2 @setfilename stabs.info
9 * Stabs:: The "stabs" debugging information format.
15 This document describes the stabs debugging symbol tables.
17 Copyright 1992, 1993 Free Software Foundation, Inc.
18 Contributed by Cygnus Support. Written by Julia Menapace, Jim Kingdon,
21 Permission is granted to make and distribute verbatim copies of
22 this manual provided the copyright notice and this permission notice
23 are preserved on all copies.
26 Permission is granted to process this file through Tex and print the
27 results, provided the printed document carries copying permission
28 notice identical to this one except for the removal of this paragraph
29 (this paragraph not being relevant to the printed manual).
32 Permission is granted to copy or distribute modified versions of this
33 manual under the terms of the GPL (for which purpose this text may be
34 regarded as a program in the language TeX).
37 @setchapternewpage odd
40 @title The ``stabs'' debug format
41 @author Julia Menapace, Jim Kingdon, David MacKenzie
42 @author Cygnus Support
45 \def\$#1${{#1}} % Kluge: collect RCS revision info without $...$
46 \xdef\manvers{\$Revision$} % For use in headers, footers too
48 \hfill Cygnus Support\par
50 \hfill \TeX{}info \texinfoversion\par
54 @vskip 0pt plus 1filll
55 Copyright @copyright{} 1992, 1993 Free Software Foundation, Inc.
56 Contributed by Cygnus Support.
58 Permission is granted to make and distribute verbatim copies of
59 this manual provided the copyright notice and this permission notice
60 are preserved on all copies.
66 @top The "stabs" representation of debugging information
68 This document describes the stabs debugging format.
71 * Overview:: Overview of stabs
72 * Program Structure:: Encoding of the structure of the program
73 * Constants:: Constants
75 * Types:: Type definitions
76 * Symbol Tables:: Symbol information in symbol tables
77 * Cplusplus:: Appendixes:
78 * Stab Types:: Symbol types in a.out files
79 * Symbol Descriptors:: Table of symbol descriptors
80 * Type Descriptors:: Table of type descriptors
81 * Expanded Reference:: Reference information by stab type
82 * Questions:: Questions and anomolies
83 * XCOFF Differences:: Differences between GNU stabs in a.out
84 and GNU stabs in XCOFF
85 * Sun Differences:: Differences between GNU stabs and Sun
87 * Stabs In ELF:: Stabs in an ELF file.
88 * Symbol Types Index:: Index of symbolic stab symbol type names.
94 @chapter Overview of Stabs
96 @dfn{Stabs} refers to a format for information that describes a program
97 to a debugger. This format was apparently invented by
99 the University of California at Berkeley, for the @code{pdx} Pascal
100 debugger; the format has spread widely since then.
102 This document is one of the few published sources of documentation on
103 stabs. It is believed to be comprehensive for stabs used by C. The
104 lists of symbol descriptors (@pxref{Symbol Descriptors}) and type
105 descriptors (@pxref{Type Descriptors}) are believed to be completely
106 comprehensive. Stabs for COBOL-specific features and for variant
107 records (used by Pascal and Modula-2) are poorly documented here.
109 Other sources of information on stabs are @cite{Dbx and Dbxtool
110 Interfaces}, 2nd edition, by Sun, 1988, and @cite{AIX Version 3.2 Files
111 Reference}, Fourth Edition, September 1992, "dbx Stabstring Grammar" in
112 the a.out section, page 2-31. This document is believed to incorporate
113 the information from those two sources except where it explicitly directs
114 you to them for more information.
117 * Flow:: Overview of debugging information flow
118 * Stabs Format:: Overview of stab format
119 * String Field:: The string field
120 * C Example:: A simple example in C source
121 * Assembly Code:: The simple example at the assembly level
125 @section Overview of Debugging Information Flow
127 The GNU C compiler compiles C source in a @file{.c} file into assembly
128 language in a @file{.s} file, which the assembler translates into
129 a @file{.o} file, which the linker combines with other @file{.o} files and
130 libraries to produce an executable file.
132 With the @samp{-g} option, GCC puts in the @file{.s} file additional
133 debugging information, which is slightly transformed by the assembler
134 and linker, and carried through into the final executable. This
135 debugging information describes features of the source file like line
136 numbers, the types and scopes of variables, and function names,
137 parameters, and scopes.
139 For some object file formats, the debugging information is encapsulated
140 in assembler directives known collectively as @dfn{stab} (symbol table)
141 directives, which are interspersed with the generated code. Stabs are
142 the native format for debugging information in the a.out and XCOFF
143 object file formats. The GNU tools can also emit stabs in the COFF and
144 ECOFF object file formats.
146 The assembler adds the information from stabs to the symbol information
147 it places by default in the symbol table and the string table of the
148 @file{.o} file it is building. The linker consolidates the @file{.o}
149 files into one executable file, with one symbol table and one string
150 table. Debuggers use the symbol and string tables in the executable as
151 a source of debugging information about the program.
154 @section Overview of Stab Format
156 There are three overall formats for stab assembler directives,
157 differentiated by the first word of the stab. The name of the directive
158 describes which combination of four possible data fields follows. It is
159 either @code{.stabs} (string), @code{.stabn} (number), or @code{.stabd}
160 (dot). IBM's XCOFF assembler uses @code{.stabx} (and some other
161 directives such as @code{.file} and @code{.bi}) instead of
162 @code{.stabs}, @code{.stabn} or @code{.stabd}.
164 The overall format of each class of stab is:
167 .stabs "@var{string}",@var{type},@var{other},@var{desc},@var{value}
168 .stabn @var{type},@var{other},@var{desc},@var{value}
169 .stabd @var{type},@var{other},@var{desc}
170 .stabx "@var{string}",@var{value},@var{type},@var{sdb-type}
173 @c what is the correct term for "current file location"? My AIX
174 @c assembler manual calls it "the value of the current location counter".
175 For @code{.stabn} and @code{.stabd}, there is no @var{string} (the
176 @code{n_strx} field is zero; see @ref{Symbol Tables}). For
177 @code{.stabd}, the @var{value} field is implicit and has the value of
178 the current file location. For @code{.stabx}, the @var{sdb-type} field
179 is unused for stabs and can always be set to zero. The @var{other}
180 field is almost always unused and can be set to zero.
182 The number in the @var{type} field gives some basic information about
183 which type of stab this is (or whether it @emph{is} a stab, as opposed
184 to an ordinary symbol). Each valid type number defines a different stab
185 type; further, the stab type defines the exact interpretation of, and
186 possible values for, any remaining @var{string}, @var{desc}, or
187 @var{value} fields present in the stab. @xref{Stab Types}, for a list
188 in numeric order of the valid @var{type} field values for stab directives.
191 @section The String Field
193 For most stabs the string field holds the meat of the
194 debugging information. The flexible nature of this field
195 is what makes stabs extensible. For some stab types the string field
196 contains only a name. For other stab types the contents can be a great
199 The overall format of the string field for most stab types is:
202 "@var{name}:@var{symbol-descriptor} @var{type-information}"
205 @var{name} is the name of the symbol represented by the stab; it can
206 contain a pair of colons (@pxref{Nested Symbols}). @var{name} can be
207 omitted, which means the stab represents an unnamed object. For
208 example, @samp{:t10=*2} defines type 10 as a pointer to type 2, but does
209 not give the type a name. Omitting the @var{name} field is supported by
210 AIX dbx and GDB after about version 4.8, but not other debuggers. GCC
211 sometimes uses a single space as the name instead of omitting the name
212 altogether; apparently that is supported by most debuggers.
214 The @var{symbol-descriptor} following the @samp{:} is an alphabetic
215 character that tells more specifically what kind of symbol the stab
216 represents. If the @var{symbol-descriptor} is omitted, but type
217 information follows, then the stab represents a local variable. For a
218 list of symbol descriptors, see @ref{Symbol Descriptors}. The @samp{c}
219 symbol descriptor is an exception in that it is not followed by type
220 information. @xref{Constants}.
222 @var{type-information} is either a @var{type-number}, or
223 @samp{@var{type-number}=}. A @var{type-number} alone is a type
224 reference, referring directly to a type that has already been defined.
226 The @samp{@var{type-number}=} form is a type definition, where the
227 number represents a new type which is about to be defined. The type
228 definition may refer to other types by number, and those type numbers
229 may be followed by @samp{=} and nested definitions. Also, the Lucid
230 compiler will repeat @samp{@var{type-number}=} more than once if it
231 wants to define several type numbers at once.
233 In a type definition, if the character that follows the equals sign is
234 non-numeric then it is a @var{type-descriptor}, and tells what kind of
235 type is about to be defined. Any other values following the
236 @var{type-descriptor} vary, depending on the @var{type-descriptor}.
237 @xref{Type Descriptors}, for a list of @var{type-descriptor} values. If
238 a number follows the @samp{=} then the number is a @var{type-reference}.
239 For a full description of types, @ref{Types}.
241 There is an AIX extension for type attributes. Following the @samp{=}
242 are any number of type attributes. Each one starts with @samp{@@} and
243 ends with @samp{;}. Debuggers, including AIX's dbx and GDB 4.10, skip
244 any type attributes they do not recognize. GDB 4.9 and other versions
245 of dbx may not do this. Because of a conflict with C++
246 (@pxref{Cplusplus}), new attributes should not be defined which begin
247 with a digit, @samp{(}, or @samp{-}; GDB may be unable to distinguish
248 those from the C++ type descriptor @samp{@@}. The attributes are:
251 @item a@var{boundary}
252 @var{boundary} is an integer specifying the alignment. I assume it
253 applies to all variables of this type.
256 Pointer class (for checking). Not sure what this means, or how
257 @var{integer} is interpreted.
260 Indicate this is a packed type, meaning that structure fields or array
261 elements are placed more closely in memory, to save memory at the
265 Size in bits of a variable of this type. This is fully supported by GDB
269 Indicate that this type is a string instead of an array of characters,
270 or a bitstring instead of a set. It doesn't change the layout of the
271 data being represented, but does enable the debugger to know which type
275 All of this can make the string field quite long. All versions of GDB,
276 and some versions of dbx, can handle arbitrarily long strings. But many
277 versions of dbx (or assemblers or linkers, I'm not sure which)
278 cretinously limit the strings to about 80 characters, so compilers which
279 must work with such systems need to split the @code{.stabs} directive
280 into several @code{.stabs} directives. Each stab duplicates every field
281 except the string field. The string field of every stab except the last
282 is marked as continued with a backslash at the end (in the assembly code
283 this may be written as a double backslash, depending on the assembler).
284 Removing the backslashes and concatenating the string fields of each
285 stab produces the original, long string.
288 @section A Simple Example in C Source
290 To get the flavor of how stabs describe source information for a C
291 program, let's look at the simple program:
296 printf("Hello world");
300 When compiled with @samp{-g}, the program above yields the following
301 @file{.s} file. Line numbers have been added to make it easier to refer
302 to parts of the @file{.s} file in the description of the stabs that
306 @section The Simple Example at the Assembly Level
308 This simple ``hello world'' example demonstrates several of the stab
309 types used to describe C language source files.
313 2 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0
314 3 .stabs "hello.c",100,0,0,Ltext0
317 6 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0
318 7 .stabs "char:t2=r2;0;127;",128,0,0,0
319 8 .stabs "long int:t3=r1;-2147483648;2147483647;",128,0,0,0
320 9 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
321 10 .stabs "long unsigned int:t5=r1;0;-1;",128,0,0,0
322 11 .stabs "short int:t6=r1;-32768;32767;",128,0,0,0
323 12 .stabs "long long int:t7=r1;0;-1;",128,0,0,0
324 13 .stabs "short unsigned int:t8=r1;0;65535;",128,0,0,0
325 14 .stabs "long long unsigned int:t9=r1;0;-1;",128,0,0,0
326 15 .stabs "signed char:t10=r1;-128;127;",128,0,0,0
327 16 .stabs "unsigned char:t11=r1;0;255;",128,0,0,0
328 17 .stabs "float:t12=r1;4;0;",128,0,0,0
329 18 .stabs "double:t13=r1;8;0;",128,0,0,0
330 19 .stabs "long double:t14=r1;8;0;",128,0,0,0
331 20 .stabs "void:t15=15",128,0,0,0
334 23 .ascii "Hello, world!\12\0"
349 38 sethi %hi(LC0),%o1
350 39 or %o1,%lo(LC0),%o0
361 50 .stabs "main:F1",36,0,0,_main
362 51 .stabn 192,0,0,LBB2
363 52 .stabn 224,0,0,LBE2
366 @node Program Structure
367 @chapter Encoding the Structure of the Program
369 The elements of the program structure that stabs encode include the name
370 of the main function, the names of the source and include files, the
371 line numbers, procedure names and types, and the beginnings and ends of
375 * Main Program:: Indicate what the main program is
376 * Source Files:: The path and name of the source file
377 * Include Files:: Names of include files
380 * Nested Procedures::
385 @section Main Program
388 Most languages allow the main program to have any name. The
389 @code{N_MAIN} stab type tells the debugger the name that is used in this
390 program. Only the string field is significant; it is the name of
391 a function which is the main program. Most C compilers do not use this
392 stab (they expect the debugger to assume that the name is @code{main}),
393 but some C compilers emit an @code{N_MAIN} stab for the @code{main}
397 @section Paths and Names of the Source Files
400 Before any other stabs occur, there must be a stab specifying the source
401 file. This information is contained in a symbol of stab type
402 @code{N_SO}; the string field contains the name of the file. The
403 value of the symbol is the start address of the portion of the
404 text section corresponding to that file.
406 With the Sun Solaris2 compiler, the desc field contains a
407 source-language code.
408 @c Do the debuggers use it? What are the codes? -djm
410 Some compilers (for example, GCC2 and SunOS4 @file{/bin/cc}) also
411 include the directory in which the source was compiled, in a second
412 @code{N_SO} symbol preceding the one containing the file name. This
413 symbol can be distinguished by the fact that it ends in a slash. Code
414 from the @code{cfront} C++ compiler can have additional @code{N_SO} symbols for
415 nonexistent source files after the @code{N_SO} for the real source file;
416 these are believed to contain no useful information.
421 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0 # @r{100 is N_SO}
422 .stabs "hello.c",100,0,0,Ltext0
427 Instead of @code{N_SO} symbols, XCOFF uses a @code{.file} assembler
428 directive which assembles to a standard COFF @code{.file} symbol;
429 explaining this in detail is outside the scope of this document.
432 @section Names of Include Files
434 There are several schemes for dealing with include files: the
435 traditional @code{N_SOL} approach, Sun's @code{N_BINCL} approach, and the
436 XCOFF @code{C_BINCL} approach (which despite the similar name has little in
437 common with @code{N_BINCL}).
440 An @code{N_SOL} symbol specifies which include file subsequent symbols
441 refer to. The string field is the name of the file and the value is the
442 text address corresponding to the end of the previous include file and
443 the start of this one. To specify the main source file again, use an
444 @code{N_SOL} symbol with the name of the main source file.
449 The @code{N_BINCL} approach works as follows. An @code{N_BINCL} symbol
450 specifies the start of an include file. In an object file, only the
451 string is significant; the Sun linker puts data into some of the
452 other fields. The end of the include file is marked by an
453 @code{N_EINCL} symbol (which has no string field). In an object
454 file, there is no significant data in the @code{N_EINCL} symbol; the Sun
455 linker puts data into some of the fields. @code{N_BINCL} and
456 @code{N_EINCL} can be nested.
458 If the linker detects that two source files have identical stabs between
459 an @code{N_BINCL} and @code{N_EINCL} pair (as will generally be the case
460 for a header file), then it only puts out the stabs once. Each
461 additional occurance is replaced by an @code{N_EXCL} symbol. I believe
462 the Sun (SunOS4, not sure about Solaris) linker is the only one which
463 supports this feature.
464 @c What do the fields of N_EXCL contain? -djm
468 For the start of an include file in XCOFF, use the @file{.bi} assembler
469 directive, which generates a @code{C_BINCL} symbol. A @file{.ei}
470 directive, which generates a @code{C_EINCL} symbol, denotes the end of
471 the include file. Both directives are followed by the name of the
472 source file in quotes, which becomes the string for the symbol.
473 The value of each symbol, produced automatically by the assembler
474 and linker, is the offset into the executable of the beginning
475 (inclusive, as you'd expect) or end (inclusive, as you would not expect)
476 of the portion of the COFF line table that corresponds to this include
477 file. @code{C_BINCL} and @code{C_EINCL} do not nest.
480 @section Line Numbers
483 An @code{N_SLINE} symbol represents the start of a source line. The
484 desc field contains the line number and the value
485 contains the code address for the start of that source line. On most
486 machines the address is absolute; for Sun's stabs-in-ELF and GNU's
487 stabs-in-SOM, it is relative to the function in which the @code{N_SLINE}
492 GNU documents @code{N_DSLINE} and @code{N_BSLINE} symbols for line
493 numbers in the data or bss segments, respectively. They are identical
494 to @code{N_SLINE} but are relocated differently by the linker. They
495 were intended to be used to describe the source location of a variable
496 declaration, but I believe that GCC2 actually puts the line number in
497 the desc field of the stab for the variable itself. GDB has been
498 ignoring these symbols (unless they contain a string field) since
501 For single source lines that generate discontiguous code, such as flow
502 of control statements, there may be more than one line number entry for
503 the same source line. In this case there is a line number entry at the
504 start of each code range, each with the same line number.
506 XCOFF does not use stabs for line numbers. Instead, it uses COFF line
507 numbers (which are outside the scope of this document). Standard COFF
508 line numbers cannot deal with include files, but in XCOFF this is fixed
509 with the @code{C_BINCL} method of marking include files (@pxref{Include
515 @findex N_FUN, for functions
517 @findex N_STSYM, for functions (Sun acc)
518 @findex N_GSYM, for functions (Sun acc)
519 All of the following stabs normally use the @code{N_FUN} symbol type.
520 However, Sun's @code{acc} compiler on SunOS4 uses @code{N_GSYM} and
521 @code{N_STSYM}, which means that the value of the stab for the function
522 is useless and the debugger must get the address of the function from
523 the non-stab symbols instead. BSD Fortran is said to use @code{N_FNAME}
524 with the same restriction; the value of the symbol is not useful (I'm
525 not sure it really does use this, because GDB doesn't handle this and no
528 A function is represented by an @samp{F} symbol descriptor for a global
529 (extern) function, and @samp{f} for a static (local) function. The
530 value is the address of the start of the function. For @code{a.out}, it
531 is already relocated. For stabs in ELF, the SunPRO compiler version
532 2.0.1 and GCC put out an address which gets relocated by the linker. In
533 a future release SunPRO is planning to put out zero, in which case the
534 address can be found from the ELF (non-stab) symbol. Because looking
535 things up in the ELF symbols would probably be slow, I'm not sure how to
536 find which symbol of that name is the right one, and this doesn't
537 provide any way to deal with nested functions, it would probably be
538 better to make the value of the stab an address relative to the start of
539 the file. See @ref{Stabs In ELF} for more information on linker
540 relocation of stabs in ELF files.
542 The type information of the stab represents the return type of the
543 function; thus @samp{foo:f5} means that foo is a function returning type
544 5. There is no need to try to get the line number of the start of the
545 function from the stab for the function; it is in the next
546 @code{N_SLINE} symbol.
548 @c FIXME: verify whether the "I suspect" below is true or not.
549 Some compilers (such as Sun's Solaris compiler) support an extension for
550 specifying the types of the arguments. I suspect this extension is not
551 used for old (non-prototyped) function definitions in C. If the
552 extension is in use, the type information of the stab for the function
553 is followed by type information for each argument, with each argument
554 preceded by @samp{;}. An argument type of 0 means that additional
555 arguments are being passed, whose types and number may vary (@samp{...}
556 in ANSI C). GDB has tolerated this extension (parsed the syntax, if not
557 necessarily used the information) since at least version 4.8; I don't
558 know whether all versions of dbx tolerate it. The argument types given
559 here are not redundant with the symbols for the formal parameters
560 (@pxref{Parameters}); they are the types of the arguments as they are
561 passed, before any conversions might take place. For example, if a C
562 function which is declared without a prototype takes a @code{float}
563 argument, the value is passed as a @code{double} but then converted to a
564 @code{float}. Debuggers need to use the types given in the arguments
565 when printing values, but when calling the function they need to use the
566 types given in the symbol defining the function.
568 If the return type and types of arguments of a function which is defined
569 in another source file are specified (i.e., a function prototype in ANSI
570 C), traditionally compilers emit no stab; the only way for the debugger
571 to find the information is if the source file where the function is
572 defined was also compiled with debugging symbols. As an extension the
573 Solaris compiler uses symbol descriptor @samp{P} followed by the return
574 type of the function, followed by the arguments, each preceded by
575 @samp{;}, as in a stab with symbol descriptor @samp{f} or @samp{F}.
576 This use of symbol descriptor @samp{P} can be distinguished from its use
577 for register parameters (@pxref{Register Parameters}) by the fact that it has
578 symbol type @code{N_FUN}.
580 The AIX documentation also defines symbol descriptor @samp{J} as an
581 internal function. I assume this means a function nested within another
582 function. It also says symbol descriptor @samp{m} is a module in
583 Modula-2 or extended Pascal.
585 Procedures (functions which do not return values) are represented as
586 functions returning the @code{void} type in C. I don't see why this couldn't
587 be used for all languages (inventing a @code{void} type for this purpose if
588 necessary), but the AIX documentation defines @samp{I}, @samp{P}, and
589 @samp{Q} for internal, global, and static procedures, respectively.
590 These symbol descriptors are unusual in that they are not followed by
593 The following example shows a stab for a function @code{main} which
594 returns type number @code{1}. The @code{_main} specified for the value
595 is a reference to an assembler label which is used to fill in the start
596 address of the function.
599 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
602 The stab representing a procedure is located immediately following the
603 code of the procedure. This stab is in turn directly followed by a
604 group of other stabs describing elements of the procedure. These other
605 stabs describe the procedure's parameters, its block local variables, and
608 @node Nested Procedures
609 @section Nested Procedures
611 For any of the symbol descriptors representing procedures, after the
612 symbol descriptor and the type information is optionally a scope
613 specifier. This consists of a comma, the name of the procedure, another
614 comma, and the name of the enclosing procedure. The first name is local
615 to the scope specified, and seems to be redundant with the name of the
616 symbol (before the @samp{:}). This feature is used by GCC, and
617 presumably Pascal, Modula-2, etc., compilers, for nested functions.
619 If procedures are nested more than one level deep, only the immediately
620 containing scope is specified. For example, this code:
632 return baz (x + 2 * y);
634 return x + bar (3 * x);
642 .stabs "baz:f1,baz,bar",36,0,0,_baz.15 # @r{36 is N_FUN}
643 .stabs "bar:f1,bar,foo",36,0,0,_bar.12
644 .stabs "foo:F1",36,0,0,_foo
647 @node Block Structure
648 @section Block Structure
652 The program's block structure is represented by the @code{N_LBRAC} (left
653 brace) and the @code{N_RBRAC} (right brace) stab types. The variables
654 defined inside a block precede the @code{N_LBRAC} symbol for most
655 compilers, including GCC. Other compilers, such as the Convex, Acorn
656 RISC machine, and Sun @code{acc} compilers, put the variables after the
657 @code{N_LBRAC} symbol. The values of the @code{N_LBRAC} and
658 @code{N_RBRAC} symbols are the start and end addresses of the code of
659 the block, respectively. For most machines, they are relative to the
660 starting address of this source file. For the Gould NP1, they are
661 absolute. For Sun's stabs-in-ELF and GNU's stabs-in-SOM, they are relative
662 to the function in which they occur.
664 The @code{N_LBRAC} and @code{N_RBRAC} stabs that describe the block
665 scope of a procedure are located after the @code{N_FUN} stab that
666 represents the procedure itself.
668 Sun documents the desc field of @code{N_LBRAC} and
669 @code{N_RBRAC} symbols as containing the nesting level of the block.
670 However, dbx seems to not care, and GCC always sets desc to
676 The @samp{c} symbol descriptor indicates that this stab represents a
677 constant. This symbol descriptor is an exception to the general rule
678 that symbol descriptors are followed by type information. Instead, it
679 is followed by @samp{=} and one of the following:
683 Boolean constant. @var{value} is a numeric value; I assume it is 0 for
687 Character constant. @var{value} is the numeric value of the constant.
689 @item e @var{type-information} , @var{value}
690 Constant whose value can be represented as integral.
691 @var{type-information} is the type of the constant, as it would appear
692 after a symbol descriptor (@pxref{String Field}). @var{value} is the
693 numeric value of the constant. GDB 4.9 does not actually get the right
694 value if @var{value} does not fit in a host @code{int}, but it does not
695 do anything violent, and future debuggers could be extended to accept
696 integers of any size (whether unsigned or not). This constant type is
697 usually documented as being only for enumeration constants, but GDB has
698 never imposed that restriction; I don't know about other debuggers.
701 Integer constant. @var{value} is the numeric value. The type is some
702 sort of generic integer type (for GDB, a host @code{int}); to specify
703 the type explicitly, use @samp{e} instead.
706 Real constant. @var{value} is the real value, which can be @samp{INF}
707 (optionally preceded by a sign) for infinity, @samp{QNAN} for a quiet
708 NaN (not-a-number), or @samp{SNAN} for a signalling NaN. If it is a
709 normal number the format is that accepted by the C library function
713 String constant. @var{string} is a string enclosed in either @samp{'}
714 (in which case @samp{'} characters within the string are represented as
715 @samp{\'} or @samp{"} (in which case @samp{"} characters within the
716 string are represented as @samp{\"}).
718 @item S @var{type-information} , @var{elements} , @var{bits} , @var{pattern}
719 Set constant. @var{type-information} is the type of the constant, as it
720 would appear after a symbol descriptor (@pxref{String Field}).
721 @var{elements} is the number of elements in the set (does this means
722 how many bits of @var{pattern} are actually used, which would be
723 redundant with the type, or perhaps the number of bits set in
724 @var{pattern}? I don't get it), @var{bits} is the number of bits in the
725 constant (meaning it specifies the length of @var{pattern}, I think),
726 and @var{pattern} is a hexadecimal representation of the set. AIX
727 documentation refers to a limit of 32 bytes, but I see no reason why
728 this limit should exist. This form could probably be used for arbitrary
729 constants, not just sets; the only catch is that @var{pattern} should be
730 understood to be target, not host, byte order and format.
733 The boolean, character, string, and set constants are not supported by
734 GDB 4.9, but it ignores them. GDB 4.8 and earlier gave an error
735 message and refused to read symbols from the file containing the
738 The above information is followed by @samp{;}.
743 Different types of stabs describe the various ways that variables can be
744 allocated: on the stack, globally, in registers, in common blocks,
745 statically, or as arguments to a function.
748 * Stack Variables:: Variables allocated on the stack.
749 * Global Variables:: Variables used by more than one source file.
750 * Register Variables:: Variables in registers.
751 * Common Blocks:: Variables statically allocated together.
752 * Statics:: Variables local to one source file.
753 * Based Variables:: Fortran pointer based variables.
754 * Parameters:: Variables for arguments to functions.
757 @node Stack Variables
758 @section Automatic Variables Allocated on the Stack
760 If a variable's scope is local to a function and its lifetime is only as
761 long as that function executes (C calls such variables
762 @dfn{automatic}), it can be allocated in a register (@pxref{Register
763 Variables}) or on the stack.
766 Each variable allocated on the stack has a stab with the symbol
767 descriptor omitted. Since type information should begin with a digit,
768 @samp{-}, or @samp{(}, only those characters precluded from being used
769 for symbol descriptors. However, the Acorn RISC machine (ARM) is said
770 to get this wrong: it puts out a mere type definition here, without the
771 preceding @samp{@var{type-number}=}. This is a bad idea; there is no
772 guarantee that type descriptors are distinct from symbol descriptors.
773 Stabs for stack variables use the @code{N_LSYM} stab type.
775 The value of the stab is the offset of the variable within the
776 local variables. On most machines this is an offset from the frame
777 pointer and is negative. The location of the stab specifies which block
778 it is defined in; see @ref{Block Structure}.
780 For example, the following C code:
790 produces the following stabs:
793 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
794 .stabs "x:1",128,0,0,-12 # @r{128 is N_LSYM}
795 .stabn 192,0,0,LBB2 # @r{192 is N_LBRAC}
796 .stabn 224,0,0,LBE2 # @r{224 is N_RBRAC}
799 @xref{Procedures} for more information on the @code{N_FUN} stab, and
800 @ref{Block Structure} for more information on the @code{N_LBRAC} and
801 @code{N_RBRAC} stabs.
803 @node Global Variables
804 @section Global Variables
807 A variable whose scope is not specific to just one source file is
808 represented by the @samp{G} symbol descriptor. These stabs use the
809 @code{N_GSYM} stab type. The type information for the stab
810 (@pxref{String Field}) gives the type of the variable.
812 For example, the following source code:
819 yields the following assembly code:
822 .stabs "g_foo:G2",32,0,0,0 # @r{32 is N_GSYM}
829 The address of the variable represented by the @code{N_GSYM} is not
830 contained in the @code{N_GSYM} stab. The debugger gets this information
831 from the external symbol for the global variable. In the example above,
832 the @code{.global _g_foo} and @code{_g_foo:} lines tell the assembler to
833 produce an external symbol.
835 @node Register Variables
836 @section Register Variables
839 @c According to an old version of this manual, AIX uses C_RPSYM instead
840 @c of C_RSYM. I am skeptical; this should be verified.
841 Register variables have their own stab type, @code{N_RSYM}, and their
842 own symbol descriptor, @samp{r}. The stab's value is the
843 number of the register where the variable data will be stored.
844 @c .stabs "name:type",N_RSYM,0,RegSize,RegNumber (Sun doc)
846 AIX defines a separate symbol descriptor @samp{d} for floating point
847 registers. This seems unnecessary; why not just just give floating
848 point registers different register numbers? I have not verified whether
849 the compiler actually uses @samp{d}.
851 If the register is explicitly allocated to a global variable, but not
855 register int g_bar asm ("%g5");
859 then the stab may be emitted at the end of the object file, with
860 the other bss symbols.
863 @section Common Blocks
865 A common block is a statically allocated section of memory which can be
866 referred to by several source files. It may contain several variables.
867 I believe Fortran is the only language with this feature.
873 A @code{N_BCOMM} stab begins a common block and an @code{N_ECOMM} stab
874 ends it. The only field that is significant in these two stabs is the
875 string, which names a normal (non-debugging) symbol that gives the
876 address of the common block. According to IBM documentation, only the
877 @code{N_BCOMM} has the name of the common block (even though their
878 compiler actually puts it both places).
882 The stabs for the members of the common block are between the
883 @code{N_BCOMM} and the @code{N_ECOMM}; the value of each stab is the
884 offset within the common block of that variable. IBM uses the
885 @code{C_ECOML} stab type, and there is a corresponding @code{N_ECOML}
886 stab type, but Sun's Fortran compiler uses @code{N_GSYM} instead. The
887 variables within a common block use the @samp{V} symbol descriptor (I
888 believe this is true of all Fortran variables). Other stabs (at least
889 type declarations using @code{C_DECL}) can also be between the
890 @code{N_BCOMM} and the @code{N_ECOMM}.
893 @section Static Variables
895 Initialized static variables are represented by the @samp{S} and
896 @samp{V} symbol descriptors. @samp{S} means file scope static, and
897 @samp{V} means procedure scope static.
899 @c This is probably not worth mentioning; it is only true on the sparc
900 @c for `double' variables which although declared const are actually in
901 @c the data segment (the text segment can't guarantee 8 byte alignment).
903 @c 2.4.5 has a bug in that it uses @code{N_FUN}, so neither dbx nor GDB can
904 @c find the variables)
907 @findex N_FUN, for variables
909 In a.out files, @code{N_STSYM} means the data section, @code{N_FUN}
910 means the text section, and @code{N_LCSYM} means the bss section. For
911 those systems with a read-only data section separate from the text
912 section (Solaris), @code{N_ROSYM} means the read-only data section.
914 For example, the source lines:
917 static const int var_const = 5;
918 static int var_init = 2;
919 static int var_noinit;
923 yield the following stabs:
926 .stabs "var_const:S1",36,0,0,_var_const # @r{36 is N_FUN}
928 .stabs "var_init:S1",38,0,0,_var_init # @r{38 is N_STSYM}
930 .stabs "var_noinit:S1",40,0,0,_var_noinit # @r{40 is N_LCSYM}
933 In XCOFF files, each symbol has a section number, so the stab type
934 need not indicate the section.
936 In ECOFF files, the storage class is used to specify the section, so the
937 stab type need not indicate the section.
939 In ELF files, for the SunPRO compiler version 2.0.1, symbol descriptor
940 @samp{S} means that the address is absolute (the linker relocates it)
941 and symbol descriptor @samp{V} means that the address is relative to the
942 start of the relevant section for that compilation unit. SunPRO has
943 plans to have the linker stop relocating stabs; I suspect that their the
944 debugger gets the address from the corresponding ELF (not stab) symbol.
945 I'm not sure how to find which symbol of that name is the right one.
946 The clean way to do all this would be to have a the value of a symbol
947 descriptor @samp{S} symbol be an offset relative to the start of the
948 file, just like everything else, but that introduces obvious
949 compatibility problems. For more information on linker stab relocation,
952 @node Based Variables
953 @section Fortran Based Variables
955 Fortran (at least, the Sun and SGI dialects of FORTRAN-77) has a feature
956 which allows allocating arrays with @code{malloc}, but which avoids
957 blurring the line between arrays and pointers the way that C does. In
958 stabs such a variable uses the @samp{b} symbol descriptor.
960 For example, the Fortran declarations
963 real foo, foo10(10), foo10_5(10,5)
965 pointer (foo10p, foo10)
966 pointer (foo105p, foo10_5)
974 foo10_5:bar3;1;5;ar3;1;10;6
977 In this example, @code{real} is type 6 and type 3 is an integral type
978 which is the type of the subscripts of the array (probably
981 The @samp{b} symbol descriptor is like @samp{V} in that it denotes a
982 statically allocated symbol whose scope is local to a function; see
983 @xref{Statics}. The value of the symbol, instead of being the address
984 of the variable itself, is the address of a pointer to that variable.
985 So in the above example, the value of the @code{foo} stab is the address
986 of a pointer to a real, the value of the @code{foo10} stab is the
987 address of a pointer to a 10-element array of reals, and the value of
988 the @code{foo10_5} stab is the address of a pointer to a 5-element array
989 of 10-element arrays of reals.
994 Formal parameters to a function are represented by a stab (or sometimes
995 two; see below) for each parameter. The stabs are in the order in which
996 the debugger should print the parameters (i.e., the order in which the
997 parameters are declared in the source file). The exact form of the stab
998 depends on how the parameter is being passed.
1001 Parameters passed on the stack use the symbol descriptor @samp{p} and
1002 the @code{N_PSYM} symbol type. The value of the symbol is an offset
1003 used to locate the parameter on the stack; its exact meaning is
1004 machine-dependent, but on most machines it is an offset from the frame
1007 As a simple example, the code:
1018 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
1019 .stabs "argc:p1",160,0,0,68 # @r{160 is N_PSYM}
1020 .stabs "argv:p20=*21=*2",160,0,0,72
1023 The type definition of @code{argv} is interesting because it contains
1024 several type definitions. Type 21 is pointer to type 2 (char) and
1025 @code{argv} (type 20) is pointer to type 21.
1027 @c FIXME: figure out what these mean and describe them coherently.
1028 The following symbol descriptors are also said to go with @code{N_PSYM}.
1029 The value of the symbol is said to be an offset from the argument
1030 pointer (I'm not sure whether this is true or not).
1034 pF Fortran function parameter
1035 X (function result variable)
1039 * Register Parameters::
1040 * Local Variable Parameters::
1041 * Reference Parameters::
1042 * Conformant Arrays::
1045 @node Register Parameters
1046 @subsection Passing Parameters in Registers
1048 If the parameter is passed in a register, then traditionally there are
1049 two symbols for each argument:
1052 .stabs "arg:p1" . . . ; N_PSYM
1053 .stabs "arg:r1" . . . ; N_RSYM
1056 Debuggers use the second one to find the value, and the first one to
1057 know that it is an argument.
1060 @findex N_RSYM, for parameters
1061 Because that approach is kind of ugly, some compilers use symbol
1062 descriptor @samp{P} or @samp{R} to indicate an argument which is in a
1063 register. Symbol type @code{C_RPSYM} is used with @samp{R} and
1064 @code{N_RSYM} is used with @samp{P}. The symbol's value is
1065 the register number. @samp{P} and @samp{R} mean the same thing; the
1066 difference is that @samp{P} is a GNU invention and @samp{R} is an IBM
1067 (XCOFF) invention. As of version 4.9, GDB should handle either one.
1069 There is at least one case where GCC uses a @samp{p} and @samp{r} pair
1070 rather than @samp{P}; this is where the argument is passed in the
1071 argument list and then loaded into a register.
1073 According to the AIX documentation, symbol descriptor @samp{D} is for a
1074 parameter passed in a floating point register. This seems
1075 unnecessary---why not just use @samp{R} with a register number which
1076 indicates that it's a floating point register? I haven't verified
1077 whether the system actually does what the documentation indicates.
1079 @c FIXME: On the hppa this is for any type > 8 bytes, I think, and not
1080 @c for small structures (investigate).
1081 On the sparc and hppa, for a @samp{P} symbol whose type is a structure
1082 or union, the register contains the address of the structure. On the
1083 sparc, this is also true of a @samp{p} and @samp{r} pair (using Sun
1084 @code{cc}) or a @samp{p} symbol. However, if a (small) structure is
1085 really in a register, @samp{r} is used. And, to top it all off, on the
1086 hppa it might be a structure which was passed on the stack and loaded
1087 into a register and for which there is a @samp{p} and @samp{r} pair! I
1088 believe that symbol descriptor @samp{i} is supposed to deal with this
1089 case (it is said to mean "value parameter by reference, indirect
1090 access"; I don't know the source for this information), but I don't know
1091 details or what compilers or debuggers use it, if any (not GDB or GCC).
1092 It is not clear to me whether this case needs to be dealt with
1093 differently than parameters passed by reference (@pxref{Reference Parameters}).
1095 @node Local Variable Parameters
1096 @subsection Storing Parameters as Local Variables
1098 There is a case similar to an argument in a register, which is an
1099 argument that is actually stored as a local variable. Sometimes this
1100 happens when the argument was passed in a register and then the compiler
1101 stores it as a local variable. If possible, the compiler should claim
1102 that it's in a register, but this isn't always done.
1104 If a parameter is passed as one type and converted to a smaller type by
1105 the prologue (for example, the parameter is declared as a @code{float},
1106 but the calling conventions specify that it is passed as a
1107 @code{double}), then GCC2 (sometimes) uses a pair of symbols. The first
1108 symbol uses symbol descriptor @samp{p} and the type which is passed.
1109 The second symbol has the type and location which the parameter actually
1110 has after the prologue. For example, suppose the following C code
1111 appears with no prototypes involved:
1120 if @code{f} is passed as a double at stack offset 8, and the prologue
1121 converts it to a float in register number 0, then the stabs look like:
1124 .stabs "f:p13",160,0,3,8 # @r{160 is @code{N_PSYM}, here 13 is @code{double}}
1125 .stabs "f:r12",64,0,3,0 # @r{64 is @code{N_RSYM}, here 12 is @code{float}}
1128 In both stabs 3 is the line number where @code{f} is declared
1129 (@pxref{Line Numbers}).
1131 @findex N_LSYM, for parameter
1132 GCC, at least on the 960, has another solution to the same problem. It
1133 uses a single @samp{p} symbol descriptor for an argument which is stored
1134 as a local variable but uses @code{N_LSYM} instead of @code{N_PSYM}. In
1135 this case, the value of the symbol is an offset relative to the local
1136 variables for that function, not relative to the arguments; on some
1137 machines those are the same thing, but not on all.
1139 @c This is mostly just background info; the part that logically belongs
1140 @c here is the last sentence.
1141 On the VAX or on other machines in which the calling convention includes
1142 the number of words of arguments actually passed, the debugger (GDB at
1143 least) uses the parameter symbols to keep track of whether it needs to
1144 print nameless arguments in addition to the formal parameters which it
1145 has printed because each one has a stab. For example, in
1148 extern int fprintf (FILE *stream, char *format, @dots{});
1150 fprintf (stdout, "%d\n", x);
1153 there are stabs for @code{stream} and @code{format}. On most machines,
1154 the debugger can only print those two arguments (because it has no way
1155 of knowing that additional arguments were passed), but on the VAX or
1156 other machines with a calling convention which indicates the number of
1157 words of arguments, the debugger can print all three arguments. To do
1158 so, the parameter symbol (symbol descriptor @samp{p}) (not necessarily
1159 @samp{r} or symbol descriptor omitted symbols) needs to contain the
1160 actual type as passed (for example, @code{double} not @code{float} if it
1161 is passed as a double and converted to a float).
1163 @node Reference Parameters
1164 @subsection Passing Parameters by Reference
1166 If the parameter is passed by reference (e.g., Pascal @code{VAR}
1167 parameters), then the symbol descriptor is @samp{v} if it is in the
1168 argument list, or @samp{a} if it in a register. Other than the fact
1169 that these contain the address of the parameter rather than the
1170 parameter itself, they are identical to @samp{p} and @samp{R},
1171 respectively. I believe @samp{a} is an AIX invention; @samp{v} is
1172 supported by all stabs-using systems as far as I know.
1174 @node Conformant Arrays
1175 @subsection Passing Conformant Array Parameters
1177 @c Is this paragraph correct? It is based on piecing together patchy
1178 @c information and some guesswork
1179 Conformant arrays are a feature of Modula-2, and perhaps other
1180 languages, in which the size of an array parameter is not known to the
1181 called function until run-time. Such parameters have two stabs: a
1182 @samp{x} for the array itself, and a @samp{C}, which represents the size
1183 of the array. The value of the @samp{x} stab is the offset in the
1184 argument list where the address of the array is stored (it this right?
1185 it is a guess); the value of the @samp{C} stab is the offset in the
1186 argument list where the size of the array (in elements? in bytes?) is
1190 @chapter Defining Types
1192 The examples so far have described types as references to previously
1193 defined types, or defined in terms of subranges of or pointers to
1194 previously defined types. This chapter describes the other type
1195 descriptors that may follow the @samp{=} in a type definition.
1198 * Builtin Types:: Integers, floating point, void, etc.
1199 * Miscellaneous Types:: Pointers, sets, files, etc.
1200 * Cross-References:: Referring to a type not yet defined.
1201 * Subranges:: A type with a specific range.
1202 * Arrays:: An aggregate type of same-typed elements.
1203 * Strings:: Like an array but also has a length.
1204 * Enumerations:: Like an integer but the values have names.
1205 * Structures:: An aggregate type of different-typed elements.
1206 * Typedefs:: Giving a type a name.
1207 * Unions:: Different types sharing storage.
1212 @section Builtin Types
1214 Certain types are built in (@code{int}, @code{short}, @code{void},
1215 @code{float}, etc.); the debugger recognizes these types and knows how
1216 to handle them. Thus, don't be surprised if some of the following ways
1217 of specifying builtin types do not specify everything that a debugger
1218 would need to know about the type---in some cases they merely specify
1219 enough information to distinguish the type from other types.
1221 The traditional way to define builtin types is convolunted, so new ways
1222 have been invented to describe them. Sun's @code{acc} uses special
1223 builtin type descriptors (@samp{b} and @samp{R}), and IBM uses negative
1224 type numbers. GDB accepts all three ways, as of version 4.8; dbx just
1225 accepts the traditional builtin types and perhaps one of the other two
1226 formats. The following sections describe each of these formats.
1229 * Traditional Builtin Types:: Put on your seatbelts and prepare for kludgery
1230 * Builtin Type Descriptors:: Builtin types with special type descriptors
1231 * Negative Type Numbers:: Builtin types using negative type numbers
1234 @node Traditional Builtin Types
1235 @subsection Traditional Builtin Types
1237 This is the traditional, convoluted method for defining builtin types.
1238 There are several classes of such type definitions: integer, floating
1239 point, and @code{void}.
1242 * Traditional Integer Types::
1243 * Traditional Other Types::
1246 @node Traditional Integer Types
1247 @subsubsection Traditional Integer Types
1249 Often types are defined as subranges of themselves. If the bounding values
1250 fit within an @code{int}, then they are given normally. For example:
1253 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0 # @r{128 is N_LSYM}
1254 .stabs "char:t2=r2;0;127;",128,0,0,0
1257 Builtin types can also be described as subranges of @code{int}:
1260 .stabs "unsigned short:t6=r1;0;65535;",128,0,0,0
1263 If the lower bound of a subrange is 0 and the upper bound is -1,
1264 the type is an unsigned integral type whose bounds are too
1265 big to describe in an @code{int}. Traditionally this is only used for
1266 @code{unsigned int} and @code{unsigned long}:
1269 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
1272 For larger types, GCC 2.4.5 puts out bounds in octal, with one or more
1273 leading zeroes. In this case a negative bound consists of a number
1274 which is a 1 bit (for the sign bit) followed by a 0 bit for each bit in
1275 the number (except the sign bit), and a positive bound is one which is a
1276 1 bit for each bit in the number (except possibly the sign bit). All
1277 known versions of dbx and GDB version 4 accept this (at least in the
1278 sense of not refusing to process the file), but GDB 3.5 refuses to read
1279 the whole file containing such symbols. So GCC 2.3.3 did not output the
1280 proper size for these types. As an example of octal bounds, the string
1281 fields of the stabs for 64 bit integer types look like:
1283 @c .stabs directives, etc., omitted to make it fit on the page.
1285 long int:t3=r1;001000000000000000000000;000777777777777777777777;
1286 long unsigned int:t5=r1;000000000000000000000000;001777777777777777777777;
1289 If the lower bound of a subrange is 0 and the upper bound is negative,
1290 the type is an unsigned integral type whose size in bytes is the
1291 absolute value of the upper bound. I believe this is a Convex
1292 convention for @code{unsigned long long}.
1294 If the lower bound of a subrange is negative and the upper bound is 0,
1295 the type is a signed integral type whose size in bytes is
1296 the absolute value of the lower bound. I believe this is a Convex
1297 convention for @code{long long}. To distinguish this from a legitimate
1298 subrange, the type should be a subrange of itself. I'm not sure whether
1299 this is the case for Convex.
1301 @node Traditional Other Types
1302 @subsubsection Traditional Other Types
1304 If the upper bound of a subrange is 0 and the lower bound is positive,
1305 the type is a floating point type, and the lower bound of the subrange
1306 indicates the number of bytes in the type:
1309 .stabs "float:t12=r1;4;0;",128,0,0,0
1310 .stabs "double:t13=r1;8;0;",128,0,0,0
1313 However, GCC writes @code{long double} the same way it writes
1314 @code{double}, so there is no way to distinguish.
1317 .stabs "long double:t14=r1;8;0;",128,0,0,0
1320 Complex types are defined the same way as floating-point types; there is
1321 no way to distinguish a single-precision complex from a double-precision
1322 floating-point type.
1324 The C @code{void} type is defined as itself:
1327 .stabs "void:t15=15",128,0,0,0
1330 I'm not sure how a boolean type is represented.
1332 @node Builtin Type Descriptors
1333 @subsection Defining Builtin Types Using Builtin Type Descriptors
1335 This is the method used by Sun's @code{acc} for defining builtin types.
1336 These are the type descriptors to define builtin types:
1339 @c FIXME: clean up description of width and offset, once we figure out
1341 @item b @var{signed} @var{char-flag} @var{width} ; @var{offset} ; @var{nbits} ;
1342 Define an integral type. @var{signed} is @samp{u} for unsigned or
1343 @samp{s} for signed. @var{char-flag} is @samp{c} which indicates this
1344 is a character type, or is omitted. I assume this is to distinguish an
1345 integral type from a character type of the same size, for example it
1346 might make sense to set it for the C type @code{wchar_t} so the debugger
1347 can print such variables differently (Solaris does not do this). Sun
1348 sets it on the C types @code{signed char} and @code{unsigned char} which
1349 arguably is wrong. @var{width} and @var{offset} appear to be for small
1350 objects stored in larger ones, for example a @code{short} in an
1351 @code{int} register. @var{width} is normally the number of bytes in the
1352 type. @var{offset} seems to always be zero. @var{nbits} is the number
1353 of bits in the type.
1355 Note that type descriptor @samp{b} used for builtin types conflicts with
1356 its use for Pascal space types (@pxref{Miscellaneous Types}); they can
1357 be distinguished because the character following the type descriptor
1358 will be a digit, @samp{(}, or @samp{-} for a Pascal space type, or
1359 @samp{u} or @samp{s} for a builtin type.
1362 Documented by AIX to define a wide character type, but their compiler
1363 actually uses negative type numbers (@pxref{Negative Type Numbers}).
1365 @item R @var{fp-type} ; @var{bytes} ;
1366 Define a floating point type. @var{fp-type} has one of the following values:
1370 IEEE 32-bit (single precision) floating point format.
1373 IEEE 64-bit (double precision) floating point format.
1375 @item 3 (NF_COMPLEX)
1376 @item 4 (NF_COMPLEX16)
1377 @item 5 (NF_COMPLEX32)
1378 @c "GDB source" really means @file{include/aout/stab_gnu.h}, but trying
1379 @c to put that here got an overfull hbox.
1380 These are for complex numbers. A comment in the GDB source describes
1381 them as Fortran @code{complex}, @code{double complex}, and
1382 @code{complex*16}, respectively, but what does that mean? (i.e., Single
1383 precision? Double precison?).
1385 @item 6 (NF_LDOUBLE)
1386 Long double. This should probably only be used for Sun format
1387 @code{long double}, and new codes should be used for other floating
1388 point formats (@code{NF_DOUBLE} can be used if a @code{long double} is
1389 really just an IEEE double, of course).
1392 @var{bytes} is the number of bytes occupied by the type. This allows a
1393 debugger to perform some operations with the type even if it doesn't
1394 understand @var{fp-type}.
1396 @item g @var{type-information} ; @var{nbits}
1397 Documented by AIX to define a floating type, but their compiler actually
1398 uses negative type numbers (@pxref{Negative Type Numbers}).
1400 @item c @var{type-information} ; @var{nbits}
1401 Documented by AIX to define a complex type, but their compiler actually
1402 uses negative type numbers (@pxref{Negative Type Numbers}).
1405 The C @code{void} type is defined as a signed integral type 0 bits long:
1407 .stabs "void:t19=bs0;0;0",128,0,0,0
1409 The Solaris compiler seems to omit the trailing semicolon in this case.
1410 Getting sloppy in this way is not a swift move because if a type is
1411 embedded in a more complex expression it is necessary to be able to tell
1414 I'm not sure how a boolean type is represented.
1416 @node Negative Type Numbers
1417 @subsection Negative Type Numbers
1419 This is the method used in XCOFF for defining builtin types.
1420 Since the debugger knows about the builtin types anyway, the idea of
1421 negative type numbers is simply to give a special type number which
1422 indicates the builtin type. There is no stab defining these types.
1424 There are several subtle issues with negative type numbers.
1426 One is the size of the type. A builtin type (for example the C types
1427 @code{int} or @code{long}) might have different sizes depending on
1428 compiler options, the target architecture, the ABI, etc. This issue
1429 doesn't come up for IBM tools since (so far) they just target the
1430 RS/6000; the sizes indicated below for each size are what the IBM
1431 RS/6000 tools use. To deal with differing sizes, either define separate
1432 negative type numbers for each size (which works but requires changing
1433 the debugger, and, unless you get both AIX dbx and GDB to accept the
1434 change, introduces an incompatibility), or use a type attribute
1435 (@pxref{String Field}) to define a new type with the appropriate size
1436 (which merely requires a debugger which understands type attributes,
1437 like AIX dbx). For example,
1440 .stabs "boolean:t10=@@s8;-16",128,0,0,0
1443 defines an 8-bit boolean type, and
1446 .stabs "boolean:t10=@@s64;-16",128,0,0,0
1449 defines a 64-bit boolean type.
1451 A similar issue is the format of the type. This comes up most often for
1452 floating-point types, which could have various formats (particularly
1453 extended doubles, which vary quite a bit even among IEEE systems).
1454 Again, it is best to define a new negative type number for each
1455 different format; changing the format based on the target system has
1456 various problems. One such problem is that the Alpha has both VAX and
1457 IEEE floating types. One can easily imagine one library using the VAX
1458 types and another library in the same executable using the IEEE types.
1459 Another example is that the interpretation of whether a boolean is true
1460 or false can be based on the least significant bit, most significant
1461 bit, whether it is zero, etc., and different compilers (or different
1462 options to the same compiler) might provide different kinds of boolean.
1464 The last major issue is the names of the types. The name of a given
1465 type depends @emph{only} on the negative type number given; these do not
1466 vary depending on the language, the target system, or anything else.
1467 One can always define separate type numbers---in the following list you
1468 will see for example separate @code{int} and @code{integer*4} types
1469 which are identical except for the name. But compatibility can be
1470 maintained by not inventing new negative type numbers and instead just
1471 defining a new type with a new name. For example:
1474 .stabs "CARDINAL:t10=-8",128,0,0,0
1477 Here is the list of negative type numbers. The phrase @dfn{integral
1478 type} is used to mean twos-complement (I strongly suspect that all
1479 machines which use stabs use twos-complement; most machines use
1480 twos-complement these days).
1484 @code{int}, 32 bit signed integral type.
1487 @code{char}, 8 bit type holding a character. Both GDB and dbx on AIX
1488 treat this as signed. GCC uses this type whether @code{char} is signed
1489 or not, which seems like a bad idea. The AIX compiler (@code{xlc}) seems to
1490 avoid this type; it uses -5 instead for @code{char}.
1493 @code{short}, 16 bit signed integral type.
1496 @code{long}, 32 bit signed integral type.
1499 @code{unsigned char}, 8 bit unsigned integral type.
1502 @code{signed char}, 8 bit signed integral type.
1505 @code{unsigned short}, 16 bit unsigned integral type.
1508 @code{unsigned int}, 32 bit unsigned integral type.
1511 @code{unsigned}, 32 bit unsigned integral type.
1514 @code{unsigned long}, 32 bit unsigned integral type.
1517 @code{void}, type indicating the lack of a value.
1520 @code{float}, IEEE single precision.
1523 @code{double}, IEEE double precision.
1526 @code{long double}, IEEE double precision. The compiler claims the size
1527 will increase in a future release, and for binary compatibility you have
1528 to avoid using @code{long double}. I hope when they increase it they
1529 use a new negative type number.
1532 @code{integer}. 32 bit signed integral type.
1535 @code{boolean}. 32 bit type. How is the truth value encoded? Is it
1536 the least significant bit or is it a question of whether the whole value
1537 is zero or non-zero?
1540 @code{short real}. IEEE single precision.
1543 @code{real}. IEEE double precision.
1546 @code{stringptr}. @xref{Strings}.
1549 @code{character}, 8 bit unsigned character type.
1552 @code{logical*1}, 8 bit type. This Fortran type has a split
1553 personality in that it is used for boolean variables, but can also be
1554 used for unsigned integers. 0 is false, 1 is true, and other values are
1558 @code{logical*2}, 16 bit type. This Fortran type has a split
1559 personality in that it is used for boolean variables, but can also be
1560 used for unsigned integers. 0 is false, 1 is true, and other values are
1564 @code{logical*4}, 32 bit type. This Fortran type has a split
1565 personality in that it is used for boolean variables, but can also be
1566 used for unsigned integers. 0 is false, 1 is true, and other values are
1570 @code{logical}, 32 bit type. This Fortran type has a split
1571 personality in that it is used for boolean variables, but can also be
1572 used for unsigned integers. 0 is false, 1 is true, and other values are
1576 @code{complex}. A complex type consisting of two IEEE single-precision
1577 floating point values.
1580 @code{complex}. A complex type consisting of two IEEE double-precision
1581 floating point values.
1584 @code{integer*1}, 8 bit signed integral type.
1587 @code{integer*2}, 16 bit signed integral type.
1590 @code{integer*4}, 32 bit signed integral type.
1593 @code{wchar}. Wide character, 16 bits wide, unsigned (what format?
1597 @node Miscellaneous Types
1598 @section Miscellaneous Types
1601 @item b @var{type-information} ; @var{bytes}
1602 Pascal space type. This is documented by IBM; what does it mean?
1604 This use of the @samp{b} type descriptor can be distinguished
1605 from its use for builtin integral types (@pxref{Builtin Type
1606 Descriptors}) because the character following the type descriptor is
1607 always a digit, @samp{(}, or @samp{-}.
1609 @item B @var{type-information}
1610 A volatile-qualified version of @var{type-information}. This is
1611 a Sun extension. References and stores to a variable with a
1612 volatile-qualified type must not be optimized or cached; they
1613 must occur as the user specifies them.
1615 @item d @var{type-information}
1616 File of type @var{type-information}. As far as I know this is only used
1619 @item k @var{type-information}
1620 A const-qualified version of @var{type-information}. This is a Sun
1621 extension. A variable with a const-qualified type cannot be modified.
1623 @item M @var{type-information} ; @var{length}
1624 Multiple instance type. The type seems to composed of @var{length}
1625 repetitions of @var{type-information}, for example @code{character*3} is
1626 represented by @samp{M-2;3}, where @samp{-2} is a reference to a
1627 character type (@pxref{Negative Type Numbers}). I'm not sure how this
1628 differs from an array. This appears to be a Fortran feature.
1629 @var{length} is a bound, like those in range types; see @ref{Subranges}.
1631 @item S @var{type-information}
1632 Pascal set type. @var{type-information} must be a small type such as an
1633 enumeration or a subrange, and the type is a bitmask whose length is
1634 specified by the number of elements in @var{type-information}.
1636 In CHILL, if it is a bitstring instead of a set, also use the @samp{S}
1637 type attribute (@pxref{String Field}).
1639 @item * @var{type-information}
1640 Pointer to @var{type-information}.
1643 @node Cross-References
1644 @section Cross-References to Other Types
1646 A type can be used before it is defined; one common way to deal with
1647 that situation is just to use a type reference to a type which has not
1650 Another way is with the @samp{x} type descriptor, which is followed by
1651 @samp{s} for a structure tag, @samp{u} for a union tag, or @samp{e} for
1652 a enumerator tag, followed by the name of the tag, followed by @samp{:}.
1653 If the name contains @samp{::} between a @samp{<} and @samp{>} pair (for
1654 C++ templates), such a @samp{::} does not end the name---only a single
1655 @samp{:} ends the name; see @ref{Nested Symbols}.
1657 For example, the following C declarations:
1668 .stabs "bar:G16=*17=xsfoo:",32,0,0,0
1671 Not all debuggers support the @samp{x} type descriptor, so on some
1672 machines GCC does not use it. I believe that for the above example it
1673 would just emit a reference to type 17 and never define it, but I
1674 haven't verified that.
1676 Modula-2 imported types, at least on AIX, use the @samp{i} type
1677 descriptor, which is followed by the name of the module from which the
1678 type is imported, followed by @samp{:}, followed by the name of the
1679 type. There is then optionally a comma followed by type information for
1680 the type. This differs from merely naming the type (@pxref{Typedefs}) in
1681 that it identifies the module; I don't understand whether the name of
1682 the type given here is always just the same as the name we are giving
1683 it, or whether this type descriptor is used with a nameless stab
1684 (@pxref{String Field}), or what. The symbol ends with @samp{;}.
1687 @section Subrange Types
1689 The @samp{r} type descriptor defines a type as a subrange of another
1690 type. It is followed by type information for the type of which it is a
1691 subrange, a semicolon, an integral lower bound, a semicolon, an
1692 integral upper bound, and a semicolon. The AIX documentation does not
1693 specify the trailing semicolon, in an effort to specify array indexes
1694 more cleanly, but a subrange which is not an array index has always
1695 included a trailing semicolon (@pxref{Arrays}).
1697 Instead of an integer, either bound can be one of the following:
1700 @item A @var{offset}
1701 The bound is passed by reference on the stack at offset @var{offset}
1702 from the argument list. @xref{Parameters}, for more information on such
1705 @item T @var{offset}
1706 The bound is passed by value on the stack at offset @var{offset} from
1709 @item a @var{register-number}
1710 The bound is pased by reference in register number
1711 @var{register-number}.
1713 @item t @var{register-number}
1714 The bound is passed by value in register number @var{register-number}.
1720 Subranges are also used for builtin types; see @ref{Traditional Builtin Types}.
1723 @section Array Types
1725 Arrays use the @samp{a} type descriptor. Following the type descriptor
1726 is the type of the index and the type of the array elements. If the
1727 index type is a range type, it ends in a semicolon; otherwise
1728 (for example, if it is a type reference), there does not
1729 appear to be any way to tell where the types are separated. In an
1730 effort to clean up this mess, IBM documents the two types as being
1731 separated by a semicolon, and a range type as not ending in a semicolon
1732 (but this is not right for range types which are not array indexes,
1733 @pxref{Subranges}). I think probably the best solution is to specify
1734 that a semicolon ends a range type, and that the index type and element
1735 type of an array are separated by a semicolon, but that if the index
1736 type is a range type, the extra semicolon can be omitted. GDB (at least
1737 through version 4.9) doesn't support any kind of index type other than a
1738 range anyway; I'm not sure about dbx.
1740 It is well established, and widely used, that the type of the index,
1741 unlike most types found in the stabs, is merely a type definition, not
1742 type information (@pxref{String Field}) (that is, it need not start with
1743 @samp{@var{type-number}=} if it is defining a new type). According to a
1744 comment in GDB, this is also true of the type of the array elements; it
1745 gives @samp{ar1;1;10;ar1;1;10;4} as a legitimate way to express a two
1746 dimensional array. According to AIX documentation, the element type
1747 must be type information. GDB accepts either.
1749 The type of the index is often a range type, expressed as the type
1750 descriptor @samp{r} and some parameters. It defines the size of the
1751 array. In the example below, the range @samp{r1;0;2;} defines an index
1752 type which is a subrange of type 1 (integer), with a lower bound of 0
1753 and an upper bound of 2. This defines the valid range of subscripts of
1754 a three-element C array.
1756 For example, the definition:
1759 char char_vec[3] = @{'a','b','c'@};
1763 produces the output:
1766 .stabs "char_vec:G19=ar1;0;2;2",32,0,0,0
1775 If an array is @dfn{packed}, the elements are spaced more
1776 closely than normal, saving memory at the expense of speed. For
1777 example, an array of 3-byte objects might, if unpacked, have each
1778 element aligned on a 4-byte boundary, but if packed, have no padding.
1779 One way to specify that something is packed is with type attributes
1780 (@pxref{String Field}). In the case of arrays, another is to use the
1781 @samp{P} type descriptor instead of @samp{a}. Other than specifying a
1782 packed array, @samp{P} is identical to @samp{a}.
1784 @c FIXME-what is it? A pointer?
1785 An open array is represented by the @samp{A} type descriptor followed by
1786 type information specifying the type of the array elements.
1788 @c FIXME: what is the format of this type? A pointer to a vector of pointers?
1789 An N-dimensional dynamic array is represented by
1792 D @var{dimensions} ; @var{type-information}
1795 @c Does dimensions really have this meaning? The AIX documentation
1797 @var{dimensions} is the number of dimensions; @var{type-information}
1798 specifies the type of the array elements.
1800 @c FIXME: what is the format of this type? A pointer to some offsets in
1802 A subarray of an N-dimensional array is represented by
1805 E @var{dimensions} ; @var{type-information}
1808 @c Does dimensions really have this meaning? The AIX documentation
1810 @var{dimensions} is the number of dimensions; @var{type-information}
1811 specifies the type of the array elements.
1816 Some languages, like C or the original Pascal, do not have string types,
1817 they just have related things like arrays of characters. But most
1818 Pascals and various other languages have string types, which are
1819 indicated as follows:
1822 @item n @var{type-information} ; @var{bytes}
1823 @var{bytes} is the maximum length. I'm not sure what
1824 @var{type-information} is; I suspect that it means that this is a string
1825 of @var{type-information} (thus allowing a string of integers, a string
1826 of wide characters, etc., as well as a string of characters). Not sure
1827 what the format of this type is. This is an AIX feature.
1829 @item z @var{type-information} ; @var{bytes}
1830 Just like @samp{n} except that this is a gstring, not an ordinary
1831 string. I don't know the difference.
1834 Pascal Stringptr. What is this? This is an AIX feature.
1837 Languages, such as CHILL which have a string type which is basically
1838 just an array of characters use the @samp{S} type attribute
1839 (@pxref{String Field}).
1842 @section Enumerations
1844 Enumerations are defined with the @samp{e} type descriptor.
1846 @c FIXME: Where does this information properly go? Perhaps it is
1847 @c redundant with something we already explain.
1848 The source line below declares an enumeration type at file scope.
1849 The type definition is located after the @code{N_RBRAC} that marks the end of
1850 the previous procedure's block scope, and before the @code{N_FUN} that marks
1851 the beginning of the next procedure's block scope. Therefore it does not
1852 describe a block local symbol, but a file local one.
1857 enum e_places @{first,second=3,last@};
1861 generates the following stab:
1864 .stabs "e_places:T22=efirst:0,second:3,last:4,;",128,0,0,0
1867 The symbol descriptor (@samp{T}) says that the stab describes a
1868 structure, enumeration, or union tag. The type descriptor @samp{e},
1869 following the @samp{22=} of the type definition narrows it down to an
1870 enumeration type. Following the @samp{e} is a list of the elements of
1871 the enumeration. The format is @samp{@var{name}:@var{value},}. The
1872 list of elements ends with @samp{;}. The fact that @var{value} is
1873 specified as an integer can cause problems if the value is large. GCC
1874 2.5.2 tries to output it in octal in that case with a leading zero,
1875 which is probably a good thing, although GDB 4.11 supports octal only in
1876 cases where decimal is perfectly good. Negative decimal values are
1877 supported by both GDB and dbx.
1879 There is no standard way to specify the size of an enumeration type; it
1880 is determined by the architecture (normally all enumerations types are
1881 32 bits). Type attributes can be used to specify an enumeration type of
1882 another size for debuggers which support them; see @ref{String Field}.
1887 The encoding of structures in stabs can be shown with an example.
1889 The following source code declares a structure tag and defines an
1890 instance of the structure in global scope. Then a @code{typedef} equates the
1891 structure tag with a new type. Seperate stabs are generated for the
1892 structure tag, the structure @code{typedef}, and the structure instance. The
1893 stabs for the tag and the @code{typedef} are emited when the definitions are
1894 encountered. Since the structure elements are not initialized, the
1895 stab and code for the structure variable itself is located at the end
1896 of the program in the bss section.
1903 struct s_tag* s_next;
1906 typedef struct s_tag s_typedef;
1909 The structure tag has an @code{N_LSYM} stab type because, like the
1910 enumeration, the symbol has file scope. Like the enumeration, the
1911 symbol descriptor is @samp{T}, for enumeration, structure, or tag type.
1912 The type descriptor @samp{s} following the @samp{16=} of the type
1913 definition narrows the symbol type to structure.
1915 Following the @samp{s} type descriptor is the number of bytes the
1916 structure occupies, followed by a description of each structure element.
1917 The structure element descriptions are of the form @var{name:type, bit
1918 offset from the start of the struct, number of bits in the element}.
1920 @c FIXME: phony line break. Can probably be fixed by using an example
1921 @c with fewer fields.
1924 .stabs "s_tag:T16=s20s_int:1,0,32;s_float:12,32,32;
1925 s_char_vec:17=ar1;0;7;2,64,64;s_next:18=*16,128,32;;",128,0,0,0
1928 In this example, the first two structure elements are previously defined
1929 types. For these, the type following the @samp{@var{name}:} part of the
1930 element description is a simple type reference. The other two structure
1931 elements are new types. In this case there is a type definition
1932 embedded after the @samp{@var{name}:}. The type definition for the
1933 array element looks just like a type definition for a standalone array.
1934 The @code{s_next} field is a pointer to the same kind of structure that
1935 the field is an element of. So the definition of structure type 16
1936 contains a type definition for an element which is a pointer to type 16.
1938 If a field is a static member (this is a C++ feature in which a single
1939 variable appears to be a field of every structure of a given type) it
1940 still starts out with the field name, a colon, and the type, but then
1941 instead of a comma, bit position, comma, and bit size, there is a colon
1942 followed by the name of the variable which each such field refers to.
1944 If the structure has methods (a C++ feature), they follow the non-method
1945 fields; see @ref{Cplusplus}.
1948 @section Giving a Type a Name
1950 To give a type a name, use the @samp{t} symbol descriptor. The type
1951 is specified by the type information (@pxref{String Field}) for the stab.
1955 .stabs "s_typedef:t16",128,0,0,0 # @r{128 is N_LSYM}
1958 specifies that @code{s_typedef} refers to type number 16. Such stabs
1959 have symbol type @code{N_LSYM} (or @code{C_DECL} for XCOFF).
1961 If you are specifying the tag name for a structure, union, or
1962 enumeration, use the @samp{T} symbol descriptor instead. I believe C is
1963 the only language with this feature.
1965 If the type is an opaque type (I believe this is a Modula-2 feature),
1966 AIX provides a type descriptor to specify it. The type descriptor is
1967 @samp{o} and is followed by a name. I don't know what the name
1968 means---is it always the same as the name of the type, or is this type
1969 descriptor used with a nameless stab (@pxref{String Field})? There
1970 optionally follows a comma followed by type information which defines
1971 the type of this type. If omitted, a semicolon is used in place of the
1972 comma and the type information, and the type is much like a generic
1973 pointer type---it has a known size but little else about it is
1987 This code generates a stab for a union tag and a stab for a union
1988 variable. Both use the @code{N_LSYM} stab type. If a union variable is
1989 scoped locally to the procedure in which it is defined, its stab is
1990 located immediately preceding the @code{N_LBRAC} for the procedure's block
1993 The stab for the union tag, however, is located preceding the code for
1994 the procedure in which it is defined. The stab type is @code{N_LSYM}. This
1995 would seem to imply that the union type is file scope, like the struct
1996 type @code{s_tag}. This is not true. The contents and position of the stab
1997 for @code{u_type} do not convey any infomation about its procedure local
2000 @c FIXME: phony line break. Can probably be fixed by using an example
2001 @c with fewer fields.
2004 .stabs "u_tag:T23=u4u_int:1,0,32;u_float:12,0,32;u_char:21,0,32;;",
2008 The symbol descriptor @samp{T}, following the @samp{name:} means that
2009 the stab describes an enumeration, structure, or union tag. The type
2010 descriptor @samp{u}, following the @samp{23=} of the type definition,
2011 narrows it down to a union type definition. Following the @samp{u} is
2012 the number of bytes in the union. After that is a list of union element
2013 descriptions. Their format is @var{name:type, bit offset into the
2014 union, number of bytes for the element;}.
2016 The stab for the union variable is:
2019 .stabs "an_u:23",128,0,0,-20 # @r{128 is N_LSYM}
2022 @samp{-20} specifies where the variable is stored (@pxref{Stack
2025 @node Function Types
2026 @section Function Types
2028 Various types can be defined for function variables. These types are
2029 not used in defining functions (@pxref{Procedures}); they are used for
2030 things like pointers to functions.
2032 The simple, traditional, type is type descriptor @samp{f} is followed by
2033 type information for the return type of the function, followed by a
2036 This does not deal with functions for which the number and types of the
2037 parameters are part of the type, as in Modula-2 or ANSI C. AIX provides
2038 extensions to specify these, using the @samp{f}, @samp{F}, @samp{p}, and
2039 @samp{R} type descriptors.
2041 First comes the type descriptor. If it is @samp{f} or @samp{F}, this
2042 type involves a function rather than a procedure, and the type
2043 information for the return type of the function follows, followed by a
2044 comma. Then comes the number of parameters to the function and a
2045 semicolon. Then, for each parameter, there is the name of the parameter
2046 followed by a colon (this is only present for type descriptors @samp{R}
2047 and @samp{F} which represent Pascal function or procedure parameters),
2048 type information for the parameter, a comma, 0 if passed by reference or
2049 1 if passed by value, and a semicolon. The type definition ends with a
2052 For example, this variable definition:
2059 generates the following code:
2062 .stabs "g_pf:G24=*25=f1",32,0,0,0
2063 .common _g_pf,4,"bss"
2066 The variable defines a new type, 24, which is a pointer to another new
2067 type, 25, which is a function returning @code{int}.
2070 @chapter Symbol Information in Symbol Tables
2072 This chapter describes the format of symbol table entries
2073 and how stab assembler directives map to them. It also describes the
2074 transformations that the assembler and linker make on data from stabs.
2077 * Symbol Table Format::
2078 * Transformations On Symbol Tables::
2081 @node Symbol Table Format
2082 @section Symbol Table Format
2084 Each time the assembler encounters a stab directive, it puts
2085 each field of the stab into a corresponding field in a symbol table
2086 entry of its output file. If the stab contains a string field, the
2087 symbol table entry for that stab points to a string table entry
2088 containing the string data from the stab. Assembler labels become
2089 relocatable addresses. Symbol table entries in a.out have the format:
2091 @c FIXME: should refer to external, not internal.
2093 struct internal_nlist @{
2094 unsigned long n_strx; /* index into string table of name */
2095 unsigned char n_type; /* type of symbol */
2096 unsigned char n_other; /* misc info (usually empty) */
2097 unsigned short n_desc; /* description field */
2098 bfd_vma n_value; /* value of symbol */
2102 If the stab has a string, the @code{n_strx} field holds the offset in
2103 bytes of the string within the string table. The string is terminated
2104 by a NUL character. If the stab lacks a string (for example, it was
2105 produced by a @code{.stabn} or @code{.stabd} directive), the
2106 @code{n_strx} field is zero.
2108 Symbol table entries with @code{n_type} field values greater than 0x1f
2109 originated as stabs generated by the compiler (with one random
2110 exception). The other entries were placed in the symbol table of the
2111 executable by the assembler or the linker.
2113 @node Transformations On Symbol Tables
2114 @section Transformations on Symbol Tables
2116 The linker concatenates object files and does fixups of externally
2119 You can see the transformations made on stab data by the assembler and
2120 linker by examining the symbol table after each pass of the build. To
2121 do this, use @samp{nm -ap}, which dumps the symbol table, including
2122 debugging information, unsorted. For stab entries the columns are:
2123 @var{value}, @var{other}, @var{desc}, @var{type}, @var{string}. For
2124 assembler and linker symbols, the columns are: @var{value}, @var{type},
2127 The low 5 bits of the stab type tell the linker how to relocate the
2128 value of the stab. Thus for stab types like @code{N_RSYM} and
2129 @code{N_LSYM}, where the value is an offset or a register number, the
2130 low 5 bits are @code{N_ABS}, which tells the linker not to relocate the
2133 Where the value of a stab contains an assembly language label,
2134 it is transformed by each build step. The assembler turns it into a
2135 relocatable address and the linker turns it into an absolute address.
2138 * Transformations On Static Variables::
2139 * Transformations On Global Variables::
2140 * ELF and SOM Transformations:: In ELF, things are a bit different.
2143 @node Transformations On Static Variables
2144 @subsection Transformations on Static Variables
2146 This source line defines a static variable at file scope:
2149 static int s_g_repeat
2153 The following stab describes the symbol:
2156 .stabs "s_g_repeat:S1",38,0,0,_s_g_repeat
2160 The assembler transforms the stab into this symbol table entry in the
2161 @file{.o} file. The location is expressed as a data segment offset.
2164 00000084 - 00 0000 STSYM s_g_repeat:S1
2168 In the symbol table entry from the executable, the linker has made the
2169 relocatable address absolute.
2172 0000e00c - 00 0000 STSYM s_g_repeat:S1
2175 @node Transformations On Global Variables
2176 @subsection Transformations on Global Variables
2178 Stabs for global variables do not contain location information. In
2179 this case, the debugger finds location information in the assembler or
2180 linker symbol table entry describing the variable. The source line:
2190 .stabs "g_foo:G2",32,0,0,0
2193 The variable is represented by two symbol table entries in the object
2194 file (see below). The first one originated as a stab. The second one
2195 is an external symbol. The upper case @samp{D} signifies that the
2196 @code{n_type} field of the symbol table contains 7, @code{N_DATA} with
2197 local linkage. The stab's value is zero since the value is not used for
2198 @code{N_GSYM} stabs. The value of the linker symbol is the relocatable
2199 address corresponding to the variable.
2202 00000000 - 00 0000 GSYM g_foo:G2
2207 These entries as transformed by the linker. The linker symbol table
2208 entry now holds an absolute address:
2211 00000000 - 00 0000 GSYM g_foo:G2
2216 @node ELF and SOM Transformations
2217 @subsection Transformations of Stabs in ELF and SOM Files
2219 For ELF and SOM files, use @code{objdump --stabs} instead of @code{nm} to show
2220 the stabs in an object or executable file. @code{objdump} is a GNU
2221 utility; Sun does not provide any equivalent.
2223 The following example is for a stab whose value is an address is
2224 relative to the compilation unit (@pxref{Stabs In ELF}). For example,
2231 appears within a function, then the assembly language output from the
2237 .stabs "ld:V(0,3)",0x26,0,4,.L18-Ddata.data # @r{0x26 is N_STSYM}
2244 Because the value is formed by subtracting one symbol from another, the
2245 value is absolute, not relocatable, and so the object file contains
2248 Symnum n_type n_othr n_desc n_value n_strx String
2249 31 STSYM 0 4 00000004 680 ld:V(0,3)
2252 without any relocations, and the executable file also contains
2255 Symnum n_type n_othr n_desc n_value n_strx String
2256 31 STSYM 0 4 00000004 680 ld:V(0,3)
2260 @chapter GNU C++ Stabs
2263 * Class Names:: C++ class names are both tags and typedefs.
2264 * Nested Symbols:: C++ symbol names can be within other types.
2265 * Basic Cplusplus Types::
2268 * Methods:: Method definition
2270 * Method Modifiers::
2273 * Virtual Base Classes::
2277 Type descriptors added for C++ descriptions:
2281 method type (@code{##} if minimal debug)
2284 Member (class and variable) type. It is followed by type information
2285 for the offset basetype, a comma, and type information for the type of
2286 the field being pointed to. (FIXME: this is acknowledged to be
2287 gibberish. Can anyone say what really goes here?).
2289 Note that there is a conflict between this and type attributes
2290 (@pxref{String Field}); both use type descriptor @samp{@@}.
2291 Fortunately, the @samp{@@} type descriptor used in this C++ sense always
2292 will be followed by a digit, @samp{(}, or @samp{-}, and type attributes
2293 never start with those things.
2297 @section C++ Class Names
2299 In C++, a class name which is declared with @code{class}, @code{struct},
2300 or @code{union}, is not only a tag, as in C, but also a type name. Thus
2301 there should be stabs with both @samp{t} and @samp{T} symbol descriptors
2304 To save space, there is a special abbreviation for this case. If the
2305 @samp{T} symbol descriptor is followed by @samp{t}, then the stab
2306 defines both a type name and a tag.
2308 For example, the C++ code
2311 struct foo @{int x;@};
2314 can be represented as either
2317 .stabs "foo:T19=s4x:1,0,32;;",128,0,0,0 # @r{128 is N_LSYM}
2318 .stabs "foo:t19",128,0,0,0
2324 .stabs "foo:Tt19=s4x:1,0,32;;",128,0,0,0
2327 @node Nested Symbols
2328 @section Defining a Symbol Within Another Type
2330 In C++, a symbol (such as a type name) can be defined within another type.
2331 @c FIXME: Needs example.
2333 In stabs, this is sometimes represented by making the name of a symbol
2334 which contains @samp{::}. Such a pair of colons does not end the name
2335 of the symbol, the way a single colon would (@pxref{String Field}). I'm
2336 not sure how consistently used or well thought out this mechanism is.
2337 So that a pair of colons in this position always has this meaning,
2338 @samp{:} cannot be used as a symbol descriptor.
2340 For example, if the string for a stab is @samp{foo::bar::baz:t5=*6},
2341 then @code{foo::bar::baz} is the name of the symbol, @samp{t} is the
2342 symbol descriptor, and @samp{5=*6} is the type information.
2344 @node Basic Cplusplus Types
2345 @section Basic Types For C++
2347 << the examples that follow are based on a01.C >>
2350 C++ adds two more builtin types to the set defined for C. These are
2351 the unknown type and the vtable record type. The unknown type, type
2352 16, is defined in terms of itself like the void type.
2354 The vtable record type, type 17, is defined as a structure type and
2355 then as a structure tag. The structure has four fields: delta, index,
2356 pfn, and delta2. pfn is the function pointer.
2358 << In boilerplate $vtbl_ptr_type, what are the fields delta,
2359 index, and delta2 used for? >>
2361 This basic type is present in all C++ programs even if there are no
2362 virtual methods defined.
2365 .stabs "struct_name:sym_desc(type)type_def(17)=type_desc(struct)struct_bytes(8)
2366 elem_name(delta):type_ref(short int),bit_offset(0),field_bits(16);
2367 elem_name(index):type_ref(short int),bit_offset(16),field_bits(16);
2368 elem_name(pfn):type_def(18)=type_desc(ptr to)type_ref(void),
2369 bit_offset(32),field_bits(32);
2370 elem_name(delta2):type_def(short int);bit_offset(32),field_bits(16);;"
2375 .stabs "$vtbl_ptr_type:t17=s8
2376 delta:6,0,16;index:6,16,16;pfn:18=*15,32,32;delta2:6,32,16;;"
2381 .stabs "name:sym_dec(struct tag)type_ref($vtbl_ptr_type)",N_LSYM,NIL,NIL,NIL
2385 .stabs "$vtbl_ptr_type:T17",128,0,0,0
2388 @node Simple Classes
2389 @section Simple Class Definition
2391 The stabs describing C++ language features are an extension of the
2392 stabs describing C. Stabs representing C++ class types elaborate
2393 extensively on the stab format used to describe structure types in C.
2394 Stabs representing class type variables look just like stabs
2395 representing C language variables.
2397 Consider the following very simple class definition.
2403 int Ameth(int in, char other);
2407 The class @code{baseA} is represented by two stabs. The first stab describes
2408 the class as a structure type. The second stab describes a structure
2409 tag of the class type. Both stabs are of stab type @code{N_LSYM}. Since the
2410 stab is not located between an @code{N_FUN} and an @code{N_LBRAC} stab this indicates
2411 that the class is defined at file scope. If it were, then the @code{N_LSYM}
2412 would signify a local variable.
2414 A stab describing a C++ class type is similar in format to a stab
2415 describing a C struct, with each class member shown as a field in the
2416 structure. The part of the struct format describing fields is
2417 expanded to include extra information relevent to C++ class members.
2418 In addition, if the class has multiple base classes or virtual
2419 functions the struct format outside of the field parts is also
2422 In this simple example the field part of the C++ class stab
2423 representing member data looks just like the field part of a C struct
2424 stab. The section on protections describes how its format is
2425 sometimes extended for member data.
2427 The field part of a C++ class stab representing a member function
2428 differs substantially from the field part of a C struct stab. It
2429 still begins with @samp{name:} but then goes on to define a new type number
2430 for the member function, describe its return type, its argument types,
2431 its protection level, any qualifiers applied to the method definition,
2432 and whether the method is virtual or not. If the method is virtual
2433 then the method description goes on to give the vtable index of the
2434 method, and the type number of the first base class defining the
2437 When the field name is a method name it is followed by two colons rather
2438 than one. This is followed by a new type definition for the method.
2439 This is a number followed by an equal sign and the type descriptor
2440 @samp{#}, indicating a method type, and a second @samp{#}, indicating
2441 that this is the @dfn{minimal} type of method definition used by GCC2,
2442 not larger method definitions used by earlier versions of GCC. This is
2443 followed by a type reference showing the return type of the method and a
2446 The format of an overloaded operator method name differs from that of
2447 other methods. It is @samp{op$::@var{operator-name}.} where
2448 @var{operator-name} is the operator name such as @samp{+} or @samp{+=}.
2449 The name ends with a period, and any characters except the period can
2450 occur in the @var{operator-name} string.
2452 The next part of the method description represents the arguments to the
2453 method, preceeded by a colon and ending with a semi-colon. The types of
2454 the arguments are expressed in the same way argument types are expressed
2455 in C++ name mangling. In this example an @code{int} and a @code{char}
2458 This is followed by a number, a letter, and an asterisk or period,
2459 followed by another semicolon. The number indicates the protections
2460 that apply to the member function. Here the 2 means public. The
2461 letter encodes any qualifier applied to the method definition. In
2462 this case, @samp{A} means that it is a normal function definition. The dot
2463 shows that the method is not virtual. The sections that follow
2464 elaborate further on these fields and describe the additional
2465 information present for virtual methods.
2469 .stabs "class_name:sym_desc(type)type_def(20)=type_desc(struct)struct_bytes(4)
2470 field_name(Adat):type(int),bit_offset(0),field_bits(32);
2472 method_name(Ameth)::type_def(21)=type_desc(method)return_type(int);
2473 :arg_types(int char);
2474 protection(public)qualifier(normal)virtual(no);;"
2479 .stabs "baseA:t20=s4Adat:1,0,32;Ameth::21=##1;:ic;2A.;;",128,0,0,0
2481 .stabs "class_name:sym_desc(struct tag)",N_LSYM,NIL,NIL,NIL
2483 .stabs "baseA:T20",128,0,0,0
2486 @node Class Instance
2487 @section Class Instance
2489 As shown above, describing even a simple C++ class definition is
2490 accomplished by massively extending the stab format used in C to
2491 describe structure types. However, once the class is defined, C stabs
2492 with no modifications can be used to describe class instances. The
2502 yields the following stab describing the class instance. It looks no
2503 different from a standard C stab describing a local variable.
2506 .stabs "name:type_ref(baseA)", N_LSYM, NIL, NIL, frame_ptr_offset
2510 .stabs "AbaseA:20",128,0,0,-20
2514 @section Method Definition
2516 The class definition shown above declares Ameth. The C++ source below
2521 baseA::Ameth(int in, char other)
2528 This method definition yields three stabs following the code of the
2529 method. One stab describes the method itself and following two describe
2530 its parameters. Although there is only one formal argument all methods
2531 have an implicit argument which is the @code{this} pointer. The @code{this}
2532 pointer is a pointer to the object on which the method was called. Note
2533 that the method name is mangled to encode the class name and argument
2534 types. Name mangling is described in the @sc{arm} (@cite{The Annotated
2535 C++ Reference Manual}, by Ellis and Stroustrup, @sc{isbn}
2536 0-201-51459-1); @file{gpcompare.texi} in Cygnus GCC distributions
2537 describes the differences between GNU mangling and @sc{arm}
2539 @c FIXME: Use @xref, especially if this is generally installed in the
2541 @c FIXME: This information should be in a net release, either of GCC or
2542 @c GDB. But gpcompare.texi doesn't seem to be in the FSF GCC.
2545 .stabs "name:symbol_desriptor(global function)return_type(int)",
2546 N_FUN, NIL, NIL, code_addr_of_method_start
2548 .stabs "Ameth__5baseAic:F1",36,0,0,_Ameth__5baseAic
2551 Here is the stab for the @code{this} pointer implicit argument. The
2552 name of the @code{this} pointer is always @code{this}. Type 19, the
2553 @code{this} pointer is defined as a pointer to type 20, @code{baseA},
2554 but a stab defining @code{baseA} has not yet been emited. Since the
2555 compiler knows it will be emited shortly, here it just outputs a cross
2556 reference to the undefined symbol, by prefixing the symbol name with
2560 .stabs "name:sym_desc(register param)type_def(19)=
2561 type_desc(ptr to)type_ref(baseA)=
2562 type_desc(cross-reference to)baseA:",N_RSYM,NIL,NIL,register_number
2564 .stabs "this:P19=*20=xsbaseA:",64,0,0,8
2567 The stab for the explicit integer argument looks just like a parameter
2568 to a C function. The last field of the stab is the offset from the
2569 argument pointer, which in most systems is the same as the frame
2573 .stabs "name:sym_desc(value parameter)type_ref(int)",
2574 N_PSYM,NIL,NIL,offset_from_arg_ptr
2576 .stabs "in:p1",160,0,0,72
2579 << The examples that follow are based on A1.C >>
2582 @section Protections
2584 In the simple class definition shown above all member data and
2585 functions were publicly accessable. The example that follows
2586 contrasts public, protected and privately accessable fields and shows
2587 how these protections are encoded in C++ stabs.
2589 If the character following the @samp{@var{field-name}:} part of the
2590 string is @samp{/}, then the next character is the visibility. @samp{0}
2591 means private, @samp{1} means protected, and @samp{2} means public.
2592 Debuggers should ignore visibility characters they do not recognize, and
2593 assume a reasonable default (such as public) (GDB 4.11 does not, but
2594 this should be fixed in the next GDB release). If no visibility is
2595 specified the field is public. The visibility @samp{9} means that the
2596 field has been optimized out and is public (there is no way to specify
2597 an optimized out field with a private or protected visibility).
2598 Visibility @samp{9} is not supported by GDB 4.11; this should be fixed
2599 in the next GDB release.
2601 The following C++ source:
2615 generates the following stab:
2619 .stabs "vis:T19=s12priv:/01,0,32;prot:/12,32,8;pub:12,64,32;;",128,0,0,0
2622 @samp{vis:T19=s12} indicates that type number 19 is a 12 byte structure
2623 named @code{vis} The @code{priv} field has public visibility
2624 (@samp{/0}), type int (@samp{1}), and offset and size @samp{,0,32;}.
2625 The @code{prot} field has protected visibility (@samp{/1}), type char
2626 (@samp{2}) and offset and size @samp{,32,8;}. The @code{pub} field has
2627 type float (@samp{12}), and offset and size @samp{,64,32;}.
2629 Protections for member functions are signified by one digit embeded in
2630 the field part of the stab describing the method. The digit is 0 if
2631 private, 1 if protected and 2 if public. Consider the C++ class
2635 class all_methods @{
2637 int priv_meth(int in)@{return in;@};
2639 char protMeth(char in)@{return in;@};
2641 float pubMeth(float in)@{return in;@};
2645 It generates the following stab. The digit in question is to the left
2646 of an @samp{A} in each case. Notice also that in this case two symbol
2647 descriptors apply to the class name struct tag and struct type.
2650 .stabs "class_name:sym_desc(struct tag&type)type_def(21)=
2651 sym_desc(struct)struct_bytes(1)
2652 meth_name::type_def(22)=sym_desc(method)returning(int);
2653 :args(int);protection(private)modifier(normal)virtual(no);
2654 meth_name::type_def(23)=sym_desc(method)returning(char);
2655 :args(char);protection(protected)modifier(normal)virual(no);
2656 meth_name::type_def(24)=sym_desc(method)returning(float);
2657 :args(float);protection(public)modifier(normal)virtual(no);;",
2662 .stabs "all_methods:Tt21=s1priv_meth::22=##1;:i;0A.;protMeth::23=##2;:c;1A.;
2663 pubMeth::24=##12;:f;2A.;;",128,0,0,0
2666 @node Method Modifiers
2667 @section Method Modifiers (@code{const}, @code{volatile}, @code{const volatile})
2671 In the class example described above all the methods have the normal
2672 modifier. This method modifier information is located just after the
2673 protection information for the method. This field has four possible
2674 character values. Normal methods use @samp{A}, const methods use
2675 @samp{B}, volatile methods use @samp{C}, and const volatile methods use
2676 @samp{D}. Consider the class definition below:
2681 int ConstMeth (int arg) const @{ return arg; @};
2682 char VolatileMeth (char arg) volatile @{ return arg; @};
2683 float ConstVolMeth (float arg) const volatile @{return arg; @};
2687 This class is described by the following stab:
2690 .stabs "class(A):sym_desc(struct)type_def(20)=type_desc(struct)struct_bytes(1)
2691 meth_name(ConstMeth)::type_def(21)sym_desc(method)
2692 returning(int);:arg(int);protection(public)modifier(const)virtual(no);
2693 meth_name(VolatileMeth)::type_def(22)=sym_desc(method)
2694 returning(char);:arg(char);protection(public)modifier(volatile)virt(no)
2695 meth_name(ConstVolMeth)::type_def(23)=sym_desc(method)
2696 returning(float);:arg(float);protection(public)modifer(const volatile)
2697 virtual(no);;", @dots{}
2701 .stabs "A:T20=s1ConstMeth::21=##1;:i;2B.;VolatileMeth::22=##2;:c;2C.;
2702 ConstVolMeth::23=##12;:f;2D.;;",128,0,0,0
2705 @node Virtual Methods
2706 @section Virtual Methods
2708 << The following examples are based on a4.C >>
2710 The presence of virtual methods in a class definition adds additional
2711 data to the class description. The extra data is appended to the
2712 description of the virtual method and to the end of the class
2713 description. Consider the class definition below:
2719 virtual int A_virt (int arg) @{ return arg; @};
2723 This results in the stab below describing class A. It defines a new
2724 type (20) which is an 8 byte structure. The first field of the class
2725 struct is @samp{Adat}, an integer, starting at structure offset 0 and
2728 The second field in the class struct is not explicitly defined by the
2729 C++ class definition but is implied by the fact that the class
2730 contains a virtual method. This field is the vtable pointer. The
2731 name of the vtable pointer field starts with @samp{$vf} and continues with a
2732 type reference to the class it is part of. In this example the type
2733 reference for class A is 20 so the name of its vtable pointer field is
2734 @samp{$vf20}, followed by the usual colon.
2736 Next there is a type definition for the vtable pointer type (21).
2737 This is in turn defined as a pointer to another new type (22).
2739 Type 22 is the vtable itself, which is defined as an array, indexed by
2740 a range of integers between 0 and 1, and whose elements are of type
2741 17. Type 17 was the vtable record type defined by the boilerplate C++
2742 type definitions, as shown earlier.
2744 The bit offset of the vtable pointer field is 32. The number of bits
2745 in the field are not specified when the field is a vtable pointer.
2747 Next is the method definition for the virtual member function @code{A_virt}.
2748 Its description starts out using the same format as the non-virtual
2749 member functions described above, except instead of a dot after the
2750 @samp{A} there is an asterisk, indicating that the function is virtual.
2751 Since is is virtual some addition information is appended to the end
2752 of the method description.
2754 The first number represents the vtable index of the method. This is a
2755 32 bit unsigned number with the high bit set, followed by a
2758 The second number is a type reference to the first base class in the
2759 inheritence hierarchy defining the virtual member function. In this
2760 case the class stab describes a base class so the virtual function is
2761 not overriding any other definition of the method. Therefore the
2762 reference is to the type number of the class that the stab is
2765 This is followed by three semi-colons. One marks the end of the
2766 current sub-section, one marks the end of the method field, and the
2767 third marks the end of the struct definition.
2769 For classes containing virtual functions the very last section of the
2770 string part of the stab holds a type reference to the first base
2771 class. This is preceeded by @samp{~%} and followed by a final semi-colon.
2774 .stabs "class_name(A):type_def(20)=sym_desc(struct)struct_bytes(8)
2775 field_name(Adat):type_ref(int),bit_offset(0),field_bits(32);
2776 field_name(A virt func ptr):type_def(21)=type_desc(ptr to)type_def(22)=
2777 sym_desc(array)index_type_ref(range of int from 0 to 1);
2778 elem_type_ref(vtbl elem type),
2780 meth_name(A_virt)::typedef(23)=sym_desc(method)returning(int);
2781 :arg_type(int),protection(public)normal(yes)virtual(yes)
2782 vtable_index(1);class_first_defining(A);;;~%first_base(A);",
2786 @c FIXME: bogus line break.
2788 .stabs "A:t20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
2789 A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
2793 @section Inheritence
2795 Stabs describing C++ derived classes include additional sections that
2796 describe the inheritence hierarchy of the class. A derived class stab
2797 also encodes the number of base classes. For each base class it tells
2798 if the base class is virtual or not, and if the inheritence is private
2799 or public. It also gives the offset into the object of the portion of
2800 the object corresponding to each base class.
2802 This additional information is embeded in the class stab following the
2803 number of bytes in the struct. First the number of base classes
2804 appears bracketed by an exclamation point and a comma.
2806 Then for each base type there repeats a series: a virtual character, a
2807 visibilty character, a number, a comma, another number, and a
2810 The virtual character is @samp{1} if the base class is virtual and
2811 @samp{0} if not. The visibility character is @samp{2} if the derivation
2812 is public, @samp{1} if it is protected, and @samp{0} if it is private.
2813 Debuggers should ignore virtual or visibility characters they do not
2814 recognize, and assume a reasonable default (such as public and
2815 non-virtual) (GDB 4.11 does not, but this should be fixed in the next
2818 The number following the virtual and visibility characters is the offset
2819 from the start of the object to the part of the object pertaining to the
2822 After the comma, the second number is a type_descriptor for the base
2823 type. Finally a semi-colon ends the series, which repeats for each
2826 The source below defines three base classes @code{A}, @code{B}, and
2827 @code{C} and the derived class @code{D}.
2834 virtual int A_virt (int arg) @{ return arg; @};
2840 virtual int B_virt (int arg) @{return arg; @};
2846 virtual int C_virt (int arg) @{return arg; @};
2849 class D : A, virtual B, public C @{
2852 virtual int A_virt (int arg ) @{ return arg+1; @};
2853 virtual int B_virt (int arg) @{ return arg+2; @};
2854 virtual int C_virt (int arg) @{ return arg+3; @};
2855 virtual int D_virt (int arg) @{ return arg; @};
2859 Class stabs similar to the ones described earlier are generated for
2862 @c FIXME!!! the linebreaks in the following example probably make the
2863 @c examples literally unusable, but I don't know any other way to get
2864 @c them on the page.
2865 @c One solution would be to put some of the type definitions into
2866 @c separate stabs, even if that's not exactly what the compiler actually
2869 .stabs "A:T20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
2870 A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
2872 .stabs "B:Tt25=s8Bdat:1,0,32;$vf25:21,32;B_virt::26=##1;
2873 :i;2A*-2147483647;25;;;~%25;",128,0,0,0
2875 .stabs "C:Tt28=s8Cdat:1,0,32;$vf28:21,32;C_virt::29=##1;
2876 :i;2A*-2147483647;28;;;~%28;",128,0,0,0
2879 In the stab describing derived class @code{D} below, the information about
2880 the derivation of this class is encoded as follows.
2883 .stabs "derived_class_name:symbol_descriptors(struct tag&type)=
2884 type_descriptor(struct)struct_bytes(32)!num_bases(3),
2885 base_virtual(no)inheritence_public(no)base_offset(0),
2886 base_class_type_ref(A);
2887 base_virtual(yes)inheritence_public(no)base_offset(NIL),
2888 base_class_type_ref(B);
2889 base_virtual(no)inheritence_public(yes)base_offset(64),
2890 base_class_type_ref(C); @dots{}
2893 @c FIXME! fake linebreaks.
2895 .stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:
2896 1,160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt:
2897 :32:i;2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;
2898 28;;D_virt::32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
2901 @node Virtual Base Classes
2902 @section Virtual Base Classes
2904 A derived class object consists of a concatination in memory of the data
2905 areas defined by each base class, starting with the leftmost and ending
2906 with the rightmost in the list of base classes. The exception to this
2907 rule is for virtual inheritence. In the example above, class @code{D}
2908 inherits virtually from base class @code{B}. This means that an
2909 instance of a @code{D} object will not contain its own @code{B} part but
2910 merely a pointer to a @code{B} part, known as a virtual base pointer.
2912 In a derived class stab, the base offset part of the derivation
2913 information, described above, shows how the base class parts are
2914 ordered. The base offset for a virtual base class is always given as 0.
2915 Notice that the base offset for @code{B} is given as 0 even though
2916 @code{B} is not the first base class. The first base class @code{A}
2919 The field information part of the stab for class @code{D} describes the field
2920 which is the pointer to the virtual base class @code{B}. The vbase pointer
2921 name is @samp{$vb} followed by a type reference to the virtual base class.
2922 Since the type id for @code{B} in this example is 25, the vbase pointer name
2925 @c FIXME!! fake linebreaks below
2927 .stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:1,
2928 160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt::32:i;
2929 2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;28;;D_virt:
2930 :32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
2933 Following the name and a semicolon is a type reference describing the
2934 type of the virtual base class pointer, in this case 24. Type 24 was
2935 defined earlier as the type of the @code{B} class @code{this} pointer. The
2936 @code{this} pointer for a class is a pointer to the class type.
2939 .stabs "this:P24=*25=xsB:",64,0,0,8
2942 Finally the field offset part of the vbase pointer field description
2943 shows that the vbase pointer is the first field in the @code{D} object,
2944 before any data fields defined by the class. The layout of a @code{D}
2945 class object is a follows, @code{Adat} at 0, the vtable pointer for
2946 @code{A} at 32, @code{Cdat} at 64, the vtable pointer for C at 96, the
2947 virtual base pointer for @code{B} at 128, and @code{Ddat} at 160.
2950 @node Static Members
2951 @section Static Members
2953 The data area for a class is a concatenation of the space used by the
2954 data members of the class. If the class has virtual methods, a vtable
2955 pointer follows the class data. The field offset part of each field
2956 description in the class stab shows this ordering.
2958 << How is this reflected in stabs? See Cygnus bug #677 for some info. >>
2961 @appendix Table of Stab Types
2963 The following are all the possible values for the stab type field, for
2964 @code{a.out} files, in numeric order. This does not apply to XCOFF, but
2965 it does apply to stabs in ELF and stabs in SOM. Stabs in ECOFF use these
2966 values but add 0x8f300 to distinguish them from non-stab symbols.
2968 The symbolic names are defined in the file @file{include/aout/stabs.def}.
2971 * Non-Stab Symbol Types:: Types from 0 to 0x1f
2972 * Stab Symbol Types:: Types from 0x20 to 0xff
2975 @node Non-Stab Symbol Types
2976 @appendixsec Non-Stab Symbol Types
2978 The following types are used by the linker and assembler, not by stab
2979 directives. Since this document does not attempt to describe aspects of
2980 object file format other than the debugging format, no details are
2983 @c Try to get most of these to fit on a single line.
2993 File scope absolute symbol
2995 @item 0x3 N_ABS | N_EXT
2996 External absolute symbol
2999 File scope text symbol
3001 @item 0x5 N_TEXT | N_EXT
3002 External text symbol
3005 File scope data symbol
3007 @item 0x7 N_DATA | N_EXT
3008 External data symbol
3011 File scope BSS symbol
3013 @item 0x9 N_BSS | N_EXT
3017 Same as @code{N_FN}, for Sequent compilers
3020 Symbol is indirected to another symbol
3023 Common---visible after shared library dynamic link
3026 Absolute set element
3029 Text segment set element
3032 Data segment set element
3035 BSS segment set element
3038 Pointer to set vector
3040 @item 0x1e N_WARNING
3041 Print a warning message during linking
3044 File name of a @file{.o} file
3047 @node Stab Symbol Types
3048 @appendixsec Stab Symbol Types
3050 The following symbol types indicate that this is a stab. This is the
3051 full list of stab numbers, including stab types that are used in
3052 languages other than C.
3056 Global symbol; see @ref{Global Variables}.
3059 Function name (for BSD Fortran); see @ref{Procedures}.
3062 Function name (@pxref{Procedures}) or text segment variable
3066 Data segment file-scope variable; see @ref{Statics}.
3069 BSS segment file-scope variable; see @ref{Statics}.
3072 Name of main routine; see @ref{Main Program}.
3075 Variable in @code{.rodata} section; see @ref{Statics}.
3078 Global symbol (for Pascal); see @ref{N_PC}.
3081 Number of symbols (according to Ultrix V4.0); see @ref{N_NSYMS}.
3084 No DST map; see @ref{N_NOMAP}.
3086 @c FIXME: describe this solaris feature in the body of the text (see
3087 @c comments in include/aout/stab.def).
3089 Object file (Solaris2).
3091 @c See include/aout/stab.def for (a little) more info.
3093 Debugger options (Solaris2).
3096 Register variable; see @ref{Register Variables}.
3099 Modula-2 compilation unit; see @ref{N_M2C}.
3102 Line number in text segment; see @ref{Line Numbers}.
3105 Line number in data segment; see @ref{Line Numbers}.
3108 Line number in bss segment; see @ref{Line Numbers}.
3111 Sun source code browser, path to @file{.cb} file; see @ref{N_BROWS}.
3114 GNU Modula2 definition module dependency; see @ref{N_DEFD}.
3117 Function start/body/end line numbers (Solaris2).
3120 GNU C++ exception variable; see @ref{N_EHDECL}.
3123 Modula2 info "for imc" (according to Ultrix V4.0); see @ref{N_MOD2}.
3126 GNU C++ @code{catch} clause; see @ref{N_CATCH}.
3129 Structure of union element; see @ref{N_SSYM}.
3132 Last stab for module (Solaris2).
3135 Path and name of source file; see @ref{Source Files}.
3138 Stack variable (@pxref{Stack Variables}) or type (@pxref{Typedefs}).
3141 Beginning of an include file (Sun only); see @ref{Include Files}.
3144 Name of include file; see @ref{Include Files}.
3147 Parameter variable; see @ref{Parameters}.
3150 End of an include file; see @ref{Include Files}.
3153 Alternate entry point; see @ref{N_ENTRY}.
3156 Beginning of a lexical block; see @ref{Block Structure}.
3159 Place holder for a deleted include file; see @ref{Include Files}.
3162 Modula2 scope information (Sun linker); see @ref{N_SCOPE}.
3165 End of a lexical block; see @ref{Block Structure}.
3168 Begin named common block; see @ref{Common Blocks}.
3171 End named common block; see @ref{Common Blocks}.
3174 Member of a common block; see @ref{Common Blocks}.
3176 @c FIXME: How does this really work? Move it to main body of document.
3178 Pascal @code{with} statement: type,,0,0,offset (Solaris2).
3181 Gould non-base registers; see @ref{Gould}.
3184 Gould non-base registers; see @ref{Gould}.
3187 Gould non-base registers; see @ref{Gould}.
3190 Gould non-base registers; see @ref{Gould}.
3193 Gould non-base registers; see @ref{Gould}.
3196 @c Restore the default table indent
3201 @node Symbol Descriptors
3202 @appendix Table of Symbol Descriptors
3204 The symbol descriptor is the character which follows the colon in many
3205 stabs, and which tells what kind of stab it is. @xref{String Field},
3206 for more information about their use.
3208 @c Please keep this alphabetical
3210 @c In TeX, this looks great, digit is in italics. But makeinfo insists
3211 @c on putting it in `', not realizing that @var should override @code.
3212 @c I don't know of any way to make makeinfo do the right thing. Seems
3213 @c like a makeinfo bug to me.
3217 Variable on the stack; see @ref{Stack Variables}.
3220 C++ nested symbol; see @xref{Nested Symbols}
3223 Parameter passed by reference in register; see @ref{Reference Parameters}.
3226 Based variable; see @ref{Based Variables}.
3229 Constant; see @ref{Constants}.
3232 Conformant array bound (Pascal, maybe other languages); @ref{Conformant
3233 Arrays}. Name of a caught exception (GNU C++). These can be
3234 distinguished because the latter uses @code{N_CATCH} and the former uses
3235 another symbol type.
3238 Floating point register variable; see @ref{Register Variables}.
3241 Parameter in floating point register; see @ref{Register Parameters}.
3244 File scope function; see @ref{Procedures}.
3247 Global function; see @ref{Procedures}.
3250 Global variable; see @ref{Global Variables}.
3253 @xref{Register Parameters}.
3256 Internal (nested) procedure; see @ref{Nested Procedures}.
3259 Internal (nested) function; see @ref{Nested Procedures}.
3262 Label name (documented by AIX, no further information known).
3265 Module; see @ref{Procedures}.
3268 Argument list parameter; see @ref{Parameters}.
3274 Fortran Function parameter; see @ref{Parameters}.
3277 Unfortunately, three separate meanings have been independently invented
3278 for this symbol descriptor. At least the GNU and Sun uses can be
3279 distinguished by the symbol type. Global Procedure (AIX) (symbol type
3280 used unknown); see @ref{Procedures}. Register parameter (GNU) (symbol
3281 type @code{N_PSYM}); see @ref{Parameters}. Prototype of function
3282 referenced by this file (Sun @code{acc}) (symbol type @code{N_FUN}).
3285 Static Procedure; see @ref{Procedures}.
3288 Register parameter; see @ref{Register Parameters}.
3291 Register variable; see @ref{Register Variables}.
3294 File scope variable; see @ref{Statics}.
3297 Type name; see @ref{Typedefs}.
3300 Enumeration, structure, or union tag; see @ref{Typedefs}.
3303 Parameter passed by reference; see @ref{Reference Parameters}.
3306 Procedure scope static variable; see @ref{Statics}.
3309 Conformant array; see @ref{Conformant Arrays}.
3312 Function return variable; see @ref{Parameters}.
3315 @node Type Descriptors
3316 @appendix Table of Type Descriptors
3318 The type descriptor is the character which follows the type number and
3319 an equals sign. It specifies what kind of type is being defined.
3320 @xref{String Field}, for more information about their use.
3325 Type reference; see @ref{String Field}.
3328 Reference to builtin type; see @ref{Negative Type Numbers}.
3331 Method (C++); see @ref{Cplusplus}.
3334 Pointer; see @ref{Miscellaneous Types}.
3340 Type Attributes (AIX); see @ref{String Field}. Member (class and variable)
3341 type (GNU C++); see @ref{Cplusplus}.
3344 Array; see @ref{Arrays}.
3347 Open array; see @ref{Arrays}.
3350 Pascal space type (AIX); see @ref{Miscellaneous Types}. Builtin integer
3351 type (Sun); see @ref{Builtin Type Descriptors}.
3354 Volatile-qualified type; see @ref{Miscellaneous Types}.
3357 Complex builtin type; see @ref{Builtin Type Descriptors}.
3360 COBOL Picture type. See AIX documentation for details.
3363 File type; see @ref{Miscellaneous Types}.
3366 N-dimensional dynamic array; see @ref{Arrays}.
3369 Enumeration type; see @ref{Enumerations}.
3372 N-dimensional subarray; see @ref{Arrays}.
3375 Function type; see @ref{Function Types}.
3378 Pascal function parameter; see @ref{Function Types}
3381 Builtin floating point type; see @ref{Builtin Type Descriptors}.
3384 COBOL Group. See AIX documentation for details.
3387 Imported type; see @ref{Cross-References}.
3390 Const-qualified type; see @ref{Miscellaneous Types}.
3393 COBOL File Descriptor. See AIX documentation for details.
3396 Multiple instance type; see @ref{Miscellaneous Types}.
3399 String type; see @ref{Strings}.
3402 Stringptr; see @ref{Strings}.
3405 Opaque type; see @ref{Typedefs}.
3408 Procedure; see @ref{Function Types}.
3411 Packed array; see @ref{Arrays}.
3414 Range type; see @ref{Subranges}.
3417 Builtin floating type; see @ref{Builtin Type Descriptors} (Sun). Pascal
3418 subroutine parameter; see @ref{Function Types} (AIX). Detecting this
3419 conflict is possible with careful parsing (hint: a Pascal subroutine
3420 parameter type will always contain a comma, and a builtin type
3421 descriptor never will).
3424 Structure type; see @ref{Structures}.
3427 Set type; see @ref{Miscellaneous Types}.
3430 Union; see @ref{Unions}.
3433 Variant record. This is a Pascal and Modula-2 feature which is like a
3434 union within a struct in C. See AIX documentation for details.
3437 Wide character; see @ref{Builtin Type Descriptors}.
3440 Cross-reference; see @ref{Cross-References}.
3443 gstring; see @ref{Strings}.
3446 @node Expanded Reference
3447 @appendix Expanded Reference by Stab Type
3449 @c FIXME: This appendix should go away; see N_PSYM or N_SO for an example.
3451 For a full list of stab types, and cross-references to where they are
3452 described, see @ref{Stab Types}. This appendix just duplicates certain
3453 information from the main body of this document; eventually the
3454 information will all be in one place.
3458 The first line is the symbol type (see @file{include/aout/stab.def}).
3460 The second line describes the language constructs the symbol type
3463 The third line is the stab format with the significant stab fields
3464 named and the rest NIL.
3466 Subsequent lines expand upon the meaning and possible values for each
3467 significant stab field. @samp{#} stands in for the type descriptor.
3469 Finally, any further information.
3472 * N_PC:: Pascal global symbol
3473 * N_NSYMS:: Number of symbols
3474 * N_NOMAP:: No DST map
3475 * N_M2C:: Modula-2 compilation unit
3476 * N_BROWS:: Path to .cb file for Sun source code browser
3477 * N_DEFD:: GNU Modula2 definition module dependency
3478 * N_EHDECL:: GNU C++ exception variable
3479 * N_MOD2:: Modula2 information "for imc"
3480 * N_CATCH:: GNU C++ "catch" clause
3481 * N_SSYM:: Structure or union element
3482 * N_ENTRY:: Alternate entry point
3483 * N_SCOPE:: Modula2 scope information (Sun only)
3484 * Gould:: non-base register symbols used on Gould systems
3485 * N_LENG:: Length of preceding entry
3491 @deffn @code{.stabs} N_PC
3493 Global symbol (for Pascal).
3496 "name" -> "symbol_name" <<?>>
3497 value -> supposedly the line number (stab.def is skeptical)
3501 @file{stabdump.c} says:
3503 global pascal symbol: name,,0,subtype,line
3511 @deffn @code{.stabn} N_NSYMS
3513 Number of symbols (according to Ultrix V4.0).
3516 0, files,,funcs,lines (stab.def)
3523 @deffn @code{.stabs} N_NOMAP
3525 No DST map for symbol (according to Ultrix V4.0). I think this means a
3526 variable has been optimized out.
3529 name, ,0,type,ignored (stab.def)
3536 @deffn @code{.stabs} N_M2C
3538 Modula-2 compilation unit.
3541 "string" -> "unit_name,unit_time_stamp[,code_time_stamp]"
3543 value -> 0 (main unit)
3547 See @cite{Dbx and Dbxtool Interfaces}, 2nd edition, by Sun, 1988, for
3555 @deffn @code{.stabs} N_BROWS
3557 Sun source code browser, path to @file{.cb} file
3560 "path to associated @file{.cb} file"
3562 Note: N_BROWS has the same value as N_BSLINE.
3568 @deffn @code{.stabn} N_DEFD
3570 GNU Modula2 definition module dependency.
3572 GNU Modula-2 definition module dependency. The value is the
3573 modification time of the definition file. The other field is non-zero
3574 if it is imported with the GNU M2 keyword @code{%INITIALIZE}. Perhaps
3575 @code{N_M2C} can be used if there are enough empty fields?
3581 @deffn @code{.stabs} N_EHDECL
3583 GNU C++ exception variable <<?>>.
3585 "@var{string} is variable name"
3587 Note: conflicts with @code{N_MOD2}.
3593 @deffn @code{.stab?} N_MOD2
3595 Modula2 info "for imc" (according to Ultrix V4.0)
3597 Note: conflicts with @code{N_EHDECL} <<?>>
3603 @deffn @code{.stabn} N_CATCH
3605 GNU C++ @code{catch} clause
3607 GNU C++ @code{catch} clause. The value is its address. The desc field
3608 is nonzero if this entry is immediately followed by a @code{CAUGHT} stab
3609 saying what exception was caught. Multiple @code{CAUGHT} stabs means
3610 that multiple exceptions can be caught here. If desc is 0, it means all
3611 exceptions are caught here.
3617 @deffn @code{.stabn} N_SSYM
3619 Structure or union element.
3621 The value is the offset in the structure.
3623 <<?looking at structs and unions in C I didn't see these>>
3629 @deffn @code{.stabn} N_ENTRY
3631 Alternate entry point.
3632 The value is its address.
3639 @deffn @code{.stab?} N_SCOPE
3641 Modula2 scope information (Sun linker)
3646 @section Non-base registers on Gould systems
3648 @deffn @code{.stab?} N_NBTEXT
3649 @deffnx @code{.stab?} N_NBDATA
3650 @deffnx @code{.stab?} N_NBBSS
3651 @deffnx @code{.stab?} N_NBSTS
3652 @deffnx @code{.stab?} N_NBLCS
3658 These are used on Gould systems for non-base registers syms.
3660 However, the following values are not the values used by Gould; they are
3661 the values which GNU has been documenting for these values for a long
3662 time, without actually checking what Gould uses. I include these values
3663 only because perhaps some someone actually did something with the GNU
3664 information (I hope not, why GNU knowingly assigned wrong values to
3665 these in the header file is a complete mystery to me).
3668 240 0xf0 N_NBTEXT ??
3669 242 0xf2 N_NBDATA ??
3679 @deffn @code{.stabn} N_LENG
3681 Second symbol entry containing a length-value for the preceding entry.
3682 The value is the length.
3686 @appendix Questions and Anomalies
3690 @c I think this is changed in GCC 2.4.5 to put the line number there.
3691 For GNU C stabs defining local and global variables (@code{N_LSYM} and
3692 @code{N_GSYM}), the desc field is supposed to contain the source
3693 line number on which the variable is defined. In reality the desc
3694 field is always 0. (This behavior is defined in @file{dbxout.c} and
3695 putting a line number in desc is controlled by @samp{#ifdef
3696 WINNING_GDB}, which defaults to false). GDB supposedly uses this
3697 information if you say @samp{list @var{var}}. In reality, @var{var} can
3698 be a variable defined in the program and GDB says @samp{function
3699 @var{var} not defined}.
3702 In GNU C stabs, there seems to be no way to differentiate tag types:
3703 structures, unions, and enums (symbol descriptor @samp{T}) and typedefs
3704 (symbol descriptor @samp{t}) defined at file scope from types defined locally
3705 to a procedure or other more local scope. They all use the @code{N_LSYM}
3706 stab type. Types defined at procedure scope are emited after the
3707 @code{N_RBRAC} of the preceding function and before the code of the
3708 procedure in which they are defined. This is exactly the same as
3709 types defined in the source file between the two procedure bodies.
3710 GDB overcompensates by placing all types in block #1, the block for
3711 symbols of file scope. This is true for default, @samp{-ansi} and
3712 @samp{-traditional} compiler options. (Bugs gcc/1063, gdb/1066.)
3715 What ends the procedure scope? Is it the proc block's @code{N_RBRAC} or the
3716 next @code{N_FUN}? (I believe its the first.)
3719 @c FIXME: This should go with the other stuff about global variables.
3720 Global variable stabs don't have location information. This comes
3721 from the external symbol for the same variable. The external symbol
3722 has a leading underbar on the _name of the variable and the stab does
3723 not. How do we know these two symbol table entries are talking about
3724 the same symbol when their names are different? (Answer: the debugger
3725 knows that external symbols have leading underbars).
3727 @c FIXME: This is absurdly vague; there all kinds of differences, some
3728 @c of which are the same between gnu & sun, and some of which aren't.
3729 @c In particular, I'm pretty sure GCC works with Sun dbx by default.
3731 @c Can GCC be configured to output stabs the way the Sun compiler
3732 @c does, so that their native debugging tools work? <NO?> It doesn't by
3733 @c default. GDB reads either format of stab. (GCC or SunC). How about
3737 @node XCOFF Differences
3738 @appendix Differences Between GNU Stabs in a.out and GNU Stabs in XCOFF
3740 @c FIXME: Merge *all* these into the main body of the document.
3741 The AIX/RS6000 native object file format is XCOFF with stabs. This
3742 appendix only covers those differences which are not covered in the main
3743 body of this document.
3747 BSD a.out stab types correspond to AIX XCOFF storage classes. In general
3748 the mapping is @code{N_@var{stabtype}} becomes @code{C_@var{stabtype}}.
3749 Some stab types in a.out are not supported in XCOFF; most of these use
3752 @c FIXME: Get C_* types for the block, figure out whether it is always
3753 @c used (I suspect not), explain clearly, and move to node Statics.
3754 Exception: initialised static @code{N_STSYM} and un-initialized static
3755 @code{N_LCSYM} both map to the @code{C_STSYM} storage class. But the
3756 distinction is preserved because in XCOFF @code{N_STSYM} and
3757 @code{N_LCSYM} must be emited in a named static block. Begin the block
3758 with @samp{.bs s[RW] data_section_name} for @code{N_STSYM} or @samp{.bs
3759 s bss_section_name} for @code{N_LCSYM}. End the block with @samp{.es}.
3761 @c FIXME: I think they are trying to say something about whether the
3762 @c assembler defaults the value to the location counter.
3764 If the XCOFF stab is an @code{N_FUN} (@code{C_FUN}) then follow the
3765 string field with @samp{,.} instead of just @samp{,}.
3768 I think that's it for @file{.s} file differences. They could stand to be
3769 better presented. This is just a list of what I have noticed so far.
3770 There are a @emph{lot} of differences in the information in the symbol
3771 tables of the executable and object files.
3773 Mapping of a.out stab types to XCOFF storage classes:
3776 stab type storage class
3777 -------------------------------
3813 @node Sun Differences
3814 @appendix Differences Between GNU Stabs and Sun Native Stabs
3816 @c FIXME: Merge all this stuff into the main body of the document.
3820 GNU C stabs define @emph{all} types, file or procedure scope, as
3821 @code{N_LSYM}. Sun doc talks about using @code{N_GSYM} too.
3824 Sun C stabs use type number pairs in the format
3825 (@var{file-number},@var{type-number}) where @var{file-number} is a
3826 number starting with 1 and incremented for each sub-source file in the
3827 compilation. @var{type-number} is a number starting with 1 and
3828 incremented for each new type defined in the compilation. GNU C stabs
3829 use the type number alone, with no source file number.
3833 @appendix Using Stabs With The ELF Object File Format
3835 The ELF object file format allows tools to create object files with
3836 custom sections containing any arbitrary data. To use stabs in ELF
3837 object files, the tools create two custom sections, a section named
3838 @code{.stab} which contains an array of fixed length structures, one
3839 struct per stab, and a section named @code{.stabstr} containing all the
3840 variable length strings that are referenced by stabs in the @code{.stab}
3841 section. The byte order of the stabs binary data matches the byte order
3842 of the ELF file itself, as determined from the @code{EI_DATA} field in
3843 the @code{e_ident} member of the ELF header.
3845 The first stab in the @code{.stab} section for each compilation unit is
3846 synthetic, generated entirely by the assembler, with no corresponding
3847 @code{.stab} directive as input to the assembler. This stab contains
3848 the following fields:
3852 Offset in the @code{.stabstr} section to the source filename.
3858 Unused field, always zero.
3861 Count of upcoming symbols, i.e., the number of remaining stabs for this
3865 Size of the string table fragment associated with this source file, in
3869 The @code{.stabstr} section always starts with a null byte (so that string
3870 offsets of zero reference a null string), followed by random length strings,
3871 each of which is null byte terminated.
3873 The ELF section header for the @code{.stab} section has its
3874 @code{sh_link} member set to the section number of the @code{.stabstr}
3875 section, and the @code{.stabstr} section has its ELF section
3876 header @code{sh_type} member set to @code{SHT_STRTAB} to mark it as a
3879 To keep linking fast, you don't want the linker to have to relocate very
3880 many stabs. Thus Sun has invented a scheme in which addresses in the
3881 @code{n_value} field are relative to the source file (or some entity
3882 smaller than a source file, like a function). To find the address of
3883 each section corresponding to a given source file, the compiler puts out
3884 symbols giving the address of each section for a given source file.
3885 Since these are ELF (not stab) symbols, the linker relocates them
3886 correctly without having to touch the stabs section. They are named
3887 @code{Bbss.bss} for the bss section, @code{Ddata.data} for the data
3888 section, and @code{Drodata.rodata} for the rodata section. For the text
3889 section, there is no such symbol (but there should be, see below). For
3890 an example of how these symbols work, @xref{ELF and SOM Transformations}. GCC
3891 does not provide these symbols; it instead relies on the stabs getting
3892 relocated. Thus addresses which would normally be relative to
3893 @code{Bbss.bss}, etc., are already relocated. The Sun linker provided
3894 with Solaris 2.2 and earlier relocates stabs using normal ELF relocation
3895 information, as it would do for any section. Sun has been threatening
3896 to kludge their linker to not do this (to speed up linking), even though
3897 the correct way to avoid having the linker do these relocations is to
3898 have the compiler no longer output relocatable values. Last I heard
3899 they had been talked out of the linker kludge. See Sun point patch
3900 101052-01 and Sun bug 1142109. With the Sun compiler this affects
3901 @samp{S} symbol descriptor stabs (@pxref{Statics}) and functions
3902 (@pxref{Procedures}). In the latter case, to adopt the clean solution
3903 (making the value of the stab relative to the start of the compilation
3904 unit), it would be necessary to invent a @code{Ttext.text} symbol,
3905 analogous to the @code{Bbss.bss}, etc., symbols. I recommend this
3906 rather than using a zero value and getting the address from the ELF
3909 Finding the correct @code{Bbss.bss}, etc., symbol is difficult, because
3910 the linker simply concatenates the @code{.stab} sections from each
3911 @file{.o} file without including any information about which part of a
3912 @code{.stab} section comes from which @file{.o} file. The way GDB does
3913 this is to look for an ELF @code{STT_FILE} symbol which has the same
3914 name as the last component of the file name from the @code{N_SO} symbol
3915 in the stabs (for example, if the file name is @file{../../gdb/main.c},
3916 it looks for an ELF @code{STT_FILE} symbol named @code{main.c}). This
3917 loses if different files have the same name (they could be in different
3918 directories, a library could have been copied from one system to
3919 another, etc.). It would be much cleaner to have the @code{Bbss.bss}
3920 symbols in the stabs themselves. Having the linker relocate them there
3921 is no more work than having the linker relocate ELF symbols, and it
3922 solves the problem of having to associate the ELF and stab symbols.
3923 However, no one has yet designed or implemented such a scheme.
3925 @node Symbol Types Index
3926 @unnumbered Symbol Types Index