1\input texinfo
2@setfilename stabs.info
3@setchapternewpage odd
4@settitle STABS
5
6@c man begin INCLUDE
7@include gdb-cfg.texi
8@c man end
9
10@c @finalout
11
12@c This is a dir.info fragment to support semi-automated addition of
13@c manuals to an info tree.
14@dircategory Software development
15@direntry
16* Stabs: (stabs).                 The "stabs" debugging information format.
17@end direntry
18
19@copying
20Copyright @copyright{} 1992--2020 Free Software Foundation, Inc.
21Contributed by Cygnus Support.  Written by Julia Menapace, Jim Kingdon,
22and David MacKenzie.
23
24Permission is granted to copy, distribute and/or modify this document
25under the terms of the GNU Free Documentation License, Version 1.3 or
26any later version published by the Free Software Foundation; with no
27Invariant Sections, with no Front-Cover Texts, and with no Back-Cover
28Texts.  A copy of the license is included in the section entitled ``GNU
29Free Documentation License''.
30@end copying
31
32@ifnottex
33This document describes the stabs debugging symbol tables.
34
35@insertcopying
36@end ifnottex
37
38@titlepage
39@title The ``stabs'' debug format
40@author Julia Menapace, Jim Kingdon, David MacKenzie
41@author Cygnus Support
42@page
43@tex
44\def\$#1${{#1}}  % Kludge: collect RCS revision info without $...$
45\xdef\manvers{\$Revision: 1.9 $}  % For use in headers, footers too
46{\parskip=0pt
47\hfill Cygnus Support\par
48\hfill \manvers\par
49\hfill \TeX{}info \texinfoversion\par
50}
51@end tex
52
53@vskip 0pt plus 1filll
54@insertcopying
55@end titlepage
56
57@ifnottex
58@node Top
59@top The "stabs" representation of debugging information
60
61This document describes the stabs debugging format.
62
63@menu
64* Overview::			Overview of stabs
65* Program Structure::		Encoding of the structure of the program
66* Constants::			Constants
67* Variables::
68* Types::			Type definitions
69* Macro define and undefine::	Representation of #define and #undef
70* Symbol Tables::		Symbol information in symbol tables
71* Cplusplus::			Stabs specific to C++
72* Stab Types::			Symbol types in a.out files
73* Symbol Descriptors::		Table of symbol descriptors
74* Type Descriptors::		Table of type descriptors
75* Expanded Reference::		Reference information by stab type
76* Questions::			Questions and anomalies
77* Stab Sections::		In some object file formats, stabs are
78                                in sections.
79* GNU Free Documentation License::  The license for this documentation
80* Symbol Types Index::          Index of symbolic stab symbol type names.
81@end menu
82@end ifnottex
83
84@contents
85
86@node Overview
87@chapter Overview of Stabs
88
89@dfn{Stabs} refers to a format for information that describes a program
90to a debugger.  This format was apparently invented by
91Peter Kessler at
92the University of California at Berkeley, for the @code{pdx} Pascal
93debugger; the format has spread widely since then.
94
95This document is one of the few published sources of documentation on
96stabs.  It is believed to be comprehensive for stabs used by C.  The
97lists of symbol descriptors (@pxref{Symbol Descriptors}) and type
98descriptors (@pxref{Type Descriptors}) are believed to be completely
99comprehensive.  Stabs for COBOL-specific features and for variant
100records (used by Pascal and Modula-2) are poorly documented here.
101
102@c FIXME: Need to document all OS9000 stuff in GDB; see all references
103@c to os9k_stabs in stabsread.c.
104
105Other sources of information on stabs are @cite{Dbx and Dbxtool
106Interfaces}, 2nd edition, by Sun, 1988, and @cite{AIX Version 3.2 Files
107Reference}, Fourth Edition, September 1992, "dbx Stabstring Grammar" in
108the a.out section, page 2-31.  This document is believed to incorporate
109the information from those two sources except where it explicitly directs
110you to them for more information.
111
112@menu
113* Flow::			Overview of debugging information flow
114* Stabs Format::		Overview of stab format
115* String Field::		The string field
116* C Example::			A simple example in C source
117* Assembly Code::		The simple example at the assembly level
118@end menu
119
120@node Flow
121@section Overview of Debugging Information Flow
122
123The GNU C compiler compiles C source in a @file{.c} file into assembly
124language in a @file{.s} file, which the assembler translates into
125a @file{.o} file, which the linker combines with other @file{.o} files and
126libraries to produce an executable file.
127
128With the @samp{-g} option, GCC puts in the @file{.s} file additional
129debugging information, which is slightly transformed by the assembler
130and linker, and carried through into the final executable.  This
131debugging information describes features of the source file like line
132numbers, the types and scopes of variables, and function names,
133parameters, and scopes.
134
135For some object file formats, the debugging information is encapsulated
136in assembler directives known collectively as @dfn{stab} (symbol table)
137directives, which are interspersed with the generated code.  Stabs are
138the native format for debugging information in the a.out and XCOFF
139object file formats.  The GNU tools can also emit stabs in the COFF and
140ECOFF object file formats.
141
142The assembler adds the information from stabs to the symbol information
143it places by default in the symbol table and the string table of the
144@file{.o} file it is building.  The linker consolidates the @file{.o}
145files into one executable file, with one symbol table and one string
146table.  Debuggers use the symbol and string tables in the executable as
147a source of debugging information about the program.
148
149@node Stabs Format
150@section Overview of Stab Format
151
152There are three overall formats for stab assembler directives,
153differentiated by the first word of the stab.  The name of the directive
154describes which combination of four possible data fields follows.  It is
155either @code{.stabs} (string), @code{.stabn} (number), or @code{.stabd}
156(dot).  IBM's XCOFF assembler uses @code{.stabx} (and some other
157directives such as @code{.file} and @code{.bi}) instead of
158@code{.stabs}, @code{.stabn} or @code{.stabd}.
159
160The overall format of each class of stab is:
161
162@example
163.stabs "@var{string}",@var{type},@var{other},@var{desc},@var{value}
164.stabn @var{type},@var{other},@var{desc},@var{value}
165.stabd @var{type},@var{other},@var{desc}
166.stabx "@var{string}",@var{value},@var{type},@var{sdb-type}
167@end example
168
169@c what is the correct term for "current file location"?  My AIX
170@c assembler manual calls it "the value of the current location counter".
171For @code{.stabn} and @code{.stabd}, there is no @var{string} (the
172@code{n_strx} field is zero; see @ref{Symbol Tables}).  For
173@code{.stabd}, the @var{value} field is implicit and has the value of
174the current file location.  For @code{.stabx}, the @var{sdb-type} field
175is unused for stabs and can always be set to zero.  The @var{other}
176field is almost always unused and can be set to zero.
177
178The number in the @var{type} field gives some basic information about
179which type of stab this is (or whether it @emph{is} a stab, as opposed
180to an ordinary symbol).  Each valid type number defines a different stab
181type; further, the stab type defines the exact interpretation of, and
182possible values for, any remaining @var{string}, @var{desc}, or
183@var{value} fields present in the stab.  @xref{Stab Types}, for a list
184in numeric order of the valid @var{type} field values for stab directives.
185
186@node String Field
187@section The String Field
188
189For most stabs the string field holds the meat of the
190debugging information.  The flexible nature of this field
191is what makes stabs extensible.  For some stab types the string field
192contains only a name.  For other stab types the contents can be a great
193deal more complex.
194
195The overall format of the string field for most stab types is:
196
197@example
198"@var{name}:@var{symbol-descriptor} @var{type-information}"
199@end example
200
201@var{name} is the name of the symbol represented by the stab; it can
202contain a pair of colons (@pxref{Nested Symbols}).  @var{name} can be
203omitted, which means the stab represents an unnamed object.  For
204example, @samp{:t10=*2} defines type 10 as a pointer to type 2, but does
205not give the type a name.  Omitting the @var{name} field is supported by
206AIX dbx and GDB after about version 4.8, but not other debuggers.  GCC
207sometimes uses a single space as the name instead of omitting the name
208altogether; apparently that is supported by most debuggers.
209
210The @var{symbol-descriptor} following the @samp{:} is an alphabetic
211character that tells more specifically what kind of symbol the stab
212represents. If the @var{symbol-descriptor} is omitted, but type
213information follows, then the stab represents a local variable.  For a
214list of symbol descriptors, see @ref{Symbol Descriptors}.  The @samp{c}
215symbol descriptor is an exception in that it is not followed by type
216information.  @xref{Constants}.
217
218@var{type-information} is either a @var{type-number}, or
219@samp{@var{type-number}=}.  A @var{type-number} alone is a type
220reference, referring directly to a type that has already been defined.
221
222The @samp{@var{type-number}=} form is a type definition, where the
223number represents a new type which is about to be defined.  The type
224definition may refer to other types by number, and those type numbers
225may be followed by @samp{=} and nested definitions.  Also, the Lucid
226compiler will repeat @samp{@var{type-number}=} more than once if it
227wants to define several type numbers at once.
228
229In a type definition, if the character that follows the equals sign is
230non-numeric then it is a @var{type-descriptor}, and tells what kind of
231type is about to be defined.  Any other values following the
232@var{type-descriptor} vary, depending on the @var{type-descriptor}.
233@xref{Type Descriptors}, for a list of @var{type-descriptor} values.  If
234a number follows the @samp{=} then the number is a @var{type-reference}.
235For a full description of types, @ref{Types}.
236
237A @var{type-number} is often a single number.  The GNU and Sun tools
238additionally permit a @var{type-number} to be a pair
239(@var{file-number},@var{filetype-number}) (the parentheses appear in the
240string, and serve to distinguish the two cases).  The @var{file-number}
241is 0 for the base source file, 1 for the first included file, 2 for the
242next, and so on.  The @var{filetype-number} is a number starting with
2431 which is incremented for each new type defined in the file.
244(Separating the file number and the type number permits the
245@code{N_BINCL} optimization to succeed more often; see @ref{Include
246Files}).
247
248There is an AIX extension for type attributes.  Following the @samp{=}
249are any number of type attributes.  Each one starts with @samp{@@} and
250ends with @samp{;}.  Debuggers, including AIX's dbx and GDB 4.10, skip
251any type attributes they do not recognize.  GDB 4.9 and other versions
252of dbx may not do this.  Because of a conflict with C@t{++}
253(@pxref{Cplusplus}), new attributes should not be defined which begin
254with a digit, @samp{(}, or @samp{-}; GDB may be unable to distinguish
255those from the C@t{++} type descriptor @samp{@@}.  The attributes are:
256
257@table @code
258@item a@var{boundary}
259@var{boundary} is an integer specifying the alignment.  I assume it
260applies to all variables of this type.
261
262@item p@var{integer}
263Pointer class (for checking).  Not sure what this means, or how
264@var{integer} is interpreted.
265
266@item P
267Indicate this is a packed type, meaning that structure fields or array
268elements are placed more closely in memory, to save memory at the
269expense of speed.
270
271@item s@var{size}
272Size in bits of a variable of this type.  This is fully supported by GDB
2734.11 and later.
274
275@item S
276Indicate that this type is a string instead of an array of characters,
277or a bitstring instead of a set.  It doesn't change the layout of the
278data being represented, but does enable the debugger to know which type
279it is.
280
281@item V
282Indicate that this type is a vector instead of an array.  The only
283major difference between vectors and arrays is that vectors are
284passed by value instead of by reference (vector coprocessor extension).
285
286@end table
287
288All of this can make the string field quite long.  All versions of GDB,
289and some versions of dbx, can handle arbitrarily long strings.  But many
290versions of dbx (or assemblers or linkers, I'm not sure which)
291cretinously limit the strings to about 80 characters, so compilers which
292must work with such systems need to split the @code{.stabs} directive
293into several @code{.stabs} directives.  Each stab duplicates every field
294except the string field.  The string field of every stab except the last
295is marked as continued with a backslash at the end (in the assembly code
296this may be written as a double backslash, depending on the assembler).
297Removing the backslashes and concatenating the string fields of each
298stab produces the original, long string.  Just to be incompatible (or so
299they don't have to worry about what the assembler does with
300backslashes), AIX can use @samp{?} instead of backslash.
301
302@node C Example
303@section A Simple Example in C Source
304
305To get the flavor of how stabs describe source information for a C
306program, let's look at the simple program:
307
308@example
309main()
310@{
311        printf("Hello world");
312@}
313@end example
314
315When compiled with @samp{-g}, the program above yields the following
316@file{.s} file.  Line numbers have been added to make it easier to refer
317to parts of the @file{.s} file in the description of the stabs that
318follows.
319
320@node Assembly Code
321@section The Simple Example at the Assembly Level
322
323This simple ``hello world'' example demonstrates several of the stab
324types used to describe C language source files.
325
326@example
3271  gcc2_compiled.:
3282  .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0
3293  .stabs "hello.c",100,0,0,Ltext0
3304  .text
3315  Ltext0:
3326  .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0
3337  .stabs "char:t2=r2;0;127;",128,0,0,0
3348  .stabs "long int:t3=r1;-2147483648;2147483647;",128,0,0,0
3359  .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
33610 .stabs "long unsigned int:t5=r1;0;-1;",128,0,0,0
33711 .stabs "short int:t6=r1;-32768;32767;",128,0,0,0
33812 .stabs "long long int:t7=r1;0;-1;",128,0,0,0
33913 .stabs "short unsigned int:t8=r1;0;65535;",128,0,0,0
34014 .stabs "long long unsigned int:t9=r1;0;-1;",128,0,0,0
34115 .stabs "signed char:t10=r1;-128;127;",128,0,0,0
34216 .stabs "unsigned char:t11=r1;0;255;",128,0,0,0
34317 .stabs "float:t12=r1;4;0;",128,0,0,0
34418 .stabs "double:t13=r1;8;0;",128,0,0,0
34519 .stabs "long double:t14=r1;8;0;",128,0,0,0
34620 .stabs "void:t15=15",128,0,0,0
34721      .align 4
34822 LC0:
34923      .ascii "Hello, world!\12\0"
35024      .align 4
35125      .global _main
35226      .proc 1
35327 _main:
35428 .stabn 68,0,4,LM1
35529 LM1:
35630      !#PROLOGUE# 0
35731      save %sp,-136,%sp
35832      !#PROLOGUE# 1
35933      call ___main,0
36034      nop
36135 .stabn 68,0,5,LM2
36236 LM2:
36337 LBB2:
36438      sethi %hi(LC0),%o1
36539      or %o1,%lo(LC0),%o0
36640      call _printf,0
36741      nop
36842 .stabn 68,0,6,LM3
36943 LM3:
37044 LBE2:
37145 .stabn 68,0,6,LM4
37246 LM4:
37347 L1:
37448      ret
37549      restore
37650 .stabs "main:F1",36,0,0,_main
37751 .stabn 192,0,0,LBB2
37852 .stabn 224,0,0,LBE2
379@end example
380
381@node Program Structure
382@chapter Encoding the Structure of the Program
383
384The elements of the program structure that stabs encode include the name
385of the main function, the names of the source and include files, the
386line numbers, procedure names and types, and the beginnings and ends of
387blocks of code.
388
389@menu
390* Main Program::		Indicate what the main program is
391* Source Files::		The path and name of the source file
392* Include Files::               Names of include files
393* Line Numbers::
394* Procedures::
395* Nested Procedures::
396* Block Structure::
397* Alternate Entry Points::      Entering procedures except at the beginning.
398@end menu
399
400@node Main Program
401@section Main Program
402
403@findex N_MAIN
404Most languages allow the main program to have any name.  The
405@code{N_MAIN} stab type tells the debugger the name that is used in this
406program.  Only the string field is significant; it is the name of
407a function which is the main program.  Most C compilers do not use this
408stab (they expect the debugger to assume that the name is @code{main}),
409but some C compilers emit an @code{N_MAIN} stab for the @code{main}
410function.  I'm not sure how XCOFF handles this.
411
412@node Source Files
413@section Paths and Names of the Source Files
414
415@findex N_SO
416Before any other stabs occur, there must be a stab specifying the source
417file.  This information is contained in a symbol of stab type
418@code{N_SO}; the string field contains the name of the file.  The
419value of the symbol is the start address of the portion of the
420text section corresponding to that file.
421
422Some compilers use the desc field to indicate the language of the
423source file.  Sun's compilers started this usage, and the first
424constants are derived from their documentation.  Languages added
425by gcc/gdb start at 0x32 to avoid conflict with languages Sun may
426add in the future.  A desc field with a value 0 indicates that no
427language has been specified via this mechanism.
428
429@table @asis
430@item @code{N_SO_AS} (0x1)
431Assembly language
432@item @code{N_SO_C}  (0x2)
433K&R traditional C
434@item @code{N_SO_ANSI_C} (0x3)
435ANSI C
436@item @code{N_SO_CC}  (0x4)
437C++
438@item @code{N_SO_FORTRAN} (0x5)
439Fortran
440@item @code{N_SO_PASCAL} (0x6)
441Pascal
442@item @code{N_SO_FORTRAN90} (0x7)
443Fortran90
444@item @code{N_SO_OBJC} (0x32)
445Objective-C
446@item @code{N_SO_OBJCPLUS} (0x33)
447Objective-C++
448@end table
449
450Some compilers (for example, GCC2 and SunOS4 @file{/bin/cc}) also
451include the directory in which the source was compiled, in a second
452@code{N_SO} symbol preceding the one containing the file name.  This
453symbol can be distinguished by the fact that it ends in a slash.  Code
454from the @code{cfront} C@t{++} compiler can have additional @code{N_SO} symbols for
455nonexistent source files after the @code{N_SO} for the real source file;
456these are believed to contain no useful information.
457
458For example:
459
460@example
461.stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0     # @r{100 is N_SO}
462.stabs "hello.c",100,0,0,Ltext0
463        .text
464Ltext0:
465@end example
466
467@findex C_FILE
468Instead of @code{N_SO} symbols, XCOFF uses a @code{.file} assembler
469directive which assembles to a @code{C_FILE} symbol; explaining this in
470detail is outside the scope of this document.
471
472@c FIXME: Exactly when should the empty N_SO be used?  Why?
473If it is useful to indicate the end of a source file, this is done with
474an @code{N_SO} symbol with an empty string for the name.  The value is
475the address of the end of the text section for the file.  For some
476systems, there is no indication of the end of a source file, and you
477just need to figure it ended when you see an @code{N_SO} for a different
478source file, or a symbol ending in @code{.o} (which at least some
479linkers insert to mark the start of a new @code{.o} file).
480
481@node Include Files
482@section Names of Include Files
483
484There are several schemes for dealing with include files: the
485traditional @code{N_SOL} approach, Sun's @code{N_BINCL} approach, and the
486XCOFF @code{C_BINCL} approach (which despite the similar name has little in
487common with @code{N_BINCL}).
488
489@findex N_SOL
490An @code{N_SOL} symbol specifies which include file subsequent symbols
491refer to.  The string field is the name of the file and the value is the
492text address corresponding to the end of the previous include file and
493the start of this one.  To specify the main source file again, use an
494@code{N_SOL} symbol with the name of the main source file.
495
496@findex N_BINCL
497@findex N_EINCL
498@findex N_EXCL
499The @code{N_BINCL} approach works as follows.  An @code{N_BINCL} symbol
500specifies the start of an include file.  In an object file, only the
501string is significant; the linker puts data into some of the other
502fields.  The end of the include file is marked by an @code{N_EINCL}
503symbol (which has no string field).  In an object file, there is no
504significant data in the @code{N_EINCL} symbol.  @code{N_BINCL} and
505@code{N_EINCL} can be nested.
506
507If the linker detects that two source files have identical stabs between
508an @code{N_BINCL} and @code{N_EINCL} pair (as will generally be the case
509for a header file), then it only puts out the stabs once.  Each
510additional occurrence is replaced by an @code{N_EXCL} symbol.  I believe
511the GNU linker and the Sun (both SunOS4 and Solaris) linker are the only
512ones which supports this feature.
513
514A linker which supports this feature will set the value of a
515@code{N_BINCL} symbol to the total of all the characters in the stabs
516strings included in the header file, omitting any file numbers.  The
517value of an @code{N_EXCL} symbol is the same as the value of the
518@code{N_BINCL} symbol it replaces.  This information can be used to
519match up @code{N_EXCL} and @code{N_BINCL} symbols which have the same
520filename.  The @code{N_EINCL} value, and the values of the other and
521description fields for all three, appear to always be zero.
522
523@findex C_BINCL
524@findex C_EINCL
525For the start of an include file in XCOFF, use the @file{.bi} assembler
526directive, which generates a @code{C_BINCL} symbol.  A @file{.ei}
527directive, which generates a @code{C_EINCL} symbol, denotes the end of
528the include file.  Both directives are followed by the name of the
529source file in quotes, which becomes the string for the symbol.
530The value of each symbol, produced automatically by the assembler
531and linker, is the offset into the executable of the beginning
532(inclusive, as you'd expect) or end (inclusive, as you would not expect)
533of the portion of the COFF line table that corresponds to this include
534file.  @code{C_BINCL} and @code{C_EINCL} do not nest.
535
536@node Line Numbers
537@section Line Numbers
538
539@findex N_SLINE
540An @code{N_SLINE} symbol represents the start of a source line.  The
541desc field contains the line number and the value contains the code
542address for the start of that source line.  On most machines the address
543is absolute; for stabs in sections (@pxref{Stab Sections}), it is
544relative to the function in which the @code{N_SLINE} symbol occurs.
545
546@findex N_DSLINE
547@findex N_BSLINE
548GNU documents @code{N_DSLINE} and @code{N_BSLINE} symbols for line
549numbers in the data or bss segments, respectively.  They are identical
550to @code{N_SLINE} but are relocated differently by the linker.  They
551were intended to be used to describe the source location of a variable
552declaration, but I believe that GCC2 actually puts the line number in
553the desc field of the stab for the variable itself.  GDB has been
554ignoring these symbols (unless they contain a string field) since
555at least GDB 3.5.
556
557For single source lines that generate discontiguous code, such as flow
558of control statements, there may be more than one line number entry for
559the same source line.  In this case there is a line number entry at the
560start of each code range, each with the same line number.
561
562XCOFF does not use stabs for line numbers.  Instead, it uses COFF line
563numbers (which are outside the scope of this document).  Standard COFF
564line numbers cannot deal with include files, but in XCOFF this is fixed
565with the @code{C_BINCL} method of marking include files (@pxref{Include
566Files}).
567
568@node Procedures
569@section Procedures
570
571@findex N_FUN, for functions
572@findex N_FNAME
573@findex N_STSYM, for functions (Sun acc)
574@findex N_GSYM, for functions (Sun acc)
575All of the following stabs normally use the @code{N_FUN} symbol type.
576However, Sun's @code{acc} compiler on SunOS4 uses @code{N_GSYM} and
577@code{N_STSYM}, which means that the value of the stab for the function
578is useless and the debugger must get the address of the function from
579the non-stab symbols instead.  On systems where non-stab symbols have
580leading underscores, the stabs will lack underscores and the debugger
581needs to know about the leading underscore to match up the stab and the
582non-stab symbol.  BSD Fortran is said to use @code{N_FNAME} with the
583same restriction; the value of the symbol is not useful (I'm not sure it
584really does use this, because GDB doesn't handle this and no one has
585complained).
586
587@findex C_FUN
588A function is represented by an @samp{F} symbol descriptor for a global
589(extern) function, and @samp{f} for a static (local) function.  For
590a.out, the value of the symbol is the address of the start of the
591function; it is already relocated.  For stabs in ELF, the SunPRO
592compiler version 2.0.1 and GCC put out an address which gets relocated
593by the linker.  In a future release SunPRO is planning to put out zero,
594in which case the address can be found from the ELF (non-stab) symbol.
595Because looking things up in the ELF symbols would probably be slow, I'm
596not sure how to find which symbol of that name is the right one, and
597this doesn't provide any way to deal with nested functions, it would
598probably be better to make the value of the stab an address relative to
599the start of the file, or just absolute.  See @ref{ELF Linker
600Relocation} for more information on linker relocation of stabs in ELF
601files.  For XCOFF, the stab uses the @code{C_FUN} storage class and the
602value of the stab is meaningless; the address of the function can be
603found from the csect symbol (XTY_LD/XMC_PR).
604
605The type information of the stab represents the return type of the
606function; thus @samp{foo:f5} means that foo is a function returning type
6075.  There is no need to try to get the line number of the start of the
608function from the stab for the function; it is in the next
609@code{N_SLINE} symbol.
610
611@c FIXME: verify whether the "I suspect" below is true or not.
612Some compilers (such as Sun's Solaris compiler) support an extension for
613specifying the types of the arguments.  I suspect this extension is not
614used for old (non-prototyped) function definitions in C.  If the
615extension is in use, the type information of the stab for the function
616is followed by type information for each argument, with each argument
617preceded by @samp{;}.  An argument type of 0 means that additional
618arguments are being passed, whose types and number may vary (@samp{...}
619in ANSI C).  GDB has tolerated this extension (parsed the syntax, if not
620necessarily used the information) since at least version 4.8; I don't
621know whether all versions of dbx tolerate it.  The argument types given
622here are not redundant with the symbols for the formal parameters
623(@pxref{Parameters}); they are the types of the arguments as they are
624passed, before any conversions might take place.  For example, if a C
625function which is declared without a prototype takes a @code{float}
626argument, the value is passed as a @code{double} but then converted to a
627@code{float}.  Debuggers need to use the types given in the arguments
628when printing values, but when calling the function they need to use the
629types given in the symbol defining the function.
630
631If the return type and types of arguments of a function which is defined
632in another source file are specified (i.e., a function prototype in ANSI
633C), traditionally compilers emit no stab; the only way for the debugger
634to find the information is if the source file where the function is
635defined was also compiled with debugging symbols.  As an extension the
636Solaris compiler uses symbol descriptor @samp{P} followed by the return
637type of the function, followed by the arguments, each preceded by
638@samp{;}, as in a stab with symbol descriptor @samp{f} or @samp{F}.
639This use of symbol descriptor @samp{P} can be distinguished from its use
640for register parameters (@pxref{Register Parameters}) by the fact that it has
641symbol type @code{N_FUN}.
642
643The AIX documentation also defines symbol descriptor @samp{J} as an
644internal function.  I assume this means a function nested within another
645function.  It also says symbol descriptor @samp{m} is a module in
646Modula-2 or extended Pascal.
647
648Procedures (functions which do not return values) are represented as
649functions returning the @code{void} type in C.  I don't see why this couldn't
650be used for all languages (inventing a @code{void} type for this purpose if
651necessary), but the AIX documentation defines @samp{I}, @samp{P}, and
652@samp{Q} for internal, global, and static procedures, respectively.
653These symbol descriptors are unusual in that they are not followed by
654type information.
655
656The following example shows a stab for a function @code{main} which
657returns type number @code{1}.  The @code{_main} specified for the value
658is a reference to an assembler label which is used to fill in the start
659address of the function.
660
661@example
662.stabs "main:F1",36,0,0,_main      # @r{36 is N_FUN}
663@end example
664
665The stab representing a procedure is located immediately following the
666code of the procedure.  This stab is in turn directly followed by a
667group of other stabs describing elements of the procedure.  These other
668stabs describe the procedure's parameters, its block local variables, and
669its block structure.
670
671If functions can appear in different sections, then the debugger may not
672be able to find the end of a function.  Recent versions of GCC will mark
673the end of a function with an @code{N_FUN} symbol with an empty string
674for the name.  The value is the address of the end of the current
675function.  Without such a symbol, there is no indication of the address
676of the end of a function, and you must assume that it ended at the
677starting address of the next function or at the end of the text section
678for the program.
679
680@node Nested Procedures
681@section Nested Procedures
682
683For any of the symbol descriptors representing procedures, after the
684symbol descriptor and the type information is optionally a scope
685specifier.  This consists of a comma, the name of the procedure, another
686comma, and the name of the enclosing procedure.  The first name is local
687to the scope specified, and seems to be redundant with the name of the
688symbol (before the @samp{:}).  This feature is used by GCC, and
689presumably Pascal, Modula-2, etc., compilers, for nested functions.
690
691If procedures are nested more than one level deep, only the immediately
692containing scope is specified.  For example, this code:
693
694@example
695int
696foo (int x)
697@{
698  int bar (int y)
699    @{
700      int baz (int z)
701        @{
702          return x + y + z;
703        @}
704      return baz (x + 2 * y);
705    @}
706  return x + bar (3 * x);
707@}
708@end example
709
710@noindent
711produces the stabs:
712
713@example
714.stabs "baz:f1,baz,bar",36,0,0,_baz.15         # @r{36 is N_FUN}
715.stabs "bar:f1,bar,foo",36,0,0,_bar.12
716.stabs "foo:F1",36,0,0,_foo
717@end example
718
719@node Block Structure
720@section Block Structure
721
722@findex N_LBRAC
723@findex N_RBRAC
724@c For GCC 2.5.8 or so stabs-in-coff, these are absolute instead of
725@c function relative (as documented below).  But GDB has never been able
726@c to deal with that (it had wanted them to be relative to the file, but
727@c I just fixed that (between GDB 4.12 and 4.13)), so it is function
728@c relative just like ELF and SOM and the below documentation.
729The program's block structure is represented by the @code{N_LBRAC} (left
730brace) and the @code{N_RBRAC} (right brace) stab types.  The variables
731defined inside a block precede the @code{N_LBRAC} symbol for most
732compilers, including GCC.  Other compilers, such as the Convex, Acorn
733RISC machine, and Sun @code{acc} compilers, put the variables after the
734@code{N_LBRAC} symbol.  The values of the @code{N_LBRAC} and
735@code{N_RBRAC} symbols are the start and end addresses of the code of
736the block, respectively.  For most machines, they are relative to the
737starting address of this source file.  For the Gould NP1, they are
738absolute.  For stabs in sections (@pxref{Stab Sections}), they are
739relative to the function in which they occur.
740
741The @code{N_LBRAC} and @code{N_RBRAC} stabs that describe the block
742scope of a procedure are located after the @code{N_FUN} stab that
743represents the procedure itself.
744
745Sun documents the desc field of @code{N_LBRAC} and
746@code{N_RBRAC} symbols as containing the nesting level of the block.
747However, dbx seems to not care, and GCC always sets desc to
748zero.
749
750@findex .bb
751@findex .be
752@findex C_BLOCK
753For XCOFF, block scope is indicated with @code{C_BLOCK} symbols.  If the
754name of the symbol is @samp{.bb}, then it is the beginning of the block;
755if the name of the symbol is @samp{.be}; it is the end of the block.
756
757@node Alternate Entry Points
758@section Alternate Entry Points
759
760@findex N_ENTRY
761@findex C_ENTRY
762Some languages, like Fortran, have the ability to enter procedures at
763some place other than the beginning.  One can declare an alternate entry
764point.  The @code{N_ENTRY} stab is for this; however, the Sun FORTRAN
765compiler doesn't use it.  According to AIX documentation, only the name
766of a @code{C_ENTRY} stab is significant; the address of the alternate
767entry point comes from the corresponding external symbol.  A previous
768revision of this document said that the value of an @code{N_ENTRY} stab
769was the address of the alternate entry point, but I don't know the
770source for that information.
771
772@node Constants
773@chapter Constants
774
775The @samp{c} symbol descriptor indicates that this stab represents a
776constant.  This symbol descriptor is an exception to the general rule
777that symbol descriptors are followed by type information.  Instead, it
778is followed by @samp{=} and one of the following:
779
780@table @code
781@item b @var{value}
782Boolean constant.  @var{value} is a numeric value; I assume it is 0 for
783false or 1 for true.
784
785@item c @var{value}
786Character constant.  @var{value} is the numeric value of the constant.
787
788@item e @var{type-information} , @var{value}
789Constant whose value can be represented as integral.
790@var{type-information} is the type of the constant, as it would appear
791after a symbol descriptor (@pxref{String Field}).  @var{value} is the
792numeric value of the constant.  GDB 4.9 does not actually get the right
793value if @var{value} does not fit in a host @code{int}, but it does not
794do anything violent, and future debuggers could be extended to accept
795integers of any size (whether unsigned or not).  This constant type is
796usually documented as being only for enumeration constants, but GDB has
797never imposed that restriction; I don't know about other debuggers.
798
799@item i @var{value}
800Integer constant.  @var{value} is the numeric value.  The type is some
801sort of generic integer type (for GDB, a host @code{int}); to specify
802the type explicitly, use @samp{e} instead.
803
804@item r @var{value}
805Real constant.  @var{value} is the real value, which can be @samp{INF}
806(optionally preceded by a sign) for infinity, @samp{QNAN} for a quiet
807NaN (not-a-number), or @samp{SNAN} for a signalling NaN.  If it is a
808normal number the format is that accepted by the C library function
809@code{atof}.
810
811@item s @var{string}
812String constant.  @var{string} is a string enclosed in either @samp{'}
813(in which case @samp{'} characters within the string are represented as
814@samp{\'} or @samp{"} (in which case @samp{"} characters within the
815string are represented as @samp{\"}).
816
817@item S @var{type-information} , @var{elements} , @var{bits} , @var{pattern}
818Set constant.  @var{type-information} is the type of the constant, as it
819would appear after a symbol descriptor (@pxref{String Field}).
820@var{elements} is the number of elements in the set (does this means
821how many bits of @var{pattern} are actually used, which would be
822redundant with the type, or perhaps the number of bits set in
823@var{pattern}?  I don't get it), @var{bits} is the number of bits in the
824constant (meaning it specifies the length of @var{pattern}, I think),
825and @var{pattern} is a hexadecimal representation of the set.  AIX
826documentation refers to a limit of 32 bytes, but I see no reason why
827this limit should exist.  This form could probably be used for arbitrary
828constants, not just sets; the only catch is that @var{pattern} should be
829understood to be target, not host, byte order and format.
830@end table
831
832The boolean, character, string, and set constants are not supported by
833GDB 4.9, but it ignores them.  GDB 4.8 and earlier gave an error
834message and refused to read symbols from the file containing the
835constants.
836
837The above information is followed by @samp{;}.
838
839@node Variables
840@chapter Variables
841
842Different types of stabs describe the various ways that variables can be
843allocated: on the stack, globally, in registers, in common blocks,
844statically, or as arguments to a function.
845
846@menu
847* Stack Variables::		Variables allocated on the stack.
848* Global Variables::		Variables used by more than one source file.
849* Register Variables::		Variables in registers.
850* Common Blocks::		Variables statically allocated together.
851* Statics::			Variables local to one source file.
852* Based Variables::		Fortran pointer based variables.
853* Parameters::			Variables for arguments to functions.
854@end menu
855
856@node Stack Variables
857@section Automatic Variables Allocated on the Stack
858
859If a variable's scope is local to a function and its lifetime is only as
860long as that function executes (C calls such variables
861@dfn{automatic}), it can be allocated in a register (@pxref{Register
862Variables}) or on the stack.
863
864@findex N_LSYM, for stack variables
865@findex C_LSYM
866Each variable allocated on the stack has a stab with the symbol
867descriptor omitted.  Since type information should begin with a digit,
868@samp{-}, or @samp{(}, only those characters precluded from being used
869for symbol descriptors.  However, the Acorn RISC machine (ARM) is said
870to get this wrong: it puts out a mere type definition here, without the
871preceding @samp{@var{type-number}=}.  This is a bad idea; there is no
872guarantee that type descriptors are distinct from symbol descriptors.
873Stabs for stack variables use the @code{N_LSYM} stab type, or
874@code{C_LSYM} for XCOFF.
875
876The value of the stab is the offset of the variable within the
877local variables.  On most machines this is an offset from the frame
878pointer and is negative.  The location of the stab specifies which block
879it is defined in; see @ref{Block Structure}.
880
881For example, the following C code:
882
883@example
884int
885main ()
886@{
887  int x;
888@}
889@end example
890
891produces the following stabs:
892
893@example
894.stabs "main:F1",36,0,0,_main   # @r{36 is N_FUN}
895.stabs "x:1",128,0,0,-12        # @r{128 is N_LSYM}
896.stabn 192,0,0,LBB2             # @r{192 is N_LBRAC}
897.stabn 224,0,0,LBE2             # @r{224 is N_RBRAC}
898@end example
899
900See @ref{Procedures} for more information on the @code{N_FUN} stab, and
901@ref{Block Structure} for more information on the @code{N_LBRAC} and
902@code{N_RBRAC} stabs.
903
904@node Global Variables
905@section Global Variables
906
907@findex N_GSYM
908@findex C_GSYM
909@c FIXME: verify for sure that it really is C_GSYM on XCOFF
910A variable whose scope is not specific to just one source file is
911represented by the @samp{G} symbol descriptor.  These stabs use the
912@code{N_GSYM} stab type (C_GSYM for XCOFF).  The type information for
913the stab (@pxref{String Field}) gives the type of the variable.
914
915For example, the following source code:
916
917@example
918char g_foo = 'c';
919@end example
920
921@noindent
922yields the following assembly code:
923
924@example
925.stabs "g_foo:G2",32,0,0,0     # @r{32 is N_GSYM}
926     .global _g_foo
927     .data
928_g_foo:
929     .byte 99
930@end example
931
932The address of the variable represented by the @code{N_GSYM} is not
933contained in the @code{N_GSYM} stab.  The debugger gets this information
934from the external symbol for the global variable.  In the example above,
935the @code{.global _g_foo} and @code{_g_foo:} lines tell the assembler to
936produce an external symbol.
937
938Some compilers, like GCC, output @code{N_GSYM} stabs only once, where
939the variable is defined.  Other compilers, like SunOS4 /bin/cc, output a
940@code{N_GSYM} stab for each compilation unit which references the
941variable.
942
943@node Register Variables
944@section Register Variables
945
946@findex N_RSYM
947@findex C_RSYM
948@c According to an old version of this manual, AIX uses C_RPSYM instead
949@c of C_RSYM.  I am skeptical; this should be verified.
950Register variables have their own stab type, @code{N_RSYM}
951(@code{C_RSYM} for XCOFF), and their own symbol descriptor, @samp{r}.
952The stab's value is the number of the register where the variable data
953will be stored.
954@c .stabs "name:type",N_RSYM,0,RegSize,RegNumber (Sun doc)
955
956AIX defines a separate symbol descriptor @samp{d} for floating point
957registers.  This seems unnecessary; why not just just give floating
958point registers different register numbers?  I have not verified whether
959the compiler actually uses @samp{d}.
960
961If the register is explicitly allocated to a global variable, but not
962initialized, as in:
963
964@example
965register int g_bar asm ("%g5");
966@end example
967
968@noindent
969then the stab may be emitted at the end of the object file, with
970the other bss symbols.
971
972@node Common Blocks
973@section Common Blocks
974
975A common block is a statically allocated section of memory which can be
976referred to by several source files.  It may contain several variables.
977I believe Fortran is the only language with this feature.
978
979@findex N_BCOMM
980@findex N_ECOMM
981@findex C_BCOMM
982@findex C_ECOMM
983A @code{N_BCOMM} stab begins a common block and an @code{N_ECOMM} stab
984ends it.  The only field that is significant in these two stabs is the
985string, which names a normal (non-debugging) symbol that gives the
986address of the common block.  According to IBM documentation, only the
987@code{N_BCOMM} has the name of the common block (even though their
988compiler actually puts it both places).
989
990@findex N_ECOML
991@findex C_ECOML
992The stabs for the members of the common block are between the
993@code{N_BCOMM} and the @code{N_ECOMM}; the value of each stab is the
994offset within the common block of that variable.  IBM uses the
995@code{C_ECOML} stab type, and there is a corresponding @code{N_ECOML}
996stab type, but Sun's Fortran compiler uses @code{N_GSYM} instead.  The
997variables within a common block use the @samp{V} symbol descriptor (I
998believe this is true of all Fortran variables).  Other stabs (at least
999type declarations using @code{C_DECL}) can also be between the
1000@code{N_BCOMM} and the @code{N_ECOMM}.
1001
1002@node Statics
1003@section Static Variables
1004
1005Initialized static variables are represented by the @samp{S} and
1006@samp{V} symbol descriptors.  @samp{S} means file scope static, and
1007@samp{V} means procedure scope static.  One exception: in XCOFF, IBM's
1008xlc compiler always uses @samp{V}, and whether it is file scope or not
1009is distinguished by whether the stab is located within a function.
1010
1011@c This is probably not worth mentioning; it is only true on the sparc
1012@c for `double' variables which although declared const are actually in
1013@c the data segment (the text segment can't guarantee 8 byte alignment).
1014@c (although GCC
1015@c 2.4.5 has a bug in that it uses @code{N_FUN}, so neither dbx nor GDB can
1016@c find the variables)
1017@findex N_STSYM
1018@findex N_LCSYM
1019@findex N_FUN, for variables
1020@findex N_ROSYM
1021In a.out files, @code{N_STSYM} means the data section, @code{N_FUN}
1022means the text section, and @code{N_LCSYM} means the bss section.  For
1023those systems with a read-only data section separate from the text
1024section (Solaris), @code{N_ROSYM} means the read-only data section.
1025
1026For example, the source lines:
1027
1028@example
1029static const int var_const = 5;
1030static int var_init = 2;
1031static int var_noinit;
1032@end example
1033
1034@noindent
1035yield the following stabs:
1036
1037@example
1038.stabs "var_const:S1",36,0,0,_var_const      # @r{36 is N_FUN}
1039@dots{}
1040.stabs "var_init:S1",38,0,0,_var_init        # @r{38 is N_STSYM}
1041@dots{}
1042.stabs "var_noinit:S1",40,0,0,_var_noinit    # @r{40 is N_LCSYM}
1043@end example
1044
1045@findex C_STSYM
1046@findex C_BSTAT
1047@findex C_ESTAT
1048In XCOFF files, the stab type need not indicate the section;
1049@code{C_STSYM} can be used for all statics.  Also, each static variable
1050is enclosed in a static block.  A @code{C_BSTAT} (emitted with a
1051@samp{.bs} assembler directive) symbol begins the static block; its
1052value is the symbol number of the csect symbol whose value is the
1053address of the static block, its section is the section of the variables
1054in that static block, and its name is @samp{.bs}.  A @code{C_ESTAT}
1055(emitted with a @samp{.es} assembler directive) symbol ends the static
1056block; its name is @samp{.es} and its value and section are ignored.
1057
1058In ECOFF files, the storage class is used to specify the section, so the
1059stab type need not indicate the section.
1060
1061In ELF files, for the SunPRO compiler version 2.0.1, symbol descriptor
1062@samp{S} means that the address is absolute (the linker relocates it)
1063and symbol descriptor @samp{V} means that the address is relative to the
1064start of the relevant section for that compilation unit.  SunPRO has
1065plans to have the linker stop relocating stabs; I suspect that their the
1066debugger gets the address from the corresponding ELF (not stab) symbol.
1067I'm not sure how to find which symbol of that name is the right one.
1068The clean way to do all this would be to have the value of a symbol
1069descriptor @samp{S} symbol be an offset relative to the start of the
1070file, just like everything else, but that introduces obvious
1071compatibility problems.  For more information on linker stab relocation,
1072@xref{ELF Linker Relocation}.
1073
1074@node Based Variables
1075@section Fortran Based Variables
1076
1077Fortran (at least, the Sun and SGI dialects of FORTRAN-77) has a feature
1078which allows allocating arrays with @code{malloc}, but which avoids
1079blurring the line between arrays and pointers the way that C does.  In
1080stabs such a variable uses the @samp{b} symbol descriptor.
1081
1082For example, the Fortran declarations
1083
1084@example
1085real foo, foo10(10), foo10_5(10,5)
1086pointer (foop, foo)
1087pointer (foo10p, foo10)
1088pointer (foo105p, foo10_5)
1089@end example
1090
1091produce the stabs
1092
1093@example
1094foo:b6
1095foo10:bar3;1;10;6
1096foo10_5:bar3;1;5;ar3;1;10;6
1097@end example
1098
1099In this example, @code{real} is type 6 and type 3 is an integral type
1100which is the type of the subscripts of the array (probably
1101@code{integer}).
1102
1103The @samp{b} symbol descriptor is like @samp{V} in that it denotes a
1104statically allocated symbol whose scope is local to a function; see
1105@xref{Statics}.  The value of the symbol, instead of being the address
1106of the variable itself, is the address of a pointer to that variable.
1107So in the above example, the value of the @code{foo} stab is the address
1108of a pointer to a real, the value of the @code{foo10} stab is the
1109address of a pointer to a 10-element array of reals, and the value of
1110the @code{foo10_5} stab is the address of a pointer to a 5-element array
1111of 10-element arrays of reals.
1112
1113@node Parameters
1114@section Parameters
1115
1116Formal parameters to a function are represented by a stab (or sometimes
1117two; see below) for each parameter.  The stabs are in the order in which
1118the debugger should print the parameters (i.e., the order in which the
1119parameters are declared in the source file).  The exact form of the stab
1120depends on how the parameter is being passed.
1121
1122@findex N_PSYM
1123@findex C_PSYM
1124Parameters passed on the stack use the symbol descriptor @samp{p} and
1125the @code{N_PSYM} symbol type (or @code{C_PSYM} for XCOFF).  The value
1126of the symbol is an offset used to locate the parameter on the stack;
1127its exact meaning is machine-dependent, but on most machines it is an
1128offset from the frame pointer.
1129
1130As a simple example, the code:
1131
1132@example
1133main (argc, argv)
1134     int argc;
1135     char **argv;
1136@end example
1137
1138produces the stabs:
1139
1140@example
1141.stabs "main:F1",36,0,0,_main                 # @r{36 is N_FUN}
1142.stabs "argc:p1",160,0,0,68                   # @r{160 is N_PSYM}
1143.stabs "argv:p20=*21=*2",160,0,0,72
1144@end example
1145
1146The type definition of @code{argv} is interesting because it contains
1147several type definitions.  Type 21 is pointer to type 2 (char) and
1148@code{argv} (type 20) is pointer to type 21.
1149
1150@c FIXME: figure out what these mean and describe them coherently.
1151The following symbol descriptors are also said to go with @code{N_PSYM}.
1152The value of the symbol is said to be an offset from the argument
1153pointer (I'm not sure whether this is true or not).
1154
1155@example
1156pP (<<??>>)
1157pF Fortran function parameter
1158X  (function result variable)
1159@end example
1160
1161@menu
1162* Register Parameters::
1163* Local Variable Parameters::
1164* Reference Parameters::
1165* Conformant Arrays::
1166@end menu
1167
1168@node Register Parameters
1169@subsection Passing Parameters in Registers
1170
1171If the parameter is passed in a register, then traditionally there are
1172two symbols for each argument:
1173
1174@example
1175.stabs "arg:p1" . . .       ; N_PSYM
1176.stabs "arg:r1" . . .       ; N_RSYM
1177@end example
1178
1179Debuggers use the second one to find the value, and the first one to
1180know that it is an argument.
1181
1182@findex C_RPSYM
1183@findex N_RSYM, for parameters
1184Because that approach is kind of ugly, some compilers use symbol
1185descriptor @samp{P} or @samp{R} to indicate an argument which is in a
1186register.  Symbol type @code{C_RPSYM} is used in XCOFF and @code{N_RSYM}
1187is used otherwise.  The symbol's value is the register number.  @samp{P}
1188and @samp{R} mean the same thing; the difference is that @samp{P} is a
1189GNU invention and @samp{R} is an IBM (XCOFF) invention.  As of version
11904.9, GDB should handle either one.
1191
1192There is at least one case where GCC uses a @samp{p} and @samp{r} pair
1193rather than @samp{P}; this is where the argument is passed in the
1194argument list and then loaded into a register.
1195
1196According to the AIX documentation, symbol descriptor @samp{D} is for a
1197parameter passed in a floating point register.  This seems
1198unnecessary---why not just use @samp{R} with a register number which
1199indicates that it's a floating point register?  I haven't verified
1200whether the system actually does what the documentation indicates.
1201
1202@c FIXME: On the hppa this is for any type > 8 bytes, I think, and not
1203@c for small structures (investigate).
1204On the sparc and hppa, for a @samp{P} symbol whose type is a structure
1205or union, the register contains the address of the structure.  On the
1206sparc, this is also true of a @samp{p} and @samp{r} pair (using Sun
1207@code{cc}) or a @samp{p} symbol.  However, if a (small) structure is
1208really in a register, @samp{r} is used.  And, to top it all off, on the
1209hppa it might be a structure which was passed on the stack and loaded
1210into a register and for which there is a @samp{p} and @samp{r} pair!  I
1211believe that symbol descriptor @samp{i} is supposed to deal with this
1212case (it is said to mean "value parameter by reference, indirect
1213access"; I don't know the source for this information), but I don't know
1214details or what compilers or debuggers use it, if any (not GDB or GCC).
1215It is not clear to me whether this case needs to be dealt with
1216differently than parameters passed by reference (@pxref{Reference Parameters}).
1217
1218@node Local Variable Parameters
1219@subsection Storing Parameters as Local Variables
1220
1221There is a case similar to an argument in a register, which is an
1222argument that is actually stored as a local variable.  Sometimes this
1223happens when the argument was passed in a register and then the compiler
1224stores it as a local variable.  If possible, the compiler should claim
1225that it's in a register, but this isn't always done.
1226
1227If a parameter is passed as one type and converted to a smaller type by
1228the prologue (for example, the parameter is declared as a @code{float},
1229but the calling conventions specify that it is passed as a
1230@code{double}), then GCC2 (sometimes) uses a pair of symbols.  The first
1231symbol uses symbol descriptor @samp{p} and the type which is passed.
1232The second symbol has the type and location which the parameter actually
1233has after the prologue.  For example, suppose the following C code
1234appears with no prototypes involved:
1235
1236@example
1237void
1238subr (f)
1239     float f;
1240@{
1241@end example
1242
1243if @code{f} is passed as a double at stack offset 8, and the prologue
1244converts it to a float in register number 0, then the stabs look like:
1245
1246@example
1247.stabs "f:p13",160,0,3,8   # @r{160 is @code{N_PSYM}, here 13 is @code{double}}
1248.stabs "f:r12",64,0,3,0    # @r{64 is @code{N_RSYM}, here 12 is @code{float}}
1249@end example
1250
1251In both stabs 3 is the line number where @code{f} is declared
1252(@pxref{Line Numbers}).
1253
1254@findex N_LSYM, for parameter
1255GCC, at least on the 960, has another solution to the same problem.  It
1256uses a single @samp{p} symbol descriptor for an argument which is stored
1257as a local variable but uses @code{N_LSYM} instead of @code{N_PSYM}.  In
1258this case, the value of the symbol is an offset relative to the local
1259variables for that function, not relative to the arguments; on some
1260machines those are the same thing, but not on all.
1261
1262@c This is mostly just background info; the part that logically belongs
1263@c here is the last sentence.
1264On the VAX or on other machines in which the calling convention includes
1265the number of words of arguments actually passed, the debugger (GDB at
1266least) uses the parameter symbols to keep track of whether it needs to
1267print nameless arguments in addition to the formal parameters which it
1268has printed because each one has a stab.  For example, in
1269
1270@example
1271extern int fprintf (FILE *stream, char *format, @dots{});
1272@dots{}
1273fprintf (stdout, "%d\n", x);
1274@end example
1275
1276there are stabs for @code{stream} and @code{format}.  On most machines,
1277the debugger can only print those two arguments (because it has no way
1278of knowing that additional arguments were passed), but on the VAX or
1279other machines with a calling convention which indicates the number of
1280words of arguments, the debugger can print all three arguments.  To do
1281so, the parameter symbol (symbol descriptor @samp{p}) (not necessarily
1282@samp{r} or symbol descriptor omitted symbols) needs to contain the
1283actual type as passed (for example, @code{double} not @code{float} if it
1284is passed as a double and converted to a float).
1285
1286@node Reference Parameters
1287@subsection Passing Parameters by Reference
1288
1289If the parameter is passed by reference (e.g., Pascal @code{VAR}
1290parameters), then the symbol descriptor is @samp{v} if it is in the
1291argument list, or @samp{a} if it in a register.  Other than the fact
1292that these contain the address of the parameter rather than the
1293parameter itself, they are identical to @samp{p} and @samp{R},
1294respectively.  I believe @samp{a} is an AIX invention; @samp{v} is
1295supported by all stabs-using systems as far as I know.
1296
1297@node Conformant Arrays
1298@subsection Passing Conformant Array Parameters
1299
1300@c Is this paragraph correct?  It is based on piecing together patchy
1301@c information and some guesswork
1302Conformant arrays are a feature of Modula-2, and perhaps other
1303languages, in which the size of an array parameter is not known to the
1304called function until run-time.  Such parameters have two stabs: a
1305@samp{x} for the array itself, and a @samp{C}, which represents the size
1306of the array.  The value of the @samp{x} stab is the offset in the
1307argument list where the address of the array is stored (it this right?
1308it is a guess); the value of the @samp{C} stab is the offset in the
1309argument list where the size of the array (in elements? in bytes?) is
1310stored.
1311
1312@node Types
1313@chapter Defining Types
1314
1315The examples so far have described types as references to previously
1316defined types, or defined in terms of subranges of or pointers to
1317previously defined types.  This chapter describes the other type
1318descriptors that may follow the @samp{=} in a type definition.
1319
1320@menu
1321* Builtin Types::		Integers, floating point, void, etc.
1322* Miscellaneous Types::		Pointers, sets, files, etc.
1323* Cross-References::		Referring to a type not yet defined.
1324* Subranges::			A type with a specific range.
1325* Arrays::			An aggregate type of same-typed elements.
1326* Strings::			Like an array but also has a length.
1327* Enumerations::		Like an integer but the values have names.
1328* Structures::			An aggregate type of different-typed elements.
1329* Typedefs::			Giving a type a name.
1330* Unions::			Different types sharing storage.
1331* Function Types::
1332@end menu
1333
1334@node Builtin Types
1335@section Builtin Types
1336
1337Certain types are built in (@code{int}, @code{short}, @code{void},
1338@code{float}, etc.); the debugger recognizes these types and knows how
1339to handle them.  Thus, don't be surprised if some of the following ways
1340of specifying builtin types do not specify everything that a debugger
1341would need to know about the type---in some cases they merely specify
1342enough information to distinguish the type from other types.
1343
1344The traditional way to define builtin types is convoluted, so new ways
1345have been invented to describe them.  Sun's @code{acc} uses special
1346builtin type descriptors (@samp{b} and @samp{R}), and IBM uses negative
1347type numbers.  GDB accepts all three ways, as of version 4.8; dbx just
1348accepts the traditional builtin types and perhaps one of the other two
1349formats.  The following sections describe each of these formats.
1350
1351@menu
1352* Traditional Builtin Types::	Put on your seat belts and prepare for kludgery
1353* Builtin Type Descriptors::	Builtin types with special type descriptors
1354* Negative Type Numbers::	Builtin types using negative type numbers
1355@end menu
1356
1357@node Traditional Builtin Types
1358@subsection Traditional Builtin Types
1359
1360This is the traditional, convoluted method for defining builtin types.
1361There are several classes of such type definitions: integer, floating
1362point, and @code{void}.
1363
1364@menu
1365* Traditional Integer Types::
1366* Traditional Other Types::
1367@end menu
1368
1369@node Traditional Integer Types
1370@subsubsection Traditional Integer Types
1371
1372Often types are defined as subranges of themselves.  If the bounding values
1373fit within an @code{int}, then they are given normally.  For example:
1374
1375@example
1376.stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0    # @r{128 is N_LSYM}
1377.stabs "char:t2=r2;0;127;",128,0,0,0
1378@end example
1379
1380Builtin types can also be described as subranges of @code{int}:
1381
1382@example
1383.stabs "unsigned short:t6=r1;0;65535;",128,0,0,0
1384@end example
1385
1386If the lower bound of a subrange is 0 and the upper bound is -1,
1387the type is an unsigned integral type whose bounds are too
1388big to describe in an @code{int}.  Traditionally this is only used for
1389@code{unsigned int} and @code{unsigned long}:
1390
1391@example
1392.stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
1393@end example
1394
1395For larger types, GCC 2.4.5 puts out bounds in octal, with one or more
1396leading zeroes.  In this case a negative bound consists of a number
1397which is a 1 bit (for the sign bit) followed by a 0 bit for each bit in
1398the number (except the sign bit), and a positive bound is one which is a
13991 bit for each bit in the number (except possibly the sign bit).  All
1400known versions of dbx and GDB version 4 accept this (at least in the
1401sense of not refusing to process the file), but GDB 3.5 refuses to read
1402the whole file containing such symbols.  So GCC 2.3.3 did not output the
1403proper size for these types.  As an example of octal bounds, the string
1404fields of the stabs for 64 bit integer types look like:
1405
1406@c .stabs directives, etc., omitted to make it fit on the page.
1407@example
1408long int:t3=r1;001000000000000000000000;000777777777777777777777;
1409long unsigned int:t5=r1;000000000000000000000000;001777777777777777777777;
1410@end example
1411
1412If the lower bound of a subrange is 0 and the upper bound is negative,
1413the type is an unsigned integral type whose size in bytes is the
1414absolute value of the upper bound.  I believe this is a Convex
1415convention for @code{unsigned long long}.
1416
1417If the lower bound of a subrange is negative and the upper bound is 0,
1418the type is a signed integral type whose size in bytes is
1419the absolute value of the lower bound.  I believe this is a Convex
1420convention for @code{long long}.  To distinguish this from a legitimate
1421subrange, the type should be a subrange of itself.  I'm not sure whether
1422this is the case for Convex.
1423
1424@node Traditional Other Types
1425@subsubsection Traditional Other Types
1426
1427If the upper bound of a subrange is 0 and the lower bound is positive,
1428the type is a floating point type, and the lower bound of the subrange
1429indicates the number of bytes in the type:
1430
1431@example
1432.stabs "float:t12=r1;4;0;",128,0,0,0
1433.stabs "double:t13=r1;8;0;",128,0,0,0
1434@end example
1435
1436However, GCC writes @code{long double} the same way it writes
1437@code{double}, so there is no way to distinguish.
1438
1439@example
1440.stabs "long double:t14=r1;8;0;",128,0,0,0
1441@end example
1442
1443Complex types are defined the same way as floating-point types; there is
1444no way to distinguish a single-precision complex from a double-precision
1445floating-point type.
1446
1447The C @code{void} type is defined as itself:
1448
1449@example
1450.stabs "void:t15=15",128,0,0,0
1451@end example
1452
1453I'm not sure how a boolean type is represented.
1454
1455@node Builtin Type Descriptors
1456@subsection Defining Builtin Types Using Builtin Type Descriptors
1457
1458This is the method used by Sun's @code{acc} for defining builtin types.
1459These are the type descriptors to define builtin types:
1460
1461@table @code
1462@c FIXME: clean up description of width and offset, once we figure out
1463@c what they mean
1464@item b @var{signed} @var{char-flag} @var{width} ; @var{offset} ; @var{nbits} ;
1465Define an integral type.  @var{signed} is @samp{u} for unsigned or
1466@samp{s} for signed.  @var{char-flag} is @samp{c} which indicates this
1467is a character type, or is omitted.  I assume this is to distinguish an
1468integral type from a character type of the same size, for example it
1469might make sense to set it for the C type @code{wchar_t} so the debugger
1470can print such variables differently (Solaris does not do this).  Sun
1471sets it on the C types @code{signed char} and @code{unsigned char} which
1472arguably is wrong.  @var{width} and @var{offset} appear to be for small
1473objects stored in larger ones, for example a @code{short} in an
1474@code{int} register.  @var{width} is normally the number of bytes in the
1475type.  @var{offset} seems to always be zero.  @var{nbits} is the number
1476of bits in the type.
1477
1478Note that type descriptor @samp{b} used for builtin types conflicts with
1479its use for Pascal space types (@pxref{Miscellaneous Types}); they can
1480be distinguished because the character following the type descriptor
1481will be a digit, @samp{(}, or @samp{-} for a Pascal space type, or
1482@samp{u} or @samp{s} for a builtin type.
1483
1484@item w
1485Documented by AIX to define a wide character type, but their compiler
1486actually uses negative type numbers (@pxref{Negative Type Numbers}).
1487
1488@item R @var{fp-type} ; @var{bytes} ;
1489Define a floating point type.  @var{fp-type} has one of the following values:
1490
1491@table @code
1492@item 1 (NF_SINGLE)
1493IEEE 32-bit (single precision) floating point format.
1494
1495@item 2 (NF_DOUBLE)
1496IEEE 64-bit (double precision) floating point format.
1497
1498@item 3 (NF_COMPLEX)
1499@item 4 (NF_COMPLEX16)
1500@item 5 (NF_COMPLEX32)
1501@c "GDB source" really means @file{include/aout/stab_gnu.h}, but trying
1502@c to put that here got an overfull hbox.
1503These are for complex numbers.  A comment in the GDB source describes
1504them as Fortran @code{complex}, @code{double complex}, and
1505@code{complex*16}, respectively, but what does that mean?  (i.e., Single
1506precision?  Double precision?).
1507
1508@item 6 (NF_LDOUBLE)
1509Long double.  This should probably only be used for Sun format
1510@code{long double}, and new codes should be used for other floating
1511point formats (@code{NF_DOUBLE} can be used if a @code{long double} is
1512really just an IEEE double, of course).
1513@end table
1514
1515@var{bytes} is the number of bytes occupied by the type.  This allows a
1516debugger to perform some operations with the type even if it doesn't
1517understand @var{fp-type}.
1518
1519@item g @var{type-information} ; @var{nbits}
1520Documented by AIX to define a floating type, but their compiler actually
1521uses negative type numbers (@pxref{Negative Type Numbers}).
1522
1523@item c @var{type-information} ; @var{nbits}
1524Documented by AIX to define a complex type, but their compiler actually
1525uses negative type numbers (@pxref{Negative Type Numbers}).
1526@end table
1527
1528The C @code{void} type is defined as a signed integral type 0 bits long:
1529@example
1530.stabs "void:t19=bs0;0;0",128,0,0,0
1531@end example
1532The Solaris compiler seems to omit the trailing semicolon in this case.
1533Getting sloppy in this way is not a swift move because if a type is
1534embedded in a more complex expression it is necessary to be able to tell
1535where it ends.
1536
1537I'm not sure how a boolean type is represented.
1538
1539@node Negative Type Numbers
1540@subsection Negative Type Numbers
1541
1542This is the method used in XCOFF for defining builtin types.
1543Since the debugger knows about the builtin types anyway, the idea of
1544negative type numbers is simply to give a special type number which
1545indicates the builtin type.  There is no stab defining these types.
1546
1547There are several subtle issues with negative type numbers.
1548
1549One is the size of the type.  A builtin type (for example the C types
1550@code{int} or @code{long}) might have different sizes depending on
1551compiler options, the target architecture, the ABI, etc.  This issue
1552doesn't come up for IBM tools since (so far) they just target the
1553RS/6000; the sizes indicated below for each size are what the IBM
1554RS/6000 tools use.  To deal with differing sizes, either define separate
1555negative type numbers for each size (which works but requires changing
1556the debugger, and, unless you get both AIX dbx and GDB to accept the
1557change, introduces an incompatibility), or use a type attribute
1558(@pxref{String Field}) to define a new type with the appropriate size
1559(which merely requires a debugger which understands type attributes,
1560like AIX dbx or GDB).  For example,
1561
1562@example
1563.stabs "boolean:t10=@@s8;-16",128,0,0,0
1564@end example
1565
1566defines an 8-bit boolean type, and
1567
1568@example
1569.stabs "boolean:t10=@@s64;-16",128,0,0,0
1570@end example
1571
1572defines a 64-bit boolean type.
1573
1574A similar issue is the format of the type.  This comes up most often for
1575floating-point types, which could have various formats (particularly
1576extended doubles, which vary quite a bit even among IEEE systems).
1577Again, it is best to define a new negative type number for each
1578different format; changing the format based on the target system has
1579various problems.  One such problem is that the Alpha has both VAX and
1580IEEE floating types.  One can easily imagine one library using the VAX
1581types and another library in the same executable using the IEEE types.
1582Another example is that the interpretation of whether a boolean is true
1583or false can be based on the least significant bit, most significant
1584bit, whether it is zero, etc., and different compilers (or different
1585options to the same compiler) might provide different kinds of boolean.
1586
1587The last major issue is the names of the types.  The name of a given
1588type depends @emph{only} on the negative type number given; these do not
1589vary depending on the language, the target system, or anything else.
1590One can always define separate type numbers---in the following list you
1591will see for example separate @code{int} and @code{integer*4} types
1592which are identical except for the name.  But compatibility can be
1593maintained by not inventing new negative type numbers and instead just
1594defining a new type with a new name.  For example:
1595
1596@example
1597.stabs "CARDINAL:t10=-8",128,0,0,0
1598@end example
1599
1600Here is the list of negative type numbers.  The phrase @dfn{integral
1601type} is used to mean twos-complement (I strongly suspect that all
1602machines which use stabs use twos-complement; most machines use
1603twos-complement these days).
1604
1605@table @code
1606@item -1
1607@code{int}, 32 bit signed integral type.
1608
1609@item -2
1610@code{char}, 8 bit type holding a character.   Both GDB and dbx on AIX
1611treat this as signed.  GCC uses this type whether @code{char} is signed
1612or not, which seems like a bad idea.  The AIX compiler (@code{xlc}) seems to
1613avoid this type; it uses -5 instead for @code{char}.
1614
1615@item -3
1616@code{short}, 16 bit signed integral type.
1617
1618@item -4
1619@code{long}, 32 bit signed integral type.
1620
1621@item -5
1622@code{unsigned char}, 8 bit unsigned integral type.
1623
1624@item -6
1625@code{signed char}, 8 bit signed integral type.
1626
1627@item -7
1628@code{unsigned short}, 16 bit unsigned integral type.
1629
1630@item -8
1631@code{unsigned int}, 32 bit unsigned integral type.
1632
1633@item -9
1634@code{unsigned}, 32 bit unsigned integral type.
1635
1636@item -10
1637@code{unsigned long}, 32 bit unsigned integral type.
1638
1639@item -11
1640@code{void}, type indicating the lack of a value.
1641
1642@item -12
1643@code{float}, IEEE single precision.
1644
1645@item -13
1646@code{double}, IEEE double precision.
1647
1648@item -14
1649@code{long double}, IEEE double precision.  The compiler claims the size
1650will increase in a future release, and for binary compatibility you have
1651to avoid using @code{long double}.  I hope when they increase it they
1652use a new negative type number.
1653
1654@item -15
1655@code{integer}.  32 bit signed integral type.
1656
1657@item -16
1658@code{boolean}.  32 bit type.  GDB and GCC assume that zero is false,
1659one is true, and other values have unspecified meaning.  I hope this
1660agrees with how the IBM tools use the type.
1661
1662@item -17
1663@code{short real}.  IEEE single precision.
1664
1665@item -18
1666@code{real}.  IEEE double precision.
1667
1668@item -19
1669@code{stringptr}.  @xref{Strings}.
1670
1671@item -20
1672@code{character}, 8 bit unsigned character type.
1673
1674@item -21
1675@code{logical*1}, 8 bit type.  This Fortran type has a split
1676personality in that it is used for boolean variables, but can also be
1677used for unsigned integers.  0 is false, 1 is true, and other values are
1678non-boolean.
1679
1680@item -22
1681@code{logical*2}, 16 bit type.  This Fortran type has a split
1682personality in that it is used for boolean variables, but can also be
1683used for unsigned integers.  0 is false, 1 is true, and other values are
1684non-boolean.
1685
1686@item -23
1687@code{logical*4}, 32 bit type.  This Fortran type has a split
1688personality in that it is used for boolean variables, but can also be
1689used for unsigned integers.  0 is false, 1 is true, and other values are
1690non-boolean.
1691
1692@item -24
1693@code{logical}, 32 bit type.  This Fortran type has a split
1694personality in that it is used for boolean variables, but can also be
1695used for unsigned integers.  0 is false, 1 is true, and other values are
1696non-boolean.
1697
1698@item -25
1699@code{complex}.  A complex type consisting of two IEEE single-precision
1700floating point values.
1701
1702@item -26
1703@code{complex}.  A complex type consisting of two IEEE double-precision
1704floating point values.
1705
1706@item -27
1707@code{integer*1}, 8 bit signed integral type.
1708
1709@item -28
1710@code{integer*2}, 16 bit signed integral type.
1711
1712@item -29
1713@code{integer*4}, 32 bit signed integral type.
1714
1715@item -30
1716@code{wchar}.  Wide character, 16 bits wide, unsigned (what format?
1717Unicode?).
1718
1719@item -31
1720@code{long long}, 64 bit signed integral type.
1721
1722@item -32
1723@code{unsigned long long}, 64 bit unsigned integral type.
1724
1725@item -33
1726@code{logical*8}, 64 bit unsigned integral type.
1727
1728@item -34
1729@code{integer*8}, 64 bit signed integral type.
1730@end table
1731
1732@node Miscellaneous Types
1733@section Miscellaneous Types
1734
1735@table @code
1736@item b @var{type-information} ; @var{bytes}
1737Pascal space type.  This is documented by IBM; what does it mean?
1738
1739This use of the @samp{b} type descriptor can be distinguished
1740from its use for builtin integral types (@pxref{Builtin Type
1741Descriptors}) because the character following the type descriptor is
1742always a digit, @samp{(}, or @samp{-}.
1743
1744@item B @var{type-information}
1745A volatile-qualified version of @var{type-information}.  This is
1746a Sun extension.  References and stores to a variable with a
1747volatile-qualified type must not be optimized or cached; they
1748must occur as the user specifies them.
1749
1750@item d @var{type-information}
1751File of type @var{type-information}.  As far as I know this is only used
1752by Pascal.
1753
1754@item k @var{type-information}
1755A const-qualified version of @var{type-information}.  This is a Sun
1756extension.  A variable with a const-qualified type cannot be modified.
1757
1758@item M @var{type-information} ; @var{length}
1759Multiple instance type.  The type seems to composed of @var{length}
1760repetitions of @var{type-information}, for example @code{character*3} is
1761represented by @samp{M-2;3}, where @samp{-2} is a reference to a
1762character type (@pxref{Negative Type Numbers}).  I'm not sure how this
1763differs from an array.  This appears to be a Fortran feature.
1764@var{length} is a bound, like those in range types; see @ref{Subranges}.
1765
1766@item S @var{type-information}
1767Pascal set type.  @var{type-information} must be a small type such as an
1768enumeration or a subrange, and the type is a bitmask whose length is
1769specified by the number of elements in @var{type-information}.
1770
1771In CHILL, if it is a bitstring instead of a set, also use the @samp{S}
1772type attribute (@pxref{String Field}).
1773
1774@item * @var{type-information}
1775Pointer to @var{type-information}.
1776@end table
1777
1778@node Cross-References
1779@section Cross-References to Other Types
1780
1781A type can be used before it is defined; one common way to deal with
1782that situation is just to use a type reference to a type which has not
1783yet been defined.
1784
1785Another way is with the @samp{x} type descriptor, which is followed by
1786@samp{s} for a structure tag, @samp{u} for a union tag, or @samp{e} for
1787a enumerator tag, followed by the name of the tag, followed by @samp{:}.
1788If the name contains @samp{::} between a @samp{<} and @samp{>} pair (for
1789C@t{++} templates), such a @samp{::} does not end the name---only a single
1790@samp{:} ends the name; see @ref{Nested Symbols}.
1791
1792For example, the following C declarations:
1793
1794@example
1795struct foo;
1796struct foo *bar;
1797@end example
1798
1799@noindent
1800produce:
1801
1802@example
1803.stabs "bar:G16=*17=xsfoo:",32,0,0,0
1804@end example
1805
1806Not all debuggers support the @samp{x} type descriptor, so on some
1807machines GCC does not use it.  I believe that for the above example it
1808would just emit a reference to type 17 and never define it, but I
1809haven't verified that.
1810
1811Modula-2 imported types, at least on AIX, use the @samp{i} type
1812descriptor, which is followed by the name of the module from which the
1813type is imported, followed by @samp{:}, followed by the name of the
1814type.  There is then optionally a comma followed by type information for
1815the type.  This differs from merely naming the type (@pxref{Typedefs}) in
1816that it identifies the module; I don't understand whether the name of
1817the type given here is always just the same as the name we are giving
1818it, or whether this type descriptor is used with a nameless stab
1819(@pxref{String Field}), or what.  The symbol ends with @samp{;}.
1820
1821@node Subranges
1822@section Subrange Types
1823
1824The @samp{r} type descriptor defines a type as a subrange of another
1825type.  It is followed by type information for the type of which it is a
1826subrange, a semicolon, an integral lower bound, a semicolon, an
1827integral upper bound, and a semicolon.  The AIX documentation does not
1828specify the trailing semicolon, in an effort to specify array indexes
1829more cleanly, but a subrange which is not an array index has always
1830included a trailing semicolon (@pxref{Arrays}).
1831
1832Instead of an integer, either bound can be one of the following:
1833
1834@table @code
1835@item A @var{offset}
1836The bound is passed by reference on the stack at offset @var{offset}
1837from the argument list.  @xref{Parameters}, for more information on such
1838offsets.
1839
1840@item T @var{offset}
1841The bound is passed by value on the stack at offset @var{offset} from
1842the argument list.
1843
1844@item a @var{register-number}
1845The bound is passed by reference in register number
1846@var{register-number}.
1847
1848@item t @var{register-number}
1849The bound is passed by value in register number @var{register-number}.
1850
1851@item J
1852There is no bound.
1853@end table
1854
1855Subranges are also used for builtin types; see @ref{Traditional Builtin Types}.
1856
1857@node Arrays
1858@section Array Types
1859
1860Arrays use the @samp{a} type descriptor.  Following the type descriptor
1861is the type of the index and the type of the array elements.  If the
1862index type is a range type, it ends in a semicolon; otherwise
1863(for example, if it is a type reference), there does not
1864appear to be any way to tell where the types are separated.  In an
1865effort to clean up this mess, IBM documents the two types as being
1866separated by a semicolon, and a range type as not ending in a semicolon
1867(but this is not right for range types which are not array indexes,
1868@pxref{Subranges}).  I think probably the best solution is to specify
1869that a semicolon ends a range type, and that the index type and element
1870type of an array are separated by a semicolon, but that if the index
1871type is a range type, the extra semicolon can be omitted.  GDB (at least
1872through version 4.9) doesn't support any kind of index type other than a
1873range anyway; I'm not sure about dbx.
1874
1875It is well established, and widely used, that the type of the index,
1876unlike most types found in the stabs, is merely a type definition, not
1877type information (@pxref{String Field}) (that is, it need not start with
1878@samp{@var{type-number}=} if it is defining a new type).  According to a
1879comment in GDB, this is also true of the type of the array elements; it
1880gives @samp{ar1;1;10;ar1;1;10;4} as a legitimate way to express a two
1881dimensional array.  According to AIX documentation, the element type
1882must be type information.  GDB accepts either.
1883
1884The type of the index is often a range type, expressed as the type
1885descriptor @samp{r} and some parameters.  It defines the size of the
1886array.  In the example below, the range @samp{r1;0;2;} defines an index
1887type which is a subrange of type 1 (integer), with a lower bound of 0
1888and an upper bound of 2.  This defines the valid range of subscripts of
1889a three-element C array.
1890
1891For example, the definition:
1892
1893@example
1894char char_vec[3] = @{'a','b','c'@};
1895@end example
1896
1897@noindent
1898produces the output:
1899
1900@example
1901.stabs "char_vec:G19=ar1;0;2;2",32,0,0,0
1902     .global _char_vec
1903     .align 4
1904_char_vec:
1905     .byte 97
1906     .byte 98
1907     .byte 99
1908@end example
1909
1910If an array is @dfn{packed}, the elements are spaced more
1911closely than normal, saving memory at the expense of speed.  For
1912example, an array of 3-byte objects might, if unpacked, have each
1913element aligned on a 4-byte boundary, but if packed, have no padding.
1914One way to specify that something is packed is with type attributes
1915(@pxref{String Field}).  In the case of arrays, another is to use the
1916@samp{P} type descriptor instead of @samp{a}.  Other than specifying a
1917packed array, @samp{P} is identical to @samp{a}.
1918
1919@c FIXME-what is it?  A pointer?
1920An open array is represented by the @samp{A} type descriptor followed by
1921type information specifying the type of the array elements.
1922
1923@c FIXME: what is the format of this type?  A pointer to a vector of pointers?
1924An N-dimensional dynamic array is represented by
1925
1926@example
1927D @var{dimensions} ; @var{type-information}
1928@end example
1929
1930@c Does dimensions really have this meaning?  The AIX documentation
1931@c doesn't say.
1932@var{dimensions} is the number of dimensions; @var{type-information}
1933specifies the type of the array elements.
1934
1935@c FIXME: what is the format of this type?  A pointer to some offsets in
1936@c another array?
1937A subarray of an N-dimensional array is represented by
1938
1939@example
1940E @var{dimensions} ; @var{type-information}
1941@end example
1942
1943@c Does dimensions really have this meaning?  The AIX documentation
1944@c doesn't say.
1945@var{dimensions} is the number of dimensions; @var{type-information}
1946specifies the type of the array elements.
1947
1948@node Strings
1949@section Strings
1950
1951Some languages, like C or the original Pascal, do not have string types,
1952they just have related things like arrays of characters.  But most
1953Pascals and various other languages have string types, which are
1954indicated as follows:
1955
1956@table @code
1957@item n @var{type-information} ; @var{bytes}
1958@var{bytes} is the maximum length.  I'm not sure what
1959@var{type-information} is; I suspect that it means that this is a string
1960of @var{type-information} (thus allowing a string of integers, a string
1961of wide characters, etc., as well as a string of characters).  Not sure
1962what the format of this type is.  This is an AIX feature.
1963
1964@item z @var{type-information} ; @var{bytes}
1965Just like @samp{n} except that this is a gstring, not an ordinary
1966string.  I don't know the difference.
1967
1968@item N
1969Pascal Stringptr.  What is this?  This is an AIX feature.
1970@end table
1971
1972Languages, such as CHILL which have a string type which is basically
1973just an array of characters use the @samp{S} type attribute
1974(@pxref{String Field}).
1975
1976@node Enumerations
1977@section Enumerations
1978
1979Enumerations are defined with the @samp{e} type descriptor.
1980
1981@c FIXME: Where does this information properly go?  Perhaps it is
1982@c redundant with something we already explain.
1983The source line below declares an enumeration type at file scope.
1984The type definition is located after the @code{N_RBRAC} that marks the end of
1985the previous procedure's block scope, and before the @code{N_FUN} that marks
1986the beginning of the next procedure's block scope.  Therefore it does not
1987describe a block local symbol, but a file local one.
1988
1989The source line:
1990
1991@example
1992enum e_places @{first,second=3,last@};
1993@end example
1994
1995@noindent
1996generates the following stab:
1997
1998@example
1999.stabs "e_places:T22=efirst:0,second:3,last:4,;",128,0,0,0
2000@end example
2001
2002The symbol descriptor (@samp{T}) says that the stab describes a
2003structure, enumeration, or union tag.  The type descriptor @samp{e},
2004following the @samp{22=} of the type definition narrows it down to an
2005enumeration type.  Following the @samp{e} is a list of the elements of
2006the enumeration.  The format is @samp{@var{name}:@var{value},}.  The
2007list of elements ends with @samp{;}.  The fact that @var{value} is
2008specified as an integer can cause problems if the value is large.  GCC
20092.5.2 tries to output it in octal in that case with a leading zero,
2010which is probably a good thing, although GDB 4.11 supports octal only in
2011cases where decimal is perfectly good.  Negative decimal values are
2012supported by both GDB and dbx.
2013
2014There is no standard way to specify the size of an enumeration type; it
2015is determined by the architecture (normally all enumerations types are
201632 bits).  Type attributes can be used to specify an enumeration type of
2017another size for debuggers which support them; see @ref{String Field}.
2018
2019Enumeration types are unusual in that they define symbols for the
2020enumeration values (@code{first}, @code{second}, and @code{third} in the
2021above example), and even though these symbols are visible in the file as
2022a whole (rather than being in a more local namespace like structure
2023member names), they are defined in the type definition for the
2024enumeration type rather than each having their own symbol.  In order to
2025be fast, GDB will only get symbols from such types (in its initial scan
2026of the stabs) if the type is the first thing defined after a @samp{T} or
2027@samp{t} symbol descriptor (the above example fulfills this
2028requirement).  If the type does not have a name, the compiler should
2029emit it in a nameless stab (@pxref{String Field}); GCC does this.
2030
2031@node Structures
2032@section Structures
2033
2034The encoding of structures in stabs can be shown with an example.
2035
2036The following source code declares a structure tag and defines an
2037instance of the structure in global scope. Then a @code{typedef} equates the
2038structure tag with a new type.  Separate stabs are generated for the
2039structure tag, the structure @code{typedef}, and the structure instance.  The
2040stabs for the tag and the @code{typedef} are emitted when the definitions are
2041encountered.  Since the structure elements are not initialized, the
2042stab and code for the structure variable itself is located at the end
2043of the program in the bss section.
2044
2045@example
2046struct s_tag @{
2047  int   s_int;
2048  float s_float;
2049  char  s_char_vec[8];
2050  struct s_tag* s_next;
2051@} g_an_s;
2052
2053typedef struct s_tag s_typedef;
2054@end example
2055
2056The structure tag has an @code{N_LSYM} stab type because, like the
2057enumeration, the symbol has file scope.  Like the enumeration, the
2058symbol descriptor is @samp{T}, for enumeration, structure, or tag type.
2059The type descriptor @samp{s} following the @samp{16=} of the type
2060definition narrows the symbol type to structure.
2061
2062Following the @samp{s} type descriptor is the number of bytes the
2063structure occupies, followed by a description of each structure element.
2064The structure element descriptions are of the form
2065@samp{@var{name}:@var{type}, @var{bit offset from the start of the
2066struct}, @var{number of bits in the element}}.
2067
2068@c FIXME: phony line break.  Can probably be fixed by using an example
2069@c with fewer fields.
2070@example
2071# @r{128 is N_LSYM}
2072.stabs "s_tag:T16=s20s_int:1,0,32;s_float:12,32,32;
2073        s_char_vec:17=ar1;0;7;2,64,64;s_next:18=*16,128,32;;",128,0,0,0
2074@end example
2075
2076In this example, the first two structure elements are previously defined
2077types.  For these, the type following the @samp{@var{name}:} part of the
2078element description is a simple type reference.  The other two structure
2079elements are new types.  In this case there is a type definition
2080embedded after the @samp{@var{name}:}.  The type definition for the
2081array element looks just like a type definition for a stand-alone array.
2082The @code{s_next} field is a pointer to the same kind of structure that
2083the field is an element of.  So the definition of structure type 16
2084contains a type definition for an element which is a pointer to type 16.
2085
2086If a field is a static member (this is a C@t{++} feature in which a single
2087variable appears to be a field of every structure of a given type) it
2088still starts out with the field name, a colon, and the type, but then
2089instead of a comma, bit position, comma, and bit size, there is a colon
2090followed by the name of the variable which each such field refers to.
2091
2092If the structure has methods (a C@t{++} feature), they follow the non-method
2093fields; see @ref{Cplusplus}.
2094
2095@node Typedefs
2096@section Giving a Type a Name
2097
2098@findex N_LSYM, for types
2099@findex C_DECL, for types
2100To give a type a name, use the @samp{t} symbol descriptor.  The type
2101is specified by the type information (@pxref{String Field}) for the stab.
2102For example,
2103
2104@example
2105.stabs "s_typedef:t16",128,0,0,0     # @r{128 is N_LSYM}
2106@end example
2107
2108specifies that @code{s_typedef} refers to type number 16.  Such stabs
2109have symbol type @code{N_LSYM} (or @code{C_DECL} for XCOFF).  (The Sun
2110documentation mentions using @code{N_GSYM} in some cases).
2111
2112If you are specifying the tag name for a structure, union, or
2113enumeration, use the @samp{T} symbol descriptor instead.  I believe C is
2114the only language with this feature.
2115
2116If the type is an opaque type (I believe this is a Modula-2 feature),
2117AIX provides a type descriptor to specify it.  The type descriptor is
2118@samp{o} and is followed by a name.  I don't know what the name
2119means---is it always the same as the name of the type, or is this type
2120descriptor used with a nameless stab (@pxref{String Field})?  There
2121optionally follows a comma followed by type information which defines
2122the type of this type.  If omitted, a semicolon is used in place of the
2123comma and the type information, and the type is much like a generic
2124pointer type---it has a known size but little else about it is
2125specified.
2126
2127@node Unions
2128@section Unions
2129
2130@example
2131union u_tag @{
2132  int  u_int;
2133  float u_float;
2134  char* u_char;
2135@} an_u;
2136@end example
2137
2138This code generates a stab for a union tag and a stab for a union
2139variable.  Both use the @code{N_LSYM} stab type.  If a union variable is
2140scoped locally to the procedure in which it is defined, its stab is
2141located immediately preceding the @code{N_LBRAC} for the procedure's block
2142start.
2143
2144The stab for the union tag, however, is located preceding the code for
2145the procedure in which it is defined.  The stab type is @code{N_LSYM}.  This
2146would seem to imply that the union type is file scope, like the struct
2147type @code{s_tag}.  This is not true.  The contents and position of the stab
2148for @code{u_type} do not convey any information about its procedure local
2149scope.
2150
2151@c FIXME: phony line break.  Can probably be fixed by using an example
2152@c with fewer fields.
2153@smallexample
2154# @r{128 is N_LSYM}
2155.stabs "u_tag:T23=u4u_int:1,0,32;u_float:12,0,32;u_char:21,0,32;;",
2156       128,0,0,0
2157@end smallexample
2158
2159The symbol descriptor @samp{T}, following the @samp{name:} means that
2160the stab describes an enumeration, structure, or union tag.  The type
2161descriptor @samp{u}, following the @samp{23=} of the type definition,
2162narrows it down to a union type definition.  Following the @samp{u} is
2163the number of bytes in the union.  After that is a list of union element
2164descriptions.  Their format is @samp{@var{name}:@var{type}, @var{bit
2165offset into the union}, @var{number of bytes for the element};}.
2166
2167The stab for the union variable is:
2168
2169@example
2170.stabs "an_u:23",128,0,0,-20     # @r{128 is N_LSYM}
2171@end example
2172
2173@samp{-20} specifies where the variable is stored (@pxref{Stack
2174Variables}).
2175
2176@node Function Types
2177@section Function Types
2178
2179Various types can be defined for function variables.  These types are
2180not used in defining functions (@pxref{Procedures}); they are used for
2181things like pointers to functions.
2182
2183The simple, traditional, type is type descriptor @samp{f} is followed by
2184type information for the return type of the function, followed by a
2185semicolon.
2186
2187This does not deal with functions for which the number and types of the
2188parameters are part of the type, as in Modula-2 or ANSI C.  AIX provides
2189extensions to specify these, using the @samp{f}, @samp{F}, @samp{p}, and
2190@samp{R} type descriptors.
2191
2192First comes the type descriptor.  If it is @samp{f} or @samp{F}, this
2193type involves a function rather than a procedure, and the type
2194information for the return type of the function follows, followed by a
2195comma.  Then comes the number of parameters to the function and a
2196semicolon.  Then, for each parameter, there is the name of the parameter
2197followed by a colon (this is only present for type descriptors @samp{R}
2198and @samp{F} which represent Pascal function or procedure parameters),
2199type information for the parameter, a comma, 0 if passed by reference or
22001 if passed by value, and a semicolon.  The type definition ends with a
2201semicolon.
2202
2203For example, this variable definition:
2204
2205@example
2206int (*g_pf)();
2207@end example
2208
2209@noindent
2210generates the following code:
2211
2212@example
2213.stabs "g_pf:G24=*25=f1",32,0,0,0
2214    .common _g_pf,4,"bss"
2215@end example
2216
2217The variable defines a new type, 24, which is a pointer to another new
2218type, 25, which is a function returning @code{int}.
2219
2220@node Macro define and undefine
2221@chapter Representation of #define and #undef
2222
2223This section describes the stabs support for macro define and undefine
2224information, supported on some systems.  (e.g., with @option{-g3}
2225@option{-gstabs} when using GCC).
2226
2227A @code{#define @var{macro-name} @var{macro-body}} is represented with
2228an @code{N_MAC_DEFINE} stab with a string field of
2229@code{@var{macro-name} @var{macro-body}}.
2230@findex N_MAC_DEFINE
2231
2232An @code{#undef @var{macro-name}} is represented with an
2233@code{N_MAC_UNDEF} stabs with a string field of simply
2234@code{@var{macro-name}}.
2235@findex N_MAC_UNDEF
2236
2237For both @code{N_MAC_DEFINE} and @code{N_MAC_UNDEF}, the desc field is
2238the line number within the file where the corresponding @code{#define}
2239or @code{#undef} occurred.
2240
2241For example, the following C code:
2242
2243@example
2244    #define NONE	42
2245    #define TWO(a, b)	(a + (a) + 2 * b)
2246    #define ONE(c)	(c + 19)
2247
2248    main(int argc, char *argv[])
2249    @{
2250      func(NONE, TWO(10, 11));
2251      func(NONE, ONE(23));
2252
2253    #undef ONE
2254    #define ONE(c)	(c + 23)
2255
2256      func(NONE, ONE(-23));
2257
2258      return (0);
2259    @}
2260
2261    int global;
2262
2263    func(int arg1, int arg2)
2264    @{
2265      global = arg1 + arg2;
2266    @}
2267@end example
2268
2269@noindent
2270produces the following stabs (as well as many others):
2271
2272@example
2273    .stabs	"NONE 42",54,0,1,0
2274    .stabs	"TWO(a,b) (a + (a) + 2 * b)",54,0,2,0
2275    .stabs	"ONE(c) (c + 19)",54,0,3,0
2276    .stabs	"ONE",58,0,10,0
2277    .stabs	"ONE(c) (c + 23)",54,0,11,0
2278@end example
2279
2280@noindent
2281NOTE: In the above example, @code{54} is @code{N_MAC_DEFINE} and
2282@code{58} is @code{N_MAC_UNDEF}.
2283
2284@node Symbol Tables
2285@chapter Symbol Information in Symbol Tables
2286
2287This chapter describes the format of symbol table entries
2288and how stab assembler directives map to them.  It also describes the
2289transformations that the assembler and linker make on data from stabs.
2290
2291@menu
2292* Symbol Table Format::
2293* Transformations On Symbol Tables::
2294@end menu
2295
2296@node Symbol Table Format
2297@section Symbol Table Format
2298
2299Each time the assembler encounters a stab directive, it puts
2300each field of the stab into a corresponding field in a symbol table
2301entry of its output file.  If the stab contains a string field, the
2302symbol table entry for that stab points to a string table entry
2303containing the string data from the stab.  Assembler labels become
2304relocatable addresses.  Symbol table entries in a.out have the format:
2305
2306@c FIXME: should refer to external, not internal.
2307@example
2308struct internal_nlist @{
2309  unsigned long n_strx;         /* index into string table of name */
2310  unsigned char n_type;         /* type of symbol */
2311  unsigned char n_other;        /* misc info (usually empty) */
2312  unsigned short n_desc;        /* description field */
2313  bfd_vma n_value;              /* value of symbol */
2314@};
2315@end example
2316
2317If the stab has a string, the @code{n_strx} field holds the offset in
2318bytes of the string within the string table.  The string is terminated
2319by a NUL character.  If the stab lacks a string (for example, it was
2320produced by a @code{.stabn} or @code{.stabd} directive), the
2321@code{n_strx} field is zero.
2322
2323Symbol table entries with @code{n_type} field values greater than 0x1f
2324originated as stabs generated by the compiler (with one random
2325exception).  The other entries were placed in the symbol table of the
2326executable by the assembler or the linker.
2327
2328@node Transformations On Symbol Tables
2329@section Transformations on Symbol Tables
2330
2331The linker concatenates object files and does fixups of externally
2332defined symbols.
2333
2334You can see the transformations made on stab data by the assembler and
2335linker by examining the symbol table after each pass of the build.  To
2336do this, use @samp{nm -ap}, which dumps the symbol table, including
2337debugging information, unsorted.  For stab entries the columns are:
2338@var{value}, @var{other}, @var{desc}, @var{type}, @var{string}.  For
2339assembler and linker symbols, the columns are: @var{value}, @var{type},
2340@var{string}.
2341
2342The low 5 bits of the stab type tell the linker how to relocate the
2343value of the stab.  Thus for stab types like @code{N_RSYM} and
2344@code{N_LSYM}, where the value is an offset or a register number, the
2345low 5 bits are @code{N_ABS}, which tells the linker not to relocate the
2346value.
2347
2348Where the value of a stab contains an assembly language label,
2349it is transformed by each build step.  The assembler turns it into a
2350relocatable address and the linker turns it into an absolute address.
2351
2352@menu
2353* Transformations On Static Variables::
2354* Transformations On Global Variables::
2355* Stab Section Transformations::	   For some object file formats,
2356                                           things are a bit different.
2357@end menu
2358
2359@node Transformations On Static Variables
2360@subsection Transformations on Static Variables
2361
2362This source line defines a static variable at file scope:
2363
2364@example
2365static int s_g_repeat
2366@end example
2367
2368@noindent
2369The following stab describes the symbol:
2370
2371@example
2372.stabs "s_g_repeat:S1",38,0,0,_s_g_repeat
2373@end example
2374
2375@noindent
2376The assembler transforms the stab into this symbol table entry in the
2377@file{.o} file.  The location is expressed as a data segment offset.
2378
2379@example
238000000084 - 00 0000 STSYM s_g_repeat:S1
2381@end example
2382
2383@noindent
2384In the symbol table entry from the executable, the linker has made the
2385relocatable address absolute.
2386
2387@example
23880000e00c - 00 0000 STSYM s_g_repeat:S1
2389@end example
2390
2391@node Transformations On Global Variables
2392@subsection Transformations on Global Variables
2393
2394Stabs for global variables do not contain location information. In
2395this case, the debugger finds location information in the assembler or
2396linker symbol table entry describing the variable.  The source line:
2397
2398@example
2399char g_foo = 'c';
2400@end example
2401
2402@noindent
2403generates the stab:
2404
2405@example
2406.stabs "g_foo:G2",32,0,0,0
2407@end example
2408
2409The variable is represented by two symbol table entries in the object
2410file (see below).  The first one originated as a stab.  The second one
2411is an external symbol.  The upper case @samp{D} signifies that the
2412@code{n_type} field of the symbol table contains 7, @code{N_DATA} with
2413local linkage.  The stab's value is zero since the value is not used for
2414@code{N_GSYM} stabs.  The value of the linker symbol is the relocatable
2415address corresponding to the variable.
2416
2417@example
241800000000 - 00 0000  GSYM g_foo:G2
241900000080 D _g_foo
2420@end example
2421
2422@noindent
2423These entries as transformed by the linker.  The linker symbol table
2424entry now holds an absolute address:
2425
2426@example
242700000000 - 00 0000  GSYM g_foo:G2
2428@dots{}
24290000e008 D _g_foo
2430@end example
2431
2432@node Stab Section Transformations
2433@subsection Transformations of Stabs in separate sections
2434
2435For object file formats using stabs in separate sections (@pxref{Stab
2436Sections}), use @code{objdump --stabs} instead of @code{nm} to show the
2437stabs in an object or executable file.  @code{objdump} is a GNU utility;
2438Sun does not provide any equivalent.
2439
2440The following example is for a stab whose value is an address is
2441relative to the compilation unit (@pxref{ELF Linker Relocation}).  For
2442example, if the source line
2443
2444@example
2445static int ld = 5;
2446@end example
2447
2448appears within a function, then the assembly language output from the
2449compiler contains:
2450
2451@example
2452.Ddata.data:
2453@dots{}
2454        .stabs "ld:V(0,3)",0x26,0,4,.L18-Ddata.data    # @r{0x26 is N_STSYM}
2455@dots{}
2456.L18:
2457        .align 4
2458        .word 0x5
2459@end example
2460
2461Because the value is formed by subtracting one symbol from another, the
2462value is absolute, not relocatable, and so the object file contains
2463
2464@example
2465Symnum n_type n_othr n_desc n_value  n_strx String
246631     STSYM  0      4      00000004 680    ld:V(0,3)
2467@end example
2468
2469without any relocations, and the executable file also contains
2470
2471@example
2472Symnum n_type n_othr n_desc n_value  n_strx String
247331     STSYM  0      4      00000004 680    ld:V(0,3)
2474@end example
2475
2476@node Cplusplus
2477@chapter GNU C@t{++} Stabs
2478
2479@menu
2480* Class Names::			C++ class names are both tags and typedefs.
2481* Nested Symbols::		C++ symbol names can be within other types.
2482* Basic Cplusplus Types::
2483* Simple Classes::
2484* Class Instance::
2485* Methods::			Method definition
2486* Method Type Descriptor::      The @samp{#} type descriptor
2487* Member Type Descriptor::      The @samp{@@} type descriptor
2488* Protections::
2489* Method Modifiers::
2490* Virtual Methods::
2491* Inheritance::
2492* Virtual Base Classes::
2493* Static Members::
2494@end menu
2495
2496@node Class Names
2497@section C@t{++} Class Names
2498
2499In C@t{++}, a class name which is declared with @code{class}, @code{struct},
2500or @code{union}, is not only a tag, as in C, but also a type name.  Thus
2501there should be stabs with both @samp{t} and @samp{T} symbol descriptors
2502(@pxref{Typedefs}).
2503
2504To save space, there is a special abbreviation for this case.  If the
2505@samp{T} symbol descriptor is followed by @samp{t}, then the stab
2506defines both a type name and a tag.
2507
2508For example, the C@t{++} code
2509
2510@example
2511struct foo @{int x;@};
2512@end example
2513
2514can be represented as either
2515
2516@example
2517.stabs "foo:T19=s4x:1,0,32;;",128,0,0,0       # @r{128 is N_LSYM}
2518.stabs "foo:t19",128,0,0,0
2519@end example
2520
2521or
2522
2523@example
2524.stabs "foo:Tt19=s4x:1,0,32;;",128,0,0,0
2525@end example
2526
2527@node Nested Symbols
2528@section Defining a Symbol Within Another Type
2529
2530In C@t{++}, a symbol (such as a type name) can be defined within another type.
2531@c FIXME: Needs example.
2532
2533In stabs, this is sometimes represented by making the name of a symbol
2534which contains @samp{::}.  Such a pair of colons does not end the name
2535of the symbol, the way a single colon would (@pxref{String Field}).  I'm
2536not sure how consistently used or well thought out this mechanism is.
2537So that a pair of colons in this position always has this meaning,
2538@samp{:} cannot be used as a symbol descriptor.
2539
2540For example, if the string for a stab is @samp{foo::bar::baz:t5=*6},
2541then @code{foo::bar::baz} is the name of the symbol, @samp{t} is the
2542symbol descriptor, and @samp{5=*6} is the type information.
2543
2544@node Basic Cplusplus Types
2545@section Basic Types For C@t{++}
2546
2547<< the examples that follow are based on a01.C >>
2548
2549
2550C@t{++} adds two more builtin types to the set defined for C.  These are
2551the unknown type and the vtable record type.  The unknown type, type
255216, is defined in terms of itself like the void type.
2553
2554The vtable record type, type 17, is defined as a structure type and
2555then as a structure tag.  The structure has four fields: delta, index,
2556pfn, and delta2.  pfn is the function pointer.
2557
2558<< In boilerplate $vtbl_ptr_type, what are the fields delta,
2559index, and delta2 used for? >>
2560
2561This basic type is present in all C@t{++} programs even if there are no
2562virtual methods defined.
2563
2564@display
2565.stabs "struct_name:sym_desc(type)type_def(17)=type_desc(struct)struct_bytes(8)
2566        elem_name(delta):type_ref(short int),bit_offset(0),field_bits(16);
2567        elem_name(index):type_ref(short int),bit_offset(16),field_bits(16);
2568        elem_name(pfn):type_def(18)=type_desc(ptr to)type_ref(void),
2569                                    bit_offset(32),field_bits(32);
2570        elem_name(delta2):type_def(short int);bit_offset(32),field_bits(16);;"
2571        N_LSYM, NIL, NIL
2572@end display
2573
2574@smallexample
2575.stabs "$vtbl_ptr_type:t17=s8
2576        delta:6,0,16;index:6,16,16;pfn:18=*15,32,32;delta2:6,32,16;;"
2577        ,128,0,0,0
2578@end smallexample
2579
2580@display
2581.stabs "name:sym_dec(struct tag)type_ref($vtbl_ptr_type)",N_LSYM,NIL,NIL,NIL
2582@end display
2583
2584@example
2585.stabs "$vtbl_ptr_type:T17",128,0,0,0
2586@end example
2587
2588@node Simple Classes
2589@section Simple Class Definition
2590
2591The stabs describing C@t{++} language features are an extension of the
2592stabs describing C.  Stabs representing C@t{++} class types elaborate
2593extensively on the stab format used to describe structure types in C.
2594Stabs representing class type variables look just like stabs
2595representing C language variables.
2596
2597Consider the following very simple class definition.
2598
2599@example
2600class baseA @{
2601public:
2602        int Adat;
2603        int Ameth(int in, char other);
2604@};
2605@end example
2606
2607The class @code{baseA} is represented by two stabs.  The first stab describes
2608the class as a structure type.  The second stab describes a structure
2609tag of the class type.  Both stabs are of stab type @code{N_LSYM}.  Since the
2610stab is not located between an @code{N_FUN} and an @code{N_LBRAC} stab this indicates
2611that the class is defined at file scope.  If it were, then the @code{N_LSYM}
2612would signify a local variable.
2613
2614A stab describing a C@t{++} class type is similar in format to a stab
2615describing a C struct, with each class member shown as a field in the
2616structure.  The part of the struct format describing fields is
2617expanded to include extra information relevant to C@t{++} class members.
2618In addition, if the class has multiple base classes or virtual
2619functions the struct format outside of the field parts is also
2620augmented.
2621
2622In this simple example the field part of the C@t{++} class stab
2623representing member data looks just like the field part of a C struct
2624stab.  The section on protections describes how its format is
2625sometimes extended for member data.
2626
2627The field part of a C@t{++} class stab representing a member function
2628differs substantially from the field part of a C struct stab.  It
2629still begins with @samp{name:} but then goes on to define a new type number
2630for the member function, describe its return type, its argument types,
2631its protection level, any qualifiers applied to the method definition,
2632and whether the method is virtual or not.  If the method is virtual
2633then the method description goes on to give the vtable index of the
2634method, and the type number of the first base class defining the
2635method.
2636
2637When the field name is a method name it is followed by two colons rather
2638than one.  This is followed by a new type definition for the method.
2639This is a number followed by an equal sign and the type of the method.
2640Normally this will be a type declared using the @samp{#} type
2641descriptor; see @ref{Method Type Descriptor}; static member functions
2642are declared using the @samp{f} type descriptor instead; see
2643@ref{Function Types}.
2644
2645The format of an overloaded operator method name differs from that of
2646other methods.  It is @samp{op$::@var{operator-name}.} where
2647@var{operator-name} is the operator name such as @samp{+} or @samp{+=}.
2648The name ends with a period, and any characters except the period can
2649occur in the @var{operator-name} string.
2650
2651The next part of the method description represents the arguments to the
2652method, preceded by a colon and ending with a semi-colon.  The types of
2653the arguments are expressed in the same way argument types are expressed
2654in C@t{++} name mangling.  In this example an @code{int} and a @code{char}
2655map to @samp{ic}.
2656
2657This is followed by a number, a letter, and an asterisk or period,
2658followed by another semicolon.  The number indicates the protections
2659that apply to the member function.  Here the 2 means public.  The
2660letter encodes any qualifier applied to the method definition.  In
2661this case, @samp{A} means that it is a normal function definition.  The dot
2662shows that the method is not virtual.  The sections that follow
2663elaborate further on these fields and describe the additional
2664information present for virtual methods.
2665
2666
2667@display
2668.stabs "class_name:sym_desc(type)type_def(20)=type_desc(struct)struct_bytes(4)
2669        field_name(Adat):type(int),bit_offset(0),field_bits(32);
2670
2671        method_name(Ameth)::type_def(21)=type_desc(method)return_type(int);
2672        :arg_types(int char);
2673        protection(public)qualifier(normal)virtual(no);;"
2674        N_LSYM,NIL,NIL,NIL
2675@end display
2676
2677@smallexample
2678.stabs "baseA:t20=s4Adat:1,0,32;Ameth::21=##1;:ic;2A.;;",128,0,0,0
2679
2680.stabs "class_name:sym_desc(struct tag)",N_LSYM,NIL,NIL,NIL
2681
2682.stabs "baseA:T20",128,0,0,0
2683@end smallexample
2684
2685@node Class Instance
2686@section Class Instance
2687
2688As shown above, describing even a simple C@t{++} class definition is
2689accomplished by massively extending the stab format used in C to
2690describe structure types.  However, once the class is defined, C stabs
2691with no modifications can be used to describe class instances.  The
2692following source:
2693
2694@example
2695main () @{
2696        baseA AbaseA;
2697@}
2698@end example
2699
2700@noindent
2701yields the following stab describing the class instance.  It looks no
2702different from a standard C stab describing a local variable.
2703
2704@display
2705.stabs "name:type_ref(baseA)", N_LSYM, NIL, NIL, frame_ptr_offset
2706@end display
2707
2708@example
2709.stabs "AbaseA:20",128,0,0,-20
2710@end example
2711
2712@node Methods
2713@section Method Definition
2714
2715The class definition shown above declares Ameth.  The C@t{++} source below
2716defines Ameth:
2717
2718@example
2719int
2720baseA::Ameth(int in, char other)
2721@{
2722        return in;
2723@};
2724@end example
2725
2726
2727This method definition yields three stabs following the code of the
2728method.  One stab describes the method itself and following two describe
2729its parameters.  Although there is only one formal argument all methods
2730have an implicit argument which is the @code{this} pointer.  The @code{this}
2731pointer is a pointer to the object on which the method was called.  Note
2732that the method name is mangled to encode the class name and argument
2733types.  Name mangling is described in the @sc{arm} (@cite{The Annotated
2734C++ Reference Manual}, by Ellis and Stroustrup, @sc{isbn}
27350-201-51459-1); @file{gpcompare.texi} in Cygnus GCC distributions
2736describes the differences between GNU mangling and @sc{arm}
2737mangling.
2738@c FIXME: Use @xref, especially if this is generally installed in the
2739@c info tree.
2740@c FIXME: This information should be in a net release, either of GCC or
2741@c GDB.  But gpcompare.texi doesn't seem to be in the FSF GCC.
2742
2743@example
2744.stabs "name:symbol_descriptor(global function)return_type(int)",
2745        N_FUN, NIL, NIL, code_addr_of_method_start
2746
2747.stabs "Ameth__5baseAic:F1",36,0,0,_Ameth__5baseAic
2748@end example
2749
2750Here is the stab for the @code{this} pointer implicit argument.  The
2751name of the @code{this} pointer is always @code{this}.  Type 19, the
2752@code{this} pointer is defined as a pointer to type 20, @code{baseA},
2753but a stab defining @code{baseA} has not yet been emitted.  Since the
2754compiler knows it will be emitted shortly, here it just outputs a cross
2755reference to the undefined symbol, by prefixing the symbol name with
2756@samp{xs}.
2757
2758@example
2759.stabs "name:sym_desc(register param)type_def(19)=
2760        type_desc(ptr to)type_ref(baseA)=
2761        type_desc(cross-reference to)baseA:",N_RSYM,NIL,NIL,register_number
2762
2763.stabs "this:P19=*20=xsbaseA:",64,0,0,8
2764@end example
2765
2766The stab for the explicit integer argument looks just like a parameter
2767to a C function.  The last field of the stab is the offset from the
2768argument pointer, which in most systems is the same as the frame
2769pointer.
2770
2771@example
2772.stabs "name:sym_desc(value parameter)type_ref(int)",
2773        N_PSYM,NIL,NIL,offset_from_arg_ptr
2774
2775.stabs "in:p1",160,0,0,72
2776@end example
2777
2778<< The examples that follow are based on A1.C >>
2779
2780@node Method Type Descriptor
2781@section The @samp{#} Type Descriptor
2782
2783This is used to describe a class method.  This is a function which takes
2784an extra argument as its first argument, for the @code{this} pointer.
2785
2786If the @samp{#} is immediately followed by another @samp{#}, the second
2787one will be followed by the return type and a semicolon.  The class and
2788argument types are not specified, and must be determined by demangling
2789the name of the method if it is available.
2790
2791Otherwise, the single @samp{#} is followed by the class type, a comma,
2792the return type, a comma, and zero or more parameter types separated by
2793commas.  The list of arguments is terminated by a semicolon.  In the
2794debugging output generated by gcc, a final argument type of @code{void}
2795indicates a method which does not take a variable number of arguments.
2796If the final argument type of @code{void} does not appear, the method
2797was declared with an ellipsis.
2798
2799Note that although such a type will normally be used to describe fields
2800in structures, unions, or classes, for at least some versions of the
2801compiler it can also be used in other contexts.
2802
2803@node Member Type Descriptor
2804@section The @samp{@@} Type Descriptor
2805
2806The @samp{@@} type descriptor is used for a
2807pointer-to-non-static-member-data type.  It is followed
2808by type information for the class (or union), a comma, and type
2809information for the member data.
2810
2811The following C@t{++} source:
2812
2813@smallexample
2814typedef int A::*int_in_a;
2815@end smallexample
2816
2817generates the following stab:
2818
2819@smallexample
2820.stabs "int_in_a:t20=21=@@19,1",128,0,0,0
2821@end smallexample
2822
2823Note that there is a conflict between this and type attributes
2824(@pxref{String Field}); both use type descriptor @samp{@@}.
2825Fortunately, the @samp{@@} type descriptor used in this C@t{++} sense always
2826will be followed by a digit, @samp{(}, or @samp{-}, and type attributes
2827never start with those things.
2828
2829@node Protections
2830@section Protections
2831
2832In the simple class definition shown above all member data and
2833functions were publicly accessible.  The example that follows
2834contrasts public, protected and privately accessible fields and shows
2835how these protections are encoded in C@t{++} stabs.
2836
2837If the character following the @samp{@var{field-name}:} part of the
2838string is @samp{/}, then the next character is the visibility.  @samp{0}
2839means private, @samp{1} means protected, and @samp{2} means public.
2840Debuggers should ignore visibility characters they do not recognize, and
2841assume a reasonable default (such as public) (GDB 4.11 does not, but
2842this should be fixed in the next GDB release).  If no visibility is
2843specified the field is public.  The visibility @samp{9} means that the
2844field has been optimized out and is public (there is no way to specify
2845an optimized out field with a private or protected visibility).
2846Visibility @samp{9} is not supported by GDB 4.11; this should be fixed
2847in the next GDB release.
2848
2849The following C@t{++} source:
2850
2851@example
2852class vis @{
2853private:
2854        int   priv;
2855protected:
2856        char  prot;
2857public:
2858        float pub;
2859@};
2860@end example
2861
2862@noindent
2863generates the following stab:
2864
2865@example
2866# @r{128 is N_LSYM}
2867.stabs "vis:T19=s12priv:/01,0,32;prot:/12,32,8;pub:12,64,32;;",128,0,0,0
2868@end example
2869
2870@samp{vis:T19=s12} indicates that type number 19 is a 12 byte structure
2871named @code{vis} The @code{priv} field has public visibility
2872(@samp{/0}), type int (@samp{1}), and offset and size @samp{,0,32;}.
2873The @code{prot} field has protected visibility (@samp{/1}), type char
2874(@samp{2}) and offset and size @samp{,32,8;}.  The @code{pub} field has
2875type float (@samp{12}), and offset and size @samp{,64,32;}.
2876
2877Protections for member functions are signified by one digit embedded in
2878the field part of the stab describing the method.  The digit is 0 if
2879private, 1 if protected and 2 if public.  Consider the C@t{++} class
2880definition below:
2881
2882@example
2883class all_methods @{
2884private:
2885        int   priv_meth(int in)@{return in;@};
2886protected:
2887        char  protMeth(char in)@{return in;@};
2888public:
2889        float pubMeth(float in)@{return in;@};
2890@};
2891@end example
2892
2893It generates the following stab.  The digit in question is to the left
2894of an @samp{A} in each case.  Notice also that in this case two symbol
2895descriptors apply to the class name struct tag and struct type.
2896
2897@display
2898.stabs "class_name:sym_desc(struct tag&type)type_def(21)=
2899        sym_desc(struct)struct_bytes(1)
2900        meth_name::type_def(22)=sym_desc(method)returning(int);
2901        :args(int);protection(private)modifier(normal)virtual(no);
2902        meth_name::type_def(23)=sym_desc(method)returning(char);
2903        :args(char);protection(protected)modifier(normal)virtual(no);
2904        meth_name::type_def(24)=sym_desc(method)returning(float);
2905        :args(float);protection(public)modifier(normal)virtual(no);;",
2906        N_LSYM,NIL,NIL,NIL
2907@end display
2908
2909@smallexample
2910.stabs "all_methods:Tt21=s1priv_meth::22=##1;:i;0A.;protMeth::23=##2;:c;1A.;
2911        pubMeth::24=##12;:f;2A.;;",128,0,0,0
2912@end smallexample
2913
2914@node Method Modifiers
2915@section Method Modifiers (@code{const}, @code{volatile}, @code{const volatile})
2916
2917<< based on a6.C >>
2918
2919In the class example described above all the methods have the normal
2920modifier.  This method modifier information is located just after the
2921protection information for the method.  This field has four possible
2922character values.  Normal methods use @samp{A}, const methods use
2923@samp{B}, volatile methods use @samp{C}, and const volatile methods use
2924@samp{D}.  Consider the class definition below:
2925
2926@example
2927class A @{
2928public:
2929        int ConstMeth (int arg) const @{ return arg; @};
2930        char VolatileMeth (char arg) volatile @{ return arg; @};
2931        float ConstVolMeth (float arg) const volatile @{return arg; @};
2932@};
2933@end example
2934
2935This class is described by the following stab:
2936
2937@display
2938.stabs "class(A):sym_desc(struct)type_def(20)=type_desc(struct)struct_bytes(1)
2939        meth_name(ConstMeth)::type_def(21)sym_desc(method)
2940        returning(int);:arg(int);protection(public)modifier(const)virtual(no);
2941        meth_name(VolatileMeth)::type_def(22)=sym_desc(method)
2942        returning(char);:arg(char);protection(public)modifier(volatile)virt(no)
2943        meth_name(ConstVolMeth)::type_def(23)=sym_desc(method)
2944        returning(float);:arg(float);protection(public)modifier(const volatile)
2945        virtual(no);;", @dots{}
2946@end display
2947
2948@example
2949.stabs "A:T20=s1ConstMeth::21=##1;:i;2B.;VolatileMeth::22=##2;:c;2C.;
2950             ConstVolMeth::23=##12;:f;2D.;;",128,0,0,0
2951@end example
2952
2953@node Virtual Methods
2954@section Virtual Methods
2955
2956<< The following examples are based on a4.C >>
2957
2958The presence of virtual methods in a class definition adds additional
2959data to the class description.  The extra data is appended to the
2960description of the virtual method and to the end of the class
2961description.  Consider the class definition below:
2962
2963@example
2964class A @{
2965public:
2966        int Adat;
2967        virtual int A_virt (int arg) @{ return arg; @};
2968@};
2969@end example
2970
2971This results in the stab below describing class A.  It defines a new
2972type (20) which is an 8 byte structure.  The first field of the class
2973struct is @samp{Adat}, an integer, starting at structure offset 0 and
2974occupying 32 bits.
2975
2976The second field in the class struct is not explicitly defined by the
2977C@t{++} class definition but is implied by the fact that the class
2978contains a virtual method.  This field is the vtable pointer.  The
2979name of the vtable pointer field starts with @samp{$vf} and continues with a
2980type reference to the class it is part of.  In this example the type
2981reference for class A is 20 so the name of its vtable pointer field is
2982@samp{$vf20}, followed by the usual colon.
2983
2984Next there is a type definition for the vtable pointer type (21).
2985This is in turn defined as a pointer to another new type (22).
2986
2987Type 22 is the vtable itself, which is defined as an array, indexed by
2988a range of integers between 0 and 1, and whose elements are of type
298917.  Type 17 was the vtable record type defined by the boilerplate C@t{++}
2990type definitions, as shown earlier.
2991
2992The bit offset of the vtable pointer field is 32.  The number of bits
2993in the field are not specified when the field is a vtable pointer.
2994
2995Next is the method definition for the virtual member function @code{A_virt}.
2996Its description starts out using the same format as the non-virtual
2997member functions described above, except instead of a dot after the
2998@samp{A} there is an asterisk, indicating that the function is virtual.
2999Since is is virtual some addition information is appended to the end
3000of the method description.
3001
3002The first number represents the vtable index of the method.  This is a
300332 bit unsigned number with the high bit set, followed by a
3004semi-colon.
3005
3006The second number is a type reference to the first base class in the
3007inheritance hierarchy defining the virtual member function.  In this
3008case the class stab describes a base class so the virtual function is
3009not overriding any other definition of the method.  Therefore the
3010reference is to the type number of the class that the stab is
3011describing (20).
3012
3013This is followed by three semi-colons.  One marks the end of the
3014current sub-section, one marks the end of the method field, and the
3015third marks the end of the struct definition.
3016
3017For classes containing virtual functions the very last section of the
3018string part of the stab holds a type reference to the first base
3019class.  This is preceded by @samp{~%} and followed by a final semi-colon.
3020
3021@display
3022.stabs "class_name(A):type_def(20)=sym_desc(struct)struct_bytes(8)
3023        field_name(Adat):type_ref(int),bit_offset(0),field_bits(32);
3024        field_name(A virt func ptr):type_def(21)=type_desc(ptr to)type_def(22)=
3025        sym_desc(array)index_type_ref(range of int from 0 to 1);
3026        elem_type_ref(vtbl elem type),
3027        bit_offset(32);
3028        meth_name(A_virt)::typedef(23)=sym_desc(method)returning(int);
3029        :arg_type(int),protection(public)normal(yes)virtual(yes)
3030        vtable_index(1);class_first_defining(A);;;~%first_base(A);",
3031        N_LSYM,NIL,NIL,NIL
3032@end display
3033
3034@c FIXME: bogus line break.
3035@example
3036.stabs "A:t20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
3037        A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
3038@end example
3039
3040@node Inheritance
3041@section Inheritance
3042
3043Stabs describing C@t{++} derived classes include additional sections that
3044describe the inheritance hierarchy of the class.  A derived class stab
3045also encodes the number of base classes.  For each base class it tells
3046if the base class is virtual or not, and if the inheritance is private
3047or public.  It also gives the offset into the object of the portion of
3048the object corresponding to each base class.
3049
3050This additional information is embedded in the class stab following the
3051number of bytes in the struct.  First the number of base classes
3052appears bracketed by an exclamation point and a comma.
3053
3054Then for each base type there repeats a series: a virtual character, a
3055visibility character, a number, a comma, another number, and a
3056semi-colon.
3057
3058The virtual character is @samp{1} if the base class is virtual and
3059@samp{0} if not.  The visibility character is @samp{2} if the derivation
3060is public, @samp{1} if it is protected, and @samp{0} if it is private.
3061Debuggers should ignore virtual or visibility characters they do not
3062recognize, and assume a reasonable default (such as public and
3063non-virtual) (GDB 4.11 does not, but this should be fixed in the next
3064GDB release).
3065
3066The number following the virtual and visibility characters is the offset
3067from the start of the object to the part of the object pertaining to the
3068base class.
3069
3070After the comma, the second number is a type_descriptor for the base
3071type.  Finally a semi-colon ends the series, which repeats for each
3072base class.
3073
3074The source below defines three base classes @code{A}, @code{B}, and
3075@code{C} and the derived class @code{D}.
3076
3077
3078@example
3079class A @{
3080public:
3081        int Adat;
3082        virtual int A_virt (int arg) @{ return arg; @};
3083@};
3084
3085class B @{
3086public:
3087        int B_dat;
3088        virtual int B_virt (int arg) @{return arg; @};
3089@};
3090
3091class C @{
3092public:
3093        int Cdat;
3094        virtual int C_virt (int arg) @{return arg; @};
3095@};
3096
3097class D : A, virtual B, public C @{
3098public:
3099        int Ddat;
3100        virtual int A_virt (int arg ) @{ return arg+1; @};
3101        virtual int B_virt (int arg)  @{ return arg+2; @};
3102        virtual int C_virt (int arg)  @{ return arg+3; @};
3103        virtual int D_virt (int arg)  @{ return arg; @};
3104@};
3105@end example
3106
3107Class stabs similar to the ones described earlier are generated for
3108each base class.
3109
3110@c FIXME!!! the linebreaks in the following example probably make the
3111@c examples literally unusable, but I don't know any other way to get
3112@c them on the page.
3113@c One solution would be to put some of the type definitions into
3114@c separate stabs, even if that's not exactly what the compiler actually
3115@c emits.
3116@smallexample
3117.stabs "A:T20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
3118        A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
3119
3120.stabs "B:Tt25=s8Bdat:1,0,32;$vf25:21,32;B_virt::26=##1;
3121        :i;2A*-2147483647;25;;;~%25;",128,0,0,0
3122
3123.stabs "C:Tt28=s8Cdat:1,0,32;$vf28:21,32;C_virt::29=##1;
3124        :i;2A*-2147483647;28;;;~%28;",128,0,0,0
3125@end smallexample
3126
3127In the stab describing derived class @code{D} below, the information about
3128the derivation of this class is encoded as follows.
3129
3130@display
3131.stabs "derived_class_name:symbol_descriptors(struct tag&type)=
3132        type_descriptor(struct)struct_bytes(32)!num_bases(3),
3133        base_virtual(no)inheritance_public(no)base_offset(0),
3134        base_class_type_ref(A);
3135        base_virtual(yes)inheritance_public(no)base_offset(NIL),
3136        base_class_type_ref(B);
3137        base_virtual(no)inheritance_public(yes)base_offset(64),
3138        base_class_type_ref(C); @dots{}
3139@end display
3140
3141@c FIXME! fake linebreaks.
3142@smallexample
3143.stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:
3144        1,160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt:
3145        :32:i;2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;
3146        28;;D_virt::32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
3147@end smallexample
3148
3149@node Virtual Base Classes
3150@section Virtual Base Classes
3151
3152A derived class object consists of a concatenation in memory of the data
3153areas defined by each base class, starting with the leftmost and ending
3154with the rightmost in the list of base classes.  The exception to this
3155rule is for virtual inheritance.  In the example above, class @code{D}
3156inherits virtually from base class @code{B}.  This means that an
3157instance of a @code{D} object will not contain its own @code{B} part but
3158merely a pointer to a @code{B} part, known as a virtual base pointer.
3159
3160In a derived class stab, the base offset part of the derivation
3161information, described above, shows how the base class parts are
3162ordered.  The base offset for a virtual base class is always given as 0.
3163Notice that the base offset for @code{B} is given as 0 even though
3164@code{B} is not the first base class.  The first base class @code{A}
3165starts at offset 0.
3166
3167The field information part of the stab for class @code{D} describes the field
3168which is the pointer to the virtual base class @code{B}. The vbase pointer
3169name is @samp{$vb} followed by a type reference to the virtual base class.
3170Since the type id for @code{B} in this example is 25, the vbase pointer name
3171is @samp{$vb25}.
3172
3173@c FIXME!! fake linebreaks below
3174@smallexample
3175.stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:1,
3176       160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt::32:i;
3177       2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;28;;D_virt:
3178       :32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
3179@end smallexample
3180
3181Following the name and a semicolon is a type reference describing the
3182type of the virtual base class pointer, in this case 24.  Type 24 was
3183defined earlier as the type of the @code{B} class @code{this} pointer.  The
3184@code{this} pointer for a class is a pointer to the class type.
3185
3186@example
3187.stabs "this:P24=*25=xsB:",64,0,0,8
3188@end example
3189
3190Finally the field offset part of the vbase pointer field description
3191shows that the vbase pointer is the first field in the @code{D} object,
3192before any data fields defined by the class.  The layout of a @code{D}
3193class object is a follows, @code{Adat} at 0, the vtable pointer for
3194@code{A} at 32, @code{Cdat} at 64, the vtable pointer for C at 96, the
3195virtual base pointer for @code{B} at 128, and @code{Ddat} at 160.
3196
3197
3198@node Static Members
3199@section Static Members
3200
3201The data area for a class is a concatenation of the space used by the
3202data members of the class.  If the class has virtual methods, a vtable
3203pointer follows the class data.  The field offset part of each field
3204description in the class stab shows this ordering.
3205
3206<< How is this reflected in stabs?  See Cygnus bug #677 for some info.  >>
3207
3208@node Stab Types
3209@appendix Table of Stab Types
3210
3211The following are all the possible values for the stab type field, for
3212a.out files, in numeric order.  This does not apply to XCOFF, but
3213it does apply to stabs in sections (@pxref{Stab Sections}).  Stabs in
3214ECOFF use these values but add 0x8f300 to distinguish them from non-stab
3215symbols.
3216
3217The symbolic names are defined in the file @file{include/aout/stabs.def}.
3218
3219@menu
3220* Non-Stab Symbol Types::	Types from 0 to 0x1f
3221* Stab Symbol Types::		Types from 0x20 to 0xff
3222@end menu
3223
3224@node Non-Stab Symbol Types
3225@appendixsec Non-Stab Symbol Types
3226
3227The following types are used by the linker and assembler, not by stab
3228directives.  Since this document does not attempt to describe aspects of
3229object file format other than the debugging format, no details are
3230given.
3231
3232@c Try to get most of these to fit on a single line.
3233@iftex
3234@tableindent=1.5in
3235@end iftex
3236
3237@table @code
3238@item 0x0     N_UNDF
3239Undefined symbol
3240
3241@item 0x2     N_ABS
3242File scope absolute symbol
3243
3244@item 0x3     N_ABS | N_EXT
3245External absolute symbol
3246
3247@item 0x4     N_TEXT
3248File scope text symbol
3249
3250@item 0x5     N_TEXT | N_EXT
3251External text symbol
3252
3253@item 0x6     N_DATA
3254File scope data symbol
3255
3256@item 0x7     N_DATA | N_EXT
3257External data symbol
3258
3259@item 0x8     N_BSS
3260File scope BSS symbol
3261
3262@item 0x9     N_BSS | N_EXT
3263External BSS symbol
3264
3265@item 0x0c    N_FN_SEQ
3266Same as @code{N_FN}, for Sequent compilers
3267
3268@item 0x0a    N_INDR
3269Symbol is indirected to another symbol
3270
3271@item 0x12    N_COMM
3272Common---visible after shared library dynamic link
3273
3274@item 0x14 N_SETA
3275@itemx 0x15 N_SETA | N_EXT
3276Absolute set element
3277
3278@item 0x16 N_SETT
3279@itemx 0x17 N_SETT | N_EXT
3280Text segment set element
3281
3282@item 0x18 N_SETD
3283@itemx 0x19 N_SETD | N_EXT
3284Data segment set element
3285
3286@item 0x1a N_SETB
3287@itemx 0x1b N_SETB | N_EXT
3288BSS segment set element
3289
3290@item 0x1c N_SETV
3291@itemx 0x1d N_SETV | N_EXT
3292Pointer to set vector
3293
3294@item 0x1e N_WARNING
3295Print a warning message during linking
3296
3297@item 0x1f    N_FN
3298File name of a @file{.o} file
3299@end table
3300
3301@node Stab Symbol Types
3302@appendixsec Stab Symbol Types
3303
3304The following symbol types indicate that this is a stab.  This is the
3305full list of stab numbers, including stab types that are used in
3306languages other than C.
3307
3308@table @code
3309@item 0x20     N_GSYM
3310Global symbol; see @ref{Global Variables}.
3311
3312@item 0x22     N_FNAME
3313Function name (for BSD Fortran); see @ref{Procedures}.
3314
3315@item 0x24     N_FUN
3316Function name (@pxref{Procedures}) or text segment variable
3317(@pxref{Statics}).
3318
3319@item 0x26 N_STSYM
3320Data segment file-scope variable; see @ref{Statics}.
3321
3322@item 0x28 N_LCSYM
3323BSS segment file-scope variable; see @ref{Statics}.
3324
3325@item 0x2a N_MAIN
3326Name of main routine; see @ref{Main Program}.
3327
3328@item 0x2c N_ROSYM
3329Variable in @code{.rodata} section; see @ref{Statics}.
3330
3331@item 0x30     N_PC
3332Global symbol (for Pascal); see @ref{N_PC}.
3333
3334@item 0x32     N_NSYMS
3335Number of symbols (according to Ultrix V4.0); see @ref{N_NSYMS}.
3336
3337@item 0x34     N_NOMAP
3338No DST map; see @ref{N_NOMAP}.
3339
3340@item 0x36     N_MAC_DEFINE
3341Name and body of a @code{#define}d macro; see @ref{Macro define and undefine}.
3342
3343@c FIXME: describe this solaris feature in the body of the text (see
3344@c comments in include/aout/stab.def).
3345@item 0x38 N_OBJ
3346Object file (Solaris2).
3347
3348@item 0x3a     N_MAC_UNDEF
3349Name of an @code{#undef}ed macro; see @ref{Macro define and undefine}.
3350
3351@c See include/aout/stab.def for (a little) more info.
3352@item 0x3c N_OPT
3353Debugger options (Solaris2).
3354
3355@item 0x40     N_RSYM
3356Register variable; see @ref{Register Variables}.
3357
3358@item 0x42     N_M2C
3359Modula-2 compilation unit; see @ref{N_M2C}.
3360
3361@item 0x44     N_SLINE
3362Line number in text segment; see @ref{Line Numbers}.
3363
3364@item 0x46     N_DSLINE
3365Line number in data segment; see @ref{Line Numbers}.
3366
3367@item 0x48     N_BSLINE
3368Line number in bss segment; see @ref{Line Numbers}.
3369
3370@item 0x48     N_BROWS
3371Sun source code browser, path to @file{.cb} file; see @ref{N_BROWS}.
3372
3373@item 0x4a     N_DEFD
3374GNU Modula2 definition module dependency; see @ref{N_DEFD}.
3375
3376@item 0x4c N_FLINE
3377Function start/body/end line numbers (Solaris2).
3378
3379@item 0x50     N_EHDECL
3380GNU C@t{++} exception variable; see @ref{N_EHDECL}.
3381
3382@item 0x50     N_MOD2
3383Modula2 info "for imc" (according to Ultrix V4.0); see @ref{N_MOD2}.
3384
3385@item 0x54     N_CATCH
3386GNU C@t{++} @code{catch} clause; see @ref{N_CATCH}.
3387
3388@item 0x60     N_SSYM
3389Structure of union element; see @ref{N_SSYM}.
3390
3391@item 0x62 N_ENDM
3392Last stab for module (Solaris2).
3393
3394@item 0x64     N_SO
3395Path and name of source file; see @ref{Source Files}.
3396
3397@item 0x80 N_LSYM
3398Stack variable (@pxref{Stack Variables}) or type (@pxref{Typedefs}).
3399
3400@item 0x82     N_BINCL
3401Beginning of an include file (Sun only); see @ref{Include Files}.
3402
3403@item 0x84     N_SOL
3404Name of include file; see @ref{Include Files}.
3405
3406@item 0xa0     N_PSYM
3407Parameter variable; see @ref{Parameters}.
3408
3409@item 0xa2     N_EINCL
3410End of an include file; see @ref{Include Files}.
3411
3412@item 0xa4     N_ENTRY
3413Alternate entry point; see @ref{Alternate Entry Points}.
3414
3415@item 0xc0     N_LBRAC
3416Beginning of a lexical block; see @ref{Block Structure}.
3417
3418@item 0xc2     N_EXCL
3419Place holder for a deleted include file; see @ref{Include Files}.
3420
3421@item 0xc4     N_SCOPE
3422Modula2 scope information (Sun linker); see @ref{N_SCOPE}.
3423
3424@item 0xe0     N_RBRAC
3425End of a lexical block; see @ref{Block Structure}.
3426
3427@item 0xe2     N_BCOMM
3428Begin named common block; see @ref{Common Blocks}.
3429
3430@item 0xe4     N_ECOMM
3431End named common block; see @ref{Common Blocks}.
3432
3433@item 0xe8     N_ECOML
3434Member of a common block; see @ref{Common Blocks}.
3435
3436@c FIXME: How does this really work?  Move it to main body of document.
3437@item 0xea N_WITH
3438Pascal @code{with} statement: type,,0,0,offset (Solaris2).
3439
3440@item 0xf0     N_NBTEXT
3441Gould non-base registers; see @ref{Gould}.
3442
3443@item 0xf2     N_NBDATA
3444Gould non-base registers; see @ref{Gould}.
3445
3446@item 0xf4     N_NBBSS
3447Gould non-base registers; see @ref{Gould}.
3448
3449@item 0xf6     N_NBSTS
3450Gould non-base registers; see @ref{Gould}.
3451
3452@item 0xf8     N_NBLCS
3453Gould non-base registers; see @ref{Gould}.
3454@end table
3455
3456@c Restore the default table indent
3457@iftex
3458@tableindent=.8in
3459@end iftex
3460
3461@node Symbol Descriptors
3462@appendix Table of Symbol Descriptors
3463
3464The symbol descriptor is the character which follows the colon in many
3465stabs, and which tells what kind of stab it is.  @xref{String Field},
3466for more information about their use.
3467
3468@c Please keep this alphabetical
3469@table @code
3470@c In TeX, this looks great, digit is in italics.  But makeinfo insists
3471@c on putting it in `', not realizing that @var should override @code.
3472@c I don't know of any way to make makeinfo do the right thing.  Seems
3473@c like a makeinfo bug to me.
3474@item @var{digit}
3475@itemx (
3476@itemx -
3477Variable on the stack; see @ref{Stack Variables}.
3478
3479@item :
3480C@t{++} nested symbol; see @xref{Nested Symbols}.
3481
3482@item a
3483Parameter passed by reference in register; see @ref{Reference Parameters}.
3484
3485@item b
3486Based variable; see @ref{Based Variables}.
3487
3488@item c
3489Constant; see @ref{Constants}.
3490
3491@item C
3492Conformant array bound (Pascal, maybe other languages); @ref{Conformant
3493Arrays}.  Name of a caught exception (GNU C@t{++}).  These can be
3494distinguished because the latter uses @code{N_CATCH} and the former uses
3495another symbol type.
3496
3497@item d
3498Floating point register variable; see @ref{Register Variables}.
3499
3500@item D
3501Parameter in floating point register; see @ref{Register Parameters}.
3502
3503@item f
3504File scope function; see @ref{Procedures}.
3505
3506@item F
3507Global function; see @ref{Procedures}.
3508
3509@item G
3510Global variable; see @ref{Global Variables}.
3511
3512@item i
3513@xref{Register Parameters}.
3514
3515@item I
3516Internal (nested) procedure; see @ref{Nested Procedures}.
3517
3518@item J
3519Internal (nested) function; see @ref{Nested Procedures}.
3520
3521@item L
3522Label name (documented by AIX, no further information known).
3523
3524@item m
3525Module; see @ref{Procedures}.
3526
3527@item p
3528Argument list parameter; see @ref{Parameters}.
3529
3530@item pP
3531@xref{Parameters}.
3532
3533@item pF
3534Fortran Function parameter; see @ref{Parameters}.
3535
3536@item P
3537Unfortunately, three separate meanings have been independently invented
3538for this symbol descriptor.  At least the GNU and Sun uses can be
3539distinguished by the symbol type.  Global Procedure (AIX) (symbol type
3540used unknown); see @ref{Procedures}.  Register parameter (GNU) (symbol
3541type @code{N_PSYM}); see @ref{Parameters}.  Prototype of function
3542referenced by this file (Sun @code{acc}) (symbol type @code{N_FUN}).
3543
3544@item Q
3545Static Procedure; see @ref{Procedures}.
3546
3547@item R
3548Register parameter; see @ref{Register Parameters}.
3549
3550@item r
3551Register variable; see @ref{Register Variables}.
3552
3553@item S
3554File scope variable; see @ref{Statics}.
3555
3556@item s
3557Local variable (OS9000).
3558
3559@item t
3560Type name; see @ref{Typedefs}.
3561
3562@item T
3563Enumeration, structure, or union tag; see @ref{Typedefs}.
3564
3565@item v
3566Parameter passed by reference; see @ref{Reference Parameters}.
3567
3568@item V
3569Procedure scope static variable; see @ref{Statics}.
3570
3571@item x
3572Conformant array; see @ref{Conformant Arrays}.
3573
3574@item X
3575Function return variable; see @ref{Parameters}.
3576@end table
3577
3578@node Type Descriptors
3579@appendix Table of Type Descriptors
3580
3581The type descriptor is the character which follows the type number and
3582an equals sign.  It specifies what kind of type is being defined.
3583@xref{String Field}, for more information about their use.
3584
3585@table @code
3586@item @var{digit}
3587@itemx (
3588Type reference; see @ref{String Field}.
3589
3590@item -
3591Reference to builtin type; see @ref{Negative Type Numbers}.
3592
3593@item #
3594Method (C@t{++}); see @ref{Method Type Descriptor}.
3595
3596@item *
3597Pointer; see @ref{Miscellaneous Types}.
3598
3599@item &
3600Reference (C@t{++}).
3601
3602@item @@
3603Type Attributes (AIX); see @ref{String Field}.  Member (class and variable)
3604type (GNU C@t{++}); see @ref{Member Type Descriptor}.
3605
3606@item a
3607Array; see @ref{Arrays}.
3608
3609@item A
3610Open array; see @ref{Arrays}.
3611
3612@item b
3613Pascal space type (AIX); see @ref{Miscellaneous Types}.  Builtin integer
3614type (Sun); see @ref{Builtin Type Descriptors}.  Const and volatile
3615qualified type (OS9000).
3616
3617@item B
3618Volatile-qualified type; see @ref{Miscellaneous Types}.
3619
3620@item c
3621Complex builtin type (AIX); see @ref{Builtin Type Descriptors}.
3622Const-qualified type (OS9000).
3623
3624@item C
3625COBOL Picture type.  See AIX documentation for details.
3626
3627@item d
3628File type; see @ref{Miscellaneous Types}.
3629
3630@item D
3631N-dimensional dynamic array; see @ref{Arrays}.
3632
3633@item e
3634Enumeration type; see @ref{Enumerations}.
3635
3636@item E
3637N-dimensional subarray; see @ref{Arrays}.
3638
3639@item f
3640Function type; see @ref{Function Types}.
3641
3642@item F
3643Pascal function parameter; see @ref{Function Types}
3644
3645@item g
3646Builtin floating point type; see @ref{Builtin Type Descriptors}.
3647
3648@item G
3649COBOL Group.  See AIX documentation for details.
3650
3651@item i
3652Imported type (AIX); see @ref{Cross-References}.  Volatile-qualified
3653type (OS9000).
3654
3655@item k
3656Const-qualified type; see @ref{Miscellaneous Types}.
3657
3658@item K
3659COBOL File Descriptor.  See AIX documentation for details.
3660
3661@item M
3662Multiple instance type; see @ref{Miscellaneous Types}.
3663
3664@item n
3665String type; see @ref{Strings}.
3666
3667@item N
3668Stringptr; see @ref{Strings}.
3669
3670@item o
3671Opaque type; see @ref{Typedefs}.
3672
3673@item p
3674Procedure; see @ref{Function Types}.
3675
3676@item P
3677Packed array; see @ref{Arrays}.
3678
3679@item r
3680Range type; see @ref{Subranges}.
3681
3682@item R
3683Builtin floating type; see @ref{Builtin Type Descriptors} (Sun).  Pascal
3684subroutine parameter; see @ref{Function Types} (AIX).  Detecting this
3685conflict is possible with careful parsing (hint: a Pascal subroutine
3686parameter type will always contain a comma, and a builtin type
3687descriptor never will).
3688
3689@item s
3690Structure type; see @ref{Structures}.
3691
3692@item S
3693Set type; see @ref{Miscellaneous Types}.
3694
3695@item u
3696Union; see @ref{Unions}.
3697
3698@item v
3699Variant record.  This is a Pascal and Modula-2 feature which is like a
3700union within a struct in C.  See AIX documentation for details.
3701
3702@item w
3703Wide character; see @ref{Builtin Type Descriptors}.
3704
3705@item x
3706Cross-reference; see @ref{Cross-References}.
3707
3708@item Y
3709Used by IBM's xlC C@t{++} compiler (for structures, I think).
3710
3711@item z
3712gstring; see @ref{Strings}.
3713@end table
3714
3715@node Expanded Reference
3716@appendix Expanded Reference by Stab Type
3717
3718@c FIXME: This appendix should go away; see N_PSYM or N_SO for an example.
3719
3720For a full list of stab types, and cross-references to where they are
3721described, see @ref{Stab Types}.  This appendix just covers certain
3722stabs which are not yet described in the main body of this document;
3723eventually the information will all be in one place.
3724
3725Format of an entry:
3726
3727The first line is the symbol type (see @file{include/aout/stab.def}).
3728
3729The second line describes the language constructs the symbol type
3730represents.
3731
3732The third line is the stab format with the significant stab fields
3733named and the rest NIL.
3734
3735Subsequent lines expand upon the meaning and possible values for each
3736significant stab field.
3737
3738Finally, any further information.
3739
3740@menu
3741* N_PC::			Pascal global symbol
3742* N_NSYMS::			Number of symbols
3743* N_NOMAP::			No DST map
3744* N_M2C::			Modula-2 compilation unit
3745* N_BROWS::			Path to .cb file for Sun source code browser
3746* N_DEFD::			GNU Modula2 definition module dependency
3747* N_EHDECL::			GNU C++ exception variable
3748* N_MOD2::			Modula2 information "for imc"
3749* N_CATCH::			GNU C++ "catch" clause
3750* N_SSYM::			Structure or union element
3751* N_SCOPE::			Modula2 scope information (Sun only)
3752* Gould::			non-base register symbols used on Gould systems
3753* N_LENG::			Length of preceding entry
3754@end menu
3755
3756@node N_PC
3757@section N_PC
3758
3759@deffn @code{.stabs} N_PC
3760@findex N_PC
3761Global symbol (for Pascal).
3762
3763@example
3764"name" -> "symbol_name"  <<?>>
3765value  -> supposedly the line number (stab.def is skeptical)
3766@end example
3767
3768@display
3769@file{stabdump.c} says:
3770
3771global pascal symbol: name,,0,subtype,line
3772<< subtype? >>
3773@end display
3774@end deffn
3775
3776@node N_NSYMS
3777@section N_NSYMS
3778
3779@deffn @code{.stabn} N_NSYMS
3780@findex N_NSYMS
3781Number of symbols (according to Ultrix V4.0).
3782
3783@display
3784        0, files,,funcs,lines (stab.def)
3785@end display
3786@end deffn
3787
3788@node N_NOMAP
3789@section N_NOMAP
3790
3791@deffn @code{.stabs} N_NOMAP
3792@findex N_NOMAP
3793No DST map for symbol (according to Ultrix V4.0).  I think this means a
3794variable has been optimized out.
3795
3796@display
3797        name, ,0,type,ignored (stab.def)
3798@end display
3799@end deffn
3800
3801@node N_M2C
3802@section N_M2C
3803
3804@deffn @code{.stabs} N_M2C
3805@findex N_M2C
3806Modula-2 compilation unit.
3807
3808@example
3809"string" -> "unit_name,unit_time_stamp[,code_time_stamp]"
3810desc   -> unit_number
3811value  -> 0 (main unit)
3812          1 (any other unit)
3813@end example
3814
3815See @cite{Dbx and Dbxtool Interfaces}, 2nd edition, by Sun, 1988, for
3816more information.
3817
3818@end deffn
3819
3820@node N_BROWS
3821@section N_BROWS
3822
3823@deffn @code{.stabs} N_BROWS
3824@findex N_BROWS
3825Sun source code browser, path to @file{.cb} file
3826
3827<<?>>
3828"path to associated @file{.cb} file"
3829
3830Note: N_BROWS has the same value as N_BSLINE.
3831@end deffn
3832
3833@node N_DEFD
3834@section N_DEFD
3835
3836@deffn @code{.stabn} N_DEFD
3837@findex N_DEFD
3838GNU Modula2 definition module dependency.
3839
3840GNU Modula-2 definition module dependency.  The value is the
3841modification time of the definition file.  The other field is non-zero
3842if it is imported with the GNU M2 keyword @code{%INITIALIZE}.  Perhaps
3843@code{N_M2C} can be used if there are enough empty fields?
3844@end deffn
3845
3846@node N_EHDECL
3847@section N_EHDECL
3848
3849@deffn @code{.stabs} N_EHDECL
3850@findex N_EHDECL
3851GNU C@t{++} exception variable <<?>>.
3852
3853"@var{string} is variable name"
3854
3855Note: conflicts with @code{N_MOD2}.
3856@end deffn
3857
3858@node N_MOD2
3859@section N_MOD2
3860
3861@deffn @code{.stab?} N_MOD2
3862@findex N_MOD2
3863Modula2 info "for imc" (according to Ultrix V4.0)
3864
3865Note: conflicts with @code{N_EHDECL}  <<?>>
3866@end deffn
3867
3868@node N_CATCH
3869@section N_CATCH
3870
3871@deffn @code{.stabn} N_CATCH
3872@findex N_CATCH
3873GNU C@t{++} @code{catch} clause
3874
3875GNU C@t{++} @code{catch} clause.  The value is its address.  The desc field
3876is nonzero if this entry is immediately followed by a @code{CAUGHT} stab
3877saying what exception was caught.  Multiple @code{CAUGHT} stabs means
3878that multiple exceptions can be caught here.  If desc is 0, it means all
3879exceptions are caught here.
3880@end deffn
3881
3882@node N_SSYM
3883@section N_SSYM
3884
3885@deffn @code{.stabn} N_SSYM
3886@findex N_SSYM
3887Structure or union element.
3888
3889The value is the offset in the structure.
3890
3891<<?looking at structs and unions in C I didn't see these>>
3892@end deffn
3893
3894@node N_SCOPE
3895@section N_SCOPE
3896
3897@deffn @code{.stab?} N_SCOPE
3898@findex N_SCOPE
3899Modula2 scope information (Sun linker)
3900<<?>>
3901@end deffn
3902
3903@node Gould
3904@section Non-base registers on Gould systems
3905
3906@deffn @code{.stab?} N_NBTEXT
3907@deffnx @code{.stab?} N_NBDATA
3908@deffnx @code{.stab?} N_NBBSS
3909@deffnx @code{.stab?} N_NBSTS
3910@deffnx @code{.stab?} N_NBLCS
3911@findex N_NBTEXT
3912@findex N_NBDATA
3913@findex N_NBBSS
3914@findex N_NBSTS
3915@findex N_NBLCS
3916These are used on Gould systems for non-base registers syms.
3917
3918However, the following values are not the values used by Gould; they are
3919the values which GNU has been documenting for these values for a long
3920time, without actually checking what Gould uses.  I include these values
3921only because perhaps some someone actually did something with the GNU
3922information (I hope not, why GNU knowingly assigned wrong values to
3923these in the header file is a complete mystery to me).
3924
3925@example
3926240    0xf0     N_NBTEXT  ??
3927242    0xf2     N_NBDATA  ??
3928244    0xf4     N_NBBSS   ??
3929246    0xf6     N_NBSTS   ??
3930248    0xf8     N_NBLCS   ??
3931@end example
3932@end deffn
3933
3934@node N_LENG
3935@section N_LENG
3936
3937@deffn @code{.stabn} N_LENG
3938@findex N_LENG
3939Second symbol entry containing a length-value for the preceding entry.
3940The value is the length.
3941@end deffn
3942
3943@node Questions
3944@appendix Questions and Anomalies
3945
3946@itemize @bullet
3947@item
3948@c I think this is changed in GCC 2.4.5 to put the line number there.
3949For GNU C stabs defining local and global variables (@code{N_LSYM} and
3950@code{N_GSYM}), the desc field is supposed to contain the source
3951line number on which the variable is defined.  In reality the desc
3952field is always 0.  (This behavior is defined in @file{dbxout.c} and
3953putting a line number in desc is controlled by @samp{#ifdef
3954WINNING_GDB}, which defaults to false). GDB supposedly uses this
3955information if you say @samp{list @var{var}}.  In reality, @var{var} can
3956be a variable defined in the program and GDB says @samp{function
3957@var{var} not defined}.
3958
3959@item
3960In GNU C stabs, there seems to be no way to differentiate tag types:
3961structures, unions, and enums (symbol descriptor @samp{T}) and typedefs
3962(symbol descriptor @samp{t}) defined at file scope from types defined locally
3963to a procedure or other more local scope.  They all use the @code{N_LSYM}
3964stab type.  Types defined at procedure scope are emitted after the
3965@code{N_RBRAC} of the preceding function and before the code of the
3966procedure in which they are defined.  This is exactly the same as
3967types defined in the source file between the two procedure bodies.
3968GDB over-compensates by placing all types in block #1, the block for
3969symbols of file scope.  This is true for default, @samp{-ansi} and
3970@samp{-traditional} compiler options. (Bugs gcc/1063, gdb/1066.)
3971
3972@item
3973What ends the procedure scope?  Is it the proc block's @code{N_RBRAC} or the
3974next @code{N_FUN}?  (I believe its the first.)
3975@end itemize
3976
3977@node Stab Sections
3978@appendix Using Stabs in Their Own Sections
3979
3980Many object file formats allow tools to create object files with custom
3981sections containing any arbitrary data.  For any such object file
3982format, stabs can be embedded in special sections.  This is how stabs
3983are used with ELF and SOM, and aside from ECOFF and XCOFF, is how stabs
3984are used with COFF.
3985
3986@menu
3987* Stab Section Basics::    How to embed stabs in sections
3988* ELF Linker Relocation::  Sun ELF hacks
3989@end menu
3990
3991@node Stab Section Basics
3992@appendixsec How to Embed Stabs in Sections
3993
3994The assembler creates two custom sections, a section named @code{.stab}
3995which contains an array of fixed length structures, one struct per stab,
3996and a section named @code{.stabstr} containing all the variable length
3997strings that are referenced by stabs in the @code{.stab} section.  The
3998byte order of the stabs binary data depends on the object file format.
3999For ELF, it matches the byte order of the ELF file itself, as determined
4000from the @code{EI_DATA} field in the @code{e_ident} member of the ELF
4001header.  For SOM, it is always big-endian (is this true??? FIXME).  For
4002COFF, it matches the byte order of the COFF headers.  The meaning of the
4003fields is the same as for a.out (@pxref{Symbol Table Format}), except
4004that the @code{n_strx} field is relative to the strings for the current
4005compilation unit (which can be found using the synthetic N_UNDF stab
4006described below), rather than the entire string table.
4007
4008The first stab in the @code{.stab} section for each compilation unit is
4009synthetic, generated entirely by the assembler, with no corresponding
4010@code{.stab} directive as input to the assembler.  This stab contains
4011the following fields:
4012
4013@table @code
4014@item n_strx
4015Offset in the @code{.stabstr} section to the source filename.
4016
4017@item n_type
4018@code{N_UNDF}.
4019
4020@item n_other
4021Unused field, always zero.
4022This may eventually be used to hold overflows from the count in
4023the @code{n_desc} field.
4024
4025@item n_desc
4026Count of upcoming symbols, i.e., the number of remaining stabs for this
4027source file.
4028
4029@item n_value
4030Size of the string table fragment associated with this source file, in
4031bytes.
4032@end table
4033
4034The @code{.stabstr} section always starts with a null byte (so that string
4035offsets of zero reference a null string), followed by random length strings,
4036each of which is null byte terminated.
4037
4038The ELF section header for the @code{.stab} section has its
4039@code{sh_link} member set to the section number of the @code{.stabstr}
4040section, and the @code{.stabstr} section has its ELF section
4041header @code{sh_type} member set to @code{SHT_STRTAB} to mark it as a
4042string table.  SOM and COFF have no way of linking the sections together
4043or marking them as string tables.
4044
4045For COFF, the @code{.stab} and @code{.stabstr} sections may be simply
4046concatenated by the linker.  GDB then uses the @code{n_desc} fields to
4047figure out the extent of the original sections.  Similarly, the
4048@code{n_value} fields of the header symbols are added together in order
4049to get the actual position of the strings in a desired @code{.stabstr}
4050section.  Although this design obviates any need for the linker to
4051relocate or otherwise manipulate @code{.stab} and @code{.stabstr}
4052sections, it also requires some care to ensure that the offsets are
4053calculated correctly.  For instance, if the linker were to pad in
4054between the @code{.stabstr} sections before concatenating, then the
4055offsets to strings in the middle of the executable's @code{.stabstr}
4056section would be wrong.
4057
4058The GNU linker is able to optimize stabs information by merging
4059duplicate strings and removing duplicate header file information
4060(@pxref{Include Files}).  When some versions of the GNU linker optimize
4061stabs in sections, they remove the leading @code{N_UNDF} symbol and
4062arranges for all the @code{n_strx} fields to be relative to the start of
4063the @code{.stabstr} section.
4064
4065@node ELF Linker Relocation
4066@appendixsec Having the Linker Relocate Stabs in ELF
4067
4068This section describes some Sun hacks for Stabs in ELF; it does not
4069apply to COFF or SOM.  While @value{GDBN} no longer supports this hack
4070for Sun Stabs in ELF, this section is kept to document the issue.
4071
4072To keep linking fast, you don't want the linker to have to relocate very
4073many stabs.  Making sure this is done for @code{N_SLINE},
4074@code{N_RBRAC}, and @code{N_LBRAC} stabs is the most important thing
4075(see the descriptions of those stabs for more information).  But Sun's
4076stabs in ELF has taken this further, to make all addresses in the
4077@code{n_value} field (functions and static variables) relative to the
4078source file.  For the @code{N_SO} symbol itself, Sun simply omits the
4079address.  To find the address of each section corresponding to a given
4080source file, the compiler puts out symbols giving the address of each
4081section for a given source file.  Since these are ELF (not stab)
4082symbols, the linker relocates them correctly without having to touch the
4083stabs section.  They are named @code{Bbss.bss} for the bss section,
4084@code{Ddata.data} for the data section, and @code{Drodata.rodata} for
4085the rodata section.  For the text section, there is no such symbol (but
4086there should be, see below).  For an example of how these symbols work,
4087@xref{Stab Section Transformations}.  GCC does not provide these symbols;
4088it instead relies on the stabs getting relocated.  Thus addresses which
4089would normally be relative to @code{Bbss.bss}, etc., are already
4090relocated.  The Sun linker provided with Solaris 2.2 and earlier
4091relocates stabs using normal ELF relocation information, as it would do
4092for any section.  Sun has been threatening to kludge their linker to not
4093do this (to speed up linking), even though the correct way to avoid
4094having the linker do these relocations is to have the compiler no longer
4095output relocatable values.  Last I heard they had been talked out of the
4096linker kludge.  See Sun point patch 101052-01 and Sun bug 1142109.  With
4097the Sun compiler this affects @samp{S} symbol descriptor stabs
4098(@pxref{Statics}) and functions (@pxref{Procedures}).  In the latter
4099case, to adopt the clean solution (making the value of the stab relative
4100to the start of the compilation unit), it would be necessary to invent a
4101@code{Ttext.text} symbol, analogous to the @code{Bbss.bss}, etc.,
4102symbols.  I recommend this rather than using a zero value and getting
4103the address from the ELF symbols.
4104
4105Finding the correct @code{Bbss.bss}, etc., symbol is difficult, because
4106the linker simply concatenates the @code{.stab} sections from each
4107@file{.o} file without including any information about which part of a
4108@code{.stab} section comes from which @file{.o} file.  The way GDB use to
4109do this is to look for an ELF @code{STT_FILE} symbol which has the same
4110name as the last component of the file name from the @code{N_SO} symbol
4111in the stabs (for example, if the file name is @file{../../gdb/main.c},
4112it looks for an ELF @code{STT_FILE} symbol named @code{main.c}).  This
4113loses if different files have the same name (they could be in different
4114directories, a library could have been copied from one system to
4115another, etc.).  It would be much cleaner to have the @code{Bbss.bss}
4116symbols in the stabs themselves.  Having the linker relocate them there
4117is no more work than having the linker relocate ELF symbols, and it
4118solves the problem of having to associate the ELF and stab symbols.
4119However, no one has yet designed or implemented such a scheme.
4120
4121@node GNU Free Documentation License
4122@appendix GNU Free Documentation License
4123@include fdl.texi
4124
4125@node Symbol Types Index
4126@unnumbered Symbol Types Index
4127
4128@printindex fn
4129
4130@bye
4131