1 This is Info file gcc.info, produced by Makeinfo version 1.68 from the
4 This file documents the use and the internals of the GNU compiler.
6 Published by the Free Software Foundation 59 Temple Place - Suite 330
7 Boston, MA 02111-1307 USA
9 Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997 Free
10 Software Foundation, Inc.
12 Permission is granted to make and distribute verbatim copies of this
13 manual provided the copyright notice and this permission notice are
14 preserved on all copies.
16 Permission is granted to copy and distribute modified versions of
17 this manual under the conditions for verbatim copying, provided also
18 that the sections entitled "GNU General Public License," "Funding for
19 Free Software," and "Protect Your Freedom--Fight `Look And Feel'" are
20 included exactly as in the original, and provided that the entire
21 resulting derived work is distributed under the terms of a permission
22 notice identical to this one.
24 Permission is granted to copy and distribute translations of this
25 manual into another language, under the above conditions for modified
26 versions, except that the sections entitled "GNU General Public
27 License," "Funding for Free Software," and "Protect Your Freedom--Fight
28 `Look And Feel'", and this permission notice, may be included in
29 translations approved by the Free Software Foundation instead of in the
33 File: gcc.info, Node: Initialization, Next: Macros for Initialization, Prev: Label Output, Up: Assembler Format
35 How Initialization Functions Are Handled
36 ----------------------------------------
38 The compiled code for certain languages includes "constructors"
39 (also called "initialization routines")--functions to initialize data
40 in the program when the program is started. These functions need to be
41 called before the program is "started"--that is to say, before `main'
44 Compiling some languages generates "destructors" (also called
45 "termination routines") that should be called when the program
48 To make the initialization and termination functions work, the
49 compiler must output something in the assembler code to cause those
50 functions to be called at the appropriate time. When you port the
51 compiler to a new system, you need to specify how to do this.
53 There are two major ways that GCC currently supports the execution of
54 initialization and termination functions. Each way has two variants.
55 Much of the structure is common to all four variations.
57 The linker must build two lists of these functions--a list of
58 initialization functions, called `__CTOR_LIST__', and a list of
59 termination functions, called `__DTOR_LIST__'.
61 Each list always begins with an ignored function pointer (which may
62 hold 0, -1, or a count of the function pointers after it, depending on
63 the environment). This is followed by a series of zero or more function
64 pointers to constructors (or destructors), followed by a function
65 pointer containing zero.
67 Depending on the operating system and its executable file format,
68 either `crtstuff.c' or `libgcc2.c' traverses these lists at startup
69 time and exit time. Constructors are called in reverse order of the
70 list; destructors in forward order.
72 The best way to handle static constructors works only for object file
73 formats which provide arbitrarily-named sections. A section is set
74 aside for a list of constructors, and another for a list of destructors.
75 Traditionally these are called `.ctors' and `.dtors'. Each object file
76 that defines an initialization function also puts a word in the
77 constructor section to point to that function. The linker accumulates
78 all these words into one contiguous `.ctors' section. Termination
79 functions are handled similarly.
81 To use this method, you need appropriate definitions of the macros
82 `ASM_OUTPUT_CONSTRUCTOR' and `ASM_OUTPUT_DESTRUCTOR'. Usually you can
83 get them by including `svr4.h'.
85 When arbitrary sections are available, there are two variants,
86 depending upon how the code in `crtstuff.c' is called. On systems that
87 support an "init" section which is executed at program startup, parts
88 of `crtstuff.c' are compiled into that section. The program is linked
89 by the `gcc' driver like this:
91 ld -o OUTPUT_FILE crtbegin.o ... crtend.o -lgcc
93 The head of a function (`__do_global_ctors') appears in the init
94 section of `crtbegin.o'; the remainder of the function appears in the
95 init section of `crtend.o'. The linker will pull these two parts of
96 the section together, making a whole function. If any of the user's
97 object files linked into the middle of it contribute code, then that
98 code will be executed as part of the body of `__do_global_ctors'.
100 To use this variant, you must define the `INIT_SECTION_ASM_OP' macro
103 If no init section is available, do not define
104 `INIT_SECTION_ASM_OP'. Then `__do_global_ctors' is built into the text
105 section like all other functions, and resides in `libgcc.a'. When GCC
106 compiles any function called `main', it inserts a procedure call to
107 `__main' as the first executable code after the function prologue. The
108 `__main' function, also defined in `libgcc2.c', simply calls
111 In file formats that don't support arbitrary sections, there are
112 again two variants. In the simplest variant, the GNU linker (GNU `ld')
113 and an `a.out' format must be used. In this case,
114 `ASM_OUTPUT_CONSTRUCTOR' is defined to produce a `.stabs' entry of type
115 `N_SETT', referencing the name `__CTOR_LIST__', and with the address of
116 the void function containing the initialization code as its value. The
117 GNU linker recognizes this as a request to add the value to a "set";
118 the values are accumulated, and are eventually placed in the executable
119 as a vector in the format described above, with a leading (ignored)
120 count and a trailing zero element. `ASM_OUTPUT_DESTRUCTOR' is handled
121 similarly. Since no init section is available, the absence of
122 `INIT_SECTION_ASM_OP' causes the compilation of `main' to call `__main'
123 as above, starting the initialization process.
125 The last variant uses neither arbitrary sections nor the GNU linker.
126 This is preferable when you want to do dynamic linking and when using
127 file formats which the GNU linker does not support, such as `ECOFF'. In
128 this case, `ASM_OUTPUT_CONSTRUCTOR' does not produce an `N_SETT'
129 symbol; initialization and termination functions are recognized simply
130 by their names. This requires an extra program in the linkage step,
131 called `collect2'. This program pretends to be the linker, for use
132 with GNU CC; it does its job by running the ordinary linker, but also
133 arranges to include the vectors of initialization and termination
134 functions. These functions are called via `__main' as described above.
136 Choosing among these configuration options has been simplified by a
137 set of operating-system-dependent files in the `config' subdirectory.
138 These files define all of the relevant parameters. Usually it is
139 sufficient to include one into your specific machine-dependent
140 configuration file. These files are:
143 For operating systems using the `a.out' format.
146 For operating systems using the `MachO' format.
149 For System V Release 3 and similar systems using `COFF' format.
152 For System V Release 4 and similar systems using `ELF' format.
155 For the VMS operating system.
157 The following section describes the specific macros that control and
158 customize the handling of initialization and termination functions.
161 File: gcc.info, Node: Macros for Initialization, Next: Instruction Output, Prev: Initialization, Up: Assembler Format
163 Macros Controlling Initialization Routines
164 ------------------------------------------
166 Here are the macros that control how the compiler handles
167 initialization and termination functions:
169 `INIT_SECTION_ASM_OP'
170 If defined, a C string constant for the assembler operation to
171 identify the following data as initialization code. If not
172 defined, GNU CC will assume such a section does not exist. When
173 you are using special sections for initialization and termination
174 functions, this macro also controls how `crtstuff.c' and
175 `libgcc2.c' arrange to run the initialization functions.
178 If defined, `main' will not call `__main' as described above.
179 This macro should be defined for systems that control the contents
180 of the init section on a symbol-by-symbol basis, such as OSF/1,
181 and should not be defined explicitly for systems that support
182 `INIT_SECTION_ASM_OP'.
185 If defined, a C string constant for a switch that tells the linker
186 that the following symbol is an initialization routine.
189 If defined, a C string constant for a switch that tells the linker
190 that the following symbol is a finalization routine.
193 If defined, `main' will call `__main' despite the presence of
194 `INIT_SECTION_ASM_OP'. This macro should be defined for systems
195 where the init section is not actually run automatically, but is
196 still useful for collecting the lists of constructors and
199 `ASM_OUTPUT_CONSTRUCTOR (STREAM, NAME)'
200 Define this macro as a C statement to output on the stream STREAM
201 the assembler code to arrange to call the function named NAME at
204 Assume that NAME is the name of a C function generated
205 automatically by the compiler. This function takes no arguments.
206 Use the function `assemble_name' to output the name NAME; this
207 performs any system-specific syntactic transformations such as
208 adding an underscore.
210 If you don't define this macro, nothing special is output to
211 arrange to call the function. This is correct when the function
212 will be called in some other manner--for example, by means of the
213 `collect2' program, which looks through the symbol table to find
214 these functions by their names.
216 `ASM_OUTPUT_DESTRUCTOR (STREAM, NAME)'
217 This is like `ASM_OUTPUT_CONSTRUCTOR' but used for termination
218 functions rather than initialization functions.
220 If your system uses `collect2' as the means of processing
221 constructors, then that program normally uses `nm' to scan an object
222 file for constructor functions to be called. On certain kinds of
223 systems, you can define these macros to make `collect2' work faster
224 (and, in some cases, make it work at all):
227 Define this macro if the system uses COFF (Common Object File
228 Format) object files, so that `collect2' can assume this format
229 and scan object files directly for dynamic constructor/destructor
233 Define this macro if the system uses ROSE format object files, so
234 that `collect2' can assume this format and scan object files
235 directly for dynamic constructor/destructor functions.
237 These macros are effective only in a native compiler; `collect2' as
238 part of a cross compiler always uses `nm' for the target machine.
241 Define this macro as a C string constant containing the file name
242 to use to execute `nm'. The default is to search the path
245 If your system supports shared libraries and has a program to list
246 the dynamic dependencies of a given library or executable, you can
247 define these macros to enable support for running initialization
248 and termination functions in shared libraries:
251 Define this macro to a C string constant containing the name of the
252 program which lists dynamic dependencies, like `"ldd"' under SunOS
255 `PARSE_LDD_OUTPUT (PTR)'
256 Define this macro to be C code that extracts filenames from the
257 output of the program denoted by `LDD_SUFFIX'. PTR is a variable
258 of type `char *' that points to the beginning of a line of output
259 from `LDD_SUFFIX'. If the line lists a dynamic dependency, the
260 code must advance PTR to the beginning of the filename on that
261 line. Otherwise, it must set PTR to `NULL'.
264 File: gcc.info, Node: Instruction Output, Next: Dispatch Tables, Prev: Macros for Initialization, Up: Assembler Format
266 Output of Assembler Instructions
267 --------------------------------
269 This describes assembler instruction output.
272 A C initializer containing the assembler's names for the machine
273 registers, each one as a C string constant. This is what
274 translates register numbers in the compiler into assembler
277 `ADDITIONAL_REGISTER_NAMES'
278 If defined, a C initializer for an array of structures containing
279 a name and a register number. This macro defines additional names
280 for hard registers, thus allowing the `asm' option in declarations
281 to refer to registers using alternate names.
283 `ASM_OUTPUT_OPCODE (STREAM, PTR)'
284 Define this macro if you are using an unusual assembler that
285 requires different names for the machine instructions.
287 The definition is a C statement or statements which output an
288 assembler instruction opcode to the stdio stream STREAM. The
289 macro-operand PTR is a variable of type `char *' which points to
290 the opcode name in its "internal" form--the form that is written
291 in the machine description. The definition should output the
292 opcode name to STREAM, performing any translation you desire, and
293 increment the variable PTR to point at the end of the opcode so
294 that it will not be output twice.
296 In fact, your macro definition may process less than the entire
297 opcode name, or more than the opcode name; but if you want to
298 process text that includes `%'-sequences to substitute operands,
299 you must take care of the substitution yourself. Just be sure to
300 increment PTR over whatever text should not be output normally.
302 If you need to look at the operand values, they can be found as the
303 elements of `recog_operand'.
305 If the macro definition does nothing, the instruction is output in
308 `FINAL_PRESCAN_INSN (INSN, OPVEC, NOPERANDS)'
309 If defined, a C statement to be executed just prior to the output
310 of assembler code for INSN, to modify the extracted operands so
311 they will be output differently.
313 Here the argument OPVEC is the vector containing the operands
314 extracted from INSN, and NOPERANDS is the number of elements of
315 the vector which contain meaningful data for this insn. The
316 contents of this vector are what will be used to convert the insn
317 template into assembler code, so you can change the assembler
318 output by changing the contents of the vector.
320 This macro is useful when various assembler syntaxes share a single
321 file of instruction patterns; by defining this macro differently,
322 you can cause a large class of instructions to be output
323 differently (such as with rearranged operands). Naturally,
324 variations in assembler syntax affecting individual insn patterns
325 ought to be handled by writing conditional output routines in
328 If this macro is not defined, it is equivalent to a null statement.
330 `FINAL_PRESCAN_LABEL'
331 If defined, `FINAL_PRESCAN_INSN' will be called on each
332 `CODE_LABEL'. In that case, OPVEC will be a null pointer and
333 NOPERANDS will be zero.
335 `PRINT_OPERAND (STREAM, X, CODE)'
336 A C compound statement to output to stdio stream STREAM the
337 assembler syntax for an instruction operand X. X is an RTL
340 CODE is a value that can be used to specify one of several ways of
341 printing the operand. It is used when identical operands must be
342 printed differently depending on the context. CODE comes from the
343 `%' specification that was used to request printing of the
344 operand. If the specification was just `%DIGIT' then CODE is 0;
345 if the specification was `%LTR DIGIT' then CODE is the ASCII code
348 If X is a register, this macro should print the register's name.
349 The names can be found in an array `reg_names' whose type is `char
350 *[]'. `reg_names' is initialized from `REGISTER_NAMES'.
352 When the machine description has a specification `%PUNCT' (a `%'
353 followed by a punctuation character), this macro is called with a
354 null pointer for X and the punctuation character for CODE.
356 `PRINT_OPERAND_PUNCT_VALID_P (CODE)'
357 A C expression which evaluates to true if CODE is a valid
358 punctuation character for use in the `PRINT_OPERAND' macro. If
359 `PRINT_OPERAND_PUNCT_VALID_P' is not defined, it means that no
360 punctuation characters (except for the standard one, `%') are used
363 `PRINT_OPERAND_ADDRESS (STREAM, X)'
364 A C compound statement to output to stdio stream STREAM the
365 assembler syntax for an instruction operand that is a memory
366 reference whose address is X. X is an RTL expression.
368 On some machines, the syntax for a symbolic address depends on the
369 section that the address refers to. On these machines, define the
370 macro `ENCODE_SECTION_INFO' to store the information into the
371 `symbol_ref', and then check for it here. *Note Assembler
374 `DBR_OUTPUT_SEQEND(FILE)'
375 A C statement, to be executed after all slot-filler instructions
376 have been output. If necessary, call `dbr_sequence_length' to
377 determine the number of slots filled in a sequence (zero if not
378 currently outputting a sequence), to decide how many no-ops to
381 Don't define this macro if it has nothing to do, but it is helpful
382 in reading assembly output if the extent of the delay sequence is
383 made explicit (e.g. with white space).
385 Note that output routines for instructions with delay slots must be
386 prepared to deal with not being output as part of a sequence (i.e.
387 when the scheduling pass is not run, or when no slot fillers could
388 be found.) The variable `final_sequence' is null when not
389 processing a sequence, otherwise it contains the `sequence' rtx
396 If defined, C string expressions to be used for the `%R', `%L',
397 `%U', and `%I' options of `asm_fprintf' (see `final.c'). These
398 are useful when a single `md' file must support multiple assembler
399 formats. In that case, the various `tm.h' files can define these
403 If your target supports multiple dialects of assembler language
404 (such as different opcodes), define this macro as a C expression
405 that gives the numeric index of the assembler language dialect to
406 use, with zero as the first variant.
408 If this macro is defined, you may use constructs of the form
409 `{option0|option1|option2...}' in the output templates of patterns
410 (*note Output Template::.) or in the first argument of
411 `asm_fprintf'. This construct outputs `option0', `option1' or
412 `option2', etc., if the value of `ASSEMBLER_DIALECT' is zero, one
413 or two, etc. Any special characters within these strings retain
416 If you do not define this macro, the characters `{', `|' and `}'
417 do not have any special meaning when used in templates or operands
420 Define the macros `REGISTER_PREFIX', `LOCAL_LABEL_PREFIX',
421 `USER_LABEL_PREFIX' and `IMMEDIATE_PREFIX' if you can express the
422 variations in assemble language syntax with that mechanism. Define
423 `ASSEMBLER_DIALECT' and use the `{option0|option1}' syntax if the
424 syntax variant are larger and involve such things as different
425 opcodes or operand order.
427 `ASM_OUTPUT_REG_PUSH (STREAM, REGNO)'
428 A C expression to output to STREAM some assembler code which will
429 push hard register number REGNO onto the stack. The code need not
430 be optimal, since this macro is used only when profiling.
432 `ASM_OUTPUT_REG_POP (STREAM, REGNO)'
433 A C expression to output to STREAM some assembler code which will
434 pop hard register number REGNO off of the stack. The code need
435 not be optimal, since this macro is used only when profiling.
438 File: gcc.info, Node: Dispatch Tables, Next: Exception Region Output, Prev: Instruction Output, Up: Assembler Format
440 Output of Dispatch Tables
441 -------------------------
443 This concerns dispatch tables.
445 `ASM_OUTPUT_ADDR_DIFF_ELT (STREAM, VALUE, REL)'
446 A C statement to output to the stdio stream STREAM an assembler
447 pseudo-instruction to generate a difference between two labels.
448 VALUE and REL are the numbers of two internal labels. The
449 definitions of these labels are output using
450 `ASM_OUTPUT_INTERNAL_LABEL', and they must be printed in the same
451 way here. For example,
453 fprintf (STREAM, "\t.word L%d-L%d\n",
456 You must provide this macro on machines where the addresses in a
457 dispatch table are relative to the table's own address. If
458 defined, GNU CC will also use this macro on all machines when
461 `ASM_OUTPUT_ADDR_VEC_ELT (STREAM, VALUE)'
462 This macro should be provided on machines where the addresses in a
463 dispatch table are absolute.
465 The definition should be a C statement to output to the stdio
466 stream STREAM an assembler pseudo-instruction to generate a
467 reference to a label. VALUE is the number of an internal label
468 whose definition is output using `ASM_OUTPUT_INTERNAL_LABEL'. For
471 fprintf (STREAM, "\t.word L%d\n", VALUE)
473 `ASM_OUTPUT_CASE_LABEL (STREAM, PREFIX, NUM, TABLE)'
474 Define this if the label before a jump-table needs to be output
475 specially. The first three arguments are the same as for
476 `ASM_OUTPUT_INTERNAL_LABEL'; the fourth argument is the jump-table
477 which follows (a `jump_insn' containing an `addr_vec' or
480 This feature is used on system V to output a `swbeg' statement for
483 If this macro is not defined, these labels are output with
484 `ASM_OUTPUT_INTERNAL_LABEL'.
486 `ASM_OUTPUT_CASE_END (STREAM, NUM, TABLE)'
487 Define this if something special must be output at the end of a
488 jump-table. The definition should be a C statement to be executed
489 after the assembler code for the table is written. It should write
490 the appropriate code to stdio stream STREAM. The argument TABLE
491 is the jump-table insn, and NUM is the label-number of the
494 If this macro is not defined, nothing special is output at the end
498 File: gcc.info, Node: Exception Region Output, Next: Alignment Output, Prev: Dispatch Tables, Up: Assembler Format
500 Assembler Commands for Exception Regions
501 ----------------------------------------
503 This describes commands marking the start and the end of an exception
506 `ASM_OUTPUT_EH_REGION_BEG ()'
507 A C expression to output text to mark the start of an exception
510 This macro need not be defined on most platforms.
512 `ASM_OUTPUT_EH_REGION_END ()'
513 A C expression to output text to mark the end of an exception
516 This macro need not be defined on most platforms.
518 `EXCEPTION_SECTION ()'
519 A C expression to switch to the section in which the main
520 exception table is to be placed (*note Sections::.). The default
521 is a section named `.gcc_except_table' on machines that support
522 named sections via `ASM_OUTPUT_SECTION_NAME', otherwise if `-fpic'
523 or `-fPIC' is in effect, the `data_section', otherwise the
524 `readonly_data_section'.
526 `EH_FRAME_SECTION_ASM_OP'
527 If defined, a C string constant for the assembler operation to
528 switch to the section for exception handling frame unwind
529 information. If not defined, GNU CC will provide a default
530 definition if the target supports named sections. `crtstuff.c'
531 uses this macro to switch to the appropriate section.
533 You should define this symbol if your target supports DWARF 2 frame
534 unwind information and the default definition does not work.
537 A C expression that is nonzero if the normal exception table output
540 This macro need not be defined on most platforms.
543 Alternate runtime support for looking up an exception at runtime
544 and finding the associated handler, if the default method won't
547 This macro need not be defined on most platforms.
549 `DOESNT_NEED_UNWINDER'
550 A C expression that decides whether or not the current function
551 needs to have a function unwinder generated for it. See the file
552 `except.c' for details on when to define this, and how.
555 An rtx used to mask the return address found via RETURN_ADDR_RTX,
556 so that it does not contain any extraneous set bits in it.
559 Define this macro to 0 if your target supports DWARF 2 frame unwind
560 information, but it does not yet work with exception handling.
561 Otherwise, if your target supports this information (if it defines
562 `INCOMING_RETURN_ADDR_RTX' and either `UNALIGNED_INT_ASM_OP' or
563 `OBJECT_FORMAT_ELF'), GCC will provide a default definition of 1.
565 If this macro is defined to 1, the DWARF 2 unwinder will be the
566 default exception handling mechanism; otherwise, setjmp/longjmp
567 will be used by default.
569 If this macro is defined to anything, the DWARF 2 unwinder will be
570 used instead of inline unwinders and __unwind_function in the
574 File: gcc.info, Node: Alignment Output, Prev: Exception Region Output, Up: Assembler Format
576 Assembler Commands for Alignment
577 --------------------------------
579 This describes commands for alignment.
581 `ASM_OUTPUT_ALIGN_CODE (FILE)'
582 A C expression to output text to align the location counter in the
583 way that is desirable at a point in the code that is reached only
586 This macro need not be defined if you don't want any special
587 alignment to be done at such a time. Most machine descriptions do
588 not currently define the macro.
590 `ASM_OUTPUT_LOOP_ALIGN (FILE)'
591 A C expression to output text to align the location counter in the
592 way that is desirable at the beginning of a loop.
594 This macro need not be defined if you don't want any special
595 alignment to be done at such a time. Most machine descriptions do
596 not currently define the macro.
598 `ASM_OUTPUT_SKIP (STREAM, NBYTES)'
599 A C statement to output to the stdio stream STREAM an assembler
600 instruction to advance the location counter by NBYTES bytes.
601 Those bytes should be zero when loaded. NBYTES will be a C
602 expression of type `int'.
604 `ASM_NO_SKIP_IN_TEXT'
605 Define this macro if `ASM_OUTPUT_SKIP' should not be used in the
606 text section because it fails put zeros in the bytes that are
607 skipped. This is true on many Unix systems, where the pseudo-op
608 to skip bytes produces no-op instructions rather than zeros when
609 used in the text section.
611 `ASM_OUTPUT_ALIGN (STREAM, POWER)'
612 A C statement to output to the stdio stream STREAM an assembler
613 command to advance the location counter to a multiple of 2 to the
614 POWER bytes. POWER will be a C expression of type `int'.
617 File: gcc.info, Node: Debugging Info, Next: Cross-compilation, Prev: Assembler Format, Up: Target Macros
619 Controlling Debugging Information Format
620 ========================================
622 This describes how to specify debugging information.
626 * All Debuggers:: Macros that affect all debugging formats uniformly.
627 * DBX Options:: Macros enabling specific options in DBX format.
628 * DBX Hooks:: Hook macros for varying DBX format.
629 * File Names and DBX:: Macros controlling output of file names in DBX format.
630 * SDB and DWARF:: Macros for SDB (COFF) and DWARF formats.
633 File: gcc.info, Node: All Debuggers, Next: DBX Options, Up: Debugging Info
635 Macros Affecting All Debugging Formats
636 --------------------------------------
638 These macros affect all debugging formats.
640 `DBX_REGISTER_NUMBER (REGNO)'
641 A C expression that returns the DBX register number for the
642 compiler register number REGNO. In simple cases, the value of this
643 expression may be REGNO itself. But sometimes there are some
644 registers that the compiler knows about and DBX does not, or vice
645 versa. In such cases, some register may need to have one number in
646 the compiler and another for DBX.
648 If two registers have consecutive numbers inside GNU CC, and they
649 can be used as a pair to hold a multiword value, then they *must*
650 have consecutive numbers after renumbering with
651 `DBX_REGISTER_NUMBER'. Otherwise, debuggers will be unable to
652 access such a pair, because they expect register pairs to be
653 consecutive in their own numbering scheme.
655 If you find yourself defining `DBX_REGISTER_NUMBER' in way that
656 does not preserve register pairs, then what you must do instead is
657 redefine the actual register numbering scheme.
659 `DEBUGGER_AUTO_OFFSET (X)'
660 A C expression that returns the integer offset value for an
661 automatic variable having address X (an RTL expression). The
662 default computation assumes that X is based on the frame-pointer
663 and gives the offset from the frame-pointer. This is required for
664 targets that produce debugging output for DBX or COFF-style
665 debugging output for SDB and allow the frame-pointer to be
666 eliminated when the `-g' options is used.
668 `DEBUGGER_ARG_OFFSET (OFFSET, X)'
669 A C expression that returns the integer offset value for an
670 argument having address X (an RTL expression). The nominal offset
673 `PREFERRED_DEBUGGING_TYPE'
674 A C expression that returns the type of debugging output GNU CC
675 produces when the user specifies `-g' or `-ggdb'. Define this if
676 you have arranged for GNU CC to support more than one format of
677 debugging output. Currently, the allowable values are `DBX_DEBUG',
678 `SDB_DEBUG', `DWARF_DEBUG', `DWARF2_DEBUG', and `XCOFF_DEBUG'.
680 The value of this macro only affects the default debugging output;
681 the user can always get a specific type of output by using
682 `-gstabs', `-gcoff', `-gdwarf-1', `-gdwarf-2', or `-gxcoff'.
685 File: gcc.info, Node: DBX Options, Next: DBX Hooks, Prev: All Debuggers, Up: Debugging Info
687 Specific Options for DBX Output
688 -------------------------------
690 These are specific options for DBX output.
693 Define this macro if GNU CC should produce debugging output for DBX
694 in response to the `-g' option.
696 `XCOFF_DEBUGGING_INFO'
697 Define this macro if GNU CC should produce XCOFF format debugging
698 output in response to the `-g' option. This is a variant of DBX
701 `DEFAULT_GDB_EXTENSIONS'
702 Define this macro to control whether GNU CC should by default
703 generate GDB's extended version of DBX debugging information
704 (assuming DBX-format debugging information is enabled at all). If
705 you don't define the macro, the default is 1: always generate the
706 extended information if there is any occasion to.
709 Define this macro if all `.stabs' commands should be output while
713 A C string constant naming the assembler pseudo op to use instead
714 of `.stabs' to define an ordinary debugging symbol. If you don't
715 define this macro, `.stabs' is used. This macro applies only to
716 DBX debugging information format.
719 A C string constant naming the assembler pseudo op to use instead
720 of `.stabd' to define a debugging symbol whose value is the current
721 location. If you don't define this macro, `.stabd' is used. This
722 macro applies only to DBX debugging information format.
725 A C string constant naming the assembler pseudo op to use instead
726 of `.stabn' to define a debugging symbol with no name. If you
727 don't define this macro, `.stabn' is used. This macro applies
728 only to DBX debugging information format.
731 Define this macro if DBX on your system does not support the
732 construct `xsTAGNAME'. On some systems, this construct is used to
733 describe a forward reference to a structure named TAGNAME. On
734 other systems, this construct is not supported at all.
737 A symbol name in DBX-format debugging information is normally
738 continued (split into two separate `.stabs' directives) when it
739 exceeds a certain length (by default, 80 characters). On some
740 operating systems, DBX requires this splitting; on others,
741 splitting must not be done. You can inhibit splitting by defining
742 this macro with the value zero. You can override the default
743 splitting-length by defining this macro as an expression for the
747 Normally continuation is indicated by adding a `\' character to
748 the end of a `.stabs' string when a continuation follows. To use
749 a different character instead, define this macro as a character
750 constant for the character you want to use. Do not define this
751 macro if backslash is correct for your system.
753 `DBX_STATIC_STAB_DATA_SECTION'
754 Define this macro if it is necessary to go to the data section
755 before outputting the `.stabs' pseudo-op for a non-global static
758 `DBX_TYPE_DECL_STABS_CODE'
759 The value to use in the "code" field of the `.stabs' directive for
760 a typedef. The default is `N_LSYM'.
762 `DBX_STATIC_CONST_VAR_CODE'
763 The value to use in the "code" field of the `.stabs' directive for
764 a static variable located in the text section. DBX format does not
765 provide any "right" way to do this. The default is `N_FUN'.
767 `DBX_REGPARM_STABS_CODE'
768 The value to use in the "code" field of the `.stabs' directive for
769 a parameter passed in registers. DBX format does not provide any
770 "right" way to do this. The default is `N_RSYM'.
772 `DBX_REGPARM_STABS_LETTER'
773 The letter to use in DBX symbol data to identify a symbol as a
774 parameter passed in registers. DBX format does not customarily
775 provide any way to do this. The default is `'P''.
777 `DBX_MEMPARM_STABS_LETTER'
778 The letter to use in DBX symbol data to identify a symbol as a
779 stack parameter. The default is `'p''.
782 Define this macro if the DBX information for a function and its
783 arguments should precede the assembler code for the function.
784 Normally, in DBX format, the debugging information entirely
785 follows the assembler code.
788 Define this macro if the `N_LBRAC' symbol for a block should
789 precede the debugging information for variables and functions
790 defined in that block. Normally, in DBX format, the `N_LBRAC'
793 `DBX_BLOCKS_FUNCTION_RELATIVE'
794 Define this macro if the value of a symbol describing the scope of
795 a block (`N_LBRAC' or `N_RBRAC') should be relative to the start
796 of the enclosing function. Normally, GNU C uses an absolute
800 Define this macro if GNU C should generate `N_BINCL' and `N_EINCL'
801 stabs for included header files, as on Sun systems. This macro
802 also directs GNU C to output a type number as a pair of a file
803 number and a type number within the file. Normally, GNU C does not
804 generate `N_BINCL' or `N_EINCL' stabs, and it outputs a single
805 number for a type number.
808 File: gcc.info, Node: DBX Hooks, Next: File Names and DBX, Prev: DBX Options, Up: Debugging Info
810 Open-Ended Hooks for DBX Format
811 -------------------------------
813 These are hooks for DBX format.
815 `DBX_OUTPUT_LBRAC (STREAM, NAME)'
816 Define this macro to say how to output to STREAM the debugging
817 information for the start of a scope level for variable names. The
818 argument NAME is the name of an assembler symbol (for use with
819 `assemble_name') whose value is the address where the scope begins.
821 `DBX_OUTPUT_RBRAC (STREAM, NAME)'
822 Like `DBX_OUTPUT_LBRAC', but for the end of a scope level.
824 `DBX_OUTPUT_ENUM (STREAM, TYPE)'
825 Define this macro if the target machine requires special handling
826 to output an enumeration type. The definition should be a C
827 statement (sans semicolon) to output the appropriate information
828 to STREAM for the type TYPE.
830 `DBX_OUTPUT_FUNCTION_END (STREAM, FUNCTION)'
831 Define this macro if the target machine requires special output at
832 the end of the debugging information for a function. The
833 definition should be a C statement (sans semicolon) to output the
834 appropriate information to STREAM. FUNCTION is the
835 `FUNCTION_DECL' node for the function.
837 `DBX_OUTPUT_STANDARD_TYPES (SYMS)'
838 Define this macro if you need to control the order of output of the
839 standard data types at the beginning of compilation. The argument
840 SYMS is a `tree' which is a chain of all the predefined global
841 symbols, including names of data types.
843 Normally, DBX output starts with definitions of the types for
844 integers and characters, followed by all the other predefined
845 types of the particular language in no particular order.
847 On some machines, it is necessary to output different particular
848 types first. To do this, define `DBX_OUTPUT_STANDARD_TYPES' to
849 output those symbols in the necessary order. Any predefined types
850 that you don't explicitly output will be output afterward in no
853 Be careful not to define this macro so that it works only for C.
854 There are no global variables to access most of the built-in
855 types, because another language may have another set of types.
856 The way to output a particular type is to look through SYMS to see
857 if you can find it. Here is an example:
861 for (decl = syms; decl; decl = TREE_CHAIN (decl))
862 if (!strcmp (IDENTIFIER_POINTER (DECL_NAME (decl)),
864 dbxout_symbol (decl);
868 This does nothing if the expected type does not exist.
870 See the function `init_decl_processing' in `c-decl.c' to find the
871 names to use for all the built-in C types.
873 Here is another way of finding a particular type:
877 for (decl = syms; decl; decl = TREE_CHAIN (decl))
878 if (TREE_CODE (decl) == TYPE_DECL
879 && (TREE_CODE (TREE_TYPE (decl))
881 && TYPE_PRECISION (TREE_TYPE (decl)) == 16
882 && TYPE_UNSIGNED (TREE_TYPE (decl)))
883 /* This must be `unsigned short'. */
884 dbxout_symbol (decl);
888 `NO_DBX_FUNCTION_END'
889 Some stabs encapsulation formats (in particular ECOFF), cannot
890 handle the `.stabs "",N_FUN,,0,0,Lscope-function-1' gdb dbx
891 extention construct. On those machines, define this macro to turn
892 this feature off without disturbing the rest of the gdb extensions.
895 File: gcc.info, Node: File Names and DBX, Next: SDB and DWARF, Prev: DBX Hooks, Up: Debugging Info
897 File Names in DBX Format
898 ------------------------
900 This describes file names in DBX format.
902 `DBX_WORKING_DIRECTORY'
903 Define this if DBX wants to have the current directory recorded in
906 Note that the working directory is always recorded if GDB
907 extensions are enabled.
909 `DBX_OUTPUT_MAIN_SOURCE_FILENAME (STREAM, NAME)'
910 A C statement to output DBX debugging information to the stdio
911 stream STREAM which indicates that file NAME is the main source
912 file--the file specified as the input file for compilation. This
913 macro is called only once, at the beginning of compilation.
915 This macro need not be defined if the standard form of output for
916 DBX debugging information is appropriate.
918 `DBX_OUTPUT_MAIN_SOURCE_DIRECTORY (STREAM, NAME)'
919 A C statement to output DBX debugging information to the stdio
920 stream STREAM which indicates that the current directory during
921 compilation is named NAME.
923 This macro need not be defined if the standard form of output for
924 DBX debugging information is appropriate.
926 `DBX_OUTPUT_MAIN_SOURCE_FILE_END (STREAM, NAME)'
927 A C statement to output DBX debugging information at the end of
928 compilation of the main source file NAME.
930 If you don't define this macro, nothing special is output at the
931 end of compilation, which is correct for most machines.
933 `DBX_OUTPUT_SOURCE_FILENAME (STREAM, NAME)'
934 A C statement to output DBX debugging information to the stdio
935 stream STREAM which indicates that file NAME is the current source
936 file. This output is generated each time input shifts to a
937 different source file as a result of `#include', the end of an
938 included file, or a `#line' command.
940 This macro need not be defined if the standard form of output for
941 DBX debugging information is appropriate.
944 File: gcc.info, Node: SDB and DWARF, Prev: File Names and DBX, Up: Debugging Info
946 Macros for SDB and DWARF Output
947 -------------------------------
949 Here are macros for SDB and DWARF output.
952 Define this macro if GNU CC should produce COFF-style debugging
953 output for SDB in response to the `-g' option.
955 `DWARF_DEBUGGING_INFO'
956 Define this macro if GNU CC should produce dwarf format debugging
957 output in response to the `-g' option.
959 `DWARF2_DEBUGGING_INFO'
960 Define this macro if GNU CC should produce dwarf version 2 format
961 debugging output in response to the `-g' option.
963 To support optional call frame debugging information, you must also
964 define `INCOMING_RETURN_ADDR_RTX' and either set
965 `RTX_FRAME_RELATED_P' on the prologue insns if you use RTL for the
966 prologue, or call `dwarf2out_def_cfa' and `dwarf2out_reg_save' as
967 appropriate from `FUNCTION_PROLOGUE' if you don't.
970 Define these macros to override the assembler syntax for the
971 special SDB assembler directives. See `sdbout.c' for a list of
972 these macros and their arguments. If the standard syntax is used,
973 you need not define them yourself.
976 Some assemblers do not support a semicolon as a delimiter, even
977 between SDB assembler directives. In that case, define this macro
978 to be the delimiter to use (usually `\n'). It is not necessary to
979 define a new set of `PUT_SDB_OP' macros if this is the only change
983 Define this macro to override the usual method of constructing a
984 dummy name for anonymous structure and union types. See
985 `sdbout.c' for more information.
987 `SDB_ALLOW_UNKNOWN_REFERENCES'
988 Define this macro to allow references to unknown structure, union,
989 or enumeration tags to be emitted. Standard COFF does not allow
990 handling of unknown references, MIPS ECOFF has support for it.
992 `SDB_ALLOW_FORWARD_REFERENCES'
993 Define this macro to allow references to structure, union, or
994 enumeration tags that have not yet been seen to be handled. Some
995 assemblers choke if forward tags are used, while some require it.
998 File: gcc.info, Node: Cross-compilation, Next: Misc, Prev: Debugging Info, Up: Target Macros
1000 Cross Compilation and Floating Point
1001 ====================================
1003 While all modern machines use 2's complement representation for
1004 integers, there are a variety of representations for floating point
1005 numbers. This means that in a cross-compiler the representation of
1006 floating point numbers in the compiled program may be different from
1007 that used in the machine doing the compilation.
1009 Because different representation systems may offer different amounts
1010 of range and precision, the cross compiler cannot safely use the host
1011 machine's floating point arithmetic. Therefore, floating point
1012 constants must be represented in the target machine's format. This
1013 means that the cross compiler cannot use `atof' to parse a floating
1014 point constant; it must have its own special routine to use instead.
1015 Also, constant folding must emulate the target machine's arithmetic (or
1016 must not be done at all).
1018 The macros in the following table should be defined only if you are
1019 cross compiling between different floating point formats.
1021 Otherwise, don't define them. Then default definitions will be set
1022 up which use `double' as the data type, `==' to test for equality, etc.
1024 You don't need to worry about how many times you use an operand of
1025 any of these macros. The compiler never uses operands which have side
1029 A macro for the C data type to be used to hold a floating point
1030 value in the target machine's format. Typically this would be a
1031 `struct' containing an array of `int'.
1033 `REAL_VALUES_EQUAL (X, Y)'
1034 A macro for a C expression which compares for equality the two
1035 values, X and Y, both of type `REAL_VALUE_TYPE'.
1037 `REAL_VALUES_LESS (X, Y)'
1038 A macro for a C expression which tests whether X is less than Y,
1039 both values being of type `REAL_VALUE_TYPE' and interpreted as
1040 floating point numbers in the target machine's representation.
1042 `REAL_VALUE_LDEXP (X, SCALE)'
1043 A macro for a C expression which performs the standard library
1044 function `ldexp', but using the target machine's floating point
1045 representation. Both X and the value of the expression have type
1046 `REAL_VALUE_TYPE'. The second argument, SCALE, is an integer.
1048 `REAL_VALUE_FIX (X)'
1049 A macro whose definition is a C expression to convert the
1050 target-machine floating point value X to a signed integer. X has
1051 type `REAL_VALUE_TYPE'.
1053 `REAL_VALUE_UNSIGNED_FIX (X)'
1054 A macro whose definition is a C expression to convert the
1055 target-machine floating point value X to an unsigned integer. X
1056 has type `REAL_VALUE_TYPE'.
1058 `REAL_VALUE_RNDZINT (X)'
1059 A macro whose definition is a C expression to round the
1060 target-machine floating point value X towards zero to an integer
1061 value (but still as a floating point number). X has type
1062 `REAL_VALUE_TYPE', and so does the value.
1064 `REAL_VALUE_UNSIGNED_RNDZINT (X)'
1065 A macro whose definition is a C expression to round the
1066 target-machine floating point value X towards zero to an unsigned
1067 integer value (but still represented as a floating point number).
1068 X has type `REAL_VALUE_TYPE', and so does the value.
1070 `REAL_VALUE_ATOF (STRING, MODE)'
1071 A macro for a C expression which converts STRING, an expression of
1072 type `char *', into a floating point number in the target machine's
1073 representation for mode MODE. The value has type
1077 Define this macro if infinity is a possible floating point value,
1078 and therefore division by 0 is legitimate.
1080 `REAL_VALUE_ISINF (X)'
1081 A macro for a C expression which determines whether X, a floating
1082 point value, is infinity. The value has type `int'. By default,
1083 this is defined to call `isinf'.
1085 `REAL_VALUE_ISNAN (X)'
1086 A macro for a C expression which determines whether X, a floating
1087 point value, is a "nan" (not-a-number). The value has type `int'.
1088 By default, this is defined to call `isnan'.
1090 Define the following additional macros if you want to make floating
1091 point constant folding work while cross compiling. If you don't define
1092 them, cross compilation is still possible, but constant folding will
1093 not happen for floating point values.
1095 `REAL_ARITHMETIC (OUTPUT, CODE, X, Y)'
1096 A macro for a C statement which calculates an arithmetic operation
1097 of the two floating point values X and Y, both of type
1098 `REAL_VALUE_TYPE' in the target machine's representation, to
1099 produce a result of the same type and representation which is
1100 stored in OUTPUT (which will be a variable).
1102 The operation to be performed is specified by CODE, a tree code
1103 which will always be one of the following: `PLUS_EXPR',
1104 `MINUS_EXPR', `MULT_EXPR', `RDIV_EXPR', `MAX_EXPR', `MIN_EXPR'.
1106 The expansion of this macro is responsible for checking for
1107 overflow. If overflow happens, the macro expansion should execute
1108 the statement `return 0;', which indicates the inability to
1109 perform the arithmetic operation requested.
1111 `REAL_VALUE_NEGATE (X)'
1112 A macro for a C expression which returns the negative of the
1113 floating point value X. Both X and the value of the expression
1114 have type `REAL_VALUE_TYPE' and are in the target machine's
1115 floating point representation.
1117 There is no way for this macro to report overflow, since overflow
1118 can't happen in the negation operation.
1120 `REAL_VALUE_TRUNCATE (MODE, X)'
1121 A macro for a C expression which converts the floating point value
1124 Both X and the value of the expression are in the target machine's
1125 floating point representation and have type `REAL_VALUE_TYPE'.
1126 However, the value should have an appropriate bit pattern to be
1127 output properly as a floating constant whose precision accords
1130 There is no way for this macro to report overflow.
1132 `REAL_VALUE_TO_INT (LOW, HIGH, X)'
1133 A macro for a C expression which converts a floating point value X
1134 into a double-precision integer which is then stored into LOW and
1135 HIGH, two variables of type INT.
1137 `REAL_VALUE_FROM_INT (X, LOW, HIGH, MODE)'
1138 A macro for a C expression which converts a double-precision
1139 integer found in LOW and HIGH, two variables of type INT, into a
1140 floating point value which is then stored into X. The value is in
1141 the target machine's representation for mode MODE and has the type