_VARIABLE_ISSUE}; it can reorder the ready list and
6477 return the number of insns to be scheduled in the same cycle. Defining
6478 this hook can be useful if there are frequent situations where
6479 scheduling one insn causes other insns to become ready in the same
6480 cycle. These other insns can then be taken into account properly.
6483 @deftypefn {Target Hook} bool TARGET_SCHED_MACRO_FUSION_P (void)
6484 This hook is used to check whether target platform supports macro fusion.
6487 @deftypefn {Target Hook} bool TARGET_SCHED_MACRO_FUSION_PAIR_P (rtx_insn *@var{condgen}, rtx_insn *@var{condjmp})
6488 This hook is used to check whether two insns could be macro fused for
6489 target microarchitecture. If this hook returns true for the given insn pair
6490 (@var{condgen} and @var{condjmp}), scheduler will put them into a sched
6491 group, and they will not be scheduled apart.
6494 @deftypefn {Target Hook} void TARGET_SCHED_DEPENDENCIES_EVALUATION_HOOK (rtx_insn *@var{head}, rtx_insn *@var{tail})
6495 This hook is called after evaluation forward dependencies of insns in
6496 chain given by two parameter values (@var{head} and @var{tail}
6497 correspondingly) but before insns scheduling of the insn chain. For
6498 example, it can be used for better insn classification if it requires
6499 analysis of dependencies. This hook can use backward and forward
6500 dependencies of the insn scheduler because they are already
6504 @deftypefn {Target Hook} void TARGET_SCHED_INIT (FILE *@var{file}, int @var{verbose}, int @var{max_ready})
6505 This hook is executed by the scheduler at the beginning of each block of
6506 instructions that are to be scheduled. @var{file} is either a null
6507 pointer, or a stdio stream to write any debug output to. @var{verbose}
6508 is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6509 @var{max_ready} is the maximum number of insns in the current scheduling
6510 region that can be live at the same time. This can be used to allocate
6511 scratch space if it is needed, e.g.@: by @samp{TARGET_SCHED_REORDER}.
6514 @deftypefn {Target Hook} void TARGET_SCHED_FINISH (FILE *@var{file}, int @var{verbose})
6515 This hook is executed by the scheduler at the end of each block of
6516 instructions that are to be scheduled. It can be used to perform
6517 cleanup of any actions done by the other scheduling hooks. @var{file}
6518 is either a null pointer, or a stdio stream to write any debug output
6519 to. @var{verbose} is the verbose level provided by
6520 @option{-fsched-verbose-@var{n}}.
6523 @deftypefn {Target Hook} void TARGET_SCHED_INIT_GLOBAL (FILE *@var{file}, int @var{verbose}, int @var{old_max_uid})
6524 This hook is executed by the scheduler after function level initializations.
6525 @var{file} is either a null pointer, or a stdio stream to write any debug output to.
6526 @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6527 @var{old_max_uid} is the maximum insn uid when scheduling begins.
6530 @deftypefn {Target Hook} void TARGET_SCHED_FINISH_GLOBAL (FILE *@var{file}, int @var{verbose})
6531 This is the cleanup hook corresponding to @code{TARGET_SCHED_INIT_GLOBAL}.
6532 @var{file} is either a null pointer, or a stdio stream to write any debug output to.
6533 @var{verbose} is the verbose level provided by @option{-fsched-verbose-@var{n}}.
6536 @deftypefn {Target Hook} rtx TARGET_SCHED_DFA_PRE_CYCLE_INSN (void)
6537 The hook returns an RTL insn. The automaton state used in the
6538 pipeline hazard recognizer is changed as if the insn were scheduled
6539 when the new simulated processor cycle starts. Usage of the hook may
6540 simplify the automaton pipeline description for some @acronym{VLIW}
6541 processors. If the hook is defined, it is used only for the automaton
6542 based pipeline description. The default is not to change the state
6543 when the new simulated processor cycle starts.
6546 @deftypefn {Target Hook} void TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN (void)
6547 The hook can be used to initialize data used by the previous hook.
6550 @deftypefn {Target Hook} {rtx_insn *} TARGET_SCHED_DFA_POST_CYCLE_INSN (void)
6551 The hook is analogous to @samp{TARGET_SCHED_DFA_PRE_CYCLE_INSN} but used
6552 to changed the state as if the insn were scheduled when the new
6553 simulated processor cycle finishes.
6556 @deftypefn {Target Hook} void TARGET_SCHED_INIT_DFA_POST_CYCLE_INSN (void)
6557 The hook is analogous to @samp{TARGET_SCHED_INIT_DFA_PRE_CYCLE_INSN} but
6558 used to initialize data used by the previous hook.
6561 @deftypefn {Target Hook} void TARGET_SCHED_DFA_PRE_ADVANCE_CYCLE (void)
6562 The hook to notify target that the current simulated cycle is about to finish.
6563 The hook is analogous to @samp{TARGET_SCHED_DFA_PRE_CYCLE_INSN} but used
6564 to change the state in more complicated situations - e.g., when advancing
6565 state on a single insn is not enough.
6568 @deftypefn {Target Hook} void TARGET_SCHED_DFA_POST_ADVANCE_CYCLE (void)
6569 The hook to notify target that new simulated cycle has just started.
6570 The hook is analogous to @samp{TARGET_SCHED_DFA_POST_CYCLE_INSN} but used
6571 to change the state in more complicated situations - e.g., when advancing
6572 state on a single insn is not enough.
6575 @deftypefn {Target Hook} int TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD (void)
6576 This hook controls better choosing an insn from the ready insn queue
6577 for the @acronym{DFA}-based insn scheduler. Usually the scheduler
6578 chooses the first insn from the queue. If the hook returns a positive
6579 value, an additional scheduler code tries all permutations of
6580 @samp{TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD ()}
6581 subsequent ready insns to choose an insn whose issue will result in
6582 maximal number of issued insns on the same cycle. For the
6583 @acronym{VLIW} processor, the code could actually solve the problem of
6584 packing simple insns into the @acronym{VLIW} insn. Of course, if the
6585 rules of @acronym{VLIW} packing are described in the automaton.
6587 This code also could be used for superscalar @acronym{RISC}
6588 processors. Let us consider a superscalar @acronym{RISC} processor
6589 with 3 pipelines. Some insns can be executed in pipelines @var{A} or
6590 @var{B}, some insns can be executed only in pipelines @var{B} or
6591 @var{C}, and one insn can be executed in pipeline @var{B}. The
6592 processor may issue the 1st insn into @var{A} and the 2nd one into
6593 @var{B}. In this case, the 3rd insn will wait for freeing @var{B}
6594 until the next cycle. If the scheduler issues the 3rd insn the first,
6595 the processor could issue all 3 insns per cycle.
6597 Actually this code demonstrates advantages of the automaton based
6598 pipeline hazard recognizer. We try quickly and easy many insn
6599 schedules to choose the best one.
6601 The default is no multipass scheduling.
6604 @deftypefn {Target Hook} int TARGET_SCHED_FIRST_CYCLE_MULTIPASS_DFA_LOOKAHEAD_GUARD (rtx_insn *@var{insn}, int @var{ready_index})
6606 This hook controls what insns from the ready insn queue will be
6607 considered for the multipass insn scheduling. If the hook returns
6608 zero for @var{insn}, the insn will be considered in multipass scheduling.
6609 Positive return values will remove @var{insn} from consideration on
6610 the current round of multipass scheduling.
6611 Negative return values will remove @var{insn} from consideration for given
6613 Backends should be careful about returning non-zero for highest priority
6614 instruction at position 0 in the ready list. @var{ready_index} is passed
6615 to allow backends make correct judgements.
6617 The default is that any ready insns can be chosen to be issued.
6620 @deftypefn {Target Hook} void TARGET_SCHED_FIRST_CYCLE_MULTIPASS_BEGIN (void *@var{data}, signed char *@var{ready_try}, int @var{n_ready}, bool @var{first_cycle_insn_p})
6621 This hook prepares the target backend for a new round of multipass
6625 @deftypefn {Target Hook} void TARGET_SCHED_FIRST_CYCLE_MULTIPASS_ISSUE (void *@var{data}, signed char *@var{ready_try}, int @var{n_ready}, rtx_insn *@var{insn}, const void *@var{prev_data})
6626 This hook is called when multipass scheduling evaluates instruction INSN.
6629 @deftypefn {Target Hook} void TARGET_SCHED_FIRST_CYCLE_MULTIPASS_BACKTRACK (const void *@var{data}, signed char *@var{ready_try}, int @var{n_ready})
6630 This is called when multipass scheduling backtracks from evaluation of
6634 @deftypefn {Target Hook} void TARGET_SCHED_FIRST_CYCLE_MULTIPASS_END (const void *@var{data})
6635 This hook notifies the target about the result of the concluded current
6636 round of multipass scheduling.
6639 @deftypefn {Target Hook} void TARGET_SCHED_FIRST_CYCLE_MULTIPASS_INIT (void *@var{data})
6640 This hook initializes target-specific data used in multipass scheduling.
6643 @deftypefn {Target Hook} void TARGET_SCHED_FIRST_CYCLE_MULTIPASS_FINI (void *@var{data})
6644 This hook finalizes target-specific data used in multipass scheduling.
6647 @deftypefn {Target Hook} int TARGET_SCHED_DFA_NEW_CYCLE (FILE *@var{dump}, int @var{verbose}, rtx_insn *@var{insn}, int @var{last_clock}, int @var{clock}, int *@var{sort_p})
6648 This hook is called by the insn scheduler before issuing @var{insn}
6649 on cycle @var{clock}. If the hook returns nonzero,
6650 @var{insn} is not issued on this processor cycle. Instead,
6651 the processor cycle is advanced. If *@var{sort_p}
6652 is zero, the insn ready queue is not sorted on the new cycle
6653 start as usually. @var{dump} and @var{verbose} specify the file and
6654 verbosity level to use for debugging output.
6655 @var{last_clock} and @var{clock} are, respectively, the
6656 processor cycle on which the previous insn has been issued,
6657 and the current processor cycle.
6660 @deftypefn {Target Hook} bool TARGET_SCHED_IS_COSTLY_DEPENDENCE (struct _dep *@var{_dep}, int @var{cost}, int @var{distance})
6661 This hook is used to define which dependences are considered costly by
6662 the target, so costly that it is not advisable to schedule the insns that
6663 are involved in the dependence too close to one another. The parameters
6664 to this hook are as follows: The first parameter @var{_dep} is the dependence
6665 being evaluated. The second parameter @var{cost} is the cost of the
6666 dependence as estimated by the scheduler, and the third
6667 parameter @var{distance} is the distance in cycles between the two insns.
6668 The hook returns @code{true} if considering the distance between the two
6669 insns the dependence between them is considered costly by the target,
6670 and @code{false} otherwise.
6672 Defining this hook can be useful in multiple-issue out-of-order machines,
6673 where (a) it's practically hopeless to predict the actual data/resource
6674 delays, however: (b) there's a better chance to predict the actual grouping
6675 that will be formed, and (c) correctly emulating the grouping can be very
6676 important. In such targets one may want to allow issuing dependent insns
6677 closer to one another---i.e., closer than the dependence distance; however,
6678 not in cases of ``costly dependences'', which this hooks allows to define.
6681 @deftypefn {Target Hook} void TARGET_SCHED_H_I_D_EXTENDED (void)
6682 This hook is called by the insn scheduler after emitting a new instruction to
6683 the instruction stream. The hook notifies a target backend to extend its
6684 per instruction data structures.
6687 @deftypefn {Target Hook} {void *} TARGET_SCHED_ALLOC_SCHED_CONTEXT (void)
6688 Return a pointer to a store large enough to hold target scheduling context.
6691 @deftypefn {Target Hook} void TARGET_SCHED_INIT_SCHED_CONTEXT (void *@var{tc}, bool @var{clean_p})
6692 Initialize store pointed to by @var{tc} to hold target scheduling context.
6693 It @var{clean_p} is true then initialize @var{tc} as if scheduler is at the
6694 beginning of the block. Otherwise, copy the current context into @var{tc}.
6697 @deftypefn {Target Hook} void TARGET_SCHED_SET_SCHED_CONTEXT (void *@var{tc})
6698 Copy target scheduling context pointed to by @var{tc} to the current context.
6701 @deftypefn {Target Hook} void TARGET_SCHED_CLEAR_SCHED_CONTEXT (void *@var{tc})
6702 Deallocate internal data in target scheduling context pointed to by @var{tc}.
6705 @deftypefn {Target Hook} void TARGET_SCHED_FREE_SCHED_CONTEXT (void *@var{tc})
6706 Deallocate a store for target scheduling context pointed to by @var{tc}.
6709 @deftypefn {Target Hook} int TARGET_SCHED_SPECULATE_INSN (rtx_insn *@var{insn}, unsigned int @var{dep_status}, rtx *@var{new_pat})
6710 This hook is called by the insn scheduler when @var{insn} has only
6711 speculative dependencies and therefore can be scheduled speculatively.
6712 The hook is used to check if the pattern of @var{insn} has a speculative
6713 version and, in case of successful check, to generate that speculative
6714 pattern. The hook should return 1, if the instruction has a speculative form,
6715 or @minus{}1, if it doesn't. @var{request} describes the type of requested
6716 speculation. If the return value equals 1 then @var{new_pat} is assigned
6717 the generated speculative pattern.
6720 @deftypefn {Target Hook} bool TARGET_SCHED_NEEDS_BLOCK_P (unsigned int @var{dep_status})
6721 This hook is called by the insn scheduler during generation of recovery code
6722 for @var{insn}. It should return @code{true}, if the corresponding check
6723 instruction should branch to recovery code, or @code{false} otherwise.
6726 @deftypefn {Target Hook} rtx TARGET_SCHED_GEN_SPEC_CHECK (rtx_insn *@var{insn}, rtx_insn *@var{label}, unsigned int @var{ds})
6727 This hook is called by the insn scheduler to generate a pattern for recovery
6728 check instruction. If @var{mutate_p} is zero, then @var{insn} is a
6729 speculative instruction for which the check should be generated.
6730 @var{label} is either a label of a basic block, where recovery code should
6731 be emitted, or a null pointer, when requested check doesn't branch to
6732 recovery code (a simple check). If @var{mutate_p} is nonzero, then
6733 a pattern for a branchy check corresponding to a simple check denoted by
6734 @var{insn} should be generated. In this case @var{label} can't be null.
6737 @deftypefn {Target Hook} void TARGET_SCHED_SET_SCHED_FLAGS (struct spec_info_def *@var{spec_info})
6738 This hook is used by the insn scheduler to find out what features should be
6740 The structure *@var{spec_info} should be filled in by the target.
6741 The structure describes speculation types that can be used in the scheduler.
6744 @deftypefn {Target Hook} int TARGET_SCHED_SMS_RES_MII (struct ddg *@var{g})
6745 This hook is called by the swing modulo scheduler to calculate a
6746 resource-based lower bound which is based on the resources available in
6747 the machine and the resources required by each instruction. The target
6748 backend can use @var{g} to calculate such bound. A very simple lower
6749 bound will be used in case this hook is not implemented: the total number
6750 of instructions divided by the issue rate.
6753 @deftypefn {Target Hook} bool TARGET_SCHED_DISPATCH (rtx_insn *@var{insn}, int @var{x})
6754 This hook is called by Haifa Scheduler. It returns true if dispatch scheduling
6755 is supported in hardware and the condition specified in the parameter is true.
6758 @deftypefn {Target Hook} void TARGET_SCHED_DISPATCH_DO (rtx_insn *@var{insn}, int @var{x})
6759 This hook is called by Haifa Scheduler. It performs the operation specified
6760 in its second parameter.
6763 @deftypevr {Target Hook} bool TARGET_SCHED_EXPOSED_PIPELINE
6764 True if the processor has an exposed pipeline, which means that not just
6765 the order of instructions is important for correctness when scheduling, but
6766 also the latencies of operations.
6769 @deftypefn {Target Hook} int TARGET_SCHED_REASSOCIATION_WIDTH (unsigned int @var{opc}, machine_mode @var{mode})
6770 This hook is called by tree reassociator to determine a level of
6771 parallelism required in output calculations chain.
6775 @section Dividing the Output into Sections (Texts, Data, @dots{})
6776 @c the above section title is WAY too long. maybe cut the part between
6777 @c the (...)? --mew 10feb93
6779 An object file is divided into sections containing different types of
6780 data. In the most common case, there are three sections: the @dfn{text
6781 section}, which holds instructions and read-only data; the @dfn{data
6782 section}, which holds initialized writable data; and the @dfn{bss
6783 section}, which holds uninitialized data. Some systems have other kinds
6786 @file{varasm.c} provides several well-known sections, such as
6787 @code{text_section}, @code{data_section} and @code{bss_section}.
6788 The normal way of controlling a @code{@var{foo}_section} variable
6789 is to define the associated @code{@var{FOO}_SECTION_ASM_OP} macro,
6790 as described below. The macros are only read once, when @file{varasm.c}
6791 initializes itself, so their values must be run-time constants.
6792 They may however depend on command-line flags.
6794 @emph{Note:} Some run-time files, such @file{crtstuff.c}, also make
6795 use of the @code{@var{FOO}_SECTION_ASM_OP} macros, and expect them
6796 to be string literals.
6798 Some assemblers require a different string to be written every time a
6799 section is selected. If your assembler falls into this category, you
6800 should define the @code{TARGET_ASM_INIT_SECTIONS} hook and use
6801 @code{get_unnamed_section} to set up the sections.
6803 You must always create a @code{text_section}, either by defining
6804 @code{TEXT_SECTION_ASM_OP} or by initializing @code{text_section}
6805 in @code{TARGET_ASM_INIT_SECTIONS}. The same is true of
6806 @code{data_section} and @code{DATA_SECTION_ASM_OP}. If you do not
6807 create a distinct @code{readonly_data_section}, the default is to
6808 reuse @code{text_section}.
6810 All the other @file{varasm.c} sections are optional, and are null
6811 if the target does not provide them.
6813 @defmac TEXT_SECTION_ASM_OP
6814 A C expression whose value is a string, including spacing, containing the
6815 assembler operation that should precede instructions and read-only data.
6816 Normally @code{"\t.text"} is right.
6819 @defmac HOT_TEXT_SECTION_NAME
6820 If defined, a C string constant for the name of the section containing most
6821 frequently executed functions of the program. If not defined, GCC will provide
6822 a default definition if the target supports named sections.
6825 @defmac UNLIKELY_EXECUTED_TEXT_SECTION_NAME
6826 If defined, a C string constant for the name of the section containing unlikely
6827 executed functions in the program.
6830 @defmac DATA_SECTION_ASM_OP
6831 A C expression whose value is a string, including spacing, containing the
6832 assembler operation to identify the following data as writable initialized
6833 data. Normally @code{"\t.data"} is right.
6836 @defmac SDATA_SECTION_ASM_OP
6837 If defined, a C expression whose value is a string, including spacing,
6838 containing the assembler operation to identify the following data as
6839 initialized, writable small data.
6842 @defmac READONLY_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 read-only initialized
6848 @defmac BSS_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 uninitialized global data. If not defined, and
6852 @code{ASM_OUTPUT_ALIGNED_BSS} not defined,
6853 uninitialized global data will be output in the data section if
6854 @option{-fno-common} is passed, otherwise @code{ASM_OUTPUT_COMMON} will be
6858 @defmac SBSS_SECTION_ASM_OP
6859 If defined, a C expression whose value is a string, including spacing,
6860 containing the assembler operation to identify the following data as
6861 uninitialized, writable small data.
6864 @defmac TLS_COMMON_ASM_OP
6865 If defined, a C expression whose value is a string containing the
6866 assembler operation to identify the following data as thread-local
6867 common data. The default is @code{".tls_common"}.
6870 @defmac TLS_SECTION_ASM_FLAG
6871 If defined, a C expression whose value is a character constant
6872 containing the flag used to mark a section as a TLS section. The
6873 default is @code{'T'}.
6876 @defmac INIT_SECTION_ASM_OP
6877 If defined, a C expression whose value is a string, including spacing,
6878 containing the assembler operation to identify the following data as
6879 initialization code. If not defined, GCC will assume such a section does
6880 not exist. This section has no corresponding @code{init_section}
6881 variable; it is used entirely in runtime code.
6884 @defmac FINI_SECTION_ASM_OP
6885 If defined, a C expression whose value is a string, including spacing,
6886 containing the assembler operation to identify the following data as
6887 finalization code. If not defined, GCC will assume such a section does
6888 not exist. This section has no corresponding @code{fini_section}
6889 variable; it is used entirely in runtime code.
6892 @defmac INIT_ARRAY_SECTION_ASM_OP
6893 If defined, a C expression whose value is a string, including spacing,
6894 containing the assembler operation to identify the following data as
6895 part of the @code{.init_array} (or equivalent) section. If not
6896 defined, GCC will assume such a section does not exist. Do not define
6897 both this macro and @code{INIT_SECTION_ASM_OP}.
6900 @defmac FINI_ARRAY_SECTION_ASM_OP
6901 If defined, a C expression whose value is a string, including spacing,
6902 containing the assembler operation to identify the following data as
6903 part of the @code{.fini_array} (or equivalent) section. If not
6904 defined, GCC will assume such a section does not exist. Do not define
6905 both this macro and @code{FINI_SECTION_ASM_OP}.
6908 @defmac CRT_CALL_STATIC_FUNCTION (@var{section_op}, @var{function})
6909 If defined, an ASM statement that switches to a different section
6910 via @var{section_op}, calls @var{function}, and switches back to
6911 the text section. This is used in @file{crtstuff.c} if
6912 @code{INIT_SECTION_ASM_OP} or @code{FINI_SECTION_ASM_OP} to calls
6913 to initialization and finalization functions from the init and fini
6914 sections. By default, this macro uses a simple function call. Some
6915 ports need hand-crafted assembly code to avoid dependencies on
6916 registers initialized in the function prologue or to ensure that
6917 constant pools don't end up too far way in the text section.
6920 @defmac TARGET_LIBGCC_SDATA_SECTION
6921 If defined, a string which names the section into which small
6922 variables defined in crtstuff and libgcc should go. This is useful
6923 when the target has options for optimizing access to small data, and
6924 you want the crtstuff and libgcc routines to be conservative in what
6925 they expect of your application yet liberal in what your application
6926 expects. For example, for targets with a @code{.sdata} section (like
6927 MIPS), you could compile crtstuff with @code{-G 0} so that it doesn't
6928 require small data support from your application, but use this macro
6929 to put small data into @code{.sdata} so that your application can
6930 access these variables whether it uses small data or not.
6933 @defmac FORCE_CODE_SECTION_ALIGN
6934 If defined, an ASM statement that aligns a code section to some
6935 arbitrary boundary. This is used to force all fragments of the
6936 @code{.init} and @code{.fini} sections to have to same alignment
6937 and thus prevent the linker from having to add any padding.
6940 @defmac JUMP_TABLES_IN_TEXT_SECTION
6941 Define this macro to be an expression with a nonzero value if jump
6942 tables (for @code{tablejump} insns) should be output in the text
6943 section, along with the assembler instructions. Otherwise, the
6944 readonly data section is used.
6946 This macro is irrelevant if there is no separate readonly data section.
6949 @deftypefn {Target Hook} void TARGET_ASM_INIT_SECTIONS (void)
6950 Define this hook if you need to do something special to set up the
6951 @file{varasm.c} sections, or if your target has some special sections
6952 of its own that you need to create.
6954 GCC calls this hook after processing the command line, but before writing
6955 any assembly code, and before calling any of the section-returning hooks
6959 @deftypefn {Target Hook} int TARGET_ASM_RELOC_RW_MASK (void)
6960 Return a mask describing how relocations should be treated when
6961 selecting sections. Bit 1 should be set if global relocations
6962 should be placed in a read-write section; bit 0 should be set if
6963 local relocations should be placed in a read-write section.
6965 The default version of this function returns 3 when @option{-fpic}
6966 is in effect, and 0 otherwise. The hook is typically redefined
6967 when the target cannot support (some kinds of) dynamic relocations
6968 in read-only sections even in executables.
6971 @deftypefn {Target Hook} {section *} TARGET_ASM_SELECT_SECTION (tree @var{exp}, int @var{reloc}, unsigned HOST_WIDE_INT @var{align})
6972 Return the section into which @var{exp} should be placed. You can
6973 assume that @var{exp} is either a @code{VAR_DECL} node or a constant of
6974 some sort. @var{reloc} indicates whether the initial value of @var{exp}
6975 requires link-time relocations. Bit 0 is set when variable contains
6976 local relocations only, while bit 1 is set for global relocations.
6977 @var{align} is the constant alignment in bits.
6979 The default version of this function takes care of putting read-only
6980 variables in @code{readonly_data_section}.
6982 See also @var{USE_SELECT_SECTION_FOR_FUNCTIONS}.
6985 @defmac USE_SELECT_SECTION_FOR_FUNCTIONS
6986 Define this macro if you wish TARGET_ASM_SELECT_SECTION to be called
6987 for @code{FUNCTION_DECL}s as well as for variables and constants.
6989 In the case of a @code{FUNCTION_DECL}, @var{reloc} will be zero if the
6990 function has been determined to be likely to be called, and nonzero if
6991 it is unlikely to be called.
6994 @deftypefn {Target Hook} void TARGET_ASM_UNIQUE_SECTION (tree @var{decl}, int @var{reloc})
6995 Build up a unique section name, expressed as a @code{STRING_CST} node,
6996 and assign it to @samp{DECL_SECTION_NAME (@var{decl})}.
6997 As with @code{TARGET_ASM_SELECT_SECTION}, @var{reloc} indicates whether
6998 the initial value of @var{exp} requires link-time relocations.
7000 The default version of this function appends the symbol name to the
7001 ELF section name that would normally be used for the symbol. For
7002 example, the function @code{foo} would be placed in @code{.text.foo}.
7003 Whatever the actual target object format, this is often good enough.
7006 @deftypefn {Target Hook} {section *} TARGET_ASM_FUNCTION_RODATA_SECTION (tree @var{decl})
7007 Return the readonly data section associated with
7008 @samp{DECL_SECTION_NAME (@var{decl})}.
7009 The default version of this function selects @code{.gnu.linkonce.r.name} if
7010 the function's section is @code{.gnu.linkonce.t.name}, @code{.rodata.name}
7011 if function is in @code{.text.name}, and the normal readonly-data section
7015 @deftypevr {Target Hook} {const char *} TARGET_ASM_MERGEABLE_RODATA_PREFIX
7016 Usually, the compiler uses the prefix @code{".rodata"} to construct
7017 section names for mergeable constant data. Define this macro to override
7018 the string if a different section name should be used.
7021 @deftypefn {Target Hook} {section *} TARGET_ASM_TM_CLONE_TABLE_SECTION (void)
7022 Return the section that should be used for transactional memory clone tables.
7025 @deftypefn {Target Hook} {section *} TARGET_ASM_SELECT_RTX_SECTION (machine_mode @var{mode}, rtx @var{x}, unsigned HOST_WIDE_INT @var{align})
7026 Return the section into which a constant @var{x}, of mode @var{mode},
7027 should be placed. You can assume that @var{x} is some kind of
7028 constant in RTL@. The argument @var{mode} is redundant except in the
7029 case of a @code{const_int} rtx. @var{align} is the constant alignment
7032 The default version of this function takes care of putting symbolic
7033 constants in @code{flag_pic} mode in @code{data_section} and everything
7034 else in @code{readonly_data_section}.
7037 @deftypefn {Target Hook} tree TARGET_MANGLE_DECL_ASSEMBLER_NAME (tree @var{decl}, tree @var{id})
7038 Define this hook if you need to postprocess the assembler name generated
7039 by target-independent code. The @var{id} provided to this hook will be
7040 the computed name (e.g., the macro @code{DECL_NAME} of the @var{decl} in C,
7041 or the mangled name of the @var{decl} in C++). The return value of the
7042 hook is an @code{IDENTIFIER_NODE} for the appropriate mangled name on
7043 your target system. The default implementation of this hook just
7044 returns the @var{id} provided.
7047 @deftypefn {Target Hook} void TARGET_ENCODE_SECTION_INFO (tree @var{decl}, rtx @var{rtl}, int @var{new_decl_p})
7048 Define this hook if references to a symbol or a constant must be
7049 treated differently depending on something about the variable or
7050 function named by the symbol (such as what section it is in).
7052 The hook is executed immediately after rtl has been created for
7053 @var{decl}, which may be a variable or function declaration or
7054 an entry in the constant pool. In either case, @var{rtl} is the
7055 rtl in question. Do @emph{not} use @code{DECL_RTL (@var{decl})}
7056 in this hook; that field may not have been initialized yet.
7058 In the case of a constant, it is safe to assume that the rtl is
7059 a @code{mem} whose address is a @code{symbol_ref}. Most decls
7060 will also have this form, but that is not guaranteed. Global
7061 register variables, for instance, will have a @code{reg} for their
7062 rtl. (Normally the right thing to do with such unusual rtl is
7065 The @var{new_decl_p} argument will be true if this is the first time
7066 that @code{TARGET_ENCODE_SECTION_INFO} has been invoked on this decl. It will
7067 be false for subsequent invocations, which will happen for duplicate
7068 declarations. Whether or not anything must be done for the duplicate
7069 declaration depends on whether the hook examines @code{DECL_ATTRIBUTES}.
7070 @var{new_decl_p} is always true when the hook is called for a constant.
7072 @cindex @code{SYMBOL_REF_FLAG}, in @code{TARGET_ENCODE_SECTION_INFO}
7073 The usual thing for this hook to do is to record flags in the
7074 @code{symbol_ref}, using @code{SYMBOL_REF_FLAG} or @code{SYMBOL_REF_FLAGS}.
7075 Historically, the name string was modified if it was necessary to
7076 encode more than one bit of information, but this practice is now
7077 discouraged; use @code{SYMBOL_REF_FLAGS}.
7079 The default definition of this hook, @code{default_encode_section_info}
7080 in @file{varasm.c}, sets a number of commonly-useful bits in
7081 @code{SYMBOL_REF_FLAGS}. Check whether the default does what you need
7082 before overriding it.
7085 @deftypefn {Target Hook} {const char *} TARGET_STRIP_NAME_ENCODING (const char *@var{name})
7086 Decode @var{name} and return the real name part, sans
7087 the characters that @code{TARGET_ENCODE_SECTION_INFO}
7091 @deftypefn {Target Hook} bool TARGET_IN_SMALL_DATA_P (const_tree @var{exp})
7092 Returns true if @var{exp} should be placed into a ``small data'' section.
7093 The default version of this hook always returns false.
7096 @deftypevr {Target Hook} bool TARGET_HAVE_SRODATA_SECTION
7097 Contains the value true if the target places read-only
7098 ``small data'' into a separate section. The default value is false.
7101 @deftypefn {Target Hook} bool TARGET_PROFILE_BEFORE_PROLOGUE (void)
7102 It returns true if target wants profile code emitted before prologue.
7104 The default version of this hook use the target macro
7105 @code{PROFILE_BEFORE_PROLOGUE}.
7108 @deftypefn {Target Hook} bool TARGET_BINDS_LOCAL_P (const_tree @var{exp})
7109 Returns true if @var{exp} names an object for which name resolution
7110 rules must resolve to the current ``module'' (dynamic shared library
7111 or executable image).
7113 The default version of this hook implements the name resolution rules
7114 for ELF, which has a looser model of global name binding than other
7115 currently supported object file formats.
7118 @deftypevr {Target Hook} bool TARGET_HAVE_TLS
7119 Contains the value true if the target supports thread-local storage.
7120 The default value is false.
7125 @section Position Independent Code
7126 @cindex position independent code
7129 This section describes macros that help implement generation of position
7130 independent code. Simply defining these macros is not enough to
7131 generate valid PIC; you must also add support to the hook
7132 @code{TARGET_LEGITIMATE_ADDRESS_P} and to the macro
7133 @code{PRINT_OPERAND_ADDRESS}, as well as @code{LEGITIMIZE_ADDRESS}. You
7134 must modify the definition of @samp{movsi} to do something appropriate
7135 when the source operand contains a symbolic address. You may also
7136 need to alter the handling of switch statements so that they use
7138 @c i rearranged the order of the macros above to try to force one of
7139 @c them to the next line, to eliminate an overfull hbox. --mew 10feb93
7141 @defmac PIC_OFFSET_TABLE_REGNUM
7142 The register number of the register used to address a table of static
7143 data addresses in memory. In some cases this register is defined by a
7144 processor's ``application binary interface'' (ABI)@. When this macro
7145 is defined, RTL is generated for this register once, as with the stack
7146 pointer and frame pointer registers. If this macro is not defined, it
7147 is up to the machine-dependent files to allocate such a register (if
7148 necessary). Note that this register must be fixed when in use (e.g.@:
7149 when @code{flag_pic} is true).
7152 @defmac PIC_OFFSET_TABLE_REG_CALL_CLOBBERED
7153 A C expression that is nonzero if the register defined by
7154 @code{PIC_OFFSET_TABLE_REGNUM} is clobbered by calls. If not defined,
7155 the default is zero. Do not define
7156 this macro if @code{PIC_OFFSET_TABLE_REGNUM} is not defined.
7159 @defmac LEGITIMATE_PIC_OPERAND_P (@var{x})
7160 A C expression that is nonzero if @var{x} is a legitimate immediate
7161 operand on the target machine when generating position independent code.
7162 You can assume that @var{x} satisfies @code{CONSTANT_P}, so you need not
7163 check this. You can also assume @var{flag_pic} is true, so you need not
7164 check it either. You need not define this macro if all constants
7165 (including @code{SYMBOL_REF}) can be immediate operands when generating
7166 position independent code.
7169 @node Assembler Format
7170 @section Defining the Output Assembler Language
7172 This section describes macros whose principal purpose is to describe how
7173 to write instructions in assembler language---rather than what the
7177 * File Framework:: Structural information for the assembler file.
7178 * Data Output:: Output of constants (numbers, strings, addresses).
7179 * Uninitialized Data:: Output of uninitialized variables.
7180 * Label Output:: Output and generation of labels.
7181 * Initialization:: General principles of initialization
7182 and termination routines.
7183 * Macros for Initialization::
7184 Specific macros that control the handling of
7185 initialization and termination routines.
7186 * Instruction Output:: Output of actual instructions.
7187 * Dispatch Tables:: Output of jump tables.
7188 * Exception Region Output:: Output of exception region code.
7189 * Alignment Output:: Pseudo ops for alignment and skipping data.
7192 @node File Framework
7193 @subsection The Overall Framework of an Assembler File
7194 @cindex assembler format
7195 @cindex output of assembler code
7197 @c prevent bad page break with this line
7198 This describes the overall framework of an assembly file.
7200 @findex default_file_start
7201 @deftypefn {Target Hook} void TARGET_ASM_FILE_START (void)
7202 Output to @code{asm_out_file} any text which the assembler expects to
7203 find at the beginning of a file. The default behavior is controlled
7204 by two flags, documented below. Unless your target's assembler is
7205 quite unusual, if you override the default, you should call
7206 @code{default_file_start} at some point in your target hook. This
7207 lets other target files rely on these variables.
7210 @deftypevr {Target Hook} bool TARGET_ASM_FILE_START_APP_OFF
7211 If this flag is true, the text of the macro @code{ASM_APP_OFF} will be
7212 printed as the very first line in the assembly file, unless
7213 @option{-fverbose-asm} is in effect. (If that macro has been defined
7214 to the empty string, this variable has no effect.) With the normal
7215 definition of @code{ASM_APP_OFF}, the effect is to notify the GNU
7216 assembler that it need not bother stripping comments or extra
7217 whitespace from its input. This allows it to work a bit faster.
7219 The default is false. You should not set it to true unless you have
7220 verified that your port does not generate any extra whitespace or
7221 comments that will cause GAS to issue errors in NO_APP mode.
7224 @deftypevr {Target Hook} bool TARGET_ASM_FILE_START_FILE_DIRECTIVE
7225 If this flag is true, @code{output_file_directive} will be called
7226 for the primary source file, immediately after printing
7227 @code{ASM_APP_OFF} (if that is enabled). Most ELF assemblers expect
7228 this to be done. The default is false.
7231 @deftypefn {Target Hook} void TARGET_ASM_FILE_END (void)
7232 Output to @code{asm_out_file} any text which the assembler expects
7233 to find at the end of a file. The default is to output nothing.
7236 @deftypefun void file_end_indicate_exec_stack ()
7237 Some systems use a common convention, the @samp{.note.GNU-stack}
7238 special section, to indicate whether or not an object file relies on
7239 the stack being executable. If your system uses this convention, you
7240 should define @code{TARGET_ASM_FILE_END} to this function. If you
7241 need to do other things in that hook, have your hook function call
7245 @deftypefn {Target Hook} void TARGET_ASM_LTO_START (void)
7246 Output to @code{asm_out_file} any text which the assembler expects
7247 to find at the start of an LTO section. The default is to output
7251 @deftypefn {Target Hook} void TARGET_ASM_LTO_END (void)
7252 Output to @code{asm_out_file} any text which the assembler expects
7253 to find at the end of an LTO section. The default is to output
7257 @deftypefn {Target Hook} void TARGET_ASM_CODE_END (void)
7258 Output to @code{asm_out_file} any text which is needed before emitting
7259 unwind info and debug info at the end of a file. Some targets emit
7260 here PIC setup thunks that cannot be emitted at the end of file,
7261 because they couldn't have unwind info then. The default is to output
7265 @defmac ASM_COMMENT_START
7266 A C string constant describing how to begin a comment in the target
7267 assembler language. The compiler assumes that the comment will end at
7268 the end of the line.
7272 A C string constant for text to be output before each @code{asm}
7273 statement or group of consecutive ones. Normally this is
7274 @code{"#APP"}, which is a comment that has no effect on most
7275 assemblers but tells the GNU assembler that it must check the lines
7276 that follow for all valid assembler constructs.
7280 A C string constant for text to be output after each @code{asm}
7281 statement or group of consecutive ones. Normally this is
7282 @code{"#NO_APP"}, which tells the GNU assembler to resume making the
7283 time-saving assumptions that are valid for ordinary compiler output.
7286 @defmac ASM_OUTPUT_SOURCE_FILENAME (@var{stream}, @var{name})
7287 A C statement to output COFF information or DWARF debugging information
7288 which indicates that filename @var{name} is the current source file to
7289 the stdio stream @var{stream}.
7291 This macro need not be defined if the standard form of output
7292 for the file format in use is appropriate.
7295 @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_SOURCE_FILENAME (FILE *@var{file}, const char *@var{name})
7296 Output COFF information or DWARF debugging information which indicates that filename @var{name} is the current source file to the stdio stream @var{file}.
7298 This target hook need not be defined if the standard form of output for the file format in use is appropriate.
7301 @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_IDENT (const char *@var{name})
7302 Output a string based on @var{name}, suitable for the @samp{#ident} directive, or the equivalent directive or pragma in non-C-family languages. If this hook is not defined, nothing is output for the @samp{#ident} directive.
7305 @defmac OUTPUT_QUOTED_STRING (@var{stream}, @var{string})
7306 A C statement to output the string @var{string} to the stdio stream
7307 @var{stream}. If you do not call the function @code{output_quoted_string}
7308 in your config files, GCC will only call it to output filenames to
7309 the assembler source. So you can use it to canonicalize the format
7310 of the filename using this macro.
7313 @deftypefn {Target Hook} void TARGET_ASM_NAMED_SECTION (const char *@var{name}, unsigned int @var{flags}, tree @var{decl})
7314 Output assembly directives to switch to section @var{name}. The section
7315 should have attributes as specified by @var{flags}, which is a bit mask
7316 of the @code{SECTION_*} flags defined in @file{output.h}. If @var{decl}
7317 is non-NULL, it is the @code{VAR_DECL} or @code{FUNCTION_DECL} with which
7318 this section is associated.
7321 @deftypefn {Target Hook} {section *} TARGET_ASM_FUNCTION_SECTION (tree @var{decl}, enum node_frequency @var{freq}, bool @var{startup}, bool @var{exit})
7322 Return preferred text (sub)section for function @var{decl}.
7323 Main purpose of this function is to separate cold, normal and hot
7324 functions. @var{startup} is true when function is known to be used only
7325 at startup (from static constructors or it is @code{main()}).
7326 @var{exit} is true when function is known to be used only at exit
7327 (from static destructors).
7328 Return NULL if function should go to default text section.
7331 @deftypefn {Target Hook} void TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS (FILE *@var{file}, tree @var{decl}, bool @var{new_is_cold})
7332 Used by the target to emit any assembler directives or additional labels needed when a function is partitioned between different sections. Output should be written to @var{file}. The function decl is available as @var{decl} and the new section is `cold' if @var{new_is_cold} is @code{true}.
7335 @deftypevr {Common Target Hook} bool TARGET_HAVE_NAMED_SECTIONS
7336 This flag is true if the target supports @code{TARGET_ASM_NAMED_SECTION}.
7337 It must not be modified by command-line option processing.
7340 @anchor{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}
7341 @deftypevr {Target Hook} bool TARGET_HAVE_SWITCHABLE_BSS_SECTIONS
7342 This flag is true if we can create zeroed data by switching to a BSS
7343 section and then using @code{ASM_OUTPUT_SKIP} to allocate the space.
7344 This is true on most ELF targets.
7347 @deftypefn {Target Hook} {unsigned int} TARGET_SECTION_TYPE_FLAGS (tree @var{decl}, const char *@var{name}, int @var{reloc})
7348 Choose a set of section attributes for use by @code{TARGET_ASM_NAMED_SECTION}
7349 based on a variable or function decl, a section name, and whether or not the
7350 declaration's initializer may contain runtime relocations. @var{decl} may be
7351 null, in which case read-write data should be assumed.
7353 The default version of this function handles choosing code vs data,
7354 read-only vs read-write data, and @code{flag_pic}. You should only
7355 need to override this if your target has special flags that might be
7356 set via @code{__attribute__}.
7359 @deftypefn {Target Hook} int TARGET_ASM_RECORD_GCC_SWITCHES (print_switch_type @var{type}, const char *@var{text})
7360 Provides the target with the ability to record the gcc command line
7361 switches that have been passed to the compiler, and options that are
7362 enabled. The @var{type} argument specifies what is being recorded.
7363 It can take the following values:
7366 @item SWITCH_TYPE_PASSED
7367 @var{text} is a command line switch that has been set by the user.
7369 @item SWITCH_TYPE_ENABLED
7370 @var{text} is an option which has been enabled. This might be as a
7371 direct result of a command line switch, or because it is enabled by
7372 default or because it has been enabled as a side effect of a different
7373 command line switch. For example, the @option{-O2} switch enables
7374 various different individual optimization passes.
7376 @item SWITCH_TYPE_DESCRIPTIVE
7377 @var{text} is either NULL or some descriptive text which should be
7378 ignored. If @var{text} is NULL then it is being used to warn the
7379 target hook that either recording is starting or ending. The first
7380 time @var{type} is SWITCH_TYPE_DESCRIPTIVE and @var{text} is NULL, the
7381 warning is for start up and the second time the warning is for
7382 wind down. This feature is to allow the target hook to make any
7383 necessary preparations before it starts to record switches and to
7384 perform any necessary tidying up after it has finished recording
7387 @item SWITCH_TYPE_LINE_START
7388 This option can be ignored by this target hook.
7390 @item SWITCH_TYPE_LINE_END
7391 This option can be ignored by this target hook.
7394 The hook's return value must be zero. Other return values may be
7395 supported in the future.
7397 By default this hook is set to NULL, but an example implementation is
7398 provided for ELF based targets. Called @var{elf_record_gcc_switches},
7399 it records the switches as ASCII text inside a new, string mergeable
7400 section in the assembler output file. The name of the new section is
7401 provided by the @code{TARGET_ASM_RECORD_GCC_SWITCHES_SECTION} target
7405 @deftypevr {Target Hook} {const char *} TARGET_ASM_RECORD_GCC_SWITCHES_SECTION
7406 This is the name of the section that will be created by the example
7407 ELF implementation of the @code{TARGET_ASM_RECORD_GCC_SWITCHES} target
7413 @subsection Output of Data
7416 @deftypevr {Target Hook} {const char *} TARGET_ASM_BYTE_OP
7417 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_HI_OP
7418 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_SI_OP
7419 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_DI_OP
7420 @deftypevrx {Target Hook} {const char *} TARGET_ASM_ALIGNED_TI_OP
7421 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_HI_OP
7422 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_SI_OP
7423 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_DI_OP
7424 @deftypevrx {Target Hook} {const char *} TARGET_ASM_UNALIGNED_TI_OP
7425 These hooks specify assembly directives for creating certain kinds
7426 of integer object. The @code{TARGET_ASM_BYTE_OP} directive creates a
7427 byte-sized object, the @code{TARGET_ASM_ALIGNED_HI_OP} one creates an
7428 aligned two-byte object, and so on. Any of the hooks may be
7429 @code{NULL}, indicating that no suitable directive is available.
7431 The compiler will print these strings at the start of a new line,
7432 followed immediately by the object's initial value. In most cases,
7433 the string should contain a tab, a pseudo-op, and then another tab.
7436 @deftypefn {Target Hook} bool TARGET_ASM_INTEGER (rtx @var{x}, unsigned int @var{size}, int @var{aligned_p})
7437 The @code{assemble_integer} function uses this hook to output an
7438 integer object. @var{x} is the object's value, @var{size} is its size
7439 in bytes and @var{aligned_p} indicates whether it is aligned. The
7440 function should return @code{true} if it was able to output the
7441 object. If it returns false, @code{assemble_integer} will try to
7442 split the object into smaller parts.
7444 The default implementation of this hook will use the
7445 @code{TARGET_ASM_BYTE_OP} family of strings, returning @code{false}
7446 when the relevant string is @code{NULL}.
7449 @deftypefn {Target Hook} void TARGET_ASM_DECL_END (void)
7450 Define this hook if the target assembler requires a special marker to
7451 terminate an initialized variable declaration.
7454 @deftypefn {Target Hook} bool TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA (FILE *@var{file}, rtx @var{x})
7455 A target hook to recognize @var{rtx} patterns that @code{output_addr_const}
7456 can't deal with, and output assembly code to @var{file} corresponding to
7457 the pattern @var{x}. This may be used to allow machine-dependent
7458 @code{UNSPEC}s to appear within constants.
7460 If target hook fails to recognize a pattern, it must return @code{false},
7461 so that a standard error message is printed. If it prints an error message
7462 itself, by calling, for example, @code{output_operand_lossage}, it may just
7466 @defmac ASM_OUTPUT_ASCII (@var{stream}, @var{ptr}, @var{len})
7467 A C statement to output to the stdio stream @var{stream} an assembler
7468 instruction to assemble a string constant containing the @var{len}
7469 bytes at @var{ptr}. @var{ptr} will be a C expression of type
7470 @code{char *} and @var{len} a C expression of type @code{int}.
7472 If the assembler has a @code{.ascii} pseudo-op as found in the
7473 Berkeley Unix assembler, do not define the macro
7474 @code{ASM_OUTPUT_ASCII}.
7477 @defmac ASM_OUTPUT_FDESC (@var{stream}, @var{decl}, @var{n})
7478 A C statement to output word @var{n} of a function descriptor for
7479 @var{decl}. This must be defined if @code{TARGET_VTABLE_USES_DESCRIPTORS}
7480 is defined, and is otherwise unused.
7483 @defmac CONSTANT_POOL_BEFORE_FUNCTION
7484 You may define this macro as a C expression. You should define the
7485 expression to have a nonzero value if GCC should output the constant
7486 pool for a function before the code for the function, or a zero value if
7487 GCC should output the constant pool after the function. If you do
7488 not define this macro, the usual case, GCC will output the constant
7489 pool before the function.
7492 @defmac ASM_OUTPUT_POOL_PROLOGUE (@var{file}, @var{funname}, @var{fundecl}, @var{size})
7493 A C statement to output assembler commands to define the start of the
7494 constant pool for a function. @var{funname} is a string giving
7495 the name of the function. Should the return type of the function
7496 be required, it can be obtained via @var{fundecl}. @var{size}
7497 is the size, in bytes, of the constant pool that will be written
7498 immediately after this call.
7500 If no constant-pool prefix is required, the usual case, this macro need
7504 @defmac ASM_OUTPUT_SPECIAL_POOL_ENTRY (@var{file}, @var{x}, @var{mode}, @var{align}, @var{labelno}, @var{jumpto})
7505 A C statement (with or without semicolon) to output a constant in the
7506 constant pool, if it needs special treatment. (This macro need not do
7507 anything for RTL expressions that can be output normally.)
7509 The argument @var{file} is the standard I/O stream to output the
7510 assembler code on. @var{x} is the RTL expression for the constant to
7511 output, and @var{mode} is the machine mode (in case @var{x} is a
7512 @samp{const_int}). @var{align} is the required alignment for the value
7513 @var{x}; you should output an assembler directive to force this much
7516 The argument @var{labelno} is a number to use in an internal label for
7517 the address of this pool entry. The definition of this macro is
7518 responsible for outputting the label definition at the proper place.
7519 Here is how to do this:
7522 @code{(*targetm.asm_out.internal_label)} (@var{file}, "LC", @var{labelno});
7525 When you output a pool entry specially, you should end with a
7526 @code{goto} to the label @var{jumpto}. This will prevent the same pool
7527 entry from being output a second time in the usual manner.
7529 You need not define this macro if it would do nothing.
7532 @defmac ASM_OUTPUT_POOL_EPILOGUE (@var{file} @var{funname} @var{fundecl} @var{size})
7533 A C statement to output assembler commands to at the end of the constant
7534 pool for a function. @var{funname} is a string giving the name of the
7535 function. Should the return type of the function be required, you can
7536 obtain it via @var{fundecl}. @var{size} is the size, in bytes, of the
7537 constant pool that GCC wrote immediately before this call.
7539 If no constant-pool epilogue is required, the usual case, you need not
7543 @defmac IS_ASM_LOGICAL_LINE_SEPARATOR (@var{C}, @var{STR})
7544 Define this macro as a C expression which is nonzero if @var{C} is
7545 used as a logical line separator by the assembler. @var{STR} points
7546 to the position in the string where @var{C} was found; this can be used if
7547 a line separator uses multiple characters.
7549 If you do not define this macro, the default is that only
7550 the character @samp{;} is treated as a logical line separator.
7553 @deftypevr {Target Hook} {const char *} TARGET_ASM_OPEN_PAREN
7554 @deftypevrx {Target Hook} {const char *} TARGET_ASM_CLOSE_PAREN
7555 These target hooks are C string constants, describing the syntax in the
7556 assembler for grouping arithmetic expressions. If not overridden, they
7557 default to normal parentheses, which is correct for most assemblers.
7560 These macros are provided by @file{real.h} for writing the definitions
7561 of @code{ASM_OUTPUT_DOUBLE} and the like:
7563 @defmac REAL_VALUE_TO_TARGET_SINGLE (@var{x}, @var{l})
7564 @defmacx REAL_VALUE_TO_TARGET_DOUBLE (@var{x}, @var{l})
7565 @defmacx REAL_VALUE_TO_TARGET_LONG_DOUBLE (@var{x}, @var{l})
7566 @defmacx REAL_VALUE_TO_TARGET_DECIMAL32 (@var{x}, @var{l})
7567 @defmacx REAL_VALUE_TO_TARGET_DECIMAL64 (@var{x}, @var{l})
7568 @defmacx REAL_VALUE_TO_TARGET_DECIMAL128 (@var{x}, @var{l})
7569 These translate @var{x}, of type @code{REAL_VALUE_TYPE}, to the
7570 target's floating point representation, and store its bit pattern in
7571 the variable @var{l}. For @code{REAL_VALUE_TO_TARGET_SINGLE} and
7572 @code{REAL_VALUE_TO_TARGET_DECIMAL32}, this variable should be a
7573 simple @code{long int}. For the others, it should be an array of
7574 @code{long int}. The number of elements in this array is determined
7575 by the size of the desired target floating point data type: 32 bits of
7576 it go in each @code{long int} array element. Each array element holds
7577 32 bits of the result, even if @code{long int} is wider than 32 bits
7578 on the host machine.
7580 The array element values are designed so that you can print them out
7581 using @code{fprintf} in the order they should appear in the target
7585 @node Uninitialized Data
7586 @subsection Output of Uninitialized Variables
7588 Each of the macros in this section is used to do the whole job of
7589 outputting a single uninitialized variable.
7591 @defmac ASM_OUTPUT_COMMON (@var{stream}, @var{name}, @var{size}, @var{rounded})
7592 A C statement (sans semicolon) to output to the stdio stream
7593 @var{stream} the assembler definition of a common-label named
7594 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7595 is the size rounded up to whatever alignment the caller wants. It is
7596 possible that @var{size} may be zero, for instance if a struct with no
7597 other member than a zero-length array is defined. In this case, the
7598 backend must output a symbol definition that allocates at least one
7599 byte, both so that the address of the resulting object does not compare
7600 equal to any other, and because some object formats cannot even express
7601 the concept of a zero-sized common symbol, as that is how they represent
7602 an ordinary undefined external.
7604 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7605 output the name itself; before and after that, output the additional
7606 assembler syntax for defining the name, and a newline.
7608 This macro controls how the assembler definitions of uninitialized
7609 common global variables are output.
7612 @defmac ASM_OUTPUT_ALIGNED_COMMON (@var{stream}, @var{name}, @var{size}, @var{alignment})
7613 Like @code{ASM_OUTPUT_COMMON} except takes the required alignment as a
7614 separate, explicit argument. If you define this macro, it is used in
7615 place of @code{ASM_OUTPUT_COMMON}, and gives you more flexibility in
7616 handling the required alignment of the variable. The alignment is specified
7617 as the number of bits.
7620 @defmac ASM_OUTPUT_ALIGNED_DECL_COMMON (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7621 Like @code{ASM_OUTPUT_ALIGNED_COMMON} except that @var{decl} of the
7622 variable to be output, if there is one, or @code{NULL_TREE} if there
7623 is no corresponding variable. If you define this macro, GCC will use it
7624 in place of both @code{ASM_OUTPUT_COMMON} and
7625 @code{ASM_OUTPUT_ALIGNED_COMMON}. Define this macro when you need to see
7626 the variable's decl in order to chose what to output.
7629 @defmac ASM_OUTPUT_ALIGNED_BSS (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7630 A C statement (sans semicolon) to output to the stdio stream
7631 @var{stream} the assembler definition of uninitialized global @var{decl} named
7632 @var{name} whose size is @var{size} bytes. The variable @var{alignment}
7633 is the alignment specified as the number of bits.
7635 Try to use function @code{asm_output_aligned_bss} defined in file
7636 @file{varasm.c} when defining this macro. If unable, use the expression
7637 @code{assemble_name (@var{stream}, @var{name})} to output the name itself;
7638 before and after that, output the additional assembler syntax for defining
7639 the name, and a newline.
7641 There are two ways of handling global BSS@. One is to define this macro.
7642 The other is to have @code{TARGET_ASM_SELECT_SECTION} return a
7643 switchable BSS section (@pxref{TARGET_HAVE_SWITCHABLE_BSS_SECTIONS}).
7644 You do not need to do both.
7646 Some languages do not have @code{common} data, and require a
7647 non-common form of global BSS in order to handle uninitialized globals
7648 efficiently. C++ is one example of this. However, if the target does
7649 not support global BSS, the front end may choose to make globals
7650 common in order to save space in the object file.
7653 @defmac ASM_OUTPUT_LOCAL (@var{stream}, @var{name}, @var{size}, @var{rounded})
7654 A C statement (sans semicolon) to output to the stdio stream
7655 @var{stream} the assembler definition of a local-common-label named
7656 @var{name} whose size is @var{size} bytes. The variable @var{rounded}
7657 is the size rounded up to whatever alignment the caller wants.
7659 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7660 output the name itself; before and after that, output the additional
7661 assembler syntax for defining the name, and a newline.
7663 This macro controls how the assembler definitions of uninitialized
7664 static variables are output.
7667 @defmac ASM_OUTPUT_ALIGNED_LOCAL (@var{stream}, @var{name}, @var{size}, @var{alignment})
7668 Like @code{ASM_OUTPUT_LOCAL} except takes the required alignment as a
7669 separate, explicit argument. If you define this macro, it is used in
7670 place of @code{ASM_OUTPUT_LOCAL}, and gives you more flexibility in
7671 handling the required alignment of the variable. The alignment is specified
7672 as the number of bits.
7675 @defmac ASM_OUTPUT_ALIGNED_DECL_LOCAL (@var{stream}, @var{decl}, @var{name}, @var{size}, @var{alignment})
7676 Like @code{ASM_OUTPUT_ALIGNED_DECL} except that @var{decl} of the
7677 variable to be output, if there is one, or @code{NULL_TREE} if there
7678 is no corresponding variable. If you define this macro, GCC will use it
7679 in place of both @code{ASM_OUTPUT_DECL} and
7680 @code{ASM_OUTPUT_ALIGNED_DECL}. Define this macro when you need to see
7681 the variable's decl in order to chose what to output.
7685 @subsection Output and Generation of Labels
7687 @c prevent bad page break with this line
7688 This is about outputting labels.
7690 @findex assemble_name
7691 @defmac ASM_OUTPUT_LABEL (@var{stream}, @var{name})
7692 A C statement (sans semicolon) to output to the stdio stream
7693 @var{stream} the assembler definition of a label named @var{name}.
7694 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7695 output the name itself; before and after that, output the additional
7696 assembler syntax for defining the name, and a newline. A default
7697 definition of this macro is provided which is correct for most systems.
7700 @defmac ASM_OUTPUT_FUNCTION_LABEL (@var{stream}, @var{name}, @var{decl})
7701 A C statement (sans semicolon) to output to the stdio stream
7702 @var{stream} the assembler definition of a label named @var{name} of
7704 Use the expression @code{assemble_name (@var{stream}, @var{name})} to
7705 output the name itself; before and after that, output the additional
7706 assembler syntax for defining the name, and a newline. A default
7707 definition of this macro is provided which is correct for most systems.
7709 If this macro is not defined, then the function name is defined in the
7710 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
7713 @findex assemble_name_raw
7714 @defmac ASM_OUTPUT_INTERNAL_LABEL (@var{stream}, @var{name})
7715 Identical to @code{ASM_OUTPUT_LABEL}, except that @var{name} is known
7716 to refer to a compiler-generated label. The default definition uses
7717 @code{assemble_name_raw}, which is like @code{assemble_name} except
7718 that it is more efficient.
7722 A C string containing the appropriate assembler directive to specify the
7723 size of a symbol, without any arguments. On systems that use ELF, the
7724 default (in @file{config/elfos.h}) is @samp{"\t.size\t"}; on other
7725 systems, the default is not to define this macro.
7727 Define this macro only if it is correct to use the default definitions
7728 of @code{ASM_OUTPUT_SIZE_DIRECTIVE} and @code{ASM_OUTPUT_MEASURED_SIZE}
7729 for your system. If you need your own custom definitions of those
7730 macros, or if you do not need explicit symbol sizes at all, do not
7734 @defmac ASM_OUTPUT_SIZE_DIRECTIVE (@var{stream}, @var{name}, @var{size})
7735 A C statement (sans semicolon) to output to the stdio stream
7736 @var{stream} a directive telling the assembler that the size of the
7737 symbol @var{name} is @var{size}. @var{size} is a @code{HOST_WIDE_INT}.
7738 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
7742 @defmac ASM_OUTPUT_MEASURED_SIZE (@var{stream}, @var{name})
7743 A C statement (sans semicolon) to output to the stdio stream
7744 @var{stream} a directive telling the assembler to calculate the size of
7745 the symbol @var{name} by subtracting its address from the current
7748 If you define @code{SIZE_ASM_OP}, a default definition of this macro is
7749 provided. The default assumes that the assembler recognizes a special
7750 @samp{.} symbol as referring to the current address, and can calculate
7751 the difference between this and another symbol. If your assembler does
7752 not recognize @samp{.} or cannot do calculations with it, you will need
7753 to redefine @code{ASM_OUTPUT_MEASURED_SIZE} to use some other technique.
7756 @defmac NO_DOLLAR_IN_LABEL
7757 Define this macro if the assembler does not accept the character
7758 @samp{$} in label names. By default constructors and destructors in
7759 G++ have @samp{$} in the identifiers. If this macro is defined,
7760 @samp{.} is used instead.
7763 @defmac NO_DOT_IN_LABEL
7764 Define this macro if the assembler does not accept the character
7765 @samp{.} in label names. By default constructors and destructors in G++
7766 have names that use @samp{.}. If this macro is defined, these names
7767 are rewritten to avoid @samp{.}.
7771 A C string containing the appropriate assembler directive to specify the
7772 type of a symbol, without any arguments. On systems that use ELF, the
7773 default (in @file{config/elfos.h}) is @samp{"\t.type\t"}; on other
7774 systems, the default is not to define this macro.
7776 Define this macro only if it is correct to use the default definition of
7777 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
7778 custom definition of this macro, or if you do not need explicit symbol
7779 types at all, do not define this macro.
7782 @defmac TYPE_OPERAND_FMT
7783 A C string which specifies (using @code{printf} syntax) the format of
7784 the second operand to @code{TYPE_ASM_OP}. On systems that use ELF, the
7785 default (in @file{config/elfos.h}) is @samp{"@@%s"}; on other systems,
7786 the default is not to define this macro.
7788 Define this macro only if it is correct to use the default definition of
7789 @code{ASM_OUTPUT_TYPE_DIRECTIVE} for your system. If you need your own
7790 custom definition of this macro, or if you do not need explicit symbol
7791 types at all, do not define this macro.
7794 @defmac ASM_OUTPUT_TYPE_DIRECTIVE (@var{stream}, @var{type})
7795 A C statement (sans semicolon) to output to the stdio stream
7796 @var{stream} a directive telling the assembler that the type of the
7797 symbol @var{name} is @var{type}. @var{type} is a C string; currently,
7798 that string is always either @samp{"function"} or @samp{"object"}, but
7799 you should not count on this.
7801 If you define @code{TYPE_ASM_OP} and @code{TYPE_OPERAND_FMT}, a default
7802 definition of this macro is provided.
7805 @defmac ASM_DECLARE_FUNCTION_NAME (@var{stream}, @var{name}, @var{decl})
7806 A C statement (sans semicolon) to output to the stdio stream
7807 @var{stream} any text necessary for declaring the name @var{name} of a
7808 function which is being defined. This macro is responsible for
7809 outputting the label definition (perhaps using
7810 @code{ASM_OUTPUT_FUNCTION_LABEL}). The argument @var{decl} is the
7811 @code{FUNCTION_DECL} tree node representing the function.
7813 If this macro is not defined, then the function name is defined in the
7814 usual manner as a label (by means of @code{ASM_OUTPUT_FUNCTION_LABEL}).
7816 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in the definition
7820 @defmac ASM_DECLARE_FUNCTION_SIZE (@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 size of a function
7823 which is being defined. The argument @var{name} is the name of the
7824 function. The argument @var{decl} is the @code{FUNCTION_DECL} tree node
7825 representing the function.
7827 If this macro is not defined, then the function size is not defined.
7829 You may wish to use @code{ASM_OUTPUT_MEASURED_SIZE} in the definition
7833 @defmac ASM_DECLARE_OBJECT_NAME (@var{stream}, @var{name}, @var{decl})
7834 A C statement (sans semicolon) to output to the stdio stream
7835 @var{stream} any text necessary for declaring the name @var{name} of an
7836 initialized variable which is being defined. This macro must output the
7837 label definition (perhaps using @code{ASM_OUTPUT_LABEL}). The argument
7838 @var{decl} is the @code{VAR_DECL} tree node representing the variable.
7840 If this macro is not defined, then the variable name is defined in the
7841 usual manner as a label (by means of @code{ASM_OUTPUT_LABEL}).
7843 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} and/or
7844 @code{ASM_OUTPUT_SIZE_DIRECTIVE} in the definition of this macro.
7847 @deftypefn {Target Hook} void TARGET_ASM_DECLARE_CONSTANT_NAME (FILE *@var{file}, const char *@var{name}, const_tree @var{expr}, HOST_WIDE_INT @var{size})
7848 A target hook to output to the stdio stream @var{file} any text necessary
7849 for declaring the name @var{name} of a constant which is being defined. This
7850 target hook is responsible for outputting the label definition (perhaps using
7851 @code{assemble_label}). The argument @var{exp} is the value of the constant,
7852 and @var{size} is the size of the constant in bytes. The @var{name}
7853 will be an internal label.
7855 The default version of this target hook, define the @var{name} in the
7856 usual manner as a label (by means of @code{assemble_label}).
7858 You may wish to use @code{ASM_OUTPUT_TYPE_DIRECTIVE} in this target hook.
7861 @defmac ASM_DECLARE_REGISTER_GLOBAL (@var{stream}, @var{decl}, @var{regno}, @var{name})
7862 A C statement (sans semicolon) to output to the stdio stream
7863 @var{stream} any text necessary for claiming a register @var{regno}
7864 for a global variable @var{decl} with name @var{name}.
7866 If you don't define this macro, that is equivalent to defining it to do
7870 @defmac ASM_FINISH_DECLARE_OBJECT (@var{stream}, @var{decl}, @var{toplevel}, @var{atend})
7871 A C statement (sans semicolon) to finish up declaring a variable name
7872 once the compiler has processed its initializer fully and thus has had a
7873 chance to determine the size of an array when controlled by an
7874 initializer. This is used on systems where it's necessary to declare
7875 something about the size of the object.
7877 If you don't define this macro, that is equivalent to defining it to do
7880 You may wish to use @code{ASM_OUTPUT_SIZE_DIRECTIVE} and/or
7881 @code{ASM_OUTPUT_MEASURED_SIZE} in the definition of this macro.
7884 @deftypefn {Target Hook} void TARGET_ASM_GLOBALIZE_LABEL (FILE *@var{stream}, const char *@var{name})
7885 This target hook is a function to output to the stdio stream
7886 @var{stream} some commands that will make the label @var{name} global;
7887 that is, available for reference from other files.
7889 The default implementation relies on a proper definition of
7890 @code{GLOBAL_ASM_OP}.
7893 @deftypefn {Target Hook} void TARGET_ASM_GLOBALIZE_DECL_NAME (FILE *@var{stream}, tree @var{decl})
7894 This target hook is a function to output to the stdio stream
7895 @var{stream} some commands that will make the name associated with @var{decl}
7896 global; that is, available for reference from other files.
7898 The default implementation uses the TARGET_ASM_GLOBALIZE_LABEL target hook.
7901 @deftypefn {Target Hook} void TARGET_ASM_ASSEMBLE_UNDEFINED_DECL (FILE *@var{stream}, const char *@var{name}, const_tree @var{decl})
7902 This target hook is a function to output to the stdio stream
7903 @var{stream} some commands that will declare the name associated with
7904 @var{decl} which is not defined in the current translation unit. Most
7905 assemblers do not require anything to be output in this case.
7908 @defmac ASM_WEAKEN_LABEL (@var{stream}, @var{name})
7909 A C statement (sans semicolon) to output to the stdio stream
7910 @var{stream} some commands that will make the label @var{name} weak;
7911 that is, available for reference from other files but only used if
7912 no other definition is available. Use the expression
7913 @code{assemble_name (@var{stream}, @var{name})} to output the name
7914 itself; before and after that, output the additional assembler syntax
7915 for making that name weak, and a newline.
7917 If you don't define this macro or @code{ASM_WEAKEN_DECL}, GCC will not
7918 support weak symbols and you should not define the @code{SUPPORTS_WEAK}
7922 @defmac ASM_WEAKEN_DECL (@var{stream}, @var{decl}, @var{name}, @var{value})
7923 Combines (and replaces) the function of @code{ASM_WEAKEN_LABEL} and
7924 @code{ASM_OUTPUT_WEAK_ALIAS}, allowing access to the associated function
7925 or variable decl. If @var{value} is not @code{NULL}, this C statement
7926 should output to the stdio stream @var{stream} assembler code which
7927 defines (equates) the weak symbol @var{name} to have the value
7928 @var{value}. If @var{value} is @code{NULL}, it should output commands
7929 to make @var{name} weak.
7932 @defmac ASM_OUTPUT_WEAKREF (@var{stream}, @var{decl}, @var{name}, @var{value})
7933 Outputs a directive that enables @var{name} to be used to refer to
7934 symbol @var{value} with weak-symbol semantics. @code{decl} is the
7935 declaration of @code{name}.
7938 @defmac SUPPORTS_WEAK
7939 A preprocessor constant expression which evaluates to true if the target
7940 supports weak symbols.
7942 If you don't define this macro, @file{defaults.h} provides a default
7943 definition. If either @code{ASM_WEAKEN_LABEL} or @code{ASM_WEAKEN_DECL}
7944 is defined, the default definition is @samp{1}; otherwise, it is @samp{0}.
7947 @defmac TARGET_SUPPORTS_WEAK
7948 A C expression which evaluates to true if the target supports weak symbols.
7950 If you don't define this macro, @file{defaults.h} provides a default
7951 definition. The default definition is @samp{(SUPPORTS_WEAK)}. Define
7952 this macro if you want to control weak symbol support with a compiler
7953 flag such as @option{-melf}.
7956 @defmac MAKE_DECL_ONE_ONLY (@var{decl})
7957 A C statement (sans semicolon) to mark @var{decl} to be emitted as a
7958 public symbol such that extra copies in multiple translation units will
7959 be discarded by the linker. Define this macro if your object file
7960 format provides support for this concept, such as the @samp{COMDAT}
7961 section flags in the Microsoft Windows PE/COFF format, and this support
7962 requires changes to @var{decl}, such as putting it in a separate section.
7965 @defmac SUPPORTS_ONE_ONLY
7966 A C expression which evaluates to true if the target supports one-only
7969 If you don't define this macro, @file{varasm.c} provides a default
7970 definition. If @code{MAKE_DECL_ONE_ONLY} is defined, the default
7971 definition is @samp{1}; otherwise, it is @samp{0}. Define this macro if
7972 you want to control one-only symbol support with a compiler flag, or if
7973 setting the @code{DECL_ONE_ONLY} flag is enough to mark a declaration to
7974 be emitted as one-only.
7977 @deftypefn {Target Hook} void TARGET_ASM_ASSEMBLE_VISIBILITY (tree @var{decl}, int @var{visibility})
7978 This target hook is a function to output to @var{asm_out_file} some
7979 commands that will make the symbol(s) associated with @var{decl} have
7980 hidden, protected or internal visibility as specified by @var{visibility}.
7983 @defmac TARGET_WEAK_NOT_IN_ARCHIVE_TOC
7984 A C expression that evaluates to true if the target's linker expects
7985 that weak symbols do not appear in a static archive's table of contents.
7986 The default is @code{0}.
7988 Leaving weak symbols out of an archive's table of contents means that,
7989 if a symbol will only have a definition in one translation unit and
7990 will have undefined references from other translation units, that
7991 symbol should not be weak. Defining this macro to be nonzero will
7992 thus have the effect that certain symbols that would normally be weak
7993 (explicit template instantiations, and vtables for polymorphic classes
7994 with noninline key methods) will instead be nonweak.
7996 The C++ ABI requires this macro to be zero. Define this macro for
7997 targets where full C++ ABI compliance is impossible and where linker
7998 restrictions require weak symbols to be left out of a static archive's
8002 @defmac ASM_OUTPUT_EXTERNAL (@var{stream}, @var{decl}, @var{name})
8003 A C statement (sans semicolon) to output to the stdio stream
8004 @var{stream} any text necessary for declaring the name of an external
8005 symbol named @var{name} which is referenced in this compilation but
8006 not defined. The value of @var{decl} is the tree node for the
8009 This macro need not be defined if it does not need to output anything.
8010 The GNU assembler and most Unix assemblers don't require anything.
8013 @deftypefn {Target Hook} void TARGET_ASM_EXTERNAL_LIBCALL (rtx @var{symref})
8014 This target hook is a function to output to @var{asm_out_file} an assembler
8015 pseudo-op to declare a library function name external. The name of the
8016 library function is given by @var{symref}, which is a @code{symbol_ref}.
8019 @deftypefn {Target Hook} void TARGET_ASM_MARK_DECL_PRESERVED (const char *@var{symbol})
8020 This target hook is a function to output to @var{asm_out_file} an assembler
8021 directive to annotate @var{symbol} as used. The Darwin target uses the
8022 .no_dead_code_strip directive.
8025 @defmac ASM_OUTPUT_LABELREF (@var{stream}, @var{name})
8026 A C statement (sans semicolon) to output to the stdio stream
8027 @var{stream} a reference in assembler syntax to a label named
8028 @var{name}. This should add @samp{_} to the front of the name, if that
8029 is customary on your operating system, as it is in most Berkeley Unix
8030 systems. This macro is used in @code{assemble_name}.
8033 @deftypefn {Target Hook} tree TARGET_MANGLE_ASSEMBLER_NAME (const char *@var{name})
8034 Given a symbol @var{name}, perform same mangling as @code{varasm.c}'s @code{assemble_name}, but in memory rather than to a file stream, returning result as an @code{IDENTIFIER_NODE}. Required for correct LTO symtabs. The default implementation calls the @code{TARGET_STRIP_NAME_ENCODING} hook and then prepends the @code{USER_LABEL_PREFIX}, if any.
8037 @defmac ASM_OUTPUT_SYMBOL_REF (@var{stream}, @var{sym})
8038 A C statement (sans semicolon) to output a reference to
8039 @code{SYMBOL_REF} @var{sym}. If not defined, @code{assemble_name}
8040 will be used to output the name of the symbol. This macro may be used
8041 to modify the way a symbol is referenced depending on information
8042 encoded by @code{TARGET_ENCODE_SECTION_INFO}.
8045 @defmac ASM_OUTPUT_LABEL_REF (@var{stream}, @var{buf})
8046 A C statement (sans semicolon) to output a reference to @var{buf}, the
8047 result of @code{ASM_GENERATE_INTERNAL_LABEL}. If not defined,
8048 @code{assemble_name} will be used to output the name of the symbol.
8049 This macro is not used by @code{output_asm_label}, or the @code{%l}
8050 specifier that calls it; the intention is that this macro should be set
8051 when it is necessary to output a label differently when its address is
8055 @deftypefn {Target Hook} void TARGET_ASM_INTERNAL_LABEL (FILE *@var{stream}, const char *@var{prefix}, unsigned long @var{labelno})
8056 A function to output to the stdio stream @var{stream} a label whose
8057 name is made from the string @var{prefix} and the number @var{labelno}.
8059 It is absolutely essential that these labels be distinct from the labels
8060 used for user-level functions and variables. Otherwise, certain programs
8061 will have name conflicts with internal labels.
8063 It is desirable to exclude internal labels from the symbol table of the
8064 object file. Most assemblers have a naming convention for labels that
8065 should be excluded; on many systems, the letter @samp{L} at the
8066 beginning of a label has this effect. You should find out what
8067 convention your system uses, and follow it.
8069 The default version of this function utilizes @code{ASM_GENERATE_INTERNAL_LABEL}.
8072 @defmac ASM_OUTPUT_DEBUG_LABEL (@var{stream}, @var{prefix}, @var{num})
8073 A C statement to output to the stdio stream @var{stream} a debug info
8074 label whose name is made from the string @var{prefix} and the number
8075 @var{num}. This is useful for VLIW targets, where debug info labels
8076 may need to be treated differently than branch target labels. On some
8077 systems, branch target labels must be at the beginning of instruction
8078 bundles, but debug info labels can occur in the middle of instruction
8081 If this macro is not defined, then @code{(*targetm.asm_out.internal_label)} will be
8085 @defmac ASM_GENERATE_INTERNAL_LABEL (@var{string}, @var{prefix}, @var{num})
8086 A C statement to store into the string @var{string} a label whose name
8087 is made from the string @var{prefix} and the number @var{num}.
8089 This string, when output subsequently by @code{assemble_name}, should
8090 produce the output that @code{(*targetm.asm_out.internal_label)} would produce
8091 with the same @var{prefix} and @var{num}.
8093 If the string begins with @samp{*}, then @code{assemble_name} will
8094 output the rest of the string unchanged. It is often convenient for
8095 @code{ASM_GENERATE_INTERNAL_LABEL} to use @samp{*} in this way. If the
8096 string doesn't start with @samp{*}, then @code{ASM_OUTPUT_LABELREF} gets
8097 to output the string, and may change it. (Of course,
8098 @code{ASM_OUTPUT_LABELREF} is also part of your machine description, so
8099 you should know what it does on your machine.)
8102 @defmac ASM_FORMAT_PRIVATE_NAME (@var{outvar}, @var{name}, @var{number})
8103 A C expression to assign to @var{outvar} (which is a variable of type
8104 @code{char *}) a newly allocated string made from the string
8105 @var{name} and the number @var{number}, with some suitable punctuation
8106 added. Use @code{alloca} to get space for the string.
8108 The string will be used as an argument to @code{ASM_OUTPUT_LABELREF} to
8109 produce an assembler label for an internal static variable whose name is
8110 @var{name}. Therefore, the string must be such as to result in valid
8111 assembler code. The argument @var{number} is different each time this
8112 macro is executed; it prevents conflicts between similarly-named
8113 internal static variables in different scopes.
8115 Ideally this string should not be a valid C identifier, to prevent any
8116 conflict with the user's own symbols. Most assemblers allow periods
8117 or percent signs in assembler symbols; putting at least one of these
8118 between the name and the number will suffice.
8120 If this macro is not defined, a default definition will be provided
8121 which is correct for most systems.
8124 @defmac ASM_OUTPUT_DEF (@var{stream}, @var{name}, @var{value})
8125 A C statement to output to the stdio stream @var{stream} assembler code
8126 which defines (equates) the symbol @var{name} to have the value @var{value}.
8129 If @code{SET_ASM_OP} is defined, a default definition is provided which is
8130 correct for most systems.
8133 @defmac ASM_OUTPUT_DEF_FROM_DECLS (@var{stream}, @var{decl_of_name}, @var{decl_of_value})
8134 A C statement to output to the stdio stream @var{stream} assembler code
8135 which defines (equates) the symbol whose tree node is @var{decl_of_name}
8136 to have the value of the tree node @var{decl_of_value}. This macro will
8137 be used in preference to @samp{ASM_OUTPUT_DEF} if it is defined and if
8138 the tree nodes are available.
8141 If @code{SET_ASM_OP} is defined, a default definition is provided which is
8142 correct for most systems.
8145 @defmac TARGET_DEFERRED_OUTPUT_DEFS (@var{decl_of_name}, @var{decl_of_value})
8146 A C statement that evaluates to true if the assembler code which defines
8147 (equates) the symbol whose tree node is @var{decl_of_name} to have the value
8148 of the tree node @var{decl_of_value} should be emitted near the end of the
8149 current compilation unit. The default is to not defer output of defines.
8150 This macro affects defines output by @samp{ASM_OUTPUT_DEF} and
8151 @samp{ASM_OUTPUT_DEF_FROM_DECLS}.
8154 @defmac ASM_OUTPUT_WEAK_ALIAS (@var{stream}, @var{name}, @var{value})
8155 A C statement to output to the stdio stream @var{stream} assembler code
8156 which defines (equates) the weak symbol @var{name} to have the value
8157 @var{value}. If @var{value} is @code{NULL}, it defines @var{name} as
8158 an undefined weak symbol.
8160 Define this macro if the target only supports weak aliases; define
8161 @code{ASM_OUTPUT_DEF} instead if possible.
8164 @defmac OBJC_GEN_METHOD_LABEL (@var{buf}, @var{is_inst}, @var{class_name}, @var{cat_name}, @var{sel_name})
8165 Define this macro to override the default assembler names used for
8166 Objective-C methods.
8168 The default name is a unique method number followed by the name of the
8169 class (e.g.@: @samp{_1_Foo}). For methods in categories, the name of
8170 the category is also included in the assembler name (e.g.@:
8173 These names are safe on most systems, but make debugging difficult since
8174 the method's selector is not present in the name. Therefore, particular
8175 systems define other ways of computing names.
8177 @var{buf} is an expression of type @code{char *} which gives you a
8178 buffer in which to store the name; its length is as long as
8179 @var{class_name}, @var{cat_name} and @var{sel_name} put together, plus
8180 50 characters extra.
8182 The argument @var{is_inst} specifies whether the method is an instance
8183 method or a class method; @var{class_name} is the name of the class;
8184 @var{cat_name} is the name of the category (or @code{NULL} if the method is not
8185 in a category); and @var{sel_name} is the name of the selector.
8187 On systems where the assembler can handle quoted names, you can use this
8188 macro to provide more human-readable names.
8191 @node Initialization
8192 @subsection How Initialization Functions Are Handled
8193 @cindex initialization routines
8194 @cindex termination routines
8195 @cindex constructors, output of
8196 @cindex destructors, output of
8198 The compiled code for certain languages includes @dfn{constructors}
8199 (also called @dfn{initialization routines})---functions to initialize
8200 data in the program when the program is started. These functions need
8201 to be called before the program is ``started''---that is to say, before
8202 @code{main} is called.
8204 Compiling some languages generates @dfn{destructors} (also called
8205 @dfn{termination routines}) that should be called when the program
8208 To make the initialization and termination functions work, the compiler
8209 must output something in the assembler code to cause those functions to
8210 be called at the appropriate time. When you port the compiler to a new
8211 system, you need to specify how to do this.
8213 There are two major ways that GCC currently supports the execution of
8214 initialization and termination functions. Each way has two variants.
8215 Much of the structure is common to all four variations.
8217 @findex __CTOR_LIST__
8218 @findex __DTOR_LIST__
8219 The linker must build two lists of these functions---a list of
8220 initialization functions, called @code{__CTOR_LIST__}, and a list of
8221 termination functions, called @code{__DTOR_LIST__}.
8223 Each list always begins with an ignored function pointer (which may hold
8224 0, @minus{}1, or a count of the function pointers after it, depending on
8225 the environment). This is followed by a series of zero or more function
8226 pointers to constructors (or destructors), followed by a function
8227 pointer containing zero.
8229 Depending on the operating system and its executable file format, either
8230 @file{crtstuff.c} or @file{libgcc2.c} traverses these lists at startup
8231 time and exit time. Constructors are called in reverse order of the
8232 list; destructors in forward order.
8234 The best way to handle static constructors works only for object file
8235 formats which provide arbitrarily-named sections. A section is set
8236 aside for a list of constructors, and another for a list of destructors.
8237 Traditionally these are called @samp{.ctors} and @samp{.dtors}. Each
8238 object file that defines an initialization function also puts a word in
8239 the constructor section to point to that function. The linker
8240 accumulates all these words into one contiguous @samp{.ctors} section.
8241 Termination functions are handled similarly.
8243 This method will be chosen as the default by @file{target-def.h} if
8244 @code{TARGET_ASM_NAMED_SECTION} is defined. A target that does not
8245 support arbitrary sections, but does support special designated
8246 constructor and destructor sections may define @code{CTORS_SECTION_ASM_OP}
8247 and @code{DTORS_SECTION_ASM_OP} to achieve the same effect.
8249 When arbitrary sections are available, there are two variants, depending
8250 upon how the code in @file{crtstuff.c} is called. On systems that
8251 support a @dfn{.init} section which is executed at program startup,
8252 parts of @file{crtstuff.c} are compiled into that section. The
8253 program is linked by the @command{gcc} driver like this:
8256 ld -o @var{output_file} crti.o crtbegin.o @dots{} -lgcc crtend.o crtn.o
8259 The prologue of a function (@code{__init}) appears in the @code{.init}
8260 section of @file{crti.o}; the epilogue appears in @file{crtn.o}. Likewise
8261 for the function @code{__fini} in the @dfn{.fini} section. Normally these
8262 files are provided by the operating system or by the GNU C library, but
8263 are provided by GCC for a few targets.
8265 The objects @file{crtbegin.o} and @file{crtend.o} are (for most targets)
8266 compiled from @file{crtstuff.c}. They contain, among other things, code
8267 fragments within the @code{.init} and @code{.fini} sections that branch
8268 to routines in the @code{.text} section. The linker will pull all parts
8269 of a section together, which results in a complete @code{__init} function
8270 that invokes the routines we need at startup.
8272 To use this variant, you must define the @code{INIT_SECTION_ASM_OP}
8275 If no init section is available, when GCC compiles any function called
8276 @code{main} (or more accurately, any function designated as a program
8277 entry point by the language front end calling @code{expand_main_function}),
8278 it inserts a procedure call to @code{__main} as the first executable code
8279 after the function prologue. The @code{__main} function is defined
8280 in @file{libgcc2.c} and runs the global constructors.
8282 In file formats that don't support arbitrary sections, there are again
8283 two variants. In the simplest variant, the GNU linker (GNU @code{ld})
8284 and an `a.out' format must be used. In this case,
8285 @code{TARGET_ASM_CONSTRUCTOR} is defined to produce a @code{.stabs}
8286 entry of type @samp{N_SETT}, referencing the name @code{__CTOR_LIST__},
8287 and with the address of the void function containing the initialization
8288 code as its value. The GNU linker recognizes this as a request to add
8289 the value to a @dfn{set}; the values are accumulated, and are eventually
8290 placed in the executable as a vector in the format described above, with
8291 a leading (ignored) count and a trailing zero element.
8292 @code{TARGET_ASM_DESTRUCTOR} is handled similarly. Since no init
8293 section is available, the absence of @code{INIT_SECTION_ASM_OP} causes
8294 the compilation of @code{main} to call @code{__main} as above, starting
8295 the initialization process.
8297 The last variant uses neither arbitrary sections nor the GNU linker.
8298 This is preferable when you want to do dynamic linking and when using
8299 file formats which the GNU linker does not support, such as `ECOFF'@. In
8300 this case, @code{TARGET_HAVE_CTORS_DTORS} is false, initialization and
8301 termination functions are recognized simply by their names. This requires
8302 an extra program in the linkage step, called @command{collect2}. This program
8303 pretends to be the linker, for use with GCC; it does its job by running
8304 the ordinary linker, but also arranges to include the vectors of
8305 initialization and termination functions. These functions are called
8306 via @code{__main} as described above. In order to use this method,
8307 @code{use_collect2} must be defined in the target in @file{config.gcc}.
8310 The following section describes the specific macros that control and
8311 customize the handling of initialization and termination functions.
8314 @node Macros for Initialization
8315 @subsection Macros Controlling Initialization Routines
8317 Here are the macros that control how the compiler handles initialization
8318 and termination functions:
8320 @defmac INIT_SECTION_ASM_OP
8321 If defined, a C string constant, including spacing, for the assembler
8322 operation to identify the following data as initialization code. If not
8323 defined, GCC will assume such a section does not exist. When you are
8324 using special sections for initialization and termination functions, this
8325 macro also controls how @file{crtstuff.c} and @file{libgcc2.c} arrange to
8326 run the initialization functions.
8329 @defmac HAS_INIT_SECTION
8330 If defined, @code{main} will not call @code{__main} as described above.
8331 This macro should be defined for systems that control start-up code
8332 on a symbol-by-symbol basis, such as OSF/1, and should not
8333 be defined explicitly for systems that support @code{INIT_SECTION_ASM_OP}.
8336 @defmac LD_INIT_SWITCH
8337 If defined, a C string constant for a switch that tells the linker that
8338 the following symbol is an initialization routine.
8341 @defmac LD_FINI_SWITCH
8342 If defined, a C string constant for a switch that tells the linker that
8343 the following symbol is a finalization routine.
8346 @defmac COLLECT_SHARED_INIT_FUNC (@var{stream}, @var{func})
8347 If defined, a C statement that will write a function that can be
8348 automatically called when a shared library is loaded. The function
8349 should call @var{func}, which takes no arguments. If not defined, and
8350 the object format requires an explicit initialization function, then a
8351 function called @code{_GLOBAL__DI} will be generated.
8353 This function and the following one are used by collect2 when linking a
8354 shared library that needs constructors or destructors, or has DWARF2
8355 exception tables embedded in the code.
8358 @defmac COLLECT_SHARED_FINI_FUNC (@var{stream}, @var{func})
8359 If defined, a C statement that will write a function that can be
8360 automatically called when a shared library is unloaded. The function
8361 should call @var{func}, which takes no arguments. If not defined, and
8362 the object format requires an explicit finalization function, then a
8363 function called @code{_GLOBAL__DD} will be generated.
8366 @defmac INVOKE__main
8367 If defined, @code{main} will call @code{__main} despite the presence of
8368 @code{INIT_SECTION_ASM_OP}. This macro should be defined for systems
8369 where the init section is not actually run automatically, but is still
8370 useful for collecting the lists of constructors and destructors.
8373 @defmac SUPPORTS_INIT_PRIORITY
8374 If nonzero, the C++ @code{init_priority} attribute is supported and the
8375 compiler should emit instructions to control the order of initialization
8376 of objects. If zero, the compiler will issue an error message upon
8377 encountering an @code{init_priority} attribute.
8380 @deftypevr {Target Hook} bool TARGET_HAVE_CTORS_DTORS
8381 This value is true if the target supports some ``native'' method of
8382 collecting constructors and destructors to be run at startup and exit.
8383 It is false if we must use @command{collect2}.
8386 @deftypefn {Target Hook} void TARGET_ASM_CONSTRUCTOR (rtx @var{symbol}, int @var{priority})
8387 If defined, a function that outputs assembler code to arrange to call
8388 the function referenced by @var{symbol} at initialization time.
8390 Assume that @var{symbol} is a @code{SYMBOL_REF} for a function taking
8391 no arguments and with no return value. If the target supports initialization
8392 priorities, @var{priority} is a value between 0 and @code{MAX_INIT_PRIORITY};
8393 otherwise it must be @code{DEFAULT_INIT_PRIORITY}.
8395 If this macro is not defined by the target, a suitable default will
8396 be chosen if (1) the target supports arbitrary section names, (2) the
8397 target defines @code{CTORS_SECTION_ASM_OP}, or (3) @code{USE_COLLECT2}
8401 @deftypefn {Target Hook} void TARGET_ASM_DESTRUCTOR (rtx @var{symbol}, int @var{priority})
8402 This is like @code{TARGET_ASM_CONSTRUCTOR} but used for termination
8403 functions rather than initialization functions.
8406 If @code{TARGET_HAVE_CTORS_DTORS} is true, the initialization routine
8407 generated for the generated object file will have static linkage.
8409 If your system uses @command{collect2} as the means of processing
8410 constructors, then that program normally uses @command{nm} to scan
8411 an object file for constructor functions to be called.
8413 On certain kinds of systems, you can define this macro to make
8414 @command{collect2} work faster (and, in some cases, make it work at all):
8416 @defmac OBJECT_FORMAT_COFF
8417 Define this macro if the system uses COFF (Common Object File Format)
8418 object files, so that @command{collect2} can assume this format and scan
8419 object files directly for dynamic constructor/destructor functions.
8421 This macro is effective only in a native compiler; @command{collect2} as
8422 part of a cross compiler always uses @command{nm} for the target machine.
8425 @defmac REAL_NM_FILE_NAME
8426 Define this macro as a C string constant containing the file name to use
8427 to execute @command{nm}. The default is to search the path normally for
8432 @command{collect2} calls @command{nm} to scan object files for static
8433 constructors and destructors and LTO info. By default, @option{-n} is
8434 passed. Define @code{NM_FLAGS} to a C string constant if other options
8435 are needed to get the same output format as GNU @command{nm -n}
8439 If your system supports shared libraries and has a program to list the
8440 dynamic dependencies of a given library or executable, you can define
8441 these macros to enable support for running initialization and
8442 termination functions in shared libraries:
8445 Define this macro to a C string constant containing the name of the program
8446 which lists dynamic dependencies, like @command{ldd} under SunOS 4.
8449 @defmac PARSE_LDD_OUTPUT (@var{ptr})
8450 Define this macro to be C code that extracts filenames from the output
8451 of the program denoted by @code{LDD_SUFFIX}. @var{ptr} is a variable
8452 of type @code{char *} that points to the beginning of a line of output
8453 from @code{LDD_SUFFIX}. If the line lists a dynamic dependency, the
8454 code must advance @var{ptr} to the beginning of the filename on that
8455 line. Otherwise, it must set @var{ptr} to @code{NULL}.
8458 @defmac SHLIB_SUFFIX
8459 Define this macro to a C string constant containing the default shared
8460 library extension of the target (e.g., @samp{".so"}). @command{collect2}
8461 strips version information after this suffix when generating global
8462 constructor and destructor names. This define is only needed on targets
8463 that use @command{collect2} to process constructors and destructors.
8466 @node Instruction Output
8467 @subsection Output of Assembler Instructions
8469 @c prevent bad page break with this line
8470 This describes assembler instruction output.
8472 @defmac REGISTER_NAMES
8473 A C initializer containing the assembler's names for the machine
8474 registers, each one as a C string constant. This is what translates
8475 register numbers in the compiler into assembler language.
8478 @defmac ADDITIONAL_REGISTER_NAMES
8479 If defined, a C initializer for an array of structures containing a name
8480 and a register number. This macro defines additional names for hard
8481 registers, thus allowing the @code{asm} option in declarations to refer
8482 to registers using alternate names.
8485 @defmac OVERLAPPING_REGISTER_NAMES
8486 If defined, a C initializer for an array of structures containing a
8487 name, a register number and a count of the number of consecutive
8488 machine registers the name overlaps. This macro defines additional
8489 names for hard registers, thus allowing the @code{asm} option in
8490 declarations to refer to registers using alternate names. Unlike
8491 @code{ADDITIONAL_REGISTER_NAMES}, this macro should be used when the
8492 register name implies multiple underlying registers.
8494 This macro should be used when it is important that a clobber in an
8495 @code{asm} statement clobbers all the underlying values implied by the
8496 register name. For example, on ARM, clobbering the double-precision
8497 VFP register ``d0'' implies clobbering both single-precision registers
8501 @defmac ASM_OUTPUT_OPCODE (@var{stream}, @var{ptr})
8502 Define this macro if you are using an unusual assembler that
8503 requires different names for the machine instructions.
8505 The definition is a C statement or statements which output an
8506 assembler instruction opcode to the stdio stream @var{stream}. The
8507 macro-operand @var{ptr} is a variable of type @code{char *} which
8508 points to the opcode name in its ``internal'' form---the form that is
8509 written in the machine description. The definition should output the
8510 opcode name to @var{stream}, performing any translation you desire, and
8511 increment the variable @var{ptr} to point at the end of the opcode
8512 so that it will not be output twice.
8514 In fact, your macro definition may process less than the entire opcode
8515 name, or more than the opcode name; but if you want to process text
8516 that includes @samp{%}-sequences to substitute operands, you must take
8517 care of the substitution yourself. Just be sure to increment
8518 @var{ptr} over whatever text should not be output normally.
8520 @findex recog_data.operand
8521 If you need to look at the operand values, they can be found as the
8522 elements of @code{recog_data.operand}.
8524 If the macro definition does nothing, the instruction is output
8528 @defmac FINAL_PRESCAN_INSN (@var{insn}, @var{opvec}, @var{noperands})
8529 If defined, a C statement to be executed just prior to the output of
8530 assembler code for @var{insn}, to modify the extracted operands so
8531 they will be output differently.
8533 Here the argument @var{opvec} is the vector containing the operands
8534 extracted from @var{insn}, and @var{noperands} is the number of
8535 elements of the vector which contain meaningful data for this insn.
8536 The contents of this vector are what will be used to convert the insn
8537 template into assembler code, so you can change the assembler output
8538 by changing the contents of the vector.
8540 This macro is useful when various assembler syntaxes share a single
8541 file of instruction patterns; by defining this macro differently, you
8542 can cause a large class of instructions to be output differently (such
8543 as with rearranged operands). Naturally, variations in assembler
8544 syntax affecting individual insn patterns ought to be handled by
8545 writing conditional output routines in those patterns.
8547 If this macro is not defined, it is equivalent to a null statement.
8550 @deftypefn {Target Hook} void TARGET_ASM_FINAL_POSTSCAN_INSN (FILE *@var{file}, rtx_insn *@var{insn}, rtx *@var{opvec}, int @var{noperands})
8551 If defined, this target hook is a function which is executed just after the
8552 output of assembler code for @var{insn}, to change the mode of the assembler
8555 Here the argument @var{opvec} is the vector containing the operands
8556 extracted from @var{insn}, and @var{noperands} is the number of
8557 elements of the vector which contain meaningful data for this insn.
8558 The contents of this vector are what was used to convert the insn
8559 template into assembler code, so you can change the assembler mode
8560 by checking the contents of the vector.
8563 @defmac PRINT_OPERAND (@var{stream}, @var{x}, @var{code})
8564 A C compound statement to output to stdio stream @var{stream} the
8565 assembler syntax for an instruction operand @var{x}. @var{x} is an
8568 @var{code} is a value that can be used to specify one of several ways
8569 of printing the operand. It is used when identical operands must be
8570 printed differently depending on the context. @var{code} comes from
8571 the @samp{%} specification that was used to request printing of the
8572 operand. If the specification was just @samp{%@var{digit}} then
8573 @var{code} is 0; if the specification was @samp{%@var{ltr}
8574 @var{digit}} then @var{code} is the ASCII code for @var{ltr}.
8577 If @var{x} is a register, this macro should print the register's name.
8578 The names can be found in an array @code{reg_names} whose type is
8579 @code{char *[]}. @code{reg_names} is initialized from
8580 @code{REGISTER_NAMES}.
8582 When the machine description has a specification @samp{%@var{punct}}
8583 (a @samp{%} followed by a punctuation character), this macro is called
8584 with a null pointer for @var{x} and the punctuation character for
8588 @defmac PRINT_OPERAND_PUNCT_VALID_P (@var{code})
8589 A C expression which evaluates to true if @var{code} is a valid
8590 punctuation character for use in the @code{PRINT_OPERAND} macro. If
8591 @code{PRINT_OPERAND_PUNCT_VALID_P} is not defined, it means that no
8592 punctuation characters (except for the standard one, @samp{%}) are used
8596 @defmac PRINT_OPERAND_ADDRESS (@var{stream}, @var{x})
8597 A C compound statement to output to stdio stream @var{stream} the
8598 assembler syntax for an instruction operand that is a memory reference
8599 whose address is @var{x}. @var{x} is an RTL expression.
8601 @cindex @code{TARGET_ENCODE_SECTION_INFO} usage
8602 On some machines, the syntax for a symbolic address depends on the
8603 section that the address refers to. On these machines, define the hook
8604 @code{TARGET_ENCODE_SECTION_INFO} to store the information into the
8605 @code{symbol_ref}, and then check for it here. @xref{Assembler
8609 @findex dbr_sequence_length
8610 @defmac DBR_OUTPUT_SEQEND (@var{file})
8611 A C statement, to be executed after all slot-filler instructions have
8612 been output. If necessary, call @code{dbr_sequence_length} to
8613 determine the number of slots filled in a sequence (zero if not
8614 currently outputting a sequence), to decide how many no-ops to output,
8617 Don't define this macro if it has nothing to do, but it is helpful in
8618 reading assembly output if the extent of the delay sequence is made
8619 explicit (e.g.@: with white space).
8622 @findex final_sequence
8623 Note that output routines for instructions with delay slots must be
8624 prepared to deal with not being output as part of a sequence
8625 (i.e.@: when the scheduling pass is not run, or when no slot fillers could be
8626 found.) The variable @code{final_sequence} is null when not
8627 processing a sequence, otherwise it contains the @code{sequence} rtx
8631 @defmac REGISTER_PREFIX
8632 @defmacx LOCAL_LABEL_PREFIX
8633 @defmacx USER_LABEL_PREFIX
8634 @defmacx IMMEDIATE_PREFIX
8635 If defined, C string expressions to be used for the @samp{%R}, @samp{%L},
8636 @samp{%U}, and @samp{%I} options of @code{asm_fprintf} (see
8637 @file{final.c}). These are useful when a single @file{md} file must
8638 support multiple assembler formats. In that case, the various @file{tm.h}
8639 files can define these macros differently.
8642 @defmac ASM_FPRINTF_EXTENSIONS (@var{file}, @var{argptr}, @var{format})
8643 If defined this macro should expand to a series of @code{case}
8644 statements which will be parsed inside the @code{switch} statement of
8645 the @code{asm_fprintf} function. This allows targets to define extra
8646 printf formats which may useful when generating their assembler
8647 statements. Note that uppercase letters are reserved for future
8648 generic extensions to asm_fprintf, and so are not available to target
8649 specific code. The output file is given by the parameter @var{file}.
8650 The varargs input pointer is @var{argptr} and the rest of the format
8651 string, starting the character after the one that is being switched
8652 upon, is pointed to by @var{format}.
8655 @defmac ASSEMBLER_DIALECT
8656 If your target supports multiple dialects of assembler language (such as
8657 different opcodes), define this macro as a C expression that gives the
8658 numeric index of the assembler language dialect to use, with zero as the
8661 If this macro is defined, you may use constructs of the form
8663 @samp{@{option0|option1|option2@dots{}@}}
8666 in the output templates of patterns (@pxref{Output Template}) or in the
8667 first argument of @code{asm_fprintf}. This construct outputs
8668 @samp{option0}, @samp{option1}, @samp{option2}, etc., if the value of
8669 @code{ASSEMBLER_DIALECT} is zero, one, two, etc. Any special characters
8670 within these strings retain their usual meaning. If there are fewer
8671 alternatives within the braces than the value of
8672 @code{ASSEMBLER_DIALECT}, the construct outputs nothing. If it's needed
8673 to print curly braces or @samp{|} character in assembler output directly,
8674 @samp{%@{}, @samp{%@}} and @samp{%|} can be used.
8676 If you do not define this macro, the characters @samp{@{}, @samp{|} and
8677 @samp{@}} do not have any special meaning when used in templates or
8678 operands to @code{asm_fprintf}.
8680 Define the macros @code{REGISTER_PREFIX}, @code{LOCAL_LABEL_PREFIX},
8681 @code{USER_LABEL_PREFIX} and @code{IMMEDIATE_PREFIX} if you can express
8682 the variations in assembler language syntax with that mechanism. Define
8683 @code{ASSEMBLER_DIALECT} and use the @samp{@{option0|option1@}} syntax
8684 if the syntax variant are larger and involve such things as different
8685 opcodes or operand order.
8688 @defmac ASM_OUTPUT_REG_PUSH (@var{stream}, @var{regno})
8689 A C expression to output to @var{stream} some assembler code
8690 which will push hard register number @var{regno} onto the stack.
8691 The code need not be optimal, since this macro is used only when
8695 @defmac ASM_OUTPUT_REG_POP (@var{stream}, @var{regno})
8696 A C expression to output to @var{stream} some assembler code
8697 which will pop hard register number @var{regno} off of the stack.
8698 The code need not be optimal, since this macro is used only when
8702 @node Dispatch Tables
8703 @subsection Output of Dispatch Tables
8705 @c prevent bad page break with this line
8706 This concerns dispatch tables.
8708 @cindex dispatch table
8709 @defmac ASM_OUTPUT_ADDR_DIFF_ELT (@var{stream}, @var{body}, @var{value}, @var{rel})
8710 A C statement to output to the stdio stream @var{stream} an assembler
8711 pseudo-instruction to generate a difference between two labels.
8712 @var{value} and @var{rel} are the numbers of two internal labels. The
8713 definitions of these labels are output using
8714 @code{(*targetm.asm_out.internal_label)}, and they must be printed in the same
8715 way here. For example,
8718 fprintf (@var{stream}, "\t.word L%d-L%d\n",
8719 @var{value}, @var{rel})
8722 You must provide this macro on machines where the addresses in a
8723 dispatch table are relative to the table's own address. If defined, GCC
8724 will also use this macro on all machines when producing PIC@.
8725 @var{body} is the body of the @code{ADDR_DIFF_VEC}; it is provided so that the
8726 mode and flags can be read.
8729 @defmac ASM_OUTPUT_ADDR_VEC_ELT (@var{stream}, @var{value})
8730 This macro should be provided on machines where the addresses
8731 in a dispatch table are absolute.
8733 The definition should be a C statement to output to the stdio stream
8734 @var{stream} an assembler pseudo-instruction to generate a reference to
8735 a label. @var{value} is the number of an internal label whose
8736 definition is output using @code{(*targetm.asm_out.internal_label)}.
8740 fprintf (@var{stream}, "\t.word L%d\n", @var{value})
8744 @defmac ASM_OUTPUT_CASE_LABEL (@var{stream}, @var{prefix}, @var{num}, @var{table})
8745 Define this if the label before a jump-table needs to be output
8746 specially. The first three arguments are the same as for
8747 @code{(*targetm.asm_out.internal_label)}; the fourth argument is the
8748 jump-table which follows (a @code{jump_table_data} containing an
8749 @code{addr_vec} or @code{addr_diff_vec}).
8751 This feature is used on system V to output a @code{swbeg} statement
8754 If this macro is not defined, these labels are output with
8755 @code{(*targetm.asm_out.internal_label)}.
8758 @defmac ASM_OUTPUT_CASE_END (@var{stream}, @var{num}, @var{table})
8759 Define this if something special must be output at the end of a
8760 jump-table. The definition should be a C statement to be executed
8761 after the assembler code for the table is written. It should write
8762 the appropriate code to stdio stream @var{stream}. The argument
8763 @var{table} is the jump-table insn, and @var{num} is the label-number
8764 of the preceding label.
8766 If this macro is not defined, nothing special is output at the end of
8770 @deftypefn {Target Hook} void TARGET_ASM_EMIT_UNWIND_LABEL (FILE *@var{stream}, tree @var{decl}, int @var{for_eh}, int @var{empty})
8771 This target hook emits a label at the beginning of each FDE@. It
8772 should be defined on targets where FDEs need special labels, and it
8773 should write the appropriate label, for the FDE associated with the
8774 function declaration @var{decl}, to the stdio stream @var{stream}.
8775 The third argument, @var{for_eh}, is a boolean: true if this is for an
8776 exception table. The fourth argument, @var{empty}, is a boolean:
8777 true if this is a placeholder label for an omitted FDE@.
8779 The default is that FDEs are not given nonlocal labels.
8782 @deftypefn {Target Hook} void TARGET_ASM_EMIT_EXCEPT_TABLE_LABEL (FILE *@var{stream})
8783 This target hook emits a label at the beginning of the exception table.
8784 It should be defined on targets where it is desirable for the table
8785 to be broken up according to function.
8787 The default is that no label is emitted.
8790 @deftypefn {Target Hook} void TARGET_ASM_EMIT_EXCEPT_PERSONALITY (rtx @var{personality})
8791 If the target implements @code{TARGET_ASM_UNWIND_EMIT}, this hook may be used to emit a directive to install a personality hook into the unwind info. This hook should not be used if dwarf2 unwind info is used.
8794 @deftypefn {Target Hook} void TARGET_ASM_UNWIND_EMIT (FILE *@var{stream}, rtx_insn *@var{insn})
8795 This target hook emits assembly directives required to unwind the
8796 given instruction. This is only used when @code{TARGET_EXCEPT_UNWIND_INFO}
8797 returns @code{UI_TARGET}.
8800 @deftypevr {Target Hook} bool TARGET_ASM_UNWIND_EMIT_BEFORE_INSN
8801 True if the @code{TARGET_ASM_UNWIND_EMIT} hook should be called before the assembly for @var{insn} has been emitted, false if the hook should be called afterward.
8804 @node Exception Region Output
8805 @subsection Assembler Commands for Exception Regions
8807 @c prevent bad page break with this line
8809 This describes commands marking the start and the end of an exception
8812 @defmac EH_FRAME_SECTION_NAME
8813 If defined, a C string constant for the name of the section containing
8814 exception handling frame unwind information. If not defined, GCC will
8815 provide a default definition if the target supports named sections.
8816 @file{crtstuff.c} uses this macro to switch to the appropriate section.
8818 You should define this symbol if your target supports DWARF 2 frame
8819 unwind information and the default definition does not work.
8822 @defmac EH_FRAME_IN_DATA_SECTION
8823 If defined, DWARF 2 frame unwind information will be placed in the
8824 data section even though the target supports named sections. This
8825 might be necessary, for instance, if the system linker does garbage
8826 collection and sections cannot be marked as not to be collected.
8828 Do not define this macro unless @code{TARGET_ASM_NAMED_SECTION} is
8832 @defmac EH_TABLES_CAN_BE_READ_ONLY
8833 Define this macro to 1 if your target is such that no frame unwind
8834 information encoding used with non-PIC code will ever require a
8835 runtime relocation, but the linker may not support merging read-only
8836 and read-write sections into a single read-write section.
8839 @defmac MASK_RETURN_ADDR
8840 An rtx used to mask the return address found via @code{RETURN_ADDR_RTX}, so
8841 that it does not contain any extraneous set bits in it.
8844 @defmac DWARF2_UNWIND_INFO
8845 Define this macro to 0 if your target supports DWARF 2 frame unwind
8846 information, but it does not yet work with exception handling.
8847 Otherwise, if your target supports this information (if it defines
8848 @code{INCOMING_RETURN_ADDR_RTX} and @code{OBJECT_FORMAT_ELF}),
8849 GCC will provide a default definition of 1.
8852 @deftypefn {Common Target Hook} {enum unwind_info_type} TARGET_EXCEPT_UNWIND_INFO (struct gcc_options *@var{opts})
8853 This hook defines the mechanism that will be used for exception handling
8854 by the target. If the target has ABI specified unwind tables, the hook
8855 should return @code{UI_TARGET}. If the target is to use the
8856 @code{setjmp}/@code{longjmp}-based exception handling scheme, the hook
8857 should return @code{UI_SJLJ}. If the target supports DWARF 2 frame unwind
8858 information, the hook should return @code{UI_DWARF2}.
8860 A target may, if exceptions are disabled, choose to return @code{UI_NONE}.
8861 This may end up simplifying other parts of target-specific code. The
8862 default implementation of this hook never returns @code{UI_NONE}.
8864 Note that the value returned by this hook should be constant. It should
8865 not depend on anything except the command-line switches described by
8866 @var{opts}. In particular, the
8867 setting @code{UI_SJLJ} must be fixed at compiler start-up as C pre-processor
8868 macros and builtin functions related to exception handling are set up
8869 depending on this setting.
8871 The default implementation of the hook first honors the
8872 @option{--enable-sjlj-exceptions} configure option, then
8873 @code{DWARF2_UNWIND_INFO}, and finally defaults to @code{UI_SJLJ}. If
8874 @code{DWARF2_UNWIND_INFO} depends on command-line options, the target
8875 must define this hook so that @var{opts} is used correctly.
8878 @deftypevr {Common Target Hook} bool TARGET_UNWIND_TABLES_DEFAULT
8879 This variable should be set to @code{true} if the target ABI requires unwinding
8880 tables even when exceptions are not used. It must not be modified by
8881 command-line option processing.
8884 @defmac DONT_USE_BUILTIN_SETJMP
8885 Define this macro to 1 if the @code{setjmp}/@code{longjmp}-based scheme
8886 should use the @code{setjmp}/@code{longjmp} functions from the C library
8887 instead of the @code{__builtin_setjmp}/@code{__builtin_longjmp} machinery.
8890 @defmac JMP_BUF_SIZE
8891 This macro has no effect unless @code{DONT_USE_BUILTIN_SETJMP} is also
8892 defined. Define this macro if the default size of @code{jmp_buf} buffer
8893 for the @code{setjmp}/@code{longjmp}-based exception handling mechanism
8894 is not large enough, or if it is much too large.
8895 The default size is @code{FIRST_PSEUDO_REGISTER * sizeof(void *)}.
8898 @defmac DWARF_CIE_DATA_ALIGNMENT
8899 This macro need only be defined if the target might save registers in the
8900 function prologue at an offset to the stack pointer that is not aligned to
8901 @code{UNITS_PER_WORD}. The definition should be the negative minimum
8902 alignment if @code{STACK_GROWS_DOWNWARD} is defined, and the positive
8903 minimum alignment otherwise. @xref{SDB and DWARF}. Only applicable if
8904 the target supports DWARF 2 frame unwind information.
8907 @deftypevr {Target Hook} bool TARGET_TERMINATE_DW2_EH_FRAME_INFO
8908 Contains the value true if the target should add a zero word onto the
8909 end of a Dwarf-2 frame info section when used for exception handling.
8910 Default value is false if @code{EH_FRAME_SECTION_NAME} is defined, and
8914 @deftypefn {Target Hook} rtx TARGET_DWARF_REGISTER_SPAN (rtx @var{reg})
8915 Given a register, this hook should return a parallel of registers to
8916 represent where to find the register pieces. Define this hook if the
8917 register and its mode are represented in Dwarf in non-contiguous
8918 locations, or if the register should be represented in more than one
8919 register in Dwarf. Otherwise, this hook should return @code{NULL_RTX}.
8920 If not defined, the default is to return @code{NULL_RTX}.
8923 @deftypefn {Target Hook} machine_mode TARGET_DWARF_FRAME_REG_MODE (int @var{regno})
8924 Given a register, this hook should return the mode which the
8925 corresponding Dwarf frame register should have. This is normally
8926 used to return a smaller mode than the raw mode to prevent call
8927 clobbered parts of a register altering the frame register size
8930 @deftypefn {Target Hook} void TARGET_INIT_DWARF_REG_SIZES_EXTRA (tree @var{address})
8931 If some registers are represented in Dwarf-2 unwind information in
8932 multiple pieces, define this hook to fill in information about the
8933 sizes of those pieces in the table used by the unwinder at runtime.
8934 It will be called by @code{expand_builtin_init_dwarf_reg_sizes} after
8935 filling in a single size corresponding to each hard register;
8936 @var{address} is the address of the table.
8939 @deftypefn {Target Hook} bool TARGET_ASM_TTYPE (rtx @var{sym})
8940 This hook is used to output a reference from a frame unwinding table to
8941 the type_info object identified by @var{sym}. It should return @code{true}
8942 if the reference was output. Returning @code{false} will cause the
8943 reference to be output using the normal Dwarf2 routines.
8946 @deftypevr {Target Hook} bool TARGET_ARM_EABI_UNWINDER
8947 This flag should be set to @code{true} on targets that use an ARM EABI
8948 based unwinding library, and @code{false} on other targets. This effects
8949 the format of unwinding tables, and how the unwinder in entered after
8950 running a cleanup. The default is @code{false}.
8953 @node Alignment Output
8954 @subsection Assembler Commands for Alignment
8956 @c prevent bad page break with this line
8957 This describes commands for alignment.
8959 @defmac JUMP_ALIGN (@var{label})
8960 The alignment (log base 2) to put in front of @var{label}, which is
8961 a common destination of jumps and has no fallthru incoming edge.
8963 This macro need not be defined if you don't want any special alignment
8964 to be done at such a time. Most machine descriptions do not currently
8967 Unless it's necessary to inspect the @var{label} parameter, it is better
8968 to set the variable @var{align_jumps} in the target's
8969 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
8970 selection in @var{align_jumps} in a @code{JUMP_ALIGN} implementation.
8973 @deftypefn {Target Hook} int TARGET_ASM_JUMP_ALIGN_MAX_SKIP (rtx_insn *@var{label})
8974 The maximum number of bytes to skip before @var{label} when applying
8975 @code{JUMP_ALIGN}. This works only if
8976 @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
8979 @defmac LABEL_ALIGN_AFTER_BARRIER (@var{label})
8980 The alignment (log base 2) to put in front of @var{label}, which follows
8983 This macro need not be defined if you don't want any special alignment
8984 to be done at such a time. Most machine descriptions do not currently
8988 @deftypefn {Target Hook} int TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP (rtx_insn *@var{label})
8989 The maximum number of bytes to skip before @var{label} when applying
8990 @code{LABEL_ALIGN_AFTER_BARRIER}. This works only if
8991 @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is defined.
8994 @defmac LOOP_ALIGN (@var{label})
8995 The alignment (log base 2) to put in front of @var{label} that heads
8996 a frequently executed basic block (usually the header of a loop).
8998 This macro need not be defined if you don't want any special alignment
8999 to be done at such a time. Most machine descriptions do not currently
9002 Unless it's necessary to inspect the @var{label} parameter, it is better
9003 to set the variable @code{align_loops} in the target's
9004 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
9005 selection in @code{align_loops} in a @code{LOOP_ALIGN} implementation.
9008 @deftypefn {Target Hook} int TARGET_ASM_LOOP_ALIGN_MAX_SKIP (rtx_insn *@var{label})
9009 The maximum number of bytes to skip when applying @code{LOOP_ALIGN} to
9010 @var{label}. This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN} is
9014 @defmac LABEL_ALIGN (@var{label})
9015 The alignment (log base 2) to put in front of @var{label}.
9016 If @code{LABEL_ALIGN_AFTER_BARRIER} / @code{LOOP_ALIGN} specify a different alignment,
9017 the maximum of the specified values is used.
9019 Unless it's necessary to inspect the @var{label} parameter, it is better
9020 to set the variable @code{align_labels} in the target's
9021 @code{TARGET_OPTION_OVERRIDE}. Otherwise, you should try to honor the user's
9022 selection in @code{align_labels} in a @code{LABEL_ALIGN} implementation.
9025 @deftypefn {Target Hook} int TARGET_ASM_LABEL_ALIGN_MAX_SKIP (rtx_insn *@var{label})
9026 The maximum number of bytes to skip when applying @code{LABEL_ALIGN}
9027 to @var{label}. This works only if @code{ASM_OUTPUT_MAX_SKIP_ALIGN}
9031 @defmac ASM_OUTPUT_SKIP (@var{stream}, @var{nbytes})
9032 A C statement to output to the stdio stream @var{stream} an assembler
9033 instruction to advance the location counter by @var{nbytes} bytes.
9034 Those bytes should be zero when loaded. @var{nbytes} will be a C
9035 expression of type @code{unsigned HOST_WIDE_INT}.
9038 @defmac ASM_NO_SKIP_IN_TEXT
9039 Define this macro if @code{ASM_OUTPUT_SKIP} should not be used in the
9040 text section because it fails to put zeros in the bytes that are skipped.
9041 This is true on many Unix systems, where the pseudo--op to skip bytes
9042 produces no-op instructions rather than zeros when used in the text
9046 @defmac ASM_OUTPUT_ALIGN (@var{stream}, @var{power})
9047 A C statement to output to the stdio stream @var{stream} an assembler
9048 command to advance the location counter to a multiple of 2 to the
9049 @var{power} bytes. @var{power} will be a C expression of type @code{int}.
9052 @defmac ASM_OUTPUT_ALIGN_WITH_NOP (@var{stream}, @var{power})
9053 Like @code{ASM_OUTPUT_ALIGN}, except that the ``nop'' instruction is used
9054 for padding, if necessary.
9057 @defmac ASM_OUTPUT_MAX_SKIP_ALIGN (@var{stream}, @var{power}, @var{max_skip})
9058 A C statement to output to the stdio stream @var{stream} an assembler
9059 command to advance the location counter to a multiple of 2 to the
9060 @var{power} bytes, but only if @var{max_skip} or fewer bytes are needed to
9061 satisfy the alignment request. @var{power} and @var{max_skip} will be
9062 a C expression of type @code{int}.
9066 @node Debugging Info
9067 @section Controlling Debugging Information Format
9069 @c prevent bad page break with this line
9070 This describes how to specify debugging information.
9073 * All Debuggers:: Macros that affect all debugging formats uniformly.
9074 * DBX Options:: Macros enabling specific options in DBX format.
9075 * DBX Hooks:: Hook macros for varying DBX format.
9076 * File Names and DBX:: Macros controlling output of file names in DBX format.
9077 * SDB and DWARF:: Macros for SDB (COFF) and DWARF formats.
9078 * VMS Debug:: Macros for VMS debug format.
9082 @subsection Macros Affecting All Debugging Formats
9084 @c prevent bad page break with this line
9085 These macros affect all debugging formats.
9087 @defmac DBX_REGISTER_NUMBER (@var{regno})
9088 A C expression that returns the DBX register number for the compiler
9089 register number @var{regno}. In the default macro provided, the value
9090 of this expression will be @var{regno} itself. But sometimes there are
9091 some registers that the compiler knows about and DBX does not, or vice
9092 versa. In such cases, some register may need to have one number in the
9093 compiler and another for DBX@.
9095 If two registers have consecutive numbers inside GCC, and they can be
9096 used as a pair to hold a multiword value, then they @emph{must} have
9097 consecutive numbers after renumbering with @code{DBX_REGISTER_NUMBER}.
9098 Otherwise, debuggers will be unable to access such a pair, because they
9099 expect register pairs to be consecutive in their own numbering scheme.
9101 If you find yourself defining @code{DBX_REGISTER_NUMBER} in way that
9102 does not preserve register pairs, then what you must do instead is
9103 redefine the actual register numbering scheme.
9106 @defmac DEBUGGER_AUTO_OFFSET (@var{x})
9107 A C expression that returns the integer offset value for an automatic
9108 variable having address @var{x} (an RTL expression). The default
9109 computation assumes that @var{x} is based on the frame-pointer and
9110 gives the offset from the frame-pointer. This is required for targets
9111 that produce debugging output for DBX or COFF-style debugging output
9112 for SDB and allow the frame-pointer to be eliminated when the
9113 @option{-g} options is used.
9116 @defmac DEBUGGER_ARG_OFFSET (@var{offset}, @var{x})
9117 A C expression that returns the integer offset value for an argument
9118 having address @var{x} (an RTL expression). The nominal offset is
9122 @defmac PREFERRED_DEBUGGING_TYPE
9123 A C expression that returns the type of debugging output GCC should
9124 produce when the user specifies just @option{-g}. Define
9125 this if you have arranged for GCC to support more than one format of
9126 debugging output. Currently, the allowable values are @code{DBX_DEBUG},
9127 @code{SDB_DEBUG}, @code{DWARF_DEBUG}, @code{DWARF2_DEBUG},
9128 @code{XCOFF_DEBUG}, @code{VMS_DEBUG}, and @code{VMS_AND_DWARF2_DEBUG}.
9130 When the user specifies @option{-ggdb}, GCC normally also uses the
9131 value of this macro to select the debugging output format, but with two
9132 exceptions. If @code{DWARF2_DEBUGGING_INFO} is defined, GCC uses the
9133 value @code{DWARF2_DEBUG}. Otherwise, if @code{DBX_DEBUGGING_INFO} is
9134 defined, GCC uses @code{DBX_DEBUG}.
9136 The value of this macro only affects the default debugging output; the
9137 user can always get a specific type of output by using @option{-gstabs},
9138 @option{-gcoff}, @option{-gdwarf-2}, @option{-gxcoff}, or @option{-gvms}.
9142 @subsection Specific Options for DBX Output
9144 @c prevent bad page break with this line
9145 These are specific options for DBX output.
9147 @defmac DBX_DEBUGGING_INFO
9148 Define this macro if GCC should produce debugging output for DBX
9149 in response to the @option{-g} option.
9152 @defmac XCOFF_DEBUGGING_INFO
9153 Define this macro if GCC should produce XCOFF format debugging output
9154 in response to the @option{-g} option. This is a variant of DBX format.
9157 @defmac DEFAULT_GDB_EXTENSIONS
9158 Define this macro to control whether GCC should by default generate
9159 GDB's extended version of DBX debugging information (assuming DBX-format
9160 debugging information is enabled at all). If you don't define the
9161 macro, the default is 1: always generate the extended information
9162 if there is any occasion to.
9165 @defmac DEBUG_SYMS_TEXT
9166 Define this macro if all @code{.stabs} commands should be output while
9167 in the text section.
9170 @defmac ASM_STABS_OP
9171 A C string constant, including spacing, naming the assembler pseudo op to
9172 use instead of @code{"\t.stabs\t"} to define an ordinary debugging symbol.
9173 If you don't define this macro, @code{"\t.stabs\t"} is used. This macro
9174 applies only to DBX debugging information format.
9177 @defmac ASM_STABD_OP
9178 A C string constant, including spacing, naming the assembler pseudo op to
9179 use instead of @code{"\t.stabd\t"} to define a debugging symbol whose
9180 value is the current location. If you don't define this macro,
9181 @code{"\t.stabd\t"} is used. This macro applies only to DBX debugging
9185 @defmac ASM_STABN_OP
9186 A C string constant, including spacing, naming the assembler pseudo op to
9187 use instead of @code{"\t.stabn\t"} to define a debugging symbol with no
9188 name. If you don't define this macro, @code{"\t.stabn\t"} is used. This
9189 macro applies only to DBX debugging information format.
9192 @defmac DBX_NO_XREFS
9193 Define this macro if DBX on your system does not support the construct
9194 @samp{xs@var{tagname}}. On some systems, this construct is used to
9195 describe a forward reference to a structure named @var{tagname}.
9196 On other systems, this construct is not supported at all.
9199 @defmac DBX_CONTIN_LENGTH
9200 A symbol name in DBX-format debugging information is normally
9201 continued (split into two separate @code{.stabs} directives) when it
9202 exceeds a certain length (by default, 80 characters). On some
9203 operating systems, DBX requires this splitting; on others, splitting
9204 must not be done. You can inhibit splitting by defining this macro
9205 with the value zero. You can override the default splitting-length by
9206 defining this macro as an expression for the length you desire.
9209 @defmac DBX_CONTIN_CHAR
9210 Normally continuation is indicated by adding a @samp{\} character to
9211 the end of a @code{.stabs} string when a continuation follows. To use
9212 a different character instead, define this macro as a character
9213 constant for the character you want to use. Do not define this macro
9214 if backslash is correct for your system.
9217 @defmac DBX_STATIC_STAB_DATA_SECTION
9218 Define this macro if it is necessary to go to the data section before
9219 outputting the @samp{.stabs} pseudo-op for a non-global static
9223 @defmac DBX_TYPE_DECL_STABS_CODE
9224 The value to use in the ``code'' field of the @code{.stabs} directive
9225 for a typedef. The default is @code{N_LSYM}.
9228 @defmac DBX_STATIC_CONST_VAR_CODE
9229 The value to use in the ``code'' field of the @code{.stabs} directive
9230 for a static variable located in the text section. DBX format does not
9231 provide any ``right'' way to do this. The default is @code{N_FUN}.
9234 @defmac DBX_REGPARM_STABS_CODE
9235 The value to use in the ``code'' field of the @code{.stabs} directive
9236 for a parameter passed in registers. DBX format does not provide any
9237 ``right'' way to do this. The default is @code{N_RSYM}.
9240 @defmac DBX_REGPARM_STABS_LETTER
9241 The letter to use in DBX symbol data to identify a symbol as a parameter
9242 passed in registers. DBX format does not customarily provide any way to
9243 do this. The default is @code{'P'}.
9246 @defmac DBX_FUNCTION_FIRST
9247 Define this macro if the DBX information for a function and its
9248 arguments should precede the assembler code for the function. Normally,
9249 in DBX format, the debugging information entirely follows the assembler
9253 @defmac DBX_BLOCKS_FUNCTION_RELATIVE
9254 Define this macro, with value 1, if the value of a symbol describing
9255 the scope of a block (@code{N_LBRAC} or @code{N_RBRAC}) should be
9256 relative to the start of the enclosing function. Normally, GCC uses
9257 an absolute address.
9260 @defmac DBX_LINES_FUNCTION_RELATIVE
9261 Define this macro, with value 1, if the value of a symbol indicating
9262 the current line number (@code{N_SLINE}) should be relative to the
9263 start of the enclosing function. Normally, GCC uses an absolute address.
9266 @defmac DBX_USE_BINCL
9267 Define this macro if GCC should generate @code{N_BINCL} and
9268 @code{N_EINCL} stabs for included header files, as on Sun systems. This
9269 macro also directs GCC to output a type number as a pair of a file
9270 number and a type number within the file. Normally, GCC does not
9271 generate @code{N_BINCL} or @code{N_EINCL} stabs, and it outputs a single
9272 number for a type number.
9276 @subsection Open-Ended Hooks for DBX Format
9278 @c prevent bad page break with this line
9279 These are hooks for DBX format.
9281 @defmac DBX_OUTPUT_SOURCE_LINE (@var{stream}, @var{line}, @var{counter})
9282 A C statement to output DBX debugging information before code for line
9283 number @var{line} of the current source file to the stdio stream
9284 @var{stream}. @var{counter} is the number of time the macro was
9285 invoked, including the current invocation; it is intended to generate
9286 unique labels in the assembly output.
9288 This macro should not be defined if the default output is correct, or
9289 if it can be made correct by defining @code{DBX_LINES_FUNCTION_RELATIVE}.
9292 @defmac NO_DBX_FUNCTION_END
9293 Some stabs encapsulation formats (in particular ECOFF), cannot handle the
9294 @code{.stabs "",N_FUN,,0,0,Lscope-function-1} gdb dbx extension construct.
9295 On those machines, define this macro to turn this feature off without
9296 disturbing the rest of the gdb extensions.
9299 @defmac NO_DBX_BNSYM_ENSYM
9300 Some assemblers cannot handle the @code{.stabd BNSYM/ENSYM,0,0} gdb dbx
9301 extension construct. On those machines, define this macro to turn this
9302 feature off without disturbing the rest of the gdb extensions.
9305 @node File Names and DBX
9306 @subsection File Names in DBX Format
9308 @c prevent bad page break with this line
9309 This describes file names in DBX format.
9311 @defmac DBX_OUTPUT_MAIN_SOURCE_FILENAME (@var{stream}, @var{name})
9312 A C statement to output DBX debugging information to the stdio stream
9313 @var{stream}, which indicates that file @var{name} is the main source
9314 file---the file specified as the input file for compilation.
9315 This macro is called only once, at the beginning of compilation.
9317 This macro need not be defined if the standard form of output
9318 for DBX debugging information is appropriate.
9320 It may be necessary to refer to a label equal to the beginning of the
9321 text section. You can use @samp{assemble_name (stream, ltext_label_name)}
9322 to do so. If you do this, you must also set the variable
9323 @var{used_ltext_label_name} to @code{true}.
9326 @defmac NO_DBX_MAIN_SOURCE_DIRECTORY
9327 Define this macro, with value 1, if GCC should not emit an indication
9328 of the current directory for compilation and current source language at
9329 the beginning of the file.
9332 @defmac NO_DBX_GCC_MARKER
9333 Define this macro, with value 1, if GCC should not emit an indication
9334 that this object file was compiled by GCC@. The default is to emit
9335 an @code{N_OPT} stab at the beginning of every source file, with
9336 @samp{gcc2_compiled.} for the string and value 0.
9339 @defmac DBX_OUTPUT_MAIN_SOURCE_FILE_END (@var{stream}, @var{name})
9340 A C statement to output DBX debugging information at the end of
9341 compilation of the main source file @var{name}. Output should be
9342 written to the stdio stream @var{stream}.
9344 If you don't define this macro, nothing special is output at the end
9345 of compilation, which is correct for most machines.
9348 @defmac DBX_OUTPUT_NULL_N_SO_AT_MAIN_SOURCE_FILE_END
9349 Define this macro @emph{instead of} defining
9350 @code{DBX_OUTPUT_MAIN_SOURCE_FILE_END}, if what needs to be output at
9351 the end of compilation is an @code{N_SO} stab with an empty string,
9352 whose value is the highest absolute text address in the file.
9357 @subsection Macros for SDB and DWARF Output
9359 @c prevent bad page break with this line
9360 Here are macros for SDB and DWARF output.
9362 @defmac SDB_DEBUGGING_INFO
9363 Define this macro if GCC should produce COFF-style debugging output
9364 for SDB in response to the @option{-g} option.
9367 @defmac DWARF2_DEBUGGING_INFO
9368 Define this macro if GCC should produce dwarf version 2 format
9369 debugging output in response to the @option{-g} option.
9371 @deftypefn {Target Hook} int TARGET_DWARF_CALLING_CONVENTION (const_tree @var{function})
9372 Define this to enable the dwarf attribute @code{DW_AT_calling_convention} to
9373 be emitted for each function. Instead of an integer return the enum
9374 value for the @code{DW_CC_} tag.
9377 To support optional call frame debugging information, you must also
9378 define @code{INCOMING_RETURN_ADDR_RTX} and either set
9379 @code{RTX_FRAME_RELATED_P} on the prologue insns if you use RTL for the
9380 prologue, or call @code{dwarf2out_def_cfa} and @code{dwarf2out_reg_save}
9381 as appropriate from @code{TARGET_ASM_FUNCTION_PROLOGUE} if you don't.
9384 @defmac DWARF2_FRAME_INFO
9385 Define this macro to a nonzero value if GCC should always output
9386 Dwarf 2 frame information. If @code{TARGET_EXCEPT_UNWIND_INFO}
9387 (@pxref{Exception Region Output}) returns @code{UI_DWARF2}, and
9388 exceptions are enabled, GCC will output this information not matter
9389 how you define @code{DWARF2_FRAME_INFO}.
9392 @deftypefn {Target Hook} {enum unwind_info_type} TARGET_DEBUG_UNWIND_INFO (void)
9393 This hook defines the mechanism that will be used for describing frame
9394 unwind information to the debugger. Normally the hook will return
9395 @code{UI_DWARF2} if DWARF 2 debug information is enabled, and
9396 return @code{UI_NONE} otherwise.
9398 A target may return @code{UI_DWARF2} even when DWARF 2 debug information
9399 is disabled in order to always output DWARF 2 frame information.
9401 A target may return @code{UI_TARGET} if it has ABI specified unwind tables.
9402 This will suppress generation of the normal debug frame unwind information.
9405 @defmac DWARF2_ASM_LINE_DEBUG_INFO
9406 Define this macro to be a nonzero value if the assembler can generate Dwarf 2
9407 line debug info sections. This will result in much more compact line number
9408 tables, and hence is desirable if it works.
9411 @deftypevr {Target Hook} bool TARGET_WANT_DEBUG_PUB_SECTIONS
9412 True if the @code{.debug_pubtypes} and @code{.debug_pubnames} sections should be emitted. These sections are not used on most platforms, and in particular GDB does not use them.
9415 @deftypevr {Target Hook} bool TARGET_FORCE_AT_COMP_DIR
9416 True if the @code{DW_AT_comp_dir} attribute should be emitted for each compilation unit. This attribute is required for the darwin linker to emit debug information.
9419 @deftypevr {Target Hook} bool TARGET_DELAY_SCHED2
9420 True if sched2 is not to be run at its normal place.
9421 This usually means it will be run as part of machine-specific reorg.
9424 @deftypevr {Target Hook} bool TARGET_DELAY_VARTRACK
9425 True if vartrack is not to be run at its normal place.
9426 This usually means it will be run as part of machine-specific reorg.
9429 @deftypevr {Target Hook} bool TARGET_NO_REGISTER_ALLOCATION
9430 True if register allocation and the passes
9431 following it should not be run. Usually true only for virtual assembler
9435 @defmac ASM_OUTPUT_DWARF_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
9436 A C statement to issue assembly directives that create a difference
9437 @var{lab1} minus @var{lab2}, using an integer of the given @var{size}.
9440 @defmac ASM_OUTPUT_DWARF_VMS_DELTA (@var{stream}, @var{size}, @var{label1}, @var{label2})
9441 A C statement to issue assembly directives that create a difference
9442 between the two given labels in system defined units, e.g. instruction
9443 slots on IA64 VMS, using an integer of the given size.
9446 @defmac ASM_OUTPUT_DWARF_OFFSET (@var{stream}, @var{size}, @var{label}, @var{section})
9447 A C statement to issue assembly directives that create a
9448 section-relative reference to the given @var{label}, using an integer of the
9449 given @var{size}. The label is known to be defined in the given @var{section}.
9452 @defmac ASM_OUTPUT_DWARF_PCREL (@var{stream}, @var{size}, @var{label})
9453 A C statement to issue assembly directives that create a self-relative
9454 reference to the given @var{label}, using an integer of the given @var{size}.
9457 @defmac ASM_OUTPUT_DWARF_TABLE_REF (@var{label})
9458 A C statement to issue assembly directives that create a reference to
9459 the DWARF table identifier @var{label} from the current section. This
9460 is used on some systems to avoid garbage collecting a DWARF table which
9461 is referenced by a function.
9464 @deftypefn {Target Hook} void TARGET_ASM_OUTPUT_DWARF_DTPREL (FILE *@var{file}, int @var{size}, rtx @var{x})
9465 If defined, this target hook is a function which outputs a DTP-relative
9466 reference to the given TLS symbol of the specified size.
9469 @defmac PUT_SDB_@dots{}
9470 Define these macros to override the assembler syntax for the special
9471 SDB assembler directives. See @file{sdbout.c} for a list of these
9472 macros and their arguments. If the standard syntax is used, you need
9473 not define them yourself.
9477 Some assemblers do not support a semicolon as a delimiter, even between
9478 SDB assembler directives. In that case, define this macro to be the
9479 delimiter to use (usually @samp{\n}). It is not necessary to define
9480 a new set of @code{PUT_SDB_@var{op}} macros if this is the only change
9484 @defmac SDB_ALLOW_UNKNOWN_REFERENCES
9485 Define this macro to allow references to unknown structure,
9486 union, or enumeration tags to be emitted. Standard COFF does not
9487 allow handling of unknown references, MIPS ECOFF has support for
9491 @defmac SDB_ALLOW_FORWARD_REFERENCES
9492 Define this macro to allow references to structure, union, or
9493 enumeration tags that have not yet been seen to be handled. Some
9494 assemblers choke if forward tags are used, while some require it.
9497 @defmac SDB_OUTPUT_SOURCE_LINE (@var{stream}, @var{line})
9498 A C statement to output SDB debugging information before code for line
9499 number @var{line} of the current source file to the stdio stream
9500 @var{stream}. The default is to emit an @code{.ln} directive.
9505 @subsection Macros for VMS Debug Format
9507 @c prevent bad page break with this line
9508 Here are macros for VMS debug format.
9510 @defmac VMS_DEBUGGING_INFO
9511 Define this macro if GCC should produce debugging output for VMS
9512 in response to the @option{-g} option. The default behavior for VMS
9513 is to generate minimal debug info for a traceback in the absence of
9514 @option{-g} unless explicitly overridden with @option{-g0}. This
9515 behavior is controlled by @code{TARGET_OPTION_OPTIMIZATION} and
9516 @code{TARGET_OPTION_OVERRIDE}.
9519 @node Floating Point
9520 @section Cross Compilation and Floating Point
9521 @cindex cross compilation and floating point
9522 @cindex floating point and cross compilation
9524 While all modern machines use twos-complement representation for integers,
9525 there are a variety of representations for floating point numbers. This
9526 means that in a cross-compiler the representation of floating point numbers
9527 in the compiled program may be different from that used in the machine
9528 doing the compilation.
9530 Because different representation systems may offer different amounts of
9531 range and precision, all floating point constants must be represented in
9532 the target machine's format. Therefore, the cross compiler cannot
9533 safely use the host machine's floating point arithmetic; it must emulate
9534 the target's arithmetic. To ensure consistency, GCC always uses
9535 emulation to work with floating point values, even when the host and
9536 target floating point formats are identical.
9538 The following macros are provided by @file{real.h} for the compiler to
9539 use. All parts of the compiler which generate or optimize
9540 floating-point calculations must use these macros. They may evaluate
9541 their operands more than once, so operands must not have side effects.
9543 @defmac REAL_VALUE_TYPE
9544 The C data type to be used to hold a floating point value in the target
9545 machine's format. Typically this is a @code{struct} containing an
9546 array of @code{HOST_WIDE_INT}, but all code should treat it as an opaque
9550 @deftypefn Macro int REAL_VALUES_EQUAL (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
9551 Compares for equality the two values, @var{x} and @var{y}. If the target
9552 floating point format supports negative zeroes and/or NaNs,
9553 @samp{REAL_VALUES_EQUAL (-0.0, 0.0)} is true, and
9554 @samp{REAL_VALUES_EQUAL (NaN, NaN)} is false.
9557 @deftypefn Macro int REAL_VALUES_LESS (REAL_VALUE_TYPE @var{x}, REAL_VALUE_TYPE @var{y})
9558 Tests whether @var{x} is less than @var{y}.
9561 @deftypefn Macro HOST_WIDE_INT REAL_VALUE_FIX (REAL_VALUE_TYPE @var{x})
9562 Truncates @var{x} to a signed integer, rounding toward zero.
9565 @deftypefn Macro {unsigned HOST_WIDE_INT} REAL_VALUE_UNSIGNED_FIX (REAL_VALUE_TYPE @var{x})
9566 Truncates @var{x} to an unsigned integer, rounding toward zero. If
9567 @var{x} is negative, returns zero.
9570 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ATOF (const char *@var{string}, machine_mode @var{mode})
9571 Converts @var{string} into a floating point number in the target machine's
9572 representation for mode @var{mode}. This routine can handle both
9573 decimal and hexadecimal floating point constants, using the syntax
9574 defined by the C language for both.
9577 @deftypefn Macro int REAL_VALUE_NEGATIVE (REAL_VALUE_TYPE @var{x})
9578 Returns 1 if @var{x} is negative (including negative zero), 0 otherwise.
9581 @deftypefn Macro int REAL_VALUE_ISINF (REAL_VALUE_TYPE @var{x})
9582 Determines whether @var{x} represents infinity (positive or negative).
9585 @deftypefn Macro int REAL_VALUE_ISNAN (REAL_VALUE_TYPE @var{x})
9586 Determines whether @var{x} represents a ``NaN'' (not-a-number).
9589 @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})
9590 Calculates an arithmetic operation on the two floating point values
9591 @var{x} and @var{y}, storing the result in @var{output} (which must be a
9594 The operation to be performed is specified by @var{code}. Only the
9595 following codes are supported: @code{PLUS_EXPR}, @code{MINUS_EXPR},
9596 @code{MULT_EXPR}, @code{RDIV_EXPR}, @code{MAX_EXPR}, @code{MIN_EXPR}.
9598 If @code{REAL_ARITHMETIC} is asked to evaluate division by zero and the
9599 target's floating point format cannot represent infinity, it will call
9600 @code{abort}. Callers should check for this situation first, using
9601 @code{MODE_HAS_INFINITIES}. @xref{Storage Layout}.
9604 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_NEGATE (REAL_VALUE_TYPE @var{x})
9605 Returns the negative of the floating point value @var{x}.
9608 @deftypefn Macro REAL_VALUE_TYPE REAL_VALUE_ABS (REAL_VALUE_TYPE @var{x})
9609 Returns the absolute value of @var{x}.
9612 @node Mode Switching
9613 @section Mode Switching Instructions
9614 @cindex mode switching
9615 The following macros control mode switching optimizations:
9617 @defmac OPTIMIZE_MODE_SWITCHING (@var{entity})
9618 Define this macro if the port needs extra instructions inserted for mode
9619 switching in an optimizing compilation.
9621 For an example, the SH4 can perform both single and double precision
9622 floating point operations, but to perform a single precision operation,
9623 the FPSCR PR bit has to be cleared, while for a double precision
9624 operation, this bit has to be set. Changing the PR bit requires a general
9625 purpose register as a scratch register, hence these FPSCR sets have to
9626 be inserted before reload, i.e.@: you can't put this into instruction emitting
9627 or @code{TARGET_MACHINE_DEPENDENT_REORG}.
9629 You can have multiple entities that are mode-switched, and select at run time
9630 which entities actually need it. @code{OPTIMIZE_MODE_SWITCHING} should
9631 return nonzero for any @var{entity} that needs mode-switching.
9632 If you define this macro, you also have to define
9633 @code{NUM_MODES_FOR_MODE_SWITCHING}, @code{TARGET_MODE_NEEDED},
9634 @code{TARGET_MODE_PRIORITY} and @code{TARGET_MODE_EMIT}.
9635 @code{TARGET_MODE_AFTER}, @code{TARGET_MODE_ENTRY}, and @code{TARGET_MODE_EXIT}
9639 @defmac NUM_MODES_FOR_MODE_SWITCHING
9640 If you define @code{OPTIMIZE_MODE_SWITCHING}, you have to define this as
9641 initializer for an array of integers. Each initializer element
9642 N refers to an entity that needs mode switching, and specifies the number
9643 of different modes that might need to be set for this entity.
9644 The position of the initializer in the initializer---starting counting at
9645 zero---determines the integer that is used to refer to the mode-switched
9647 In macros that take mode arguments / yield a mode result, modes are
9648 represented as numbers 0 @dots{} N @minus{} 1. N is used to specify that no mode
9649 switch is needed / supplied.
9652 @deftypefn {Target Hook} void TARGET_MODE_EMIT (int @var{entity}, int @var{mode}, int @var{prev_mode}, HARD_REG_SET @var{regs_live})
9653 Generate one or more insns to set @var{entity} to @var{mode}. @var{hard_reg_live} is the set of hard registers live at the point where the insn(s) are to be inserted. @var{prev_moxde} indicates the mode to switch from. Sets of a lower numbered entity will be emitted before sets of a higher numbered entity to a mode of the same or lower priority.
9656 @deftypefn {Target Hook} int TARGET_MODE_NEEDED (int @var{entity}, rtx_insn *@var{insn})
9657 @var{entity} is an integer specifying a mode-switched entity. If @code{OPTIMIZE_MODE_SWITCHING} is defined, you must define this macro to return an integer value not larger than the corresponding element in @code{NUM_MODES_FOR_MODE_SWITCHING}, to denote the mode that @var{entity} must be switched into prior to the execution of @var{insn}.
9660 @deftypefn {Target Hook} int TARGET_MODE_AFTER (int @var{entity}, int @var{mode}, rtx_insn *@var{insn})
9661 @var{entity} is an integer specifying a mode-switched entity. If this macro is defined, it is evaluated for every @var{insn} during mode switching. It determines the mode that an insn results in (if different from the incoming mode).
9664 @deftypefn {Target Hook} int TARGET_MODE_ENTRY (int @var{entity})
9665 If this macro is defined, it is evaluated for every @var{entity} that needs mode switching. It should evaluate to an integer, which is a mode that @var{entity} is assumed to be switched to at function entry. If @code{TARGET_MODE_ENTRY} is defined then @code{TARGET_MODE_EXIT} must be defined.
9668 @deftypefn {Target Hook} int TARGET_MODE_EXIT (int @var{entity})
9669 If this macro is defined, it is evaluated for every @var{entity} that needs mode switching. It should evaluate to an integer, which is a mode that @var{entity} is assumed to be switched to at function exit. If @code{TARGET_MODE_EXIT} is defined then @code{TARGET_MODE_ENTRY} must be defined.
9672 @deftypefn {Target Hook} int TARGET_MODE_PRIORITY (int @var{entity}, int @var{n})
9673 This macro specifies the order in which modes for @var{entity} are processed. 0 is the highest priority, @code{NUM_MODES_FOR_MODE_SWITCHING[@var{entity}] - 1} the lowest. The value of the macro should be an integer designating a mode for @var{entity}. For any fixed @var{entity}, @code{mode_priority} (@var{entity}, @var{n}) shall be a bijection in 0 @dots{} @code{num_modes_for_mode_switching[@var{entity}] - 1}.
9676 @node Target Attributes
9677 @section Defining target-specific uses of @code{__attribute__}
9678 @cindex target attributes
9679 @cindex machine attributes
9680 @cindex attributes, target-specific
9682 Target-specific attributes may be defined for functions, data and types.
9683 These are described using the following target hooks; they also need to
9684 be documented in @file{extend.texi}.
9686 @deftypevr {Target Hook} {const struct attribute_spec *} TARGET_ATTRIBUTE_TABLE
9687 If defined, this target hook points to an array of @samp{struct
9688 attribute_spec} (defined in @file{tree.h}) specifying the machine
9689 specific attributes for this target and some of the restrictions on the
9690 entities to which these attributes are applied and the arguments they
9694 @deftypefn {Target Hook} bool TARGET_ATTRIBUTE_TAKES_IDENTIFIER_P (const_tree @var{name})
9695 If defined, this target hook is a function which returns true if the
9696 machine-specific attribute named @var{name} expects an identifier
9697 given as its first argument to be passed on as a plain identifier, not
9698 subjected to name lookup. If this is not defined, the default is
9699 false for all machine-specific attributes.
9702 @deftypefn {Target Hook} int TARGET_COMP_TYPE_ATTRIBUTES (const_tree @var{type1}, const_tree @var{type2})
9703 If defined, this target hook is a function which returns zero if the attributes on
9704 @var{type1} and @var{type2} are incompatible, one if they are compatible,
9705 and two if they are nearly compatible (which causes a warning to be
9706 generated). If this is not defined, machine-specific attributes are
9707 supposed always to be compatible.
9710 @deftypefn {Target Hook} void TARGET_SET_DEFAULT_TYPE_ATTRIBUTES (tree @var{type})
9711 If defined, this target hook is a function which assigns default attributes to
9712 the newly defined @var{type}.
9715 @deftypefn {Target Hook} tree TARGET_MERGE_TYPE_ATTRIBUTES (tree @var{type1}, tree @var{type2})
9716 Define this target hook if the merging of type attributes needs special
9717 handling. If defined, the result is a list of the combined
9718 @code{TYPE_ATTRIBUTES} of @var{type1} and @var{type2}. It is assumed
9719 that @code{comptypes} has already been called and returned 1. This
9720 function may call @code{merge_attributes} to handle machine-independent
9724 @deftypefn {Target Hook} tree TARGET_MERGE_DECL_ATTRIBUTES (tree @var{olddecl}, tree @var{newdecl})
9725 Define this target hook if the merging of decl attributes needs special
9726 handling. If defined, the result is a list of the combined
9727 @code{DECL_ATTRIBUTES} of @var{olddecl} and @var{newdecl}.
9728 @var{newdecl} is a duplicate declaration of @var{olddecl}. Examples of
9729 when this is needed are when one attribute overrides another, or when an
9730 attribute is nullified by a subsequent definition. This function may
9731 call @code{merge_attributes} to handle machine-independent merging.
9733 @findex TARGET_DLLIMPORT_DECL_ATTRIBUTES
9734 If the only target-specific handling you require is @samp{dllimport}
9735 for Microsoft Windows targets, you should define the macro
9736 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES} to @code{1}. The compiler
9737 will then define a function called
9738 @code{merge_dllimport_decl_attributes} which can then be defined as
9739 the expansion of @code{TARGET_MERGE_DECL_ATTRIBUTES}. You can also
9740 add @code{handle_dll_attribute} in the attribute table for your port
9741 to perform initial processing of the @samp{dllimport} and
9742 @samp{dllexport} attributes. This is done in @file{i386/cygwin.h} and
9743 @file{i386/i386.c}, for example.
9746 @deftypefn {Target Hook} bool TARGET_VALID_DLLIMPORT_ATTRIBUTE_P (const_tree @var{decl})
9747 @var{decl} is a variable or function with @code{__attribute__((dllimport))} specified. Use this hook if the target needs to add extra validation checks to @code{handle_dll_attribute}.
9750 @defmac TARGET_DECLSPEC
9751 Define this macro to a nonzero value if you want to treat
9752 @code{__declspec(X)} as equivalent to @code{__attribute((X))}. By
9753 default, this behavior is enabled only for targets that define
9754 @code{TARGET_DLLIMPORT_DECL_ATTRIBUTES}. The current implementation
9755 of @code{__declspec} is via a built-in macro, but you should not rely
9756 on this implementation detail.
9759 @deftypefn {Target Hook} void TARGET_INSERT_ATTRIBUTES (tree @var{node}, tree *@var{attr_ptr})
9760 Define this target hook if you want to be able to add attributes to a decl
9761 when it is being created. This is normally useful for back ends which
9762 wish to implement a pragma by using the attributes which correspond to
9763 the pragma's effect. The @var{node} argument is the decl which is being
9764 created. The @var{attr_ptr} argument is a pointer to the attribute list
9765 for this decl. The list itself should not be modified, since it may be
9766 shared with other decls, but attributes may be chained on the head of
9767 the list and @code{*@var{attr_ptr}} modified to point to the new
9768 attributes, or a copy of the list may be made if further changes are
9772 @deftypefn {Target Hook} bool TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P (const_tree @var{fndecl})
9774 This target hook returns @code{true} if it is OK to inline @var{fndecl}
9775 into the current function, despite its having target-specific
9776 attributes, @code{false} otherwise. By default, if a function has a
9777 target specific attribute attached to it, it will not be inlined.
9780 @deftypefn {Target Hook} bool TARGET_OPTION_VALID_ATTRIBUTE_P (tree @var{fndecl}, tree @var{name}, tree @var{args}, int @var{flags})
9781 This hook is called to parse @code{attribute(target("..."))}, which
9782 allows setting target-specific options on individual functions.
9783 These function-specific options may differ
9784 from the options specified on the command line. The hook should return
9785 @code{true} if the options are valid.
9787 The hook should set the @code{DECL_FUNCTION_SPECIFIC_TARGET} field in
9788 the function declaration to hold a pointer to a target-specific
9789 @code{struct cl_target_option} structure.
9792 @deftypefn {Target Hook} void TARGET_OPTION_SAVE (struct cl_target_option *@var{ptr}, struct gcc_options *@var{opts})
9793 This hook is called to save any additional target-specific information
9794 in the @code{struct cl_target_option} structure for function-specific
9795 options from the @code{struct gcc_options} structure.
9796 @xref{Option file format}.
9799 @deftypefn {Target Hook} void TARGET_OPTION_RESTORE (struct gcc_options *@var{opts}, struct cl_target_option *@var{ptr})
9800 This hook is called to restore any additional target-specific
9801 information in the @code{struct cl_target_option} structure for
9802 function-specific options to the @code{struct gcc_options} structure.
9805 @deftypefn {Target Hook} void TARGET_OPTION_PRINT (FILE *@var{file}, int @var{indent}, struct cl_target_option *@var{ptr})
9806 This hook is called to print any additional target-specific
9807 information in the @code{struct cl_target_option} structure for
9808 function-specific options.
9811 @deftypefn {Target Hook} bool TARGET_OPTION_PRAGMA_PARSE (tree @var{args}, tree @var{pop_target})
9812 This target hook parses the options for @code{#pragma GCC target}, which
9813 sets the target-specific options for functions that occur later in the
9814 input stream. The options accepted should be the same as those handled by the
9815 @code{TARGET_OPTION_VALID_ATTRIBUTE_P} hook.
9818 @deftypefn {Target Hook} void TARGET_OPTION_OVERRIDE (void)
9819 Sometimes certain combinations of command options do not make sense on
9820 a particular target machine. You can override the hook
9821 @code{TARGET_OPTION_OVERRIDE} to take account of this. This hooks is called
9822 once just after all the command options have been parsed.
9824 Don't use this hook to turn on various extra optimizations for
9825 @option{-O}. That is what @code{TARGET_OPTION_OPTIMIZATION} is for.
9827 If you need to do something whenever the optimization level is
9828 changed via the optimize attribute or pragma, see
9829 @code{TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE}
9832 @deftypefn {Target Hook} bool TARGET_OPTION_FUNCTION_VERSIONS (tree @var{decl1}, tree @var{decl2})
9833 This target hook returns @code{true} if @var{DECL1} and @var{DECL2} are
9834 versions of the same function. @var{DECL1} and @var{DECL2} are function
9835 versions if and only if they have the same function signature and
9836 different target specific attributes, that is, they are compiled for
9837 different target machines.
9840 @deftypefn {Target Hook} bool TARGET_CAN_INLINE_P (tree @var{caller}, tree @var{callee})
9841 This target hook returns @code{false} if the @var{caller} function
9842 cannot inline @var{callee}, based on target specific information. By
9843 default, inlining is not allowed if the callee function has function
9844 specific target options and the caller does not use the same options.
9848 @section Emulating TLS
9849 @cindex Emulated TLS
9851 For targets whose psABI does not provide Thread Local Storage via
9852 specific relocations and instruction sequences, an emulation layer is
9853 used. A set of target hooks allows this emulation layer to be
9854 configured for the requirements of a particular target. For instance
9855 the psABI may in fact specify TLS support in terms of an emulation
9858 The emulation layer works by creating a control object for every TLS
9859 object. To access the TLS object, a lookup function is provided
9860 which, when given the address of the control object, will return the
9861 address of the current thread's instance of the TLS object.
9863 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_GET_ADDRESS
9864 Contains the name of the helper function that uses a TLS control
9865 object to locate a TLS instance. The default causes libgcc's
9866 emulated TLS helper function to be used.
9869 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_REGISTER_COMMON
9870 Contains the name of the helper function that should be used at
9871 program startup to register TLS objects that are implicitly
9872 initialized to zero. If this is @code{NULL}, all TLS objects will
9873 have explicit initializers. The default causes libgcc's emulated TLS
9874 registration function to be used.
9877 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_VAR_SECTION
9878 Contains the name of the section in which TLS control variables should
9879 be placed. The default of @code{NULL} allows these to be placed in
9883 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_TMPL_SECTION
9884 Contains the name of the section in which TLS initializers should be
9885 placed. The default of @code{NULL} allows these to be placed in any
9889 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_VAR_PREFIX
9890 Contains the prefix to be prepended to TLS control variable names.
9891 The default of @code{NULL} uses a target-specific prefix.
9894 @deftypevr {Target Hook} {const char *} TARGET_EMUTLS_TMPL_PREFIX
9895 Contains the prefix to be prepended to TLS initializer objects. The
9896 default of @code{NULL} uses a target-specific prefix.
9899 @deftypefn {Target Hook} tree TARGET_EMUTLS_VAR_FIELDS (tree @var{type}, tree *@var{name})
9900 Specifies a function that generates the FIELD_DECLs for a TLS control
9901 object type. @var{type} is the RECORD_TYPE the fields are for and
9902 @var{name} should be filled with the structure tag, if the default of
9903 @code{__emutls_object} is unsuitable. The default creates a type suitable
9904 for libgcc's emulated TLS function.
9907 @deftypefn {Target Hook} tree TARGET_EMUTLS_VAR_INIT (tree @var{var}, tree @var{decl}, tree @var{tmpl_addr})
9908 Specifies a function that generates the CONSTRUCTOR to initialize a
9909 TLS control object. @var{var} is the TLS control object, @var{decl}
9910 is the TLS object and @var{tmpl_addr} is the address of the
9911 initializer. The default initializes libgcc's emulated TLS control object.
9914 @deftypevr {Target Hook} bool TARGET_EMUTLS_VAR_ALIGN_FIXED
9915 Specifies whether the alignment of TLS control variable objects is
9916 fixed and should not be increased as some backends may do to optimize
9917 single objects. The default is false.
9920 @deftypevr {Target Hook} bool TARGET_EMUTLS_DEBUG_FORM_TLS_ADDRESS
9921 Specifies whether a DWARF @code{DW_OP_form_tls_address} location descriptor
9922 may be used to describe emulated TLS control objects.
9925 @node MIPS Coprocessors
9926 @section Defining coprocessor specifics for MIPS targets.
9927 @cindex MIPS coprocessor-definition macros
9929 The MIPS specification allows MIPS implementations to have as many as 4
9930 coprocessors, each with as many as 32 private registers. GCC supports
9931 accessing these registers and transferring values between the registers
9932 and memory using asm-ized variables. For example:
9935 register unsigned int cp0count asm ("c0r1");
9941 (``c0r1'' is the default name of register 1 in coprocessor 0; alternate
9942 names may be added as described below, or the default names may be
9943 overridden entirely in @code{SUBTARGET_CONDITIONAL_REGISTER_USAGE}.)
9945 Coprocessor registers are assumed to be epilogue-used; sets to them will
9946 be preserved even if it does not appear that the register is used again
9947 later in the function.
9949 Another note: according to the MIPS spec, coprocessor 1 (if present) is
9950 the FPU@. One accesses COP1 registers through standard mips
9951 floating-point support; they are not included in this mechanism.
9954 @section Parameters for Precompiled Header Validity Checking
9955 @cindex parameters, precompiled headers
9957 @deftypefn {Target Hook} {void *} TARGET_GET_PCH_VALIDITY (size_t *@var{sz})
9958 This hook returns a pointer to the data needed by
9959 @code{TARGET_PCH_VALID_P} and sets
9960 @samp{*@var{sz}} to the size of the data in bytes.
9963 @deftypefn {Target Hook} {const char *} TARGET_PCH_VALID_P (const void *@var{data}, size_t @var{sz})
9964 This hook checks whether the options used to create a PCH file are
9965 compatible with the current settings. It returns @code{NULL}
9966 if so and a suitable error message if not. Error messages will
9967 be presented to the user and must be localized using @samp{_(@var{msg})}.
9969 @var{data} is the data that was returned by @code{TARGET_GET_PCH_VALIDITY}
9970 when the PCH file was created and @var{sz} is the size of that data in bytes.
9971 It's safe to assume that the data was created by the same version of the
9972 compiler, so no format checking is needed.
9974 The default definition of @code{default_pch_valid_p} should be
9975 suitable for most targets.
9978 @deftypefn {Target Hook} {const char *} TARGET_CHECK_PCH_TARGET_FLAGS (int @var{pch_flags})
9979 If this hook is nonnull, the default implementation of
9980 @code{TARGET_PCH_VALID_P} will use it to check for compatible values
9981 of @code{target_flags}. @var{pch_flags} specifies the value that
9982 @code{target_flags} had when the PCH file was created. The return
9983 value is the same as for @code{TARGET_PCH_VALID_P}.
9986 @deftypefn {Target Hook} void TARGET_PREPARE_PCH_SAVE (void)
9987 Called before writing out a PCH file. If the target has some
9988 garbage-collected data that needs to be in a particular state on PCH loads,
9989 it can use this hook to enforce that state. Very few targets need
9990 to do anything here.
9994 @section C++ ABI parameters
9995 @cindex parameters, c++ abi
9997 @deftypefn {Target Hook} tree TARGET_CXX_GUARD_TYPE (void)
9998 Define this hook to override the integer type used for guard variables.
9999 These are used to implement one-time construction of static objects. The
10000 default is long_long_integer_type_node.
10003 @deftypefn {Target Hook} bool TARGET_CXX_GUARD_MASK_BIT (void)
10004 This hook determines how guard variables are used. It should return
10005 @code{false} (the default) if the first byte should be used. A return value of
10006 @code{true} indicates that only the least significant bit should be used.
10009 @deftypefn {Target Hook} tree TARGET_CXX_GET_COOKIE_SIZE (tree @var{type})
10010 This hook returns the size of the cookie to use when allocating an array
10011 whose elements have the indicated @var{type}. Assumes that it is already
10012 known that a cookie is needed. The default is
10013 @code{max(sizeof (size_t), alignof(type))}, as defined in section 2.7 of the
10014 IA64/Generic C++ ABI@.
10017 @deftypefn {Target Hook} bool TARGET_CXX_COOKIE_HAS_SIZE (void)
10018 This hook should return @code{true} if the element size should be stored in
10019 array cookies. The default is to return @code{false}.
10022 @deftypefn {Target Hook} int TARGET_CXX_IMPORT_EXPORT_CLASS (tree @var{type}, int @var{import_export})
10023 If defined by a backend this hook allows the decision made to export
10024 class @var{type} to be overruled. Upon entry @var{import_export}
10025 will contain 1 if the class is going to be exported, @minus{}1 if it is going
10026 to be imported and 0 otherwise. This function should return the
10027 modified value and perform any other actions necessary to support the
10028 backend's targeted operating system.
10031 @deftypefn {Target Hook} bool TARGET_CXX_CDTOR_RETURNS_THIS (void)
10032 This hook should return @code{true} if constructors and destructors return
10033 the address of the object created/destroyed. The default is to return
10037 @deftypefn {Target Hook} bool TARGET_CXX_KEY_METHOD_MAY_BE_INLINE (void)
10038 This hook returns true if the key method for a class (i.e., the method
10039 which, if defined in the current translation unit, causes the virtual
10040 table to be emitted) may be an inline function. Under the standard
10041 Itanium C++ ABI the key method may be an inline function so long as
10042 the function is not declared inline in the class definition. Under
10043 some variants of the ABI, an inline function can never be the key
10044 method. The default is to return @code{true}.
10047 @deftypefn {Target Hook} void TARGET_CXX_DETERMINE_CLASS_DATA_VISIBILITY (tree @var{decl})
10048 @var{decl} is a virtual table, virtual table table, typeinfo object, or other similar implicit class data object that will be emitted with external linkage in this translation unit. No ELF visibility has been explicitly specified. If the target needs to specify a visibility other than that of the containing class, use this hook to set @code{DECL_VISIBILITY} and @code{DECL_VISIBILITY_SPECIFIED}.
10051 @deftypefn {Target Hook} bool TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT (void)
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 @deftypefn {Target Hook} bool TARGET_CXX_LIBRARY_RTTI_COMDAT (void)
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 @deftypefn {Target Hook} bool TARGET_CXX_USE_AEABI_ATEXIT (void)
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 @deftypefn {Target Hook} bool TARGET_CXX_USE_ATEXIT_FOR_CXA_ATEXIT (void)
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 @deftypefn {Target Hook} void TARGET_CXX_ADJUST_CLASS_AT_DEFINITION (tree @var{type})
10080 @var{type} is a C++ class (i.e., RECORD_TYPE or UNION_TYPE) that has just been defined. Use this hook to make adjustments to the class (eg, tweak visibility or perform any other required target modifications).
10083 @deftypefn {Target Hook} tree TARGET_CXX_DECL_MANGLING_CONTEXT (const_tree @var{decl})
10084 Return target-specific mangling context of @var{decl} or @code{NULL_TREE}.
10087 @node Named Address Spaces
10088 @section Adding support for named address spaces
10089 @cindex named address spaces
10091 The draft technical report of the ISO/IEC JTC1 S22 WG14 N1275
10092 standards committee, @cite{Programming Languages - C - Extensions to
10093 support embedded processors}, specifies a syntax for embedded
10094 processors to specify alternate address spaces. You can configure a
10095 GCC port to support section 5.1 of the draft report to add support for
10096 address spaces other than the default address space. These address
10097 spaces are new keywords that are similar to the @code{volatile} and
10098 @code{const} type attributes.
10100 Pointers to named address spaces can have a different size than
10101 pointers to the generic address space.
10103 For example, the SPU port uses the @code{__ea} address space to refer
10104 to memory in the host processor, rather than memory local to the SPU
10105 processor. Access to memory in the @code{__ea} address space involves
10106 issuing DMA operations to move data between the host processor and the
10107 local processor memory address space. Pointers in the @code{__ea}
10108 address space are either 32 bits or 64 bits based on the
10109 @option{-mea32} or @option{-mea64} switches (native SPU pointers are
10112 Internally, address spaces are represented as a small integer in the
10113 range 0 to 15 with address space 0 being reserved for the generic
10116 To register a named address space qualifier keyword with the C front end,
10117 the target may call the @code{c_register_addr_space} routine. For example,
10118 the SPU port uses the following to declare @code{__ea} as the keyword for
10119 named address space #1:
10121 #define ADDR_SPACE_EA 1
10122 c_register_addr_space ("__ea", ADDR_SPACE_EA);
10125 @deftypefn {Target Hook} machine_mode TARGET_ADDR_SPACE_POINTER_MODE (addr_space_t @var{address_space})
10126 Define this to return the machine mode to use for pointers to
10127 @var{address_space} if the target supports named address spaces.
10128 The default version of this hook returns @code{ptr_mode} for the
10129 generic address space only.
10132 @deftypefn {Target Hook} machine_mode TARGET_ADDR_SPACE_ADDRESS_MODE (addr_space_t @var{address_space})
10133 Define this to return the machine mode to use for addresses in
10134 @var{address_space} if the target supports named address spaces.
10135 The default version of this hook returns @code{Pmode} for the
10136 generic address space only.
10139 @deftypefn {Target Hook} bool TARGET_ADDR_SPACE_VALID_POINTER_MODE (machine_mode @var{mode}, addr_space_t @var{as})
10140 Define this to return nonzero if the port can handle pointers
10141 with machine mode @var{mode} to address space @var{as}. This target
10142 hook is the same as the @code{TARGET_VALID_POINTER_MODE} target hook,
10143 except that it includes explicit named address space support. The default
10144 version of this hook returns true for the modes returned by either the
10145 @code{TARGET_ADDR_SPACE_POINTER_MODE} or @code{TARGET_ADDR_SPACE_ADDRESS_MODE}
10146 target hooks for the given address space.
10149 @deftypefn {Target Hook} bool TARGET_ADDR_SPACE_LEGITIMATE_ADDRESS_P (machine_mode @var{mode}, rtx @var{exp}, bool @var{strict}, addr_space_t @var{as})
10150 Define this to return true if @var{exp} is a valid address for mode
10151 @var{mode} in the named address space @var{as}. The @var{strict}
10152 parameter says whether strict addressing is in effect after reload has
10153 finished. This target hook is the same as the
10154 @code{TARGET_LEGITIMATE_ADDRESS_P} target hook, except that it includes
10155 explicit named address space support.
10158 @deftypefn {Target Hook} rtx TARGET_ADDR_SPACE_LEGITIMIZE_ADDRESS (rtx @var{x}, rtx @var{oldx}, machine_mode @var{mode}, addr_space_t @var{as})
10159 Define this to modify an invalid address @var{x} to be a valid address
10160 with mode @var{mode} in the named address space @var{as}. This target
10161 hook is the same as the @code{TARGET_LEGITIMIZE_ADDRESS} target hook,
10162 except that it includes explicit named address space support.
10165 @deftypefn {Target Hook} bool TARGET_ADDR_SPACE_SUBSET_P (addr_space_t @var{subset}, addr_space_t @var{superset})
10166 Define this to return whether the @var{subset} named address space is
10167 contained within the @var{superset} named address space. Pointers to
10168 a named address space that is a subset of another named address space
10169 will be converted automatically without a cast if used together in
10170 arithmetic operations. Pointers to a superset address space can be
10171 converted to pointers to a subset address space via explicit casts.
10174 @deftypefn {Target Hook} rtx TARGET_ADDR_SPACE_CONVERT (rtx @var{op}, tree @var{from_type}, tree @var{to_type})
10175 Define this to convert the pointer expression represented by the RTL
10176 @var{op} with type @var{from_type} that points to a named address
10177 space to a new pointer expression with type @var{to_type} that points
10178 to a different named address space. When this hook it called, it is
10179 guaranteed that one of the two address spaces is a subset of the other,
10180 as determined by the @code{TARGET_ADDR_SPACE_SUBSET_P} target hook.
10184 @section Miscellaneous Parameters
10185 @cindex parameters, miscellaneous
10187 @c prevent bad page break with this line
10188 Here are several miscellaneous parameters.
10190 @defmac HAS_LONG_COND_BRANCH
10191 Define this boolean macro to indicate whether or not your architecture
10192 has conditional branches that can span all of memory. It is used in
10193 conjunction with an optimization that partitions hot and cold basic
10194 blocks into separate sections of the executable. If this macro is
10195 set to false, gcc will convert any conditional branches that attempt
10196 to cross between sections into unconditional branches or indirect jumps.
10199 @defmac HAS_LONG_UNCOND_BRANCH
10200 Define this boolean macro to indicate whether or not your architecture
10201 has unconditional branches that can span all of memory. It is used in
10202 conjunction with an optimization that partitions hot and cold basic
10203 blocks into separate sections of the executable. If this macro is
10204 set to false, gcc will convert any unconditional branches that attempt
10205 to cross between sections into indirect jumps.
10208 @defmac CASE_VECTOR_MODE
10209 An alias for a machine mode name. This is the machine mode that
10210 elements of a jump-table should have.
10213 @defmac CASE_VECTOR_SHORTEN_MODE (@var{min_offset}, @var{max_offset}, @var{body})
10214 Optional: return the preferred mode for an @code{addr_diff_vec}
10215 when the minimum and maximum offset are known. If you define this,
10216 it enables extra code in branch shortening to deal with @code{addr_diff_vec}.
10217 To make this work, you also have to define @code{INSN_ALIGN} and
10218 make the alignment for @code{addr_diff_vec} explicit.
10219 The @var{body} argument is provided so that the offset_unsigned and scale
10220 flags can be updated.
10223 @defmac CASE_VECTOR_PC_RELATIVE
10224 Define this macro to be a C expression to indicate when jump-tables
10225 should contain relative addresses. You need not define this macro if
10226 jump-tables never contain relative addresses, or jump-tables should
10227 contain relative addresses only when @option{-fPIC} or @option{-fPIC}
10231 @deftypefn {Target Hook} {unsigned int} TARGET_CASE_VALUES_THRESHOLD (void)
10232 This function return the smallest number of different values for which it
10233 is best to use a jump-table instead of a tree of conditional branches.
10234 The default is four for machines with a @code{casesi} instruction and
10235 five otherwise. This is best for most machines.
10238 @defmac WORD_REGISTER_OPERATIONS
10239 Define this macro if operations between registers with integral mode
10240 smaller than a word are always performed on the entire register.
10241 Most RISC machines have this property and most CISC machines do not.
10244 @defmac LOAD_EXTEND_OP (@var{mem_mode})
10245 Define this macro to be a C expression indicating when insns that read
10246 memory in @var{mem_mode}, an integral mode narrower than a word, set the
10247 bits outside of @var{mem_mode} to be either the sign-extension or the
10248 zero-extension of the data read. Return @code{SIGN_EXTEND} for values
10249 of @var{mem_mode} for which the
10250 insn sign-extends, @code{ZERO_EXTEND} for which it zero-extends, and
10251 @code{UNKNOWN} for other modes.
10253 This macro is not called with @var{mem_mode} non-integral or with a width
10254 greater than or equal to @code{BITS_PER_WORD}, so you may return any
10255 value in this case. Do not define this macro if it would always return
10256 @code{UNKNOWN}. On machines where this macro is defined, you will normally
10257 define it as the constant @code{SIGN_EXTEND} or @code{ZERO_EXTEND}.
10259 You may return a non-@code{UNKNOWN} value even if for some hard registers
10260 the sign extension is not performed, if for the @code{REGNO_REG_CLASS}
10261 of these hard registers @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero
10262 when the @var{from} mode is @var{mem_mode} and the @var{to} mode is any
10263 integral mode larger than this but not larger than @code{word_mode}.
10265 You must return @code{UNKNOWN} if for some hard registers that allow this
10266 mode, @code{CANNOT_CHANGE_MODE_CLASS} says that they cannot change to
10267 @code{word_mode}, but that they can change to another integral mode that
10268 is larger then @var{mem_mode} but still smaller than @code{word_mode}.
10271 @defmac SHORT_IMMEDIATES_SIGN_EXTEND
10272 Define this macro if loading short immediate values into registers sign
10276 @deftypefn {Target Hook} {unsigned int} TARGET_MIN_DIVISIONS_FOR_RECIP_MUL (machine_mode @var{mode})
10277 When @option{-ffast-math} is in effect, GCC tries to optimize
10278 divisions by the same divisor, by turning them into multiplications by
10279 the reciprocal. This target hook specifies the minimum number of divisions
10280 that should be there for GCC to perform the optimization for a variable
10281 of mode @var{mode}. The default implementation returns 3 if the machine
10282 has an instruction for the division, and 2 if it does not.
10286 The maximum number of bytes that a single instruction can move quickly
10287 between memory and registers or between two memory locations.
10290 @defmac MAX_MOVE_MAX
10291 The maximum number of bytes that a single instruction can move quickly
10292 between memory and registers or between two memory locations. If this
10293 is undefined, the default is @code{MOVE_MAX}. Otherwise, it is the
10294 constant value that is the largest value that @code{MOVE_MAX} can have
10298 @defmac SHIFT_COUNT_TRUNCATED
10299 A C expression that is nonzero if on this machine the number of bits
10300 actually used for the count of a shift operation is equal to the number
10301 of bits needed to represent the size of the object being shifted. When
10302 this macro is nonzero, the compiler will assume that it is safe to omit
10303 a sign-extend, zero-extend, and certain bitwise `and' instructions that
10304 truncates the count of a shift operation. On machines that have
10305 instructions that act on bit-fields at variable positions, which may
10306 include `bit test' instructions, a nonzero @code{SHIFT_COUNT_TRUNCATED}
10307 also enables deletion of truncations of the values that serve as
10308 arguments to bit-field instructions.
10310 If both types of instructions truncate the count (for shifts) and
10311 position (for bit-field operations), or if no variable-position bit-field
10312 instructions exist, you should define this macro.
10314 However, on some machines, such as the 80386 and the 680x0, truncation
10315 only applies to shift operations and not the (real or pretended)
10316 bit-field operations. Define @code{SHIFT_COUNT_TRUNCATED} to be zero on
10317 such machines. Instead, add patterns to the @file{md} file that include
10318 the implied truncation of the shift instructions.
10320 You need not define this macro if it would always have the value of zero.
10323 @anchor{TARGET_SHIFT_TRUNCATION_MASK}
10324 @deftypefn {Target Hook} {unsigned HOST_WIDE_INT} TARGET_SHIFT_TRUNCATION_MASK (machine_mode @var{mode})
10325 This function describes how the standard shift patterns for @var{mode}
10326 deal with shifts by negative amounts or by more than the width of the mode.
10327 @xref{shift patterns}.
10329 On many machines, the shift patterns will apply a mask @var{m} to the
10330 shift count, meaning that a fixed-width shift of @var{x} by @var{y} is
10331 equivalent to an arbitrary-width shift of @var{x} by @var{y & m}. If
10332 this is true for mode @var{mode}, the function should return @var{m},
10333 otherwise it should return 0. A return value of 0 indicates that no
10334 particular behavior is guaranteed.
10336 Note that, unlike @code{SHIFT_COUNT_TRUNCATED}, this function does
10337 @emph{not} apply to general shift rtxes; it applies only to instructions
10338 that are generated by the named shift patterns.
10340 The default implementation of this function returns
10341 @code{GET_MODE_BITSIZE (@var{mode}) - 1} if @code{SHIFT_COUNT_TRUNCATED}
10342 and 0 otherwise. This definition is always safe, but if
10343 @code{SHIFT_COUNT_TRUNCATED} is false, and some shift patterns
10344 nevertheless truncate the shift count, you may get better code
10348 @defmac TRULY_NOOP_TRUNCATION (@var{outprec}, @var{inprec})
10349 A C expression which is nonzero if on this machine it is safe to
10350 ``convert'' an integer of @var{inprec} bits to one of @var{outprec}
10351 bits (where @var{outprec} is smaller than @var{inprec}) by merely
10352 operating on it as if it had only @var{outprec} bits.
10354 On many machines, this expression can be 1.
10356 @c rearranged this, removed the phrase "it is reported that". this was
10357 @c to fix an overfull hbox. --mew 10feb93
10358 When @code{TRULY_NOOP_TRUNCATION} returns 1 for a pair of sizes for
10359 modes for which @code{MODES_TIEABLE_P} is 0, suboptimal code can result.
10360 If this is the case, making @code{TRULY_NOOP_TRUNCATION} return 0 in
10361 such cases may improve things.
10364 @deftypefn {Target Hook} int TARGET_MODE_REP_EXTENDED (machine_mode @var{mode}, machine_mode @var{rep_mode})
10365 The representation of an integral mode can be such that the values
10366 are always extended to a wider integral mode. Return
10367 @code{SIGN_EXTEND} if values of @var{mode} are represented in
10368 sign-extended form to @var{rep_mode}. Return @code{UNKNOWN}
10369 otherwise. (Currently, none of the targets use zero-extended
10370 representation this way so unlike @code{LOAD_EXTEND_OP},
10371 @code{TARGET_MODE_REP_EXTENDED} is expected to return either
10372 @code{SIGN_EXTEND} or @code{UNKNOWN}. Also no target extends
10373 @var{mode} to @var{rep_mode} so that @var{rep_mode} is not the next
10374 widest integral mode and currently we take advantage of this fact.)
10376 Similarly to @code{LOAD_EXTEND_OP} you may return a non-@code{UNKNOWN}
10377 value even if the extension is not performed on certain hard registers
10378 as long as for the @code{REGNO_REG_CLASS} of these hard registers
10379 @code{CANNOT_CHANGE_MODE_CLASS} returns nonzero.
10381 Note that @code{TARGET_MODE_REP_EXTENDED} and @code{LOAD_EXTEND_OP}
10382 describe two related properties. If you define
10383 @code{TARGET_MODE_REP_EXTENDED (mode, word_mode)} you probably also want
10384 to define @code{LOAD_EXTEND_OP (mode)} to return the same type of
10387 In order to enforce the representation of @code{mode},
10388 @code{TRULY_NOOP_TRUNCATION} should return false when truncating to
10392 @defmac STORE_FLAG_VALUE
10393 A C expression describing the value returned by a comparison operator
10394 with an integral mode and stored by a store-flag instruction
10395 (@samp{cstore@var{mode}4}) when the condition is true. This description must
10396 apply to @emph{all} the @samp{cstore@var{mode}4} patterns and all the
10397 comparison operators whose results have a @code{MODE_INT} mode.
10399 A value of 1 or @minus{}1 means that the instruction implementing the
10400 comparison operator returns exactly 1 or @minus{}1 when the comparison is true
10401 and 0 when the comparison is false. Otherwise, the value indicates
10402 which bits of the result are guaranteed to be 1 when the comparison is
10403 true. This value is interpreted in the mode of the comparison
10404 operation, which is given by the mode of the first operand in the
10405 @samp{cstore@var{mode}4} pattern. Either the low bit or the sign bit of
10406 @code{STORE_FLAG_VALUE} be on. Presently, only those bits are used by
10409 If @code{STORE_FLAG_VALUE} is neither 1 or @minus{}1, the compiler will
10410 generate code that depends only on the specified bits. It can also
10411 replace comparison operators with equivalent operations if they cause
10412 the required bits to be set, even if the remaining bits are undefined.
10413 For example, on a machine whose comparison operators return an
10414 @code{SImode} value and where @code{STORE_FLAG_VALUE} is defined as
10415 @samp{0x80000000}, saying that just the sign bit is relevant, the
10419 (ne:SI (and:SI @var{x} (const_int @var{power-of-2})) (const_int 0))
10423 can be converted to
10426 (ashift:SI @var{x} (const_int @var{n}))
10430 where @var{n} is the appropriate shift count to move the bit being
10431 tested into the sign bit.
10433 There is no way to describe a machine that always sets the low-order bit
10434 for a true value, but does not guarantee the value of any other bits,
10435 but we do not know of any machine that has such an instruction. If you
10436 are trying to port GCC to such a machine, include an instruction to
10437 perform a logical-and of the result with 1 in the pattern for the
10438 comparison operators and let us know at @email{gcc@@gcc.gnu.org}.
10440 Often, a machine will have multiple instructions that obtain a value
10441 from a comparison (or the condition codes). Here are rules to guide the
10442 choice of value for @code{STORE_FLAG_VALUE}, and hence the instructions
10447 Use the shortest sequence that yields a valid definition for
10448 @code{STORE_FLAG_VALUE}. It is more efficient for the compiler to
10449 ``normalize'' the value (convert it to, e.g., 1 or 0) than for the
10450 comparison operators to do so because there may be opportunities to
10451 combine the normalization with other operations.
10454 For equal-length sequences, use a value of 1 or @minus{}1, with @minus{}1 being
10455 slightly preferred on machines with expensive jumps and 1 preferred on
10459 As a second choice, choose a value of @samp{0x80000001} if instructions
10460 exist that set both the sign and low-order bits but do not define the
10464 Otherwise, use a value of @samp{0x80000000}.
10467 Many machines can produce both the value chosen for
10468 @code{STORE_FLAG_VALUE} and its negation in the same number of
10469 instructions. On those machines, you should also define a pattern for
10470 those cases, e.g., one matching
10473 (set @var{A} (neg:@var{m} (ne:@var{m} @var{B} @var{C})))
10476 Some machines can also perform @code{and} or @code{plus} operations on
10477 condition code values with less instructions than the corresponding
10478 @samp{cstore@var{mode}4} insn followed by @code{and} or @code{plus}. On those
10479 machines, define the appropriate patterns. Use the names @code{incscc}
10480 and @code{decscc}, respectively, for the patterns which perform
10481 @code{plus} or @code{minus} operations on condition code values. See
10482 @file{rs6000.md} for some examples. The GNU Superoptimizer can be used to
10483 find such instruction sequences on other machines.
10485 If this macro is not defined, the default value, 1, is used. You need
10486 not define @code{STORE_FLAG_VALUE} if the machine has no store-flag
10487 instructions, or if the value generated by these instructions is 1.
10490 @defmac FLOAT_STORE_FLAG_VALUE (@var{mode})
10491 A C expression that gives a nonzero @code{REAL_VALUE_TYPE} value that is
10492 returned when comparison operators with floating-point results are true.
10493 Define this macro on machines that have comparison operations that return
10494 floating-point values. If there are no such operations, do not define
10498 @defmac VECTOR_STORE_FLAG_VALUE (@var{mode})
10499 A C expression that gives a rtx representing the nonzero true element
10500 for vector comparisons. The returned rtx should be valid for the inner
10501 mode of @var{mode} which is guaranteed to be a vector mode. Define
10502 this macro on machines that have vector comparison operations that
10503 return a vector result. If there are no such operations, do not define
10504 this macro. Typically, this macro is defined as @code{const1_rtx} or
10505 @code{constm1_rtx}. This macro may return @code{NULL_RTX} to prevent
10506 the compiler optimizing such vector comparison operations for the
10510 @defmac CLZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
10511 @defmacx CTZ_DEFINED_VALUE_AT_ZERO (@var{mode}, @var{value})
10512 A C expression that indicates whether the architecture defines a value
10513 for @code{clz} or @code{ctz} with a zero operand.
10514 A result of @code{0} indicates the value is undefined.
10515 If the value is defined for only the RTL expression, the macro should
10516 evaluate to @code{1}; if the value applies also to the corresponding optab
10517 entry (which is normally the case if it expands directly into
10518 the corresponding RTL), then the macro should evaluate to @code{2}.
10519 In the cases where the value is defined, @var{value} should be set to
10522 If this macro is not defined, the value of @code{clz} or
10523 @code{ctz} at zero is assumed to be undefined.
10525 This macro must be defined if the target's expansion for @code{ffs}
10526 relies on a particular value to get correct results. Otherwise it
10527 is not necessary, though it may be used to optimize some corner cases, and
10528 to provide a default expansion for the @code{ffs} optab.
10530 Note that regardless of this macro the ``definedness'' of @code{clz}
10531 and @code{ctz} at zero do @emph{not} extend to the builtin functions
10532 visible to the user. Thus one may be free to adjust the value at will
10533 to match the target expansion of these operations without fear of
10538 An alias for the machine mode for pointers. On most machines, define
10539 this to be the integer mode corresponding to the width of a hardware
10540 pointer; @code{SImode} on 32-bit machine or @code{DImode} on 64-bit machines.
10541 On some machines you must define this to be one of the partial integer
10542 modes, such as @code{PSImode}.
10544 The width of @code{Pmode} must be at least as large as the value of
10545 @code{POINTER_SIZE}. If it is not equal, you must define the macro
10546 @code{POINTERS_EXTEND_UNSIGNED} to specify how pointers are extended
10550 @defmac FUNCTION_MODE
10551 An alias for the machine mode used for memory references to functions
10552 being called, in @code{call} RTL expressions. On most CISC machines,
10553 where an instruction can begin at any byte address, this should be
10554 @code{QImode}. On most RISC machines, where all instructions have fixed
10555 size and alignment, this should be a mode with the same size and alignment
10556 as the machine instruction words - typically @code{SImode} or @code{HImode}.
10559 @defmac STDC_0_IN_SYSTEM_HEADERS
10560 In normal operation, the preprocessor expands @code{__STDC__} to the
10561 constant 1, to signify that GCC conforms to ISO Standard C@. On some
10562 hosts, like Solaris, the system compiler uses a different convention,
10563 where @code{__STDC__} is normally 0, but is 1 if the user specifies
10564 strict conformance to the C Standard.
10566 Defining @code{STDC_0_IN_SYSTEM_HEADERS} makes GNU CPP follows the host
10567 convention when processing system header files, but when processing user
10568 files @code{__STDC__} will always expand to 1.
10571 @deftypefn {C Target Hook} {const char *} TARGET_C_PREINCLUDE (void)
10572 Define this hook to return the name of a header file to be included at the start of all compilations, as if it had been included with @code{#include <@var{file}>}. If this hook returns @code{NULL}, or is not defined, or the header is not found, or if the user specifies @option{-ffreestanding} or @option{-nostdinc}, no header is included.
10574 This hook can be used together with a header provided by the system C library to implement ISO C requirements for certain macros to be predefined that describe properties of the whole implementation rather than just the compiler.
10577 @deftypefn {C Target Hook} bool TARGET_CXX_IMPLICIT_EXTERN_C (const char*@var{})
10578 Define this hook to add target-specific C++ implicit extern C functions. If this function returns true for the name of a file-scope function, that function implicitly gets extern "C" linkage rather than whatever language linkage the declaration would normally have. An example of such function is WinMain on Win32 targets.
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 @defmac TARGET_DEFAULT_PACK_STRUCT
10651 If your target requires a structure packing default other than 0 (meaning
10652 the machine default), define this macro to the necessary value (in bytes).
10653 This must be a value that would also be valid to use with
10654 @samp{#pragma pack()} (that is, a small power of two).
10657 @defmac DOLLARS_IN_IDENTIFIERS
10658 Define this macro to control use of the character @samp{$} in
10659 identifier names for the C family of languages. 0 means @samp{$} is
10660 not allowed by default; 1 means it is allowed. 1 is the default;
10661 there is no need to define this macro in that case.
10664 @defmac INSN_SETS_ARE_DELAYED (@var{insn})
10665 Define this macro as a C expression that is nonzero if it is safe for the
10666 delay slot scheduler to place instructions in the delay slot of @var{insn},
10667 even if they appear to use a resource set or clobbered in @var{insn}.
10668 @var{insn} is always a @code{jump_insn} or an @code{insn}; GCC knows that
10669 every @code{call_insn} has this behavior. On machines where some @code{insn}
10670 or @code{jump_insn} is really a function call and hence has this behavior,
10671 you should define this macro.
10673 You need not define this macro if it would always return zero.
10676 @defmac INSN_REFERENCES_ARE_DELAYED (@var{insn})
10677 Define this macro as a C expression that is nonzero if it is safe for the
10678 delay slot scheduler to place instructions in the delay slot of @var{insn},
10679 even if they appear to set or clobber a resource referenced in @var{insn}.
10680 @var{insn} is always a @code{jump_insn} or an @code{insn}. On machines where
10681 some @code{insn} or @code{jump_insn} is really a function call and its operands
10682 are registers whose use is actually in the subroutine it calls, you should
10683 define this macro. Doing so allows the delay slot scheduler to move
10684 instructions which copy arguments into the argument registers into the delay
10685 slot of @var{insn}.
10687 You need not define this macro if it would always return zero.
10690 @defmac MULTIPLE_SYMBOL_SPACES
10691 Define this macro as a C expression that is nonzero if, in some cases,
10692 global symbols from one translation unit may not be bound to undefined
10693 symbols in another translation unit without user intervention. For
10694 instance, under Microsoft Windows symbols must be explicitly imported
10695 from shared libraries (DLLs).
10697 You need not define this macro if it would always evaluate to zero.
10700 @deftypefn {Target Hook} tree TARGET_MD_ASM_CLOBBERS (tree @var{outputs}, tree @var{inputs}, tree @var{clobbers})
10701 This target hook should add to @var{clobbers} @code{STRING_CST} trees for
10702 any hard regs the port wishes to automatically clobber for an asm.
10703 It should return the result of the last @code{tree_cons} used to add a
10704 clobber. The @var{outputs}, @var{inputs} and @var{clobber} lists are the
10705 corresponding parameters to the asm and may be inspected to avoid
10706 clobbering a register that is an input or output of the asm. You can use
10707 @code{tree_overlaps_hard_reg_set}, declared in @file{tree.h}, to test
10708 for overlap with regards to asm-declared registers.
10711 @defmac MATH_LIBRARY
10712 Define this macro as a C string constant for the linker argument to link
10713 in the system math library, minus the initial @samp{"-l"}, or
10714 @samp{""} if the target does not have a
10715 separate math library.
10717 You need only define this macro if the default of @samp{"m"} is wrong.
10720 @defmac LIBRARY_PATH_ENV
10721 Define this macro as a C string constant for the environment variable that
10722 specifies where the linker should look for libraries.
10724 You need only define this macro if the default of @samp{"LIBRARY_PATH"}
10728 @defmac TARGET_POSIX_IO
10729 Define this macro if the target supports the following POSIX@ file
10730 functions, access, mkdir and file locking with fcntl / F_SETLKW@.
10731 Defining @code{TARGET_POSIX_IO} will enable the test coverage code
10732 to use file locking when exiting a program, which avoids race conditions
10733 if the program has forked. It will also create directories at run-time
10734 for cross-profiling.
10737 @defmac MAX_CONDITIONAL_EXECUTE
10739 A C expression for the maximum number of instructions to execute via
10740 conditional execution instructions instead of a branch. A value of
10741 @code{BRANCH_COST}+1 is the default if the machine does not use cc0, and
10742 1 if it does use cc0.
10745 @defmac IFCVT_MODIFY_TESTS (@var{ce_info}, @var{true_expr}, @var{false_expr})
10746 Used if the target needs to perform machine-dependent modifications on the
10747 conditionals used for turning basic blocks into conditionally executed code.
10748 @var{ce_info} points to a data structure, @code{struct ce_if_block}, which
10749 contains information about the currently processed blocks. @var{true_expr}
10750 and @var{false_expr} are the tests that are used for converting the
10751 then-block and the else-block, respectively. Set either @var{true_expr} or
10752 @var{false_expr} to a null pointer if the tests cannot be converted.
10755 @defmac IFCVT_MODIFY_MULTIPLE_TESTS (@var{ce_info}, @var{bb}, @var{true_expr}, @var{false_expr})
10756 Like @code{IFCVT_MODIFY_TESTS}, but used when converting more complicated
10757 if-statements into conditions combined by @code{and} and @code{or} operations.
10758 @var{bb} contains the basic block that contains the test that is currently
10759 being processed and about to be turned into a condition.
10762 @defmac IFCVT_MODIFY_INSN (@var{ce_info}, @var{pattern}, @var{insn})
10763 A C expression to modify the @var{PATTERN} of an @var{INSN} that is to
10764 be converted to conditional execution format. @var{ce_info} points to
10765 a data structure, @code{struct ce_if_block}, which contains information
10766 about the currently processed blocks.
10769 @defmac IFCVT_MODIFY_FINAL (@var{ce_info})
10770 A C expression to perform any final machine dependent modifications in
10771 converting code to conditional execution. The involved basic blocks
10772 can be found in the @code{struct ce_if_block} structure that is pointed
10773 to by @var{ce_info}.
10776 @defmac IFCVT_MODIFY_CANCEL (@var{ce_info})
10777 A C expression to cancel any machine dependent modifications in
10778 converting code to conditional execution. The involved basic blocks
10779 can be found in the @code{struct ce_if_block} structure that is pointed
10780 to by @var{ce_info}.
10783 @defmac IFCVT_MACHDEP_INIT (@var{ce_info})
10784 A C expression to initialize any machine specific data for if-conversion
10785 of the if-block in the @code{struct ce_if_block} structure that is pointed
10786 to by @var{ce_info}.
10789 @deftypefn {Target Hook} void TARGET_MACHINE_DEPENDENT_REORG (void)
10790 If non-null, this hook performs a target-specific pass over the
10791 instruction stream. The compiler will run it at all optimization levels,
10792 just before the point at which it normally does delayed-branch scheduling.
10794 The exact purpose of the hook varies from target to target. Some use
10795 it to do transformations that are necessary for correctness, such as
10796 laying out in-function constant pools or avoiding hardware hazards.
10797 Others use it as an opportunity to do some machine-dependent optimizations.
10799 You need not implement the hook if it has nothing to do. The default
10800 definition is null.
10803 @deftypefn {Target Hook} void TARGET_INIT_BUILTINS (void)
10804 Define this hook if you have any machine-specific built-in functions
10805 that need to be defined. It should be a function that performs the
10808 Machine specific built-in functions can be useful to expand special machine
10809 instructions that would otherwise not normally be generated because
10810 they have no equivalent in the source language (for example, SIMD vector
10811 instructions or prefetch instructions).
10813 To create a built-in function, call the function
10814 @code{lang_hooks.builtin_function}
10815 which is defined by the language front end. You can use any type nodes set
10816 up by @code{build_common_tree_nodes};
10817 only language front ends that use those two functions will call
10818 @samp{TARGET_INIT_BUILTINS}.
10821 @deftypefn {Target Hook} tree TARGET_BUILTIN_DECL (unsigned @var{code}, bool @var{initialize_p})
10822 Define this hook if you have any machine-specific built-in functions
10823 that need to be defined. It should be a function that returns the
10824 builtin function declaration for the builtin function code @var{code}.
10825 If there is no such builtin and it cannot be initialized at this time
10826 if @var{initialize_p} is true the function should return @code{NULL_TREE}.
10827 If @var{code} is out of range the function should return
10828 @code{error_mark_node}.
10831 @deftypefn {Target Hook} rtx TARGET_EXPAND_BUILTIN (tree @var{exp}, rtx @var{target}, rtx @var{subtarget}, machine_mode @var{mode}, int @var{ignore})
10833 Expand a call to a machine specific built-in function that was set up by
10834 @samp{TARGET_INIT_BUILTINS}. @var{exp} is the expression for the
10835 function call; the result should go to @var{target} if that is
10836 convenient, and have mode @var{mode} if that is convenient.
10837 @var{subtarget} may be used as the target for computing one of
10838 @var{exp}'s operands. @var{ignore} is nonzero if the value is to be
10839 ignored. This function should return the result of the call to the
10843 @deftypefn {Target Hook} tree TARGET_BUILTIN_CHKP_FUNCTION (unsigned @var{fcode})
10844 This hook allows target to redefine built-in functions used by
10845 Pointer Bounds Checker for code instrumentation. Hook should return
10846 fndecl of function implementing generic builtin whose code is
10847 passed in @var{fcode}. Currently following built-in functions are
10848 obtained using this hook:
10849 @deftypefn {Built-in Function} __bounds_type __chkp_bndmk (const void *@var{lb}, size_t @var{size})
10850 Function code - BUILT_IN_CHKP_BNDMK. This built-in function is used
10851 by Pointer Bounds Checker to create bound values. @var{lb} holds low
10852 bound of the resulting bounds. @var{size} holds size of created bounds.
10855 @deftypefn {Built-in Function} void __chkp_bndstx (const void *@var{ptr}, __bounds_type @var{b}, const void **@var{loc})
10856 Function code - @code{BUILT_IN_CHKP_BNDSTX}. This built-in function is used
10857 by Pointer Bounds Checker to store bounds @var{b} for pointer @var{ptr}
10858 when @var{ptr} is stored by address @var{loc}.
10861 @deftypefn {Built-in Function} __bounds_type __chkp_bndldx (const void **@var{loc}, const void *@var{ptr})
10862 Function code - @code{BUILT_IN_CHKP_BNDLDX}. This built-in function is used
10863 by Pointer Bounds Checker to get bounds of pointer @var{ptr} loaded by
10867 @deftypefn {Built-in Function} void __chkp_bndcl (const void *@var{ptr}, __bounds_type @var{b})
10868 Function code - @code{BUILT_IN_CHKP_BNDCL}. This built-in function is used
10869 by Pointer Bounds Checker to perform check for pointer @var{ptr} against
10870 lower bound of bounds @var{b}.
10873 @deftypefn {Built-in Function} void __chkp_bndcu (const void *@var{ptr}, __bounds_type @var{b})
10874 Function code - @code{BUILT_IN_CHKP_BNDCU}. This built-in function is used
10875 by Pointer Bounds Checker to perform check for pointer @var{ptr} against
10876 upper bound of bounds @var{b}.
10879 @deftypefn {Built-in Function} __bounds_type __chkp_bndret (void *@var{ptr})
10880 Function code - @code{BUILT_IN_CHKP_BNDRET}. This built-in function is used
10881 by Pointer Bounds Checker to obtain bounds returned by a call statement.
10882 @var{ptr} passed to built-in is @code{SSA_NAME} returned by the call.
10885 @deftypefn {Built-in Function} __bounds_type __chkp_intersect (__bounds_type @var{b1}, __bounds_type @var{b2})
10886 Function code - @code{BUILT_IN_CHKP_INTERSECT}. This built-in function
10887 returns intersection of bounds @var{b1} and @var{b2}.
10890 @deftypefn {Built-in Function} __bounds_type __chkp_narrow (const void *@var{ptr}, __bounds_type @var{b}, size_t @var{s})
10891 Function code - @code{BUILT_IN_CHKP_NARROW}. This built-in function
10892 returns intersection of bounds @var{b} and
10893 [@var{ptr}, @var{ptr} + @var{s} - @code{1}].
10896 @deftypefn {Built-in Function} size_t __chkp_sizeof (const void *@var{ptr})
10897 Function code - @code{BUILT_IN_CHKP_SIZEOF}. This built-in function
10898 returns size of object referenced by @var{ptr}. @var{ptr} is always
10899 @code{ADDR_EXPR} of @code{VAR_DECL}. This built-in is used by
10900 Pointer Bounds Checker when bounds of object cannot be computed statically
10901 (e.g. object has incomplete type).
10904 @deftypefn {Built-in Function} const void *__chkp_extract_lower (__bounds_type @var{b})
10905 Function code - @code{BUILT_IN_CHKP_EXTRACT_LOWER}. This built-in function
10906 returns lower bound of bounds @var{b}.
10909 @deftypefn {Built-in Function} const void *__chkp_extract_upper (__bounds_type @var{b})
10910 Function code - @code{BUILT_IN_CHKP_EXTRACT_UPPER}. This built-in function
10911 returns upper bound of bounds @var{b}.
10914 @deftypefn {Target Hook} tree TARGET_CHKP_BOUND_TYPE (void)
10915 Return type to be used for bounds
10917 @deftypefn {Target Hook} {enum machine_mode} TARGET_CHKP_BOUND_MODE (void)
10918 Return mode to be used for bounds.
10920 @deftypefn {Target Hook} tree TARGET_CHKP_MAKE_BOUNDS_CONSTANT (HOST_WIDE_INT @var{lb}, HOST_WIDE_INT @var{ub})
10921 Return constant used to statically initialize constant bounds
10922 with specified lower bound @var{lb} and upper bounds @var{ub}.
10924 @deftypefn {Target Hook} int TARGET_CHKP_INITIALIZE_BOUNDS (tree @var{var}, tree @var{lb}, tree @var{ub}, tree *@var{stmts})
10925 Generate a list of statements @var{stmts} to initialize pointer
10926 bounds variable @var{var} with bounds @var{lb} and @var{ub}. Return
10927 the number of generated statements.
10930 @deftypefn {Target Hook} tree TARGET_RESOLVE_OVERLOADED_BUILTIN (unsigned int @var{loc}, tree @var{fndecl}, void *@var{arglist})
10931 Select a replacement for a machine specific built-in function that
10932 was set up by @samp{TARGET_INIT_BUILTINS}. This is done
10933 @emph{before} regular type checking, and so allows the target to
10934 implement a crude form of function overloading. @var{fndecl} is the
10935 declaration of the built-in function. @var{arglist} is the list of
10936 arguments passed to the built-in function. The result is a
10937 complete expression that implements the operation, usually
10938 another @code{CALL_EXPR}.
10939 @var{arglist} really has type @samp{VEC(tree,gc)*}
10942 @deftypefn {Target Hook} tree TARGET_FOLD_BUILTIN (tree @var{fndecl}, int @var{n_args}, tree *@var{argp}, bool @var{ignore})
10943 Fold a call to a machine specific built-in function that was set up by
10944 @samp{TARGET_INIT_BUILTINS}. @var{fndecl} is the declaration of the
10945 built-in function. @var{n_args} is the number of arguments passed to
10946 the function; the arguments themselves are pointed to by @var{argp}.
10947 The result is another tree, valid for both GIMPLE and GENERIC,
10948 containing a simplified expression for the call's result. If
10949 @var{ignore} is true the value will be ignored.
10952 @deftypefn {Target Hook} bool TARGET_GIMPLE_FOLD_BUILTIN (gimple_stmt_iterator *@var{gsi})
10953 Fold a call to a machine specific built-in function that was set up
10954 by @samp{TARGET_INIT_BUILTINS}. @var{gsi} points to the gimple
10955 statement holding the function call. Returns true if any change
10956 was made to the GIMPLE stream.
10959 @deftypefn {Target Hook} int TARGET_COMPARE_VERSION_PRIORITY (tree @var{decl1}, tree @var{decl2})
10960 This hook is used to compare the target attributes in two functions to
10961 determine which function's features get higher priority. This is used
10962 during function multi-versioning to figure out the order in which two
10963 versions must be dispatched. A function version with a higher priority
10964 is checked for dispatching earlier. @var{decl1} and @var{decl2} are
10965 the two function decls that will be compared.
10968 @deftypefn {Target Hook} tree TARGET_GET_FUNCTION_VERSIONS_DISPATCHER (void *@var{decl})
10969 This hook is used to get the dispatcher function for a set of function
10970 versions. The dispatcher function is called to invoke the right function
10971 version at run-time. @var{decl} is one version from a set of semantically
10972 identical versions.
10975 @deftypefn {Target Hook} tree TARGET_GENERATE_VERSION_DISPATCHER_BODY (void *@var{arg})
10976 This hook is used to generate the dispatcher logic to invoke the right
10977 function version at run-time for a given set of function versions.
10978 @var{arg} points to the callgraph node of the dispatcher function whose
10979 body must be generated.
10982 @deftypefn {Target Hook} bool TARGET_CAN_USE_DOLOOP_P (const widest_int @var{&iterations}, const widest_int @var{&iterations_max}, unsigned int @var{loop_depth}, bool @var{entered_at_top})
10983 Return true if it is possible to use low-overhead loops (@code{doloop_end}
10984 and @code{doloop_begin}) for a particular loop. @var{iterations} gives the
10985 exact number of iterations, or 0 if not known. @var{iterations_max} gives
10986 the maximum number of iterations, or 0 if not known. @var{loop_depth} is
10987 the nesting depth of the loop, with 1 for innermost loops, 2 for loops that
10988 contain innermost loops, and so on. @var{entered_at_top} is true if the
10989 loop is only entered from the top.
10991 This hook is only used if @code{doloop_end} is available. The default
10992 implementation returns true. You can use @code{can_use_doloop_if_innermost}
10993 if the loop must be the innermost, and if there are no other restrictions.
10996 @deftypefn {Target Hook} {const char *} TARGET_INVALID_WITHIN_DOLOOP (const rtx_insn *@var{insn})
10998 Take an instruction in @var{insn} and return NULL if it is valid within a
10999 low-overhead loop, otherwise return a string explaining why doloop
11000 could not be applied.
11002 Many targets use special registers for low-overhead looping. For any
11003 instruction that clobbers these this function should return a string indicating
11004 the reason why the doloop could not be applied.
11005 By default, the RTL loop optimizer does not use a present doloop pattern for
11006 loops containing function calls or branch on table instructions.
11009 @deftypefn {Target Hook} bool TARGET_LEGITIMATE_COMBINED_INSN (rtx_insn *@var{insn})
11010 Take an instruction in @var{insn} and return @code{false} if the instruction is not appropriate as a combination of two or more instructions. The default is to accept all instructions.
11013 @defmac MD_CAN_REDIRECT_BRANCH (@var{branch1}, @var{branch2})
11015 Take a branch insn in @var{branch1} and another in @var{branch2}.
11016 Return true if redirecting @var{branch1} to the destination of
11017 @var{branch2} is possible.
11019 On some targets, branches may have a limited range. Optimizing the
11020 filling of delay slots can result in branches being redirected, and this
11021 may in turn cause a branch offset to overflow.
11024 @deftypefn {Target Hook} bool TARGET_CAN_FOLLOW_JUMP (const rtx_insn *@var{follower}, const rtx_insn *@var{followee})
11025 FOLLOWER and FOLLOWEE are JUMP_INSN instructions; return true if FOLLOWER may be modified to follow FOLLOWEE; false, if it can't. For example, on some targets, certain kinds of branches can't be made to follow through a hot/cold partitioning.
11028 @deftypefn {Target Hook} bool TARGET_COMMUTATIVE_P (const_rtx @var{x}, int @var{outer_code})
11029 This target hook returns @code{true} if @var{x} is considered to be commutative.
11030 Usually, this is just COMMUTATIVE_P (@var{x}), but the HP PA doesn't consider
11031 PLUS to be commutative inside a MEM@. @var{outer_code} is the rtx code
11032 of the enclosing rtl, if known, otherwise it is UNKNOWN.
11035 @deftypefn {Target Hook} rtx TARGET_ALLOCATE_INITIAL_VALUE (rtx @var{hard_reg})
11037 When the initial value of a hard register has been copied in a pseudo
11038 register, it is often not necessary to actually allocate another register
11039 to this pseudo register, because the original hard register or a stack slot
11040 it has been saved into can be used. @code{TARGET_ALLOCATE_INITIAL_VALUE}
11041 is called at the start of register allocation once for each hard register
11042 that had its initial value copied by using
11043 @code{get_func_hard_reg_initial_val} or @code{get_hard_reg_initial_val}.
11044 Possible values are @code{NULL_RTX}, if you don't want
11045 to do any special allocation, a @code{REG} rtx---that would typically be
11046 the hard register itself, if it is known not to be clobbered---or a
11048 If you are returning a @code{MEM}, this is only a hint for the allocator;
11049 it might decide to use another register anyways.
11050 You may use @code{current_function_is_leaf} or
11051 @code{REG_N_SETS} in the hook to determine if the hard
11052 register in question will not be clobbered.
11053 The default value of this hook is @code{NULL}, which disables any special
11057 @deftypefn {Target Hook} int TARGET_UNSPEC_MAY_TRAP_P (const_rtx @var{x}, unsigned @var{flags})
11058 This target hook returns nonzero if @var{x}, an @code{unspec} or
11059 @code{unspec_volatile} operation, might cause a trap. Targets can use
11060 this hook to enhance precision of analysis for @code{unspec} and
11061 @code{unspec_volatile} operations. You may call @code{may_trap_p_1}
11062 to analyze inner elements of @var{x} in which case @var{flags} should be
11066 @deftypefn {Target Hook} void TARGET_SET_CURRENT_FUNCTION (tree @var{decl})
11067 The compiler invokes this hook whenever it changes its current function
11068 context (@code{cfun}). You can define this function if
11069 the back end needs to perform any initialization or reset actions on a
11070 per-function basis. For example, it may be used to implement function
11071 attributes that affect register usage or code generation patterns.
11072 The argument @var{decl} is the declaration for the new function context,
11073 and may be null to indicate that the compiler has left a function context
11074 and is returning to processing at the top level.
11075 The default hook function does nothing.
11077 GCC sets @code{cfun} to a dummy function context during initialization of
11078 some parts of the back end. The hook function is not invoked in this
11079 situation; you need not worry about the hook being invoked recursively,
11080 or when the back end is in a partially-initialized state.
11081 @code{cfun} might be @code{NULL} to indicate processing at top level,
11082 outside of any function scope.
11085 @defmac TARGET_OBJECT_SUFFIX
11086 Define this macro to be a C string representing the suffix for object
11087 files on your target machine. If you do not define this macro, GCC will
11088 use @samp{.o} as the suffix for object files.
11091 @defmac TARGET_EXECUTABLE_SUFFIX
11092 Define this macro to be a C string representing the suffix to be
11093 automatically added to executable files on your target machine. If you
11094 do not define this macro, GCC will use the null string as the suffix for
11098 @defmac COLLECT_EXPORT_LIST
11099 If defined, @code{collect2} will scan the individual object files
11100 specified on its command line and create an export list for the linker.
11101 Define this macro for systems like AIX, where the linker discards
11102 object files that are not referenced from @code{main} and uses export
11106 @defmac MODIFY_JNI_METHOD_CALL (@var{mdecl})
11107 Define this macro to a C expression representing a variant of the
11108 method call @var{mdecl}, if Java Native Interface (JNI) methods
11109 must be invoked differently from other methods on your target.
11110 For example, on 32-bit Microsoft Windows, JNI methods must be invoked using
11111 the @code{stdcall} calling convention and this macro is then
11112 defined as this expression:
11115 build_type_attribute_variant (@var{mdecl},
11117 (get_identifier ("stdcall"),
11122 @deftypefn {Target Hook} bool TARGET_CANNOT_MODIFY_JUMPS_P (void)
11123 This target hook returns @code{true} past the point in which new jump
11124 instructions could be created. On machines that require a register for
11125 every jump such as the SHmedia ISA of SH5, this point would typically be
11126 reload, so this target hook should be defined to a function such as:
11130 cannot_modify_jumps_past_reload_p ()
11132 return (reload_completed || reload_in_progress);
11137 @deftypefn {Target Hook} reg_class_t TARGET_BRANCH_TARGET_REGISTER_CLASS (void)
11138 This target hook returns a register class for which branch target register
11139 optimizations should be applied. All registers in this class should be
11140 usable interchangeably. After reload, registers in this class will be
11141 re-allocated and loads will be hoisted out of loops and be subjected
11142 to inter-block scheduling.
11145 @deftypefn {Target Hook} bool TARGET_BRANCH_TARGET_REGISTER_CALLEE_SAVED (bool @var{after_prologue_epilogue_gen})
11146 Branch target register optimization will by default exclude callee-saved
11148 that are not already live during the current function; if this target hook
11149 returns true, they will be included. The target code must than make sure
11150 that all target registers in the class returned by
11151 @samp{TARGET_BRANCH_TARGET_REGISTER_CLASS} that might need saving are
11152 saved. @var{after_prologue_epilogue_gen} indicates if prologues and
11153 epilogues have already been generated. Note, even if you only return
11154 true when @var{after_prologue_epilogue_gen} is false, you still are likely
11155 to have to make special provisions in @code{INITIAL_ELIMINATION_OFFSET}
11156 to reserve space for caller-saved target registers.
11159 @deftypefn {Target Hook} bool TARGET_HAVE_CONDITIONAL_EXECUTION (void)
11160 This target hook returns true if the target supports conditional execution.
11161 This target hook is required only when the target has several different
11162 modes and they have different conditional execution capability, such as ARM.
11165 @deftypefn {Target Hook} unsigned TARGET_LOOP_UNROLL_ADJUST (unsigned @var{nunroll}, struct loop *@var{loop})
11166 This target hook returns a new value for the number of times @var{loop}
11167 should be unrolled. The parameter @var{nunroll} is the number of times
11168 the loop is to be unrolled. The parameter @var{loop} is a pointer to
11169 the loop, which is going to be checked for unrolling. This target hook
11170 is required only when the target has special constraints like maximum
11171 number of memory accesses.
11174 @defmac POWI_MAX_MULTS
11175 If defined, this macro is interpreted as a signed integer C expression
11176 that specifies the maximum number of floating point multiplications
11177 that should be emitted when expanding exponentiation by an integer
11178 constant inline. When this value is defined, exponentiation requiring
11179 more than this number of multiplications is implemented by calling the
11180 system library's @code{pow}, @code{powf} or @code{powl} routines.
11181 The default value places no upper bound on the multiplication count.
11184 @deftypefn Macro void TARGET_EXTRA_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
11185 This target hook should register any extra include files for the
11186 target. The parameter @var{stdinc} indicates if normal include files
11187 are present. The parameter @var{sysroot} is the system root directory.
11188 The parameter @var{iprefix} is the prefix for the gcc directory.
11191 @deftypefn Macro void TARGET_EXTRA_PRE_INCLUDES (const char *@var{sysroot}, const char *@var{iprefix}, int @var{stdinc})
11192 This target hook should register any extra include files for the
11193 target before any standard headers. The parameter @var{stdinc}
11194 indicates if normal include files are present. The parameter
11195 @var{sysroot} is the system root directory. The parameter
11196 @var{iprefix} is the prefix for the gcc directory.
11199 @deftypefn Macro void TARGET_OPTF (char *@var{path})
11200 This target hook should register special include paths for the target.
11201 The parameter @var{path} is the include to register. On Darwin
11202 systems, this is used for Framework includes, which have semantics
11203 that are different from @option{-I}.
11206 @defmac bool TARGET_USE_LOCAL_THUNK_ALIAS_P (tree @var{fndecl})
11207 This target macro returns @code{true} if it is safe to use a local alias
11208 for a virtual function @var{fndecl} when constructing thunks,
11209 @code{false} otherwise. By default, the macro returns @code{true} for all
11210 functions, if a target supports aliases (i.e.@: defines
11211 @code{ASM_OUTPUT_DEF}), @code{false} otherwise,
11214 @defmac TARGET_FORMAT_TYPES
11215 If defined, this macro is the name of a global variable containing
11216 target-specific format checking information for the @option{-Wformat}
11217 option. The default is to have no target-specific format checks.
11220 @defmac TARGET_N_FORMAT_TYPES
11221 If defined, this macro is the number of entries in
11222 @code{TARGET_FORMAT_TYPES}.
11225 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES
11226 If defined, this macro is the name of a global variable containing
11227 target-specific format overrides for the @option{-Wformat} option. The
11228 default is to have no target-specific format overrides. If defined,
11229 @code{TARGET_FORMAT_TYPES} must be defined, too.
11232 @defmac TARGET_OVERRIDES_FORMAT_ATTRIBUTES_COUNT
11233 If defined, this macro specifies the number of entries in
11234 @code{TARGET_OVERRIDES_FORMAT_ATTRIBUTES}.
11237 @defmac TARGET_OVERRIDES_FORMAT_INIT
11238 If defined, this macro specifies the optional initialization
11239 routine for target specific customizations of the system printf
11240 and scanf formatter settings.
11243 @deftypevr {Target Hook} bool TARGET_RELAXED_ORDERING
11244 If set to @code{true}, means that the target's memory model does not
11245 guarantee that loads which do not depend on one another will access
11246 main memory in the order of the instruction stream; if ordering is
11247 important, an explicit memory barrier must be used. This is true of
11248 many recent processors which implement a policy of ``relaxed,''
11249 ``weak,'' or ``release'' memory consistency, such as Alpha, PowerPC,
11250 and ia64. The default is @code{false}.
11253 @deftypefn {Target Hook} {const char *} TARGET_INVALID_ARG_FOR_UNPROTOTYPED_FN (const_tree @var{typelist}, const_tree @var{funcdecl}, const_tree @var{val})
11254 If defined, this macro returns the diagnostic message when it is
11255 illegal to pass argument @var{val} to function @var{funcdecl}
11256 with prototype @var{typelist}.
11259 @deftypefn {Target Hook} {const char *} TARGET_INVALID_CONVERSION (const_tree @var{fromtype}, const_tree @var{totype})
11260 If defined, this macro returns the diagnostic message when it is
11261 invalid to convert from @var{fromtype} to @var{totype}, or @code{NULL}
11262 if validity should be determined by the front end.
11265 @deftypefn {Target Hook} {const char *} TARGET_INVALID_UNARY_OP (int @var{op}, const_tree @var{type})
11266 If defined, this macro returns the diagnostic message when it is
11267 invalid to apply operation @var{op} (where unary plus is denoted by
11268 @code{CONVERT_EXPR}) to an operand of type @var{type}, or @code{NULL}
11269 if validity should be determined by the front end.
11272 @deftypefn {Target Hook} {const char *} TARGET_INVALID_BINARY_OP (int @var{op}, const_tree @var{type1}, const_tree @var{type2})
11273 If defined, this macro returns the diagnostic message when it is
11274 invalid to apply operation @var{op} to operands of types @var{type1}
11275 and @var{type2}, or @code{NULL} if validity should be determined by
11279 @deftypefn {Target Hook} {const char *} TARGET_INVALID_PARAMETER_TYPE (const_tree @var{type})
11280 If defined, this macro returns the diagnostic message when it is
11281 invalid for functions to include parameters of type @var{type},
11282 or @code{NULL} if validity should be determined by
11283 the front end. This is currently used only by the C and C++ front ends.
11286 @deftypefn {Target Hook} {const char *} TARGET_INVALID_RETURN_TYPE (const_tree @var{type})
11287 If defined, this macro returns the diagnostic message when it is
11288 invalid for functions to have return type @var{type},
11289 or @code{NULL} if validity should be determined by
11290 the front end. This is currently used only by the C and C++ front ends.
11293 @deftypefn {Target Hook} tree TARGET_PROMOTED_TYPE (const_tree @var{type})
11294 If defined, this target hook returns the type to which values of
11295 @var{type} should be promoted when they appear in expressions,
11296 analogous to the integer promotions, or @code{NULL_TREE} to use the
11297 front end's normal promotion rules. This hook is useful when there are
11298 target-specific types with special promotion rules.
11299 This is currently used only by the C and C++ front ends.
11302 @deftypefn {Target Hook} tree TARGET_CONVERT_TO_TYPE (tree @var{type}, tree @var{expr})
11303 If defined, this hook returns the result of converting @var{expr} to
11304 @var{type}. It should return the converted expression,
11305 or @code{NULL_TREE} to apply the front end's normal conversion rules.
11306 This hook is useful when there are target-specific types with special
11308 This is currently used only by the C and C++ front ends.
11311 @defmac TARGET_USE_JCR_SECTION
11312 This macro determines whether to use the JCR section to register Java
11313 classes. By default, TARGET_USE_JCR_SECTION is defined to 1 if both
11314 SUPPORTS_WEAK and TARGET_HAVE_NAMED_SECTIONS are true, else 0.
11318 This macro determines the size of the objective C jump buffer for the
11319 NeXT runtime. By default, OBJC_JBLEN is defined to an innocuous value.
11322 @defmac LIBGCC2_UNWIND_ATTRIBUTE
11323 Define this macro if any target-specific attributes need to be attached
11324 to the functions in @file{libgcc} that provide low-level support for
11325 call stack unwinding. It is used in declarations in @file{unwind-generic.h}
11326 and the associated definitions of those functions.
11329 @deftypefn {Target Hook} void TARGET_UPDATE_STACK_BOUNDARY (void)
11330 Define this macro to update the current function stack boundary if
11334 @deftypefn {Target Hook} rtx TARGET_GET_DRAP_RTX (void)
11335 This hook should return an rtx for Dynamic Realign Argument Pointer (DRAP) if a
11336 different argument pointer register is needed to access the function's
11337 argument list due to stack realignment. Return @code{NULL} if no DRAP
11341 @deftypefn {Target Hook} bool TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS (void)
11342 When optimization is disabled, this hook indicates whether or not
11343 arguments should be allocated to stack slots. Normally, GCC allocates
11344 stacks slots for arguments when not optimizing in order to make
11345 debugging easier. However, when a function is declared with
11346 @code{__attribute__((naked))}, there is no stack frame, and the compiler
11347 cannot safely move arguments from the registers in which they are passed
11348 to the stack. Therefore, this hook should return true in general, but
11349 false for naked functions. The default implementation always returns true.
11352 @deftypevr {Target Hook} {unsigned HOST_WIDE_INT} TARGET_CONST_ANCHOR
11353 On some architectures it can take multiple instructions to synthesize
11354 a constant. If there is another constant already in a register that
11355 is close enough in value then it is preferable that the new constant
11356 is computed from this register using immediate addition or
11357 subtraction. We accomplish this through CSE. Besides the value of
11358 the constant we also add a lower and an upper constant anchor to the
11359 available expressions. These are then queried when encountering new
11360 constants. The anchors are computed by rounding the constant up and
11361 down to a multiple of the value of @code{TARGET_CONST_ANCHOR}.
11362 @code{TARGET_CONST_ANCHOR} should be the maximum positive value
11363 accepted by immediate-add plus one. We currently assume that the
11364 value of @code{TARGET_CONST_ANCHOR} is a power of 2. For example, on
11365 MIPS, where add-immediate takes a 16-bit signed value,
11366 @code{TARGET_CONST_ANCHOR} is set to @samp{0x8000}. The default value
11367 is zero, which disables this optimization.
11370 @deftypefn {Target Hook} {unsigned HOST_WIDE_INT} TARGET_ASAN_SHADOW_OFFSET (void)
11371 Return the offset bitwise ored into shifted address to get corresponding
11372 Address Sanitizer shadow memory address. NULL if Address Sanitizer is not
11373 supported by the target.
11376 @deftypefn {Target Hook} {unsigned HOST_WIDE_INT} TARGET_MEMMODEL_CHECK (unsigned HOST_WIDE_INT @var{val})
11377 Validate target specific memory model mask bits. When NULL no target specific
11378 memory model bits are allowed.
11381 @deftypevr {Target Hook} {unsigned char} TARGET_ATOMIC_TEST_AND_SET_TRUEVAL
11382 This value should be set if the result written by @code{atomic_test_and_set} is not exactly 1, i.e. the @code{bool} @code{true}.
11385 @deftypefn {Target Hook} bool TARGET_HAS_IFUNC_P (void)
11386 It returns true if the target supports GNU indirect functions.
11387 The support includes the assembler, linker and dynamic linker.
11388 The default value of this hook is based on target's libc.
11391 @deftypefn {Target Hook} {unsigned int} TARGET_ATOMIC_ALIGN_FOR_MODE (machine_mode @var{mode})
11392 If defined, this function returns an appropriate alignment in bits for an atomic object of machine_mode @var{mode}. If 0 is returned then the default alignment for the specified mode is used.
11395 @deftypefn {Target Hook} void TARGET_ATOMIC_ASSIGN_EXPAND_FENV (tree *@var{hold}, tree *@var{clear}, tree *@var{update})
11396 ISO C11 requires atomic compound assignments that may raise floating-point exceptions to raise exceptions corresponding to the arithmetic operation whose result was successfully stored in a compare-and-exchange sequence. This requires code equivalent to calls to @code{feholdexcept}, @code{feclearexcept} and @code{feupdateenv} to be generated at appropriate points in the compare-and-exchange sequence. This hook should set @code{*@var{hold}} to an expression equivalent to the call to @code{feholdexcept}, @code{*@var{clear}} to an expression equivalent to the call to @code{feclearexcept} and @code{*@var{update}} to an expression equivalent to the call to @code{feupdateenv}. The three expressions are @code{NULL_TREE} on entry to the hook and may be left as @code{NULL_TREE} if no code is required in a particular place. The default implementation leaves all three expressions as @code{NULL_TREE}. The @code{__atomic_feraiseexcept} function from @code{libatomic} may be of use as part of the code generated in @code{*@var{update}}.
11399 @defmac TARGET_SUPPORTS_WIDE_INT
11401 On older ports, large integers are stored in @code{CONST_DOUBLE} rtl
11402 objects. Newer ports define @code{TARGET_SUPPORTS_WIDE_INT} to be nonzero
11403 to indicate that large integers are stored in
11404 @code{CONST_WIDE_INT} rtl objects. The @code{CONST_WIDE_INT} allows
11405 very large integer constants to be represented. @code{CONST_DOUBLE}
11406 is limited to twice the size of the host's @code{HOST_WIDE_INT}
11409 Converting a port mostly requires looking for the places where
11410 @code{CONST_DOUBLE}s are used with @code{VOIDmode} and replacing that
11411 code with code that accesses @code{CONST_WIDE_INT}s. @samp{"grep -i
11412 const_double"} at the port level gets you to 95% of the changes that
11413 need to be made. There are a few places that require a deeper look.
11417 There is no equivalent to @code{hval} and @code{lval} for
11418 @code{CONST_WIDE_INT}s. This would be difficult to express in the md
11419 language since there are a variable number of elements.
11421 Most ports only check that @code{hval} is either 0 or -1 to see if the
11422 value is small. As mentioned above, this will no longer be necessary
11423 since small constants are always @code{CONST_INT}. Of course there
11424 are still a few exceptions, the alpha's constraint used by the zap
11425 instruction certainly requires careful examination by C code.
11426 However, all the current code does is pass the hval and lval to C
11427 code, so evolving the c code to look at the @code{CONST_WIDE_INT} is
11428 not really a large change.
11431 Because there is no standard template that ports use to materialize
11432 constants, there is likely to be some futzing that is unique to each
11436 The rtx costs may have to be adjusted to properly account for larger
11437 constants that are represented as @code{CONST_WIDE_INT}.
11440 All and all it does not take long to convert ports that the
11441 maintainer is familiar with.
11445 @deftypefn {Target Hook} void TARGET_RECORD_OFFLOAD_SYMBOL (tree)
11446 Used when offloaded functions are seen in the compilation unit and no named
11447 sections are available. It is called once for each symbol that must be
11448 recorded in the offload function and variable table.