2 @setfilename stabs.info
9 * Stabs: (stabs). The "stabs" debugging information format.
15 This document describes the stabs debugging symbol tables.
17 Copyright 1992, 1993 Free Software Foundation, Inc.
18 Contributed by Cygnus Support. Written by Julia Menapace, Jim Kingdon,
21 Permission is granted to make and distribute verbatim copies of
22 this manual provided the copyright notice and this permission notice
23 are preserved on all copies.
26 Permission is granted to process this file through Tex and print the
27 results, provided the printed document carries copying permission
28 notice identical to this one except for the removal of this paragraph
29 (this paragraph not being relevant to the printed manual).
32 Permission is granted to copy or distribute modified versions of this
33 manual under the terms of the GPL (for which purpose this text may be
34 regarded as a program in the language TeX).
37 @setchapternewpage odd
40 @title The ``stabs'' debug format
41 @author Julia Menapace, Jim Kingdon, David MacKenzie
42 @author Cygnus Support
45 \def\$#1${{#1}} % Kluge: collect RCS revision info without $...$
46 \xdef\manvers{\$Revision$} % For use in headers, footers too
48 \hfill Cygnus Support\par
50 \hfill \TeX{}info \texinfoversion\par
54 @vskip 0pt plus 1filll
55 Copyright @copyright{} 1992, 1993 Free Software Foundation, Inc.
56 Contributed by Cygnus Support.
58 Permission is granted to make and distribute verbatim copies of
59 this manual provided the copyright notice and this permission notice
60 are preserved on all copies.
66 @top The "stabs" representation of debugging information
68 This document describes the stabs debugging format.
71 * Overview:: Overview of stabs
72 * Program Structure:: Encoding of the structure of the program
73 * Constants:: Constants
75 * Types:: Type definitions
76 * Symbol Tables:: Symbol information in symbol tables
77 * Cplusplus:: Stabs specific to C++
78 * Stab Types:: Symbol types in a.out files
79 * Symbol Descriptors:: Table of symbol descriptors
80 * Type Descriptors:: Table of type descriptors
81 * Expanded Reference:: Reference information by stab type
82 * Questions:: Questions and anomolies
83 * Sun Differences:: Differences between GNU stabs and Sun
85 * Stab Sections:: In some object file formats, stabs are
87 * Symbol Types Index:: Index of symbolic stab symbol type names.
93 @chapter Overview of Stabs
95 @dfn{Stabs} refers to a format for information that describes a program
96 to a debugger. This format was apparently invented by
98 the University of California at Berkeley, for the @code{pdx} Pascal
99 debugger; the format has spread widely since then.
101 This document is one of the few published sources of documentation on
102 stabs. It is believed to be comprehensive for stabs used by C. The
103 lists of symbol descriptors (@pxref{Symbol Descriptors}) and type
104 descriptors (@pxref{Type Descriptors}) are believed to be completely
105 comprehensive. Stabs for COBOL-specific features and for variant
106 records (used by Pascal and Modula-2) are poorly documented here.
108 @c FIXME: Need to document all OS9000 stuff in GDB; see all references
109 @c to os9k_stabs in stabsread.c.
111 Other sources of information on stabs are @cite{Dbx and Dbxtool
112 Interfaces}, 2nd edition, by Sun, 1988, and @cite{AIX Version 3.2 Files
113 Reference}, Fourth Edition, September 1992, "dbx Stabstring Grammar" in
114 the a.out section, page 2-31. This document is believed to incorporate
115 the information from those two sources except where it explicitly directs
116 you to them for more information.
119 * Flow:: Overview of debugging information flow
120 * Stabs Format:: Overview of stab format
121 * String Field:: The string field
122 * C Example:: A simple example in C source
123 * Assembly Code:: The simple example at the assembly level
127 @section Overview of Debugging Information Flow
129 The GNU C compiler compiles C source in a @file{.c} file into assembly
130 language in a @file{.s} file, which the assembler translates into
131 a @file{.o} file, which the linker combines with other @file{.o} files and
132 libraries to produce an executable file.
134 With the @samp{-g} option, GCC puts in the @file{.s} file additional
135 debugging information, which is slightly transformed by the assembler
136 and linker, and carried through into the final executable. This
137 debugging information describes features of the source file like line
138 numbers, the types and scopes of variables, and function names,
139 parameters, and scopes.
141 For some object file formats, the debugging information is encapsulated
142 in assembler directives known collectively as @dfn{stab} (symbol table)
143 directives, which are interspersed with the generated code. Stabs are
144 the native format for debugging information in the a.out and XCOFF
145 object file formats. The GNU tools can also emit stabs in the COFF and
146 ECOFF object file formats.
148 The assembler adds the information from stabs to the symbol information
149 it places by default in the symbol table and the string table of the
150 @file{.o} file it is building. The linker consolidates the @file{.o}
151 files into one executable file, with one symbol table and one string
152 table. Debuggers use the symbol and string tables in the executable as
153 a source of debugging information about the program.
156 @section Overview of Stab Format
158 There are three overall formats for stab assembler directives,
159 differentiated by the first word of the stab. The name of the directive
160 describes which combination of four possible data fields follows. It is
161 either @code{.stabs} (string), @code{.stabn} (number), or @code{.stabd}
162 (dot). IBM's XCOFF assembler uses @code{.stabx} (and some other
163 directives such as @code{.file} and @code{.bi}) instead of
164 @code{.stabs}, @code{.stabn} or @code{.stabd}.
166 The overall format of each class of stab is:
169 .stabs "@var{string}",@var{type},@var{other},@var{desc},@var{value}
170 .stabn @var{type},@var{other},@var{desc},@var{value}
171 .stabd @var{type},@var{other},@var{desc}
172 .stabx "@var{string}",@var{value},@var{type},@var{sdb-type}
175 @c what is the correct term for "current file location"? My AIX
176 @c assembler manual calls it "the value of the current location counter".
177 For @code{.stabn} and @code{.stabd}, there is no @var{string} (the
178 @code{n_strx} field is zero; see @ref{Symbol Tables}). For
179 @code{.stabd}, the @var{value} field is implicit and has the value of
180 the current file location. For @code{.stabx}, the @var{sdb-type} field
181 is unused for stabs and can always be set to zero. The @var{other}
182 field is almost always unused and can be set to zero.
184 The number in the @var{type} field gives some basic information about
185 which type of stab this is (or whether it @emph{is} a stab, as opposed
186 to an ordinary symbol). Each valid type number defines a different stab
187 type; further, the stab type defines the exact interpretation of, and
188 possible values for, any remaining @var{string}, @var{desc}, or
189 @var{value} fields present in the stab. @xref{Stab Types}, for a list
190 in numeric order of the valid @var{type} field values for stab directives.
193 @section The String Field
195 For most stabs the string field holds the meat of the
196 debugging information. The flexible nature of this field
197 is what makes stabs extensible. For some stab types the string field
198 contains only a name. For other stab types the contents can be a great
201 The overall format of the string field for most stab types is:
204 "@var{name}:@var{symbol-descriptor} @var{type-information}"
207 @var{name} is the name of the symbol represented by the stab; it can
208 contain a pair of colons (@pxref{Nested Symbols}). @var{name} can be
209 omitted, which means the stab represents an unnamed object. For
210 example, @samp{:t10=*2} defines type 10 as a pointer to type 2, but does
211 not give the type a name. Omitting the @var{name} field is supported by
212 AIX dbx and GDB after about version 4.8, but not other debuggers. GCC
213 sometimes uses a single space as the name instead of omitting the name
214 altogether; apparently that is supported by most debuggers.
216 The @var{symbol-descriptor} following the @samp{:} is an alphabetic
217 character that tells more specifically what kind of symbol the stab
218 represents. If the @var{symbol-descriptor} is omitted, but type
219 information follows, then the stab represents a local variable. For a
220 list of symbol descriptors, see @ref{Symbol Descriptors}. The @samp{c}
221 symbol descriptor is an exception in that it is not followed by type
222 information. @xref{Constants}.
224 @var{type-information} is either a @var{type-number}, or
225 @samp{@var{type-number}=}. A @var{type-number} alone is a type
226 reference, referring directly to a type that has already been defined.
228 The @samp{@var{type-number}=} form is a type definition, where the
229 number represents a new type which is about to be defined. The type
230 definition may refer to other types by number, and those type numbers
231 may be followed by @samp{=} and nested definitions. Also, the Lucid
232 compiler will repeat @samp{@var{type-number}=} more than once if it
233 wants to define several type numbers at once.
235 In a type definition, if the character that follows the equals sign is
236 non-numeric then it is a @var{type-descriptor}, and tells what kind of
237 type is about to be defined. Any other values following the
238 @var{type-descriptor} vary, depending on the @var{type-descriptor}.
239 @xref{Type Descriptors}, for a list of @var{type-descriptor} values. If
240 a number follows the @samp{=} then the number is a @var{type-reference}.
241 For a full description of types, @ref{Types}.
243 There is an AIX extension for type attributes. Following the @samp{=}
244 are any number of type attributes. Each one starts with @samp{@@} and
245 ends with @samp{;}. Debuggers, including AIX's dbx and GDB 4.10, skip
246 any type attributes they do not recognize. GDB 4.9 and other versions
247 of dbx may not do this. Because of a conflict with C++
248 (@pxref{Cplusplus}), new attributes should not be defined which begin
249 with a digit, @samp{(}, or @samp{-}; GDB may be unable to distinguish
250 those from the C++ type descriptor @samp{@@}. The attributes are:
253 @item a@var{boundary}
254 @var{boundary} is an integer specifying the alignment. I assume it
255 applies to all variables of this type.
258 Pointer class (for checking). Not sure what this means, or how
259 @var{integer} is interpreted.
262 Indicate this is a packed type, meaning that structure fields or array
263 elements are placed more closely in memory, to save memory at the
267 Size in bits of a variable of this type. This is fully supported by GDB
271 Indicate that this type is a string instead of an array of characters,
272 or a bitstring instead of a set. It doesn't change the layout of the
273 data being represented, but does enable the debugger to know which type
277 All of this can make the string field quite long. All versions of GDB,
278 and some versions of dbx, can handle arbitrarily long strings. But many
279 versions of dbx (or assemblers or linkers, I'm not sure which)
280 cretinously limit the strings to about 80 characters, so compilers which
281 must work with such systems need to split the @code{.stabs} directive
282 into several @code{.stabs} directives. Each stab duplicates every field
283 except the string field. The string field of every stab except the last
284 is marked as continued with a backslash at the end (in the assembly code
285 this may be written as a double backslash, depending on the assembler).
286 Removing the backslashes and concatenating the string fields of each
287 stab produces the original, long string. Just to be incompatible (or so
288 they don't have to worry about what the assembler does with
289 backslashes), AIX can use @samp{?} instead of backslash.
292 @section A Simple Example in C Source
294 To get the flavor of how stabs describe source information for a C
295 program, let's look at the simple program:
300 printf("Hello world");
304 When compiled with @samp{-g}, the program above yields the following
305 @file{.s} file. Line numbers have been added to make it easier to refer
306 to parts of the @file{.s} file in the description of the stabs that
310 @section The Simple Example at the Assembly Level
312 This simple ``hello world'' example demonstrates several of the stab
313 types used to describe C language source files.
317 2 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0
318 3 .stabs "hello.c",100,0,0,Ltext0
321 6 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0
322 7 .stabs "char:t2=r2;0;127;",128,0,0,0
323 8 .stabs "long int:t3=r1;-2147483648;2147483647;",128,0,0,0
324 9 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
325 10 .stabs "long unsigned int:t5=r1;0;-1;",128,0,0,0
326 11 .stabs "short int:t6=r1;-32768;32767;",128,0,0,0
327 12 .stabs "long long int:t7=r1;0;-1;",128,0,0,0
328 13 .stabs "short unsigned int:t8=r1;0;65535;",128,0,0,0
329 14 .stabs "long long unsigned int:t9=r1;0;-1;",128,0,0,0
330 15 .stabs "signed char:t10=r1;-128;127;",128,0,0,0
331 16 .stabs "unsigned char:t11=r1;0;255;",128,0,0,0
332 17 .stabs "float:t12=r1;4;0;",128,0,0,0
333 18 .stabs "double:t13=r1;8;0;",128,0,0,0
334 19 .stabs "long double:t14=r1;8;0;",128,0,0,0
335 20 .stabs "void:t15=15",128,0,0,0
338 23 .ascii "Hello, world!\12\0"
353 38 sethi %hi(LC0),%o1
354 39 or %o1,%lo(LC0),%o0
365 50 .stabs "main:F1",36,0,0,_main
366 51 .stabn 192,0,0,LBB2
367 52 .stabn 224,0,0,LBE2
370 @node Program Structure
371 @chapter Encoding the Structure of the Program
373 The elements of the program structure that stabs encode include the name
374 of the main function, the names of the source and include files, the
375 line numbers, procedure names and types, and the beginnings and ends of
379 * Main Program:: Indicate what the main program is
380 * Source Files:: The path and name of the source file
381 * Include Files:: Names of include files
384 * Nested Procedures::
386 * Alternate Entry Points:: Entering procedures except at the beginning.
390 @section Main Program
393 Most languages allow the main program to have any name. The
394 @code{N_MAIN} stab type tells the debugger the name that is used in this
395 program. Only the string field is significant; it is the name of
396 a function which is the main program. Most C compilers do not use this
397 stab (they expect the debugger to assume that the name is @code{main}),
398 but some C compilers emit an @code{N_MAIN} stab for the @code{main}
399 function. I'm not sure how XCOFF handles this.
402 @section Paths and Names of the Source Files
405 Before any other stabs occur, there must be a stab specifying the source
406 file. This information is contained in a symbol of stab type
407 @code{N_SO}; the string field contains the name of the file. The
408 value of the symbol is the start address of the portion of the
409 text section corresponding to that file.
411 With the Sun Solaris2 compiler, the desc field contains a
412 source-language code.
413 @c Do the debuggers use it? What are the codes? -djm
415 Some compilers (for example, GCC2 and SunOS4 @file{/bin/cc}) also
416 include the directory in which the source was compiled, in a second
417 @code{N_SO} symbol preceding the one containing the file name. This
418 symbol can be distinguished by the fact that it ends in a slash. Code
419 from the @code{cfront} C++ compiler can have additional @code{N_SO} symbols for
420 nonexistent source files after the @code{N_SO} for the real source file;
421 these are believed to contain no useful information.
426 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0 # @r{100 is N_SO}
427 .stabs "hello.c",100,0,0,Ltext0
433 Instead of @code{N_SO} symbols, XCOFF uses a @code{.file} assembler
434 directive which assembles to a @code{C_FILE} symbol; explaining this in
435 detail is outside the scope of this document.
437 @c FIXME: Exactly when should the empty N_SO be used? Why?
438 If it is useful to indicate the end of a source file, this is done with
439 an @code{N_SO} symbol with an empty string for the name. The value is
440 the address of the end of the text section for the file. For some
441 systems, there is no indication of the end of a source file, and you
442 just need to figure it ended when you see an @code{N_SO} for a different
443 source file, or a symbol ending in @code{.o} (which at least some
444 linkers insert to mark the start of a new @code{.o} file).
447 @section Names of Include Files
449 There are several schemes for dealing with include files: the
450 traditional @code{N_SOL} approach, Sun's @code{N_BINCL} approach, and the
451 XCOFF @code{C_BINCL} approach (which despite the similar name has little in
452 common with @code{N_BINCL}).
455 An @code{N_SOL} symbol specifies which include file subsequent symbols
456 refer to. The string field is the name of the file and the value is the
457 text address corresponding to the end of the previous include file and
458 the start of this one. To specify the main source file again, use an
459 @code{N_SOL} symbol with the name of the main source file.
464 The @code{N_BINCL} approach works as follows. An @code{N_BINCL} symbol
465 specifies the start of an include file. In an object file, only the
466 string is significant; the Sun linker puts data into some of the other
467 fields. The end of the include file is marked by an @code{N_EINCL}
468 symbol (which has no string field). In an object file, there is no
469 significant data in the @code{N_EINCL} symbol. @code{N_BINCL} and
470 @code{N_EINCL} can be nested.
472 If the linker detects that two source files have identical stabs between
473 an @code{N_BINCL} and @code{N_EINCL} pair (as will generally be the case
474 for a header file), then it only puts out the stabs once. Each
475 additional occurance is replaced by an @code{N_EXCL} symbol. I believe
476 the Sun (SunOS4, not sure about Solaris) linker is the only one which
477 supports this feature.
479 The SunOS4 linker sets the value of a @code{N_BINCL} symbol to the total
480 of all the characters in the stabs strings included in the header file,
481 omitting the file number. The value of an @code{N_EXCL} symbol is the
482 same as the value of the @code{N_BINCL} symbol it replaces. I do not
483 know if this information is used by anything. The @code{N_EINCL} value,
484 and the values of the other and description fields for all three, appear
489 For the start of an include file in XCOFF, use the @file{.bi} assembler
490 directive, which generates a @code{C_BINCL} symbol. A @file{.ei}
491 directive, which generates a @code{C_EINCL} symbol, denotes the end of
492 the include file. Both directives are followed by the name of the
493 source file in quotes, which becomes the string for the symbol.
494 The value of each symbol, produced automatically by the assembler
495 and linker, is the offset into the executable of the beginning
496 (inclusive, as you'd expect) or end (inclusive, as you would not expect)
497 of the portion of the COFF line table that corresponds to this include
498 file. @code{C_BINCL} and @code{C_EINCL} do not nest.
501 @section Line Numbers
504 An @code{N_SLINE} symbol represents the start of a source line. The
505 desc field contains the line number and the value contains the code
506 address for the start of that source line. On most machines the address
507 is absolute; for stabs in sections (@pxref{Stab Sections}), it is
508 relative to the function in which the @code{N_SLINE} symbol occurs.
512 GNU documents @code{N_DSLINE} and @code{N_BSLINE} symbols for line
513 numbers in the data or bss segments, respectively. They are identical
514 to @code{N_SLINE} but are relocated differently by the linker. They
515 were intended to be used to describe the source location of a variable
516 declaration, but I believe that GCC2 actually puts the line number in
517 the desc field of the stab for the variable itself. GDB has been
518 ignoring these symbols (unless they contain a string field) since
521 For single source lines that generate discontiguous code, such as flow
522 of control statements, there may be more than one line number entry for
523 the same source line. In this case there is a line number entry at the
524 start of each code range, each with the same line number.
526 XCOFF does not use stabs for line numbers. Instead, it uses COFF line
527 numbers (which are outside the scope of this document). Standard COFF
528 line numbers cannot deal with include files, but in XCOFF this is fixed
529 with the @code{C_BINCL} method of marking include files (@pxref{Include
535 @findex N_FUN, for functions
537 @findex N_STSYM, for functions (Sun acc)
538 @findex N_GSYM, for functions (Sun acc)
539 All of the following stabs normally use the @code{N_FUN} symbol type.
540 However, Sun's @code{acc} compiler on SunOS4 uses @code{N_GSYM} and
541 @code{N_STSYM}, which means that the value of the stab for the function
542 is useless and the debugger must get the address of the function from
543 the non-stab symbols instead. On systems where non-stab symbols have
544 leading underscores, the stabs will lack underscores and the debugger
545 needs to know about the leading underscore to match up the stab and the
546 non-stab symbol. BSD Fortran is said to use @code{N_FNAME} with the
547 same restriction; the value of the symbol is not useful (I'm not sure it
548 really does use this, because GDB doesn't handle this and no one has
552 A function is represented by an @samp{F} symbol descriptor for a global
553 (extern) function, and @samp{f} for a static (local) function. For
554 a.out, the value of the symbol is the address of the start of the
555 function; it is already relocated. For stabs in ELF, the SunPRO
556 compiler version 2.0.1 and GCC put out an address which gets relocated
557 by the linker. In a future release SunPRO is planning to put out zero,
558 in which case the address can be found from the ELF (non-stab) symbol.
559 Because looking things up in the ELF symbols would probably be slow, I'm
560 not sure how to find which symbol of that name is the right one, and
561 this doesn't provide any way to deal with nested functions, it would
562 probably be better to make the value of the stab an address relative to
563 the start of the file, or just absolute. See @ref{ELF Linker
564 Relocation} for more information on linker relocation of stabs in ELF
565 files. For XCOFF, the stab uses the @code{C_FUN} storage class and the
566 value of the stab is meaningless; the address of the function can be
567 found from the csect symbol (XTY_LD/XMC_PR).
569 The type information of the stab represents the return type of the
570 function; thus @samp{foo:f5} means that foo is a function returning type
571 5. There is no need to try to get the line number of the start of the
572 function from the stab for the function; it is in the next
573 @code{N_SLINE} symbol.
575 @c FIXME: verify whether the "I suspect" below is true or not.
576 Some compilers (such as Sun's Solaris compiler) support an extension for
577 specifying the types of the arguments. I suspect this extension is not
578 used for old (non-prototyped) function definitions in C. If the
579 extension is in use, the type information of the stab for the function
580 is followed by type information for each argument, with each argument
581 preceded by @samp{;}. An argument type of 0 means that additional
582 arguments are being passed, whose types and number may vary (@samp{...}
583 in ANSI C). GDB has tolerated this extension (parsed the syntax, if not
584 necessarily used the information) since at least version 4.8; I don't
585 know whether all versions of dbx tolerate it. The argument types given
586 here are not redundant with the symbols for the formal parameters
587 (@pxref{Parameters}); they are the types of the arguments as they are
588 passed, before any conversions might take place. For example, if a C
589 function which is declared without a prototype takes a @code{float}
590 argument, the value is passed as a @code{double} but then converted to a
591 @code{float}. Debuggers need to use the types given in the arguments
592 when printing values, but when calling the function they need to use the
593 types given in the symbol defining the function.
595 If the return type and types of arguments of a function which is defined
596 in another source file are specified (i.e., a function prototype in ANSI
597 C), traditionally compilers emit no stab; the only way for the debugger
598 to find the information is if the source file where the function is
599 defined was also compiled with debugging symbols. As an extension the
600 Solaris compiler uses symbol descriptor @samp{P} followed by the return
601 type of the function, followed by the arguments, each preceded by
602 @samp{;}, as in a stab with symbol descriptor @samp{f} or @samp{F}.
603 This use of symbol descriptor @samp{P} can be distinguished from its use
604 for register parameters (@pxref{Register Parameters}) by the fact that it has
605 symbol type @code{N_FUN}.
607 The AIX documentation also defines symbol descriptor @samp{J} as an
608 internal function. I assume this means a function nested within another
609 function. It also says symbol descriptor @samp{m} is a module in
610 Modula-2 or extended Pascal.
612 Procedures (functions which do not return values) are represented as
613 functions returning the @code{void} type in C. I don't see why this couldn't
614 be used for all languages (inventing a @code{void} type for this purpose if
615 necessary), but the AIX documentation defines @samp{I}, @samp{P}, and
616 @samp{Q} for internal, global, and static procedures, respectively.
617 These symbol descriptors are unusual in that they are not followed by
620 The following example shows a stab for a function @code{main} which
621 returns type number @code{1}. The @code{_main} specified for the value
622 is a reference to an assembler label which is used to fill in the start
623 address of the function.
626 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
629 The stab representing a procedure is located immediately following the
630 code of the procedure. This stab is in turn directly followed by a
631 group of other stabs describing elements of the procedure. These other
632 stabs describe the procedure's parameters, its block local variables, and
635 @node Nested Procedures
636 @section Nested Procedures
638 For any of the symbol descriptors representing procedures, after the
639 symbol descriptor and the type information is optionally a scope
640 specifier. This consists of a comma, the name of the procedure, another
641 comma, and the name of the enclosing procedure. The first name is local
642 to the scope specified, and seems to be redundant with the name of the
643 symbol (before the @samp{:}). This feature is used by GCC, and
644 presumably Pascal, Modula-2, etc., compilers, for nested functions.
646 If procedures are nested more than one level deep, only the immediately
647 containing scope is specified. For example, this code:
659 return baz (x + 2 * y);
661 return x + bar (3 * x);
669 .stabs "baz:f1,baz,bar",36,0,0,_baz.15 # @r{36 is N_FUN}
670 .stabs "bar:f1,bar,foo",36,0,0,_bar.12
671 .stabs "foo:F1",36,0,0,_foo
674 @node Block Structure
675 @section Block Structure
679 @c For GCC 2.5.8 or so stabs-in-coff, these are absolute instead of
680 @c function relative (as documented below). But GDB has never been able
681 @c to deal with that (it had wanted them to be relative to the file, but
682 @c I just fixed that (between GDB 4.12 and 4.13)), so it is function
683 @c relative just like ELF and SOM and the below documentation.
684 The program's block structure is represented by the @code{N_LBRAC} (left
685 brace) and the @code{N_RBRAC} (right brace) stab types. The variables
686 defined inside a block precede the @code{N_LBRAC} symbol for most
687 compilers, including GCC. Other compilers, such as the Convex, Acorn
688 RISC machine, and Sun @code{acc} compilers, put the variables after the
689 @code{N_LBRAC} symbol. The values of the @code{N_LBRAC} and
690 @code{N_RBRAC} symbols are the start and end addresses of the code of
691 the block, respectively. For most machines, they are relative to the
692 starting address of this source file. For the Gould NP1, they are
693 absolute. For stabs in sections (@pxref{Stab Sections}), they are
694 relative to the function in which they occur.
696 The @code{N_LBRAC} and @code{N_RBRAC} stabs that describe the block
697 scope of a procedure are located after the @code{N_FUN} stab that
698 represents the procedure itself.
700 Sun documents the desc field of @code{N_LBRAC} and
701 @code{N_RBRAC} symbols as containing the nesting level of the block.
702 However, dbx seems to not care, and GCC always sets desc to
708 For XCOFF, block scope is indicated with @code{C_BLOCK} symbols. If the
709 name of the symbol is @samp{.bb}, then it is the beginning of the block;
710 if the name of the symbol is @samp{.be}; it is the end of the block.
712 @node Alternate Entry Points
713 @section Alternate Entry Points
717 Some languages, like Fortran, have the ability to enter procedures at
718 some place other than the beginning. One can declare an alternate entry
719 point. The @code{N_ENTRY} stab is for this; however, the Sun FORTRAN
720 compiler doesn't use it. According to AIX documentation, only the name
721 of a @code{C_ENTRY} stab is significant; the address of the alternate
722 entry point comes from the corresponding external symbol. A previous
723 revision of this document said that the value of an @code{N_ENTRY} stab
724 was the address of the alternate entry point, but I don't know the
725 source for that information.
730 The @samp{c} symbol descriptor indicates that this stab represents a
731 constant. This symbol descriptor is an exception to the general rule
732 that symbol descriptors are followed by type information. Instead, it
733 is followed by @samp{=} and one of the following:
737 Boolean constant. @var{value} is a numeric value; I assume it is 0 for
741 Character constant. @var{value} is the numeric value of the constant.
743 @item e @var{type-information} , @var{value}
744 Constant whose value can be represented as integral.
745 @var{type-information} is the type of the constant, as it would appear
746 after a symbol descriptor (@pxref{String Field}). @var{value} is the
747 numeric value of the constant. GDB 4.9 does not actually get the right
748 value if @var{value} does not fit in a host @code{int}, but it does not
749 do anything violent, and future debuggers could be extended to accept
750 integers of any size (whether unsigned or not). This constant type is
751 usually documented as being only for enumeration constants, but GDB has
752 never imposed that restriction; I don't know about other debuggers.
755 Integer constant. @var{value} is the numeric value. The type is some
756 sort of generic integer type (for GDB, a host @code{int}); to specify
757 the type explicitly, use @samp{e} instead.
760 Real constant. @var{value} is the real value, which can be @samp{INF}
761 (optionally preceded by a sign) for infinity, @samp{QNAN} for a quiet
762 NaN (not-a-number), or @samp{SNAN} for a signalling NaN. If it is a
763 normal number the format is that accepted by the C library function
767 String constant. @var{string} is a string enclosed in either @samp{'}
768 (in which case @samp{'} characters within the string are represented as
769 @samp{\'} or @samp{"} (in which case @samp{"} characters within the
770 string are represented as @samp{\"}).
772 @item S @var{type-information} , @var{elements} , @var{bits} , @var{pattern}
773 Set constant. @var{type-information} is the type of the constant, as it
774 would appear after a symbol descriptor (@pxref{String Field}).
775 @var{elements} is the number of elements in the set (does this means
776 how many bits of @var{pattern} are actually used, which would be
777 redundant with the type, or perhaps the number of bits set in
778 @var{pattern}? I don't get it), @var{bits} is the number of bits in the
779 constant (meaning it specifies the length of @var{pattern}, I think),
780 and @var{pattern} is a hexadecimal representation of the set. AIX
781 documentation refers to a limit of 32 bytes, but I see no reason why
782 this limit should exist. This form could probably be used for arbitrary
783 constants, not just sets; the only catch is that @var{pattern} should be
784 understood to be target, not host, byte order and format.
787 The boolean, character, string, and set constants are not supported by
788 GDB 4.9, but it ignores them. GDB 4.8 and earlier gave an error
789 message and refused to read symbols from the file containing the
792 The above information is followed by @samp{;}.
797 Different types of stabs describe the various ways that variables can be
798 allocated: on the stack, globally, in registers, in common blocks,
799 statically, or as arguments to a function.
802 * Stack Variables:: Variables allocated on the stack.
803 * Global Variables:: Variables used by more than one source file.
804 * Register Variables:: Variables in registers.
805 * Common Blocks:: Variables statically allocated together.
806 * Statics:: Variables local to one source file.
807 * Based Variables:: Fortran pointer based variables.
808 * Parameters:: Variables for arguments to functions.
811 @node Stack Variables
812 @section Automatic Variables Allocated on the Stack
814 If a variable's scope is local to a function and its lifetime is only as
815 long as that function executes (C calls such variables
816 @dfn{automatic}), it can be allocated in a register (@pxref{Register
817 Variables}) or on the stack.
819 @findex N_LSYM, for stack variables
821 Each variable allocated on the stack has a stab with the symbol
822 descriptor omitted. Since type information should begin with a digit,
823 @samp{-}, or @samp{(}, only those characters precluded from being used
824 for symbol descriptors. However, the Acorn RISC machine (ARM) is said
825 to get this wrong: it puts out a mere type definition here, without the
826 preceding @samp{@var{type-number}=}. This is a bad idea; there is no
827 guarantee that type descriptors are distinct from symbol descriptors.
828 Stabs for stack variables use the @code{N_LSYM} stab type, or
829 @code{C_LSYM} for XCOFF.
831 The value of the stab is the offset of the variable within the
832 local variables. On most machines this is an offset from the frame
833 pointer and is negative. The location of the stab specifies which block
834 it is defined in; see @ref{Block Structure}.
836 For example, the following C code:
846 produces the following stabs:
849 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
850 .stabs "x:1",128,0,0,-12 # @r{128 is N_LSYM}
851 .stabn 192,0,0,LBB2 # @r{192 is N_LBRAC}
852 .stabn 224,0,0,LBE2 # @r{224 is N_RBRAC}
855 @xref{Procedures} for more information on the @code{N_FUN} stab, and
856 @ref{Block Structure} for more information on the @code{N_LBRAC} and
857 @code{N_RBRAC} stabs.
859 @node Global Variables
860 @section Global Variables
864 @c FIXME: verify for sure that it really is C_GSYM on XCOFF
865 A variable whose scope is not specific to just one source file is
866 represented by the @samp{G} symbol descriptor. These stabs use the
867 @code{N_GSYM} stab type (C_GSYM for XCOFF). The type information for
868 the stab (@pxref{String Field}) gives the type of the variable.
870 For example, the following source code:
877 yields the following assembly code:
880 .stabs "g_foo:G2",32,0,0,0 # @r{32 is N_GSYM}
887 The address of the variable represented by the @code{N_GSYM} is not
888 contained in the @code{N_GSYM} stab. The debugger gets this information
889 from the external symbol for the global variable. In the example above,
890 the @code{.global _g_foo} and @code{_g_foo:} lines tell the assembler to
891 produce an external symbol.
893 Some compilers, like GCC, output @code{N_GSYM} stabs only once, where
894 the variable is defined. Other compilers, like SunOS4 /bin/cc, output a
895 @code{N_GSYM} stab for each compilation unit which references the
898 @node Register Variables
899 @section Register Variables
903 @c According to an old version of this manual, AIX uses C_RPSYM instead
904 @c of C_RSYM. I am skeptical; this should be verified.
905 Register variables have their own stab type, @code{N_RSYM}
906 (@code{C_RSYM} for XCOFF), and their own symbol descriptor, @samp{r}.
907 The stab's value is the number of the register where the variable data
909 @c .stabs "name:type",N_RSYM,0,RegSize,RegNumber (Sun doc)
911 AIX defines a separate symbol descriptor @samp{d} for floating point
912 registers. This seems unnecessary; why not just just give floating
913 point registers different register numbers? I have not verified whether
914 the compiler actually uses @samp{d}.
916 If the register is explicitly allocated to a global variable, but not
920 register int g_bar asm ("%g5");
924 then the stab may be emitted at the end of the object file, with
925 the other bss symbols.
928 @section Common Blocks
930 A common block is a statically allocated section of memory which can be
931 referred to by several source files. It may contain several variables.
932 I believe Fortran is the only language with this feature.
938 A @code{N_BCOMM} stab begins a common block and an @code{N_ECOMM} stab
939 ends it. The only field that is significant in these two stabs is the
940 string, which names a normal (non-debugging) symbol that gives the
941 address of the common block. According to IBM documentation, only the
942 @code{N_BCOMM} has the name of the common block (even though their
943 compiler actually puts it both places).
947 The stabs for the members of the common block are between the
948 @code{N_BCOMM} and the @code{N_ECOMM}; the value of each stab is the
949 offset within the common block of that variable. IBM uses the
950 @code{C_ECOML} stab type, and there is a corresponding @code{N_ECOML}
951 stab type, but Sun's Fortran compiler uses @code{N_GSYM} instead. The
952 variables within a common block use the @samp{V} symbol descriptor (I
953 believe this is true of all Fortran variables). Other stabs (at least
954 type declarations using @code{C_DECL}) can also be between the
955 @code{N_BCOMM} and the @code{N_ECOMM}.
958 @section Static Variables
960 Initialized static variables are represented by the @samp{S} and
961 @samp{V} symbol descriptors. @samp{S} means file scope static, and
962 @samp{V} means procedure scope static. One exception: in XCOFF, IBM's
963 xlc compiler always uses @samp{V}, and whether it is file scope or not
964 is distinguished by whether the stab is located within a function.
966 @c This is probably not worth mentioning; it is only true on the sparc
967 @c for `double' variables which although declared const are actually in
968 @c the data segment (the text segment can't guarantee 8 byte alignment).
970 @c 2.4.5 has a bug in that it uses @code{N_FUN}, so neither dbx nor GDB can
971 @c find the variables)
974 @findex N_FUN, for variables
976 In a.out files, @code{N_STSYM} means the data section, @code{N_FUN}
977 means the text section, and @code{N_LCSYM} means the bss section. For
978 those systems with a read-only data section separate from the text
979 section (Solaris), @code{N_ROSYM} means the read-only data section.
981 For example, the source lines:
984 static const int var_const = 5;
985 static int var_init = 2;
986 static int var_noinit;
990 yield the following stabs:
993 .stabs "var_const:S1",36,0,0,_var_const # @r{36 is N_FUN}
995 .stabs "var_init:S1",38,0,0,_var_init # @r{38 is N_STSYM}
997 .stabs "var_noinit:S1",40,0,0,_var_noinit # @r{40 is N_LCSYM}
1003 In XCOFF files, the stab type need not indicate the section;
1004 @code{C_STSYM} can be used for all statics. Also, each static variable
1005 is enclosed in a static block. A @code{C_BSTAT} (emitted with a
1006 @samp{.bs} assembler directive) symbol begins the static block; its
1007 value is the symbol number of the csect symbol whose value is the
1008 address of the static block, its section is the section of the variables
1009 in that static block, and its name is @samp{.bs}. A @code{C_ESTAT}
1010 (emitted with a @samp{.es} assembler directive) symbol ends the static
1011 block; its name is @samp{.es} and its value and section are ignored.
1013 In ECOFF files, the storage class is used to specify the section, so the
1014 stab type need not indicate the section.
1016 In ELF files, for the SunPRO compiler version 2.0.1, symbol descriptor
1017 @samp{S} means that the address is absolute (the linker relocates it)
1018 and symbol descriptor @samp{V} means that the address is relative to the
1019 start of the relevant section for that compilation unit. SunPRO has
1020 plans to have the linker stop relocating stabs; I suspect that their the
1021 debugger gets the address from the corresponding ELF (not stab) symbol.
1022 I'm not sure how to find which symbol of that name is the right one.
1023 The clean way to do all this would be to have a the value of a symbol
1024 descriptor @samp{S} symbol be an offset relative to the start of the
1025 file, just like everything else, but that introduces obvious
1026 compatibility problems. For more information on linker stab relocation,
1027 @xref{ELF Linker Relocation}.
1029 @node Based Variables
1030 @section Fortran Based Variables
1032 Fortran (at least, the Sun and SGI dialects of FORTRAN-77) has a feature
1033 which allows allocating arrays with @code{malloc}, but which avoids
1034 blurring the line between arrays and pointers the way that C does. In
1035 stabs such a variable uses the @samp{b} symbol descriptor.
1037 For example, the Fortran declarations
1040 real foo, foo10(10), foo10_5(10,5)
1042 pointer (foo10p, foo10)
1043 pointer (foo105p, foo10_5)
1051 foo10_5:bar3;1;5;ar3;1;10;6
1054 In this example, @code{real} is type 6 and type 3 is an integral type
1055 which is the type of the subscripts of the array (probably
1058 The @samp{b} symbol descriptor is like @samp{V} in that it denotes a
1059 statically allocated symbol whose scope is local to a function; see
1060 @xref{Statics}. The value of the symbol, instead of being the address
1061 of the variable itself, is the address of a pointer to that variable.
1062 So in the above example, the value of the @code{foo} stab is the address
1063 of a pointer to a real, the value of the @code{foo10} stab is the
1064 address of a pointer to a 10-element array of reals, and the value of
1065 the @code{foo10_5} stab is the address of a pointer to a 5-element array
1066 of 10-element arrays of reals.
1071 Formal parameters to a function are represented by a stab (or sometimes
1072 two; see below) for each parameter. The stabs are in the order in which
1073 the debugger should print the parameters (i.e., the order in which the
1074 parameters are declared in the source file). The exact form of the stab
1075 depends on how the parameter is being passed.
1079 Parameters passed on the stack use the symbol descriptor @samp{p} and
1080 the @code{N_PSYM} symbol type (or @code{C_PSYM} for XCOFF). The value
1081 of the symbol is an offset used to locate the parameter on the stack;
1082 its exact meaning is machine-dependent, but on most machines it is an
1083 offset from the frame pointer.
1085 As a simple example, the code:
1096 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
1097 .stabs "argc:p1",160,0,0,68 # @r{160 is N_PSYM}
1098 .stabs "argv:p20=*21=*2",160,0,0,72
1101 The type definition of @code{argv} is interesting because it contains
1102 several type definitions. Type 21 is pointer to type 2 (char) and
1103 @code{argv} (type 20) is pointer to type 21.
1105 @c FIXME: figure out what these mean and describe them coherently.
1106 The following symbol descriptors are also said to go with @code{N_PSYM}.
1107 The value of the symbol is said to be an offset from the argument
1108 pointer (I'm not sure whether this is true or not).
1112 pF Fortran function parameter
1113 X (function result variable)
1117 * Register Parameters::
1118 * Local Variable Parameters::
1119 * Reference Parameters::
1120 * Conformant Arrays::
1123 @node Register Parameters
1124 @subsection Passing Parameters in Registers
1126 If the parameter is passed in a register, then traditionally there are
1127 two symbols for each argument:
1130 .stabs "arg:p1" . . . ; N_PSYM
1131 .stabs "arg:r1" . . . ; N_RSYM
1134 Debuggers use the second one to find the value, and the first one to
1135 know that it is an argument.
1138 @findex N_RSYM, for parameters
1139 Because that approach is kind of ugly, some compilers use symbol
1140 descriptor @samp{P} or @samp{R} to indicate an argument which is in a
1141 register. Symbol type @code{C_RPSYM} is used in XCOFF and @code{N_RSYM}
1142 is used otherwise. The symbol's value is the register number. @samp{P}
1143 and @samp{R} mean the same thing; the difference is that @samp{P} is a
1144 GNU invention and @samp{R} is an IBM (XCOFF) invention. As of version
1145 4.9, GDB should handle either one.
1147 There is at least one case where GCC uses a @samp{p} and @samp{r} pair
1148 rather than @samp{P}; this is where the argument is passed in the
1149 argument list and then loaded into a register.
1151 According to the AIX documentation, symbol descriptor @samp{D} is for a
1152 parameter passed in a floating point register. This seems
1153 unnecessary---why not just use @samp{R} with a register number which
1154 indicates that it's a floating point register? I haven't verified
1155 whether the system actually does what the documentation indicates.
1157 @c FIXME: On the hppa this is for any type > 8 bytes, I think, and not
1158 @c for small structures (investigate).
1159 On the sparc and hppa, for a @samp{P} symbol whose type is a structure
1160 or union, the register contains the address of the structure. On the
1161 sparc, this is also true of a @samp{p} and @samp{r} pair (using Sun
1162 @code{cc}) or a @samp{p} symbol. However, if a (small) structure is
1163 really in a register, @samp{r} is used. And, to top it all off, on the
1164 hppa it might be a structure which was passed on the stack and loaded
1165 into a register and for which there is a @samp{p} and @samp{r} pair! I
1166 believe that symbol descriptor @samp{i} is supposed to deal with this
1167 case (it is said to mean "value parameter by reference, indirect
1168 access"; I don't know the source for this information), but I don't know
1169 details or what compilers or debuggers use it, if any (not GDB or GCC).
1170 It is not clear to me whether this case needs to be dealt with
1171 differently than parameters passed by reference (@pxref{Reference Parameters}).
1173 @node Local Variable Parameters
1174 @subsection Storing Parameters as Local Variables
1176 There is a case similar to an argument in a register, which is an
1177 argument that is actually stored as a local variable. Sometimes this
1178 happens when the argument was passed in a register and then the compiler
1179 stores it as a local variable. If possible, the compiler should claim
1180 that it's in a register, but this isn't always done.
1182 If a parameter is passed as one type and converted to a smaller type by
1183 the prologue (for example, the parameter is declared as a @code{float},
1184 but the calling conventions specify that it is passed as a
1185 @code{double}), then GCC2 (sometimes) uses a pair of symbols. The first
1186 symbol uses symbol descriptor @samp{p} and the type which is passed.
1187 The second symbol has the type and location which the parameter actually
1188 has after the prologue. For example, suppose the following C code
1189 appears with no prototypes involved:
1198 if @code{f} is passed as a double at stack offset 8, and the prologue
1199 converts it to a float in register number 0, then the stabs look like:
1202 .stabs "f:p13",160,0,3,8 # @r{160 is @code{N_PSYM}, here 13 is @code{double}}
1203 .stabs "f:r12",64,0,3,0 # @r{64 is @code{N_RSYM}, here 12 is @code{float}}
1206 In both stabs 3 is the line number where @code{f} is declared
1207 (@pxref{Line Numbers}).
1209 @findex N_LSYM, for parameter
1210 GCC, at least on the 960, has another solution to the same problem. It
1211 uses a single @samp{p} symbol descriptor for an argument which is stored
1212 as a local variable but uses @code{N_LSYM} instead of @code{N_PSYM}. In
1213 this case, the value of the symbol is an offset relative to the local
1214 variables for that function, not relative to the arguments; on some
1215 machines those are the same thing, but not on all.
1217 @c This is mostly just background info; the part that logically belongs
1218 @c here is the last sentence.
1219 On the VAX or on other machines in which the calling convention includes
1220 the number of words of arguments actually passed, the debugger (GDB at
1221 least) uses the parameter symbols to keep track of whether it needs to
1222 print nameless arguments in addition to the formal parameters which it
1223 has printed because each one has a stab. For example, in
1226 extern int fprintf (FILE *stream, char *format, @dots{});
1228 fprintf (stdout, "%d\n", x);
1231 there are stabs for @code{stream} and @code{format}. On most machines,
1232 the debugger can only print those two arguments (because it has no way
1233 of knowing that additional arguments were passed), but on the VAX or
1234 other machines with a calling convention which indicates the number of
1235 words of arguments, the debugger can print all three arguments. To do
1236 so, the parameter symbol (symbol descriptor @samp{p}) (not necessarily
1237 @samp{r} or symbol descriptor omitted symbols) needs to contain the
1238 actual type as passed (for example, @code{double} not @code{float} if it
1239 is passed as a double and converted to a float).
1241 @node Reference Parameters
1242 @subsection Passing Parameters by Reference
1244 If the parameter is passed by reference (e.g., Pascal @code{VAR}
1245 parameters), then the symbol descriptor is @samp{v} if it is in the
1246 argument list, or @samp{a} if it in a register. Other than the fact
1247 that these contain the address of the parameter rather than the
1248 parameter itself, they are identical to @samp{p} and @samp{R},
1249 respectively. I believe @samp{a} is an AIX invention; @samp{v} is
1250 supported by all stabs-using systems as far as I know.
1252 @node Conformant Arrays
1253 @subsection Passing Conformant Array Parameters
1255 @c Is this paragraph correct? It is based on piecing together patchy
1256 @c information and some guesswork
1257 Conformant arrays are a feature of Modula-2, and perhaps other
1258 languages, in which the size of an array parameter is not known to the
1259 called function until run-time. Such parameters have two stabs: a
1260 @samp{x} for the array itself, and a @samp{C}, which represents the size
1261 of the array. The value of the @samp{x} stab is the offset in the
1262 argument list where the address of the array is stored (it this right?
1263 it is a guess); the value of the @samp{C} stab is the offset in the
1264 argument list where the size of the array (in elements? in bytes?) is
1268 @chapter Defining Types
1270 The examples so far have described types as references to previously
1271 defined types, or defined in terms of subranges of or pointers to
1272 previously defined types. This chapter describes the other type
1273 descriptors that may follow the @samp{=} in a type definition.
1276 * Builtin Types:: Integers, floating point, void, etc.
1277 * Miscellaneous Types:: Pointers, sets, files, etc.
1278 * Cross-References:: Referring to a type not yet defined.
1279 * Subranges:: A type with a specific range.
1280 * Arrays:: An aggregate type of same-typed elements.
1281 * Strings:: Like an array but also has a length.
1282 * Enumerations:: Like an integer but the values have names.
1283 * Structures:: An aggregate type of different-typed elements.
1284 * Typedefs:: Giving a type a name.
1285 * Unions:: Different types sharing storage.
1290 @section Builtin Types
1292 Certain types are built in (@code{int}, @code{short}, @code{void},
1293 @code{float}, etc.); the debugger recognizes these types and knows how
1294 to handle them. Thus, don't be surprised if some of the following ways
1295 of specifying builtin types do not specify everything that a debugger
1296 would need to know about the type---in some cases they merely specify
1297 enough information to distinguish the type from other types.
1299 The traditional way to define builtin types is convolunted, so new ways
1300 have been invented to describe them. Sun's @code{acc} uses special
1301 builtin type descriptors (@samp{b} and @samp{R}), and IBM uses negative
1302 type numbers. GDB accepts all three ways, as of version 4.8; dbx just
1303 accepts the traditional builtin types and perhaps one of the other two
1304 formats. The following sections describe each of these formats.
1307 * Traditional Builtin Types:: Put on your seatbelts and prepare for kludgery
1308 * Builtin Type Descriptors:: Builtin types with special type descriptors
1309 * Negative Type Numbers:: Builtin types using negative type numbers
1312 @node Traditional Builtin Types
1313 @subsection Traditional Builtin Types
1315 This is the traditional, convoluted method for defining builtin types.
1316 There are several classes of such type definitions: integer, floating
1317 point, and @code{void}.
1320 * Traditional Integer Types::
1321 * Traditional Other Types::
1324 @node Traditional Integer Types
1325 @subsubsection Traditional Integer Types
1327 Often types are defined as subranges of themselves. If the bounding values
1328 fit within an @code{int}, then they are given normally. For example:
1331 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0 # @r{128 is N_LSYM}
1332 .stabs "char:t2=r2;0;127;",128,0,0,0
1335 Builtin types can also be described as subranges of @code{int}:
1338 .stabs "unsigned short:t6=r1;0;65535;",128,0,0,0
1341 If the lower bound of a subrange is 0 and the upper bound is -1,
1342 the type is an unsigned integral type whose bounds are too
1343 big to describe in an @code{int}. Traditionally this is only used for
1344 @code{unsigned int} and @code{unsigned long}:
1347 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
1350 For larger types, GCC 2.4.5 puts out bounds in octal, with one or more
1351 leading zeroes. In this case a negative bound consists of a number
1352 which is a 1 bit (for the sign bit) followed by a 0 bit for each bit in
1353 the number (except the sign bit), and a positive bound is one which is a
1354 1 bit for each bit in the number (except possibly the sign bit). All
1355 known versions of dbx and GDB version 4 accept this (at least in the
1356 sense of not refusing to process the file), but GDB 3.5 refuses to read
1357 the whole file containing such symbols. So GCC 2.3.3 did not output the
1358 proper size for these types. As an example of octal bounds, the string
1359 fields of the stabs for 64 bit integer types look like:
1361 @c .stabs directives, etc., omitted to make it fit on the page.
1363 long int:t3=r1;001000000000000000000000;000777777777777777777777;
1364 long unsigned int:t5=r1;000000000000000000000000;001777777777777777777777;
1367 If the lower bound of a subrange is 0 and the upper bound is negative,
1368 the type is an unsigned integral type whose size in bytes is the
1369 absolute value of the upper bound. I believe this is a Convex
1370 convention for @code{unsigned long long}.
1372 If the lower bound of a subrange is negative and the upper bound is 0,
1373 the type is a signed integral type whose size in bytes is
1374 the absolute value of the lower bound. I believe this is a Convex
1375 convention for @code{long long}. To distinguish this from a legitimate
1376 subrange, the type should be a subrange of itself. I'm not sure whether
1377 this is the case for Convex.
1379 @node Traditional Other Types
1380 @subsubsection Traditional Other Types
1382 If the upper bound of a subrange is 0 and the lower bound is positive,
1383 the type is a floating point type, and the lower bound of the subrange
1384 indicates the number of bytes in the type:
1387 .stabs "float:t12=r1;4;0;",128,0,0,0
1388 .stabs "double:t13=r1;8;0;",128,0,0,0
1391 However, GCC writes @code{long double} the same way it writes
1392 @code{double}, so there is no way to distinguish.
1395 .stabs "long double:t14=r1;8;0;",128,0,0,0
1398 Complex types are defined the same way as floating-point types; there is
1399 no way to distinguish a single-precision complex from a double-precision
1400 floating-point type.
1402 The C @code{void} type is defined as itself:
1405 .stabs "void:t15=15",128,0,0,0
1408 I'm not sure how a boolean type is represented.
1410 @node Builtin Type Descriptors
1411 @subsection Defining Builtin Types Using Builtin Type Descriptors
1413 This is the method used by Sun's @code{acc} for defining builtin types.
1414 These are the type descriptors to define builtin types:
1417 @c FIXME: clean up description of width and offset, once we figure out
1419 @item b @var{signed} @var{char-flag} @var{width} ; @var{offset} ; @var{nbits} ;
1420 Define an integral type. @var{signed} is @samp{u} for unsigned or
1421 @samp{s} for signed. @var{char-flag} is @samp{c} which indicates this
1422 is a character type, or is omitted. I assume this is to distinguish an
1423 integral type from a character type of the same size, for example it
1424 might make sense to set it for the C type @code{wchar_t} so the debugger
1425 can print such variables differently (Solaris does not do this). Sun
1426 sets it on the C types @code{signed char} and @code{unsigned char} which
1427 arguably is wrong. @var{width} and @var{offset} appear to be for small
1428 objects stored in larger ones, for example a @code{short} in an
1429 @code{int} register. @var{width} is normally the number of bytes in the
1430 type. @var{offset} seems to always be zero. @var{nbits} is the number
1431 of bits in the type.
1433 Note that type descriptor @samp{b} used for builtin types conflicts with
1434 its use for Pascal space types (@pxref{Miscellaneous Types}); they can
1435 be distinguished because the character following the type descriptor
1436 will be a digit, @samp{(}, or @samp{-} for a Pascal space type, or
1437 @samp{u} or @samp{s} for a builtin type.
1440 Documented by AIX to define a wide character type, but their compiler
1441 actually uses negative type numbers (@pxref{Negative Type Numbers}).
1443 @item R @var{fp-type} ; @var{bytes} ;
1444 Define a floating point type. @var{fp-type} has one of the following values:
1448 IEEE 32-bit (single precision) floating point format.
1451 IEEE 64-bit (double precision) floating point format.
1453 @item 3 (NF_COMPLEX)
1454 @item 4 (NF_COMPLEX16)
1455 @item 5 (NF_COMPLEX32)
1456 @c "GDB source" really means @file{include/aout/stab_gnu.h}, but trying
1457 @c to put that here got an overfull hbox.
1458 These are for complex numbers. A comment in the GDB source describes
1459 them as Fortran @code{complex}, @code{double complex}, and
1460 @code{complex*16}, respectively, but what does that mean? (i.e., Single
1461 precision? Double precison?).
1463 @item 6 (NF_LDOUBLE)
1464 Long double. This should probably only be used for Sun format
1465 @code{long double}, and new codes should be used for other floating
1466 point formats (@code{NF_DOUBLE} can be used if a @code{long double} is
1467 really just an IEEE double, of course).
1470 @var{bytes} is the number of bytes occupied by the type. This allows a
1471 debugger to perform some operations with the type even if it doesn't
1472 understand @var{fp-type}.
1474 @item g @var{type-information} ; @var{nbits}
1475 Documented by AIX to define a floating type, but their compiler actually
1476 uses negative type numbers (@pxref{Negative Type Numbers}).
1478 @item c @var{type-information} ; @var{nbits}
1479 Documented by AIX to define a complex type, but their compiler actually
1480 uses negative type numbers (@pxref{Negative Type Numbers}).
1483 The C @code{void} type is defined as a signed integral type 0 bits long:
1485 .stabs "void:t19=bs0;0;0",128,0,0,0
1487 The Solaris compiler seems to omit the trailing semicolon in this case.
1488 Getting sloppy in this way is not a swift move because if a type is
1489 embedded in a more complex expression it is necessary to be able to tell
1492 I'm not sure how a boolean type is represented.
1494 @node Negative Type Numbers
1495 @subsection Negative Type Numbers
1497 This is the method used in XCOFF for defining builtin types.
1498 Since the debugger knows about the builtin types anyway, the idea of
1499 negative type numbers is simply to give a special type number which
1500 indicates the builtin type. There is no stab defining these types.
1502 There are several subtle issues with negative type numbers.
1504 One is the size of the type. A builtin type (for example the C types
1505 @code{int} or @code{long}) might have different sizes depending on
1506 compiler options, the target architecture, the ABI, etc. This issue
1507 doesn't come up for IBM tools since (so far) they just target the
1508 RS/6000; the sizes indicated below for each size are what the IBM
1509 RS/6000 tools use. To deal with differing sizes, either define separate
1510 negative type numbers for each size (which works but requires changing
1511 the debugger, and, unless you get both AIX dbx and GDB to accept the
1512 change, introduces an incompatibility), or use a type attribute
1513 (@pxref{String Field}) to define a new type with the appropriate size
1514 (which merely requires a debugger which understands type attributes,
1515 like AIX dbx or GDB). For example,
1518 .stabs "boolean:t10=@@s8;-16",128,0,0,0
1521 defines an 8-bit boolean type, and
1524 .stabs "boolean:t10=@@s64;-16",128,0,0,0
1527 defines a 64-bit boolean type.
1529 A similar issue is the format of the type. This comes up most often for
1530 floating-point types, which could have various formats (particularly
1531 extended doubles, which vary quite a bit even among IEEE systems).
1532 Again, it is best to define a new negative type number for each
1533 different format; changing the format based on the target system has
1534 various problems. One such problem is that the Alpha has both VAX and
1535 IEEE floating types. One can easily imagine one library using the VAX
1536 types and another library in the same executable using the IEEE types.
1537 Another example is that the interpretation of whether a boolean is true
1538 or false can be based on the least significant bit, most significant
1539 bit, whether it is zero, etc., and different compilers (or different
1540 options to the same compiler) might provide different kinds of boolean.
1542 The last major issue is the names of the types. The name of a given
1543 type depends @emph{only} on the negative type number given; these do not
1544 vary depending on the language, the target system, or anything else.
1545 One can always define separate type numbers---in the following list you
1546 will see for example separate @code{int} and @code{integer*4} types
1547 which are identical except for the name. But compatibility can be
1548 maintained by not inventing new negative type numbers and instead just
1549 defining a new type with a new name. For example:
1552 .stabs "CARDINAL:t10=-8",128,0,0,0
1555 Here is the list of negative type numbers. The phrase @dfn{integral
1556 type} is used to mean twos-complement (I strongly suspect that all
1557 machines which use stabs use twos-complement; most machines use
1558 twos-complement these days).
1562 @code{int}, 32 bit signed integral type.
1565 @code{char}, 8 bit type holding a character. Both GDB and dbx on AIX
1566 treat this as signed. GCC uses this type whether @code{char} is signed
1567 or not, which seems like a bad idea. The AIX compiler (@code{xlc}) seems to
1568 avoid this type; it uses -5 instead for @code{char}.
1571 @code{short}, 16 bit signed integral type.
1574 @code{long}, 32 bit signed integral type.
1577 @code{unsigned char}, 8 bit unsigned integral type.
1580 @code{signed char}, 8 bit signed integral type.
1583 @code{unsigned short}, 16 bit unsigned integral type.
1586 @code{unsigned int}, 32 bit unsigned integral type.
1589 @code{unsigned}, 32 bit unsigned integral type.
1592 @code{unsigned long}, 32 bit unsigned integral type.
1595 @code{void}, type indicating the lack of a value.
1598 @code{float}, IEEE single precision.
1601 @code{double}, IEEE double precision.
1604 @code{long double}, IEEE double precision. The compiler claims the size
1605 will increase in a future release, and for binary compatibility you have
1606 to avoid using @code{long double}. I hope when they increase it they
1607 use a new negative type number.
1610 @code{integer}. 32 bit signed integral type.
1613 @code{boolean}. 32 bit type. GDB and GCC assume that zero is false,
1614 one is true, and other values have unspecified meaning. I hope this
1615 agrees with how the IBM tools use the type.
1618 @code{short real}. IEEE single precision.
1621 @code{real}. IEEE double precision.
1624 @code{stringptr}. @xref{Strings}.
1627 @code{character}, 8 bit unsigned character type.
1630 @code{logical*1}, 8 bit type. This Fortran type has a split
1631 personality in that it is used for boolean variables, but can also be
1632 used for unsigned integers. 0 is false, 1 is true, and other values are
1636 @code{logical*2}, 16 bit type. This Fortran type has a split
1637 personality in that it is used for boolean variables, but can also be
1638 used for unsigned integers. 0 is false, 1 is true, and other values are
1642 @code{logical*4}, 32 bit type. This Fortran type has a split
1643 personality in that it is used for boolean variables, but can also be
1644 used for unsigned integers. 0 is false, 1 is true, and other values are
1648 @code{logical}, 32 bit type. This Fortran type has a split
1649 personality in that it is used for boolean variables, but can also be
1650 used for unsigned integers. 0 is false, 1 is true, and other values are
1654 @code{complex}. A complex type consisting of two IEEE single-precision
1655 floating point values.
1658 @code{complex}. A complex type consisting of two IEEE double-precision
1659 floating point values.
1662 @code{integer*1}, 8 bit signed integral type.
1665 @code{integer*2}, 16 bit signed integral type.
1668 @code{integer*4}, 32 bit signed integral type.
1671 @code{wchar}. Wide character, 16 bits wide, unsigned (what format?
1675 @code{long long}, 64 bit signed integral type.
1678 @code{unsigned long long}, 64 bit unsigned integral type.
1681 @code{logical*8}, 64 bit unsigned integral type.
1684 @code{integer*8}, 64 bit signed integral type.
1687 @node Miscellaneous Types
1688 @section Miscellaneous Types
1691 @item b @var{type-information} ; @var{bytes}
1692 Pascal space type. This is documented by IBM; what does it mean?
1694 This use of the @samp{b} type descriptor can be distinguished
1695 from its use for builtin integral types (@pxref{Builtin Type
1696 Descriptors}) because the character following the type descriptor is
1697 always a digit, @samp{(}, or @samp{-}.
1699 @item B @var{type-information}
1700 A volatile-qualified version of @var{type-information}. This is
1701 a Sun extension. References and stores to a variable with a
1702 volatile-qualified type must not be optimized or cached; they
1703 must occur as the user specifies them.
1705 @item d @var{type-information}
1706 File of type @var{type-information}. As far as I know this is only used
1709 @item k @var{type-information}
1710 A const-qualified version of @var{type-information}. This is a Sun
1711 extension. A variable with a const-qualified type cannot be modified.
1713 @item M @var{type-information} ; @var{length}
1714 Multiple instance type. The type seems to composed of @var{length}
1715 repetitions of @var{type-information}, for example @code{character*3} is
1716 represented by @samp{M-2;3}, where @samp{-2} is a reference to a
1717 character type (@pxref{Negative Type Numbers}). I'm not sure how this
1718 differs from an array. This appears to be a Fortran feature.
1719 @var{length} is a bound, like those in range types; see @ref{Subranges}.
1721 @item S @var{type-information}
1722 Pascal set type. @var{type-information} must be a small type such as an
1723 enumeration or a subrange, and the type is a bitmask whose length is
1724 specified by the number of elements in @var{type-information}.
1726 In CHILL, if it is a bitstring instead of a set, also use the @samp{S}
1727 type attribute (@pxref{String Field}).
1729 @item * @var{type-information}
1730 Pointer to @var{type-information}.
1733 @node Cross-References
1734 @section Cross-References to Other Types
1736 A type can be used before it is defined; one common way to deal with
1737 that situation is just to use a type reference to a type which has not
1740 Another way is with the @samp{x} type descriptor, which is followed by
1741 @samp{s} for a structure tag, @samp{u} for a union tag, or @samp{e} for
1742 a enumerator tag, followed by the name of the tag, followed by @samp{:}.
1743 If the name contains @samp{::} between a @samp{<} and @samp{>} pair (for
1744 C++ templates), such a @samp{::} does not end the name---only a single
1745 @samp{:} ends the name; see @ref{Nested Symbols}.
1747 For example, the following C declarations:
1758 .stabs "bar:G16=*17=xsfoo:",32,0,0,0
1761 Not all debuggers support the @samp{x} type descriptor, so on some
1762 machines GCC does not use it. I believe that for the above example it
1763 would just emit a reference to type 17 and never define it, but I
1764 haven't verified that.
1766 Modula-2 imported types, at least on AIX, use the @samp{i} type
1767 descriptor, which is followed by the name of the module from which the
1768 type is imported, followed by @samp{:}, followed by the name of the
1769 type. There is then optionally a comma followed by type information for
1770 the type. This differs from merely naming the type (@pxref{Typedefs}) in
1771 that it identifies the module; I don't understand whether the name of
1772 the type given here is always just the same as the name we are giving
1773 it, or whether this type descriptor is used with a nameless stab
1774 (@pxref{String Field}), or what. The symbol ends with @samp{;}.
1777 @section Subrange Types
1779 The @samp{r} type descriptor defines a type as a subrange of another
1780 type. It is followed by type information for the type of which it is a
1781 subrange, a semicolon, an integral lower bound, a semicolon, an
1782 integral upper bound, and a semicolon. The AIX documentation does not
1783 specify the trailing semicolon, in an effort to specify array indexes
1784 more cleanly, but a subrange which is not an array index has always
1785 included a trailing semicolon (@pxref{Arrays}).
1787 Instead of an integer, either bound can be one of the following:
1790 @item A @var{offset}
1791 The bound is passed by reference on the stack at offset @var{offset}
1792 from the argument list. @xref{Parameters}, for more information on such
1795 @item T @var{offset}
1796 The bound is passed by value on the stack at offset @var{offset} from
1799 @item a @var{register-number}
1800 The bound is pased by reference in register number
1801 @var{register-number}.
1803 @item t @var{register-number}
1804 The bound is passed by value in register number @var{register-number}.
1810 Subranges are also used for builtin types; see @ref{Traditional Builtin Types}.
1813 @section Array Types
1815 Arrays use the @samp{a} type descriptor. Following the type descriptor
1816 is the type of the index and the type of the array elements. If the
1817 index type is a range type, it ends in a semicolon; otherwise
1818 (for example, if it is a type reference), there does not
1819 appear to be any way to tell where the types are separated. In an
1820 effort to clean up this mess, IBM documents the two types as being
1821 separated by a semicolon, and a range type as not ending in a semicolon
1822 (but this is not right for range types which are not array indexes,
1823 @pxref{Subranges}). I think probably the best solution is to specify
1824 that a semicolon ends a range type, and that the index type and element
1825 type of an array are separated by a semicolon, but that if the index
1826 type is a range type, the extra semicolon can be omitted. GDB (at least
1827 through version 4.9) doesn't support any kind of index type other than a
1828 range anyway; I'm not sure about dbx.
1830 It is well established, and widely used, that the type of the index,
1831 unlike most types found in the stabs, is merely a type definition, not
1832 type information (@pxref{String Field}) (that is, it need not start with
1833 @samp{@var{type-number}=} if it is defining a new type). According to a
1834 comment in GDB, this is also true of the type of the array elements; it
1835 gives @samp{ar1;1;10;ar1;1;10;4} as a legitimate way to express a two
1836 dimensional array. According to AIX documentation, the element type
1837 must be type information. GDB accepts either.
1839 The type of the index is often a range type, expressed as the type
1840 descriptor @samp{r} and some parameters. It defines the size of the
1841 array. In the example below, the range @samp{r1;0;2;} defines an index
1842 type which is a subrange of type 1 (integer), with a lower bound of 0
1843 and an upper bound of 2. This defines the valid range of subscripts of
1844 a three-element C array.
1846 For example, the definition:
1849 char char_vec[3] = @{'a','b','c'@};
1853 produces the output:
1856 .stabs "char_vec:G19=ar1;0;2;2",32,0,0,0
1865 If an array is @dfn{packed}, the elements are spaced more
1866 closely than normal, saving memory at the expense of speed. For
1867 example, an array of 3-byte objects might, if unpacked, have each
1868 element aligned on a 4-byte boundary, but if packed, have no padding.
1869 One way to specify that something is packed is with type attributes
1870 (@pxref{String Field}). In the case of arrays, another is to use the
1871 @samp{P} type descriptor instead of @samp{a}. Other than specifying a
1872 packed array, @samp{P} is identical to @samp{a}.
1874 @c FIXME-what is it? A pointer?
1875 An open array is represented by the @samp{A} type descriptor followed by
1876 type information specifying the type of the array elements.
1878 @c FIXME: what is the format of this type? A pointer to a vector of pointers?
1879 An N-dimensional dynamic array is represented by
1882 D @var{dimensions} ; @var{type-information}
1885 @c Does dimensions really have this meaning? The AIX documentation
1887 @var{dimensions} is the number of dimensions; @var{type-information}
1888 specifies the type of the array elements.
1890 @c FIXME: what is the format of this type? A pointer to some offsets in
1892 A subarray of an N-dimensional array is represented by
1895 E @var{dimensions} ; @var{type-information}
1898 @c Does dimensions really have this meaning? The AIX documentation
1900 @var{dimensions} is the number of dimensions; @var{type-information}
1901 specifies the type of the array elements.
1906 Some languages, like C or the original Pascal, do not have string types,
1907 they just have related things like arrays of characters. But most
1908 Pascals and various other languages have string types, which are
1909 indicated as follows:
1912 @item n @var{type-information} ; @var{bytes}
1913 @var{bytes} is the maximum length. I'm not sure what
1914 @var{type-information} is; I suspect that it means that this is a string
1915 of @var{type-information} (thus allowing a string of integers, a string
1916 of wide characters, etc., as well as a string of characters). Not sure
1917 what the format of this type is. This is an AIX feature.
1919 @item z @var{type-information} ; @var{bytes}
1920 Just like @samp{n} except that this is a gstring, not an ordinary
1921 string. I don't know the difference.
1924 Pascal Stringptr. What is this? This is an AIX feature.
1927 Languages, such as CHILL which have a string type which is basically
1928 just an array of characters use the @samp{S} type attribute
1929 (@pxref{String Field}).
1932 @section Enumerations
1934 Enumerations are defined with the @samp{e} type descriptor.
1936 @c FIXME: Where does this information properly go? Perhaps it is
1937 @c redundant with something we already explain.
1938 The source line below declares an enumeration type at file scope.
1939 The type definition is located after the @code{N_RBRAC} that marks the end of
1940 the previous procedure's block scope, and before the @code{N_FUN} that marks
1941 the beginning of the next procedure's block scope. Therefore it does not
1942 describe a block local symbol, but a file local one.
1947 enum e_places @{first,second=3,last@};
1951 generates the following stab:
1954 .stabs "e_places:T22=efirst:0,second:3,last:4,;",128,0,0,0
1957 The symbol descriptor (@samp{T}) says that the stab describes a
1958 structure, enumeration, or union tag. The type descriptor @samp{e},
1959 following the @samp{22=} of the type definition narrows it down to an
1960 enumeration type. Following the @samp{e} is a list of the elements of
1961 the enumeration. The format is @samp{@var{name}:@var{value},}. The
1962 list of elements ends with @samp{;}. The fact that @var{value} is
1963 specified as an integer can cause problems if the value is large. GCC
1964 2.5.2 tries to output it in octal in that case with a leading zero,
1965 which is probably a good thing, although GDB 4.11 supports octal only in
1966 cases where decimal is perfectly good. Negative decimal values are
1967 supported by both GDB and dbx.
1969 There is no standard way to specify the size of an enumeration type; it
1970 is determined by the architecture (normally all enumerations types are
1971 32 bits). Type attributes can be used to specify an enumeration type of
1972 another size for debuggers which support them; see @ref{String Field}.
1974 Enumeration types are unusual in that they define symbols for the
1975 enumeration values (@code{first}, @code{second}, and @code{third} in the
1976 above example), and even though these symbols are visible in the file as
1977 a whole (rather than being in a more local namespace like structure
1978 member names), they are defined in the type definition for the
1979 enumeration type rather than each having their own symbol. In order to
1980 be fast, GDB will only get symbols from such types (in its initial scan
1981 of the stabs) if the type is the first thing defined after a @samp{T} or
1982 @samp{t} symbol descriptor (the above example fulfills this
1983 requirement). If the type does not have a name, the compiler should
1984 emit it in a nameless stab (@pxref{String Field}); GCC does this.
1989 The encoding of structures in stabs can be shown with an example.
1991 The following source code declares a structure tag and defines an
1992 instance of the structure in global scope. Then a @code{typedef} equates the
1993 structure tag with a new type. Seperate stabs are generated for the
1994 structure tag, the structure @code{typedef}, and the structure instance. The
1995 stabs for the tag and the @code{typedef} are emited when the definitions are
1996 encountered. Since the structure elements are not initialized, the
1997 stab and code for the structure variable itself is located at the end
1998 of the program in the bss section.
2005 struct s_tag* s_next;
2008 typedef struct s_tag s_typedef;
2011 The structure tag has an @code{N_LSYM} stab type because, like the
2012 enumeration, the symbol has file scope. Like the enumeration, the
2013 symbol descriptor is @samp{T}, for enumeration, structure, or tag type.
2014 The type descriptor @samp{s} following the @samp{16=} of the type
2015 definition narrows the symbol type to structure.
2017 Following the @samp{s} type descriptor is the number of bytes the
2018 structure occupies, followed by a description of each structure element.
2019 The structure element descriptions are of the form @var{name:type, bit
2020 offset from the start of the struct, number of bits in the element}.
2022 @c FIXME: phony line break. Can probably be fixed by using an example
2023 @c with fewer fields.
2026 .stabs "s_tag:T16=s20s_int:1,0,32;s_float:12,32,32;
2027 s_char_vec:17=ar1;0;7;2,64,64;s_next:18=*16,128,32;;",128,0,0,0
2030 In this example, the first two structure elements are previously defined
2031 types. For these, the type following the @samp{@var{name}:} part of the
2032 element description is a simple type reference. The other two structure
2033 elements are new types. In this case there is a type definition
2034 embedded after the @samp{@var{name}:}. The type definition for the
2035 array element looks just like a type definition for a standalone array.
2036 The @code{s_next} field is a pointer to the same kind of structure that
2037 the field is an element of. So the definition of structure type 16
2038 contains a type definition for an element which is a pointer to type 16.
2040 If a field is a static member (this is a C++ feature in which a single
2041 variable appears to be a field of every structure of a given type) it
2042 still starts out with the field name, a colon, and the type, but then
2043 instead of a comma, bit position, comma, and bit size, there is a colon
2044 followed by the name of the variable which each such field refers to.
2046 If the structure has methods (a C++ feature), they follow the non-method
2047 fields; see @ref{Cplusplus}.
2050 @section Giving a Type a Name
2052 @findex N_LSYM, for types
2053 @findex C_DECL, for types
2054 To give a type a name, use the @samp{t} symbol descriptor. The type
2055 is specified by the type information (@pxref{String Field}) for the stab.
2059 .stabs "s_typedef:t16",128,0,0,0 # @r{128 is N_LSYM}
2062 specifies that @code{s_typedef} refers to type number 16. Such stabs
2063 have symbol type @code{N_LSYM} (or @code{C_DECL} for XCOFF).
2065 If you are specifying the tag name for a structure, union, or
2066 enumeration, use the @samp{T} symbol descriptor instead. I believe C is
2067 the only language with this feature.
2069 If the type is an opaque type (I believe this is a Modula-2 feature),
2070 AIX provides a type descriptor to specify it. The type descriptor is
2071 @samp{o} and is followed by a name. I don't know what the name
2072 means---is it always the same as the name of the type, or is this type
2073 descriptor used with a nameless stab (@pxref{String Field})? There
2074 optionally follows a comma followed by type information which defines
2075 the type of this type. If omitted, a semicolon is used in place of the
2076 comma and the type information, and the type is much like a generic
2077 pointer type---it has a known size but little else about it is
2091 This code generates a stab for a union tag and a stab for a union
2092 variable. Both use the @code{N_LSYM} stab type. If a union variable is
2093 scoped locally to the procedure in which it is defined, its stab is
2094 located immediately preceding the @code{N_LBRAC} for the procedure's block
2097 The stab for the union tag, however, is located preceding the code for
2098 the procedure in which it is defined. The stab type is @code{N_LSYM}. This
2099 would seem to imply that the union type is file scope, like the struct
2100 type @code{s_tag}. This is not true. The contents and position of the stab
2101 for @code{u_type} do not convey any infomation about its procedure local
2104 @c FIXME: phony line break. Can probably be fixed by using an example
2105 @c with fewer fields.
2108 .stabs "u_tag:T23=u4u_int:1,0,32;u_float:12,0,32;u_char:21,0,32;;",
2112 The symbol descriptor @samp{T}, following the @samp{name:} means that
2113 the stab describes an enumeration, structure, or union tag. The type
2114 descriptor @samp{u}, following the @samp{23=} of the type definition,
2115 narrows it down to a union type definition. Following the @samp{u} is
2116 the number of bytes in the union. After that is a list of union element
2117 descriptions. Their format is @var{name:type, bit offset into the
2118 union, number of bytes for the element;}.
2120 The stab for the union variable is:
2123 .stabs "an_u:23",128,0,0,-20 # @r{128 is N_LSYM}
2126 @samp{-20} specifies where the variable is stored (@pxref{Stack
2129 @node Function Types
2130 @section Function Types
2132 Various types can be defined for function variables. These types are
2133 not used in defining functions (@pxref{Procedures}); they are used for
2134 things like pointers to functions.
2136 The simple, traditional, type is type descriptor @samp{f} is followed by
2137 type information for the return type of the function, followed by a
2140 This does not deal with functions for which the number and types of the
2141 parameters are part of the type, as in Modula-2 or ANSI C. AIX provides
2142 extensions to specify these, using the @samp{f}, @samp{F}, @samp{p}, and
2143 @samp{R} type descriptors.
2145 First comes the type descriptor. If it is @samp{f} or @samp{F}, this
2146 type involves a function rather than a procedure, and the type
2147 information for the return type of the function follows, followed by a
2148 comma. Then comes the number of parameters to the function and a
2149 semicolon. Then, for each parameter, there is the name of the parameter
2150 followed by a colon (this is only present for type descriptors @samp{R}
2151 and @samp{F} which represent Pascal function or procedure parameters),
2152 type information for the parameter, a comma, 0 if passed by reference or
2153 1 if passed by value, and a semicolon. The type definition ends with a
2156 For example, this variable definition:
2163 generates the following code:
2166 .stabs "g_pf:G24=*25=f1",32,0,0,0
2167 .common _g_pf,4,"bss"
2170 The variable defines a new type, 24, which is a pointer to another new
2171 type, 25, which is a function returning @code{int}.
2174 @chapter Symbol Information in Symbol Tables
2176 This chapter describes the format of symbol table entries
2177 and how stab assembler directives map to them. It also describes the
2178 transformations that the assembler and linker make on data from stabs.
2181 * Symbol Table Format::
2182 * Transformations On Symbol Tables::
2185 @node Symbol Table Format
2186 @section Symbol Table Format
2188 Each time the assembler encounters a stab directive, it puts
2189 each field of the stab into a corresponding field in a symbol table
2190 entry of its output file. If the stab contains a string field, the
2191 symbol table entry for that stab points to a string table entry
2192 containing the string data from the stab. Assembler labels become
2193 relocatable addresses. Symbol table entries in a.out have the format:
2195 @c FIXME: should refer to external, not internal.
2197 struct internal_nlist @{
2198 unsigned long n_strx; /* index into string table of name */
2199 unsigned char n_type; /* type of symbol */
2200 unsigned char n_other; /* misc info (usually empty) */
2201 unsigned short n_desc; /* description field */
2202 bfd_vma n_value; /* value of symbol */
2206 If the stab has a string, the @code{n_strx} field holds the offset in
2207 bytes of the string within the string table. The string is terminated
2208 by a NUL character. If the stab lacks a string (for example, it was
2209 produced by a @code{.stabn} or @code{.stabd} directive), the
2210 @code{n_strx} field is zero.
2212 Symbol table entries with @code{n_type} field values greater than 0x1f
2213 originated as stabs generated by the compiler (with one random
2214 exception). The other entries were placed in the symbol table of the
2215 executable by the assembler or the linker.
2217 @node Transformations On Symbol Tables
2218 @section Transformations on Symbol Tables
2220 The linker concatenates object files and does fixups of externally
2223 You can see the transformations made on stab data by the assembler and
2224 linker by examining the symbol table after each pass of the build. To
2225 do this, use @samp{nm -ap}, which dumps the symbol table, including
2226 debugging information, unsorted. For stab entries the columns are:
2227 @var{value}, @var{other}, @var{desc}, @var{type}, @var{string}. For
2228 assembler and linker symbols, the columns are: @var{value}, @var{type},
2231 The low 5 bits of the stab type tell the linker how to relocate the
2232 value of the stab. Thus for stab types like @code{N_RSYM} and
2233 @code{N_LSYM}, where the value is an offset or a register number, the
2234 low 5 bits are @code{N_ABS}, which tells the linker not to relocate the
2237 Where the value of a stab contains an assembly language label,
2238 it is transformed by each build step. The assembler turns it into a
2239 relocatable address and the linker turns it into an absolute address.
2242 * Transformations On Static Variables::
2243 * Transformations On Global Variables::
2244 * Stab Section Transformations:: For some object file formats,
2245 things are a bit different.
2248 @node Transformations On Static Variables
2249 @subsection Transformations on Static Variables
2251 This source line defines a static variable at file scope:
2254 static int s_g_repeat
2258 The following stab describes the symbol:
2261 .stabs "s_g_repeat:S1",38,0,0,_s_g_repeat
2265 The assembler transforms the stab into this symbol table entry in the
2266 @file{.o} file. The location is expressed as a data segment offset.
2269 00000084 - 00 0000 STSYM s_g_repeat:S1
2273 In the symbol table entry from the executable, the linker has made the
2274 relocatable address absolute.
2277 0000e00c - 00 0000 STSYM s_g_repeat:S1
2280 @node Transformations On Global Variables
2281 @subsection Transformations on Global Variables
2283 Stabs for global variables do not contain location information. In
2284 this case, the debugger finds location information in the assembler or
2285 linker symbol table entry describing the variable. The source line:
2295 .stabs "g_foo:G2",32,0,0,0
2298 The variable is represented by two symbol table entries in the object
2299 file (see below). The first one originated as a stab. The second one
2300 is an external symbol. The upper case @samp{D} signifies that the
2301 @code{n_type} field of the symbol table contains 7, @code{N_DATA} with
2302 local linkage. The stab's value is zero since the value is not used for
2303 @code{N_GSYM} stabs. The value of the linker symbol is the relocatable
2304 address corresponding to the variable.
2307 00000000 - 00 0000 GSYM g_foo:G2
2312 These entries as transformed by the linker. The linker symbol table
2313 entry now holds an absolute address:
2316 00000000 - 00 0000 GSYM g_foo:G2
2321 @node Stab Section Transformations
2322 @subsection Transformations of Stabs in separate sections
2324 For object file formats using stabs in separate sections (@pxref{Stab
2325 Sections}), use @code{objdump --stabs} instead of @code{nm} to show the
2326 stabs in an object or executable file. @code{objdump} is a GNU utility;
2327 Sun does not provide any equivalent.
2329 The following example is for a stab whose value is an address is
2330 relative to the compilation unit (@pxref{ELF Linker Relocation}). For
2331 example, if the source line
2337 appears within a function, then the assembly language output from the
2343 .stabs "ld:V(0,3)",0x26,0,4,.L18-Ddata.data # @r{0x26 is N_STSYM}
2350 Because the value is formed by subtracting one symbol from another, the
2351 value is absolute, not relocatable, and so the object file contains
2354 Symnum n_type n_othr n_desc n_value n_strx String
2355 31 STSYM 0 4 00000004 680 ld:V(0,3)
2358 without any relocations, and the executable file also contains
2361 Symnum n_type n_othr n_desc n_value n_strx String
2362 31 STSYM 0 4 00000004 680 ld:V(0,3)
2366 @chapter GNU C++ Stabs
2369 * Class Names:: C++ class names are both tags and typedefs.
2370 * Nested Symbols:: C++ symbol names can be within other types.
2371 * Basic Cplusplus Types::
2374 * Methods:: Method definition
2375 * Method Type Descriptor:: The @samp{#} type descriptor
2376 * Member Type Descriptor:: The @samp{@@} type descriptor
2378 * Method Modifiers::
2381 * Virtual Base Classes::
2386 @section C++ Class Names
2388 In C++, a class name which is declared with @code{class}, @code{struct},
2389 or @code{union}, is not only a tag, as in C, but also a type name. Thus
2390 there should be stabs with both @samp{t} and @samp{T} symbol descriptors
2393 To save space, there is a special abbreviation for this case. If the
2394 @samp{T} symbol descriptor is followed by @samp{t}, then the stab
2395 defines both a type name and a tag.
2397 For example, the C++ code
2400 struct foo @{int x;@};
2403 can be represented as either
2406 .stabs "foo:T19=s4x:1,0,32;;",128,0,0,0 # @r{128 is N_LSYM}
2407 .stabs "foo:t19",128,0,0,0
2413 .stabs "foo:Tt19=s4x:1,0,32;;",128,0,0,0
2416 @node Nested Symbols
2417 @section Defining a Symbol Within Another Type
2419 In C++, a symbol (such as a type name) can be defined within another type.
2420 @c FIXME: Needs example.
2422 In stabs, this is sometimes represented by making the name of a symbol
2423 which contains @samp{::}. Such a pair of colons does not end the name
2424 of the symbol, the way a single colon would (@pxref{String Field}). I'm
2425 not sure how consistently used or well thought out this mechanism is.
2426 So that a pair of colons in this position always has this meaning,
2427 @samp{:} cannot be used as a symbol descriptor.
2429 For example, if the string for a stab is @samp{foo::bar::baz:t5=*6},
2430 then @code{foo::bar::baz} is the name of the symbol, @samp{t} is the
2431 symbol descriptor, and @samp{5=*6} is the type information.
2433 @node Basic Cplusplus Types
2434 @section Basic Types For C++
2436 << the examples that follow are based on a01.C >>
2439 C++ adds two more builtin types to the set defined for C. These are
2440 the unknown type and the vtable record type. The unknown type, type
2441 16, is defined in terms of itself like the void type.
2443 The vtable record type, type 17, is defined as a structure type and
2444 then as a structure tag. The structure has four fields: delta, index,
2445 pfn, and delta2. pfn is the function pointer.
2447 << In boilerplate $vtbl_ptr_type, what are the fields delta,
2448 index, and delta2 used for? >>
2450 This basic type is present in all C++ programs even if there are no
2451 virtual methods defined.
2454 .stabs "struct_name:sym_desc(type)type_def(17)=type_desc(struct)struct_bytes(8)
2455 elem_name(delta):type_ref(short int),bit_offset(0),field_bits(16);
2456 elem_name(index):type_ref(short int),bit_offset(16),field_bits(16);
2457 elem_name(pfn):type_def(18)=type_desc(ptr to)type_ref(void),
2458 bit_offset(32),field_bits(32);
2459 elem_name(delta2):type_def(short int);bit_offset(32),field_bits(16);;"
2464 .stabs "$vtbl_ptr_type:t17=s8
2465 delta:6,0,16;index:6,16,16;pfn:18=*15,32,32;delta2:6,32,16;;"
2470 .stabs "name:sym_dec(struct tag)type_ref($vtbl_ptr_type)",N_LSYM,NIL,NIL,NIL
2474 .stabs "$vtbl_ptr_type:T17",128,0,0,0
2477 @node Simple Classes
2478 @section Simple Class Definition
2480 The stabs describing C++ language features are an extension of the
2481 stabs describing C. Stabs representing C++ class types elaborate
2482 extensively on the stab format used to describe structure types in C.
2483 Stabs representing class type variables look just like stabs
2484 representing C language variables.
2486 Consider the following very simple class definition.
2492 int Ameth(int in, char other);
2496 The class @code{baseA} is represented by two stabs. The first stab describes
2497 the class as a structure type. The second stab describes a structure
2498 tag of the class type. Both stabs are of stab type @code{N_LSYM}. Since the
2499 stab is not located between an @code{N_FUN} and an @code{N_LBRAC} stab this indicates
2500 that the class is defined at file scope. If it were, then the @code{N_LSYM}
2501 would signify a local variable.
2503 A stab describing a C++ class type is similar in format to a stab
2504 describing a C struct, with each class member shown as a field in the
2505 structure. The part of the struct format describing fields is
2506 expanded to include extra information relevent to C++ class members.
2507 In addition, if the class has multiple base classes or virtual
2508 functions the struct format outside of the field parts is also
2511 In this simple example the field part of the C++ class stab
2512 representing member data looks just like the field part of a C struct
2513 stab. The section on protections describes how its format is
2514 sometimes extended for member data.
2516 The field part of a C++ class stab representing a member function
2517 differs substantially from the field part of a C struct stab. It
2518 still begins with @samp{name:} but then goes on to define a new type number
2519 for the member function, describe its return type, its argument types,
2520 its protection level, any qualifiers applied to the method definition,
2521 and whether the method is virtual or not. If the method is virtual
2522 then the method description goes on to give the vtable index of the
2523 method, and the type number of the first base class defining the
2526 When the field name is a method name it is followed by two colons rather
2527 than one. This is followed by a new type definition for the method.
2528 This is a number followed by an equal sign and the type of the method.
2529 Normally this will be a type declared using the @samp{#} type
2530 descriptor; see @ref{Method Type Descriptor}; static member functions
2531 are declared using the @samp{f} type descriptor instead; see
2532 @ref{Function Types}.
2534 The format of an overloaded operator method name differs from that of
2535 other methods. It is @samp{op$::@var{operator-name}.} where
2536 @var{operator-name} is the operator name such as @samp{+} or @samp{+=}.
2537 The name ends with a period, and any characters except the period can
2538 occur in the @var{operator-name} string.
2540 The next part of the method description represents the arguments to the
2541 method, preceeded by a colon and ending with a semi-colon. The types of
2542 the arguments are expressed in the same way argument types are expressed
2543 in C++ name mangling. In this example an @code{int} and a @code{char}
2546 This is followed by a number, a letter, and an asterisk or period,
2547 followed by another semicolon. The number indicates the protections
2548 that apply to the member function. Here the 2 means public. The
2549 letter encodes any qualifier applied to the method definition. In
2550 this case, @samp{A} means that it is a normal function definition. The dot
2551 shows that the method is not virtual. The sections that follow
2552 elaborate further on these fields and describe the additional
2553 information present for virtual methods.
2557 .stabs "class_name:sym_desc(type)type_def(20)=type_desc(struct)struct_bytes(4)
2558 field_name(Adat):type(int),bit_offset(0),field_bits(32);
2560 method_name(Ameth)::type_def(21)=type_desc(method)return_type(int);
2561 :arg_types(int char);
2562 protection(public)qualifier(normal)virtual(no);;"
2567 .stabs "baseA:t20=s4Adat:1,0,32;Ameth::21=##1;:ic;2A.;;",128,0,0,0
2569 .stabs "class_name:sym_desc(struct tag)",N_LSYM,NIL,NIL,NIL
2571 .stabs "baseA:T20",128,0,0,0
2574 @node Class Instance
2575 @section Class Instance
2577 As shown above, describing even a simple C++ class definition is
2578 accomplished by massively extending the stab format used in C to
2579 describe structure types. However, once the class is defined, C stabs
2580 with no modifications can be used to describe class instances. The
2590 yields the following stab describing the class instance. It looks no
2591 different from a standard C stab describing a local variable.
2594 .stabs "name:type_ref(baseA)", N_LSYM, NIL, NIL, frame_ptr_offset
2598 .stabs "AbaseA:20",128,0,0,-20
2602 @section Method Definition
2604 The class definition shown above declares Ameth. The C++ source below
2609 baseA::Ameth(int in, char other)
2616 This method definition yields three stabs following the code of the
2617 method. One stab describes the method itself and following two describe
2618 its parameters. Although there is only one formal argument all methods
2619 have an implicit argument which is the @code{this} pointer. The @code{this}
2620 pointer is a pointer to the object on which the method was called. Note
2621 that the method name is mangled to encode the class name and argument
2622 types. Name mangling is described in the @sc{arm} (@cite{The Annotated
2623 C++ Reference Manual}, by Ellis and Stroustrup, @sc{isbn}
2624 0-201-51459-1); @file{gpcompare.texi} in Cygnus GCC distributions
2625 describes the differences between GNU mangling and @sc{arm}
2627 @c FIXME: Use @xref, especially if this is generally installed in the
2629 @c FIXME: This information should be in a net release, either of GCC or
2630 @c GDB. But gpcompare.texi doesn't seem to be in the FSF GCC.
2633 .stabs "name:symbol_desriptor(global function)return_type(int)",
2634 N_FUN, NIL, NIL, code_addr_of_method_start
2636 .stabs "Ameth__5baseAic:F1",36,0,0,_Ameth__5baseAic
2639 Here is the stab for the @code{this} pointer implicit argument. The
2640 name of the @code{this} pointer is always @code{this}. Type 19, the
2641 @code{this} pointer is defined as a pointer to type 20, @code{baseA},
2642 but a stab defining @code{baseA} has not yet been emited. Since the
2643 compiler knows it will be emited shortly, here it just outputs a cross
2644 reference to the undefined symbol, by prefixing the symbol name with
2648 .stabs "name:sym_desc(register param)type_def(19)=
2649 type_desc(ptr to)type_ref(baseA)=
2650 type_desc(cross-reference to)baseA:",N_RSYM,NIL,NIL,register_number
2652 .stabs "this:P19=*20=xsbaseA:",64,0,0,8
2655 The stab for the explicit integer argument looks just like a parameter
2656 to a C function. The last field of the stab is the offset from the
2657 argument pointer, which in most systems is the same as the frame
2661 .stabs "name:sym_desc(value parameter)type_ref(int)",
2662 N_PSYM,NIL,NIL,offset_from_arg_ptr
2664 .stabs "in:p1",160,0,0,72
2667 << The examples that follow are based on A1.C >>
2669 @node Method Type Descriptor
2670 @section The @samp{#} Type Descriptor
2672 This is like the @samp{f} type descriptor for functions (@pxref{Function
2673 Types}), except that a function which uses the @samp{#} type descriptor
2674 takes an extra argument as its first argument, for the @code{this}
2675 pointer. The @samp{#} type descriptor is optionally followed by the
2676 types of the arguments, then another @samp{#}. If the types of the
2677 arguments are omitted, so that the second @samp{#} immediately follows
2678 the @samp{#} which is the type descriptor, the arguments are being
2679 omitted (to save space) and can be deduced from the mangled name of the
2680 method. After the second @samp{#} there is type information for the
2681 return type of the method and a semicolon.
2683 Note that although such a type will normally be used to describe fields
2684 in structures, unions, or classes, for at least some versions of the
2685 compiler it can also be used in other contexts.
2687 @node Member Type Descriptor
2688 @section The @samp{@@} Type Descriptor
2690 The @samp{@@} type descriptor is for a member (class and variable) type.
2691 It is followed by type information for the offset basetype, a comma, and
2692 type information for the type of the field being pointed to. (FIXME:
2693 this is acknowledged to be gibberish. Can anyone say what really goes
2696 Note that there is a conflict between this and type attributes
2697 (@pxref{String Field}); both use type descriptor @samp{@@}.
2698 Fortunately, the @samp{@@} type descriptor used in this C++ sense always
2699 will be followed by a digit, @samp{(}, or @samp{-}, and type attributes
2700 never start with those things.
2703 @section Protections
2705 In the simple class definition shown above all member data and
2706 functions were publicly accessable. The example that follows
2707 contrasts public, protected and privately accessable fields and shows
2708 how these protections are encoded in C++ stabs.
2710 If the character following the @samp{@var{field-name}:} part of the
2711 string is @samp{/}, then the next character is the visibility. @samp{0}
2712 means private, @samp{1} means protected, and @samp{2} means public.
2713 Debuggers should ignore visibility characters they do not recognize, and
2714 assume a reasonable default (such as public) (GDB 4.11 does not, but
2715 this should be fixed in the next GDB release). If no visibility is
2716 specified the field is public. The visibility @samp{9} means that the
2717 field has been optimized out and is public (there is no way to specify
2718 an optimized out field with a private or protected visibility).
2719 Visibility @samp{9} is not supported by GDB 4.11; this should be fixed
2720 in the next GDB release.
2722 The following C++ source:
2736 generates the following stab:
2740 .stabs "vis:T19=s12priv:/01,0,32;prot:/12,32,8;pub:12,64,32;;",128,0,0,0
2743 @samp{vis:T19=s12} indicates that type number 19 is a 12 byte structure
2744 named @code{vis} The @code{priv} field has public visibility
2745 (@samp{/0}), type int (@samp{1}), and offset and size @samp{,0,32;}.
2746 The @code{prot} field has protected visibility (@samp{/1}), type char
2747 (@samp{2}) and offset and size @samp{,32,8;}. The @code{pub} field has
2748 type float (@samp{12}), and offset and size @samp{,64,32;}.
2750 Protections for member functions are signified by one digit embeded in
2751 the field part of the stab describing the method. The digit is 0 if
2752 private, 1 if protected and 2 if public. Consider the C++ class
2756 class all_methods @{
2758 int priv_meth(int in)@{return in;@};
2760 char protMeth(char in)@{return in;@};
2762 float pubMeth(float in)@{return in;@};
2766 It generates the following stab. The digit in question is to the left
2767 of an @samp{A} in each case. Notice also that in this case two symbol
2768 descriptors apply to the class name struct tag and struct type.
2771 .stabs "class_name:sym_desc(struct tag&type)type_def(21)=
2772 sym_desc(struct)struct_bytes(1)
2773 meth_name::type_def(22)=sym_desc(method)returning(int);
2774 :args(int);protection(private)modifier(normal)virtual(no);
2775 meth_name::type_def(23)=sym_desc(method)returning(char);
2776 :args(char);protection(protected)modifier(normal)virual(no);
2777 meth_name::type_def(24)=sym_desc(method)returning(float);
2778 :args(float);protection(public)modifier(normal)virtual(no);;",
2783 .stabs "all_methods:Tt21=s1priv_meth::22=##1;:i;0A.;protMeth::23=##2;:c;1A.;
2784 pubMeth::24=##12;:f;2A.;;",128,0,0,0
2787 @node Method Modifiers
2788 @section Method Modifiers (@code{const}, @code{volatile}, @code{const volatile})
2792 In the class example described above all the methods have the normal
2793 modifier. This method modifier information is located just after the
2794 protection information for the method. This field has four possible
2795 character values. Normal methods use @samp{A}, const methods use
2796 @samp{B}, volatile methods use @samp{C}, and const volatile methods use
2797 @samp{D}. Consider the class definition below:
2802 int ConstMeth (int arg) const @{ return arg; @};
2803 char VolatileMeth (char arg) volatile @{ return arg; @};
2804 float ConstVolMeth (float arg) const volatile @{return arg; @};
2808 This class is described by the following stab:
2811 .stabs "class(A):sym_desc(struct)type_def(20)=type_desc(struct)struct_bytes(1)
2812 meth_name(ConstMeth)::type_def(21)sym_desc(method)
2813 returning(int);:arg(int);protection(public)modifier(const)virtual(no);
2814 meth_name(VolatileMeth)::type_def(22)=sym_desc(method)
2815 returning(char);:arg(char);protection(public)modifier(volatile)virt(no)
2816 meth_name(ConstVolMeth)::type_def(23)=sym_desc(method)
2817 returning(float);:arg(float);protection(public)modifer(const volatile)
2818 virtual(no);;", @dots{}
2822 .stabs "A:T20=s1ConstMeth::21=##1;:i;2B.;VolatileMeth::22=##2;:c;2C.;
2823 ConstVolMeth::23=##12;:f;2D.;;",128,0,0,0
2826 @node Virtual Methods
2827 @section Virtual Methods
2829 << The following examples are based on a4.C >>
2831 The presence of virtual methods in a class definition adds additional
2832 data to the class description. The extra data is appended to the
2833 description of the virtual method and to the end of the class
2834 description. Consider the class definition below:
2840 virtual int A_virt (int arg) @{ return arg; @};
2844 This results in the stab below describing class A. It defines a new
2845 type (20) which is an 8 byte structure. The first field of the class
2846 struct is @samp{Adat}, an integer, starting at structure offset 0 and
2849 The second field in the class struct is not explicitly defined by the
2850 C++ class definition but is implied by the fact that the class
2851 contains a virtual method. This field is the vtable pointer. The
2852 name of the vtable pointer field starts with @samp{$vf} and continues with a
2853 type reference to the class it is part of. In this example the type
2854 reference for class A is 20 so the name of its vtable pointer field is
2855 @samp{$vf20}, followed by the usual colon.
2857 Next there is a type definition for the vtable pointer type (21).
2858 This is in turn defined as a pointer to another new type (22).
2860 Type 22 is the vtable itself, which is defined as an array, indexed by
2861 a range of integers between 0 and 1, and whose elements are of type
2862 17. Type 17 was the vtable record type defined by the boilerplate C++
2863 type definitions, as shown earlier.
2865 The bit offset of the vtable pointer field is 32. The number of bits
2866 in the field are not specified when the field is a vtable pointer.
2868 Next is the method definition for the virtual member function @code{A_virt}.
2869 Its description starts out using the same format as the non-virtual
2870 member functions described above, except instead of a dot after the
2871 @samp{A} there is an asterisk, indicating that the function is virtual.
2872 Since is is virtual some addition information is appended to the end
2873 of the method description.
2875 The first number represents the vtable index of the method. This is a
2876 32 bit unsigned number with the high bit set, followed by a
2879 The second number is a type reference to the first base class in the
2880 inheritence hierarchy defining the virtual member function. In this
2881 case the class stab describes a base class so the virtual function is
2882 not overriding any other definition of the method. Therefore the
2883 reference is to the type number of the class that the stab is
2886 This is followed by three semi-colons. One marks the end of the
2887 current sub-section, one marks the end of the method field, and the
2888 third marks the end of the struct definition.
2890 For classes containing virtual functions the very last section of the
2891 string part of the stab holds a type reference to the first base
2892 class. This is preceeded by @samp{~%} and followed by a final semi-colon.
2895 .stabs "class_name(A):type_def(20)=sym_desc(struct)struct_bytes(8)
2896 field_name(Adat):type_ref(int),bit_offset(0),field_bits(32);
2897 field_name(A virt func ptr):type_def(21)=type_desc(ptr to)type_def(22)=
2898 sym_desc(array)index_type_ref(range of int from 0 to 1);
2899 elem_type_ref(vtbl elem type),
2901 meth_name(A_virt)::typedef(23)=sym_desc(method)returning(int);
2902 :arg_type(int),protection(public)normal(yes)virtual(yes)
2903 vtable_index(1);class_first_defining(A);;;~%first_base(A);",
2907 @c FIXME: bogus line break.
2909 .stabs "A:t20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
2910 A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
2914 @section Inheritence
2916 Stabs describing C++ derived classes include additional sections that
2917 describe the inheritence hierarchy of the class. A derived class stab
2918 also encodes the number of base classes. For each base class it tells
2919 if the base class is virtual or not, and if the inheritence is private
2920 or public. It also gives the offset into the object of the portion of
2921 the object corresponding to each base class.
2923 This additional information is embeded in the class stab following the
2924 number of bytes in the struct. First the number of base classes
2925 appears bracketed by an exclamation point and a comma.
2927 Then for each base type there repeats a series: a virtual character, a
2928 visibilty character, a number, a comma, another number, and a
2931 The virtual character is @samp{1} if the base class is virtual and
2932 @samp{0} if not. The visibility character is @samp{2} if the derivation
2933 is public, @samp{1} if it is protected, and @samp{0} if it is private.
2934 Debuggers should ignore virtual or visibility characters they do not
2935 recognize, and assume a reasonable default (such as public and
2936 non-virtual) (GDB 4.11 does not, but this should be fixed in the next
2939 The number following the virtual and visibility characters is the offset
2940 from the start of the object to the part of the object pertaining to the
2943 After the comma, the second number is a type_descriptor for the base
2944 type. Finally a semi-colon ends the series, which repeats for each
2947 The source below defines three base classes @code{A}, @code{B}, and
2948 @code{C} and the derived class @code{D}.
2955 virtual int A_virt (int arg) @{ return arg; @};
2961 virtual int B_virt (int arg) @{return arg; @};
2967 virtual int C_virt (int arg) @{return arg; @};
2970 class D : A, virtual B, public C @{
2973 virtual int A_virt (int arg ) @{ return arg+1; @};
2974 virtual int B_virt (int arg) @{ return arg+2; @};
2975 virtual int C_virt (int arg) @{ return arg+3; @};
2976 virtual int D_virt (int arg) @{ return arg; @};
2980 Class stabs similar to the ones described earlier are generated for
2983 @c FIXME!!! the linebreaks in the following example probably make the
2984 @c examples literally unusable, but I don't know any other way to get
2985 @c them on the page.
2986 @c One solution would be to put some of the type definitions into
2987 @c separate stabs, even if that's not exactly what the compiler actually
2990 .stabs "A:T20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
2991 A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
2993 .stabs "B:Tt25=s8Bdat:1,0,32;$vf25:21,32;B_virt::26=##1;
2994 :i;2A*-2147483647;25;;;~%25;",128,0,0,0
2996 .stabs "C:Tt28=s8Cdat:1,0,32;$vf28:21,32;C_virt::29=##1;
2997 :i;2A*-2147483647;28;;;~%28;",128,0,0,0
3000 In the stab describing derived class @code{D} below, the information about
3001 the derivation of this class is encoded as follows.
3004 .stabs "derived_class_name:symbol_descriptors(struct tag&type)=
3005 type_descriptor(struct)struct_bytes(32)!num_bases(3),
3006 base_virtual(no)inheritence_public(no)base_offset(0),
3007 base_class_type_ref(A);
3008 base_virtual(yes)inheritence_public(no)base_offset(NIL),
3009 base_class_type_ref(B);
3010 base_virtual(no)inheritence_public(yes)base_offset(64),
3011 base_class_type_ref(C); @dots{}
3014 @c FIXME! fake linebreaks.
3016 .stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:
3017 1,160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt:
3018 :32:i;2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;
3019 28;;D_virt::32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
3022 @node Virtual Base Classes
3023 @section Virtual Base Classes
3025 A derived class object consists of a concatination in memory of the data
3026 areas defined by each base class, starting with the leftmost and ending
3027 with the rightmost in the list of base classes. The exception to this
3028 rule is for virtual inheritence. In the example above, class @code{D}
3029 inherits virtually from base class @code{B}. This means that an
3030 instance of a @code{D} object will not contain its own @code{B} part but
3031 merely a pointer to a @code{B} part, known as a virtual base pointer.
3033 In a derived class stab, the base offset part of the derivation
3034 information, described above, shows how the base class parts are
3035 ordered. The base offset for a virtual base class is always given as 0.
3036 Notice that the base offset for @code{B} is given as 0 even though
3037 @code{B} is not the first base class. The first base class @code{A}
3040 The field information part of the stab for class @code{D} describes the field
3041 which is the pointer to the virtual base class @code{B}. The vbase pointer
3042 name is @samp{$vb} followed by a type reference to the virtual base class.
3043 Since the type id for @code{B} in this example is 25, the vbase pointer name
3046 @c FIXME!! fake linebreaks below
3048 .stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:1,
3049 160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt::32:i;
3050 2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;28;;D_virt:
3051 :32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
3054 Following the name and a semicolon is a type reference describing the
3055 type of the virtual base class pointer, in this case 24. Type 24 was
3056 defined earlier as the type of the @code{B} class @code{this} pointer. The
3057 @code{this} pointer for a class is a pointer to the class type.
3060 .stabs "this:P24=*25=xsB:",64,0,0,8
3063 Finally the field offset part of the vbase pointer field description
3064 shows that the vbase pointer is the first field in the @code{D} object,
3065 before any data fields defined by the class. The layout of a @code{D}
3066 class object is a follows, @code{Adat} at 0, the vtable pointer for
3067 @code{A} at 32, @code{Cdat} at 64, the vtable pointer for C at 96, the
3068 virtual base pointer for @code{B} at 128, and @code{Ddat} at 160.
3071 @node Static Members
3072 @section Static Members
3074 The data area for a class is a concatenation of the space used by the
3075 data members of the class. If the class has virtual methods, a vtable
3076 pointer follows the class data. The field offset part of each field
3077 description in the class stab shows this ordering.
3079 << How is this reflected in stabs? See Cygnus bug #677 for some info. >>
3082 @appendix Table of Stab Types
3084 The following are all the possible values for the stab type field, for
3085 a.out files, in numeric order. This does not apply to XCOFF, but
3086 it does apply to stabs in sections (@pxref{Stab Sections}). Stabs in
3087 ECOFF use these values but add 0x8f300 to distinguish them from non-stab
3090 The symbolic names are defined in the file @file{include/aout/stabs.def}.
3093 * Non-Stab Symbol Types:: Types from 0 to 0x1f
3094 * Stab Symbol Types:: Types from 0x20 to 0xff
3097 @node Non-Stab Symbol Types
3098 @appendixsec Non-Stab Symbol Types
3100 The following types are used by the linker and assembler, not by stab
3101 directives. Since this document does not attempt to describe aspects of
3102 object file format other than the debugging format, no details are
3105 @c Try to get most of these to fit on a single line.
3115 File scope absolute symbol
3117 @item 0x3 N_ABS | N_EXT
3118 External absolute symbol
3121 File scope text symbol
3123 @item 0x5 N_TEXT | N_EXT
3124 External text symbol
3127 File scope data symbol
3129 @item 0x7 N_DATA | N_EXT
3130 External data symbol
3133 File scope BSS symbol
3135 @item 0x9 N_BSS | N_EXT
3139 Same as @code{N_FN}, for Sequent compilers
3142 Symbol is indirected to another symbol
3145 Common---visible after shared library dynamic link
3148 @itemx 0x15 N_SETA | N_EXT
3149 Absolute set element
3152 @itemx 0x17 N_SETT | N_EXT
3153 Text segment set element
3156 @itemx 0x19 N_SETD | N_EXT
3157 Data segment set element
3160 @itemx 0x1b N_SETB | N_EXT
3161 BSS segment set element
3164 @itemx 0x1d N_SETV | N_EXT
3165 Pointer to set vector
3167 @item 0x1e N_WARNING
3168 Print a warning message during linking
3171 File name of a @file{.o} file
3174 @node Stab Symbol Types
3175 @appendixsec Stab Symbol Types
3177 The following symbol types indicate that this is a stab. This is the
3178 full list of stab numbers, including stab types that are used in
3179 languages other than C.
3183 Global symbol; see @ref{Global Variables}.
3186 Function name (for BSD Fortran); see @ref{Procedures}.
3189 Function name (@pxref{Procedures}) or text segment variable
3193 Data segment file-scope variable; see @ref{Statics}.
3196 BSS segment file-scope variable; see @ref{Statics}.
3199 Name of main routine; see @ref{Main Program}.
3202 Variable in @code{.rodata} section; see @ref{Statics}.
3205 Global symbol (for Pascal); see @ref{N_PC}.
3208 Number of symbols (according to Ultrix V4.0); see @ref{N_NSYMS}.
3211 No DST map; see @ref{N_NOMAP}.
3213 @c FIXME: describe this solaris feature in the body of the text (see
3214 @c comments in include/aout/stab.def).
3216 Object file (Solaris2).
3218 @c See include/aout/stab.def for (a little) more info.
3220 Debugger options (Solaris2).
3223 Register variable; see @ref{Register Variables}.
3226 Modula-2 compilation unit; see @ref{N_M2C}.
3229 Line number in text segment; see @ref{Line Numbers}.
3232 Line number in data segment; see @ref{Line Numbers}.
3235 Line number in bss segment; see @ref{Line Numbers}.
3238 Sun source code browser, path to @file{.cb} file; see @ref{N_BROWS}.
3241 GNU Modula2 definition module dependency; see @ref{N_DEFD}.
3244 Function start/body/end line numbers (Solaris2).
3247 GNU C++ exception variable; see @ref{N_EHDECL}.
3250 Modula2 info "for imc" (according to Ultrix V4.0); see @ref{N_MOD2}.
3253 GNU C++ @code{catch} clause; see @ref{N_CATCH}.
3256 Structure of union element; see @ref{N_SSYM}.
3259 Last stab for module (Solaris2).
3262 Path and name of source file; see @ref{Source Files}.
3265 Stack variable (@pxref{Stack Variables}) or type (@pxref{Typedefs}).
3268 Beginning of an include file (Sun only); see @ref{Include Files}.
3271 Name of include file; see @ref{Include Files}.
3274 Parameter variable; see @ref{Parameters}.
3277 End of an include file; see @ref{Include Files}.
3280 Alternate entry point; see @ref{Alternate Entry Points}.
3283 Beginning of a lexical block; see @ref{Block Structure}.
3286 Place holder for a deleted include file; see @ref{Include Files}.
3289 Modula2 scope information (Sun linker); see @ref{N_SCOPE}.
3292 End of a lexical block; see @ref{Block Structure}.
3295 Begin named common block; see @ref{Common Blocks}.
3298 End named common block; see @ref{Common Blocks}.
3301 Member of a common block; see @ref{Common Blocks}.
3303 @c FIXME: How does this really work? Move it to main body of document.
3305 Pascal @code{with} statement: type,,0,0,offset (Solaris2).
3308 Gould non-base registers; see @ref{Gould}.
3311 Gould non-base registers; see @ref{Gould}.
3314 Gould non-base registers; see @ref{Gould}.
3317 Gould non-base registers; see @ref{Gould}.
3320 Gould non-base registers; see @ref{Gould}.
3323 @c Restore the default table indent
3328 @node Symbol Descriptors
3329 @appendix Table of Symbol Descriptors
3331 The symbol descriptor is the character which follows the colon in many
3332 stabs, and which tells what kind of stab it is. @xref{String Field},
3333 for more information about their use.
3335 @c Please keep this alphabetical
3337 @c In TeX, this looks great, digit is in italics. But makeinfo insists
3338 @c on putting it in `', not realizing that @var should override @code.
3339 @c I don't know of any way to make makeinfo do the right thing. Seems
3340 @c like a makeinfo bug to me.
3344 Variable on the stack; see @ref{Stack Variables}.
3347 C++ nested symbol; see @xref{Nested Symbols}
3350 Parameter passed by reference in register; see @ref{Reference Parameters}.
3353 Based variable; see @ref{Based Variables}.
3356 Constant; see @ref{Constants}.
3359 Conformant array bound (Pascal, maybe other languages); @ref{Conformant
3360 Arrays}. Name of a caught exception (GNU C++). These can be
3361 distinguished because the latter uses @code{N_CATCH} and the former uses
3362 another symbol type.
3365 Floating point register variable; see @ref{Register Variables}.
3368 Parameter in floating point register; see @ref{Register Parameters}.
3371 File scope function; see @ref{Procedures}.
3374 Global function; see @ref{Procedures}.
3377 Global variable; see @ref{Global Variables}.
3380 @xref{Register Parameters}.
3383 Internal (nested) procedure; see @ref{Nested Procedures}.
3386 Internal (nested) function; see @ref{Nested Procedures}.
3389 Label name (documented by AIX, no further information known).
3392 Module; see @ref{Procedures}.
3395 Argument list parameter; see @ref{Parameters}.
3401 Fortran Function parameter; see @ref{Parameters}.
3404 Unfortunately, three separate meanings have been independently invented
3405 for this symbol descriptor. At least the GNU and Sun uses can be
3406 distinguished by the symbol type. Global Procedure (AIX) (symbol type
3407 used unknown); see @ref{Procedures}. Register parameter (GNU) (symbol
3408 type @code{N_PSYM}); see @ref{Parameters}. Prototype of function
3409 referenced by this file (Sun @code{acc}) (symbol type @code{N_FUN}).
3412 Static Procedure; see @ref{Procedures}.
3415 Register parameter; see @ref{Register Parameters}.
3418 Register variable; see @ref{Register Variables}.
3421 File scope variable; see @ref{Statics}.
3424 Local variable (OS9000).
3427 Type name; see @ref{Typedefs}.
3430 Enumeration, structure, or union tag; see @ref{Typedefs}.
3433 Parameter passed by reference; see @ref{Reference Parameters}.
3436 Procedure scope static variable; see @ref{Statics}.
3439 Conformant array; see @ref{Conformant Arrays}.
3442 Function return variable; see @ref{Parameters}.
3445 @node Type Descriptors
3446 @appendix Table of Type Descriptors
3448 The type descriptor is the character which follows the type number and
3449 an equals sign. It specifies what kind of type is being defined.
3450 @xref{String Field}, for more information about their use.
3455 Type reference; see @ref{String Field}.
3458 Reference to builtin type; see @ref{Negative Type Numbers}.
3461 Method (C++); see @ref{Method Type Descriptor}.
3464 Pointer; see @ref{Miscellaneous Types}.
3470 Type Attributes (AIX); see @ref{String Field}. Member (class and variable)
3471 type (GNU C++); see @ref{Member Type Descriptor}.
3474 Array; see @ref{Arrays}.
3477 Open array; see @ref{Arrays}.
3480 Pascal space type (AIX); see @ref{Miscellaneous Types}. Builtin integer
3481 type (Sun); see @ref{Builtin Type Descriptors}. Const and volatile
3482 qualfied type (OS9000).
3485 Volatile-qualified type; see @ref{Miscellaneous Types}.
3488 Complex builtin type (AIX); see @ref{Builtin Type Descriptors}.
3489 Const-qualified type (OS9000).
3492 COBOL Picture type. See AIX documentation for details.
3495 File type; see @ref{Miscellaneous Types}.
3498 N-dimensional dynamic array; see @ref{Arrays}.
3501 Enumeration type; see @ref{Enumerations}.
3504 N-dimensional subarray; see @ref{Arrays}.
3507 Function type; see @ref{Function Types}.
3510 Pascal function parameter; see @ref{Function Types}
3513 Builtin floating point type; see @ref{Builtin Type Descriptors}.
3516 COBOL Group. See AIX documentation for details.
3519 Imported type (AIX); see @ref{Cross-References}. Volatile-qualified
3523 Const-qualified type; see @ref{Miscellaneous Types}.
3526 COBOL File Descriptor. See AIX documentation for details.
3529 Multiple instance type; see @ref{Miscellaneous Types}.
3532 String type; see @ref{Strings}.
3535 Stringptr; see @ref{Strings}.
3538 Opaque type; see @ref{Typedefs}.
3541 Procedure; see @ref{Function Types}.
3544 Packed array; see @ref{Arrays}.
3547 Range type; see @ref{Subranges}.
3550 Builtin floating type; see @ref{Builtin Type Descriptors} (Sun). Pascal
3551 subroutine parameter; see @ref{Function Types} (AIX). Detecting this
3552 conflict is possible with careful parsing (hint: a Pascal subroutine
3553 parameter type will always contain a comma, and a builtin type
3554 descriptor never will).
3557 Structure type; see @ref{Structures}.
3560 Set type; see @ref{Miscellaneous Types}.
3563 Union; see @ref{Unions}.
3566 Variant record. This is a Pascal and Modula-2 feature which is like a
3567 union within a struct in C. See AIX documentation for details.
3570 Wide character; see @ref{Builtin Type Descriptors}.
3573 Cross-reference; see @ref{Cross-References}.
3576 Used by IBM's xlC C++ compiler (for structures, I think).
3579 gstring; see @ref{Strings}.
3582 @node Expanded Reference
3583 @appendix Expanded Reference by Stab Type
3585 @c FIXME: This appendix should go away; see N_PSYM or N_SO for an example.
3587 For a full list of stab types, and cross-references to where they are
3588 described, see @ref{Stab Types}. This appendix just covers certain
3589 stabs which are not yet described in the main body of this document;
3590 eventually the information will all be in one place.
3594 The first line is the symbol type (see @file{include/aout/stab.def}).
3596 The second line describes the language constructs the symbol type
3599 The third line is the stab format with the significant stab fields
3600 named and the rest NIL.
3602 Subsequent lines expand upon the meaning and possible values for each
3603 significant stab field.
3605 Finally, any further information.
3608 * N_PC:: Pascal global symbol
3609 * N_NSYMS:: Number of symbols
3610 * N_NOMAP:: No DST map
3611 * N_M2C:: Modula-2 compilation unit
3612 * N_BROWS:: Path to .cb file for Sun source code browser
3613 * N_DEFD:: GNU Modula2 definition module dependency
3614 * N_EHDECL:: GNU C++ exception variable
3615 * N_MOD2:: Modula2 information "for imc"
3616 * N_CATCH:: GNU C++ "catch" clause
3617 * N_SSYM:: Structure or union element
3618 * N_SCOPE:: Modula2 scope information (Sun only)
3619 * Gould:: non-base register symbols used on Gould systems
3620 * N_LENG:: Length of preceding entry
3626 @deffn @code{.stabs} N_PC
3628 Global symbol (for Pascal).
3631 "name" -> "symbol_name" <<?>>
3632 value -> supposedly the line number (stab.def is skeptical)
3636 @file{stabdump.c} says:
3638 global pascal symbol: name,,0,subtype,line
3646 @deffn @code{.stabn} N_NSYMS
3648 Number of symbols (according to Ultrix V4.0).
3651 0, files,,funcs,lines (stab.def)
3658 @deffn @code{.stabs} N_NOMAP
3660 No DST map for symbol (according to Ultrix V4.0). I think this means a
3661 variable has been optimized out.
3664 name, ,0,type,ignored (stab.def)
3671 @deffn @code{.stabs} N_M2C
3673 Modula-2 compilation unit.
3676 "string" -> "unit_name,unit_time_stamp[,code_time_stamp]"
3678 value -> 0 (main unit)
3682 See @cite{Dbx and Dbxtool Interfaces}, 2nd edition, by Sun, 1988, for
3690 @deffn @code{.stabs} N_BROWS
3692 Sun source code browser, path to @file{.cb} file
3695 "path to associated @file{.cb} file"
3697 Note: N_BROWS has the same value as N_BSLINE.
3703 @deffn @code{.stabn} N_DEFD
3705 GNU Modula2 definition module dependency.
3707 GNU Modula-2 definition module dependency. The value is the
3708 modification time of the definition file. The other field is non-zero
3709 if it is imported with the GNU M2 keyword @code{%INITIALIZE}. Perhaps
3710 @code{N_M2C} can be used if there are enough empty fields?
3716 @deffn @code{.stabs} N_EHDECL
3718 GNU C++ exception variable <<?>>.
3720 "@var{string} is variable name"
3722 Note: conflicts with @code{N_MOD2}.
3728 @deffn @code{.stab?} N_MOD2
3730 Modula2 info "for imc" (according to Ultrix V4.0)
3732 Note: conflicts with @code{N_EHDECL} <<?>>
3738 @deffn @code{.stabn} N_CATCH
3740 GNU C++ @code{catch} clause
3742 GNU C++ @code{catch} clause. The value is its address. The desc field
3743 is nonzero if this entry is immediately followed by a @code{CAUGHT} stab
3744 saying what exception was caught. Multiple @code{CAUGHT} stabs means
3745 that multiple exceptions can be caught here. If desc is 0, it means all
3746 exceptions are caught here.
3752 @deffn @code{.stabn} N_SSYM
3754 Structure or union element.
3756 The value is the offset in the structure.
3758 <<?looking at structs and unions in C I didn't see these>>
3764 @deffn @code{.stab?} N_SCOPE
3766 Modula2 scope information (Sun linker)
3771 @section Non-base registers on Gould systems
3773 @deffn @code{.stab?} N_NBTEXT
3774 @deffnx @code{.stab?} N_NBDATA
3775 @deffnx @code{.stab?} N_NBBSS
3776 @deffnx @code{.stab?} N_NBSTS
3777 @deffnx @code{.stab?} N_NBLCS
3783 These are used on Gould systems for non-base registers syms.
3785 However, the following values are not the values used by Gould; they are
3786 the values which GNU has been documenting for these values for a long
3787 time, without actually checking what Gould uses. I include these values
3788 only because perhaps some someone actually did something with the GNU
3789 information (I hope not, why GNU knowingly assigned wrong values to
3790 these in the header file is a complete mystery to me).
3793 240 0xf0 N_NBTEXT ??
3794 242 0xf2 N_NBDATA ??
3804 @deffn @code{.stabn} N_LENG
3806 Second symbol entry containing a length-value for the preceding entry.
3807 The value is the length.
3811 @appendix Questions and Anomalies
3815 @c I think this is changed in GCC 2.4.5 to put the line number there.
3816 For GNU C stabs defining local and global variables (@code{N_LSYM} and
3817 @code{N_GSYM}), the desc field is supposed to contain the source
3818 line number on which the variable is defined. In reality the desc
3819 field is always 0. (This behavior is defined in @file{dbxout.c} and
3820 putting a line number in desc is controlled by @samp{#ifdef
3821 WINNING_GDB}, which defaults to false). GDB supposedly uses this
3822 information if you say @samp{list @var{var}}. In reality, @var{var} can
3823 be a variable defined in the program and GDB says @samp{function
3824 @var{var} not defined}.
3827 In GNU C stabs, there seems to be no way to differentiate tag types:
3828 structures, unions, and enums (symbol descriptor @samp{T}) and typedefs
3829 (symbol descriptor @samp{t}) defined at file scope from types defined locally
3830 to a procedure or other more local scope. They all use the @code{N_LSYM}
3831 stab type. Types defined at procedure scope are emited after the
3832 @code{N_RBRAC} of the preceding function and before the code of the
3833 procedure in which they are defined. This is exactly the same as
3834 types defined in the source file between the two procedure bodies.
3835 GDB overcompensates by placing all types in block #1, the block for
3836 symbols of file scope. This is true for default, @samp{-ansi} and
3837 @samp{-traditional} compiler options. (Bugs gcc/1063, gdb/1066.)
3840 What ends the procedure scope? Is it the proc block's @code{N_RBRAC} or the
3841 next @code{N_FUN}? (I believe its the first.)
3844 @node Sun Differences
3845 @appendix Differences Between GNU Stabs and Sun Native Stabs
3847 @c FIXME: Merge all this stuff into the main body of the document.
3851 GNU C stabs define @emph{all} types, file or procedure scope, as
3852 @code{N_LSYM}. Sun doc talks about using @code{N_GSYM} too.
3855 Sun C stabs use type number pairs in the format
3856 (@var{file-number},@var{type-number}) where @var{file-number} is a
3857 number starting with 1 and incremented for each sub-source file in the
3858 compilation. @var{type-number} is a number starting with 1 and
3859 incremented for each new type defined in the compilation. GNU C stabs
3860 use the type number alone, with no source file number.
3864 @appendix Using Stabs in Their Own Sections
3866 Many object file formats allow tools to create object files with custom
3867 sections containing any arbitrary data. For any such object file
3868 format, stabs can be embedded in special sections. This is how stabs
3869 are used with ELF and SOM, and aside from ECOFF and XCOFF, is how stabs
3873 * Stab Section Basics:: How to embed stabs in sections
3874 * ELF Linker Relocation:: Sun ELF hacks
3877 @node Stab Section Basics
3878 @appendixsec How to Embed Stabs in Sections
3880 The assembler creates two custom sections, a section named @code{.stab}
3881 which contains an array of fixed length structures, one struct per stab,
3882 and a section named @code{.stabstr} containing all the variable length
3883 strings that are referenced by stabs in the @code{.stab} section. The
3884 byte order of the stabs binary data depends on the object file format.
3885 For ELF, it matches the byte order of the ELF file itself, as determined
3886 from the @code{EI_DATA} field in the @code{e_ident} member of the ELF
3887 header. For SOM, it is always big-endian (is this true??? FIXME). For
3888 COFF, it matches the byte order of the COFF headers. The meaning of the
3889 fields is the same as for a.out (@pxref{Symbol Table Format}), except
3890 that the @code{n_strx} field is relative to the strings for the current
3891 compilation unit (which can be found using the synthetic N_UNDF stab
3892 described below), rather than the entire string table.
3894 The first stab in the @code{.stab} section for each compilation unit is
3895 synthetic, generated entirely by the assembler, with no corresponding
3896 @code{.stab} directive as input to the assembler. This stab contains
3897 the following fields:
3901 Offset in the @code{.stabstr} section to the source filename.
3907 Unused field, always zero.
3908 This may eventually be used to hold overflows from the count in
3909 the @code{n_desc} field.
3912 Count of upcoming symbols, i.e., the number of remaining stabs for this
3916 Size of the string table fragment associated with this source file, in
3920 The @code{.stabstr} section always starts with a null byte (so that string
3921 offsets of zero reference a null string), followed by random length strings,
3922 each of which is null byte terminated.
3924 The ELF section header for the @code{.stab} section has its
3925 @code{sh_link} member set to the section number of the @code{.stabstr}
3926 section, and the @code{.stabstr} section has its ELF section
3927 header @code{sh_type} member set to @code{SHT_STRTAB} to mark it as a
3928 string table. SOM and COFF have no way of linking the sections together
3929 or marking them as string tables.
3931 For COFF, the @code{.stab} and @code{.stabstr} sections are simply
3932 concatenated by the linker. GDB then uses the @code{n_desc} fields to
3933 figure out the extent of the original sections. Similarly, the
3934 @code{n_value} fields of the header symbols are added together in order
3935 to get the actual position of the strings in a desired @code{.stabstr}
3936 section. Although this design obviates any need for the linker to relocate
3937 or otherwise manipulate @code{.stab} and @code{.stabstr} sections, it also
3938 requires some care to ensure that the offsets are calculated correctly.
3939 For instance, if the linker were to pad in between the @code{.stabstr}
3940 sections before concatenating, then the offsets to strings in the middle
3941 of the executable's @code{.stabstr} section would be wrong.
3943 @node ELF Linker Relocation
3944 @appendixsec Having the Linker Relocate Stabs in ELF
3946 This section describes some Sun hacks for Stabs in ELF; it does not
3947 apply to COFF or SOM.
3949 To keep linking fast, you don't want the linker to have to relocate very
3950 many stabs. Making sure this is done for @code{N_SLINE},
3951 @code{N_RBRAC}, and @code{N_LBRAC} stabs is the most important thing
3952 (see the descriptions of those stabs for more information). But Sun's
3953 stabs in ELF has taken this further, to make all addresses in the
3954 @code{n_value} field (functions and static variables) relative to the
3955 source file. For the @code{N_SO} symbol itself, Sun simply omits the
3956 address. To find the address of each section corresponding to a given
3957 source file, the compiler puts out symbols giving the address of each
3958 section for a given source file. Since these are ELF (not stab)
3959 symbols, the linker relocates them correctly without having to touch the
3960 stabs section. They are named @code{Bbss.bss} for the bss section,
3961 @code{Ddata.data} for the data section, and @code{Drodata.rodata} for
3962 the rodata section. For the text section, there is no such symbol (but
3963 there should be, see below). For an example of how these symbols work,
3964 @xref{Stab Section Transformations}. GCC does not provide these symbols;
3965 it instead relies on the stabs getting relocated. Thus addresses which
3966 would normally be relative to @code{Bbss.bss}, etc., are already
3967 relocated. The Sun linker provided with Solaris 2.2 and earlier
3968 relocates stabs using normal ELF relocation information, as it would do
3969 for any section. Sun has been threatening to kludge their linker to not
3970 do this (to speed up linking), even though the correct way to avoid
3971 having the linker do these relocations is to have the compiler no longer
3972 output relocatable values. Last I heard they had been talked out of the
3973 linker kludge. See Sun point patch 101052-01 and Sun bug 1142109. With
3974 the Sun compiler this affects @samp{S} symbol descriptor stabs
3975 (@pxref{Statics}) and functions (@pxref{Procedures}). In the latter
3976 case, to adopt the clean solution (making the value of the stab relative
3977 to the start of the compilation unit), it would be necessary to invent a
3978 @code{Ttext.text} symbol, analogous to the @code{Bbss.bss}, etc.,
3979 symbols. I recommend this rather than using a zero value and getting
3980 the address from the ELF symbols.
3982 Finding the correct @code{Bbss.bss}, etc., symbol is difficult, because
3983 the linker simply concatenates the @code{.stab} sections from each
3984 @file{.o} file without including any information about which part of a
3985 @code{.stab} section comes from which @file{.o} file. The way GDB does
3986 this is to look for an ELF @code{STT_FILE} symbol which has the same
3987 name as the last component of the file name from the @code{N_SO} symbol
3988 in the stabs (for example, if the file name is @file{../../gdb/main.c},
3989 it looks for an ELF @code{STT_FILE} symbol named @code{main.c}). This
3990 loses if different files have the same name (they could be in different
3991 directories, a library could have been copied from one system to
3992 another, etc.). It would be much cleaner to have the @code{Bbss.bss}
3993 symbols in the stabs themselves. Having the linker relocate them there
3994 is no more work than having the linker relocate ELF symbols, and it
3995 solves the problem of having to associate the ELF and stab symbols.
3996 However, no one has yet designed or implemented such a scheme.
3998 @node Symbol Types Index
3999 @unnumbered Symbol Types Index