2 @setfilename stabs.info
8 @c This is a dir.info fragment to support semi-automated addition of
9 @c manuals to an info tree.
10 @dircategory Software development
12 * Stabs: (stabs). The "stabs" debugging information format.
16 Copyright @copyright{} 1992, 1993, 1994, 1995, 1997, 1998, 2000, 2001,
17 2002, 2003, 2004, 2005, 2006, 2007, 2009, 2010
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 no
25 Invariant Sections, with no Front-Cover Texts, and with no Back-Cover
26 Texts. A copy of the license is included in the section entitled ``GNU
27 Free Documentation License''.
31 This document describes the stabs debugging symbol tables.
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: 2.130 $} % For use in headers, footers too
45 \hfill Cygnus Support\par
47 \hfill \TeX{}info \texinfoversion\par
51 @vskip 0pt plus 1filll
57 @top The "stabs" representation of debugging information
59 This document describes the stabs debugging format.
62 * Overview:: Overview of stabs
63 * Program Structure:: Encoding of the structure of the program
64 * Constants:: Constants
66 * Types:: Type definitions
67 * Macro define and undefine:: Representation of #define and #undef
68 * Symbol Tables:: Symbol information in symbol tables
69 * Cplusplus:: Stabs specific to C++
70 * Stab Types:: Symbol types in a.out files
71 * Symbol Descriptors:: Table of symbol descriptors
72 * Type Descriptors:: Table of type descriptors
73 * Expanded Reference:: Reference information by stab type
74 * Questions:: Questions and anomalies
75 * Stab Sections:: In some object file formats, stabs are
77 * Symbol Types Index:: Index of symbolic stab symbol type names.
78 * GNU Free Documentation License:: The license for this documentation
85 @chapter Overview of Stabs
87 @dfn{Stabs} refers to a format for information that describes a program
88 to a debugger. This format was apparently invented by
90 the University of California at Berkeley, for the @code{pdx} Pascal
91 debugger; the format has spread widely since then.
93 This document is one of the few published sources of documentation on
94 stabs. It is believed to be comprehensive for stabs used by C. The
95 lists of symbol descriptors (@pxref{Symbol Descriptors}) and type
96 descriptors (@pxref{Type Descriptors}) are believed to be completely
97 comprehensive. Stabs for COBOL-specific features and for variant
98 records (used by Pascal and Modula-2) are poorly documented here.
100 @c FIXME: Need to document all OS9000 stuff in GDB; see all references
101 @c to os9k_stabs in stabsread.c.
103 Other sources of information on stabs are @cite{Dbx and Dbxtool
104 Interfaces}, 2nd edition, by Sun, 1988, and @cite{AIX Version 3.2 Files
105 Reference}, Fourth Edition, September 1992, "dbx Stabstring Grammar" in
106 the a.out section, page 2-31. This document is believed to incorporate
107 the information from those two sources except where it explicitly directs
108 you to them for more information.
111 * Flow:: Overview of debugging information flow
112 * Stabs Format:: Overview of stab format
113 * String Field:: The string field
114 * C Example:: A simple example in C source
115 * Assembly Code:: The simple example at the assembly level
119 @section Overview of Debugging Information Flow
121 The GNU C compiler compiles C source in a @file{.c} file into assembly
122 language in a @file{.s} file, which the assembler translates into
123 a @file{.o} file, which the linker combines with other @file{.o} files and
124 libraries to produce an executable file.
126 With the @samp{-g} option, GCC puts in the @file{.s} file additional
127 debugging information, which is slightly transformed by the assembler
128 and linker, and carried through into the final executable. This
129 debugging information describes features of the source file like line
130 numbers, the types and scopes of variables, and function names,
131 parameters, and scopes.
133 For some object file formats, the debugging information is encapsulated
134 in assembler directives known collectively as @dfn{stab} (symbol table)
135 directives, which are interspersed with the generated code. Stabs are
136 the native format for debugging information in the a.out and XCOFF
137 object file formats. The GNU tools can also emit stabs in the COFF and
138 ECOFF object file formats.
140 The assembler adds the information from stabs to the symbol information
141 it places by default in the symbol table and the string table of the
142 @file{.o} file it is building. The linker consolidates the @file{.o}
143 files into one executable file, with one symbol table and one string
144 table. Debuggers use the symbol and string tables in the executable as
145 a source of debugging information about the program.
148 @section Overview of Stab Format
150 There are three overall formats for stab assembler directives,
151 differentiated by the first word of the stab. The name of the directive
152 describes which combination of four possible data fields follows. It is
153 either @code{.stabs} (string), @code{.stabn} (number), or @code{.stabd}
154 (dot). IBM's XCOFF assembler uses @code{.stabx} (and some other
155 directives such as @code{.file} and @code{.bi}) instead of
156 @code{.stabs}, @code{.stabn} or @code{.stabd}.
158 The overall format of each class of stab is:
161 .stabs "@var{string}",@var{type},@var{other},@var{desc},@var{value}
162 .stabn @var{type},@var{other},@var{desc},@var{value}
163 .stabd @var{type},@var{other},@var{desc}
164 .stabx "@var{string}",@var{value},@var{type},@var{sdb-type}
167 @c what is the correct term for "current file location"? My AIX
168 @c assembler manual calls it "the value of the current location counter".
169 For @code{.stabn} and @code{.stabd}, there is no @var{string} (the
170 @code{n_strx} field is zero; see @ref{Symbol Tables}). For
171 @code{.stabd}, the @var{value} field is implicit and has the value of
172 the current file location. For @code{.stabx}, the @var{sdb-type} field
173 is unused for stabs and can always be set to zero. The @var{other}
174 field is almost always unused and can be set to zero.
176 The number in the @var{type} field gives some basic information about
177 which type of stab this is (or whether it @emph{is} a stab, as opposed
178 to an ordinary symbol). Each valid type number defines a different stab
179 type; further, the stab type defines the exact interpretation of, and
180 possible values for, any remaining @var{string}, @var{desc}, or
181 @var{value} fields present in the stab. @xref{Stab Types}, for a list
182 in numeric order of the valid @var{type} field values for stab directives.
185 @section The String Field
187 For most stabs the string field holds the meat of the
188 debugging information. The flexible nature of this field
189 is what makes stabs extensible. For some stab types the string field
190 contains only a name. For other stab types the contents can be a great
193 The overall format of the string field for most stab types is:
196 "@var{name}:@var{symbol-descriptor} @var{type-information}"
199 @var{name} is the name of the symbol represented by the stab; it can
200 contain a pair of colons (@pxref{Nested Symbols}). @var{name} can be
201 omitted, which means the stab represents an unnamed object. For
202 example, @samp{:t10=*2} defines type 10 as a pointer to type 2, but does
203 not give the type a name. Omitting the @var{name} field is supported by
204 AIX dbx and GDB after about version 4.8, but not other debuggers. GCC
205 sometimes uses a single space as the name instead of omitting the name
206 altogether; apparently that is supported by most debuggers.
208 The @var{symbol-descriptor} following the @samp{:} is an alphabetic
209 character that tells more specifically what kind of symbol the stab
210 represents. If the @var{symbol-descriptor} is omitted, but type
211 information follows, then the stab represents a local variable. For a
212 list of symbol descriptors, see @ref{Symbol Descriptors}. The @samp{c}
213 symbol descriptor is an exception in that it is not followed by type
214 information. @xref{Constants}.
216 @var{type-information} is either a @var{type-number}, or
217 @samp{@var{type-number}=}. A @var{type-number} alone is a type
218 reference, referring directly to a type that has already been defined.
220 The @samp{@var{type-number}=} form is a type definition, where the
221 number represents a new type which is about to be defined. The type
222 definition may refer to other types by number, and those type numbers
223 may be followed by @samp{=} and nested definitions. Also, the Lucid
224 compiler will repeat @samp{@var{type-number}=} more than once if it
225 wants to define several type numbers at once.
227 In a type definition, if the character that follows the equals sign is
228 non-numeric then it is a @var{type-descriptor}, and tells what kind of
229 type is about to be defined. Any other values following the
230 @var{type-descriptor} vary, depending on the @var{type-descriptor}.
231 @xref{Type Descriptors}, for a list of @var{type-descriptor} values. If
232 a number follows the @samp{=} then the number is a @var{type-reference}.
233 For a full description of types, @ref{Types}.
235 A @var{type-number} is often a single number. The GNU and Sun tools
236 additionally permit a @var{type-number} to be a pair
237 (@var{file-number},@var{filetype-number}) (the parentheses appear in the
238 string, and serve to distinguish the two cases). The @var{file-number}
239 is 0 for the base source file, 1 for the first included file, 2 for the
240 next, and so on. The @var{filetype-number} is a number starting with
241 1 which is incremented for each new type defined in the file.
242 (Separating the file number and the type number permits the
243 @code{N_BINCL} optimization to succeed more often; see @ref{Include
246 There is an AIX extension for type attributes. Following the @samp{=}
247 are any number of type attributes. Each one starts with @samp{@@} and
248 ends with @samp{;}. Debuggers, including AIX's dbx and GDB 4.10, skip
249 any type attributes they do not recognize. GDB 4.9 and other versions
250 of dbx may not do this. Because of a conflict with C@t{++}
251 (@pxref{Cplusplus}), new attributes should not be defined which begin
252 with a digit, @samp{(}, or @samp{-}; GDB may be unable to distinguish
253 those from the C@t{++} type descriptor @samp{@@}. The attributes are:
256 @item a@var{boundary}
257 @var{boundary} is an integer specifying the alignment. I assume it
258 applies to all variables of this type.
261 Pointer class (for checking). Not sure what this means, or how
262 @var{integer} is interpreted.
265 Indicate this is a packed type, meaning that structure fields or array
266 elements are placed more closely in memory, to save memory at the
270 Size in bits of a variable of this type. This is fully supported by GDB
274 Indicate that this type is a string instead of an array of characters,
275 or a bitstring instead of a set. It doesn't change the layout of the
276 data being represented, but does enable the debugger to know which type
280 Indicate that this type is a vector instead of an array. The only
281 major difference between vectors and arrays is that vectors are
282 passed by value instead of by reference (vector coprocessor extension).
286 All of this can make the string field quite long. All versions of GDB,
287 and some versions of dbx, can handle arbitrarily long strings. But many
288 versions of dbx (or assemblers or linkers, I'm not sure which)
289 cretinously limit the strings to about 80 characters, so compilers which
290 must work with such systems need to split the @code{.stabs} directive
291 into several @code{.stabs} directives. Each stab duplicates every field
292 except the string field. The string field of every stab except the last
293 is marked as continued with a backslash at the end (in the assembly code
294 this may be written as a double backslash, depending on the assembler).
295 Removing the backslashes and concatenating the string fields of each
296 stab produces the original, long string. Just to be incompatible (or so
297 they don't have to worry about what the assembler does with
298 backslashes), AIX can use @samp{?} instead of backslash.
301 @section A Simple Example in C Source
303 To get the flavor of how stabs describe source information for a C
304 program, let's look at the simple program:
309 printf("Hello world");
313 When compiled with @samp{-g}, the program above yields the following
314 @file{.s} file. Line numbers have been added to make it easier to refer
315 to parts of the @file{.s} file in the description of the stabs that
319 @section The Simple Example at the Assembly Level
321 This simple ``hello world'' example demonstrates several of the stab
322 types used to describe C language source files.
326 2 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0
327 3 .stabs "hello.c",100,0,0,Ltext0
330 6 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0
331 7 .stabs "char:t2=r2;0;127;",128,0,0,0
332 8 .stabs "long int:t3=r1;-2147483648;2147483647;",128,0,0,0
333 9 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
334 10 .stabs "long unsigned int:t5=r1;0;-1;",128,0,0,0
335 11 .stabs "short int:t6=r1;-32768;32767;",128,0,0,0
336 12 .stabs "long long int:t7=r1;0;-1;",128,0,0,0
337 13 .stabs "short unsigned int:t8=r1;0;65535;",128,0,0,0
338 14 .stabs "long long unsigned int:t9=r1;0;-1;",128,0,0,0
339 15 .stabs "signed char:t10=r1;-128;127;",128,0,0,0
340 16 .stabs "unsigned char:t11=r1;0;255;",128,0,0,0
341 17 .stabs "float:t12=r1;4;0;",128,0,0,0
342 18 .stabs "double:t13=r1;8;0;",128,0,0,0
343 19 .stabs "long double:t14=r1;8;0;",128,0,0,0
344 20 .stabs "void:t15=15",128,0,0,0
347 23 .ascii "Hello, world!\12\0"
362 38 sethi %hi(LC0),%o1
363 39 or %o1,%lo(LC0),%o0
374 50 .stabs "main:F1",36,0,0,_main
375 51 .stabn 192,0,0,LBB2
376 52 .stabn 224,0,0,LBE2
379 @node Program Structure
380 @chapter Encoding the Structure of the Program
382 The elements of the program structure that stabs encode include the name
383 of the main function, the names of the source and include files, the
384 line numbers, procedure names and types, and the beginnings and ends of
388 * Main Program:: Indicate what the main program is
389 * Source Files:: The path and name of the source file
390 * Include Files:: Names of include files
393 * Nested Procedures::
395 * Alternate Entry Points:: Entering procedures except at the beginning.
399 @section Main Program
402 Most languages allow the main program to have any name. The
403 @code{N_MAIN} stab type tells the debugger the name that is used in this
404 program. Only the string field is significant; it is the name of
405 a function which is the main program. Most C compilers do not use this
406 stab (they expect the debugger to assume that the name is @code{main}),
407 but some C compilers emit an @code{N_MAIN} stab for the @code{main}
408 function. I'm not sure how XCOFF handles this.
411 @section Paths and Names of the Source Files
414 Before any other stabs occur, there must be a stab specifying the source
415 file. This information is contained in a symbol of stab type
416 @code{N_SO}; the string field contains the name of the file. The
417 value of the symbol is the start address of the portion of the
418 text section corresponding to that file.
420 Some compilers use the desc field to indicate the language of the
421 source file. Sun's compilers started this usage, and the first
422 constants are derived from their documentation. Languages added
423 by gcc/gdb start at 0x32 to avoid conflict with languages Sun may
424 add in the future. A desc field with a value 0 indicates that no
425 language has been specified via this mechanism.
428 @item @code{N_SO_AS} (0x1)
430 @item @code{N_SO_C} (0x2)
432 @item @code{N_SO_ANSI_C} (0x3)
434 @item @code{N_SO_CC} (0x4)
436 @item @code{N_SO_FORTRAN} (0x5)
438 @item @code{N_SO_PASCAL} (0x6)
440 @item @code{N_SO_FORTRAN90} (0x7)
442 @item @code{N_SO_OBJC} (0x32)
444 @item @code{N_SO_OBJCPLUS} (0x33)
448 Some compilers (for example, GCC2 and SunOS4 @file{/bin/cc}) also
449 include the directory in which the source was compiled, in a second
450 @code{N_SO} symbol preceding the one containing the file name. This
451 symbol can be distinguished by the fact that it ends in a slash. Code
452 from the @code{cfront} C@t{++} compiler can have additional @code{N_SO} symbols for
453 nonexistent source files after the @code{N_SO} for the real source file;
454 these are believed to contain no useful information.
459 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0 # @r{100 is N_SO}
460 .stabs "hello.c",100,0,0,Ltext0
466 Instead of @code{N_SO} symbols, XCOFF uses a @code{.file} assembler
467 directive which assembles to a @code{C_FILE} symbol; explaining this in
468 detail is outside the scope of this document.
470 @c FIXME: Exactly when should the empty N_SO be used? Why?
471 If it is useful to indicate the end of a source file, this is done with
472 an @code{N_SO} symbol with an empty string for the name. The value is
473 the address of the end of the text section for the file. For some
474 systems, there is no indication of the end of a source file, and you
475 just need to figure it ended when you see an @code{N_SO} for a different
476 source file, or a symbol ending in @code{.o} (which at least some
477 linkers insert to mark the start of a new @code{.o} file).
480 @section Names of Include Files
482 There are several schemes for dealing with include files: the
483 traditional @code{N_SOL} approach, Sun's @code{N_BINCL} approach, and the
484 XCOFF @code{C_BINCL} approach (which despite the similar name has little in
485 common with @code{N_BINCL}).
488 An @code{N_SOL} symbol specifies which include file subsequent symbols
489 refer to. The string field is the name of the file and the value is the
490 text address corresponding to the end of the previous include file and
491 the start of this one. To specify the main source file again, use an
492 @code{N_SOL} symbol with the name of the main source file.
497 The @code{N_BINCL} approach works as follows. An @code{N_BINCL} symbol
498 specifies the start of an include file. In an object file, only the
499 string is significant; the linker puts data into some of the other
500 fields. The end of the include file is marked by an @code{N_EINCL}
501 symbol (which has no string field). In an object file, there is no
502 significant data in the @code{N_EINCL} symbol. @code{N_BINCL} and
503 @code{N_EINCL} can be nested.
505 If the linker detects that two source files have identical stabs between
506 an @code{N_BINCL} and @code{N_EINCL} pair (as will generally be the case
507 for a header file), then it only puts out the stabs once. Each
508 additional occurrence is replaced by an @code{N_EXCL} symbol. I believe
509 the GNU linker and the Sun (both SunOS4 and Solaris) linker are the only
510 ones which supports this feature.
512 A linker which supports this feature will set the value of a
513 @code{N_BINCL} symbol to the total of all the characters in the stabs
514 strings included in the header file, omitting any file numbers. The
515 value of an @code{N_EXCL} symbol is the same as the value of the
516 @code{N_BINCL} symbol it replaces. This information can be used to
517 match up @code{N_EXCL} and @code{N_BINCL} symbols which have the same
518 filename. The @code{N_EINCL} value, and the values of the other and
519 description fields for all three, appear to always be zero.
523 For the start of an include file in XCOFF, use the @file{.bi} assembler
524 directive, which generates a @code{C_BINCL} symbol. A @file{.ei}
525 directive, which generates a @code{C_EINCL} symbol, denotes the end of
526 the include file. Both directives are followed by the name of the
527 source file in quotes, which becomes the string for the symbol.
528 The value of each symbol, produced automatically by the assembler
529 and linker, is the offset into the executable of the beginning
530 (inclusive, as you'd expect) or end (inclusive, as you would not expect)
531 of the portion of the COFF line table that corresponds to this include
532 file. @code{C_BINCL} and @code{C_EINCL} do not nest.
535 @section Line Numbers
538 An @code{N_SLINE} symbol represents the start of a source line. The
539 desc field contains the line number and the value contains the code
540 address for the start of that source line. On most machines the address
541 is absolute; for stabs in sections (@pxref{Stab Sections}), it is
542 relative to the function in which the @code{N_SLINE} symbol occurs.
546 GNU documents @code{N_DSLINE} and @code{N_BSLINE} symbols for line
547 numbers in the data or bss segments, respectively. They are identical
548 to @code{N_SLINE} but are relocated differently by the linker. They
549 were intended to be used to describe the source location of a variable
550 declaration, but I believe that GCC2 actually puts the line number in
551 the desc field of the stab for the variable itself. GDB has been
552 ignoring these symbols (unless they contain a string field) since
555 For single source lines that generate discontiguous code, such as flow
556 of control statements, there may be more than one line number entry for
557 the same source line. In this case there is a line number entry at the
558 start of each code range, each with the same line number.
560 XCOFF does not use stabs for line numbers. Instead, it uses COFF line
561 numbers (which are outside the scope of this document). Standard COFF
562 line numbers cannot deal with include files, but in XCOFF this is fixed
563 with the @code{C_BINCL} method of marking include files (@pxref{Include
569 @findex N_FUN, for functions
571 @findex N_STSYM, for functions (Sun acc)
572 @findex N_GSYM, for functions (Sun acc)
573 All of the following stabs normally use the @code{N_FUN} symbol type.
574 However, Sun's @code{acc} compiler on SunOS4 uses @code{N_GSYM} and
575 @code{N_STSYM}, which means that the value of the stab for the function
576 is useless and the debugger must get the address of the function from
577 the non-stab symbols instead. On systems where non-stab symbols have
578 leading underscores, the stabs will lack underscores and the debugger
579 needs to know about the leading underscore to match up the stab and the
580 non-stab symbol. BSD Fortran is said to use @code{N_FNAME} with the
581 same restriction; the value of the symbol is not useful (I'm not sure it
582 really does use this, because GDB doesn't handle this and no one has
586 A function is represented by an @samp{F} symbol descriptor for a global
587 (extern) function, and @samp{f} for a static (local) function. For
588 a.out, the value of the symbol is the address of the start of the
589 function; it is already relocated. For stabs in ELF, the SunPRO
590 compiler version 2.0.1 and GCC put out an address which gets relocated
591 by the linker. In a future release SunPRO is planning to put out zero,
592 in which case the address can be found from the ELF (non-stab) symbol.
593 Because looking things up in the ELF symbols would probably be slow, I'm
594 not sure how to find which symbol of that name is the right one, and
595 this doesn't provide any way to deal with nested functions, it would
596 probably be better to make the value of the stab an address relative to
597 the start of the file, or just absolute. See @ref{ELF Linker
598 Relocation} for more information on linker relocation of stabs in ELF
599 files. For XCOFF, the stab uses the @code{C_FUN} storage class and the
600 value of the stab is meaningless; the address of the function can be
601 found from the csect symbol (XTY_LD/XMC_PR).
603 The type information of the stab represents the return type of the
604 function; thus @samp{foo:f5} means that foo is a function returning type
605 5. There is no need to try to get the line number of the start of the
606 function from the stab for the function; it is in the next
607 @code{N_SLINE} symbol.
609 @c FIXME: verify whether the "I suspect" below is true or not.
610 Some compilers (such as Sun's Solaris compiler) support an extension for
611 specifying the types of the arguments. I suspect this extension is not
612 used for old (non-prototyped) function definitions in C. If the
613 extension is in use, the type information of the stab for the function
614 is followed by type information for each argument, with each argument
615 preceded by @samp{;}. An argument type of 0 means that additional
616 arguments are being passed, whose types and number may vary (@samp{...}
617 in ANSI C). GDB has tolerated this extension (parsed the syntax, if not
618 necessarily used the information) since at least version 4.8; I don't
619 know whether all versions of dbx tolerate it. The argument types given
620 here are not redundant with the symbols for the formal parameters
621 (@pxref{Parameters}); they are the types of the arguments as they are
622 passed, before any conversions might take place. For example, if a C
623 function which is declared without a prototype takes a @code{float}
624 argument, the value is passed as a @code{double} but then converted to a
625 @code{float}. Debuggers need to use the types given in the arguments
626 when printing values, but when calling the function they need to use the
627 types given in the symbol defining the function.
629 If the return type and types of arguments of a function which is defined
630 in another source file are specified (i.e., a function prototype in ANSI
631 C), traditionally compilers emit no stab; the only way for the debugger
632 to find the information is if the source file where the function is
633 defined was also compiled with debugging symbols. As an extension the
634 Solaris compiler uses symbol descriptor @samp{P} followed by the return
635 type of the function, followed by the arguments, each preceded by
636 @samp{;}, as in a stab with symbol descriptor @samp{f} or @samp{F}.
637 This use of symbol descriptor @samp{P} can be distinguished from its use
638 for register parameters (@pxref{Register Parameters}) by the fact that it has
639 symbol type @code{N_FUN}.
641 The AIX documentation also defines symbol descriptor @samp{J} as an
642 internal function. I assume this means a function nested within another
643 function. It also says symbol descriptor @samp{m} is a module in
644 Modula-2 or extended Pascal.
646 Procedures (functions which do not return values) are represented as
647 functions returning the @code{void} type in C. I don't see why this couldn't
648 be used for all languages (inventing a @code{void} type for this purpose if
649 necessary), but the AIX documentation defines @samp{I}, @samp{P}, and
650 @samp{Q} for internal, global, and static procedures, respectively.
651 These symbol descriptors are unusual in that they are not followed by
654 The following example shows a stab for a function @code{main} which
655 returns type number @code{1}. The @code{_main} specified for the value
656 is a reference to an assembler label which is used to fill in the start
657 address of the function.
660 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
663 The stab representing a procedure is located immediately following the
664 code of the procedure. This stab is in turn directly followed by a
665 group of other stabs describing elements of the procedure. These other
666 stabs describe the procedure's parameters, its block local variables, and
669 If functions can appear in different sections, then the debugger may not
670 be able to find the end of a function. Recent versions of GCC will mark
671 the end of a function with an @code{N_FUN} symbol with an empty string
672 for the name. The value is the address of the end of the current
673 function. Without such a symbol, there is no indication of the address
674 of the end of a function, and you must assume that it ended at the
675 starting address of the next function or at the end of the text section
678 @node Nested Procedures
679 @section Nested Procedures
681 For any of the symbol descriptors representing procedures, after the
682 symbol descriptor and the type information is optionally a scope
683 specifier. This consists of a comma, the name of the procedure, another
684 comma, and the name of the enclosing procedure. The first name is local
685 to the scope specified, and seems to be redundant with the name of the
686 symbol (before the @samp{:}). This feature is used by GCC, and
687 presumably Pascal, Modula-2, etc., compilers, for nested functions.
689 If procedures are nested more than one level deep, only the immediately
690 containing scope is specified. For example, this code:
702 return baz (x + 2 * y);
704 return x + bar (3 * x);
712 .stabs "baz:f1,baz,bar",36,0,0,_baz.15 # @r{36 is N_FUN}
713 .stabs "bar:f1,bar,foo",36,0,0,_bar.12
714 .stabs "foo:F1",36,0,0,_foo
717 @node Block Structure
718 @section Block Structure
722 @c For GCC 2.5.8 or so stabs-in-coff, these are absolute instead of
723 @c function relative (as documented below). But GDB has never been able
724 @c to deal with that (it had wanted them to be relative to the file, but
725 @c I just fixed that (between GDB 4.12 and 4.13)), so it is function
726 @c relative just like ELF and SOM and the below documentation.
727 The program's block structure is represented by the @code{N_LBRAC} (left
728 brace) and the @code{N_RBRAC} (right brace) stab types. The variables
729 defined inside a block precede the @code{N_LBRAC} symbol for most
730 compilers, including GCC. Other compilers, such as the Convex, Acorn
731 RISC machine, and Sun @code{acc} compilers, put the variables after the
732 @code{N_LBRAC} symbol. The values of the @code{N_LBRAC} and
733 @code{N_RBRAC} symbols are the start and end addresses of the code of
734 the block, respectively. For most machines, they are relative to the
735 starting address of this source file. For the Gould NP1, they are
736 absolute. For stabs in sections (@pxref{Stab Sections}), they are
737 relative to the function in which they occur.
739 The @code{N_LBRAC} and @code{N_RBRAC} stabs that describe the block
740 scope of a procedure are located after the @code{N_FUN} stab that
741 represents the procedure itself.
743 Sun documents the desc field of @code{N_LBRAC} and
744 @code{N_RBRAC} symbols as containing the nesting level of the block.
745 However, dbx seems to not care, and GCC always sets desc to
751 For XCOFF, block scope is indicated with @code{C_BLOCK} symbols. If the
752 name of the symbol is @samp{.bb}, then it is the beginning of the block;
753 if the name of the symbol is @samp{.be}; it is the end of the block.
755 @node Alternate Entry Points
756 @section Alternate Entry Points
760 Some languages, like Fortran, have the ability to enter procedures at
761 some place other than the beginning. One can declare an alternate entry
762 point. The @code{N_ENTRY} stab is for this; however, the Sun FORTRAN
763 compiler doesn't use it. According to AIX documentation, only the name
764 of a @code{C_ENTRY} stab is significant; the address of the alternate
765 entry point comes from the corresponding external symbol. A previous
766 revision of this document said that the value of an @code{N_ENTRY} stab
767 was the address of the alternate entry point, but I don't know the
768 source for that information.
773 The @samp{c} symbol descriptor indicates that this stab represents a
774 constant. This symbol descriptor is an exception to the general rule
775 that symbol descriptors are followed by type information. Instead, it
776 is followed by @samp{=} and one of the following:
780 Boolean constant. @var{value} is a numeric value; I assume it is 0 for
784 Character constant. @var{value} is the numeric value of the constant.
786 @item e @var{type-information} , @var{value}
787 Constant whose value can be represented as integral.
788 @var{type-information} is the type of the constant, as it would appear
789 after a symbol descriptor (@pxref{String Field}). @var{value} is the
790 numeric value of the constant. GDB 4.9 does not actually get the right
791 value if @var{value} does not fit in a host @code{int}, but it does not
792 do anything violent, and future debuggers could be extended to accept
793 integers of any size (whether unsigned or not). This constant type is
794 usually documented as being only for enumeration constants, but GDB has
795 never imposed that restriction; I don't know about other debuggers.
798 Integer constant. @var{value} is the numeric value. The type is some
799 sort of generic integer type (for GDB, a host @code{int}); to specify
800 the type explicitly, use @samp{e} instead.
803 Real constant. @var{value} is the real value, which can be @samp{INF}
804 (optionally preceded by a sign) for infinity, @samp{QNAN} for a quiet
805 NaN (not-a-number), or @samp{SNAN} for a signalling NaN. If it is a
806 normal number the format is that accepted by the C library function
810 String constant. @var{string} is a string enclosed in either @samp{'}
811 (in which case @samp{'} characters within the string are represented as
812 @samp{\'} or @samp{"} (in which case @samp{"} characters within the
813 string are represented as @samp{\"}).
815 @item S @var{type-information} , @var{elements} , @var{bits} , @var{pattern}
816 Set constant. @var{type-information} is the type of the constant, as it
817 would appear after a symbol descriptor (@pxref{String Field}).
818 @var{elements} is the number of elements in the set (does this means
819 how many bits of @var{pattern} are actually used, which would be
820 redundant with the type, or perhaps the number of bits set in
821 @var{pattern}? I don't get it), @var{bits} is the number of bits in the
822 constant (meaning it specifies the length of @var{pattern}, I think),
823 and @var{pattern} is a hexadecimal representation of the set. AIX
824 documentation refers to a limit of 32 bytes, but I see no reason why
825 this limit should exist. This form could probably be used for arbitrary
826 constants, not just sets; the only catch is that @var{pattern} should be
827 understood to be target, not host, byte order and format.
830 The boolean, character, string, and set constants are not supported by
831 GDB 4.9, but it ignores them. GDB 4.8 and earlier gave an error
832 message and refused to read symbols from the file containing the
835 The above information is followed by @samp{;}.
840 Different types of stabs describe the various ways that variables can be
841 allocated: on the stack, globally, in registers, in common blocks,
842 statically, or as arguments to a function.
845 * Stack Variables:: Variables allocated on the stack.
846 * Global Variables:: Variables used by more than one source file.
847 * Register Variables:: Variables in registers.
848 * Common Blocks:: Variables statically allocated together.
849 * Statics:: Variables local to one source file.
850 * Based Variables:: Fortran pointer based variables.
851 * Parameters:: Variables for arguments to functions.
854 @node Stack Variables
855 @section Automatic Variables Allocated on the Stack
857 If a variable's scope is local to a function and its lifetime is only as
858 long as that function executes (C calls such variables
859 @dfn{automatic}), it can be allocated in a register (@pxref{Register
860 Variables}) or on the stack.
862 @findex N_LSYM, for stack variables
864 Each variable allocated on the stack has a stab with the symbol
865 descriptor omitted. Since type information should begin with a digit,
866 @samp{-}, or @samp{(}, only those characters precluded from being used
867 for symbol descriptors. However, the Acorn RISC machine (ARM) is said
868 to get this wrong: it puts out a mere type definition here, without the
869 preceding @samp{@var{type-number}=}. This is a bad idea; there is no
870 guarantee that type descriptors are distinct from symbol descriptors.
871 Stabs for stack variables use the @code{N_LSYM} stab type, or
872 @code{C_LSYM} for XCOFF.
874 The value of the stab is the offset of the variable within the
875 local variables. On most machines this is an offset from the frame
876 pointer and is negative. The location of the stab specifies which block
877 it is defined in; see @ref{Block Structure}.
879 For example, the following C code:
889 produces the following stabs:
892 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
893 .stabs "x:1",128,0,0,-12 # @r{128 is N_LSYM}
894 .stabn 192,0,0,LBB2 # @r{192 is N_LBRAC}
895 .stabn 224,0,0,LBE2 # @r{224 is N_RBRAC}
898 See @ref{Procedures} for more information on the @code{N_FUN} stab, and
899 @ref{Block Structure} for more information on the @code{N_LBRAC} and
900 @code{N_RBRAC} stabs.
902 @node Global Variables
903 @section Global Variables
907 @c FIXME: verify for sure that it really is C_GSYM on XCOFF
908 A variable whose scope is not specific to just one source file is
909 represented by the @samp{G} symbol descriptor. These stabs use the
910 @code{N_GSYM} stab type (C_GSYM for XCOFF). The type information for
911 the stab (@pxref{String Field}) gives the type of the variable.
913 For example, the following source code:
920 yields the following assembly code:
923 .stabs "g_foo:G2",32,0,0,0 # @r{32 is N_GSYM}
930 The address of the variable represented by the @code{N_GSYM} is not
931 contained in the @code{N_GSYM} stab. The debugger gets this information
932 from the external symbol for the global variable. In the example above,
933 the @code{.global _g_foo} and @code{_g_foo:} lines tell the assembler to
934 produce an external symbol.
936 Some compilers, like GCC, output @code{N_GSYM} stabs only once, where
937 the variable is defined. Other compilers, like SunOS4 /bin/cc, output a
938 @code{N_GSYM} stab for each compilation unit which references the
941 @node Register Variables
942 @section Register Variables
946 @c According to an old version of this manual, AIX uses C_RPSYM instead
947 @c of C_RSYM. I am skeptical; this should be verified.
948 Register variables have their own stab type, @code{N_RSYM}
949 (@code{C_RSYM} for XCOFF), and their own symbol descriptor, @samp{r}.
950 The stab's value is the number of the register where the variable data
952 @c .stabs "name:type",N_RSYM,0,RegSize,RegNumber (Sun doc)
954 AIX defines a separate symbol descriptor @samp{d} for floating point
955 registers. This seems unnecessary; why not just just give floating
956 point registers different register numbers? I have not verified whether
957 the compiler actually uses @samp{d}.
959 If the register is explicitly allocated to a global variable, but not
963 register int g_bar asm ("%g5");
967 then the stab may be emitted at the end of the object file, with
968 the other bss symbols.
971 @section Common Blocks
973 A common block is a statically allocated section of memory which can be
974 referred to by several source files. It may contain several variables.
975 I believe Fortran is the only language with this feature.
981 A @code{N_BCOMM} stab begins a common block and an @code{N_ECOMM} stab
982 ends it. The only field that is significant in these two stabs is the
983 string, which names a normal (non-debugging) symbol that gives the
984 address of the common block. According to IBM documentation, only the
985 @code{N_BCOMM} has the name of the common block (even though their
986 compiler actually puts it both places).
990 The stabs for the members of the common block are between the
991 @code{N_BCOMM} and the @code{N_ECOMM}; the value of each stab is the
992 offset within the common block of that variable. IBM uses the
993 @code{C_ECOML} stab type, and there is a corresponding @code{N_ECOML}
994 stab type, but Sun's Fortran compiler uses @code{N_GSYM} instead. The
995 variables within a common block use the @samp{V} symbol descriptor (I
996 believe this is true of all Fortran variables). Other stabs (at least
997 type declarations using @code{C_DECL}) can also be between the
998 @code{N_BCOMM} and the @code{N_ECOMM}.
1001 @section Static Variables
1003 Initialized static variables are represented by the @samp{S} and
1004 @samp{V} symbol descriptors. @samp{S} means file scope static, and
1005 @samp{V} means procedure scope static. One exception: in XCOFF, IBM's
1006 xlc compiler always uses @samp{V}, and whether it is file scope or not
1007 is distinguished by whether the stab is located within a function.
1009 @c This is probably not worth mentioning; it is only true on the sparc
1010 @c for `double' variables which although declared const are actually in
1011 @c the data segment (the text segment can't guarantee 8 byte alignment).
1013 @c 2.4.5 has a bug in that it uses @code{N_FUN}, so neither dbx nor GDB can
1014 @c find the variables)
1017 @findex N_FUN, for variables
1019 In a.out files, @code{N_STSYM} means the data section, @code{N_FUN}
1020 means the text section, and @code{N_LCSYM} means the bss section. For
1021 those systems with a read-only data section separate from the text
1022 section (Solaris), @code{N_ROSYM} means the read-only data section.
1024 For example, the source lines:
1027 static const int var_const = 5;
1028 static int var_init = 2;
1029 static int var_noinit;
1033 yield the following stabs:
1036 .stabs "var_const:S1",36,0,0,_var_const # @r{36 is N_FUN}
1038 .stabs "var_init:S1",38,0,0,_var_init # @r{38 is N_STSYM}
1040 .stabs "var_noinit:S1",40,0,0,_var_noinit # @r{40 is N_LCSYM}
1046 In XCOFF files, the stab type need not indicate the section;
1047 @code{C_STSYM} can be used for all statics. Also, each static variable
1048 is enclosed in a static block. A @code{C_BSTAT} (emitted with a
1049 @samp{.bs} assembler directive) symbol begins the static block; its
1050 value is the symbol number of the csect symbol whose value is the
1051 address of the static block, its section is the section of the variables
1052 in that static block, and its name is @samp{.bs}. A @code{C_ESTAT}
1053 (emitted with a @samp{.es} assembler directive) symbol ends the static
1054 block; its name is @samp{.es} and its value and section are ignored.
1056 In ECOFF files, the storage class is used to specify the section, so the
1057 stab type need not indicate the section.
1059 In ELF files, for the SunPRO compiler version 2.0.1, symbol descriptor
1060 @samp{S} means that the address is absolute (the linker relocates it)
1061 and symbol descriptor @samp{V} means that the address is relative to the
1062 start of the relevant section for that compilation unit. SunPRO has
1063 plans to have the linker stop relocating stabs; I suspect that their the
1064 debugger gets the address from the corresponding ELF (not stab) symbol.
1065 I'm not sure how to find which symbol of that name is the right one.
1066 The clean way to do all this would be to have the value of a symbol
1067 descriptor @samp{S} symbol be an offset relative to the start of the
1068 file, just like everything else, but that introduces obvious
1069 compatibility problems. For more information on linker stab relocation,
1070 @xref{ELF Linker Relocation}.
1072 @node Based Variables
1073 @section Fortran Based Variables
1075 Fortran (at least, the Sun and SGI dialects of FORTRAN-77) has a feature
1076 which allows allocating arrays with @code{malloc}, but which avoids
1077 blurring the line between arrays and pointers the way that C does. In
1078 stabs such a variable uses the @samp{b} symbol descriptor.
1080 For example, the Fortran declarations
1083 real foo, foo10(10), foo10_5(10,5)
1085 pointer (foo10p, foo10)
1086 pointer (foo105p, foo10_5)
1094 foo10_5:bar3;1;5;ar3;1;10;6
1097 In this example, @code{real} is type 6 and type 3 is an integral type
1098 which is the type of the subscripts of the array (probably
1101 The @samp{b} symbol descriptor is like @samp{V} in that it denotes a
1102 statically allocated symbol whose scope is local to a function; see
1103 @xref{Statics}. The value of the symbol, instead of being the address
1104 of the variable itself, is the address of a pointer to that variable.
1105 So in the above example, the value of the @code{foo} stab is the address
1106 of a pointer to a real, the value of the @code{foo10} stab is the
1107 address of a pointer to a 10-element array of reals, and the value of
1108 the @code{foo10_5} stab is the address of a pointer to a 5-element array
1109 of 10-element arrays of reals.
1114 Formal parameters to a function are represented by a stab (or sometimes
1115 two; see below) for each parameter. The stabs are in the order in which
1116 the debugger should print the parameters (i.e., the order in which the
1117 parameters are declared in the source file). The exact form of the stab
1118 depends on how the parameter is being passed.
1122 Parameters passed on the stack use the symbol descriptor @samp{p} and
1123 the @code{N_PSYM} symbol type (or @code{C_PSYM} for XCOFF). The value
1124 of the symbol is an offset used to locate the parameter on the stack;
1125 its exact meaning is machine-dependent, but on most machines it is an
1126 offset from the frame pointer.
1128 As a simple example, the code:
1139 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
1140 .stabs "argc:p1",160,0,0,68 # @r{160 is N_PSYM}
1141 .stabs "argv:p20=*21=*2",160,0,0,72
1144 The type definition of @code{argv} is interesting because it contains
1145 several type definitions. Type 21 is pointer to type 2 (char) and
1146 @code{argv} (type 20) is pointer to type 21.
1148 @c FIXME: figure out what these mean and describe them coherently.
1149 The following symbol descriptors are also said to go with @code{N_PSYM}.
1150 The value of the symbol is said to be an offset from the argument
1151 pointer (I'm not sure whether this is true or not).
1155 pF Fortran function parameter
1156 X (function result variable)
1160 * Register Parameters::
1161 * Local Variable Parameters::
1162 * Reference Parameters::
1163 * Conformant Arrays::
1166 @node Register Parameters
1167 @subsection Passing Parameters in Registers
1169 If the parameter is passed in a register, then traditionally there are
1170 two symbols for each argument:
1173 .stabs "arg:p1" . . . ; N_PSYM
1174 .stabs "arg:r1" . . . ; N_RSYM
1177 Debuggers use the second one to find the value, and the first one to
1178 know that it is an argument.
1181 @findex N_RSYM, for parameters
1182 Because that approach is kind of ugly, some compilers use symbol
1183 descriptor @samp{P} or @samp{R} to indicate an argument which is in a
1184 register. Symbol type @code{C_RPSYM} is used in XCOFF and @code{N_RSYM}
1185 is used otherwise. The symbol's value is the register number. @samp{P}
1186 and @samp{R} mean the same thing; the difference is that @samp{P} is a
1187 GNU invention and @samp{R} is an IBM (XCOFF) invention. As of version
1188 4.9, GDB should handle either one.
1190 There is at least one case where GCC uses a @samp{p} and @samp{r} pair
1191 rather than @samp{P}; this is where the argument is passed in the
1192 argument list and then loaded into a register.
1194 According to the AIX documentation, symbol descriptor @samp{D} is for a
1195 parameter passed in a floating point register. This seems
1196 unnecessary---why not just use @samp{R} with a register number which
1197 indicates that it's a floating point register? I haven't verified
1198 whether the system actually does what the documentation indicates.
1200 @c FIXME: On the hppa this is for any type > 8 bytes, I think, and not
1201 @c for small structures (investigate).
1202 On the sparc and hppa, for a @samp{P} symbol whose type is a structure
1203 or union, the register contains the address of the structure. On the
1204 sparc, this is also true of a @samp{p} and @samp{r} pair (using Sun
1205 @code{cc}) or a @samp{p} symbol. However, if a (small) structure is
1206 really in a register, @samp{r} is used. And, to top it all off, on the
1207 hppa it might be a structure which was passed on the stack and loaded
1208 into a register and for which there is a @samp{p} and @samp{r} pair! I
1209 believe that symbol descriptor @samp{i} is supposed to deal with this
1210 case (it is said to mean "value parameter by reference, indirect
1211 access"; I don't know the source for this information), but I don't know
1212 details or what compilers or debuggers use it, if any (not GDB or GCC).
1213 It is not clear to me whether this case needs to be dealt with
1214 differently than parameters passed by reference (@pxref{Reference Parameters}).
1216 @node Local Variable Parameters
1217 @subsection Storing Parameters as Local Variables
1219 There is a case similar to an argument in a register, which is an
1220 argument that is actually stored as a local variable. Sometimes this
1221 happens when the argument was passed in a register and then the compiler
1222 stores it as a local variable. If possible, the compiler should claim
1223 that it's in a register, but this isn't always done.
1225 If a parameter is passed as one type and converted to a smaller type by
1226 the prologue (for example, the parameter is declared as a @code{float},
1227 but the calling conventions specify that it is passed as a
1228 @code{double}), then GCC2 (sometimes) uses a pair of symbols. The first
1229 symbol uses symbol descriptor @samp{p} and the type which is passed.
1230 The second symbol has the type and location which the parameter actually
1231 has after the prologue. For example, suppose the following C code
1232 appears with no prototypes involved:
1241 if @code{f} is passed as a double at stack offset 8, and the prologue
1242 converts it to a float in register number 0, then the stabs look like:
1245 .stabs "f:p13",160,0,3,8 # @r{160 is @code{N_PSYM}, here 13 is @code{double}}
1246 .stabs "f:r12",64,0,3,0 # @r{64 is @code{N_RSYM}, here 12 is @code{float}}
1249 In both stabs 3 is the line number where @code{f} is declared
1250 (@pxref{Line Numbers}).
1252 @findex N_LSYM, for parameter
1253 GCC, at least on the 960, has another solution to the same problem. It
1254 uses a single @samp{p} symbol descriptor for an argument which is stored
1255 as a local variable but uses @code{N_LSYM} instead of @code{N_PSYM}. In
1256 this case, the value of the symbol is an offset relative to the local
1257 variables for that function, not relative to the arguments; on some
1258 machines those are the same thing, but not on all.
1260 @c This is mostly just background info; the part that logically belongs
1261 @c here is the last sentence.
1262 On the VAX or on other machines in which the calling convention includes
1263 the number of words of arguments actually passed, the debugger (GDB at
1264 least) uses the parameter symbols to keep track of whether it needs to
1265 print nameless arguments in addition to the formal parameters which it
1266 has printed because each one has a stab. For example, in
1269 extern int fprintf (FILE *stream, char *format, @dots{});
1271 fprintf (stdout, "%d\n", x);
1274 there are stabs for @code{stream} and @code{format}. On most machines,
1275 the debugger can only print those two arguments (because it has no way
1276 of knowing that additional arguments were passed), but on the VAX or
1277 other machines with a calling convention which indicates the number of
1278 words of arguments, the debugger can print all three arguments. To do
1279 so, the parameter symbol (symbol descriptor @samp{p}) (not necessarily
1280 @samp{r} or symbol descriptor omitted symbols) needs to contain the
1281 actual type as passed (for example, @code{double} not @code{float} if it
1282 is passed as a double and converted to a float).
1284 @node Reference Parameters
1285 @subsection Passing Parameters by Reference
1287 If the parameter is passed by reference (e.g., Pascal @code{VAR}
1288 parameters), then the symbol descriptor is @samp{v} if it is in the
1289 argument list, or @samp{a} if it in a register. Other than the fact
1290 that these contain the address of the parameter rather than the
1291 parameter itself, they are identical to @samp{p} and @samp{R},
1292 respectively. I believe @samp{a} is an AIX invention; @samp{v} is
1293 supported by all stabs-using systems as far as I know.
1295 @node Conformant Arrays
1296 @subsection Passing Conformant Array Parameters
1298 @c Is this paragraph correct? It is based on piecing together patchy
1299 @c information and some guesswork
1300 Conformant arrays are a feature of Modula-2, and perhaps other
1301 languages, in which the size of an array parameter is not known to the
1302 called function until run-time. Such parameters have two stabs: a
1303 @samp{x} for the array itself, and a @samp{C}, which represents the size
1304 of the array. The value of the @samp{x} stab is the offset in the
1305 argument list where the address of the array is stored (it this right?
1306 it is a guess); the value of the @samp{C} stab is the offset in the
1307 argument list where the size of the array (in elements? in bytes?) is
1311 @chapter Defining Types
1313 The examples so far have described types as references to previously
1314 defined types, or defined in terms of subranges of or pointers to
1315 previously defined types. This chapter describes the other type
1316 descriptors that may follow the @samp{=} in a type definition.
1319 * Builtin Types:: Integers, floating point, void, etc.
1320 * Miscellaneous Types:: Pointers, sets, files, etc.
1321 * Cross-References:: Referring to a type not yet defined.
1322 * Subranges:: A type with a specific range.
1323 * Arrays:: An aggregate type of same-typed elements.
1324 * Strings:: Like an array but also has a length.
1325 * Enumerations:: Like an integer but the values have names.
1326 * Structures:: An aggregate type of different-typed elements.
1327 * Typedefs:: Giving a type a name.
1328 * Unions:: Different types sharing storage.
1333 @section Builtin Types
1335 Certain types are built in (@code{int}, @code{short}, @code{void},
1336 @code{float}, etc.); the debugger recognizes these types and knows how
1337 to handle them. Thus, don't be surprised if some of the following ways
1338 of specifying builtin types do not specify everything that a debugger
1339 would need to know about the type---in some cases they merely specify
1340 enough information to distinguish the type from other types.
1342 The traditional way to define builtin types is convoluted, so new ways
1343 have been invented to describe them. Sun's @code{acc} uses special
1344 builtin type descriptors (@samp{b} and @samp{R}), and IBM uses negative
1345 type numbers. GDB accepts all three ways, as of version 4.8; dbx just
1346 accepts the traditional builtin types and perhaps one of the other two
1347 formats. The following sections describe each of these formats.
1350 * Traditional Builtin Types:: Put on your seat belts and prepare for kludgery
1351 * Builtin Type Descriptors:: Builtin types with special type descriptors
1352 * Negative Type Numbers:: Builtin types using negative type numbers
1355 @node Traditional Builtin Types
1356 @subsection Traditional Builtin Types
1358 This is the traditional, convoluted method for defining builtin types.
1359 There are several classes of such type definitions: integer, floating
1360 point, and @code{void}.
1363 * Traditional Integer Types::
1364 * Traditional Other Types::
1367 @node Traditional Integer Types
1368 @subsubsection Traditional Integer Types
1370 Often types are defined as subranges of themselves. If the bounding values
1371 fit within an @code{int}, then they are given normally. For example:
1374 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0 # @r{128 is N_LSYM}
1375 .stabs "char:t2=r2;0;127;",128,0,0,0
1378 Builtin types can also be described as subranges of @code{int}:
1381 .stabs "unsigned short:t6=r1;0;65535;",128,0,0,0
1384 If the lower bound of a subrange is 0 and the upper bound is -1,
1385 the type is an unsigned integral type whose bounds are too
1386 big to describe in an @code{int}. Traditionally this is only used for
1387 @code{unsigned int} and @code{unsigned long}:
1390 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
1393 For larger types, GCC 2.4.5 puts out bounds in octal, with one or more
1394 leading zeroes. In this case a negative bound consists of a number
1395 which is a 1 bit (for the sign bit) followed by a 0 bit for each bit in
1396 the number (except the sign bit), and a positive bound is one which is a
1397 1 bit for each bit in the number (except possibly the sign bit). All
1398 known versions of dbx and GDB version 4 accept this (at least in the
1399 sense of not refusing to process the file), but GDB 3.5 refuses to read
1400 the whole file containing such symbols. So GCC 2.3.3 did not output the
1401 proper size for these types. As an example of octal bounds, the string
1402 fields of the stabs for 64 bit integer types look like:
1404 @c .stabs directives, etc., omitted to make it fit on the page.
1406 long int:t3=r1;001000000000000000000000;000777777777777777777777;
1407 long unsigned int:t5=r1;000000000000000000000000;001777777777777777777777;
1410 If the lower bound of a subrange is 0 and the upper bound is negative,
1411 the type is an unsigned integral type whose size in bytes is the
1412 absolute value of the upper bound. I believe this is a Convex
1413 convention for @code{unsigned long long}.
1415 If the lower bound of a subrange is negative and the upper bound is 0,
1416 the type is a signed integral type whose size in bytes is
1417 the absolute value of the lower bound. I believe this is a Convex
1418 convention for @code{long long}. To distinguish this from a legitimate
1419 subrange, the type should be a subrange of itself. I'm not sure whether
1420 this is the case for Convex.
1422 @node Traditional Other Types
1423 @subsubsection Traditional Other Types
1425 If the upper bound of a subrange is 0 and the lower bound is positive,
1426 the type is a floating point type, and the lower bound of the subrange
1427 indicates the number of bytes in the type:
1430 .stabs "float:t12=r1;4;0;",128,0,0,0
1431 .stabs "double:t13=r1;8;0;",128,0,0,0
1434 However, GCC writes @code{long double} the same way it writes
1435 @code{double}, so there is no way to distinguish.
1438 .stabs "long double:t14=r1;8;0;",128,0,0,0
1441 Complex types are defined the same way as floating-point types; there is
1442 no way to distinguish a single-precision complex from a double-precision
1443 floating-point type.
1445 The C @code{void} type is defined as itself:
1448 .stabs "void:t15=15",128,0,0,0
1451 I'm not sure how a boolean type is represented.
1453 @node Builtin Type Descriptors
1454 @subsection Defining Builtin Types Using Builtin Type Descriptors
1456 This is the method used by Sun's @code{acc} for defining builtin types.
1457 These are the type descriptors to define builtin types:
1460 @c FIXME: clean up description of width and offset, once we figure out
1462 @item b @var{signed} @var{char-flag} @var{width} ; @var{offset} ; @var{nbits} ;
1463 Define an integral type. @var{signed} is @samp{u} for unsigned or
1464 @samp{s} for signed. @var{char-flag} is @samp{c} which indicates this
1465 is a character type, or is omitted. I assume this is to distinguish an
1466 integral type from a character type of the same size, for example it
1467 might make sense to set it for the C type @code{wchar_t} so the debugger
1468 can print such variables differently (Solaris does not do this). Sun
1469 sets it on the C types @code{signed char} and @code{unsigned char} which
1470 arguably is wrong. @var{width} and @var{offset} appear to be for small
1471 objects stored in larger ones, for example a @code{short} in an
1472 @code{int} register. @var{width} is normally the number of bytes in the
1473 type. @var{offset} seems to always be zero. @var{nbits} is the number
1474 of bits in the type.
1476 Note that type descriptor @samp{b} used for builtin types conflicts with
1477 its use for Pascal space types (@pxref{Miscellaneous Types}); they can
1478 be distinguished because the character following the type descriptor
1479 will be a digit, @samp{(}, or @samp{-} for a Pascal space type, or
1480 @samp{u} or @samp{s} for a builtin type.
1483 Documented by AIX to define a wide character type, but their compiler
1484 actually uses negative type numbers (@pxref{Negative Type Numbers}).
1486 @item R @var{fp-type} ; @var{bytes} ;
1487 Define a floating point type. @var{fp-type} has one of the following values:
1491 IEEE 32-bit (single precision) floating point format.
1494 IEEE 64-bit (double precision) floating point format.
1496 @item 3 (NF_COMPLEX)
1497 @item 4 (NF_COMPLEX16)
1498 @item 5 (NF_COMPLEX32)
1499 @c "GDB source" really means @file{include/aout/stab_gnu.h}, but trying
1500 @c to put that here got an overfull hbox.
1501 These are for complex numbers. A comment in the GDB source describes
1502 them as Fortran @code{complex}, @code{double complex}, and
1503 @code{complex*16}, respectively, but what does that mean? (i.e., Single
1504 precision? Double precision?).
1506 @item 6 (NF_LDOUBLE)
1507 Long double. This should probably only be used for Sun format
1508 @code{long double}, and new codes should be used for other floating
1509 point formats (@code{NF_DOUBLE} can be used if a @code{long double} is
1510 really just an IEEE double, of course).
1513 @var{bytes} is the number of bytes occupied by the type. This allows a
1514 debugger to perform some operations with the type even if it doesn't
1515 understand @var{fp-type}.
1517 @item g @var{type-information} ; @var{nbits}
1518 Documented by AIX to define a floating type, but their compiler actually
1519 uses negative type numbers (@pxref{Negative Type Numbers}).
1521 @item c @var{type-information} ; @var{nbits}
1522 Documented by AIX to define a complex type, but their compiler actually
1523 uses negative type numbers (@pxref{Negative Type Numbers}).
1526 The C @code{void} type is defined as a signed integral type 0 bits long:
1528 .stabs "void:t19=bs0;0;0",128,0,0,0
1530 The Solaris compiler seems to omit the trailing semicolon in this case.
1531 Getting sloppy in this way is not a swift move because if a type is
1532 embedded in a more complex expression it is necessary to be able to tell
1535 I'm not sure how a boolean type is represented.
1537 @node Negative Type Numbers
1538 @subsection Negative Type Numbers
1540 This is the method used in XCOFF for defining builtin types.
1541 Since the debugger knows about the builtin types anyway, the idea of
1542 negative type numbers is simply to give a special type number which
1543 indicates the builtin type. There is no stab defining these types.
1545 There are several subtle issues with negative type numbers.
1547 One is the size of the type. A builtin type (for example the C types
1548 @code{int} or @code{long}) might have different sizes depending on
1549 compiler options, the target architecture, the ABI, etc. This issue
1550 doesn't come up for IBM tools since (so far) they just target the
1551 RS/6000; the sizes indicated below for each size are what the IBM
1552 RS/6000 tools use. To deal with differing sizes, either define separate
1553 negative type numbers for each size (which works but requires changing
1554 the debugger, and, unless you get both AIX dbx and GDB to accept the
1555 change, introduces an incompatibility), or use a type attribute
1556 (@pxref{String Field}) to define a new type with the appropriate size
1557 (which merely requires a debugger which understands type attributes,
1558 like AIX dbx or GDB). For example,
1561 .stabs "boolean:t10=@@s8;-16",128,0,0,0
1564 defines an 8-bit boolean type, and
1567 .stabs "boolean:t10=@@s64;-16",128,0,0,0
1570 defines a 64-bit boolean type.
1572 A similar issue is the format of the type. This comes up most often for
1573 floating-point types, which could have various formats (particularly
1574 extended doubles, which vary quite a bit even among IEEE systems).
1575 Again, it is best to define a new negative type number for each
1576 different format; changing the format based on the target system has
1577 various problems. One such problem is that the Alpha has both VAX and
1578 IEEE floating types. One can easily imagine one library using the VAX
1579 types and another library in the same executable using the IEEE types.
1580 Another example is that the interpretation of whether a boolean is true
1581 or false can be based on the least significant bit, most significant
1582 bit, whether it is zero, etc., and different compilers (or different
1583 options to the same compiler) might provide different kinds of boolean.
1585 The last major issue is the names of the types. The name of a given
1586 type depends @emph{only} on the negative type number given; these do not
1587 vary depending on the language, the target system, or anything else.
1588 One can always define separate type numbers---in the following list you
1589 will see for example separate @code{int} and @code{integer*4} types
1590 which are identical except for the name. But compatibility can be
1591 maintained by not inventing new negative type numbers and instead just
1592 defining a new type with a new name. For example:
1595 .stabs "CARDINAL:t10=-8",128,0,0,0
1598 Here is the list of negative type numbers. The phrase @dfn{integral
1599 type} is used to mean twos-complement (I strongly suspect that all
1600 machines which use stabs use twos-complement; most machines use
1601 twos-complement these days).
1605 @code{int}, 32 bit signed integral type.
1608 @code{char}, 8 bit type holding a character. Both GDB and dbx on AIX
1609 treat this as signed. GCC uses this type whether @code{char} is signed
1610 or not, which seems like a bad idea. The AIX compiler (@code{xlc}) seems to
1611 avoid this type; it uses -5 instead for @code{char}.
1614 @code{short}, 16 bit signed integral type.
1617 @code{long}, 32 bit signed integral type.
1620 @code{unsigned char}, 8 bit unsigned integral type.
1623 @code{signed char}, 8 bit signed integral type.
1626 @code{unsigned short}, 16 bit unsigned integral type.
1629 @code{unsigned int}, 32 bit unsigned integral type.
1632 @code{unsigned}, 32 bit unsigned integral type.
1635 @code{unsigned long}, 32 bit unsigned integral type.
1638 @code{void}, type indicating the lack of a value.
1641 @code{float}, IEEE single precision.
1644 @code{double}, IEEE double precision.
1647 @code{long double}, IEEE double precision. The compiler claims the size
1648 will increase in a future release, and for binary compatibility you have
1649 to avoid using @code{long double}. I hope when they increase it they
1650 use a new negative type number.
1653 @code{integer}. 32 bit signed integral type.
1656 @code{boolean}. 32 bit type. GDB and GCC assume that zero is false,
1657 one is true, and other values have unspecified meaning. I hope this
1658 agrees with how the IBM tools use the type.
1661 @code{short real}. IEEE single precision.
1664 @code{real}. IEEE double precision.
1667 @code{stringptr}. @xref{Strings}.
1670 @code{character}, 8 bit unsigned character type.
1673 @code{logical*1}, 8 bit type. This Fortran type has a split
1674 personality in that it is used for boolean variables, but can also be
1675 used for unsigned integers. 0 is false, 1 is true, and other values are
1679 @code{logical*2}, 16 bit type. This Fortran type has a split
1680 personality in that it is used for boolean variables, but can also be
1681 used for unsigned integers. 0 is false, 1 is true, and other values are
1685 @code{logical*4}, 32 bit type. This Fortran type has a split
1686 personality in that it is used for boolean variables, but can also be
1687 used for unsigned integers. 0 is false, 1 is true, and other values are
1691 @code{logical}, 32 bit type. This Fortran type has a split
1692 personality in that it is used for boolean variables, but can also be
1693 used for unsigned integers. 0 is false, 1 is true, and other values are
1697 @code{complex}. A complex type consisting of two IEEE single-precision
1698 floating point values.
1701 @code{complex}. A complex type consisting of two IEEE double-precision
1702 floating point values.
1705 @code{integer*1}, 8 bit signed integral type.
1708 @code{integer*2}, 16 bit signed integral type.
1711 @code{integer*4}, 32 bit signed integral type.
1714 @code{wchar}. Wide character, 16 bits wide, unsigned (what format?
1718 @code{long long}, 64 bit signed integral type.
1721 @code{unsigned long long}, 64 bit unsigned integral type.
1724 @code{logical*8}, 64 bit unsigned integral type.
1727 @code{integer*8}, 64 bit signed integral type.
1730 @node Miscellaneous Types
1731 @section Miscellaneous Types
1734 @item b @var{type-information} ; @var{bytes}
1735 Pascal space type. This is documented by IBM; what does it mean?
1737 This use of the @samp{b} type descriptor can be distinguished
1738 from its use for builtin integral types (@pxref{Builtin Type
1739 Descriptors}) because the character following the type descriptor is
1740 always a digit, @samp{(}, or @samp{-}.
1742 @item B @var{type-information}
1743 A volatile-qualified version of @var{type-information}. This is
1744 a Sun extension. References and stores to a variable with a
1745 volatile-qualified type must not be optimized or cached; they
1746 must occur as the user specifies them.
1748 @item d @var{type-information}
1749 File of type @var{type-information}. As far as I know this is only used
1752 @item k @var{type-information}
1753 A const-qualified version of @var{type-information}. This is a Sun
1754 extension. A variable with a const-qualified type cannot be modified.
1756 @item M @var{type-information} ; @var{length}
1757 Multiple instance type. The type seems to composed of @var{length}
1758 repetitions of @var{type-information}, for example @code{character*3} is
1759 represented by @samp{M-2;3}, where @samp{-2} is a reference to a
1760 character type (@pxref{Negative Type Numbers}). I'm not sure how this
1761 differs from an array. This appears to be a Fortran feature.
1762 @var{length} is a bound, like those in range types; see @ref{Subranges}.
1764 @item S @var{type-information}
1765 Pascal set type. @var{type-information} must be a small type such as an
1766 enumeration or a subrange, and the type is a bitmask whose length is
1767 specified by the number of elements in @var{type-information}.
1769 In CHILL, if it is a bitstring instead of a set, also use the @samp{S}
1770 type attribute (@pxref{String Field}).
1772 @item * @var{type-information}
1773 Pointer to @var{type-information}.
1776 @node Cross-References
1777 @section Cross-References to Other Types
1779 A type can be used before it is defined; one common way to deal with
1780 that situation is just to use a type reference to a type which has not
1783 Another way is with the @samp{x} type descriptor, which is followed by
1784 @samp{s} for a structure tag, @samp{u} for a union tag, or @samp{e} for
1785 a enumerator tag, followed by the name of the tag, followed by @samp{:}.
1786 If the name contains @samp{::} between a @samp{<} and @samp{>} pair (for
1787 C@t{++} templates), such a @samp{::} does not end the name---only a single
1788 @samp{:} ends the name; see @ref{Nested Symbols}.
1790 For example, the following C declarations:
1801 .stabs "bar:G16=*17=xsfoo:",32,0,0,0
1804 Not all debuggers support the @samp{x} type descriptor, so on some
1805 machines GCC does not use it. I believe that for the above example it
1806 would just emit a reference to type 17 and never define it, but I
1807 haven't verified that.
1809 Modula-2 imported types, at least on AIX, use the @samp{i} type
1810 descriptor, which is followed by the name of the module from which the
1811 type is imported, followed by @samp{:}, followed by the name of the
1812 type. There is then optionally a comma followed by type information for
1813 the type. This differs from merely naming the type (@pxref{Typedefs}) in
1814 that it identifies the module; I don't understand whether the name of
1815 the type given here is always just the same as the name we are giving
1816 it, or whether this type descriptor is used with a nameless stab
1817 (@pxref{String Field}), or what. The symbol ends with @samp{;}.
1820 @section Subrange Types
1822 The @samp{r} type descriptor defines a type as a subrange of another
1823 type. It is followed by type information for the type of which it is a
1824 subrange, a semicolon, an integral lower bound, a semicolon, an
1825 integral upper bound, and a semicolon. The AIX documentation does not
1826 specify the trailing semicolon, in an effort to specify array indexes
1827 more cleanly, but a subrange which is not an array index has always
1828 included a trailing semicolon (@pxref{Arrays}).
1830 Instead of an integer, either bound can be one of the following:
1833 @item A @var{offset}
1834 The bound is passed by reference on the stack at offset @var{offset}
1835 from the argument list. @xref{Parameters}, for more information on such
1838 @item T @var{offset}
1839 The bound is passed by value on the stack at offset @var{offset} from
1842 @item a @var{register-number}
1843 The bound is passed by reference in register number
1844 @var{register-number}.
1846 @item t @var{register-number}
1847 The bound is passed by value in register number @var{register-number}.
1853 Subranges are also used for builtin types; see @ref{Traditional Builtin Types}.
1856 @section Array Types
1858 Arrays use the @samp{a} type descriptor. Following the type descriptor
1859 is the type of the index and the type of the array elements. If the
1860 index type is a range type, it ends in a semicolon; otherwise
1861 (for example, if it is a type reference), there does not
1862 appear to be any way to tell where the types are separated. In an
1863 effort to clean up this mess, IBM documents the two types as being
1864 separated by a semicolon, and a range type as not ending in a semicolon
1865 (but this is not right for range types which are not array indexes,
1866 @pxref{Subranges}). I think probably the best solution is to specify
1867 that a semicolon ends a range type, and that the index type and element
1868 type of an array are separated by a semicolon, but that if the index
1869 type is a range type, the extra semicolon can be omitted. GDB (at least
1870 through version 4.9) doesn't support any kind of index type other than a
1871 range anyway; I'm not sure about dbx.
1873 It is well established, and widely used, that the type of the index,
1874 unlike most types found in the stabs, is merely a type definition, not
1875 type information (@pxref{String Field}) (that is, it need not start with
1876 @samp{@var{type-number}=} if it is defining a new type). According to a
1877 comment in GDB, this is also true of the type of the array elements; it
1878 gives @samp{ar1;1;10;ar1;1;10;4} as a legitimate way to express a two
1879 dimensional array. According to AIX documentation, the element type
1880 must be type information. GDB accepts either.
1882 The type of the index is often a range type, expressed as the type
1883 descriptor @samp{r} and some parameters. It defines the size of the
1884 array. In the example below, the range @samp{r1;0;2;} defines an index
1885 type which is a subrange of type 1 (integer), with a lower bound of 0
1886 and an upper bound of 2. This defines the valid range of subscripts of
1887 a three-element C array.
1889 For example, the definition:
1892 char char_vec[3] = @{'a','b','c'@};
1896 produces the output:
1899 .stabs "char_vec:G19=ar1;0;2;2",32,0,0,0
1908 If an array is @dfn{packed}, the elements are spaced more
1909 closely than normal, saving memory at the expense of speed. For
1910 example, an array of 3-byte objects might, if unpacked, have each
1911 element aligned on a 4-byte boundary, but if packed, have no padding.
1912 One way to specify that something is packed is with type attributes
1913 (@pxref{String Field}). In the case of arrays, another is to use the
1914 @samp{P} type descriptor instead of @samp{a}. Other than specifying a
1915 packed array, @samp{P} is identical to @samp{a}.
1917 @c FIXME-what is it? A pointer?
1918 An open array is represented by the @samp{A} type descriptor followed by
1919 type information specifying the type of the array elements.
1921 @c FIXME: what is the format of this type? A pointer to a vector of pointers?
1922 An N-dimensional dynamic array is represented by
1925 D @var{dimensions} ; @var{type-information}
1928 @c Does dimensions really have this meaning? The AIX documentation
1930 @var{dimensions} is the number of dimensions; @var{type-information}
1931 specifies the type of the array elements.
1933 @c FIXME: what is the format of this type? A pointer to some offsets in
1935 A subarray of an N-dimensional array is represented by
1938 E @var{dimensions} ; @var{type-information}
1941 @c Does dimensions really have this meaning? The AIX documentation
1943 @var{dimensions} is the number of dimensions; @var{type-information}
1944 specifies the type of the array elements.
1949 Some languages, like C or the original Pascal, do not have string types,
1950 they just have related things like arrays of characters. But most
1951 Pascals and various other languages have string types, which are
1952 indicated as follows:
1955 @item n @var{type-information} ; @var{bytes}
1956 @var{bytes} is the maximum length. I'm not sure what
1957 @var{type-information} is; I suspect that it means that this is a string
1958 of @var{type-information} (thus allowing a string of integers, a string
1959 of wide characters, etc., as well as a string of characters). Not sure
1960 what the format of this type is. This is an AIX feature.
1962 @item z @var{type-information} ; @var{bytes}
1963 Just like @samp{n} except that this is a gstring, not an ordinary
1964 string. I don't know the difference.
1967 Pascal Stringptr. What is this? This is an AIX feature.
1970 Languages, such as CHILL which have a string type which is basically
1971 just an array of characters use the @samp{S} type attribute
1972 (@pxref{String Field}).
1975 @section Enumerations
1977 Enumerations are defined with the @samp{e} type descriptor.
1979 @c FIXME: Where does this information properly go? Perhaps it is
1980 @c redundant with something we already explain.
1981 The source line below declares an enumeration type at file scope.
1982 The type definition is located after the @code{N_RBRAC} that marks the end of
1983 the previous procedure's block scope, and before the @code{N_FUN} that marks
1984 the beginning of the next procedure's block scope. Therefore it does not
1985 describe a block local symbol, but a file local one.
1990 enum e_places @{first,second=3,last@};
1994 generates the following stab:
1997 .stabs "e_places:T22=efirst:0,second:3,last:4,;",128,0,0,0
2000 The symbol descriptor (@samp{T}) says that the stab describes a
2001 structure, enumeration, or union tag. The type descriptor @samp{e},
2002 following the @samp{22=} of the type definition narrows it down to an
2003 enumeration type. Following the @samp{e} is a list of the elements of
2004 the enumeration. The format is @samp{@var{name}:@var{value},}. The
2005 list of elements ends with @samp{;}. The fact that @var{value} is
2006 specified as an integer can cause problems if the value is large. GCC
2007 2.5.2 tries to output it in octal in that case with a leading zero,
2008 which is probably a good thing, although GDB 4.11 supports octal only in
2009 cases where decimal is perfectly good. Negative decimal values are
2010 supported by both GDB and dbx.
2012 There is no standard way to specify the size of an enumeration type; it
2013 is determined by the architecture (normally all enumerations types are
2014 32 bits). Type attributes can be used to specify an enumeration type of
2015 another size for debuggers which support them; see @ref{String Field}.
2017 Enumeration types are unusual in that they define symbols for the
2018 enumeration values (@code{first}, @code{second}, and @code{third} in the
2019 above example), and even though these symbols are visible in the file as
2020 a whole (rather than being in a more local namespace like structure
2021 member names), they are defined in the type definition for the
2022 enumeration type rather than each having their own symbol. In order to
2023 be fast, GDB will only get symbols from such types (in its initial scan
2024 of the stabs) if the type is the first thing defined after a @samp{T} or
2025 @samp{t} symbol descriptor (the above example fulfills this
2026 requirement). If the type does not have a name, the compiler should
2027 emit it in a nameless stab (@pxref{String Field}); GCC does this.
2032 The encoding of structures in stabs can be shown with an example.
2034 The following source code declares a structure tag and defines an
2035 instance of the structure in global scope. Then a @code{typedef} equates the
2036 structure tag with a new type. Separate stabs are generated for the
2037 structure tag, the structure @code{typedef}, and the structure instance. The
2038 stabs for the tag and the @code{typedef} are emitted when the definitions are
2039 encountered. Since the structure elements are not initialized, the
2040 stab and code for the structure variable itself is located at the end
2041 of the program in the bss section.
2048 struct s_tag* s_next;
2051 typedef struct s_tag s_typedef;
2054 The structure tag has an @code{N_LSYM} stab type because, like the
2055 enumeration, the symbol has file scope. Like the enumeration, the
2056 symbol descriptor is @samp{T}, for enumeration, structure, or tag type.
2057 The type descriptor @samp{s} following the @samp{16=} of the type
2058 definition narrows the symbol type to structure.
2060 Following the @samp{s} type descriptor is the number of bytes the
2061 structure occupies, followed by a description of each structure element.
2062 The structure element descriptions are of the form
2063 @samp{@var{name}:@var{type}, @var{bit offset from the start of the
2064 struct}, @var{number of bits in the element}}.
2066 @c FIXME: phony line break. Can probably be fixed by using an example
2067 @c with fewer fields.
2070 .stabs "s_tag:T16=s20s_int:1,0,32;s_float:12,32,32;
2071 s_char_vec:17=ar1;0;7;2,64,64;s_next:18=*16,128,32;;",128,0,0,0
2074 In this example, the first two structure elements are previously defined
2075 types. For these, the type following the @samp{@var{name}:} part of the
2076 element description is a simple type reference. The other two structure
2077 elements are new types. In this case there is a type definition
2078 embedded after the @samp{@var{name}:}. The type definition for the
2079 array element looks just like a type definition for a stand-alone array.
2080 The @code{s_next} field is a pointer to the same kind of structure that
2081 the field is an element of. So the definition of structure type 16
2082 contains a type definition for an element which is a pointer to type 16.
2084 If a field is a static member (this is a C@t{++} feature in which a single
2085 variable appears to be a field of every structure of a given type) it
2086 still starts out with the field name, a colon, and the type, but then
2087 instead of a comma, bit position, comma, and bit size, there is a colon
2088 followed by the name of the variable which each such field refers to.
2090 If the structure has methods (a C@t{++} feature), they follow the non-method
2091 fields; see @ref{Cplusplus}.
2094 @section Giving a Type a Name
2096 @findex N_LSYM, for types
2097 @findex C_DECL, for types
2098 To give a type a name, use the @samp{t} symbol descriptor. The type
2099 is specified by the type information (@pxref{String Field}) for the stab.
2103 .stabs "s_typedef:t16",128,0,0,0 # @r{128 is N_LSYM}
2106 specifies that @code{s_typedef} refers to type number 16. Such stabs
2107 have symbol type @code{N_LSYM} (or @code{C_DECL} for XCOFF). (The Sun
2108 documentation mentions using @code{N_GSYM} in some cases).
2110 If you are specifying the tag name for a structure, union, or
2111 enumeration, use the @samp{T} symbol descriptor instead. I believe C is
2112 the only language with this feature.
2114 If the type is an opaque type (I believe this is a Modula-2 feature),
2115 AIX provides a type descriptor to specify it. The type descriptor is
2116 @samp{o} and is followed by a name. I don't know what the name
2117 means---is it always the same as the name of the type, or is this type
2118 descriptor used with a nameless stab (@pxref{String Field})? There
2119 optionally follows a comma followed by type information which defines
2120 the type of this type. If omitted, a semicolon is used in place of the
2121 comma and the type information, and the type is much like a generic
2122 pointer type---it has a known size but little else about it is
2136 This code generates a stab for a union tag and a stab for a union
2137 variable. Both use the @code{N_LSYM} stab type. If a union variable is
2138 scoped locally to the procedure in which it is defined, its stab is
2139 located immediately preceding the @code{N_LBRAC} for the procedure's block
2142 The stab for the union tag, however, is located preceding the code for
2143 the procedure in which it is defined. The stab type is @code{N_LSYM}. This
2144 would seem to imply that the union type is file scope, like the struct
2145 type @code{s_tag}. This is not true. The contents and position of the stab
2146 for @code{u_type} do not convey any information about its procedure local
2149 @c FIXME: phony line break. Can probably be fixed by using an example
2150 @c with fewer fields.
2153 .stabs "u_tag:T23=u4u_int:1,0,32;u_float:12,0,32;u_char:21,0,32;;",
2157 The symbol descriptor @samp{T}, following the @samp{name:} means that
2158 the stab describes an enumeration, structure, or union tag. The type
2159 descriptor @samp{u}, following the @samp{23=} of the type definition,
2160 narrows it down to a union type definition. Following the @samp{u} is
2161 the number of bytes in the union. After that is a list of union element
2162 descriptions. Their format is @samp{@var{name}:@var{type}, @var{bit
2163 offset into the union}, @var{number of bytes for the element};}.
2165 The stab for the union variable is:
2168 .stabs "an_u:23",128,0,0,-20 # @r{128 is N_LSYM}
2171 @samp{-20} specifies where the variable is stored (@pxref{Stack
2174 @node Function Types
2175 @section Function Types
2177 Various types can be defined for function variables. These types are
2178 not used in defining functions (@pxref{Procedures}); they are used for
2179 things like pointers to functions.
2181 The simple, traditional, type is type descriptor @samp{f} is followed by
2182 type information for the return type of the function, followed by a
2185 This does not deal with functions for which the number and types of the
2186 parameters are part of the type, as in Modula-2 or ANSI C. AIX provides
2187 extensions to specify these, using the @samp{f}, @samp{F}, @samp{p}, and
2188 @samp{R} type descriptors.
2190 First comes the type descriptor. If it is @samp{f} or @samp{F}, this
2191 type involves a function rather than a procedure, and the type
2192 information for the return type of the function follows, followed by a
2193 comma. Then comes the number of parameters to the function and a
2194 semicolon. Then, for each parameter, there is the name of the parameter
2195 followed by a colon (this is only present for type descriptors @samp{R}
2196 and @samp{F} which represent Pascal function or procedure parameters),
2197 type information for the parameter, a comma, 0 if passed by reference or
2198 1 if passed by value, and a semicolon. The type definition ends with a
2201 For example, this variable definition:
2208 generates the following code:
2211 .stabs "g_pf:G24=*25=f1",32,0,0,0
2212 .common _g_pf,4,"bss"
2215 The variable defines a new type, 24, which is a pointer to another new
2216 type, 25, which is a function returning @code{int}.
2218 @node Macro define and undefine
2219 @chapter Representation of #define and #undef
2221 This section describes the stabs support for macro define and undefine
2222 information, supported on some systems. (e.g., with @option{-g3}
2223 @option{-gstabs} when using GCC).
2225 A @code{#define @var{macro-name} @var{macro-body}} is represented with
2226 an @code{N_MAC_DEFINE} stab with a string field of
2227 @code{@var{macro-name} @var{macro-body}}.
2228 @findex N_MAC_DEFINE
2230 An @code{#undef @var{macro-name}} is represented with an
2231 @code{N_MAC_UNDEF} stabs with a string field of simply
2232 @code{@var{macro-name}}.
2235 For both @code{N_MAC_DEFINE} and @code{N_MAC_UNDEF}, the desc field is
2236 the line number within the file where the corresponding @code{#define}
2237 or @code{#undef} occurred.
2239 For example, the following C code:
2243 #define TWO(a, b) (a + (a) + 2 * b)
2244 #define ONE(c) (c + 19)
2246 main(int argc, char *argv[])
2248 func(NONE, TWO(10, 11));
2249 func(NONE, ONE(23));
2252 #define ONE(c) (c + 23)
2254 func(NONE, ONE(-23));
2261 func(int arg1, int arg2)
2263 global = arg1 + arg2;
2268 produces the following stabs (as well as many others):
2271 .stabs "NONE 42",54,0,1,0
2272 .stabs "TWO(a,b) (a + (a) + 2 * b)",54,0,2,0
2273 .stabs "ONE(c) (c + 19)",54,0,3,0
2274 .stabs "ONE",58,0,10,0
2275 .stabs "ONE(c) (c + 23)",54,0,11,0
2279 NOTE: In the above example, @code{54} is @code{N_MAC_DEFINE} and
2280 @code{58} is @code{N_MAC_UNDEF}.
2283 @chapter Symbol Information in Symbol Tables
2285 This chapter describes the format of symbol table entries
2286 and how stab assembler directives map to them. It also describes the
2287 transformations that the assembler and linker make on data from stabs.
2290 * Symbol Table Format::
2291 * Transformations On Symbol Tables::
2294 @node Symbol Table Format
2295 @section Symbol Table Format
2297 Each time the assembler encounters a stab directive, it puts
2298 each field of the stab into a corresponding field in a symbol table
2299 entry of its output file. If the stab contains a string field, the
2300 symbol table entry for that stab points to a string table entry
2301 containing the string data from the stab. Assembler labels become
2302 relocatable addresses. Symbol table entries in a.out have the format:
2304 @c FIXME: should refer to external, not internal.
2306 struct internal_nlist @{
2307 unsigned long n_strx; /* index into string table of name */
2308 unsigned char n_type; /* type of symbol */
2309 unsigned char n_other; /* misc info (usually empty) */
2310 unsigned short n_desc; /* description field */
2311 bfd_vma n_value; /* value of symbol */
2315 If the stab has a string, the @code{n_strx} field holds the offset in
2316 bytes of the string within the string table. The string is terminated
2317 by a NUL character. If the stab lacks a string (for example, it was
2318 produced by a @code{.stabn} or @code{.stabd} directive), the
2319 @code{n_strx} field is zero.
2321 Symbol table entries with @code{n_type} field values greater than 0x1f
2322 originated as stabs generated by the compiler (with one random
2323 exception). The other entries were placed in the symbol table of the
2324 executable by the assembler or the linker.
2326 @node Transformations On Symbol Tables
2327 @section Transformations on Symbol Tables
2329 The linker concatenates object files and does fixups of externally
2332 You can see the transformations made on stab data by the assembler and
2333 linker by examining the symbol table after each pass of the build. To
2334 do this, use @samp{nm -ap}, which dumps the symbol table, including
2335 debugging information, unsorted. For stab entries the columns are:
2336 @var{value}, @var{other}, @var{desc}, @var{type}, @var{string}. For
2337 assembler and linker symbols, the columns are: @var{value}, @var{type},
2340 The low 5 bits of the stab type tell the linker how to relocate the
2341 value of the stab. Thus for stab types like @code{N_RSYM} and
2342 @code{N_LSYM}, where the value is an offset or a register number, the
2343 low 5 bits are @code{N_ABS}, which tells the linker not to relocate the
2346 Where the value of a stab contains an assembly language label,
2347 it is transformed by each build step. The assembler turns it into a
2348 relocatable address and the linker turns it into an absolute address.
2351 * Transformations On Static Variables::
2352 * Transformations On Global Variables::
2353 * Stab Section Transformations:: For some object file formats,
2354 things are a bit different.
2357 @node Transformations On Static Variables
2358 @subsection Transformations on Static Variables
2360 This source line defines a static variable at file scope:
2363 static int s_g_repeat
2367 The following stab describes the symbol:
2370 .stabs "s_g_repeat:S1",38,0,0,_s_g_repeat
2374 The assembler transforms the stab into this symbol table entry in the
2375 @file{.o} file. The location is expressed as a data segment offset.
2378 00000084 - 00 0000 STSYM s_g_repeat:S1
2382 In the symbol table entry from the executable, the linker has made the
2383 relocatable address absolute.
2386 0000e00c - 00 0000 STSYM s_g_repeat:S1
2389 @node Transformations On Global Variables
2390 @subsection Transformations on Global Variables
2392 Stabs for global variables do not contain location information. In
2393 this case, the debugger finds location information in the assembler or
2394 linker symbol table entry describing the variable. The source line:
2404 .stabs "g_foo:G2",32,0,0,0
2407 The variable is represented by two symbol table entries in the object
2408 file (see below). The first one originated as a stab. The second one
2409 is an external symbol. The upper case @samp{D} signifies that the
2410 @code{n_type} field of the symbol table contains 7, @code{N_DATA} with
2411 local linkage. The stab's value is zero since the value is not used for
2412 @code{N_GSYM} stabs. The value of the linker symbol is the relocatable
2413 address corresponding to the variable.
2416 00000000 - 00 0000 GSYM g_foo:G2
2421 These entries as transformed by the linker. The linker symbol table
2422 entry now holds an absolute address:
2425 00000000 - 00 0000 GSYM g_foo:G2
2430 @node Stab Section Transformations
2431 @subsection Transformations of Stabs in separate sections
2433 For object file formats using stabs in separate sections (@pxref{Stab
2434 Sections}), use @code{objdump --stabs} instead of @code{nm} to show the
2435 stabs in an object or executable file. @code{objdump} is a GNU utility;
2436 Sun does not provide any equivalent.
2438 The following example is for a stab whose value is an address is
2439 relative to the compilation unit (@pxref{ELF Linker Relocation}). For
2440 example, if the source line
2446 appears within a function, then the assembly language output from the
2452 .stabs "ld:V(0,3)",0x26,0,4,.L18-Ddata.data # @r{0x26 is N_STSYM}
2459 Because the value is formed by subtracting one symbol from another, the
2460 value is absolute, not relocatable, and so the object file contains
2463 Symnum n_type n_othr n_desc n_value n_strx String
2464 31 STSYM 0 4 00000004 680 ld:V(0,3)
2467 without any relocations, and the executable file also contains
2470 Symnum n_type n_othr n_desc n_value n_strx String
2471 31 STSYM 0 4 00000004 680 ld:V(0,3)
2475 @chapter GNU C@t{++} Stabs
2478 * Class Names:: C++ class names are both tags and typedefs.
2479 * Nested Symbols:: C++ symbol names can be within other types.
2480 * Basic Cplusplus Types::
2483 * Methods:: Method definition
2484 * Method Type Descriptor:: The @samp{#} type descriptor
2485 * Member Type Descriptor:: The @samp{@@} type descriptor
2487 * Method Modifiers::
2490 * Virtual Base Classes::
2495 @section C@t{++} Class Names
2497 In C@t{++}, a class name which is declared with @code{class}, @code{struct},
2498 or @code{union}, is not only a tag, as in C, but also a type name. Thus
2499 there should be stabs with both @samp{t} and @samp{T} symbol descriptors
2502 To save space, there is a special abbreviation for this case. If the
2503 @samp{T} symbol descriptor is followed by @samp{t}, then the stab
2504 defines both a type name and a tag.
2506 For example, the C@t{++} code
2509 struct foo @{int x;@};
2512 can be represented as either
2515 .stabs "foo:T19=s4x:1,0,32;;",128,0,0,0 # @r{128 is N_LSYM}
2516 .stabs "foo:t19",128,0,0,0
2522 .stabs "foo:Tt19=s4x:1,0,32;;",128,0,0,0
2525 @node Nested Symbols
2526 @section Defining a Symbol Within Another Type
2528 In C@t{++}, a symbol (such as a type name) can be defined within another type.
2529 @c FIXME: Needs example.
2531 In stabs, this is sometimes represented by making the name of a symbol
2532 which contains @samp{::}. Such a pair of colons does not end the name
2533 of the symbol, the way a single colon would (@pxref{String Field}). I'm
2534 not sure how consistently used or well thought out this mechanism is.
2535 So that a pair of colons in this position always has this meaning,
2536 @samp{:} cannot be used as a symbol descriptor.
2538 For example, if the string for a stab is @samp{foo::bar::baz:t5=*6},
2539 then @code{foo::bar::baz} is the name of the symbol, @samp{t} is the
2540 symbol descriptor, and @samp{5=*6} is the type information.
2542 @node Basic Cplusplus Types
2543 @section Basic Types For C@t{++}
2545 << the examples that follow are based on a01.C >>
2548 C@t{++} adds two more builtin types to the set defined for C. These are
2549 the unknown type and the vtable record type. The unknown type, type
2550 16, is defined in terms of itself like the void type.
2552 The vtable record type, type 17, is defined as a structure type and
2553 then as a structure tag. The structure has four fields: delta, index,
2554 pfn, and delta2. pfn is the function pointer.
2556 << In boilerplate $vtbl_ptr_type, what are the fields delta,
2557 index, and delta2 used for? >>
2559 This basic type is present in all C@t{++} programs even if there are no
2560 virtual methods defined.
2563 .stabs "struct_name:sym_desc(type)type_def(17)=type_desc(struct)struct_bytes(8)
2564 elem_name(delta):type_ref(short int),bit_offset(0),field_bits(16);
2565 elem_name(index):type_ref(short int),bit_offset(16),field_bits(16);
2566 elem_name(pfn):type_def(18)=type_desc(ptr to)type_ref(void),
2567 bit_offset(32),field_bits(32);
2568 elem_name(delta2):type_def(short int);bit_offset(32),field_bits(16);;"
2573 .stabs "$vtbl_ptr_type:t17=s8
2574 delta:6,0,16;index:6,16,16;pfn:18=*15,32,32;delta2:6,32,16;;"
2579 .stabs "name:sym_dec(struct tag)type_ref($vtbl_ptr_type)",N_LSYM,NIL,NIL,NIL
2583 .stabs "$vtbl_ptr_type:T17",128,0,0,0
2586 @node Simple Classes
2587 @section Simple Class Definition
2589 The stabs describing C@t{++} language features are an extension of the
2590 stabs describing C. Stabs representing C@t{++} class types elaborate
2591 extensively on the stab format used to describe structure types in C.
2592 Stabs representing class type variables look just like stabs
2593 representing C language variables.
2595 Consider the following very simple class definition.
2601 int Ameth(int in, char other);
2605 The class @code{baseA} is represented by two stabs. The first stab describes
2606 the class as a structure type. The second stab describes a structure
2607 tag of the class type. Both stabs are of stab type @code{N_LSYM}. Since the
2608 stab is not located between an @code{N_FUN} and an @code{N_LBRAC} stab this indicates
2609 that the class is defined at file scope. If it were, then the @code{N_LSYM}
2610 would signify a local variable.
2612 A stab describing a C@t{++} class type is similar in format to a stab
2613 describing a C struct, with each class member shown as a field in the
2614 structure. The part of the struct format describing fields is
2615 expanded to include extra information relevant to C@t{++} class members.
2616 In addition, if the class has multiple base classes or virtual
2617 functions the struct format outside of the field parts is also
2620 In this simple example the field part of the C@t{++} class stab
2621 representing member data looks just like the field part of a C struct
2622 stab. The section on protections describes how its format is
2623 sometimes extended for member data.
2625 The field part of a C@t{++} class stab representing a member function
2626 differs substantially from the field part of a C struct stab. It
2627 still begins with @samp{name:} but then goes on to define a new type number
2628 for the member function, describe its return type, its argument types,
2629 its protection level, any qualifiers applied to the method definition,
2630 and whether the method is virtual or not. If the method is virtual
2631 then the method description goes on to give the vtable index of the
2632 method, and the type number of the first base class defining the
2635 When the field name is a method name it is followed by two colons rather
2636 than one. This is followed by a new type definition for the method.
2637 This is a number followed by an equal sign and the type of the method.
2638 Normally this will be a type declared using the @samp{#} type
2639 descriptor; see @ref{Method Type Descriptor}; static member functions
2640 are declared using the @samp{f} type descriptor instead; see
2641 @ref{Function Types}.
2643 The format of an overloaded operator method name differs from that of
2644 other methods. It is @samp{op$::@var{operator-name}.} where
2645 @var{operator-name} is the operator name such as @samp{+} or @samp{+=}.
2646 The name ends with a period, and any characters except the period can
2647 occur in the @var{operator-name} string.
2649 The next part of the method description represents the arguments to the
2650 method, preceded by a colon and ending with a semi-colon. The types of
2651 the arguments are expressed in the same way argument types are expressed
2652 in C@t{++} name mangling. In this example an @code{int} and a @code{char}
2655 This is followed by a number, a letter, and an asterisk or period,
2656 followed by another semicolon. The number indicates the protections
2657 that apply to the member function. Here the 2 means public. The
2658 letter encodes any qualifier applied to the method definition. In
2659 this case, @samp{A} means that it is a normal function definition. The dot
2660 shows that the method is not virtual. The sections that follow
2661 elaborate further on these fields and describe the additional
2662 information present for virtual methods.
2666 .stabs "class_name:sym_desc(type)type_def(20)=type_desc(struct)struct_bytes(4)
2667 field_name(Adat):type(int),bit_offset(0),field_bits(32);
2669 method_name(Ameth)::type_def(21)=type_desc(method)return_type(int);
2670 :arg_types(int char);
2671 protection(public)qualifier(normal)virtual(no);;"
2676 .stabs "baseA:t20=s4Adat:1,0,32;Ameth::21=##1;:ic;2A.;;",128,0,0,0
2678 .stabs "class_name:sym_desc(struct tag)",N_LSYM,NIL,NIL,NIL
2680 .stabs "baseA:T20",128,0,0,0
2683 @node Class Instance
2684 @section Class Instance
2686 As shown above, describing even a simple C@t{++} class definition is
2687 accomplished by massively extending the stab format used in C to
2688 describe structure types. However, once the class is defined, C stabs
2689 with no modifications can be used to describe class instances. The
2699 yields the following stab describing the class instance. It looks no
2700 different from a standard C stab describing a local variable.
2703 .stabs "name:type_ref(baseA)", N_LSYM, NIL, NIL, frame_ptr_offset
2707 .stabs "AbaseA:20",128,0,0,-20
2711 @section Method Definition
2713 The class definition shown above declares Ameth. The C@t{++} source below
2718 baseA::Ameth(int in, char other)
2725 This method definition yields three stabs following the code of the
2726 method. One stab describes the method itself and following two describe
2727 its parameters. Although there is only one formal argument all methods
2728 have an implicit argument which is the @code{this} pointer. The @code{this}
2729 pointer is a pointer to the object on which the method was called. Note
2730 that the method name is mangled to encode the class name and argument
2731 types. Name mangling is described in the @sc{arm} (@cite{The Annotated
2732 C++ Reference Manual}, by Ellis and Stroustrup, @sc{isbn}
2733 0-201-51459-1); @file{gpcompare.texi} in Cygnus GCC distributions
2734 describes the differences between GNU mangling and @sc{arm}
2736 @c FIXME: Use @xref, especially if this is generally installed in the
2738 @c FIXME: This information should be in a net release, either of GCC or
2739 @c GDB. But gpcompare.texi doesn't seem to be in the FSF GCC.
2742 .stabs "name:symbol_descriptor(global function)return_type(int)",
2743 N_FUN, NIL, NIL, code_addr_of_method_start
2745 .stabs "Ameth__5baseAic:F1",36,0,0,_Ameth__5baseAic
2748 Here is the stab for the @code{this} pointer implicit argument. The
2749 name of the @code{this} pointer is always @code{this}. Type 19, the
2750 @code{this} pointer is defined as a pointer to type 20, @code{baseA},
2751 but a stab defining @code{baseA} has not yet been emitted. Since the
2752 compiler knows it will be emitted shortly, here it just outputs a cross
2753 reference to the undefined symbol, by prefixing the symbol name with
2757 .stabs "name:sym_desc(register param)type_def(19)=
2758 type_desc(ptr to)type_ref(baseA)=
2759 type_desc(cross-reference to)baseA:",N_RSYM,NIL,NIL,register_number
2761 .stabs "this:P19=*20=xsbaseA:",64,0,0,8
2764 The stab for the explicit integer argument looks just like a parameter
2765 to a C function. The last field of the stab is the offset from the
2766 argument pointer, which in most systems is the same as the frame
2770 .stabs "name:sym_desc(value parameter)type_ref(int)",
2771 N_PSYM,NIL,NIL,offset_from_arg_ptr
2773 .stabs "in:p1",160,0,0,72
2776 << The examples that follow are based on A1.C >>
2778 @node Method Type Descriptor
2779 @section The @samp{#} Type Descriptor
2781 This is used to describe a class method. This is a function which takes
2782 an extra argument as its first argument, for the @code{this} pointer.
2784 If the @samp{#} is immediately followed by another @samp{#}, the second
2785 one will be followed by the return type and a semicolon. The class and
2786 argument types are not specified, and must be determined by demangling
2787 the name of the method if it is available.
2789 Otherwise, the single @samp{#} is followed by the class type, a comma,
2790 the return type, a comma, and zero or more parameter types separated by
2791 commas. The list of arguments is terminated by a semicolon. In the
2792 debugging output generated by gcc, a final argument type of @code{void}
2793 indicates a method which does not take a variable number of arguments.
2794 If the final argument type of @code{void} does not appear, the method
2795 was declared with an ellipsis.
2797 Note that although such a type will normally be used to describe fields
2798 in structures, unions, or classes, for at least some versions of the
2799 compiler it can also be used in other contexts.
2801 @node Member Type Descriptor
2802 @section The @samp{@@} Type Descriptor
2804 The @samp{@@} type descriptor is used for a
2805 pointer-to-non-static-member-data type. It is followed
2806 by type information for the class (or union), a comma, and type
2807 information for the member data.
2809 The following C@t{++} source:
2812 typedef int A::*int_in_a;
2815 generates the following stab:
2818 .stabs "int_in_a:t20=21=@@19,1",128,0,0,0
2821 Note that there is a conflict between this and type attributes
2822 (@pxref{String Field}); both use type descriptor @samp{@@}.
2823 Fortunately, the @samp{@@} type descriptor used in this C@t{++} sense always
2824 will be followed by a digit, @samp{(}, or @samp{-}, and type attributes
2825 never start with those things.
2828 @section Protections
2830 In the simple class definition shown above all member data and
2831 functions were publicly accessible. The example that follows
2832 contrasts public, protected and privately accessible fields and shows
2833 how these protections are encoded in C@t{++} stabs.
2835 If the character following the @samp{@var{field-name}:} part of the
2836 string is @samp{/}, then the next character is the visibility. @samp{0}
2837 means private, @samp{1} means protected, and @samp{2} means public.
2838 Debuggers should ignore visibility characters they do not recognize, and
2839 assume a reasonable default (such as public) (GDB 4.11 does not, but
2840 this should be fixed in the next GDB release). If no visibility is
2841 specified the field is public. The visibility @samp{9} means that the
2842 field has been optimized out and is public (there is no way to specify
2843 an optimized out field with a private or protected visibility).
2844 Visibility @samp{9} is not supported by GDB 4.11; this should be fixed
2845 in the next GDB release.
2847 The following C@t{++} source:
2861 generates the following stab:
2865 .stabs "vis:T19=s12priv:/01,0,32;prot:/12,32,8;pub:12,64,32;;",128,0,0,0
2868 @samp{vis:T19=s12} indicates that type number 19 is a 12 byte structure
2869 named @code{vis} The @code{priv} field has public visibility
2870 (@samp{/0}), type int (@samp{1}), and offset and size @samp{,0,32;}.
2871 The @code{prot} field has protected visibility (@samp{/1}), type char
2872 (@samp{2}) and offset and size @samp{,32,8;}. The @code{pub} field has
2873 type float (@samp{12}), and offset and size @samp{,64,32;}.
2875 Protections for member functions are signified by one digit embedded in
2876 the field part of the stab describing the method. The digit is 0 if
2877 private, 1 if protected and 2 if public. Consider the C@t{++} class
2881 class all_methods @{
2883 int priv_meth(int in)@{return in;@};
2885 char protMeth(char in)@{return in;@};
2887 float pubMeth(float in)@{return in;@};
2891 It generates the following stab. The digit in question is to the left
2892 of an @samp{A} in each case. Notice also that in this case two symbol
2893 descriptors apply to the class name struct tag and struct type.
2896 .stabs "class_name:sym_desc(struct tag&type)type_def(21)=
2897 sym_desc(struct)struct_bytes(1)
2898 meth_name::type_def(22)=sym_desc(method)returning(int);
2899 :args(int);protection(private)modifier(normal)virtual(no);
2900 meth_name::type_def(23)=sym_desc(method)returning(char);
2901 :args(char);protection(protected)modifier(normal)virtual(no);
2902 meth_name::type_def(24)=sym_desc(method)returning(float);
2903 :args(float);protection(public)modifier(normal)virtual(no);;",
2908 .stabs "all_methods:Tt21=s1priv_meth::22=##1;:i;0A.;protMeth::23=##2;:c;1A.;
2909 pubMeth::24=##12;:f;2A.;;",128,0,0,0
2912 @node Method Modifiers
2913 @section Method Modifiers (@code{const}, @code{volatile}, @code{const volatile})
2917 In the class example described above all the methods have the normal
2918 modifier. This method modifier information is located just after the
2919 protection information for the method. This field has four possible
2920 character values. Normal methods use @samp{A}, const methods use
2921 @samp{B}, volatile methods use @samp{C}, and const volatile methods use
2922 @samp{D}. Consider the class definition below:
2927 int ConstMeth (int arg) const @{ return arg; @};
2928 char VolatileMeth (char arg) volatile @{ return arg; @};
2929 float ConstVolMeth (float arg) const volatile @{return arg; @};
2933 This class is described by the following stab:
2936 .stabs "class(A):sym_desc(struct)type_def(20)=type_desc(struct)struct_bytes(1)
2937 meth_name(ConstMeth)::type_def(21)sym_desc(method)
2938 returning(int);:arg(int);protection(public)modifier(const)virtual(no);
2939 meth_name(VolatileMeth)::type_def(22)=sym_desc(method)
2940 returning(char);:arg(char);protection(public)modifier(volatile)virt(no)
2941 meth_name(ConstVolMeth)::type_def(23)=sym_desc(method)
2942 returning(float);:arg(float);protection(public)modifier(const volatile)
2943 virtual(no);;", @dots{}
2947 .stabs "A:T20=s1ConstMeth::21=##1;:i;2B.;VolatileMeth::22=##2;:c;2C.;
2948 ConstVolMeth::23=##12;:f;2D.;;",128,0,0,0
2951 @node Virtual Methods
2952 @section Virtual Methods
2954 << The following examples are based on a4.C >>
2956 The presence of virtual methods in a class definition adds additional
2957 data to the class description. The extra data is appended to the
2958 description of the virtual method and to the end of the class
2959 description. Consider the class definition below:
2965 virtual int A_virt (int arg) @{ return arg; @};
2969 This results in the stab below describing class A. It defines a new
2970 type (20) which is an 8 byte structure. The first field of the class
2971 struct is @samp{Adat}, an integer, starting at structure offset 0 and
2974 The second field in the class struct is not explicitly defined by the
2975 C@t{++} class definition but is implied by the fact that the class
2976 contains a virtual method. This field is the vtable pointer. The
2977 name of the vtable pointer field starts with @samp{$vf} and continues with a
2978 type reference to the class it is part of. In this example the type
2979 reference for class A is 20 so the name of its vtable pointer field is
2980 @samp{$vf20}, followed by the usual colon.
2982 Next there is a type definition for the vtable pointer type (21).
2983 This is in turn defined as a pointer to another new type (22).
2985 Type 22 is the vtable itself, which is defined as an array, indexed by
2986 a range of integers between 0 and 1, and whose elements are of type
2987 17. Type 17 was the vtable record type defined by the boilerplate C@t{++}
2988 type definitions, as shown earlier.
2990 The bit offset of the vtable pointer field is 32. The number of bits
2991 in the field are not specified when the field is a vtable pointer.
2993 Next is the method definition for the virtual member function @code{A_virt}.
2994 Its description starts out using the same format as the non-virtual
2995 member functions described above, except instead of a dot after the
2996 @samp{A} there is an asterisk, indicating that the function is virtual.
2997 Since is is virtual some addition information is appended to the end
2998 of the method description.
3000 The first number represents the vtable index of the method. This is a
3001 32 bit unsigned number with the high bit set, followed by a
3004 The second number is a type reference to the first base class in the
3005 inheritance hierarchy defining the virtual member function. In this
3006 case the class stab describes a base class so the virtual function is
3007 not overriding any other definition of the method. Therefore the
3008 reference is to the type number of the class that the stab is
3011 This is followed by three semi-colons. One marks the end of the
3012 current sub-section, one marks the end of the method field, and the
3013 third marks the end of the struct definition.
3015 For classes containing virtual functions the very last section of the
3016 string part of the stab holds a type reference to the first base
3017 class. This is preceded by @samp{~%} and followed by a final semi-colon.
3020 .stabs "class_name(A):type_def(20)=sym_desc(struct)struct_bytes(8)
3021 field_name(Adat):type_ref(int),bit_offset(0),field_bits(32);
3022 field_name(A virt func ptr):type_def(21)=type_desc(ptr to)type_def(22)=
3023 sym_desc(array)index_type_ref(range of int from 0 to 1);
3024 elem_type_ref(vtbl elem type),
3026 meth_name(A_virt)::typedef(23)=sym_desc(method)returning(int);
3027 :arg_type(int),protection(public)normal(yes)virtual(yes)
3028 vtable_index(1);class_first_defining(A);;;~%first_base(A);",
3032 @c FIXME: bogus line break.
3034 .stabs "A:t20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
3035 A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
3039 @section Inheritance
3041 Stabs describing C@t{++} derived classes include additional sections that
3042 describe the inheritance hierarchy of the class. A derived class stab
3043 also encodes the number of base classes. For each base class it tells
3044 if the base class is virtual or not, and if the inheritance is private
3045 or public. It also gives the offset into the object of the portion of
3046 the object corresponding to each base class.
3048 This additional information is embedded in the class stab following the
3049 number of bytes in the struct. First the number of base classes
3050 appears bracketed by an exclamation point and a comma.
3052 Then for each base type there repeats a series: a virtual character, a
3053 visibility character, a number, a comma, another number, and a
3056 The virtual character is @samp{1} if the base class is virtual and
3057 @samp{0} if not. The visibility character is @samp{2} if the derivation
3058 is public, @samp{1} if it is protected, and @samp{0} if it is private.
3059 Debuggers should ignore virtual or visibility characters they do not
3060 recognize, and assume a reasonable default (such as public and
3061 non-virtual) (GDB 4.11 does not, but this should be fixed in the next
3064 The number following the virtual and visibility characters is the offset
3065 from the start of the object to the part of the object pertaining to the
3068 After the comma, the second number is a type_descriptor for the base
3069 type. Finally a semi-colon ends the series, which repeats for each
3072 The source below defines three base classes @code{A}, @code{B}, and
3073 @code{C} and the derived class @code{D}.
3080 virtual int A_virt (int arg) @{ return arg; @};
3086 virtual int B_virt (int arg) @{return arg; @};
3092 virtual int C_virt (int arg) @{return arg; @};
3095 class D : A, virtual B, public C @{
3098 virtual int A_virt (int arg ) @{ return arg+1; @};
3099 virtual int B_virt (int arg) @{ return arg+2; @};
3100 virtual int C_virt (int arg) @{ return arg+3; @};
3101 virtual int D_virt (int arg) @{ return arg; @};
3105 Class stabs similar to the ones described earlier are generated for
3108 @c FIXME!!! the linebreaks in the following example probably make the
3109 @c examples literally unusable, but I don't know any other way to get
3110 @c them on the page.
3111 @c One solution would be to put some of the type definitions into
3112 @c separate stabs, even if that's not exactly what the compiler actually
3115 .stabs "A:T20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
3116 A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
3118 .stabs "B:Tt25=s8Bdat:1,0,32;$vf25:21,32;B_virt::26=##1;
3119 :i;2A*-2147483647;25;;;~%25;",128,0,0,0
3121 .stabs "C:Tt28=s8Cdat:1,0,32;$vf28:21,32;C_virt::29=##1;
3122 :i;2A*-2147483647;28;;;~%28;",128,0,0,0
3125 In the stab describing derived class @code{D} below, the information about
3126 the derivation of this class is encoded as follows.
3129 .stabs "derived_class_name:symbol_descriptors(struct tag&type)=
3130 type_descriptor(struct)struct_bytes(32)!num_bases(3),
3131 base_virtual(no)inheritance_public(no)base_offset(0),
3132 base_class_type_ref(A);
3133 base_virtual(yes)inheritance_public(no)base_offset(NIL),
3134 base_class_type_ref(B);
3135 base_virtual(no)inheritance_public(yes)base_offset(64),
3136 base_class_type_ref(C); @dots{}
3139 @c FIXME! fake linebreaks.
3141 .stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:
3142 1,160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt:
3143 :32:i;2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;
3144 28;;D_virt::32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
3147 @node Virtual Base Classes
3148 @section Virtual Base Classes
3150 A derived class object consists of a concatenation in memory of the data
3151 areas defined by each base class, starting with the leftmost and ending
3152 with the rightmost in the list of base classes. The exception to this
3153 rule is for virtual inheritance. In the example above, class @code{D}
3154 inherits virtually from base class @code{B}. This means that an
3155 instance of a @code{D} object will not contain its own @code{B} part but
3156 merely a pointer to a @code{B} part, known as a virtual base pointer.
3158 In a derived class stab, the base offset part of the derivation
3159 information, described above, shows how the base class parts are
3160 ordered. The base offset for a virtual base class is always given as 0.
3161 Notice that the base offset for @code{B} is given as 0 even though
3162 @code{B} is not the first base class. The first base class @code{A}
3165 The field information part of the stab for class @code{D} describes the field
3166 which is the pointer to the virtual base class @code{B}. The vbase pointer
3167 name is @samp{$vb} followed by a type reference to the virtual base class.
3168 Since the type id for @code{B} in this example is 25, the vbase pointer name
3171 @c FIXME!! fake linebreaks below
3173 .stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:1,
3174 160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt::32:i;
3175 2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;28;;D_virt:
3176 :32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
3179 Following the name and a semicolon is a type reference describing the
3180 type of the virtual base class pointer, in this case 24. Type 24 was
3181 defined earlier as the type of the @code{B} class @code{this} pointer. The
3182 @code{this} pointer for a class is a pointer to the class type.
3185 .stabs "this:P24=*25=xsB:",64,0,0,8
3188 Finally the field offset part of the vbase pointer field description
3189 shows that the vbase pointer is the first field in the @code{D} object,
3190 before any data fields defined by the class. The layout of a @code{D}
3191 class object is a follows, @code{Adat} at 0, the vtable pointer for
3192 @code{A} at 32, @code{Cdat} at 64, the vtable pointer for C at 96, the
3193 virtual base pointer for @code{B} at 128, and @code{Ddat} at 160.
3196 @node Static Members
3197 @section Static Members
3199 The data area for a class is a concatenation of the space used by the
3200 data members of the class. If the class has virtual methods, a vtable
3201 pointer follows the class data. The field offset part of each field
3202 description in the class stab shows this ordering.
3204 << How is this reflected in stabs? See Cygnus bug #677 for some info. >>
3207 @appendix Table of Stab Types
3209 The following are all the possible values for the stab type field, for
3210 a.out files, in numeric order. This does not apply to XCOFF, but
3211 it does apply to stabs in sections (@pxref{Stab Sections}). Stabs in
3212 ECOFF use these values but add 0x8f300 to distinguish them from non-stab
3215 The symbolic names are defined in the file @file{include/aout/stabs.def}.
3218 * Non-Stab Symbol Types:: Types from 0 to 0x1f
3219 * Stab Symbol Types:: Types from 0x20 to 0xff
3222 @node Non-Stab Symbol Types
3223 @appendixsec Non-Stab Symbol Types
3225 The following types are used by the linker and assembler, not by stab
3226 directives. Since this document does not attempt to describe aspects of
3227 object file format other than the debugging format, no details are
3230 @c Try to get most of these to fit on a single line.
3240 File scope absolute symbol
3242 @item 0x3 N_ABS | N_EXT
3243 External absolute symbol
3246 File scope text symbol
3248 @item 0x5 N_TEXT | N_EXT
3249 External text symbol
3252 File scope data symbol
3254 @item 0x7 N_DATA | N_EXT
3255 External data symbol
3258 File scope BSS symbol
3260 @item 0x9 N_BSS | N_EXT
3264 Same as @code{N_FN}, for Sequent compilers
3267 Symbol is indirected to another symbol
3270 Common---visible after shared library dynamic link
3273 @itemx 0x15 N_SETA | N_EXT
3274 Absolute set element
3277 @itemx 0x17 N_SETT | N_EXT
3278 Text segment set element
3281 @itemx 0x19 N_SETD | N_EXT
3282 Data segment set element
3285 @itemx 0x1b N_SETB | N_EXT
3286 BSS segment set element
3289 @itemx 0x1d N_SETV | N_EXT
3290 Pointer to set vector
3292 @item 0x1e N_WARNING
3293 Print a warning message during linking
3296 File name of a @file{.o} file
3299 @node Stab Symbol Types
3300 @appendixsec Stab Symbol Types
3302 The following symbol types indicate that this is a stab. This is the
3303 full list of stab numbers, including stab types that are used in
3304 languages other than C.
3308 Global symbol; see @ref{Global Variables}.
3311 Function name (for BSD Fortran); see @ref{Procedures}.
3314 Function name (@pxref{Procedures}) or text segment variable
3318 Data segment file-scope variable; see @ref{Statics}.
3321 BSS segment file-scope variable; see @ref{Statics}.
3324 Name of main routine; see @ref{Main Program}.
3327 Variable in @code{.rodata} section; see @ref{Statics}.
3330 Global symbol (for Pascal); see @ref{N_PC}.
3333 Number of symbols (according to Ultrix V4.0); see @ref{N_NSYMS}.
3336 No DST map; see @ref{N_NOMAP}.
3338 @item 0x36 N_MAC_DEFINE
3339 Name and body of a @code{#define}d macro; see @ref{Macro define and undefine}.
3341 @c FIXME: describe this solaris feature in the body of the text (see
3342 @c comments in include/aout/stab.def).
3344 Object file (Solaris2).
3346 @item 0x3a N_MAC_UNDEF
3347 Name of an @code{#undef}ed macro; see @ref{Macro define and undefine}.
3349 @c See include/aout/stab.def for (a little) more info.
3351 Debugger options (Solaris2).
3354 Register variable; see @ref{Register Variables}.
3357 Modula-2 compilation unit; see @ref{N_M2C}.
3360 Line number in text segment; see @ref{Line Numbers}.
3363 Line number in data segment; see @ref{Line Numbers}.
3366 Line number in bss segment; see @ref{Line Numbers}.
3369 Sun source code browser, path to @file{.cb} file; see @ref{N_BROWS}.
3372 GNU Modula2 definition module dependency; see @ref{N_DEFD}.
3375 Function start/body/end line numbers (Solaris2).
3378 GNU C@t{++} exception variable; see @ref{N_EHDECL}.
3381 Modula2 info "for imc" (according to Ultrix V4.0); see @ref{N_MOD2}.
3384 GNU C@t{++} @code{catch} clause; see @ref{N_CATCH}.
3387 Structure of union element; see @ref{N_SSYM}.
3390 Last stab for module (Solaris2).
3393 Path and name of source file; see @ref{Source Files}.
3396 Stack variable (@pxref{Stack Variables}) or type (@pxref{Typedefs}).
3399 Beginning of an include file (Sun only); see @ref{Include Files}.
3402 Name of include file; see @ref{Include Files}.
3405 Parameter variable; see @ref{Parameters}.
3408 End of an include file; see @ref{Include Files}.
3411 Alternate entry point; see @ref{Alternate Entry Points}.
3414 Beginning of a lexical block; see @ref{Block Structure}.
3417 Place holder for a deleted include file; see @ref{Include Files}.
3420 Modula2 scope information (Sun linker); see @ref{N_SCOPE}.
3423 End of a lexical block; see @ref{Block Structure}.
3426 Begin named common block; see @ref{Common Blocks}.
3429 End named common block; see @ref{Common Blocks}.
3432 Member of a common block; see @ref{Common Blocks}.
3434 @c FIXME: How does this really work? Move it to main body of document.
3436 Pascal @code{with} statement: type,,0,0,offset (Solaris2).
3439 Gould non-base registers; see @ref{Gould}.
3442 Gould non-base registers; see @ref{Gould}.
3445 Gould non-base registers; see @ref{Gould}.
3448 Gould non-base registers; see @ref{Gould}.
3451 Gould non-base registers; see @ref{Gould}.
3454 @c Restore the default table indent
3459 @node Symbol Descriptors
3460 @appendix Table of Symbol Descriptors
3462 The symbol descriptor is the character which follows the colon in many
3463 stabs, and which tells what kind of stab it is. @xref{String Field},
3464 for more information about their use.
3466 @c Please keep this alphabetical
3468 @c In TeX, this looks great, digit is in italics. But makeinfo insists
3469 @c on putting it in `', not realizing that @var should override @code.
3470 @c I don't know of any way to make makeinfo do the right thing. Seems
3471 @c like a makeinfo bug to me.
3475 Variable on the stack; see @ref{Stack Variables}.
3478 C@t{++} nested symbol; see @xref{Nested Symbols}.
3481 Parameter passed by reference in register; see @ref{Reference Parameters}.
3484 Based variable; see @ref{Based Variables}.
3487 Constant; see @ref{Constants}.
3490 Conformant array bound (Pascal, maybe other languages); @ref{Conformant
3491 Arrays}. Name of a caught exception (GNU C@t{++}). These can be
3492 distinguished because the latter uses @code{N_CATCH} and the former uses
3493 another symbol type.
3496 Floating point register variable; see @ref{Register Variables}.
3499 Parameter in floating point register; see @ref{Register Parameters}.
3502 File scope function; see @ref{Procedures}.
3505 Global function; see @ref{Procedures}.
3508 Global variable; see @ref{Global Variables}.
3511 @xref{Register Parameters}.
3514 Internal (nested) procedure; see @ref{Nested Procedures}.
3517 Internal (nested) function; see @ref{Nested Procedures}.
3520 Label name (documented by AIX, no further information known).
3523 Module; see @ref{Procedures}.
3526 Argument list parameter; see @ref{Parameters}.
3532 Fortran Function parameter; see @ref{Parameters}.
3535 Unfortunately, three separate meanings have been independently invented
3536 for this symbol descriptor. At least the GNU and Sun uses can be
3537 distinguished by the symbol type. Global Procedure (AIX) (symbol type
3538 used unknown); see @ref{Procedures}. Register parameter (GNU) (symbol
3539 type @code{N_PSYM}); see @ref{Parameters}. Prototype of function
3540 referenced by this file (Sun @code{acc}) (symbol type @code{N_FUN}).
3543 Static Procedure; see @ref{Procedures}.
3546 Register parameter; see @ref{Register Parameters}.
3549 Register variable; see @ref{Register Variables}.
3552 File scope variable; see @ref{Statics}.
3555 Local variable (OS9000).
3558 Type name; see @ref{Typedefs}.
3561 Enumeration, structure, or union tag; see @ref{Typedefs}.
3564 Parameter passed by reference; see @ref{Reference Parameters}.
3567 Procedure scope static variable; see @ref{Statics}.
3570 Conformant array; see @ref{Conformant Arrays}.
3573 Function return variable; see @ref{Parameters}.
3576 @node Type Descriptors
3577 @appendix Table of Type Descriptors
3579 The type descriptor is the character which follows the type number and
3580 an equals sign. It specifies what kind of type is being defined.
3581 @xref{String Field}, for more information about their use.
3586 Type reference; see @ref{String Field}.
3589 Reference to builtin type; see @ref{Negative Type Numbers}.
3592 Method (C@t{++}); see @ref{Method Type Descriptor}.
3595 Pointer; see @ref{Miscellaneous Types}.
3598 Reference (C@t{++}).
3601 Type Attributes (AIX); see @ref{String Field}. Member (class and variable)
3602 type (GNU C@t{++}); see @ref{Member Type Descriptor}.
3605 Array; see @ref{Arrays}.
3608 Open array; see @ref{Arrays}.
3611 Pascal space type (AIX); see @ref{Miscellaneous Types}. Builtin integer
3612 type (Sun); see @ref{Builtin Type Descriptors}. Const and volatile
3613 qualified type (OS9000).
3616 Volatile-qualified type; see @ref{Miscellaneous Types}.
3619 Complex builtin type (AIX); see @ref{Builtin Type Descriptors}.
3620 Const-qualified type (OS9000).
3623 COBOL Picture type. See AIX documentation for details.
3626 File type; see @ref{Miscellaneous Types}.
3629 N-dimensional dynamic array; see @ref{Arrays}.
3632 Enumeration type; see @ref{Enumerations}.
3635 N-dimensional subarray; see @ref{Arrays}.
3638 Function type; see @ref{Function Types}.
3641 Pascal function parameter; see @ref{Function Types}
3644 Builtin floating point type; see @ref{Builtin Type Descriptors}.
3647 COBOL Group. See AIX documentation for details.
3650 Imported type (AIX); see @ref{Cross-References}. Volatile-qualified
3654 Const-qualified type; see @ref{Miscellaneous Types}.
3657 COBOL File Descriptor. See AIX documentation for details.
3660 Multiple instance type; see @ref{Miscellaneous Types}.
3663 String type; see @ref{Strings}.
3666 Stringptr; see @ref{Strings}.
3669 Opaque type; see @ref{Typedefs}.
3672 Procedure; see @ref{Function Types}.
3675 Packed array; see @ref{Arrays}.
3678 Range type; see @ref{Subranges}.
3681 Builtin floating type; see @ref{Builtin Type Descriptors} (Sun). Pascal
3682 subroutine parameter; see @ref{Function Types} (AIX). Detecting this
3683 conflict is possible with careful parsing (hint: a Pascal subroutine
3684 parameter type will always contain a comma, and a builtin type
3685 descriptor never will).
3688 Structure type; see @ref{Structures}.
3691 Set type; see @ref{Miscellaneous Types}.
3694 Union; see @ref{Unions}.
3697 Variant record. This is a Pascal and Modula-2 feature which is like a
3698 union within a struct in C. See AIX documentation for details.
3701 Wide character; see @ref{Builtin Type Descriptors}.
3704 Cross-reference; see @ref{Cross-References}.
3707 Used by IBM's xlC C@t{++} compiler (for structures, I think).
3710 gstring; see @ref{Strings}.
3713 @node Expanded Reference
3714 @appendix Expanded Reference by Stab Type
3716 @c FIXME: This appendix should go away; see N_PSYM or N_SO for an example.
3718 For a full list of stab types, and cross-references to where they are
3719 described, see @ref{Stab Types}. This appendix just covers certain
3720 stabs which are not yet described in the main body of this document;
3721 eventually the information will all be in one place.
3725 The first line is the symbol type (see @file{include/aout/stab.def}).
3727 The second line describes the language constructs the symbol type
3730 The third line is the stab format with the significant stab fields
3731 named and the rest NIL.
3733 Subsequent lines expand upon the meaning and possible values for each
3734 significant stab field.
3736 Finally, any further information.
3739 * N_PC:: Pascal global symbol
3740 * N_NSYMS:: Number of symbols
3741 * N_NOMAP:: No DST map
3742 * N_M2C:: Modula-2 compilation unit
3743 * N_BROWS:: Path to .cb file for Sun source code browser
3744 * N_DEFD:: GNU Modula2 definition module dependency
3745 * N_EHDECL:: GNU C++ exception variable
3746 * N_MOD2:: Modula2 information "for imc"
3747 * N_CATCH:: GNU C++ "catch" clause
3748 * N_SSYM:: Structure or union element
3749 * N_SCOPE:: Modula2 scope information (Sun only)
3750 * Gould:: non-base register symbols used on Gould systems
3751 * N_LENG:: Length of preceding entry
3757 @deffn @code{.stabs} N_PC
3759 Global symbol (for Pascal).
3762 "name" -> "symbol_name" <<?>>
3763 value -> supposedly the line number (stab.def is skeptical)
3767 @file{stabdump.c} says:
3769 global pascal symbol: name,,0,subtype,line
3777 @deffn @code{.stabn} N_NSYMS
3779 Number of symbols (according to Ultrix V4.0).
3782 0, files,,funcs,lines (stab.def)
3789 @deffn @code{.stabs} N_NOMAP
3791 No DST map for symbol (according to Ultrix V4.0). I think this means a
3792 variable has been optimized out.
3795 name, ,0,type,ignored (stab.def)
3802 @deffn @code{.stabs} N_M2C
3804 Modula-2 compilation unit.
3807 "string" -> "unit_name,unit_time_stamp[,code_time_stamp]"
3809 value -> 0 (main unit)
3813 See @cite{Dbx and Dbxtool Interfaces}, 2nd edition, by Sun, 1988, for
3821 @deffn @code{.stabs} N_BROWS
3823 Sun source code browser, path to @file{.cb} file
3826 "path to associated @file{.cb} file"
3828 Note: N_BROWS has the same value as N_BSLINE.
3834 @deffn @code{.stabn} N_DEFD
3836 GNU Modula2 definition module dependency.
3838 GNU Modula-2 definition module dependency. The value is the
3839 modification time of the definition file. The other field is non-zero
3840 if it is imported with the GNU M2 keyword @code{%INITIALIZE}. Perhaps
3841 @code{N_M2C} can be used if there are enough empty fields?
3847 @deffn @code{.stabs} N_EHDECL
3849 GNU C@t{++} exception variable <<?>>.
3851 "@var{string} is variable name"
3853 Note: conflicts with @code{N_MOD2}.
3859 @deffn @code{.stab?} N_MOD2
3861 Modula2 info "for imc" (according to Ultrix V4.0)
3863 Note: conflicts with @code{N_EHDECL} <<?>>
3869 @deffn @code{.stabn} N_CATCH
3871 GNU C@t{++} @code{catch} clause
3873 GNU C@t{++} @code{catch} clause. The value is its address. The desc field
3874 is nonzero if this entry is immediately followed by a @code{CAUGHT} stab
3875 saying what exception was caught. Multiple @code{CAUGHT} stabs means
3876 that multiple exceptions can be caught here. If desc is 0, it means all
3877 exceptions are caught here.
3883 @deffn @code{.stabn} N_SSYM
3885 Structure or union element.
3887 The value is the offset in the structure.
3889 <<?looking at structs and unions in C I didn't see these>>
3895 @deffn @code{.stab?} N_SCOPE
3897 Modula2 scope information (Sun linker)
3902 @section Non-base registers on Gould systems
3904 @deffn @code{.stab?} N_NBTEXT
3905 @deffnx @code{.stab?} N_NBDATA
3906 @deffnx @code{.stab?} N_NBBSS
3907 @deffnx @code{.stab?} N_NBSTS
3908 @deffnx @code{.stab?} N_NBLCS
3914 These are used on Gould systems for non-base registers syms.
3916 However, the following values are not the values used by Gould; they are
3917 the values which GNU has been documenting for these values for a long
3918 time, without actually checking what Gould uses. I include these values
3919 only because perhaps some someone actually did something with the GNU
3920 information (I hope not, why GNU knowingly assigned wrong values to
3921 these in the header file is a complete mystery to me).
3924 240 0xf0 N_NBTEXT ??
3925 242 0xf2 N_NBDATA ??
3935 @deffn @code{.stabn} N_LENG
3937 Second symbol entry containing a length-value for the preceding entry.
3938 The value is the length.
3942 @appendix Questions and Anomalies
3946 @c I think this is changed in GCC 2.4.5 to put the line number there.
3947 For GNU C stabs defining local and global variables (@code{N_LSYM} and
3948 @code{N_GSYM}), the desc field is supposed to contain the source
3949 line number on which the variable is defined. In reality the desc
3950 field is always 0. (This behavior is defined in @file{dbxout.c} and
3951 putting a line number in desc is controlled by @samp{#ifdef
3952 WINNING_GDB}, which defaults to false). GDB supposedly uses this
3953 information if you say @samp{list @var{var}}. In reality, @var{var} can
3954 be a variable defined in the program and GDB says @samp{function
3955 @var{var} not defined}.
3958 In GNU C stabs, there seems to be no way to differentiate tag types:
3959 structures, unions, and enums (symbol descriptor @samp{T}) and typedefs
3960 (symbol descriptor @samp{t}) defined at file scope from types defined locally
3961 to a procedure or other more local scope. They all use the @code{N_LSYM}
3962 stab type. Types defined at procedure scope are emitted after the
3963 @code{N_RBRAC} of the preceding function and before the code of the
3964 procedure in which they are defined. This is exactly the same as
3965 types defined in the source file between the two procedure bodies.
3966 GDB over-compensates by placing all types in block #1, the block for
3967 symbols of file scope. This is true for default, @samp{-ansi} and
3968 @samp{-traditional} compiler options. (Bugs gcc/1063, gdb/1066.)
3971 What ends the procedure scope? Is it the proc block's @code{N_RBRAC} or the
3972 next @code{N_FUN}? (I believe its the first.)
3976 @appendix Using Stabs in Their Own Sections
3978 Many object file formats allow tools to create object files with custom
3979 sections containing any arbitrary data. For any such object file
3980 format, stabs can be embedded in special sections. This is how stabs
3981 are used with ELF and SOM, and aside from ECOFF and XCOFF, is how stabs
3985 * Stab Section Basics:: How to embed stabs in sections
3986 * ELF Linker Relocation:: Sun ELF hacks
3989 @node Stab Section Basics
3990 @appendixsec How to Embed Stabs in Sections
3992 The assembler creates two custom sections, a section named @code{.stab}
3993 which contains an array of fixed length structures, one struct per stab,
3994 and a section named @code{.stabstr} containing all the variable length
3995 strings that are referenced by stabs in the @code{.stab} section. The
3996 byte order of the stabs binary data depends on the object file format.
3997 For ELF, it matches the byte order of the ELF file itself, as determined
3998 from the @code{EI_DATA} field in the @code{e_ident} member of the ELF
3999 header. For SOM, it is always big-endian (is this true??? FIXME). For
4000 COFF, it matches the byte order of the COFF headers. The meaning of the
4001 fields is the same as for a.out (@pxref{Symbol Table Format}), except
4002 that the @code{n_strx} field is relative to the strings for the current
4003 compilation unit (which can be found using the synthetic N_UNDF stab
4004 described below), rather than the entire string table.
4006 The first stab in the @code{.stab} section for each compilation unit is
4007 synthetic, generated entirely by the assembler, with no corresponding
4008 @code{.stab} directive as input to the assembler. This stab contains
4009 the following fields:
4013 Offset in the @code{.stabstr} section to the source filename.
4019 Unused field, always zero.
4020 This may eventually be used to hold overflows from the count in
4021 the @code{n_desc} field.
4024 Count of upcoming symbols, i.e., the number of remaining stabs for this
4028 Size of the string table fragment associated with this source file, in
4032 The @code{.stabstr} section always starts with a null byte (so that string
4033 offsets of zero reference a null string), followed by random length strings,
4034 each of which is null byte terminated.
4036 The ELF section header for the @code{.stab} section has its
4037 @code{sh_link} member set to the section number of the @code{.stabstr}
4038 section, and the @code{.stabstr} section has its ELF section
4039 header @code{sh_type} member set to @code{SHT_STRTAB} to mark it as a
4040 string table. SOM and COFF have no way of linking the sections together
4041 or marking them as string tables.
4043 For COFF, the @code{.stab} and @code{.stabstr} sections may be simply
4044 concatenated by the linker. GDB then uses the @code{n_desc} fields to
4045 figure out the extent of the original sections. Similarly, the
4046 @code{n_value} fields of the header symbols are added together in order
4047 to get the actual position of the strings in a desired @code{.stabstr}
4048 section. Although this design obviates any need for the linker to
4049 relocate or otherwise manipulate @code{.stab} and @code{.stabstr}
4050 sections, it also requires some care to ensure that the offsets are
4051 calculated correctly. For instance, if the linker were to pad in
4052 between the @code{.stabstr} sections before concatenating, then the
4053 offsets to strings in the middle of the executable's @code{.stabstr}
4054 section would be wrong.
4056 The GNU linker is able to optimize stabs information by merging
4057 duplicate strings and removing duplicate header file information
4058 (@pxref{Include Files}). When some versions of the GNU linker optimize
4059 stabs in sections, they remove the leading @code{N_UNDF} symbol and
4060 arranges for all the @code{n_strx} fields to be relative to the start of
4061 the @code{.stabstr} section.
4063 @node ELF Linker Relocation
4064 @appendixsec Having the Linker Relocate Stabs in ELF
4066 This section describes some Sun hacks for Stabs in ELF; it does not
4067 apply to COFF or SOM.
4069 To keep linking fast, you don't want the linker to have to relocate very
4070 many stabs. Making sure this is done for @code{N_SLINE},
4071 @code{N_RBRAC}, and @code{N_LBRAC} stabs is the most important thing
4072 (see the descriptions of those stabs for more information). But Sun's
4073 stabs in ELF has taken this further, to make all addresses in the
4074 @code{n_value} field (functions and static variables) relative to the
4075 source file. For the @code{N_SO} symbol itself, Sun simply omits the
4076 address. To find the address of each section corresponding to a given
4077 source file, the compiler puts out symbols giving the address of each
4078 section for a given source file. Since these are ELF (not stab)
4079 symbols, the linker relocates them correctly without having to touch the
4080 stabs section. They are named @code{Bbss.bss} for the bss section,
4081 @code{Ddata.data} for the data section, and @code{Drodata.rodata} for
4082 the rodata section. For the text section, there is no such symbol (but
4083 there should be, see below). For an example of how these symbols work,
4084 @xref{Stab Section Transformations}. GCC does not provide these symbols;
4085 it instead relies on the stabs getting relocated. Thus addresses which
4086 would normally be relative to @code{Bbss.bss}, etc., are already
4087 relocated. The Sun linker provided with Solaris 2.2 and earlier
4088 relocates stabs using normal ELF relocation information, as it would do
4089 for any section. Sun has been threatening to kludge their linker to not
4090 do this (to speed up linking), even though the correct way to avoid
4091 having the linker do these relocations is to have the compiler no longer
4092 output relocatable values. Last I heard they had been talked out of the
4093 linker kludge. See Sun point patch 101052-01 and Sun bug 1142109. With
4094 the Sun compiler this affects @samp{S} symbol descriptor stabs
4095 (@pxref{Statics}) and functions (@pxref{Procedures}). In the latter
4096 case, to adopt the clean solution (making the value of the stab relative
4097 to the start of the compilation unit), it would be necessary to invent a
4098 @code{Ttext.text} symbol, analogous to the @code{Bbss.bss}, etc.,
4099 symbols. I recommend this rather than using a zero value and getting
4100 the address from the ELF symbols.
4102 Finding the correct @code{Bbss.bss}, etc., symbol is difficult, because
4103 the linker simply concatenates the @code{.stab} sections from each
4104 @file{.o} file without including any information about which part of a
4105 @code{.stab} section comes from which @file{.o} file. The way GDB does
4106 this is to look for an ELF @code{STT_FILE} symbol which has the same
4107 name as the last component of the file name from the @code{N_SO} symbol
4108 in the stabs (for example, if the file name is @file{../../gdb/main.c},
4109 it looks for an ELF @code{STT_FILE} symbol named @code{main.c}). This
4110 loses if different files have the same name (they could be in different
4111 directories, a library could have been copied from one system to
4112 another, etc.). It would be much cleaner to have the @code{Bbss.bss}
4113 symbols in the stabs themselves. Having the linker relocate them there
4114 is no more work than having the linker relocate ELF symbols, and it
4115 solves the problem of having to associate the ELF and stab symbols.
4116 However, no one has yet designed or implemented such a scheme.
4122 @node Symbol Types Index
4123 @unnumbered Symbol Types Index