2 @setfilename stabs.info
9 * Stabs: (stabs). The "stabs" debugging information format.
15 This document describes the stabs debugging symbol tables.
17 Copyright 1992,1993,1994,1995,1997,1998,2000,2001
18 Free Software Foundation, Inc.
19 Contributed by Cygnus Support. Written by Julia Menapace, Jim Kingdon,
22 Permission is granted to copy, distribute and/or modify this document
23 under the terms of the GNU Free Documentation License, Version 1.1 or
24 any later version published by the Free Software Foundation; with the
25 Invariant Sections being ``Stabs Types'' and ``Stabs Sections'', with
26 the Front-Cover texts being ``A GNU Manual,'' and with the Back-Cover
27 Texts as in (a) below.
29 (a) The FSF's Back-Cover Text is: ``You have freedom to copy and modify
30 this GNU Manual, like GNU software. Copies published by the Free
31 Software Foundation raise funds for GNU development.''
34 @setchapternewpage odd
37 @title The ``stabs'' debug format
38 @author Julia Menapace, Jim Kingdon, David MacKenzie
39 @author Cygnus Support
42 \def\$#1${{#1}} % Kluge: collect RCS revision info without $...$
43 \xdef\manvers{\$Revision$} % For use in headers, footers too
45 \hfill Cygnus Support\par
47 \hfill \TeX{}info \texinfoversion\par
51 @vskip 0pt plus 1filll
52 Copyright @copyright{} 1992,1993,1994,1995,1997,1998,2000,2001 Free Software Foundation, Inc.
53 Contributed by Cygnus Support.
55 Permission is granted to copy, distribute and/or modify this document
56 under the terms of the GNU Free Documentation License, Version 1.1 or
57 any later version published by the Free Software Foundation; with the
58 Invariant Sections being ``Stabs Types'' and ``Stabs Sections'', with
59 the Front-Cover texts being ``A GNU Manual,'' and with the Back-Cover
60 Texts as in (a) below.
62 (a) The FSF's Back-Cover Text is: ``You have freedom to copy and modify
63 this GNU Manual, like GNU software. Copies published by the Free
64 Software Foundation raise funds for GNU development.''
70 @top The "stabs" representation of debugging information
72 This document describes the stabs debugging format.
75 * Overview:: Overview of stabs
76 * Program Structure:: Encoding of the structure of the program
77 * Constants:: Constants
79 * Types:: Type definitions
80 * Symbol Tables:: Symbol information in symbol tables
81 * Cplusplus:: Stabs specific to C++
82 * Stab Types:: Symbol types in a.out files
83 * Symbol Descriptors:: Table of symbol descriptors
84 * Type Descriptors:: Table of type descriptors
85 * Expanded Reference:: Reference information by stab type
86 * Questions:: Questions and anomalies
87 * Stab Sections:: In some object file formats, stabs are
89 * Symbol Types Index:: Index of symbolic stab symbol type names.
93 @c TeX can handle the contents at the start but makeinfo 3.12 can not
99 @chapter Overview of Stabs
101 @dfn{Stabs} refers to a format for information that describes a program
102 to a debugger. This format was apparently invented by
104 the University of California at Berkeley, for the @code{pdx} Pascal
105 debugger; the format has spread widely since then.
107 This document is one of the few published sources of documentation on
108 stabs. It is believed to be comprehensive for stabs used by C. The
109 lists of symbol descriptors (@pxref{Symbol Descriptors}) and type
110 descriptors (@pxref{Type Descriptors}) are believed to be completely
111 comprehensive. Stabs for COBOL-specific features and for variant
112 records (used by Pascal and Modula-2) are poorly documented here.
114 @c FIXME: Need to document all OS9000 stuff in GDB; see all references
115 @c to os9k_stabs in stabsread.c.
117 Other sources of information on stabs are @cite{Dbx and Dbxtool
118 Interfaces}, 2nd edition, by Sun, 1988, and @cite{AIX Version 3.2 Files
119 Reference}, Fourth Edition, September 1992, "dbx Stabstring Grammar" in
120 the a.out section, page 2-31. This document is believed to incorporate
121 the information from those two sources except where it explicitly directs
122 you to them for more information.
125 * Flow:: Overview of debugging information flow
126 * Stabs Format:: Overview of stab format
127 * String Field:: The string field
128 * C Example:: A simple example in C source
129 * Assembly Code:: The simple example at the assembly level
133 @section Overview of Debugging Information Flow
135 The GNU C compiler compiles C source in a @file{.c} file into assembly
136 language in a @file{.s} file, which the assembler translates into
137 a @file{.o} file, which the linker combines with other @file{.o} files and
138 libraries to produce an executable file.
140 With the @samp{-g} option, GCC puts in the @file{.s} file additional
141 debugging information, which is slightly transformed by the assembler
142 and linker, and carried through into the final executable. This
143 debugging information describes features of the source file like line
144 numbers, the types and scopes of variables, and function names,
145 parameters, and scopes.
147 For some object file formats, the debugging information is encapsulated
148 in assembler directives known collectively as @dfn{stab} (symbol table)
149 directives, which are interspersed with the generated code. Stabs are
150 the native format for debugging information in the a.out and XCOFF
151 object file formats. The GNU tools can also emit stabs in the COFF and
152 ECOFF object file formats.
154 The assembler adds the information from stabs to the symbol information
155 it places by default in the symbol table and the string table of the
156 @file{.o} file it is building. The linker consolidates the @file{.o}
157 files into one executable file, with one symbol table and one string
158 table. Debuggers use the symbol and string tables in the executable as
159 a source of debugging information about the program.
162 @section Overview of Stab Format
164 There are three overall formats for stab assembler directives,
165 differentiated by the first word of the stab. The name of the directive
166 describes which combination of four possible data fields follows. It is
167 either @code{.stabs} (string), @code{.stabn} (number), or @code{.stabd}
168 (dot). IBM's XCOFF assembler uses @code{.stabx} (and some other
169 directives such as @code{.file} and @code{.bi}) instead of
170 @code{.stabs}, @code{.stabn} or @code{.stabd}.
172 The overall format of each class of stab is:
175 .stabs "@var{string}",@var{type},@var{other},@var{desc},@var{value}
176 .stabn @var{type},@var{other},@var{desc},@var{value}
177 .stabd @var{type},@var{other},@var{desc}
178 .stabx "@var{string}",@var{value},@var{type},@var{sdb-type}
181 @c what is the correct term for "current file location"? My AIX
182 @c assembler manual calls it "the value of the current location counter".
183 For @code{.stabn} and @code{.stabd}, there is no @var{string} (the
184 @code{n_strx} field is zero; see @ref{Symbol Tables}). For
185 @code{.stabd}, the @var{value} field is implicit and has the value of
186 the current file location. For @code{.stabx}, the @var{sdb-type} field
187 is unused for stabs and can always be set to zero. The @var{other}
188 field is almost always unused and can be set to zero.
190 The number in the @var{type} field gives some basic information about
191 which type of stab this is (or whether it @emph{is} a stab, as opposed
192 to an ordinary symbol). Each valid type number defines a different stab
193 type; further, the stab type defines the exact interpretation of, and
194 possible values for, any remaining @var{string}, @var{desc}, or
195 @var{value} fields present in the stab. @xref{Stab Types}, for a list
196 in numeric order of the valid @var{type} field values for stab directives.
199 @section The String Field
201 For most stabs the string field holds the meat of the
202 debugging information. The flexible nature of this field
203 is what makes stabs extensible. For some stab types the string field
204 contains only a name. For other stab types the contents can be a great
207 The overall format of the string field for most stab types is:
210 "@var{name}:@var{symbol-descriptor} @var{type-information}"
213 @var{name} is the name of the symbol represented by the stab; it can
214 contain a pair of colons (@pxref{Nested Symbols}). @var{name} can be
215 omitted, which means the stab represents an unnamed object. For
216 example, @samp{:t10=*2} defines type 10 as a pointer to type 2, but does
217 not give the type a name. Omitting the @var{name} field is supported by
218 AIX dbx and GDB after about version 4.8, but not other debuggers. GCC
219 sometimes uses a single space as the name instead of omitting the name
220 altogether; apparently that is supported by most debuggers.
222 The @var{symbol-descriptor} following the @samp{:} is an alphabetic
223 character that tells more specifically what kind of symbol the stab
224 represents. If the @var{symbol-descriptor} is omitted, but type
225 information follows, then the stab represents a local variable. For a
226 list of symbol descriptors, see @ref{Symbol Descriptors}. The @samp{c}
227 symbol descriptor is an exception in that it is not followed by type
228 information. @xref{Constants}.
230 @var{type-information} is either a @var{type-number}, or
231 @samp{@var{type-number}=}. A @var{type-number} alone is a type
232 reference, referring directly to a type that has already been defined.
234 The @samp{@var{type-number}=} form is a type definition, where the
235 number represents a new type which is about to be defined. The type
236 definition may refer to other types by number, and those type numbers
237 may be followed by @samp{=} and nested definitions. Also, the Lucid
238 compiler will repeat @samp{@var{type-number}=} more than once if it
239 wants to define several type numbers at once.
241 In a type definition, if the character that follows the equals sign is
242 non-numeric then it is a @var{type-descriptor}, and tells what kind of
243 type is about to be defined. Any other values following the
244 @var{type-descriptor} vary, depending on the @var{type-descriptor}.
245 @xref{Type Descriptors}, for a list of @var{type-descriptor} values. If
246 a number follows the @samp{=} then the number is a @var{type-reference}.
247 For a full description of types, @ref{Types}.
249 A @var{type-number} is often a single number. The GNU and Sun tools
250 additionally permit a @var{type-number} to be a pair
251 (@var{file-number},@var{filetype-number}) (the parentheses appear in the
252 string, and serve to distinguish the two cases). The @var{file-number}
253 is 0 for the base source file, 1 for the first included file, 2 for the
254 next, and so on. The @var{filetype-number} is a number starting with
255 1 which is incremented for each new type defined in the file.
256 (Separating the file number and the type number permits the
257 @code{N_BINCL} optimization to succeed more often; see @ref{Include
260 There is an AIX extension for type attributes. Following the @samp{=}
261 are any number of type attributes. Each one starts with @samp{@@} and
262 ends with @samp{;}. Debuggers, including AIX's dbx and GDB 4.10, skip
263 any type attributes they do not recognize. GDB 4.9 and other versions
264 of dbx may not do this. Because of a conflict with C++
265 (@pxref{Cplusplus}), new attributes should not be defined which begin
266 with a digit, @samp{(}, or @samp{-}; GDB may be unable to distinguish
267 those from the C++ type descriptor @samp{@@}. The attributes are:
270 @item a@var{boundary}
271 @var{boundary} is an integer specifying the alignment. I assume it
272 applies to all variables of this type.
275 Pointer class (for checking). Not sure what this means, or how
276 @var{integer} is interpreted.
279 Indicate this is a packed type, meaning that structure fields or array
280 elements are placed more closely in memory, to save memory at the
284 Size in bits of a variable of this type. This is fully supported by GDB
288 Indicate that this type is a string instead of an array of characters,
289 or a bitstring instead of a set. It doesn't change the layout of the
290 data being represented, but does enable the debugger to know which type
294 All of this can make the string field quite long. All versions of GDB,
295 and some versions of dbx, can handle arbitrarily long strings. But many
296 versions of dbx (or assemblers or linkers, I'm not sure which)
297 cretinously limit the strings to about 80 characters, so compilers which
298 must work with such systems need to split the @code{.stabs} directive
299 into several @code{.stabs} directives. Each stab duplicates every field
300 except the string field. The string field of every stab except the last
301 is marked as continued with a backslash at the end (in the assembly code
302 this may be written as a double backslash, depending on the assembler).
303 Removing the backslashes and concatenating the string fields of each
304 stab produces the original, long string. Just to be incompatible (or so
305 they don't have to worry about what the assembler does with
306 backslashes), AIX can use @samp{?} instead of backslash.
309 @section A Simple Example in C Source
311 To get the flavor of how stabs describe source information for a C
312 program, let's look at the simple program:
317 printf("Hello world");
321 When compiled with @samp{-g}, the program above yields the following
322 @file{.s} file. Line numbers have been added to make it easier to refer
323 to parts of the @file{.s} file in the description of the stabs that
327 @section The Simple Example at the Assembly Level
329 This simple ``hello world'' example demonstrates several of the stab
330 types used to describe C language source files.
334 2 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0
335 3 .stabs "hello.c",100,0,0,Ltext0
338 6 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0
339 7 .stabs "char:t2=r2;0;127;",128,0,0,0
340 8 .stabs "long int:t3=r1;-2147483648;2147483647;",128,0,0,0
341 9 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
342 10 .stabs "long unsigned int:t5=r1;0;-1;",128,0,0,0
343 11 .stabs "short int:t6=r1;-32768;32767;",128,0,0,0
344 12 .stabs "long long int:t7=r1;0;-1;",128,0,0,0
345 13 .stabs "short unsigned int:t8=r1;0;65535;",128,0,0,0
346 14 .stabs "long long unsigned int:t9=r1;0;-1;",128,0,0,0
347 15 .stabs "signed char:t10=r1;-128;127;",128,0,0,0
348 16 .stabs "unsigned char:t11=r1;0;255;",128,0,0,0
349 17 .stabs "float:t12=r1;4;0;",128,0,0,0
350 18 .stabs "double:t13=r1;8;0;",128,0,0,0
351 19 .stabs "long double:t14=r1;8;0;",128,0,0,0
352 20 .stabs "void:t15=15",128,0,0,0
355 23 .ascii "Hello, world!\12\0"
370 38 sethi %hi(LC0),%o1
371 39 or %o1,%lo(LC0),%o0
382 50 .stabs "main:F1",36,0,0,_main
383 51 .stabn 192,0,0,LBB2
384 52 .stabn 224,0,0,LBE2
387 @node Program Structure
388 @chapter Encoding the Structure of the Program
390 The elements of the program structure that stabs encode include the name
391 of the main function, the names of the source and include files, the
392 line numbers, procedure names and types, and the beginnings and ends of
396 * Main Program:: Indicate what the main program is
397 * Source Files:: The path and name of the source file
398 * Include Files:: Names of include files
401 * Nested Procedures::
403 * Alternate Entry Points:: Entering procedures except at the beginning.
407 @section Main Program
410 Most languages allow the main program to have any name. The
411 @code{N_MAIN} stab type tells the debugger the name that is used in this
412 program. Only the string field is significant; it is the name of
413 a function which is the main program. Most C compilers do not use this
414 stab (they expect the debugger to assume that the name is @code{main}),
415 but some C compilers emit an @code{N_MAIN} stab for the @code{main}
416 function. I'm not sure how XCOFF handles this.
419 @section Paths and Names of the Source Files
422 Before any other stabs occur, there must be a stab specifying the source
423 file. This information is contained in a symbol of stab type
424 @code{N_SO}; the string field contains the name of the file. The
425 value of the symbol is the start address of the portion of the
426 text section corresponding to that file.
428 With the Sun Solaris2 compiler, the desc field contains a
429 source-language code.
430 @c Do the debuggers use it? What are the codes? -djm
432 Some compilers (for example, GCC2 and SunOS4 @file{/bin/cc}) also
433 include the directory in which the source was compiled, in a second
434 @code{N_SO} symbol preceding the one containing the file name. This
435 symbol can be distinguished by the fact that it ends in a slash. Code
436 from the @code{cfront} C++ compiler can have additional @code{N_SO} symbols for
437 nonexistent source files after the @code{N_SO} for the real source file;
438 these are believed to contain no useful information.
443 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0 # @r{100 is N_SO}
444 .stabs "hello.c",100,0,0,Ltext0
450 Instead of @code{N_SO} symbols, XCOFF uses a @code{.file} assembler
451 directive which assembles to a @code{C_FILE} symbol; explaining this in
452 detail is outside the scope of this document.
454 @c FIXME: Exactly when should the empty N_SO be used? Why?
455 If it is useful to indicate the end of a source file, this is done with
456 an @code{N_SO} symbol with an empty string for the name. The value is
457 the address of the end of the text section for the file. For some
458 systems, there is no indication of the end of a source file, and you
459 just need to figure it ended when you see an @code{N_SO} for a different
460 source file, or a symbol ending in @code{.o} (which at least some
461 linkers insert to mark the start of a new @code{.o} file).
464 @section Names of Include Files
466 There are several schemes for dealing with include files: the
467 traditional @code{N_SOL} approach, Sun's @code{N_BINCL} approach, and the
468 XCOFF @code{C_BINCL} approach (which despite the similar name has little in
469 common with @code{N_BINCL}).
472 An @code{N_SOL} symbol specifies which include file subsequent symbols
473 refer to. The string field is the name of the file and the value is the
474 text address corresponding to the end of the previous include file and
475 the start of this one. To specify the main source file again, use an
476 @code{N_SOL} symbol with the name of the main source file.
481 The @code{N_BINCL} approach works as follows. An @code{N_BINCL} symbol
482 specifies the start of an include file. In an object file, only the
483 string is significant; the linker puts data into some of the other
484 fields. The end of the include file is marked by an @code{N_EINCL}
485 symbol (which has no string field). In an object file, there is no
486 significant data in the @code{N_EINCL} symbol. @code{N_BINCL} and
487 @code{N_EINCL} can be nested.
489 If the linker detects that two source files have identical stabs between
490 an @code{N_BINCL} and @code{N_EINCL} pair (as will generally be the case
491 for a header file), then it only puts out the stabs once. Each
492 additional occurrence is replaced by an @code{N_EXCL} symbol. I believe
493 the GNU linker and the Sun (both SunOS4 and Solaris) linker are the only
494 ones which supports this feature.
496 A linker which supports this feature will set the value of a
497 @code{N_BINCL} symbol to the total of all the characters in the stabs
498 strings included in the header file, omitting any file numbers. The
499 value of an @code{N_EXCL} symbol is the same as the value of the
500 @code{N_BINCL} symbol it replaces. This information can be used to
501 match up @code{N_EXCL} and @code{N_BINCL} symbols which have the same
502 filename. The @code{N_EINCL} value, and the values of the other and
503 description fields for all three, appear to always be zero.
507 For the start of an include file in XCOFF, use the @file{.bi} assembler
508 directive, which generates a @code{C_BINCL} symbol. A @file{.ei}
509 directive, which generates a @code{C_EINCL} symbol, denotes the end of
510 the include file. Both directives are followed by the name of the
511 source file in quotes, which becomes the string for the symbol.
512 The value of each symbol, produced automatically by the assembler
513 and linker, is the offset into the executable of the beginning
514 (inclusive, as you'd expect) or end (inclusive, as you would not expect)
515 of the portion of the COFF line table that corresponds to this include
516 file. @code{C_BINCL} and @code{C_EINCL} do not nest.
519 @section Line Numbers
522 An @code{N_SLINE} symbol represents the start of a source line. The
523 desc field contains the line number and the value contains the code
524 address for the start of that source line. On most machines the address
525 is absolute; for stabs in sections (@pxref{Stab Sections}), it is
526 relative to the function in which the @code{N_SLINE} symbol occurs.
530 GNU documents @code{N_DSLINE} and @code{N_BSLINE} symbols for line
531 numbers in the data or bss segments, respectively. They are identical
532 to @code{N_SLINE} but are relocated differently by the linker. They
533 were intended to be used to describe the source location of a variable
534 declaration, but I believe that GCC2 actually puts the line number in
535 the desc field of the stab for the variable itself. GDB has been
536 ignoring these symbols (unless they contain a string field) since
539 For single source lines that generate discontiguous code, such as flow
540 of control statements, there may be more than one line number entry for
541 the same source line. In this case there is a line number entry at the
542 start of each code range, each with the same line number.
544 XCOFF does not use stabs for line numbers. Instead, it uses COFF line
545 numbers (which are outside the scope of this document). Standard COFF
546 line numbers cannot deal with include files, but in XCOFF this is fixed
547 with the @code{C_BINCL} method of marking include files (@pxref{Include
553 @findex N_FUN, for functions
555 @findex N_STSYM, for functions (Sun acc)
556 @findex N_GSYM, for functions (Sun acc)
557 All of the following stabs normally use the @code{N_FUN} symbol type.
558 However, Sun's @code{acc} compiler on SunOS4 uses @code{N_GSYM} and
559 @code{N_STSYM}, which means that the value of the stab for the function
560 is useless and the debugger must get the address of the function from
561 the non-stab symbols instead. On systems where non-stab symbols have
562 leading underscores, the stabs will lack underscores and the debugger
563 needs to know about the leading underscore to match up the stab and the
564 non-stab symbol. BSD Fortran is said to use @code{N_FNAME} with the
565 same restriction; the value of the symbol is not useful (I'm not sure it
566 really does use this, because GDB doesn't handle this and no one has
570 A function is represented by an @samp{F} symbol descriptor for a global
571 (extern) function, and @samp{f} for a static (local) function. For
572 a.out, the value of the symbol is the address of the start of the
573 function; it is already relocated. For stabs in ELF, the SunPRO
574 compiler version 2.0.1 and GCC put out an address which gets relocated
575 by the linker. In a future release SunPRO is planning to put out zero,
576 in which case the address can be found from the ELF (non-stab) symbol.
577 Because looking things up in the ELF symbols would probably be slow, I'm
578 not sure how to find which symbol of that name is the right one, and
579 this doesn't provide any way to deal with nested functions, it would
580 probably be better to make the value of the stab an address relative to
581 the start of the file, or just absolute. See @ref{ELF Linker
582 Relocation} for more information on linker relocation of stabs in ELF
583 files. For XCOFF, the stab uses the @code{C_FUN} storage class and the
584 value of the stab is meaningless; the address of the function can be
585 found from the csect symbol (XTY_LD/XMC_PR).
587 The type information of the stab represents the return type of the
588 function; thus @samp{foo:f5} means that foo is a function returning type
589 5. There is no need to try to get the line number of the start of the
590 function from the stab for the function; it is in the next
591 @code{N_SLINE} symbol.
593 @c FIXME: verify whether the "I suspect" below is true or not.
594 Some compilers (such as Sun's Solaris compiler) support an extension for
595 specifying the types of the arguments. I suspect this extension is not
596 used for old (non-prototyped) function definitions in C. If the
597 extension is in use, the type information of the stab for the function
598 is followed by type information for each argument, with each argument
599 preceded by @samp{;}. An argument type of 0 means that additional
600 arguments are being passed, whose types and number may vary (@samp{...}
601 in ANSI C). GDB has tolerated this extension (parsed the syntax, if not
602 necessarily used the information) since at least version 4.8; I don't
603 know whether all versions of dbx tolerate it. The argument types given
604 here are not redundant with the symbols for the formal parameters
605 (@pxref{Parameters}); they are the types of the arguments as they are
606 passed, before any conversions might take place. For example, if a C
607 function which is declared without a prototype takes a @code{float}
608 argument, the value is passed as a @code{double} but then converted to a
609 @code{float}. Debuggers need to use the types given in the arguments
610 when printing values, but when calling the function they need to use the
611 types given in the symbol defining the function.
613 If the return type and types of arguments of a function which is defined
614 in another source file are specified (i.e., a function prototype in ANSI
615 C), traditionally compilers emit no stab; the only way for the debugger
616 to find the information is if the source file where the function is
617 defined was also compiled with debugging symbols. As an extension the
618 Solaris compiler uses symbol descriptor @samp{P} followed by the return
619 type of the function, followed by the arguments, each preceded by
620 @samp{;}, as in a stab with symbol descriptor @samp{f} or @samp{F}.
621 This use of symbol descriptor @samp{P} can be distinguished from its use
622 for register parameters (@pxref{Register Parameters}) by the fact that it has
623 symbol type @code{N_FUN}.
625 The AIX documentation also defines symbol descriptor @samp{J} as an
626 internal function. I assume this means a function nested within another
627 function. It also says symbol descriptor @samp{m} is a module in
628 Modula-2 or extended Pascal.
630 Procedures (functions which do not return values) are represented as
631 functions returning the @code{void} type in C. I don't see why this couldn't
632 be used for all languages (inventing a @code{void} type for this purpose if
633 necessary), but the AIX documentation defines @samp{I}, @samp{P}, and
634 @samp{Q} for internal, global, and static procedures, respectively.
635 These symbol descriptors are unusual in that they are not followed by
638 The following example shows a stab for a function @code{main} which
639 returns type number @code{1}. The @code{_main} specified for the value
640 is a reference to an assembler label which is used to fill in the start
641 address of the function.
644 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
647 The stab representing a procedure is located immediately following the
648 code of the procedure. This stab is in turn directly followed by a
649 group of other stabs describing elements of the procedure. These other
650 stabs describe the procedure's parameters, its block local variables, and
653 If functions can appear in different sections, then the debugger may not
654 be able to find the end of a function. Recent versions of GCC will mark
655 the end of a function with an @code{N_FUN} symbol with an empty string
656 for the name. The value is the address of the end of the current
657 function. Without such a symbol, there is no indication of the address
658 of the end of a function, and you must assume that it ended at the
659 starting address of the next function or at the end of the text section
662 @node Nested Procedures
663 @section Nested Procedures
665 For any of the symbol descriptors representing procedures, after the
666 symbol descriptor and the type information is optionally a scope
667 specifier. This consists of a comma, the name of the procedure, another
668 comma, and the name of the enclosing procedure. The first name is local
669 to the scope specified, and seems to be redundant with the name of the
670 symbol (before the @samp{:}). This feature is used by GCC, and
671 presumably Pascal, Modula-2, etc., compilers, for nested functions.
673 If procedures are nested more than one level deep, only the immediately
674 containing scope is specified. For example, this code:
686 return baz (x + 2 * y);
688 return x + bar (3 * x);
696 .stabs "baz:f1,baz,bar",36,0,0,_baz.15 # @r{36 is N_FUN}
697 .stabs "bar:f1,bar,foo",36,0,0,_bar.12
698 .stabs "foo:F1",36,0,0,_foo
701 @node Block Structure
702 @section Block Structure
706 @c For GCC 2.5.8 or so stabs-in-coff, these are absolute instead of
707 @c function relative (as documented below). But GDB has never been able
708 @c to deal with that (it had wanted them to be relative to the file, but
709 @c I just fixed that (between GDB 4.12 and 4.13)), so it is function
710 @c relative just like ELF and SOM and the below documentation.
711 The program's block structure is represented by the @code{N_LBRAC} (left
712 brace) and the @code{N_RBRAC} (right brace) stab types. The variables
713 defined inside a block precede the @code{N_LBRAC} symbol for most
714 compilers, including GCC. Other compilers, such as the Convex, Acorn
715 RISC machine, and Sun @code{acc} compilers, put the variables after the
716 @code{N_LBRAC} symbol. The values of the @code{N_LBRAC} and
717 @code{N_RBRAC} symbols are the start and end addresses of the code of
718 the block, respectively. For most machines, they are relative to the
719 starting address of this source file. For the Gould NP1, they are
720 absolute. For stabs in sections (@pxref{Stab Sections}), they are
721 relative to the function in which they occur.
723 The @code{N_LBRAC} and @code{N_RBRAC} stabs that describe the block
724 scope of a procedure are located after the @code{N_FUN} stab that
725 represents the procedure itself.
727 Sun documents the desc field of @code{N_LBRAC} and
728 @code{N_RBRAC} symbols as containing the nesting level of the block.
729 However, dbx seems to not care, and GCC always sets desc to
735 For XCOFF, block scope is indicated with @code{C_BLOCK} symbols. If the
736 name of the symbol is @samp{.bb}, then it is the beginning of the block;
737 if the name of the symbol is @samp{.be}; it is the end of the block.
739 @node Alternate Entry Points
740 @section Alternate Entry Points
744 Some languages, like Fortran, have the ability to enter procedures at
745 some place other than the beginning. One can declare an alternate entry
746 point. The @code{N_ENTRY} stab is for this; however, the Sun FORTRAN
747 compiler doesn't use it. According to AIX documentation, only the name
748 of a @code{C_ENTRY} stab is significant; the address of the alternate
749 entry point comes from the corresponding external symbol. A previous
750 revision of this document said that the value of an @code{N_ENTRY} stab
751 was the address of the alternate entry point, but I don't know the
752 source for that information.
757 The @samp{c} symbol descriptor indicates that this stab represents a
758 constant. This symbol descriptor is an exception to the general rule
759 that symbol descriptors are followed by type information. Instead, it
760 is followed by @samp{=} and one of the following:
764 Boolean constant. @var{value} is a numeric value; I assume it is 0 for
768 Character constant. @var{value} is the numeric value of the constant.
770 @item e @var{type-information} , @var{value}
771 Constant whose value can be represented as integral.
772 @var{type-information} is the type of the constant, as it would appear
773 after a symbol descriptor (@pxref{String Field}). @var{value} is the
774 numeric value of the constant. GDB 4.9 does not actually get the right
775 value if @var{value} does not fit in a host @code{int}, but it does not
776 do anything violent, and future debuggers could be extended to accept
777 integers of any size (whether unsigned or not). This constant type is
778 usually documented as being only for enumeration constants, but GDB has
779 never imposed that restriction; I don't know about other debuggers.
782 Integer constant. @var{value} is the numeric value. The type is some
783 sort of generic integer type (for GDB, a host @code{int}); to specify
784 the type explicitly, use @samp{e} instead.
787 Real constant. @var{value} is the real value, which can be @samp{INF}
788 (optionally preceded by a sign) for infinity, @samp{QNAN} for a quiet
789 NaN (not-a-number), or @samp{SNAN} for a signalling NaN. If it is a
790 normal number the format is that accepted by the C library function
794 String constant. @var{string} is a string enclosed in either @samp{'}
795 (in which case @samp{'} characters within the string are represented as
796 @samp{\'} or @samp{"} (in which case @samp{"} characters within the
797 string are represented as @samp{\"}).
799 @item S @var{type-information} , @var{elements} , @var{bits} , @var{pattern}
800 Set constant. @var{type-information} is the type of the constant, as it
801 would appear after a symbol descriptor (@pxref{String Field}).
802 @var{elements} is the number of elements in the set (does this means
803 how many bits of @var{pattern} are actually used, which would be
804 redundant with the type, or perhaps the number of bits set in
805 @var{pattern}? I don't get it), @var{bits} is the number of bits in the
806 constant (meaning it specifies the length of @var{pattern}, I think),
807 and @var{pattern} is a hexadecimal representation of the set. AIX
808 documentation refers to a limit of 32 bytes, but I see no reason why
809 this limit should exist. This form could probably be used for arbitrary
810 constants, not just sets; the only catch is that @var{pattern} should be
811 understood to be target, not host, byte order and format.
814 The boolean, character, string, and set constants are not supported by
815 GDB 4.9, but it ignores them. GDB 4.8 and earlier gave an error
816 message and refused to read symbols from the file containing the
819 The above information is followed by @samp{;}.
824 Different types of stabs describe the various ways that variables can be
825 allocated: on the stack, globally, in registers, in common blocks,
826 statically, or as arguments to a function.
829 * Stack Variables:: Variables allocated on the stack.
830 * Global Variables:: Variables used by more than one source file.
831 * Register Variables:: Variables in registers.
832 * Common Blocks:: Variables statically allocated together.
833 * Statics:: Variables local to one source file.
834 * Based Variables:: Fortran pointer based variables.
835 * Parameters:: Variables for arguments to functions.
838 @node Stack Variables
839 @section Automatic Variables Allocated on the Stack
841 If a variable's scope is local to a function and its lifetime is only as
842 long as that function executes (C calls such variables
843 @dfn{automatic}), it can be allocated in a register (@pxref{Register
844 Variables}) or on the stack.
846 @findex N_LSYM, for stack variables
848 Each variable allocated on the stack has a stab with the symbol
849 descriptor omitted. Since type information should begin with a digit,
850 @samp{-}, or @samp{(}, only those characters precluded from being used
851 for symbol descriptors. However, the Acorn RISC machine (ARM) is said
852 to get this wrong: it puts out a mere type definition here, without the
853 preceding @samp{@var{type-number}=}. This is a bad idea; there is no
854 guarantee that type descriptors are distinct from symbol descriptors.
855 Stabs for stack variables use the @code{N_LSYM} stab type, or
856 @code{C_LSYM} for XCOFF.
858 The value of the stab is the offset of the variable within the
859 local variables. On most machines this is an offset from the frame
860 pointer and is negative. The location of the stab specifies which block
861 it is defined in; see @ref{Block Structure}.
863 For example, the following C code:
873 produces the following stabs:
876 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
877 .stabs "x:1",128,0,0,-12 # @r{128 is N_LSYM}
878 .stabn 192,0,0,LBB2 # @r{192 is N_LBRAC}
879 .stabn 224,0,0,LBE2 # @r{224 is N_RBRAC}
882 See @ref{Procedures} for more information on the @code{N_FUN} stab, and
883 @ref{Block Structure} for more information on the @code{N_LBRAC} and
884 @code{N_RBRAC} stabs.
886 @node Global Variables
887 @section Global Variables
891 @c FIXME: verify for sure that it really is C_GSYM on XCOFF
892 A variable whose scope is not specific to just one source file is
893 represented by the @samp{G} symbol descriptor. These stabs use the
894 @code{N_GSYM} stab type (C_GSYM for XCOFF). The type information for
895 the stab (@pxref{String Field}) gives the type of the variable.
897 For example, the following source code:
904 yields the following assembly code:
907 .stabs "g_foo:G2",32,0,0,0 # @r{32 is N_GSYM}
914 The address of the variable represented by the @code{N_GSYM} is not
915 contained in the @code{N_GSYM} stab. The debugger gets this information
916 from the external symbol for the global variable. In the example above,
917 the @code{.global _g_foo} and @code{_g_foo:} lines tell the assembler to
918 produce an external symbol.
920 Some compilers, like GCC, output @code{N_GSYM} stabs only once, where
921 the variable is defined. Other compilers, like SunOS4 /bin/cc, output a
922 @code{N_GSYM} stab for each compilation unit which references the
925 @node Register Variables
926 @section Register Variables
930 @c According to an old version of this manual, AIX uses C_RPSYM instead
931 @c of C_RSYM. I am skeptical; this should be verified.
932 Register variables have their own stab type, @code{N_RSYM}
933 (@code{C_RSYM} for XCOFF), and their own symbol descriptor, @samp{r}.
934 The stab's value is the number of the register where the variable data
936 @c .stabs "name:type",N_RSYM,0,RegSize,RegNumber (Sun doc)
938 AIX defines a separate symbol descriptor @samp{d} for floating point
939 registers. This seems unnecessary; why not just just give floating
940 point registers different register numbers? I have not verified whether
941 the compiler actually uses @samp{d}.
943 If the register is explicitly allocated to a global variable, but not
947 register int g_bar asm ("%g5");
951 then the stab may be emitted at the end of the object file, with
952 the other bss symbols.
955 @section Common Blocks
957 A common block is a statically allocated section of memory which can be
958 referred to by several source files. It may contain several variables.
959 I believe Fortran is the only language with this feature.
965 A @code{N_BCOMM} stab begins a common block and an @code{N_ECOMM} stab
966 ends it. The only field that is significant in these two stabs is the
967 string, which names a normal (non-debugging) symbol that gives the
968 address of the common block. According to IBM documentation, only the
969 @code{N_BCOMM} has the name of the common block (even though their
970 compiler actually puts it both places).
974 The stabs for the members of the common block are between the
975 @code{N_BCOMM} and the @code{N_ECOMM}; the value of each stab is the
976 offset within the common block of that variable. IBM uses the
977 @code{C_ECOML} stab type, and there is a corresponding @code{N_ECOML}
978 stab type, but Sun's Fortran compiler uses @code{N_GSYM} instead. The
979 variables within a common block use the @samp{V} symbol descriptor (I
980 believe this is true of all Fortran variables). Other stabs (at least
981 type declarations using @code{C_DECL}) can also be between the
982 @code{N_BCOMM} and the @code{N_ECOMM}.
985 @section Static Variables
987 Initialized static variables are represented by the @samp{S} and
988 @samp{V} symbol descriptors. @samp{S} means file scope static, and
989 @samp{V} means procedure scope static. One exception: in XCOFF, IBM's
990 xlc compiler always uses @samp{V}, and whether it is file scope or not
991 is distinguished by whether the stab is located within a function.
993 @c This is probably not worth mentioning; it is only true on the sparc
994 @c for `double' variables which although declared const are actually in
995 @c the data segment (the text segment can't guarantee 8 byte alignment).
997 @c 2.4.5 has a bug in that it uses @code{N_FUN}, so neither dbx nor GDB can
998 @c find the variables)
1001 @findex N_FUN, for variables
1003 In a.out files, @code{N_STSYM} means the data section, @code{N_FUN}
1004 means the text section, and @code{N_LCSYM} means the bss section. For
1005 those systems with a read-only data section separate from the text
1006 section (Solaris), @code{N_ROSYM} means the read-only data section.
1008 For example, the source lines:
1011 static const int var_const = 5;
1012 static int var_init = 2;
1013 static int var_noinit;
1017 yield the following stabs:
1020 .stabs "var_const:S1",36,0,0,_var_const # @r{36 is N_FUN}
1022 .stabs "var_init:S1",38,0,0,_var_init # @r{38 is N_STSYM}
1024 .stabs "var_noinit:S1",40,0,0,_var_noinit # @r{40 is N_LCSYM}
1030 In XCOFF files, the stab type need not indicate the section;
1031 @code{C_STSYM} can be used for all statics. Also, each static variable
1032 is enclosed in a static block. A @code{C_BSTAT} (emitted with a
1033 @samp{.bs} assembler directive) symbol begins the static block; its
1034 value is the symbol number of the csect symbol whose value is the
1035 address of the static block, its section is the section of the variables
1036 in that static block, and its name is @samp{.bs}. A @code{C_ESTAT}
1037 (emitted with a @samp{.es} assembler directive) symbol ends the static
1038 block; its name is @samp{.es} and its value and section are ignored.
1040 In ECOFF files, the storage class is used to specify the section, so the
1041 stab type need not indicate the section.
1043 In ELF files, for the SunPRO compiler version 2.0.1, symbol descriptor
1044 @samp{S} means that the address is absolute (the linker relocates it)
1045 and symbol descriptor @samp{V} means that the address is relative to the
1046 start of the relevant section for that compilation unit. SunPRO has
1047 plans to have the linker stop relocating stabs; I suspect that their the
1048 debugger gets the address from the corresponding ELF (not stab) symbol.
1049 I'm not sure how to find which symbol of that name is the right one.
1050 The clean way to do all this would be to have a the value of a symbol
1051 descriptor @samp{S} symbol be an offset relative to the start of the
1052 file, just like everything else, but that introduces obvious
1053 compatibility problems. For more information on linker stab relocation,
1054 @xref{ELF Linker Relocation}.
1056 @node Based Variables
1057 @section Fortran Based Variables
1059 Fortran (at least, the Sun and SGI dialects of FORTRAN-77) has a feature
1060 which allows allocating arrays with @code{malloc}, but which avoids
1061 blurring the line between arrays and pointers the way that C does. In
1062 stabs such a variable uses the @samp{b} symbol descriptor.
1064 For example, the Fortran declarations
1067 real foo, foo10(10), foo10_5(10,5)
1069 pointer (foo10p, foo10)
1070 pointer (foo105p, foo10_5)
1078 foo10_5:bar3;1;5;ar3;1;10;6
1081 In this example, @code{real} is type 6 and type 3 is an integral type
1082 which is the type of the subscripts of the array (probably
1085 The @samp{b} symbol descriptor is like @samp{V} in that it denotes a
1086 statically allocated symbol whose scope is local to a function; see
1087 @xref{Statics}. The value of the symbol, instead of being the address
1088 of the variable itself, is the address of a pointer to that variable.
1089 So in the above example, the value of the @code{foo} stab is the address
1090 of a pointer to a real, the value of the @code{foo10} stab is the
1091 address of a pointer to a 10-element array of reals, and the value of
1092 the @code{foo10_5} stab is the address of a pointer to a 5-element array
1093 of 10-element arrays of reals.
1098 Formal parameters to a function are represented by a stab (or sometimes
1099 two; see below) for each parameter. The stabs are in the order in which
1100 the debugger should print the parameters (i.e., the order in which the
1101 parameters are declared in the source file). The exact form of the stab
1102 depends on how the parameter is being passed.
1106 Parameters passed on the stack use the symbol descriptor @samp{p} and
1107 the @code{N_PSYM} symbol type (or @code{C_PSYM} for XCOFF). The value
1108 of the symbol is an offset used to locate the parameter on the stack;
1109 its exact meaning is machine-dependent, but on most machines it is an
1110 offset from the frame pointer.
1112 As a simple example, the code:
1123 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
1124 .stabs "argc:p1",160,0,0,68 # @r{160 is N_PSYM}
1125 .stabs "argv:p20=*21=*2",160,0,0,72
1128 The type definition of @code{argv} is interesting because it contains
1129 several type definitions. Type 21 is pointer to type 2 (char) and
1130 @code{argv} (type 20) is pointer to type 21.
1132 @c FIXME: figure out what these mean and describe them coherently.
1133 The following symbol descriptors are also said to go with @code{N_PSYM}.
1134 The value of the symbol is said to be an offset from the argument
1135 pointer (I'm not sure whether this is true or not).
1139 pF Fortran function parameter
1140 X (function result variable)
1144 * Register Parameters::
1145 * Local Variable Parameters::
1146 * Reference Parameters::
1147 * Conformant Arrays::
1150 @node Register Parameters
1151 @subsection Passing Parameters in Registers
1153 If the parameter is passed in a register, then traditionally there are
1154 two symbols for each argument:
1157 .stabs "arg:p1" . . . ; N_PSYM
1158 .stabs "arg:r1" . . . ; N_RSYM
1161 Debuggers use the second one to find the value, and the first one to
1162 know that it is an argument.
1165 @findex N_RSYM, for parameters
1166 Because that approach is kind of ugly, some compilers use symbol
1167 descriptor @samp{P} or @samp{R} to indicate an argument which is in a
1168 register. Symbol type @code{C_RPSYM} is used in XCOFF and @code{N_RSYM}
1169 is used otherwise. The symbol's value is the register number. @samp{P}
1170 and @samp{R} mean the same thing; the difference is that @samp{P} is a
1171 GNU invention and @samp{R} is an IBM (XCOFF) invention. As of version
1172 4.9, GDB should handle either one.
1174 There is at least one case where GCC uses a @samp{p} and @samp{r} pair
1175 rather than @samp{P}; this is where the argument is passed in the
1176 argument list and then loaded into a register.
1178 According to the AIX documentation, symbol descriptor @samp{D} is for a
1179 parameter passed in a floating point register. This seems
1180 unnecessary---why not just use @samp{R} with a register number which
1181 indicates that it's a floating point register? I haven't verified
1182 whether the system actually does what the documentation indicates.
1184 @c FIXME: On the hppa this is for any type > 8 bytes, I think, and not
1185 @c for small structures (investigate).
1186 On the sparc and hppa, for a @samp{P} symbol whose type is a structure
1187 or union, the register contains the address of the structure. On the
1188 sparc, this is also true of a @samp{p} and @samp{r} pair (using Sun
1189 @code{cc}) or a @samp{p} symbol. However, if a (small) structure is
1190 really in a register, @samp{r} is used. And, to top it all off, on the
1191 hppa it might be a structure which was passed on the stack and loaded
1192 into a register and for which there is a @samp{p} and @samp{r} pair! I
1193 believe that symbol descriptor @samp{i} is supposed to deal with this
1194 case (it is said to mean "value parameter by reference, indirect
1195 access"; I don't know the source for this information), but I don't know
1196 details or what compilers or debuggers use it, if any (not GDB or GCC).
1197 It is not clear to me whether this case needs to be dealt with
1198 differently than parameters passed by reference (@pxref{Reference Parameters}).
1200 @node Local Variable Parameters
1201 @subsection Storing Parameters as Local Variables
1203 There is a case similar to an argument in a register, which is an
1204 argument that is actually stored as a local variable. Sometimes this
1205 happens when the argument was passed in a register and then the compiler
1206 stores it as a local variable. If possible, the compiler should claim
1207 that it's in a register, but this isn't always done.
1209 If a parameter is passed as one type and converted to a smaller type by
1210 the prologue (for example, the parameter is declared as a @code{float},
1211 but the calling conventions specify that it is passed as a
1212 @code{double}), then GCC2 (sometimes) uses a pair of symbols. The first
1213 symbol uses symbol descriptor @samp{p} and the type which is passed.
1214 The second symbol has the type and location which the parameter actually
1215 has after the prologue. For example, suppose the following C code
1216 appears with no prototypes involved:
1225 if @code{f} is passed as a double at stack offset 8, and the prologue
1226 converts it to a float in register number 0, then the stabs look like:
1229 .stabs "f:p13",160,0,3,8 # @r{160 is @code{N_PSYM}, here 13 is @code{double}}
1230 .stabs "f:r12",64,0,3,0 # @r{64 is @code{N_RSYM}, here 12 is @code{float}}
1233 In both stabs 3 is the line number where @code{f} is declared
1234 (@pxref{Line Numbers}).
1236 @findex N_LSYM, for parameter
1237 GCC, at least on the 960, has another solution to the same problem. It
1238 uses a single @samp{p} symbol descriptor for an argument which is stored
1239 as a local variable but uses @code{N_LSYM} instead of @code{N_PSYM}. In
1240 this case, the value of the symbol is an offset relative to the local
1241 variables for that function, not relative to the arguments; on some
1242 machines those are the same thing, but not on all.
1244 @c This is mostly just background info; the part that logically belongs
1245 @c here is the last sentence.
1246 On the VAX or on other machines in which the calling convention includes
1247 the number of words of arguments actually passed, the debugger (GDB at
1248 least) uses the parameter symbols to keep track of whether it needs to
1249 print nameless arguments in addition to the formal parameters which it
1250 has printed because each one has a stab. For example, in
1253 extern int fprintf (FILE *stream, char *format, @dots{});
1255 fprintf (stdout, "%d\n", x);
1258 there are stabs for @code{stream} and @code{format}. On most machines,
1259 the debugger can only print those two arguments (because it has no way
1260 of knowing that additional arguments were passed), but on the VAX or
1261 other machines with a calling convention which indicates the number of
1262 words of arguments, the debugger can print all three arguments. To do
1263 so, the parameter symbol (symbol descriptor @samp{p}) (not necessarily
1264 @samp{r} or symbol descriptor omitted symbols) needs to contain the
1265 actual type as passed (for example, @code{double} not @code{float} if it
1266 is passed as a double and converted to a float).
1268 @node Reference Parameters
1269 @subsection Passing Parameters by Reference
1271 If the parameter is passed by reference (e.g., Pascal @code{VAR}
1272 parameters), then the symbol descriptor is @samp{v} if it is in the
1273 argument list, or @samp{a} if it in a register. Other than the fact
1274 that these contain the address of the parameter rather than the
1275 parameter itself, they are identical to @samp{p} and @samp{R},
1276 respectively. I believe @samp{a} is an AIX invention; @samp{v} is
1277 supported by all stabs-using systems as far as I know.
1279 @node Conformant Arrays
1280 @subsection Passing Conformant Array Parameters
1282 @c Is this paragraph correct? It is based on piecing together patchy
1283 @c information and some guesswork
1284 Conformant arrays are a feature of Modula-2, and perhaps other
1285 languages, in which the size of an array parameter is not known to the
1286 called function until run-time. Such parameters have two stabs: a
1287 @samp{x} for the array itself, and a @samp{C}, which represents the size
1288 of the array. The value of the @samp{x} stab is the offset in the
1289 argument list where the address of the array is stored (it this right?
1290 it is a guess); the value of the @samp{C} stab is the offset in the
1291 argument list where the size of the array (in elements? in bytes?) is
1295 @chapter Defining Types
1297 The examples so far have described types as references to previously
1298 defined types, or defined in terms of subranges of or pointers to
1299 previously defined types. This chapter describes the other type
1300 descriptors that may follow the @samp{=} in a type definition.
1303 * Builtin Types:: Integers, floating point, void, etc.
1304 * Miscellaneous Types:: Pointers, sets, files, etc.
1305 * Cross-References:: Referring to a type not yet defined.
1306 * Subranges:: A type with a specific range.
1307 * Arrays:: An aggregate type of same-typed elements.
1308 * Strings:: Like an array but also has a length.
1309 * Enumerations:: Like an integer but the values have names.
1310 * Structures:: An aggregate type of different-typed elements.
1311 * Typedefs:: Giving a type a name.
1312 * Unions:: Different types sharing storage.
1317 @section Builtin Types
1319 Certain types are built in (@code{int}, @code{short}, @code{void},
1320 @code{float}, etc.); the debugger recognizes these types and knows how
1321 to handle them. Thus, don't be surprised if some of the following ways
1322 of specifying builtin types do not specify everything that a debugger
1323 would need to know about the type---in some cases they merely specify
1324 enough information to distinguish the type from other types.
1326 The traditional way to define builtin types is convoluted, so new ways
1327 have been invented to describe them. Sun's @code{acc} uses special
1328 builtin type descriptors (@samp{b} and @samp{R}), and IBM uses negative
1329 type numbers. GDB accepts all three ways, as of version 4.8; dbx just
1330 accepts the traditional builtin types and perhaps one of the other two
1331 formats. The following sections describe each of these formats.
1334 * Traditional Builtin Types:: Put on your seat belts and prepare for kludgery
1335 * Builtin Type Descriptors:: Builtin types with special type descriptors
1336 * Negative Type Numbers:: Builtin types using negative type numbers
1339 @node Traditional Builtin Types
1340 @subsection Traditional Builtin Types
1342 This is the traditional, convoluted method for defining builtin types.
1343 There are several classes of such type definitions: integer, floating
1344 point, and @code{void}.
1347 * Traditional Integer Types::
1348 * Traditional Other Types::
1351 @node Traditional Integer Types
1352 @subsubsection Traditional Integer Types
1354 Often types are defined as subranges of themselves. If the bounding values
1355 fit within an @code{int}, then they are given normally. For example:
1358 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0 # @r{128 is N_LSYM}
1359 .stabs "char:t2=r2;0;127;",128,0,0,0
1362 Builtin types can also be described as subranges of @code{int}:
1365 .stabs "unsigned short:t6=r1;0;65535;",128,0,0,0
1368 If the lower bound of a subrange is 0 and the upper bound is -1,
1369 the type is an unsigned integral type whose bounds are too
1370 big to describe in an @code{int}. Traditionally this is only used for
1371 @code{unsigned int} and @code{unsigned long}:
1374 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
1377 For larger types, GCC 2.4.5 puts out bounds in octal, with one or more
1378 leading zeroes. In this case a negative bound consists of a number
1379 which is a 1 bit (for the sign bit) followed by a 0 bit for each bit in
1380 the number (except the sign bit), and a positive bound is one which is a
1381 1 bit for each bit in the number (except possibly the sign bit). All
1382 known versions of dbx and GDB version 4 accept this (at least in the
1383 sense of not refusing to process the file), but GDB 3.5 refuses to read
1384 the whole file containing such symbols. So GCC 2.3.3 did not output the
1385 proper size for these types. As an example of octal bounds, the string
1386 fields of the stabs for 64 bit integer types look like:
1388 @c .stabs directives, etc., omitted to make it fit on the page.
1390 long int:t3=r1;001000000000000000000000;000777777777777777777777;
1391 long unsigned int:t5=r1;000000000000000000000000;001777777777777777777777;
1394 If the lower bound of a subrange is 0 and the upper bound is negative,
1395 the type is an unsigned integral type whose size in bytes is the
1396 absolute value of the upper bound. I believe this is a Convex
1397 convention for @code{unsigned long long}.
1399 If the lower bound of a subrange is negative and the upper bound is 0,
1400 the type is a signed integral type whose size in bytes is
1401 the absolute value of the lower bound. I believe this is a Convex
1402 convention for @code{long long}. To distinguish this from a legitimate
1403 subrange, the type should be a subrange of itself. I'm not sure whether
1404 this is the case for Convex.
1406 @node Traditional Other Types
1407 @subsubsection Traditional Other Types
1409 If the upper bound of a subrange is 0 and the lower bound is positive,
1410 the type is a floating point type, and the lower bound of the subrange
1411 indicates the number of bytes in the type:
1414 .stabs "float:t12=r1;4;0;",128,0,0,0
1415 .stabs "double:t13=r1;8;0;",128,0,0,0
1418 However, GCC writes @code{long double} the same way it writes
1419 @code{double}, so there is no way to distinguish.
1422 .stabs "long double:t14=r1;8;0;",128,0,0,0
1425 Complex types are defined the same way as floating-point types; there is
1426 no way to distinguish a single-precision complex from a double-precision
1427 floating-point type.
1429 The C @code{void} type is defined as itself:
1432 .stabs "void:t15=15",128,0,0,0
1435 I'm not sure how a boolean type is represented.
1437 @node Builtin Type Descriptors
1438 @subsection Defining Builtin Types Using Builtin Type Descriptors
1440 This is the method used by Sun's @code{acc} for defining builtin types.
1441 These are the type descriptors to define builtin types:
1444 @c FIXME: clean up description of width and offset, once we figure out
1446 @item b @var{signed} @var{char-flag} @var{width} ; @var{offset} ; @var{nbits} ;
1447 Define an integral type. @var{signed} is @samp{u} for unsigned or
1448 @samp{s} for signed. @var{char-flag} is @samp{c} which indicates this
1449 is a character type, or is omitted. I assume this is to distinguish an
1450 integral type from a character type of the same size, for example it
1451 might make sense to set it for the C type @code{wchar_t} so the debugger
1452 can print such variables differently (Solaris does not do this). Sun
1453 sets it on the C types @code{signed char} and @code{unsigned char} which
1454 arguably is wrong. @var{width} and @var{offset} appear to be for small
1455 objects stored in larger ones, for example a @code{short} in an
1456 @code{int} register. @var{width} is normally the number of bytes in the
1457 type. @var{offset} seems to always be zero. @var{nbits} is the number
1458 of bits in the type.
1460 Note that type descriptor @samp{b} used for builtin types conflicts with
1461 its use for Pascal space types (@pxref{Miscellaneous Types}); they can
1462 be distinguished because the character following the type descriptor
1463 will be a digit, @samp{(}, or @samp{-} for a Pascal space type, or
1464 @samp{u} or @samp{s} for a builtin type.
1467 Documented by AIX to define a wide character type, but their compiler
1468 actually uses negative type numbers (@pxref{Negative Type Numbers}).
1470 @item R @var{fp-type} ; @var{bytes} ;
1471 Define a floating point type. @var{fp-type} has one of the following values:
1475 IEEE 32-bit (single precision) floating point format.
1478 IEEE 64-bit (double precision) floating point format.
1480 @item 3 (NF_COMPLEX)
1481 @item 4 (NF_COMPLEX16)
1482 @item 5 (NF_COMPLEX32)
1483 @c "GDB source" really means @file{include/aout/stab_gnu.h}, but trying
1484 @c to put that here got an overfull hbox.
1485 These are for complex numbers. A comment in the GDB source describes
1486 them as Fortran @code{complex}, @code{double complex}, and
1487 @code{complex*16}, respectively, but what does that mean? (i.e., Single
1488 precision? Double precision?).
1490 @item 6 (NF_LDOUBLE)
1491 Long double. This should probably only be used for Sun format
1492 @code{long double}, and new codes should be used for other floating
1493 point formats (@code{NF_DOUBLE} can be used if a @code{long double} is
1494 really just an IEEE double, of course).
1497 @var{bytes} is the number of bytes occupied by the type. This allows a
1498 debugger to perform some operations with the type even if it doesn't
1499 understand @var{fp-type}.
1501 @item g @var{type-information} ; @var{nbits}
1502 Documented by AIX to define a floating type, but their compiler actually
1503 uses negative type numbers (@pxref{Negative Type Numbers}).
1505 @item c @var{type-information} ; @var{nbits}
1506 Documented by AIX to define a complex type, but their compiler actually
1507 uses negative type numbers (@pxref{Negative Type Numbers}).
1510 The C @code{void} type is defined as a signed integral type 0 bits long:
1512 .stabs "void:t19=bs0;0;0",128,0,0,0
1514 The Solaris compiler seems to omit the trailing semicolon in this case.
1515 Getting sloppy in this way is not a swift move because if a type is
1516 embedded in a more complex expression it is necessary to be able to tell
1519 I'm not sure how a boolean type is represented.
1521 @node Negative Type Numbers
1522 @subsection Negative Type Numbers
1524 This is the method used in XCOFF for defining builtin types.
1525 Since the debugger knows about the builtin types anyway, the idea of
1526 negative type numbers is simply to give a special type number which
1527 indicates the builtin type. There is no stab defining these types.
1529 There are several subtle issues with negative type numbers.
1531 One is the size of the type. A builtin type (for example the C types
1532 @code{int} or @code{long}) might have different sizes depending on
1533 compiler options, the target architecture, the ABI, etc. This issue
1534 doesn't come up for IBM tools since (so far) they just target the
1535 RS/6000; the sizes indicated below for each size are what the IBM
1536 RS/6000 tools use. To deal with differing sizes, either define separate
1537 negative type numbers for each size (which works but requires changing
1538 the debugger, and, unless you get both AIX dbx and GDB to accept the
1539 change, introduces an incompatibility), or use a type attribute
1540 (@pxref{String Field}) to define a new type with the appropriate size
1541 (which merely requires a debugger which understands type attributes,
1542 like AIX dbx or GDB). For example,
1545 .stabs "boolean:t10=@@s8;-16",128,0,0,0
1548 defines an 8-bit boolean type, and
1551 .stabs "boolean:t10=@@s64;-16",128,0,0,0
1554 defines a 64-bit boolean type.
1556 A similar issue is the format of the type. This comes up most often for
1557 floating-point types, which could have various formats (particularly
1558 extended doubles, which vary quite a bit even among IEEE systems).
1559 Again, it is best to define a new negative type number for each
1560 different format; changing the format based on the target system has
1561 various problems. One such problem is that the Alpha has both VAX and
1562 IEEE floating types. One can easily imagine one library using the VAX
1563 types and another library in the same executable using the IEEE types.
1564 Another example is that the interpretation of whether a boolean is true
1565 or false can be based on the least significant bit, most significant
1566 bit, whether it is zero, etc., and different compilers (or different
1567 options to the same compiler) might provide different kinds of boolean.
1569 The last major issue is the names of the types. The name of a given
1570 type depends @emph{only} on the negative type number given; these do not
1571 vary depending on the language, the target system, or anything else.
1572 One can always define separate type numbers---in the following list you
1573 will see for example separate @code{int} and @code{integer*4} types
1574 which are identical except for the name. But compatibility can be
1575 maintained by not inventing new negative type numbers and instead just
1576 defining a new type with a new name. For example:
1579 .stabs "CARDINAL:t10=-8",128,0,0,0
1582 Here is the list of negative type numbers. The phrase @dfn{integral
1583 type} is used to mean twos-complement (I strongly suspect that all
1584 machines which use stabs use twos-complement; most machines use
1585 twos-complement these days).
1589 @code{int}, 32 bit signed integral type.
1592 @code{char}, 8 bit type holding a character. Both GDB and dbx on AIX
1593 treat this as signed. GCC uses this type whether @code{char} is signed
1594 or not, which seems like a bad idea. The AIX compiler (@code{xlc}) seems to
1595 avoid this type; it uses -5 instead for @code{char}.
1598 @code{short}, 16 bit signed integral type.
1601 @code{long}, 32 bit signed integral type.
1604 @code{unsigned char}, 8 bit unsigned integral type.
1607 @code{signed char}, 8 bit signed integral type.
1610 @code{unsigned short}, 16 bit unsigned integral type.
1613 @code{unsigned int}, 32 bit unsigned integral type.
1616 @code{unsigned}, 32 bit unsigned integral type.
1619 @code{unsigned long}, 32 bit unsigned integral type.
1622 @code{void}, type indicating the lack of a value.
1625 @code{float}, IEEE single precision.
1628 @code{double}, IEEE double precision.
1631 @code{long double}, IEEE double precision. The compiler claims the size
1632 will increase in a future release, and for binary compatibility you have
1633 to avoid using @code{long double}. I hope when they increase it they
1634 use a new negative type number.
1637 @code{integer}. 32 bit signed integral type.
1640 @code{boolean}. 32 bit type. GDB and GCC assume that zero is false,
1641 one is true, and other values have unspecified meaning. I hope this
1642 agrees with how the IBM tools use the type.
1645 @code{short real}. IEEE single precision.
1648 @code{real}. IEEE double precision.
1651 @code{stringptr}. @xref{Strings}.
1654 @code{character}, 8 bit unsigned character type.
1657 @code{logical*1}, 8 bit type. This Fortran type has a split
1658 personality in that it is used for boolean variables, but can also be
1659 used for unsigned integers. 0 is false, 1 is true, and other values are
1663 @code{logical*2}, 16 bit type. This Fortran type has a split
1664 personality in that it is used for boolean variables, but can also be
1665 used for unsigned integers. 0 is false, 1 is true, and other values are
1669 @code{logical*4}, 32 bit type. This Fortran type has a split
1670 personality in that it is used for boolean variables, but can also be
1671 used for unsigned integers. 0 is false, 1 is true, and other values are
1675 @code{logical}, 32 bit type. This Fortran type has a split
1676 personality in that it is used for boolean variables, but can also be
1677 used for unsigned integers. 0 is false, 1 is true, and other values are
1681 @code{complex}. A complex type consisting of two IEEE single-precision
1682 floating point values.
1685 @code{complex}. A complex type consisting of two IEEE double-precision
1686 floating point values.
1689 @code{integer*1}, 8 bit signed integral type.
1692 @code{integer*2}, 16 bit signed integral type.
1695 @code{integer*4}, 32 bit signed integral type.
1698 @code{wchar}. Wide character, 16 bits wide, unsigned (what format?
1702 @code{long long}, 64 bit signed integral type.
1705 @code{unsigned long long}, 64 bit unsigned integral type.
1708 @code{logical*8}, 64 bit unsigned integral type.
1711 @code{integer*8}, 64 bit signed integral type.
1714 @node Miscellaneous Types
1715 @section Miscellaneous Types
1718 @item b @var{type-information} ; @var{bytes}
1719 Pascal space type. This is documented by IBM; what does it mean?
1721 This use of the @samp{b} type descriptor can be distinguished
1722 from its use for builtin integral types (@pxref{Builtin Type
1723 Descriptors}) because the character following the type descriptor is
1724 always a digit, @samp{(}, or @samp{-}.
1726 @item B @var{type-information}
1727 A volatile-qualified version of @var{type-information}. This is
1728 a Sun extension. References and stores to a variable with a
1729 volatile-qualified type must not be optimized or cached; they
1730 must occur as the user specifies them.
1732 @item d @var{type-information}
1733 File of type @var{type-information}. As far as I know this is only used
1736 @item k @var{type-information}
1737 A const-qualified version of @var{type-information}. This is a Sun
1738 extension. A variable with a const-qualified type cannot be modified.
1740 @item M @var{type-information} ; @var{length}
1741 Multiple instance type. The type seems to composed of @var{length}
1742 repetitions of @var{type-information}, for example @code{character*3} is
1743 represented by @samp{M-2;3}, where @samp{-2} is a reference to a
1744 character type (@pxref{Negative Type Numbers}). I'm not sure how this
1745 differs from an array. This appears to be a Fortran feature.
1746 @var{length} is a bound, like those in range types; see @ref{Subranges}.
1748 @item S @var{type-information}
1749 Pascal set type. @var{type-information} must be a small type such as an
1750 enumeration or a subrange, and the type is a bitmask whose length is
1751 specified by the number of elements in @var{type-information}.
1753 In CHILL, if it is a bitstring instead of a set, also use the @samp{S}
1754 type attribute (@pxref{String Field}).
1756 @item * @var{type-information}
1757 Pointer to @var{type-information}.
1760 @node Cross-References
1761 @section Cross-References to Other Types
1763 A type can be used before it is defined; one common way to deal with
1764 that situation is just to use a type reference to a type which has not
1767 Another way is with the @samp{x} type descriptor, which is followed by
1768 @samp{s} for a structure tag, @samp{u} for a union tag, or @samp{e} for
1769 a enumerator tag, followed by the name of the tag, followed by @samp{:}.
1770 If the name contains @samp{::} between a @samp{<} and @samp{>} pair (for
1771 C++ templates), such a @samp{::} does not end the name---only a single
1772 @samp{:} ends the name; see @ref{Nested Symbols}.
1774 For example, the following C declarations:
1785 .stabs "bar:G16=*17=xsfoo:",32,0,0,0
1788 Not all debuggers support the @samp{x} type descriptor, so on some
1789 machines GCC does not use it. I believe that for the above example it
1790 would just emit a reference to type 17 and never define it, but I
1791 haven't verified that.
1793 Modula-2 imported types, at least on AIX, use the @samp{i} type
1794 descriptor, which is followed by the name of the module from which the
1795 type is imported, followed by @samp{:}, followed by the name of the
1796 type. There is then optionally a comma followed by type information for
1797 the type. This differs from merely naming the type (@pxref{Typedefs}) in
1798 that it identifies the module; I don't understand whether the name of
1799 the type given here is always just the same as the name we are giving
1800 it, or whether this type descriptor is used with a nameless stab
1801 (@pxref{String Field}), or what. The symbol ends with @samp{;}.
1804 @section Subrange Types
1806 The @samp{r} type descriptor defines a type as a subrange of another
1807 type. It is followed by type information for the type of which it is a
1808 subrange, a semicolon, an integral lower bound, a semicolon, an
1809 integral upper bound, and a semicolon. The AIX documentation does not
1810 specify the trailing semicolon, in an effort to specify array indexes
1811 more cleanly, but a subrange which is not an array index has always
1812 included a trailing semicolon (@pxref{Arrays}).
1814 Instead of an integer, either bound can be one of the following:
1817 @item A @var{offset}
1818 The bound is passed by reference on the stack at offset @var{offset}
1819 from the argument list. @xref{Parameters}, for more information on such
1822 @item T @var{offset}
1823 The bound is passed by value on the stack at offset @var{offset} from
1826 @item a @var{register-number}
1827 The bound is passed by reference in register number
1828 @var{register-number}.
1830 @item t @var{register-number}
1831 The bound is passed by value in register number @var{register-number}.
1837 Subranges are also used for builtin types; see @ref{Traditional Builtin Types}.
1840 @section Array Types
1842 Arrays use the @samp{a} type descriptor. Following the type descriptor
1843 is the type of the index and the type of the array elements. If the
1844 index type is a range type, it ends in a semicolon; otherwise
1845 (for example, if it is a type reference), there does not
1846 appear to be any way to tell where the types are separated. In an
1847 effort to clean up this mess, IBM documents the two types as being
1848 separated by a semicolon, and a range type as not ending in a semicolon
1849 (but this is not right for range types which are not array indexes,
1850 @pxref{Subranges}). I think probably the best solution is to specify
1851 that a semicolon ends a range type, and that the index type and element
1852 type of an array are separated by a semicolon, but that if the index
1853 type is a range type, the extra semicolon can be omitted. GDB (at least
1854 through version 4.9) doesn't support any kind of index type other than a
1855 range anyway; I'm not sure about dbx.
1857 It is well established, and widely used, that the type of the index,
1858 unlike most types found in the stabs, is merely a type definition, not
1859 type information (@pxref{String Field}) (that is, it need not start with
1860 @samp{@var{type-number}=} if it is defining a new type). According to a
1861 comment in GDB, this is also true of the type of the array elements; it
1862 gives @samp{ar1;1;10;ar1;1;10;4} as a legitimate way to express a two
1863 dimensional array. According to AIX documentation, the element type
1864 must be type information. GDB accepts either.
1866 The type of the index is often a range type, expressed as the type
1867 descriptor @samp{r} and some parameters. It defines the size of the
1868 array. In the example below, the range @samp{r1;0;2;} defines an index
1869 type which is a subrange of type 1 (integer), with a lower bound of 0
1870 and an upper bound of 2. This defines the valid range of subscripts of
1871 a three-element C array.
1873 For example, the definition:
1876 char char_vec[3] = @{'a','b','c'@};
1880 produces the output:
1883 .stabs "char_vec:G19=ar1;0;2;2",32,0,0,0
1892 If an array is @dfn{packed}, the elements are spaced more
1893 closely than normal, saving memory at the expense of speed. For
1894 example, an array of 3-byte objects might, if unpacked, have each
1895 element aligned on a 4-byte boundary, but if packed, have no padding.
1896 One way to specify that something is packed is with type attributes
1897 (@pxref{String Field}). In the case of arrays, another is to use the
1898 @samp{P} type descriptor instead of @samp{a}. Other than specifying a
1899 packed array, @samp{P} is identical to @samp{a}.
1901 @c FIXME-what is it? A pointer?
1902 An open array is represented by the @samp{A} type descriptor followed by
1903 type information specifying the type of the array elements.
1905 @c FIXME: what is the format of this type? A pointer to a vector of pointers?
1906 An N-dimensional dynamic array is represented by
1909 D @var{dimensions} ; @var{type-information}
1912 @c Does dimensions really have this meaning? The AIX documentation
1914 @var{dimensions} is the number of dimensions; @var{type-information}
1915 specifies the type of the array elements.
1917 @c FIXME: what is the format of this type? A pointer to some offsets in
1919 A subarray of an N-dimensional array is represented by
1922 E @var{dimensions} ; @var{type-information}
1925 @c Does dimensions really have this meaning? The AIX documentation
1927 @var{dimensions} is the number of dimensions; @var{type-information}
1928 specifies the type of the array elements.
1933 Some languages, like C or the original Pascal, do not have string types,
1934 they just have related things like arrays of characters. But most
1935 Pascals and various other languages have string types, which are
1936 indicated as follows:
1939 @item n @var{type-information} ; @var{bytes}
1940 @var{bytes} is the maximum length. I'm not sure what
1941 @var{type-information} is; I suspect that it means that this is a string
1942 of @var{type-information} (thus allowing a string of integers, a string
1943 of wide characters, etc., as well as a string of characters). Not sure
1944 what the format of this type is. This is an AIX feature.
1946 @item z @var{type-information} ; @var{bytes}
1947 Just like @samp{n} except that this is a gstring, not an ordinary
1948 string. I don't know the difference.
1951 Pascal Stringptr. What is this? This is an AIX feature.
1954 Languages, such as CHILL which have a string type which is basically
1955 just an array of characters use the @samp{S} type attribute
1956 (@pxref{String Field}).
1959 @section Enumerations
1961 Enumerations are defined with the @samp{e} type descriptor.
1963 @c FIXME: Where does this information properly go? Perhaps it is
1964 @c redundant with something we already explain.
1965 The source line below declares an enumeration type at file scope.
1966 The type definition is located after the @code{N_RBRAC} that marks the end of
1967 the previous procedure's block scope, and before the @code{N_FUN} that marks
1968 the beginning of the next procedure's block scope. Therefore it does not
1969 describe a block local symbol, but a file local one.
1974 enum e_places @{first,second=3,last@};
1978 generates the following stab:
1981 .stabs "e_places:T22=efirst:0,second:3,last:4,;",128,0,0,0
1984 The symbol descriptor (@samp{T}) says that the stab describes a
1985 structure, enumeration, or union tag. The type descriptor @samp{e},
1986 following the @samp{22=} of the type definition narrows it down to an
1987 enumeration type. Following the @samp{e} is a list of the elements of
1988 the enumeration. The format is @samp{@var{name}:@var{value},}. The
1989 list of elements ends with @samp{;}. The fact that @var{value} is
1990 specified as an integer can cause problems if the value is large. GCC
1991 2.5.2 tries to output it in octal in that case with a leading zero,
1992 which is probably a good thing, although GDB 4.11 supports octal only in
1993 cases where decimal is perfectly good. Negative decimal values are
1994 supported by both GDB and dbx.
1996 There is no standard way to specify the size of an enumeration type; it
1997 is determined by the architecture (normally all enumerations types are
1998 32 bits). Type attributes can be used to specify an enumeration type of
1999 another size for debuggers which support them; see @ref{String Field}.
2001 Enumeration types are unusual in that they define symbols for the
2002 enumeration values (@code{first}, @code{second}, and @code{third} in the
2003 above example), and even though these symbols are visible in the file as
2004 a whole (rather than being in a more local namespace like structure
2005 member names), they are defined in the type definition for the
2006 enumeration type rather than each having their own symbol. In order to
2007 be fast, GDB will only get symbols from such types (in its initial scan
2008 of the stabs) if the type is the first thing defined after a @samp{T} or
2009 @samp{t} symbol descriptor (the above example fulfills this
2010 requirement). If the type does not have a name, the compiler should
2011 emit it in a nameless stab (@pxref{String Field}); GCC does this.
2016 The encoding of structures in stabs can be shown with an example.
2018 The following source code declares a structure tag and defines an
2019 instance of the structure in global scope. Then a @code{typedef} equates the
2020 structure tag with a new type. Separate stabs are generated for the
2021 structure tag, the structure @code{typedef}, and the structure instance. The
2022 stabs for the tag and the @code{typedef} are emitted when the definitions are
2023 encountered. Since the structure elements are not initialized, the
2024 stab and code for the structure variable itself is located at the end
2025 of the program in the bss section.
2032 struct s_tag* s_next;
2035 typedef struct s_tag s_typedef;
2038 The structure tag has an @code{N_LSYM} stab type because, like the
2039 enumeration, the symbol has file scope. Like the enumeration, the
2040 symbol descriptor is @samp{T}, for enumeration, structure, or tag type.
2041 The type descriptor @samp{s} following the @samp{16=} of the type
2042 definition narrows the symbol type to structure.
2044 Following the @samp{s} type descriptor is the number of bytes the
2045 structure occupies, followed by a description of each structure element.
2046 The structure element descriptions are of the form @var{name:type, bit
2047 offset from the start of the struct, number of bits in the element}.
2049 @c FIXME: phony line break. Can probably be fixed by using an example
2050 @c with fewer fields.
2053 .stabs "s_tag:T16=s20s_int:1,0,32;s_float:12,32,32;
2054 s_char_vec:17=ar1;0;7;2,64,64;s_next:18=*16,128,32;;",128,0,0,0
2057 In this example, the first two structure elements are previously defined
2058 types. For these, the type following the @samp{@var{name}:} part of the
2059 element description is a simple type reference. The other two structure
2060 elements are new types. In this case there is a type definition
2061 embedded after the @samp{@var{name}:}. The type definition for the
2062 array element looks just like a type definition for a stand-alone array.
2063 The @code{s_next} field is a pointer to the same kind of structure that
2064 the field is an element of. So the definition of structure type 16
2065 contains a type definition for an element which is a pointer to type 16.
2067 If a field is a static member (this is a C++ feature in which a single
2068 variable appears to be a field of every structure of a given type) it
2069 still starts out with the field name, a colon, and the type, but then
2070 instead of a comma, bit position, comma, and bit size, there is a colon
2071 followed by the name of the variable which each such field refers to.
2073 If the structure has methods (a C++ feature), they follow the non-method
2074 fields; see @ref{Cplusplus}.
2077 @section Giving a Type a Name
2079 @findex N_LSYM, for types
2080 @findex C_DECL, for types
2081 To give a type a name, use the @samp{t} symbol descriptor. The type
2082 is specified by the type information (@pxref{String Field}) for the stab.
2086 .stabs "s_typedef:t16",128,0,0,0 # @r{128 is N_LSYM}
2089 specifies that @code{s_typedef} refers to type number 16. Such stabs
2090 have symbol type @code{N_LSYM} (or @code{C_DECL} for XCOFF). (The Sun
2091 documentation mentions using @code{N_GSYM} in some cases).
2093 If you are specifying the tag name for a structure, union, or
2094 enumeration, use the @samp{T} symbol descriptor instead. I believe C is
2095 the only language with this feature.
2097 If the type is an opaque type (I believe this is a Modula-2 feature),
2098 AIX provides a type descriptor to specify it. The type descriptor is
2099 @samp{o} and is followed by a name. I don't know what the name
2100 means---is it always the same as the name of the type, or is this type
2101 descriptor used with a nameless stab (@pxref{String Field})? There
2102 optionally follows a comma followed by type information which defines
2103 the type of this type. If omitted, a semicolon is used in place of the
2104 comma and the type information, and the type is much like a generic
2105 pointer type---it has a known size but little else about it is
2119 This code generates a stab for a union tag and a stab for a union
2120 variable. Both use the @code{N_LSYM} stab type. If a union variable is
2121 scoped locally to the procedure in which it is defined, its stab is
2122 located immediately preceding the @code{N_LBRAC} for the procedure's block
2125 The stab for the union tag, however, is located preceding the code for
2126 the procedure in which it is defined. The stab type is @code{N_LSYM}. This
2127 would seem to imply that the union type is file scope, like the struct
2128 type @code{s_tag}. This is not true. The contents and position of the stab
2129 for @code{u_type} do not convey any information about its procedure local
2132 @c FIXME: phony line break. Can probably be fixed by using an example
2133 @c with fewer fields.
2136 .stabs "u_tag:T23=u4u_int:1,0,32;u_float:12,0,32;u_char:21,0,32;;",
2140 The symbol descriptor @samp{T}, following the @samp{name:} means that
2141 the stab describes an enumeration, structure, or union tag. The type
2142 descriptor @samp{u}, following the @samp{23=} of the type definition,
2143 narrows it down to a union type definition. Following the @samp{u} is
2144 the number of bytes in the union. After that is a list of union element
2145 descriptions. Their format is @var{name:type, bit offset into the
2146 union, number of bytes for the element;}.
2148 The stab for the union variable is:
2151 .stabs "an_u:23",128,0,0,-20 # @r{128 is N_LSYM}
2154 @samp{-20} specifies where the variable is stored (@pxref{Stack
2157 @node Function Types
2158 @section Function Types
2160 Various types can be defined for function variables. These types are
2161 not used in defining functions (@pxref{Procedures}); they are used for
2162 things like pointers to functions.
2164 The simple, traditional, type is type descriptor @samp{f} is followed by
2165 type information for the return type of the function, followed by a
2168 This does not deal with functions for which the number and types of the
2169 parameters are part of the type, as in Modula-2 or ANSI C. AIX provides
2170 extensions to specify these, using the @samp{f}, @samp{F}, @samp{p}, and
2171 @samp{R} type descriptors.
2173 First comes the type descriptor. If it is @samp{f} or @samp{F}, this
2174 type involves a function rather than a procedure, and the type
2175 information for the return type of the function follows, followed by a
2176 comma. Then comes the number of parameters to the function and a
2177 semicolon. Then, for each parameter, there is the name of the parameter
2178 followed by a colon (this is only present for type descriptors @samp{R}
2179 and @samp{F} which represent Pascal function or procedure parameters),
2180 type information for the parameter, a comma, 0 if passed by reference or
2181 1 if passed by value, and a semicolon. The type definition ends with a
2184 For example, this variable definition:
2191 generates the following code:
2194 .stabs "g_pf:G24=*25=f1",32,0,0,0
2195 .common _g_pf,4,"bss"
2198 The variable defines a new type, 24, which is a pointer to another new
2199 type, 25, which is a function returning @code{int}.
2202 @chapter Symbol Information in Symbol Tables
2204 This chapter describes the format of symbol table entries
2205 and how stab assembler directives map to them. It also describes the
2206 transformations that the assembler and linker make on data from stabs.
2209 * Symbol Table Format::
2210 * Transformations On Symbol Tables::
2213 @node Symbol Table Format
2214 @section Symbol Table Format
2216 Each time the assembler encounters a stab directive, it puts
2217 each field of the stab into a corresponding field in a symbol table
2218 entry of its output file. If the stab contains a string field, the
2219 symbol table entry for that stab points to a string table entry
2220 containing the string data from the stab. Assembler labels become
2221 relocatable addresses. Symbol table entries in a.out have the format:
2223 @c FIXME: should refer to external, not internal.
2225 struct internal_nlist @{
2226 unsigned long n_strx; /* index into string table of name */
2227 unsigned char n_type; /* type of symbol */
2228 unsigned char n_other; /* misc info (usually empty) */
2229 unsigned short n_desc; /* description field */
2230 bfd_vma n_value; /* value of symbol */
2234 If the stab has a string, the @code{n_strx} field holds the offset in
2235 bytes of the string within the string table. The string is terminated
2236 by a NUL character. If the stab lacks a string (for example, it was
2237 produced by a @code{.stabn} or @code{.stabd} directive), the
2238 @code{n_strx} field is zero.
2240 Symbol table entries with @code{n_type} field values greater than 0x1f
2241 originated as stabs generated by the compiler (with one random
2242 exception). The other entries were placed in the symbol table of the
2243 executable by the assembler or the linker.
2245 @node Transformations On Symbol Tables
2246 @section Transformations on Symbol Tables
2248 The linker concatenates object files and does fixups of externally
2251 You can see the transformations made on stab data by the assembler and
2252 linker by examining the symbol table after each pass of the build. To
2253 do this, use @samp{nm -ap}, which dumps the symbol table, including
2254 debugging information, unsorted. For stab entries the columns are:
2255 @var{value}, @var{other}, @var{desc}, @var{type}, @var{string}. For
2256 assembler and linker symbols, the columns are: @var{value}, @var{type},
2259 The low 5 bits of the stab type tell the linker how to relocate the
2260 value of the stab. Thus for stab types like @code{N_RSYM} and
2261 @code{N_LSYM}, where the value is an offset or a register number, the
2262 low 5 bits are @code{N_ABS}, which tells the linker not to relocate the
2265 Where the value of a stab contains an assembly language label,
2266 it is transformed by each build step. The assembler turns it into a
2267 relocatable address and the linker turns it into an absolute address.
2270 * Transformations On Static Variables::
2271 * Transformations On Global Variables::
2272 * Stab Section Transformations:: For some object file formats,
2273 things are a bit different.
2276 @node Transformations On Static Variables
2277 @subsection Transformations on Static Variables
2279 This source line defines a static variable at file scope:
2282 static int s_g_repeat
2286 The following stab describes the symbol:
2289 .stabs "s_g_repeat:S1",38,0,0,_s_g_repeat
2293 The assembler transforms the stab into this symbol table entry in the
2294 @file{.o} file. The location is expressed as a data segment offset.
2297 00000084 - 00 0000 STSYM s_g_repeat:S1
2301 In the symbol table entry from the executable, the linker has made the
2302 relocatable address absolute.
2305 0000e00c - 00 0000 STSYM s_g_repeat:S1
2308 @node Transformations On Global Variables
2309 @subsection Transformations on Global Variables
2311 Stabs for global variables do not contain location information. In
2312 this case, the debugger finds location information in the assembler or
2313 linker symbol table entry describing the variable. The source line:
2323 .stabs "g_foo:G2",32,0,0,0
2326 The variable is represented by two symbol table entries in the object
2327 file (see below). The first one originated as a stab. The second one
2328 is an external symbol. The upper case @samp{D} signifies that the
2329 @code{n_type} field of the symbol table contains 7, @code{N_DATA} with
2330 local linkage. The stab's value is zero since the value is not used for
2331 @code{N_GSYM} stabs. The value of the linker symbol is the relocatable
2332 address corresponding to the variable.
2335 00000000 - 00 0000 GSYM g_foo:G2
2340 These entries as transformed by the linker. The linker symbol table
2341 entry now holds an absolute address:
2344 00000000 - 00 0000 GSYM g_foo:G2
2349 @node Stab Section Transformations
2350 @subsection Transformations of Stabs in separate sections
2352 For object file formats using stabs in separate sections (@pxref{Stab
2353 Sections}), use @code{objdump --stabs} instead of @code{nm} to show the
2354 stabs in an object or executable file. @code{objdump} is a GNU utility;
2355 Sun does not provide any equivalent.
2357 The following example is for a stab whose value is an address is
2358 relative to the compilation unit (@pxref{ELF Linker Relocation}). For
2359 example, if the source line
2365 appears within a function, then the assembly language output from the
2371 .stabs "ld:V(0,3)",0x26,0,4,.L18-Ddata.data # @r{0x26 is N_STSYM}
2378 Because the value is formed by subtracting one symbol from another, the
2379 value is absolute, not relocatable, and so the object file contains
2382 Symnum n_type n_othr n_desc n_value n_strx String
2383 31 STSYM 0 4 00000004 680 ld:V(0,3)
2386 without any relocations, and the executable file also contains
2389 Symnum n_type n_othr n_desc n_value n_strx String
2390 31 STSYM 0 4 00000004 680 ld:V(0,3)
2394 @chapter GNU C++ Stabs
2397 * Class Names:: C++ class names are both tags and typedefs.
2398 * Nested Symbols:: C++ symbol names can be within other types.
2399 * Basic Cplusplus Types::
2402 * Methods:: Method definition
2403 * Method Type Descriptor:: The @samp{#} type descriptor
2404 * Member Type Descriptor:: The @samp{@@} type descriptor
2406 * Method Modifiers::
2409 * Virtual Base Classes::
2414 @section C++ Class Names
2416 In C++, a class name which is declared with @code{class}, @code{struct},
2417 or @code{union}, is not only a tag, as in C, but also a type name. Thus
2418 there should be stabs with both @samp{t} and @samp{T} symbol descriptors
2421 To save space, there is a special abbreviation for this case. If the
2422 @samp{T} symbol descriptor is followed by @samp{t}, then the stab
2423 defines both a type name and a tag.
2425 For example, the C++ code
2428 struct foo @{int x;@};
2431 can be represented as either
2434 .stabs "foo:T19=s4x:1,0,32;;",128,0,0,0 # @r{128 is N_LSYM}
2435 .stabs "foo:t19",128,0,0,0
2441 .stabs "foo:Tt19=s4x:1,0,32;;",128,0,0,0
2444 @node Nested Symbols
2445 @section Defining a Symbol Within Another Type
2447 In C++, a symbol (such as a type name) can be defined within another type.
2448 @c FIXME: Needs example.
2450 In stabs, this is sometimes represented by making the name of a symbol
2451 which contains @samp{::}. Such a pair of colons does not end the name
2452 of the symbol, the way a single colon would (@pxref{String Field}). I'm
2453 not sure how consistently used or well thought out this mechanism is.
2454 So that a pair of colons in this position always has this meaning,
2455 @samp{:} cannot be used as a symbol descriptor.
2457 For example, if the string for a stab is @samp{foo::bar::baz:t5=*6},
2458 then @code{foo::bar::baz} is the name of the symbol, @samp{t} is the
2459 symbol descriptor, and @samp{5=*6} is the type information.
2461 @node Basic Cplusplus Types
2462 @section Basic Types For C++
2464 << the examples that follow are based on a01.C >>
2467 C++ adds two more builtin types to the set defined for C. These are
2468 the unknown type and the vtable record type. The unknown type, type
2469 16, is defined in terms of itself like the void type.
2471 The vtable record type, type 17, is defined as a structure type and
2472 then as a structure tag. The structure has four fields: delta, index,
2473 pfn, and delta2. pfn is the function pointer.
2475 << In boilerplate $vtbl_ptr_type, what are the fields delta,
2476 index, and delta2 used for? >>
2478 This basic type is present in all C++ programs even if there are no
2479 virtual methods defined.
2482 .stabs "struct_name:sym_desc(type)type_def(17)=type_desc(struct)struct_bytes(8)
2483 elem_name(delta):type_ref(short int),bit_offset(0),field_bits(16);
2484 elem_name(index):type_ref(short int),bit_offset(16),field_bits(16);
2485 elem_name(pfn):type_def(18)=type_desc(ptr to)type_ref(void),
2486 bit_offset(32),field_bits(32);
2487 elem_name(delta2):type_def(short int);bit_offset(32),field_bits(16);;"
2492 .stabs "$vtbl_ptr_type:t17=s8
2493 delta:6,0,16;index:6,16,16;pfn:18=*15,32,32;delta2:6,32,16;;"
2498 .stabs "name:sym_dec(struct tag)type_ref($vtbl_ptr_type)",N_LSYM,NIL,NIL,NIL
2502 .stabs "$vtbl_ptr_type:T17",128,0,0,0
2505 @node Simple Classes
2506 @section Simple Class Definition
2508 The stabs describing C++ language features are an extension of the
2509 stabs describing C. Stabs representing C++ class types elaborate
2510 extensively on the stab format used to describe structure types in C.
2511 Stabs representing class type variables look just like stabs
2512 representing C language variables.
2514 Consider the following very simple class definition.
2520 int Ameth(int in, char other);
2524 The class @code{baseA} is represented by two stabs. The first stab describes
2525 the class as a structure type. The second stab describes a structure
2526 tag of the class type. Both stabs are of stab type @code{N_LSYM}. Since the
2527 stab is not located between an @code{N_FUN} and an @code{N_LBRAC} stab this indicates
2528 that the class is defined at file scope. If it were, then the @code{N_LSYM}
2529 would signify a local variable.
2531 A stab describing a C++ class type is similar in format to a stab
2532 describing a C struct, with each class member shown as a field in the
2533 structure. The part of the struct format describing fields is
2534 expanded to include extra information relevant to C++ class members.
2535 In addition, if the class has multiple base classes or virtual
2536 functions the struct format outside of the field parts is also
2539 In this simple example the field part of the C++ class stab
2540 representing member data looks just like the field part of a C struct
2541 stab. The section on protections describes how its format is
2542 sometimes extended for member data.
2544 The field part of a C++ class stab representing a member function
2545 differs substantially from the field part of a C struct stab. It
2546 still begins with @samp{name:} but then goes on to define a new type number
2547 for the member function, describe its return type, its argument types,
2548 its protection level, any qualifiers applied to the method definition,
2549 and whether the method is virtual or not. If the method is virtual
2550 then the method description goes on to give the vtable index of the
2551 method, and the type number of the first base class defining the
2554 When the field name is a method name it is followed by two colons rather
2555 than one. This is followed by a new type definition for the method.
2556 This is a number followed by an equal sign and the type of the method.
2557 Normally this will be a type declared using the @samp{#} type
2558 descriptor; see @ref{Method Type Descriptor}; static member functions
2559 are declared using the @samp{f} type descriptor instead; see
2560 @ref{Function Types}.
2562 The format of an overloaded operator method name differs from that of
2563 other methods. It is @samp{op$::@var{operator-name}.} where
2564 @var{operator-name} is the operator name such as @samp{+} or @samp{+=}.
2565 The name ends with a period, and any characters except the period can
2566 occur in the @var{operator-name} string.
2568 The next part of the method description represents the arguments to the
2569 method, preceded by a colon and ending with a semi-colon. The types of
2570 the arguments are expressed in the same way argument types are expressed
2571 in C++ name mangling. In this example an @code{int} and a @code{char}
2574 This is followed by a number, a letter, and an asterisk or period,
2575 followed by another semicolon. The number indicates the protections
2576 that apply to the member function. Here the 2 means public. The
2577 letter encodes any qualifier applied to the method definition. In
2578 this case, @samp{A} means that it is a normal function definition. The dot
2579 shows that the method is not virtual. The sections that follow
2580 elaborate further on these fields and describe the additional
2581 information present for virtual methods.
2585 .stabs "class_name:sym_desc(type)type_def(20)=type_desc(struct)struct_bytes(4)
2586 field_name(Adat):type(int),bit_offset(0),field_bits(32);
2588 method_name(Ameth)::type_def(21)=type_desc(method)return_type(int);
2589 :arg_types(int char);
2590 protection(public)qualifier(normal)virtual(no);;"
2595 .stabs "baseA:t20=s4Adat:1,0,32;Ameth::21=##1;:ic;2A.;;",128,0,0,0
2597 .stabs "class_name:sym_desc(struct tag)",N_LSYM,NIL,NIL,NIL
2599 .stabs "baseA:T20",128,0,0,0
2602 @node Class Instance
2603 @section Class Instance
2605 As shown above, describing even a simple C++ class definition is
2606 accomplished by massively extending the stab format used in C to
2607 describe structure types. However, once the class is defined, C stabs
2608 with no modifications can be used to describe class instances. The
2618 yields the following stab describing the class instance. It looks no
2619 different from a standard C stab describing a local variable.
2622 .stabs "name:type_ref(baseA)", N_LSYM, NIL, NIL, frame_ptr_offset
2626 .stabs "AbaseA:20",128,0,0,-20
2630 @section Method Definition
2632 The class definition shown above declares Ameth. The C++ source below
2637 baseA::Ameth(int in, char other)
2644 This method definition yields three stabs following the code of the
2645 method. One stab describes the method itself and following two describe
2646 its parameters. Although there is only one formal argument all methods
2647 have an implicit argument which is the @code{this} pointer. The @code{this}
2648 pointer is a pointer to the object on which the method was called. Note
2649 that the method name is mangled to encode the class name and argument
2650 types. Name mangling is described in the @sc{arm} (@cite{The Annotated
2651 C++ Reference Manual}, by Ellis and Stroustrup, @sc{isbn}
2652 0-201-51459-1); @file{gpcompare.texi} in Cygnus GCC distributions
2653 describes the differences between GNU mangling and @sc{arm}
2655 @c FIXME: Use @xref, especially if this is generally installed in the
2657 @c FIXME: This information should be in a net release, either of GCC or
2658 @c GDB. But gpcompare.texi doesn't seem to be in the FSF GCC.
2661 .stabs "name:symbol_descriptor(global function)return_type(int)",
2662 N_FUN, NIL, NIL, code_addr_of_method_start
2664 .stabs "Ameth__5baseAic:F1",36,0,0,_Ameth__5baseAic
2667 Here is the stab for the @code{this} pointer implicit argument. The
2668 name of the @code{this} pointer is always @code{this}. Type 19, the
2669 @code{this} pointer is defined as a pointer to type 20, @code{baseA},
2670 but a stab defining @code{baseA} has not yet been emitted. Since the
2671 compiler knows it will be emitted shortly, here it just outputs a cross
2672 reference to the undefined symbol, by prefixing the symbol name with
2676 .stabs "name:sym_desc(register param)type_def(19)=
2677 type_desc(ptr to)type_ref(baseA)=
2678 type_desc(cross-reference to)baseA:",N_RSYM,NIL,NIL,register_number
2680 .stabs "this:P19=*20=xsbaseA:",64,0,0,8
2683 The stab for the explicit integer argument looks just like a parameter
2684 to a C function. The last field of the stab is the offset from the
2685 argument pointer, which in most systems is the same as the frame
2689 .stabs "name:sym_desc(value parameter)type_ref(int)",
2690 N_PSYM,NIL,NIL,offset_from_arg_ptr
2692 .stabs "in:p1",160,0,0,72
2695 << The examples that follow are based on A1.C >>
2697 @node Method Type Descriptor
2698 @section The @samp{#} Type Descriptor
2700 This is used to describe a class method. This is a function which takes
2701 an extra argument as its first argument, for the @code{this} pointer.
2703 If the @samp{#} is immediately followed by another @samp{#}, the second
2704 one will be followed by the return type and a semicolon. The class and
2705 argument types are not specified, and must be determined by demangling
2706 the name of the method if it is available.
2708 Otherwise, the single @samp{#} is followed by the class type, a comma,
2709 the return type, a comma, and zero or more parameter types separated by
2710 commas. The list of arguments is terminated by a semicolon. In the
2711 debugging output generated by gcc, a final argument type of @code{void}
2712 indicates a method which does not take a variable number of arguments.
2713 If the final argument type of @code{void} does not appear, the method
2714 was declared with an ellipsis.
2716 Note that although such a type will normally be used to describe fields
2717 in structures, unions, or classes, for at least some versions of the
2718 compiler it can also be used in other contexts.
2720 @node Member Type Descriptor
2721 @section The @samp{@@} Type Descriptor
2723 The @samp{@@} type descriptor is for a member (class and variable) type.
2724 It is followed by type information for the offset basetype, a comma, and
2725 type information for the type of the field being pointed to. (FIXME:
2726 this is acknowledged to be gibberish. Can anyone say what really goes
2729 Note that there is a conflict between this and type attributes
2730 (@pxref{String Field}); both use type descriptor @samp{@@}.
2731 Fortunately, the @samp{@@} type descriptor used in this C++ sense always
2732 will be followed by a digit, @samp{(}, or @samp{-}, and type attributes
2733 never start with those things.
2736 @section Protections
2738 In the simple class definition shown above all member data and
2739 functions were publicly accessible. The example that follows
2740 contrasts public, protected and privately accessible fields and shows
2741 how these protections are encoded in C++ stabs.
2743 If the character following the @samp{@var{field-name}:} part of the
2744 string is @samp{/}, then the next character is the visibility. @samp{0}
2745 means private, @samp{1} means protected, and @samp{2} means public.
2746 Debuggers should ignore visibility characters they do not recognize, and
2747 assume a reasonable default (such as public) (GDB 4.11 does not, but
2748 this should be fixed in the next GDB release). If no visibility is
2749 specified the field is public. The visibility @samp{9} means that the
2750 field has been optimized out and is public (there is no way to specify
2751 an optimized out field with a private or protected visibility).
2752 Visibility @samp{9} is not supported by GDB 4.11; this should be fixed
2753 in the next GDB release.
2755 The following C++ source:
2769 generates the following stab:
2773 .stabs "vis:T19=s12priv:/01,0,32;prot:/12,32,8;pub:12,64,32;;",128,0,0,0
2776 @samp{vis:T19=s12} indicates that type number 19 is a 12 byte structure
2777 named @code{vis} The @code{priv} field has public visibility
2778 (@samp{/0}), type int (@samp{1}), and offset and size @samp{,0,32;}.
2779 The @code{prot} field has protected visibility (@samp{/1}), type char
2780 (@samp{2}) and offset and size @samp{,32,8;}. The @code{pub} field has
2781 type float (@samp{12}), and offset and size @samp{,64,32;}.
2783 Protections for member functions are signified by one digit embedded in
2784 the field part of the stab describing the method. The digit is 0 if
2785 private, 1 if protected and 2 if public. Consider the C++ class
2789 class all_methods @{
2791 int priv_meth(int in)@{return in;@};
2793 char protMeth(char in)@{return in;@};
2795 float pubMeth(float in)@{return in;@};
2799 It generates the following stab. The digit in question is to the left
2800 of an @samp{A} in each case. Notice also that in this case two symbol
2801 descriptors apply to the class name struct tag and struct type.
2804 .stabs "class_name:sym_desc(struct tag&type)type_def(21)=
2805 sym_desc(struct)struct_bytes(1)
2806 meth_name::type_def(22)=sym_desc(method)returning(int);
2807 :args(int);protection(private)modifier(normal)virtual(no);
2808 meth_name::type_def(23)=sym_desc(method)returning(char);
2809 :args(char);protection(protected)modifier(normal)virtual(no);
2810 meth_name::type_def(24)=sym_desc(method)returning(float);
2811 :args(float);protection(public)modifier(normal)virtual(no);;",
2816 .stabs "all_methods:Tt21=s1priv_meth::22=##1;:i;0A.;protMeth::23=##2;:c;1A.;
2817 pubMeth::24=##12;:f;2A.;;",128,0,0,0
2820 @node Method Modifiers
2821 @section Method Modifiers (@code{const}, @code{volatile}, @code{const volatile})
2825 In the class example described above all the methods have the normal
2826 modifier. This method modifier information is located just after the
2827 protection information for the method. This field has four possible
2828 character values. Normal methods use @samp{A}, const methods use
2829 @samp{B}, volatile methods use @samp{C}, and const volatile methods use
2830 @samp{D}. Consider the class definition below:
2835 int ConstMeth (int arg) const @{ return arg; @};
2836 char VolatileMeth (char arg) volatile @{ return arg; @};
2837 float ConstVolMeth (float arg) const volatile @{return arg; @};
2841 This class is described by the following stab:
2844 .stabs "class(A):sym_desc(struct)type_def(20)=type_desc(struct)struct_bytes(1)
2845 meth_name(ConstMeth)::type_def(21)sym_desc(method)
2846 returning(int);:arg(int);protection(public)modifier(const)virtual(no);
2847 meth_name(VolatileMeth)::type_def(22)=sym_desc(method)
2848 returning(char);:arg(char);protection(public)modifier(volatile)virt(no)
2849 meth_name(ConstVolMeth)::type_def(23)=sym_desc(method)
2850 returning(float);:arg(float);protection(public)modifier(const volatile)
2851 virtual(no);;", @dots{}
2855 .stabs "A:T20=s1ConstMeth::21=##1;:i;2B.;VolatileMeth::22=##2;:c;2C.;
2856 ConstVolMeth::23=##12;:f;2D.;;",128,0,0,0
2859 @node Virtual Methods
2860 @section Virtual Methods
2862 << The following examples are based on a4.C >>
2864 The presence of virtual methods in a class definition adds additional
2865 data to the class description. The extra data is appended to the
2866 description of the virtual method and to the end of the class
2867 description. Consider the class definition below:
2873 virtual int A_virt (int arg) @{ return arg; @};
2877 This results in the stab below describing class A. It defines a new
2878 type (20) which is an 8 byte structure. The first field of the class
2879 struct is @samp{Adat}, an integer, starting at structure offset 0 and
2882 The second field in the class struct is not explicitly defined by the
2883 C++ class definition but is implied by the fact that the class
2884 contains a virtual method. This field is the vtable pointer. The
2885 name of the vtable pointer field starts with @samp{$vf} and continues with a
2886 type reference to the class it is part of. In this example the type
2887 reference for class A is 20 so the name of its vtable pointer field is
2888 @samp{$vf20}, followed by the usual colon.
2890 Next there is a type definition for the vtable pointer type (21).
2891 This is in turn defined as a pointer to another new type (22).
2893 Type 22 is the vtable itself, which is defined as an array, indexed by
2894 a range of integers between 0 and 1, and whose elements are of type
2895 17. Type 17 was the vtable record type defined by the boilerplate C++
2896 type definitions, as shown earlier.
2898 The bit offset of the vtable pointer field is 32. The number of bits
2899 in the field are not specified when the field is a vtable pointer.
2901 Next is the method definition for the virtual member function @code{A_virt}.
2902 Its description starts out using the same format as the non-virtual
2903 member functions described above, except instead of a dot after the
2904 @samp{A} there is an asterisk, indicating that the function is virtual.
2905 Since is is virtual some addition information is appended to the end
2906 of the method description.
2908 The first number represents the vtable index of the method. This is a
2909 32 bit unsigned number with the high bit set, followed by a
2912 The second number is a type reference to the first base class in the
2913 inheritance hierarchy defining the virtual member function. In this
2914 case the class stab describes a base class so the virtual function is
2915 not overriding any other definition of the method. Therefore the
2916 reference is to the type number of the class that the stab is
2919 This is followed by three semi-colons. One marks the end of the
2920 current sub-section, one marks the end of the method field, and the
2921 third marks the end of the struct definition.
2923 For classes containing virtual functions the very last section of the
2924 string part of the stab holds a type reference to the first base
2925 class. This is preceded by @samp{~%} and followed by a final semi-colon.
2928 .stabs "class_name(A):type_def(20)=sym_desc(struct)struct_bytes(8)
2929 field_name(Adat):type_ref(int),bit_offset(0),field_bits(32);
2930 field_name(A virt func ptr):type_def(21)=type_desc(ptr to)type_def(22)=
2931 sym_desc(array)index_type_ref(range of int from 0 to 1);
2932 elem_type_ref(vtbl elem type),
2934 meth_name(A_virt)::typedef(23)=sym_desc(method)returning(int);
2935 :arg_type(int),protection(public)normal(yes)virtual(yes)
2936 vtable_index(1);class_first_defining(A);;;~%first_base(A);",
2940 @c FIXME: bogus line break.
2942 .stabs "A:t20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
2943 A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
2947 @section Inheritance
2949 Stabs describing C++ derived classes include additional sections that
2950 describe the inheritance hierarchy of the class. A derived class stab
2951 also encodes the number of base classes. For each base class it tells
2952 if the base class is virtual or not, and if the inheritance is private
2953 or public. It also gives the offset into the object of the portion of
2954 the object corresponding to each base class.
2956 This additional information is embedded in the class stab following the
2957 number of bytes in the struct. First the number of base classes
2958 appears bracketed by an exclamation point and a comma.
2960 Then for each base type there repeats a series: a virtual character, a
2961 visibility character, a number, a comma, another number, and a
2964 The virtual character is @samp{1} if the base class is virtual and
2965 @samp{0} if not. The visibility character is @samp{2} if the derivation
2966 is public, @samp{1} if it is protected, and @samp{0} if it is private.
2967 Debuggers should ignore virtual or visibility characters they do not
2968 recognize, and assume a reasonable default (such as public and
2969 non-virtual) (GDB 4.11 does not, but this should be fixed in the next
2972 The number following the virtual and visibility characters is the offset
2973 from the start of the object to the part of the object pertaining to the
2976 After the comma, the second number is a type_descriptor for the base
2977 type. Finally a semi-colon ends the series, which repeats for each
2980 The source below defines three base classes @code{A}, @code{B}, and
2981 @code{C} and the derived class @code{D}.
2988 virtual int A_virt (int arg) @{ return arg; @};
2994 virtual int B_virt (int arg) @{return arg; @};
3000 virtual int C_virt (int arg) @{return arg; @};
3003 class D : A, virtual B, public C @{
3006 virtual int A_virt (int arg ) @{ return arg+1; @};
3007 virtual int B_virt (int arg) @{ return arg+2; @};
3008 virtual int C_virt (int arg) @{ return arg+3; @};
3009 virtual int D_virt (int arg) @{ return arg; @};
3013 Class stabs similar to the ones described earlier are generated for
3016 @c FIXME!!! the linebreaks in the following example probably make the
3017 @c examples literally unusable, but I don't know any other way to get
3018 @c them on the page.
3019 @c One solution would be to put some of the type definitions into
3020 @c separate stabs, even if that's not exactly what the compiler actually
3023 .stabs "A:T20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
3024 A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
3026 .stabs "B:Tt25=s8Bdat:1,0,32;$vf25:21,32;B_virt::26=##1;
3027 :i;2A*-2147483647;25;;;~%25;",128,0,0,0
3029 .stabs "C:Tt28=s8Cdat:1,0,32;$vf28:21,32;C_virt::29=##1;
3030 :i;2A*-2147483647;28;;;~%28;",128,0,0,0
3033 In the stab describing derived class @code{D} below, the information about
3034 the derivation of this class is encoded as follows.
3037 .stabs "derived_class_name:symbol_descriptors(struct tag&type)=
3038 type_descriptor(struct)struct_bytes(32)!num_bases(3),
3039 base_virtual(no)inheritance_public(no)base_offset(0),
3040 base_class_type_ref(A);
3041 base_virtual(yes)inheritance_public(no)base_offset(NIL),
3042 base_class_type_ref(B);
3043 base_virtual(no)inheritance_public(yes)base_offset(64),
3044 base_class_type_ref(C); @dots{}
3047 @c FIXME! fake linebreaks.
3049 .stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:
3050 1,160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt:
3051 :32:i;2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;
3052 28;;D_virt::32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
3055 @node Virtual Base Classes
3056 @section Virtual Base Classes
3058 A derived class object consists of a concatenation in memory of the data
3059 areas defined by each base class, starting with the leftmost and ending
3060 with the rightmost in the list of base classes. The exception to this
3061 rule is for virtual inheritance. In the example above, class @code{D}
3062 inherits virtually from base class @code{B}. This means that an
3063 instance of a @code{D} object will not contain its own @code{B} part but
3064 merely a pointer to a @code{B} part, known as a virtual base pointer.
3066 In a derived class stab, the base offset part of the derivation
3067 information, described above, shows how the base class parts are
3068 ordered. The base offset for a virtual base class is always given as 0.
3069 Notice that the base offset for @code{B} is given as 0 even though
3070 @code{B} is not the first base class. The first base class @code{A}
3073 The field information part of the stab for class @code{D} describes the field
3074 which is the pointer to the virtual base class @code{B}. The vbase pointer
3075 name is @samp{$vb} followed by a type reference to the virtual base class.
3076 Since the type id for @code{B} in this example is 25, the vbase pointer name
3079 @c FIXME!! fake linebreaks below
3081 .stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:1,
3082 160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt::32:i;
3083 2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;28;;D_virt:
3084 :32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
3087 Following the name and a semicolon is a type reference describing the
3088 type of the virtual base class pointer, in this case 24. Type 24 was
3089 defined earlier as the type of the @code{B} class @code{this} pointer. The
3090 @code{this} pointer for a class is a pointer to the class type.
3093 .stabs "this:P24=*25=xsB:",64,0,0,8
3096 Finally the field offset part of the vbase pointer field description
3097 shows that the vbase pointer is the first field in the @code{D} object,
3098 before any data fields defined by the class. The layout of a @code{D}
3099 class object is a follows, @code{Adat} at 0, the vtable pointer for
3100 @code{A} at 32, @code{Cdat} at 64, the vtable pointer for C at 96, the
3101 virtual base pointer for @code{B} at 128, and @code{Ddat} at 160.
3104 @node Static Members
3105 @section Static Members
3107 The data area for a class is a concatenation of the space used by the
3108 data members of the class. If the class has virtual methods, a vtable
3109 pointer follows the class data. The field offset part of each field
3110 description in the class stab shows this ordering.
3112 << How is this reflected in stabs? See Cygnus bug #677 for some info. >>
3115 @appendix Table of Stab Types
3117 The following are all the possible values for the stab type field, for
3118 a.out files, in numeric order. This does not apply to XCOFF, but
3119 it does apply to stabs in sections (@pxref{Stab Sections}). Stabs in
3120 ECOFF use these values but add 0x8f300 to distinguish them from non-stab
3123 The symbolic names are defined in the file @file{include/aout/stabs.def}.
3126 * Non-Stab Symbol Types:: Types from 0 to 0x1f
3127 * Stab Symbol Types:: Types from 0x20 to 0xff
3130 @node Non-Stab Symbol Types
3131 @appendixsec Non-Stab Symbol Types
3133 The following types are used by the linker and assembler, not by stab
3134 directives. Since this document does not attempt to describe aspects of
3135 object file format other than the debugging format, no details are
3138 @c Try to get most of these to fit on a single line.
3148 File scope absolute symbol
3150 @item 0x3 N_ABS | N_EXT
3151 External absolute symbol
3154 File scope text symbol
3156 @item 0x5 N_TEXT | N_EXT
3157 External text symbol
3160 File scope data symbol
3162 @item 0x7 N_DATA | N_EXT
3163 External data symbol
3166 File scope BSS symbol
3168 @item 0x9 N_BSS | N_EXT
3172 Same as @code{N_FN}, for Sequent compilers
3175 Symbol is indirected to another symbol
3178 Common---visible after shared library dynamic link
3181 @itemx 0x15 N_SETA | N_EXT
3182 Absolute set element
3185 @itemx 0x17 N_SETT | N_EXT
3186 Text segment set element
3189 @itemx 0x19 N_SETD | N_EXT
3190 Data segment set element
3193 @itemx 0x1b N_SETB | N_EXT
3194 BSS segment set element
3197 @itemx 0x1d N_SETV | N_EXT
3198 Pointer to set vector
3200 @item 0x1e N_WARNING
3201 Print a warning message during linking
3204 File name of a @file{.o} file
3207 @node Stab Symbol Types
3208 @appendixsec Stab Symbol Types
3210 The following symbol types indicate that this is a stab. This is the
3211 full list of stab numbers, including stab types that are used in
3212 languages other than C.
3216 Global symbol; see @ref{Global Variables}.
3219 Function name (for BSD Fortran); see @ref{Procedures}.
3222 Function name (@pxref{Procedures}) or text segment variable
3226 Data segment file-scope variable; see @ref{Statics}.
3229 BSS segment file-scope variable; see @ref{Statics}.
3232 Name of main routine; see @ref{Main Program}.
3235 Variable in @code{.rodata} section; see @ref{Statics}.
3238 Global symbol (for Pascal); see @ref{N_PC}.
3241 Number of symbols (according to Ultrix V4.0); see @ref{N_NSYMS}.
3244 No DST map; see @ref{N_NOMAP}.
3246 @c FIXME: describe this solaris feature in the body of the text (see
3247 @c comments in include/aout/stab.def).
3249 Object file (Solaris2).
3251 @c See include/aout/stab.def for (a little) more info.
3253 Debugger options (Solaris2).
3256 Register variable; see @ref{Register Variables}.
3259 Modula-2 compilation unit; see @ref{N_M2C}.
3262 Line number in text segment; see @ref{Line Numbers}.
3265 Line number in data segment; see @ref{Line Numbers}.
3268 Line number in bss segment; see @ref{Line Numbers}.
3271 Sun source code browser, path to @file{.cb} file; see @ref{N_BROWS}.
3274 GNU Modula2 definition module dependency; see @ref{N_DEFD}.
3277 Function start/body/end line numbers (Solaris2).
3280 GNU C++ exception variable; see @ref{N_EHDECL}.
3283 Modula2 info "for imc" (according to Ultrix V4.0); see @ref{N_MOD2}.
3286 GNU C++ @code{catch} clause; see @ref{N_CATCH}.
3289 Structure of union element; see @ref{N_SSYM}.
3292 Last stab for module (Solaris2).
3295 Path and name of source file; see @ref{Source Files}.
3298 Stack variable (@pxref{Stack Variables}) or type (@pxref{Typedefs}).
3301 Beginning of an include file (Sun only); see @ref{Include Files}.
3304 Name of include file; see @ref{Include Files}.
3307 Parameter variable; see @ref{Parameters}.
3310 End of an include file; see @ref{Include Files}.
3313 Alternate entry point; see @ref{Alternate Entry Points}.
3316 Beginning of a lexical block; see @ref{Block Structure}.
3319 Place holder for a deleted include file; see @ref{Include Files}.
3322 Modula2 scope information (Sun linker); see @ref{N_SCOPE}.
3325 End of a lexical block; see @ref{Block Structure}.
3328 Begin named common block; see @ref{Common Blocks}.
3331 End named common block; see @ref{Common Blocks}.
3334 Member of a common block; see @ref{Common Blocks}.
3336 @c FIXME: How does this really work? Move it to main body of document.
3338 Pascal @code{with} statement: type,,0,0,offset (Solaris2).
3341 Gould non-base registers; see @ref{Gould}.
3344 Gould non-base registers; see @ref{Gould}.
3347 Gould non-base registers; see @ref{Gould}.
3350 Gould non-base registers; see @ref{Gould}.
3353 Gould non-base registers; see @ref{Gould}.
3356 @c Restore the default table indent
3361 @node Symbol Descriptors
3362 @appendix Table of Symbol Descriptors
3364 The symbol descriptor is the character which follows the colon in many
3365 stabs, and which tells what kind of stab it is. @xref{String Field},
3366 for more information about their use.
3368 @c Please keep this alphabetical
3370 @c In TeX, this looks great, digit is in italics. But makeinfo insists
3371 @c on putting it in `', not realizing that @var should override @code.
3372 @c I don't know of any way to make makeinfo do the right thing. Seems
3373 @c like a makeinfo bug to me.
3377 Variable on the stack; see @ref{Stack Variables}.
3380 C++ nested symbol; see @xref{Nested Symbols}.
3383 Parameter passed by reference in register; see @ref{Reference Parameters}.
3386 Based variable; see @ref{Based Variables}.
3389 Constant; see @ref{Constants}.
3392 Conformant array bound (Pascal, maybe other languages); @ref{Conformant
3393 Arrays}. Name of a caught exception (GNU C++). These can be
3394 distinguished because the latter uses @code{N_CATCH} and the former uses
3395 another symbol type.
3398 Floating point register variable; see @ref{Register Variables}.
3401 Parameter in floating point register; see @ref{Register Parameters}.
3404 File scope function; see @ref{Procedures}.
3407 Global function; see @ref{Procedures}.
3410 Global variable; see @ref{Global Variables}.
3413 @xref{Register Parameters}.
3416 Internal (nested) procedure; see @ref{Nested Procedures}.
3419 Internal (nested) function; see @ref{Nested Procedures}.
3422 Label name (documented by AIX, no further information known).
3425 Module; see @ref{Procedures}.
3428 Argument list parameter; see @ref{Parameters}.
3434 Fortran Function parameter; see @ref{Parameters}.
3437 Unfortunately, three separate meanings have been independently invented
3438 for this symbol descriptor. At least the GNU and Sun uses can be
3439 distinguished by the symbol type. Global Procedure (AIX) (symbol type
3440 used unknown); see @ref{Procedures}. Register parameter (GNU) (symbol
3441 type @code{N_PSYM}); see @ref{Parameters}. Prototype of function
3442 referenced by this file (Sun @code{acc}) (symbol type @code{N_FUN}).
3445 Static Procedure; see @ref{Procedures}.
3448 Register parameter; see @ref{Register Parameters}.
3451 Register variable; see @ref{Register Variables}.
3454 File scope variable; see @ref{Statics}.
3457 Local variable (OS9000).
3460 Type name; see @ref{Typedefs}.
3463 Enumeration, structure, or union tag; see @ref{Typedefs}.
3466 Parameter passed by reference; see @ref{Reference Parameters}.
3469 Procedure scope static variable; see @ref{Statics}.
3472 Conformant array; see @ref{Conformant Arrays}.
3475 Function return variable; see @ref{Parameters}.
3478 @node Type Descriptors
3479 @appendix Table of Type Descriptors
3481 The type descriptor is the character which follows the type number and
3482 an equals sign. It specifies what kind of type is being defined.
3483 @xref{String Field}, for more information about their use.
3488 Type reference; see @ref{String Field}.
3491 Reference to builtin type; see @ref{Negative Type Numbers}.
3494 Method (C++); see @ref{Method Type Descriptor}.
3497 Pointer; see @ref{Miscellaneous Types}.
3503 Type Attributes (AIX); see @ref{String Field}. Member (class and variable)
3504 type (GNU C++); see @ref{Member Type Descriptor}.
3507 Array; see @ref{Arrays}.
3510 Open array; see @ref{Arrays}.
3513 Pascal space type (AIX); see @ref{Miscellaneous Types}. Builtin integer
3514 type (Sun); see @ref{Builtin Type Descriptors}. Const and volatile
3515 qualified type (OS9000).
3518 Volatile-qualified type; see @ref{Miscellaneous Types}.
3521 Complex builtin type (AIX); see @ref{Builtin Type Descriptors}.
3522 Const-qualified type (OS9000).
3525 COBOL Picture type. See AIX documentation for details.
3528 File type; see @ref{Miscellaneous Types}.
3531 N-dimensional dynamic array; see @ref{Arrays}.
3534 Enumeration type; see @ref{Enumerations}.
3537 N-dimensional subarray; see @ref{Arrays}.
3540 Function type; see @ref{Function Types}.
3543 Pascal function parameter; see @ref{Function Types}
3546 Builtin floating point type; see @ref{Builtin Type Descriptors}.
3549 COBOL Group. See AIX documentation for details.
3552 Imported type (AIX); see @ref{Cross-References}. Volatile-qualified
3556 Const-qualified type; see @ref{Miscellaneous Types}.
3559 COBOL File Descriptor. See AIX documentation for details.
3562 Multiple instance type; see @ref{Miscellaneous Types}.
3565 String type; see @ref{Strings}.
3568 Stringptr; see @ref{Strings}.
3571 Opaque type; see @ref{Typedefs}.
3574 Procedure; see @ref{Function Types}.
3577 Packed array; see @ref{Arrays}.
3580 Range type; see @ref{Subranges}.
3583 Builtin floating type; see @ref{Builtin Type Descriptors} (Sun). Pascal
3584 subroutine parameter; see @ref{Function Types} (AIX). Detecting this
3585 conflict is possible with careful parsing (hint: a Pascal subroutine
3586 parameter type will always contain a comma, and a builtin type
3587 descriptor never will).
3590 Structure type; see @ref{Structures}.
3593 Set type; see @ref{Miscellaneous Types}.
3596 Union; see @ref{Unions}.
3599 Variant record. This is a Pascal and Modula-2 feature which is like a
3600 union within a struct in C. See AIX documentation for details.
3603 Wide character; see @ref{Builtin Type Descriptors}.
3606 Cross-reference; see @ref{Cross-References}.
3609 Used by IBM's xlC C++ compiler (for structures, I think).
3612 gstring; see @ref{Strings}.
3615 @node Expanded Reference
3616 @appendix Expanded Reference by Stab Type
3618 @c FIXME: This appendix should go away; see N_PSYM or N_SO for an example.
3620 For a full list of stab types, and cross-references to where they are
3621 described, see @ref{Stab Types}. This appendix just covers certain
3622 stabs which are not yet described in the main body of this document;
3623 eventually the information will all be in one place.
3627 The first line is the symbol type (see @file{include/aout/stab.def}).
3629 The second line describes the language constructs the symbol type
3632 The third line is the stab format with the significant stab fields
3633 named and the rest NIL.
3635 Subsequent lines expand upon the meaning and possible values for each
3636 significant stab field.
3638 Finally, any further information.
3641 * N_PC:: Pascal global symbol
3642 * N_NSYMS:: Number of symbols
3643 * N_NOMAP:: No DST map
3644 * N_M2C:: Modula-2 compilation unit
3645 * N_BROWS:: Path to .cb file for Sun source code browser
3646 * N_DEFD:: GNU Modula2 definition module dependency
3647 * N_EHDECL:: GNU C++ exception variable
3648 * N_MOD2:: Modula2 information "for imc"
3649 * N_CATCH:: GNU C++ "catch" clause
3650 * N_SSYM:: Structure or union element
3651 * N_SCOPE:: Modula2 scope information (Sun only)
3652 * Gould:: non-base register symbols used on Gould systems
3653 * N_LENG:: Length of preceding entry
3659 @deffn @code{.stabs} N_PC
3661 Global symbol (for Pascal).
3664 "name" -> "symbol_name" <<?>>
3665 value -> supposedly the line number (stab.def is skeptical)
3669 @file{stabdump.c} says:
3671 global pascal symbol: name,,0,subtype,line
3679 @deffn @code{.stabn} N_NSYMS
3681 Number of symbols (according to Ultrix V4.0).
3684 0, files,,funcs,lines (stab.def)
3691 @deffn @code{.stabs} N_NOMAP
3693 No DST map for symbol (according to Ultrix V4.0). I think this means a
3694 variable has been optimized out.
3697 name, ,0,type,ignored (stab.def)
3704 @deffn @code{.stabs} N_M2C
3706 Modula-2 compilation unit.
3709 "string" -> "unit_name,unit_time_stamp[,code_time_stamp]"
3711 value -> 0 (main unit)
3715 See @cite{Dbx and Dbxtool Interfaces}, 2nd edition, by Sun, 1988, for
3723 @deffn @code{.stabs} N_BROWS
3725 Sun source code browser, path to @file{.cb} file
3728 "path to associated @file{.cb} file"
3730 Note: N_BROWS has the same value as N_BSLINE.
3736 @deffn @code{.stabn} N_DEFD
3738 GNU Modula2 definition module dependency.
3740 GNU Modula-2 definition module dependency. The value is the
3741 modification time of the definition file. The other field is non-zero
3742 if it is imported with the GNU M2 keyword @code{%INITIALIZE}. Perhaps
3743 @code{N_M2C} can be used if there are enough empty fields?
3749 @deffn @code{.stabs} N_EHDECL
3751 GNU C++ exception variable <<?>>.
3753 "@var{string} is variable name"
3755 Note: conflicts with @code{N_MOD2}.
3761 @deffn @code{.stab?} N_MOD2
3763 Modula2 info "for imc" (according to Ultrix V4.0)
3765 Note: conflicts with @code{N_EHDECL} <<?>>
3771 @deffn @code{.stabn} N_CATCH
3773 GNU C++ @code{catch} clause
3775 GNU C++ @code{catch} clause. The value is its address. The desc field
3776 is nonzero if this entry is immediately followed by a @code{CAUGHT} stab
3777 saying what exception was caught. Multiple @code{CAUGHT} stabs means
3778 that multiple exceptions can be caught here. If desc is 0, it means all
3779 exceptions are caught here.
3785 @deffn @code{.stabn} N_SSYM
3787 Structure or union element.
3789 The value is the offset in the structure.
3791 <<?looking at structs and unions in C I didn't see these>>
3797 @deffn @code{.stab?} N_SCOPE
3799 Modula2 scope information (Sun linker)
3804 @section Non-base registers on Gould systems
3806 @deffn @code{.stab?} N_NBTEXT
3807 @deffnx @code{.stab?} N_NBDATA
3808 @deffnx @code{.stab?} N_NBBSS
3809 @deffnx @code{.stab?} N_NBSTS
3810 @deffnx @code{.stab?} N_NBLCS
3816 These are used on Gould systems for non-base registers syms.
3818 However, the following values are not the values used by Gould; they are
3819 the values which GNU has been documenting for these values for a long
3820 time, without actually checking what Gould uses. I include these values
3821 only because perhaps some someone actually did something with the GNU
3822 information (I hope not, why GNU knowingly assigned wrong values to
3823 these in the header file is a complete mystery to me).
3826 240 0xf0 N_NBTEXT ??
3827 242 0xf2 N_NBDATA ??
3837 @deffn @code{.stabn} N_LENG
3839 Second symbol entry containing a length-value for the preceding entry.
3840 The value is the length.
3844 @appendix Questions and Anomalies
3848 @c I think this is changed in GCC 2.4.5 to put the line number there.
3849 For GNU C stabs defining local and global variables (@code{N_LSYM} and
3850 @code{N_GSYM}), the desc field is supposed to contain the source
3851 line number on which the variable is defined. In reality the desc
3852 field is always 0. (This behavior is defined in @file{dbxout.c} and
3853 putting a line number in desc is controlled by @samp{#ifdef
3854 WINNING_GDB}, which defaults to false). GDB supposedly uses this
3855 information if you say @samp{list @var{var}}. In reality, @var{var} can
3856 be a variable defined in the program and GDB says @samp{function
3857 @var{var} not defined}.
3860 In GNU C stabs, there seems to be no way to differentiate tag types:
3861 structures, unions, and enums (symbol descriptor @samp{T}) and typedefs
3862 (symbol descriptor @samp{t}) defined at file scope from types defined locally
3863 to a procedure or other more local scope. They all use the @code{N_LSYM}
3864 stab type. Types defined at procedure scope are emitted after the
3865 @code{N_RBRAC} of the preceding function and before the code of the
3866 procedure in which they are defined. This is exactly the same as
3867 types defined in the source file between the two procedure bodies.
3868 GDB over-compensates by placing all types in block #1, the block for
3869 symbols of file scope. This is true for default, @samp{-ansi} and
3870 @samp{-traditional} compiler options. (Bugs gcc/1063, gdb/1066.)
3873 What ends the procedure scope? Is it the proc block's @code{N_RBRAC} or the
3874 next @code{N_FUN}? (I believe its the first.)
3878 @appendix Using Stabs in Their Own Sections
3880 Many object file formats allow tools to create object files with custom
3881 sections containing any arbitrary data. For any such object file
3882 format, stabs can be embedded in special sections. This is how stabs
3883 are used with ELF and SOM, and aside from ECOFF and XCOFF, is how stabs
3887 * Stab Section Basics:: How to embed stabs in sections
3888 * ELF Linker Relocation:: Sun ELF hacks
3891 @node Stab Section Basics
3892 @appendixsec How to Embed Stabs in Sections
3894 The assembler creates two custom sections, a section named @code{.stab}
3895 which contains an array of fixed length structures, one struct per stab,
3896 and a section named @code{.stabstr} containing all the variable length
3897 strings that are referenced by stabs in the @code{.stab} section. The
3898 byte order of the stabs binary data depends on the object file format.
3899 For ELF, it matches the byte order of the ELF file itself, as determined
3900 from the @code{EI_DATA} field in the @code{e_ident} member of the ELF
3901 header. For SOM, it is always big-endian (is this true??? FIXME). For
3902 COFF, it matches the byte order of the COFF headers. The meaning of the
3903 fields is the same as for a.out (@pxref{Symbol Table Format}), except
3904 that the @code{n_strx} field is relative to the strings for the current
3905 compilation unit (which can be found using the synthetic N_UNDF stab
3906 described below), rather than the entire string table.
3908 The first stab in the @code{.stab} section for each compilation unit is
3909 synthetic, generated entirely by the assembler, with no corresponding
3910 @code{.stab} directive as input to the assembler. This stab contains
3911 the following fields:
3915 Offset in the @code{.stabstr} section to the source filename.
3921 Unused field, always zero.
3922 This may eventually be used to hold overflows from the count in
3923 the @code{n_desc} field.
3926 Count of upcoming symbols, i.e., the number of remaining stabs for this
3930 Size of the string table fragment associated with this source file, in
3934 The @code{.stabstr} section always starts with a null byte (so that string
3935 offsets of zero reference a null string), followed by random length strings,
3936 each of which is null byte terminated.
3938 The ELF section header for the @code{.stab} section has its
3939 @code{sh_link} member set to the section number of the @code{.stabstr}
3940 section, and the @code{.stabstr} section has its ELF section
3941 header @code{sh_type} member set to @code{SHT_STRTAB} to mark it as a
3942 string table. SOM and COFF have no way of linking the sections together
3943 or marking them as string tables.
3945 For COFF, the @code{.stab} and @code{.stabstr} sections may be simply
3946 concatenated by the linker. GDB then uses the @code{n_desc} fields to
3947 figure out the extent of the original sections. Similarly, the
3948 @code{n_value} fields of the header symbols are added together in order
3949 to get the actual position of the strings in a desired @code{.stabstr}
3950 section. Although this design obviates any need for the linker to
3951 relocate or otherwise manipulate @code{.stab} and @code{.stabstr}
3952 sections, it also requires some care to ensure that the offsets are
3953 calculated correctly. For instance, if the linker were to pad in
3954 between the @code{.stabstr} sections before concatenating, then the
3955 offsets to strings in the middle of the executable's @code{.stabstr}
3956 section would be wrong.
3958 The GNU linker is able to optimize stabs information by merging
3959 duplicate strings and removing duplicate header file information
3960 (@pxref{Include Files}). When some versions of the GNU linker optimize
3961 stabs in sections, they remove the leading @code{N_UNDF} symbol and
3962 arranges for all the @code{n_strx} fields to be relative to the start of
3963 the @code{.stabstr} section.
3965 @node ELF Linker Relocation
3966 @appendixsec Having the Linker Relocate Stabs in ELF
3968 This section describes some Sun hacks for Stabs in ELF; it does not
3969 apply to COFF or SOM.
3971 To keep linking fast, you don't want the linker to have to relocate very
3972 many stabs. Making sure this is done for @code{N_SLINE},
3973 @code{N_RBRAC}, and @code{N_LBRAC} stabs is the most important thing
3974 (see the descriptions of those stabs for more information). But Sun's
3975 stabs in ELF has taken this further, to make all addresses in the
3976 @code{n_value} field (functions and static variables) relative to the
3977 source file. For the @code{N_SO} symbol itself, Sun simply omits the
3978 address. To find the address of each section corresponding to a given
3979 source file, the compiler puts out symbols giving the address of each
3980 section for a given source file. Since these are ELF (not stab)
3981 symbols, the linker relocates them correctly without having to touch the
3982 stabs section. They are named @code{Bbss.bss} for the bss section,
3983 @code{Ddata.data} for the data section, and @code{Drodata.rodata} for
3984 the rodata section. For the text section, there is no such symbol (but
3985 there should be, see below). For an example of how these symbols work,
3986 @xref{Stab Section Transformations}. GCC does not provide these symbols;
3987 it instead relies on the stabs getting relocated. Thus addresses which
3988 would normally be relative to @code{Bbss.bss}, etc., are already
3989 relocated. The Sun linker provided with Solaris 2.2 and earlier
3990 relocates stabs using normal ELF relocation information, as it would do
3991 for any section. Sun has been threatening to kludge their linker to not
3992 do this (to speed up linking), even though the correct way to avoid
3993 having the linker do these relocations is to have the compiler no longer
3994 output relocatable values. Last I heard they had been talked out of the
3995 linker kludge. See Sun point patch 101052-01 and Sun bug 1142109. With
3996 the Sun compiler this affects @samp{S} symbol descriptor stabs
3997 (@pxref{Statics}) and functions (@pxref{Procedures}). In the latter
3998 case, to adopt the clean solution (making the value of the stab relative
3999 to the start of the compilation unit), it would be necessary to invent a
4000 @code{Ttext.text} symbol, analogous to the @code{Bbss.bss}, etc.,
4001 symbols. I recommend this rather than using a zero value and getting
4002 the address from the ELF symbols.
4004 Finding the correct @code{Bbss.bss}, etc., symbol is difficult, because
4005 the linker simply concatenates the @code{.stab} sections from each
4006 @file{.o} file without including any information about which part of a
4007 @code{.stab} section comes from which @file{.o} file. The way GDB does
4008 this is to look for an ELF @code{STT_FILE} symbol which has the same
4009 name as the last component of the file name from the @code{N_SO} symbol
4010 in the stabs (for example, if the file name is @file{../../gdb/main.c},
4011 it looks for an ELF @code{STT_FILE} symbol named @code{main.c}). This
4012 loses if different files have the same name (they could be in different
4013 directories, a library could have been copied from one system to
4014 another, etc.). It would be much cleaner to have the @code{Bbss.bss}
4015 symbols in the stabs themselves. Having the linker relocate them there
4016 is no more work than having the linker relocate ELF symbols, and it
4017 solves the problem of having to associate the ELF and stab symbols.
4018 However, no one has yet designed or implemented such a scheme.
4020 @node Symbol Types Index
4021 @unnumbered Symbol Types Index
4025 @c TeX can handle the contents at the start but makeinfo 3.12 can not