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 * Stab Sections:: In some object file formats, stabs are
89 * Symbol Types Index:: Index of symbolic stab symbol type names.
95 @chapter Overview of Stabs
97 @dfn{Stabs} refers to a format for information that describes a program
98 to a debugger. This format was apparently invented by
100 the University of California at Berkeley, for the @code{pdx} Pascal
101 debugger; the format has spread widely since then.
103 This document is one of the few published sources of documentation on
104 stabs. It is believed to be comprehensive for stabs used by C. The
105 lists of symbol descriptors (@pxref{Symbol Descriptors}) and type
106 descriptors (@pxref{Type Descriptors}) are believed to be completely
107 comprehensive. Stabs for COBOL-specific features and for variant
108 records (used by Pascal and Modula-2) are poorly documented here.
110 Other sources of information on stabs are @cite{Dbx and Dbxtool
111 Interfaces}, 2nd edition, by Sun, 1988, and @cite{AIX Version 3.2 Files
112 Reference}, Fourth Edition, September 1992, "dbx Stabstring Grammar" in
113 the a.out section, page 2-31. This document is believed to incorporate
114 the information from those two sources except where it explicitly directs
115 you to them for more information.
118 * Flow:: Overview of debugging information flow
119 * Stabs Format:: Overview of stab format
120 * String Field:: The string field
121 * C Example:: A simple example in C source
122 * Assembly Code:: The simple example at the assembly level
126 @section Overview of Debugging Information Flow
128 The GNU C compiler compiles C source in a @file{.c} file into assembly
129 language in a @file{.s} file, which the assembler translates into
130 a @file{.o} file, which the linker combines with other @file{.o} files and
131 libraries to produce an executable file.
133 With the @samp{-g} option, GCC puts in the @file{.s} file additional
134 debugging information, which is slightly transformed by the assembler
135 and linker, and carried through into the final executable. This
136 debugging information describes features of the source file like line
137 numbers, the types and scopes of variables, and function names,
138 parameters, and scopes.
140 For some object file formats, the debugging information is encapsulated
141 in assembler directives known collectively as @dfn{stab} (symbol table)
142 directives, which are interspersed with the generated code. Stabs are
143 the native format for debugging information in the a.out and XCOFF
144 object file formats. The GNU tools can also emit stabs in the COFF and
145 ECOFF object file formats.
147 The assembler adds the information from stabs to the symbol information
148 it places by default in the symbol table and the string table of the
149 @file{.o} file it is building. The linker consolidates the @file{.o}
150 files into one executable file, with one symbol table and one string
151 table. Debuggers use the symbol and string tables in the executable as
152 a source of debugging information about the program.
155 @section Overview of Stab Format
157 There are three overall formats for stab assembler directives,
158 differentiated by the first word of the stab. The name of the directive
159 describes which combination of four possible data fields follows. It is
160 either @code{.stabs} (string), @code{.stabn} (number), or @code{.stabd}
161 (dot). IBM's XCOFF assembler uses @code{.stabx} (and some other
162 directives such as @code{.file} and @code{.bi}) instead of
163 @code{.stabs}, @code{.stabn} or @code{.stabd}.
165 The overall format of each class of stab is:
168 .stabs "@var{string}",@var{type},@var{other},@var{desc},@var{value}
169 .stabn @var{type},@var{other},@var{desc},@var{value}
170 .stabd @var{type},@var{other},@var{desc}
171 .stabx "@var{string}",@var{value},@var{type},@var{sdb-type}
174 @c what is the correct term for "current file location"? My AIX
175 @c assembler manual calls it "the value of the current location counter".
176 For @code{.stabn} and @code{.stabd}, there is no @var{string} (the
177 @code{n_strx} field is zero; see @ref{Symbol Tables}). For
178 @code{.stabd}, the @var{value} field is implicit and has the value of
179 the current file location. For @code{.stabx}, the @var{sdb-type} field
180 is unused for stabs and can always be set to zero. The @var{other}
181 field is almost always unused and can be set to zero.
183 The number in the @var{type} field gives some basic information about
184 which type of stab this is (or whether it @emph{is} a stab, as opposed
185 to an ordinary symbol). Each valid type number defines a different stab
186 type; further, the stab type defines the exact interpretation of, and
187 possible values for, any remaining @var{string}, @var{desc}, or
188 @var{value} fields present in the stab. @xref{Stab Types}, for a list
189 in numeric order of the valid @var{type} field values for stab directives.
192 @section The String Field
194 For most stabs the string field holds the meat of the
195 debugging information. The flexible nature of this field
196 is what makes stabs extensible. For some stab types the string field
197 contains only a name. For other stab types the contents can be a great
200 The overall format of the string field for most stab types is:
203 "@var{name}:@var{symbol-descriptor} @var{type-information}"
206 @var{name} is the name of the symbol represented by the stab; it can
207 contain a pair of colons (@pxref{Nested Symbols}). @var{name} can be
208 omitted, which means the stab represents an unnamed object. For
209 example, @samp{:t10=*2} defines type 10 as a pointer to type 2, but does
210 not give the type a name. Omitting the @var{name} field is supported by
211 AIX dbx and GDB after about version 4.8, but not other debuggers. GCC
212 sometimes uses a single space as the name instead of omitting the name
213 altogether; apparently that is supported by most debuggers.
215 The @var{symbol-descriptor} following the @samp{:} is an alphabetic
216 character that tells more specifically what kind of symbol the stab
217 represents. If the @var{symbol-descriptor} is omitted, but type
218 information follows, then the stab represents a local variable. For a
219 list of symbol descriptors, see @ref{Symbol Descriptors}. The @samp{c}
220 symbol descriptor is an exception in that it is not followed by type
221 information. @xref{Constants}.
223 @var{type-information} is either a @var{type-number}, or
224 @samp{@var{type-number}=}. A @var{type-number} alone is a type
225 reference, referring directly to a type that has already been defined.
227 The @samp{@var{type-number}=} form is a type definition, where the
228 number represents a new type which is about to be defined. The type
229 definition may refer to other types by number, and those type numbers
230 may be followed by @samp{=} and nested definitions. Also, the Lucid
231 compiler will repeat @samp{@var{type-number}=} more than once if it
232 wants to define several type numbers at once.
234 In a type definition, if the character that follows the equals sign is
235 non-numeric then it is a @var{type-descriptor}, and tells what kind of
236 type is about to be defined. Any other values following the
237 @var{type-descriptor} vary, depending on the @var{type-descriptor}.
238 @xref{Type Descriptors}, for a list of @var{type-descriptor} values. If
239 a number follows the @samp{=} then the number is a @var{type-reference}.
240 For a full description of types, @ref{Types}.
242 There is an AIX extension for type attributes. Following the @samp{=}
243 are any number of type attributes. Each one starts with @samp{@@} and
244 ends with @samp{;}. Debuggers, including AIX's dbx and GDB 4.10, skip
245 any type attributes they do not recognize. GDB 4.9 and other versions
246 of dbx may not do this. Because of a conflict with C++
247 (@pxref{Cplusplus}), new attributes should not be defined which begin
248 with a digit, @samp{(}, or @samp{-}; GDB may be unable to distinguish
249 those from the C++ type descriptor @samp{@@}. The attributes are:
252 @item a@var{boundary}
253 @var{boundary} is an integer specifying the alignment. I assume it
254 applies to all variables of this type.
257 Pointer class (for checking). Not sure what this means, or how
258 @var{integer} is interpreted.
261 Indicate this is a packed type, meaning that structure fields or array
262 elements are placed more closely in memory, to save memory at the
266 Size in bits of a variable of this type. This is fully supported by GDB
270 Indicate that this type is a string instead of an array of characters,
271 or a bitstring instead of a set. It doesn't change the layout of the
272 data being represented, but does enable the debugger to know which type
276 All of this can make the string field quite long. All versions of GDB,
277 and some versions of dbx, can handle arbitrarily long strings. But many
278 versions of dbx (or assemblers or linkers, I'm not sure which)
279 cretinously limit the strings to about 80 characters, so compilers which
280 must work with such systems need to split the @code{.stabs} directive
281 into several @code{.stabs} directives. Each stab duplicates every field
282 except the string field. The string field of every stab except the last
283 is marked as continued with a backslash at the end (in the assembly code
284 this may be written as a double backslash, depending on the assembler).
285 Removing the backslashes and concatenating the string fields of each
286 stab produces the original, long string. Just to be incompatible (or so
287 they don't have to worry about what the assembler does with
288 backslashes), AIX can use @samp{?} instead of backslash.
291 @section A Simple Example in C Source
293 To get the flavor of how stabs describe source information for a C
294 program, let's look at the simple program:
299 printf("Hello world");
303 When compiled with @samp{-g}, the program above yields the following
304 @file{.s} file. Line numbers have been added to make it easier to refer
305 to parts of the @file{.s} file in the description of the stabs that
309 @section The Simple Example at the Assembly Level
311 This simple ``hello world'' example demonstrates several of the stab
312 types used to describe C language source files.
316 2 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0
317 3 .stabs "hello.c",100,0,0,Ltext0
320 6 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0
321 7 .stabs "char:t2=r2;0;127;",128,0,0,0
322 8 .stabs "long int:t3=r1;-2147483648;2147483647;",128,0,0,0
323 9 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
324 10 .stabs "long unsigned int:t5=r1;0;-1;",128,0,0,0
325 11 .stabs "short int:t6=r1;-32768;32767;",128,0,0,0
326 12 .stabs "long long int:t7=r1;0;-1;",128,0,0,0
327 13 .stabs "short unsigned int:t8=r1;0;65535;",128,0,0,0
328 14 .stabs "long long unsigned int:t9=r1;0;-1;",128,0,0,0
329 15 .stabs "signed char:t10=r1;-128;127;",128,0,0,0
330 16 .stabs "unsigned char:t11=r1;0;255;",128,0,0,0
331 17 .stabs "float:t12=r1;4;0;",128,0,0,0
332 18 .stabs "double:t13=r1;8;0;",128,0,0,0
333 19 .stabs "long double:t14=r1;8;0;",128,0,0,0
334 20 .stabs "void:t15=15",128,0,0,0
337 23 .ascii "Hello, world!\12\0"
352 38 sethi %hi(LC0),%o1
353 39 or %o1,%lo(LC0),%o0
364 50 .stabs "main:F1",36,0,0,_main
365 51 .stabn 192,0,0,LBB2
366 52 .stabn 224,0,0,LBE2
369 @node Program Structure
370 @chapter Encoding the Structure of the Program
372 The elements of the program structure that stabs encode include the name
373 of the main function, the names of the source and include files, the
374 line numbers, procedure names and types, and the beginnings and ends of
378 * Main Program:: Indicate what the main program is
379 * Source Files:: The path and name of the source file
380 * Include Files:: Names of include files
383 * Nested Procedures::
388 @section Main Program
391 Most languages allow the main program to have any name. The
392 @code{N_MAIN} stab type tells the debugger the name that is used in this
393 program. Only the string field is significant; it is the name of
394 a function which is the main program. Most C compilers do not use this
395 stab (they expect the debugger to assume that the name is @code{main}),
396 but some C compilers emit an @code{N_MAIN} stab for the @code{main}
400 @section Paths and Names of the Source Files
403 Before any other stabs occur, there must be a stab specifying the source
404 file. This information is contained in a symbol of stab type
405 @code{N_SO}; the string field contains the name of the file. The
406 value of the symbol is the start address of the portion of the
407 text section corresponding to that file.
409 With the Sun Solaris2 compiler, the desc field contains a
410 source-language code.
411 @c Do the debuggers use it? What are the codes? -djm
413 Some compilers (for example, GCC2 and SunOS4 @file{/bin/cc}) also
414 include the directory in which the source was compiled, in a second
415 @code{N_SO} symbol preceding the one containing the file name. This
416 symbol can be distinguished by the fact that it ends in a slash. Code
417 from the @code{cfront} C++ compiler can have additional @code{N_SO} symbols for
418 nonexistent source files after the @code{N_SO} for the real source file;
419 these are believed to contain no useful information.
424 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0 # @r{100 is N_SO}
425 .stabs "hello.c",100,0,0,Ltext0
430 Instead of @code{N_SO} symbols, XCOFF uses a @code{.file} assembler
431 directive which assembles to a standard COFF @code{.file} symbol;
432 explaining this in detail is outside the scope of this document.
435 @section Names of Include Files
437 There are several schemes for dealing with include files: the
438 traditional @code{N_SOL} approach, Sun's @code{N_BINCL} approach, and the
439 XCOFF @code{C_BINCL} approach (which despite the similar name has little in
440 common with @code{N_BINCL}).
443 An @code{N_SOL} symbol specifies which include file subsequent symbols
444 refer to. The string field is the name of the file and the value is the
445 text address corresponding to the end of the previous include file and
446 the start of this one. To specify the main source file again, use an
447 @code{N_SOL} symbol with the name of the main source file.
452 The @code{N_BINCL} approach works as follows. An @code{N_BINCL} symbol
453 specifies the start of an include file. In an object file, only the
454 string is significant; the Sun linker puts data into some of the
455 other fields. The end of the include file is marked by an
456 @code{N_EINCL} symbol (which has no string field). In an object
457 file, there is no significant data in the @code{N_EINCL} symbol; the Sun
458 linker puts data into some of the fields. @code{N_BINCL} and
459 @code{N_EINCL} can be nested.
461 If the linker detects that two source files have identical stabs between
462 an @code{N_BINCL} and @code{N_EINCL} pair (as will generally be the case
463 for a header file), then it only puts out the stabs once. Each
464 additional occurance is replaced by an @code{N_EXCL} symbol. I believe
465 the Sun (SunOS4, not sure about Solaris) linker is the only one which
466 supports this feature.
467 @c What do the fields of N_EXCL contain? -djm
471 For the start of an include file in XCOFF, use the @file{.bi} assembler
472 directive, which generates a @code{C_BINCL} symbol. A @file{.ei}
473 directive, which generates a @code{C_EINCL} symbol, denotes the end of
474 the include file. Both directives are followed by the name of the
475 source file in quotes, which becomes the string for the symbol.
476 The value of each symbol, produced automatically by the assembler
477 and linker, is the offset into the executable of the beginning
478 (inclusive, as you'd expect) or end (inclusive, as you would not expect)
479 of the portion of the COFF line table that corresponds to this include
480 file. @code{C_BINCL} and @code{C_EINCL} do not nest.
483 @section Line Numbers
486 An @code{N_SLINE} symbol represents the start of a source line. The
487 desc field contains the line number and the value contains the code
488 address for the start of that source line. On most machines the address
489 is absolute; for stabs in sections (@pxref{Stab Sections}), it is
490 relative to the function in which the @code{N_SLINE} symbol occurs.
494 GNU documents @code{N_DSLINE} and @code{N_BSLINE} symbols for line
495 numbers in the data or bss segments, respectively. They are identical
496 to @code{N_SLINE} but are relocated differently by the linker. They
497 were intended to be used to describe the source location of a variable
498 declaration, but I believe that GCC2 actually puts the line number in
499 the desc field of the stab for the variable itself. GDB has been
500 ignoring these symbols (unless they contain a string field) since
503 For single source lines that generate discontiguous code, such as flow
504 of control statements, there may be more than one line number entry for
505 the same source line. In this case there is a line number entry at the
506 start of each code range, each with the same line number.
508 XCOFF does not use stabs for line numbers. Instead, it uses COFF line
509 numbers (which are outside the scope of this document). Standard COFF
510 line numbers cannot deal with include files, but in XCOFF this is fixed
511 with the @code{C_BINCL} method of marking include files (@pxref{Include
517 @findex N_FUN, for functions
519 @findex N_STSYM, for functions (Sun acc)
520 @findex N_GSYM, for functions (Sun acc)
521 All of the following stabs normally use the @code{N_FUN} symbol type.
522 However, Sun's @code{acc} compiler on SunOS4 uses @code{N_GSYM} and
523 @code{N_STSYM}, which means that the value of the stab for the function
524 is useless and the debugger must get the address of the function from
525 the non-stab symbols instead. BSD Fortran is said to use @code{N_FNAME}
526 with the same restriction; the value of the symbol is not useful (I'm
527 not sure it really does use this, because GDB doesn't handle this and no
530 A function is represented by an @samp{F} symbol descriptor for a global
531 (extern) function, and @samp{f} for a static (local) function. The
532 value is the address of the start of the function. For @code{a.out}, it
533 is already relocated. For stabs in ELF, the SunPRO compiler version
534 2.0.1 and GCC put out an address which gets relocated by the linker. In
535 a future release SunPRO is planning to put out zero, in which case the
536 address can be found from the ELF (non-stab) symbol. Because looking
537 things up in the ELF symbols would probably be slow, I'm not sure how to
538 find which symbol of that name is the right one, and this doesn't
539 provide any way to deal with nested functions, it would probably be
540 better to make the value of the stab an address relative to the start of
541 the file, or just absolute. See @ref{ELF Linker Relocation} for more
542 information on linker relocation of stabs in ELF files.
544 The type information of the stab represents the return type of the
545 function; thus @samp{foo:f5} means that foo is a function returning type
546 5. There is no need to try to get the line number of the start of the
547 function from the stab for the function; it is in the next
548 @code{N_SLINE} symbol.
550 @c FIXME: verify whether the "I suspect" below is true or not.
551 Some compilers (such as Sun's Solaris compiler) support an extension for
552 specifying the types of the arguments. I suspect this extension is not
553 used for old (non-prototyped) function definitions in C. If the
554 extension is in use, the type information of the stab for the function
555 is followed by type information for each argument, with each argument
556 preceded by @samp{;}. An argument type of 0 means that additional
557 arguments are being passed, whose types and number may vary (@samp{...}
558 in ANSI C). GDB has tolerated this extension (parsed the syntax, if not
559 necessarily used the information) since at least version 4.8; I don't
560 know whether all versions of dbx tolerate it. The argument types given
561 here are not redundant with the symbols for the formal parameters
562 (@pxref{Parameters}); they are the types of the arguments as they are
563 passed, before any conversions might take place. For example, if a C
564 function which is declared without a prototype takes a @code{float}
565 argument, the value is passed as a @code{double} but then converted to a
566 @code{float}. Debuggers need to use the types given in the arguments
567 when printing values, but when calling the function they need to use the
568 types given in the symbol defining the function.
570 If the return type and types of arguments of a function which is defined
571 in another source file are specified (i.e., a function prototype in ANSI
572 C), traditionally compilers emit no stab; the only way for the debugger
573 to find the information is if the source file where the function is
574 defined was also compiled with debugging symbols. As an extension the
575 Solaris compiler uses symbol descriptor @samp{P} followed by the return
576 type of the function, followed by the arguments, each preceded by
577 @samp{;}, as in a stab with symbol descriptor @samp{f} or @samp{F}.
578 This use of symbol descriptor @samp{P} can be distinguished from its use
579 for register parameters (@pxref{Register Parameters}) by the fact that it has
580 symbol type @code{N_FUN}.
582 The AIX documentation also defines symbol descriptor @samp{J} as an
583 internal function. I assume this means a function nested within another
584 function. It also says symbol descriptor @samp{m} is a module in
585 Modula-2 or extended Pascal.
587 Procedures (functions which do not return values) are represented as
588 functions returning the @code{void} type in C. I don't see why this couldn't
589 be used for all languages (inventing a @code{void} type for this purpose if
590 necessary), but the AIX documentation defines @samp{I}, @samp{P}, and
591 @samp{Q} for internal, global, and static procedures, respectively.
592 These symbol descriptors are unusual in that they are not followed by
595 The following example shows a stab for a function @code{main} which
596 returns type number @code{1}. The @code{_main} specified for the value
597 is a reference to an assembler label which is used to fill in the start
598 address of the function.
601 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
604 The stab representing a procedure is located immediately following the
605 code of the procedure. This stab is in turn directly followed by a
606 group of other stabs describing elements of the procedure. These other
607 stabs describe the procedure's parameters, its block local variables, and
610 @node Nested Procedures
611 @section Nested Procedures
613 For any of the symbol descriptors representing procedures, after the
614 symbol descriptor and the type information is optionally a scope
615 specifier. This consists of a comma, the name of the procedure, another
616 comma, and the name of the enclosing procedure. The first name is local
617 to the scope specified, and seems to be redundant with the name of the
618 symbol (before the @samp{:}). This feature is used by GCC, and
619 presumably Pascal, Modula-2, etc., compilers, for nested functions.
621 If procedures are nested more than one level deep, only the immediately
622 containing scope is specified. For example, this code:
634 return baz (x + 2 * y);
636 return x + bar (3 * x);
644 .stabs "baz:f1,baz,bar",36,0,0,_baz.15 # @r{36 is N_FUN}
645 .stabs "bar:f1,bar,foo",36,0,0,_bar.12
646 .stabs "foo:F1",36,0,0,_foo
649 @node Block Structure
650 @section Block Structure
654 @c For GCC 2.5.8 or so stabs-in-coff, these are absolute instead of
655 @c function relative (as documented below). But GDB has never been able
656 @c to deal with that (it had wanted them to be relative to the file, but
657 @c I just fixed that (between GDB 4.12 and 4.13)), so it is function
658 @c relative just like ELF and SOM and the below documentation.
659 The program's block structure is represented by the @code{N_LBRAC} (left
660 brace) and the @code{N_RBRAC} (right brace) stab types. The variables
661 defined inside a block precede the @code{N_LBRAC} symbol for most
662 compilers, including GCC. Other compilers, such as the Convex, Acorn
663 RISC machine, and Sun @code{acc} compilers, put the variables after the
664 @code{N_LBRAC} symbol. The values of the @code{N_LBRAC} and
665 @code{N_RBRAC} symbols are the start and end addresses of the code of
666 the block, respectively. For most machines, they are relative to the
667 starting address of this source file. For the Gould NP1, they are
668 absolute. For stabs in sections (@pxref{Stab Sections}), they are
669 relative to the function in which they occur.
671 The @code{N_LBRAC} and @code{N_RBRAC} stabs that describe the block
672 scope of a procedure are located after the @code{N_FUN} stab that
673 represents the procedure itself.
675 Sun documents the desc field of @code{N_LBRAC} and
676 @code{N_RBRAC} symbols as containing the nesting level of the block.
677 However, dbx seems to not care, and GCC always sets desc to
683 The @samp{c} symbol descriptor indicates that this stab represents a
684 constant. This symbol descriptor is an exception to the general rule
685 that symbol descriptors are followed by type information. Instead, it
686 is followed by @samp{=} and one of the following:
690 Boolean constant. @var{value} is a numeric value; I assume it is 0 for
694 Character constant. @var{value} is the numeric value of the constant.
696 @item e @var{type-information} , @var{value}
697 Constant whose value can be represented as integral.
698 @var{type-information} is the type of the constant, as it would appear
699 after a symbol descriptor (@pxref{String Field}). @var{value} is the
700 numeric value of the constant. GDB 4.9 does not actually get the right
701 value if @var{value} does not fit in a host @code{int}, but it does not
702 do anything violent, and future debuggers could be extended to accept
703 integers of any size (whether unsigned or not). This constant type is
704 usually documented as being only for enumeration constants, but GDB has
705 never imposed that restriction; I don't know about other debuggers.
708 Integer constant. @var{value} is the numeric value. The type is some
709 sort of generic integer type (for GDB, a host @code{int}); to specify
710 the type explicitly, use @samp{e} instead.
713 Real constant. @var{value} is the real value, which can be @samp{INF}
714 (optionally preceded by a sign) for infinity, @samp{QNAN} for a quiet
715 NaN (not-a-number), or @samp{SNAN} for a signalling NaN. If it is a
716 normal number the format is that accepted by the C library function
720 String constant. @var{string} is a string enclosed in either @samp{'}
721 (in which case @samp{'} characters within the string are represented as
722 @samp{\'} or @samp{"} (in which case @samp{"} characters within the
723 string are represented as @samp{\"}).
725 @item S @var{type-information} , @var{elements} , @var{bits} , @var{pattern}
726 Set constant. @var{type-information} is the type of the constant, as it
727 would appear after a symbol descriptor (@pxref{String Field}).
728 @var{elements} is the number of elements in the set (does this means
729 how many bits of @var{pattern} are actually used, which would be
730 redundant with the type, or perhaps the number of bits set in
731 @var{pattern}? I don't get it), @var{bits} is the number of bits in the
732 constant (meaning it specifies the length of @var{pattern}, I think),
733 and @var{pattern} is a hexadecimal representation of the set. AIX
734 documentation refers to a limit of 32 bytes, but I see no reason why
735 this limit should exist. This form could probably be used for arbitrary
736 constants, not just sets; the only catch is that @var{pattern} should be
737 understood to be target, not host, byte order and format.
740 The boolean, character, string, and set constants are not supported by
741 GDB 4.9, but it ignores them. GDB 4.8 and earlier gave an error
742 message and refused to read symbols from the file containing the
745 The above information is followed by @samp{;}.
750 Different types of stabs describe the various ways that variables can be
751 allocated: on the stack, globally, in registers, in common blocks,
752 statically, or as arguments to a function.
755 * Stack Variables:: Variables allocated on the stack.
756 * Global Variables:: Variables used by more than one source file.
757 * Register Variables:: Variables in registers.
758 * Common Blocks:: Variables statically allocated together.
759 * Statics:: Variables local to one source file.
760 * Based Variables:: Fortran pointer based variables.
761 * Parameters:: Variables for arguments to functions.
764 @node Stack Variables
765 @section Automatic Variables Allocated on the Stack
767 If a variable's scope is local to a function and its lifetime is only as
768 long as that function executes (C calls such variables
769 @dfn{automatic}), it can be allocated in a register (@pxref{Register
770 Variables}) or on the stack.
773 Each variable allocated on the stack has a stab with the symbol
774 descriptor omitted. Since type information should begin with a digit,
775 @samp{-}, or @samp{(}, only those characters precluded from being used
776 for symbol descriptors. However, the Acorn RISC machine (ARM) is said
777 to get this wrong: it puts out a mere type definition here, without the
778 preceding @samp{@var{type-number}=}. This is a bad idea; there is no
779 guarantee that type descriptors are distinct from symbol descriptors.
780 Stabs for stack variables use the @code{N_LSYM} stab type.
782 The value of the stab is the offset of the variable within the
783 local variables. On most machines this is an offset from the frame
784 pointer and is negative. The location of the stab specifies which block
785 it is defined in; see @ref{Block Structure}.
787 For example, the following C code:
797 produces the following stabs:
800 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
801 .stabs "x:1",128,0,0,-12 # @r{128 is N_LSYM}
802 .stabn 192,0,0,LBB2 # @r{192 is N_LBRAC}
803 .stabn 224,0,0,LBE2 # @r{224 is N_RBRAC}
806 @xref{Procedures} for more information on the @code{N_FUN} stab, and
807 @ref{Block Structure} for more information on the @code{N_LBRAC} and
808 @code{N_RBRAC} stabs.
810 @node Global Variables
811 @section Global Variables
814 A variable whose scope is not specific to just one source file is
815 represented by the @samp{G} symbol descriptor. These stabs use the
816 @code{N_GSYM} stab type. The type information for the stab
817 (@pxref{String Field}) gives the type of the variable.
819 For example, the following source code:
826 yields the following assembly code:
829 .stabs "g_foo:G2",32,0,0,0 # @r{32 is N_GSYM}
836 The address of the variable represented by the @code{N_GSYM} is not
837 contained in the @code{N_GSYM} stab. The debugger gets this information
838 from the external symbol for the global variable. In the example above,
839 the @code{.global _g_foo} and @code{_g_foo:} lines tell the assembler to
840 produce an external symbol.
842 @node Register Variables
843 @section Register Variables
846 @c According to an old version of this manual, AIX uses C_RPSYM instead
847 @c of C_RSYM. I am skeptical; this should be verified.
848 Register variables have their own stab type, @code{N_RSYM}, and their
849 own symbol descriptor, @samp{r}. The stab's value is the
850 number of the register where the variable data will be stored.
851 @c .stabs "name:type",N_RSYM,0,RegSize,RegNumber (Sun doc)
853 AIX defines a separate symbol descriptor @samp{d} for floating point
854 registers. This seems unnecessary; why not just just give floating
855 point registers different register numbers? I have not verified whether
856 the compiler actually uses @samp{d}.
858 If the register is explicitly allocated to a global variable, but not
862 register int g_bar asm ("%g5");
866 then the stab may be emitted at the end of the object file, with
867 the other bss symbols.
870 @section Common Blocks
872 A common block is a statically allocated section of memory which can be
873 referred to by several source files. It may contain several variables.
874 I believe Fortran is the only language with this feature.
880 A @code{N_BCOMM} stab begins a common block and an @code{N_ECOMM} stab
881 ends it. The only field that is significant in these two stabs is the
882 string, which names a normal (non-debugging) symbol that gives the
883 address of the common block. According to IBM documentation, only the
884 @code{N_BCOMM} has the name of the common block (even though their
885 compiler actually puts it both places).
889 The stabs for the members of the common block are between the
890 @code{N_BCOMM} and the @code{N_ECOMM}; the value of each stab is the
891 offset within the common block of that variable. IBM uses the
892 @code{C_ECOML} stab type, and there is a corresponding @code{N_ECOML}
893 stab type, but Sun's Fortran compiler uses @code{N_GSYM} instead. The
894 variables within a common block use the @samp{V} symbol descriptor (I
895 believe this is true of all Fortran variables). Other stabs (at least
896 type declarations using @code{C_DECL}) can also be between the
897 @code{N_BCOMM} and the @code{N_ECOMM}.
900 @section Static Variables
902 Initialized static variables are represented by the @samp{S} and
903 @samp{V} symbol descriptors. @samp{S} means file scope static, and
904 @samp{V} means procedure scope static. One exception: in XCOFF, IBM's
905 xlc compiler always uses @samp{V}, and whether it is file scope or not
906 is distinguished by whether the stab is located within a function.
908 @c This is probably not worth mentioning; it is only true on the sparc
909 @c for `double' variables which although declared const are actually in
910 @c the data segment (the text segment can't guarantee 8 byte alignment).
912 @c 2.4.5 has a bug in that it uses @code{N_FUN}, so neither dbx nor GDB can
913 @c find the variables)
916 @findex N_FUN, for variables
918 In a.out files, @code{N_STSYM} means the data section, @code{N_FUN}
919 means the text section, and @code{N_LCSYM} means the bss section. For
920 those systems with a read-only data section separate from the text
921 section (Solaris), @code{N_ROSYM} means the read-only data section.
923 For example, the source lines:
926 static const int var_const = 5;
927 static int var_init = 2;
928 static int var_noinit;
932 yield the following stabs:
935 .stabs "var_const:S1",36,0,0,_var_const # @r{36 is N_FUN}
937 .stabs "var_init:S1",38,0,0,_var_init # @r{38 is N_STSYM}
939 .stabs "var_noinit:S1",40,0,0,_var_noinit # @r{40 is N_LCSYM}
942 In XCOFF files, the stab type need not indicate the section;
943 @code{C_STSYM} can be used for all statics. Also, each static variable
944 is enclosed in a static block. A @code{C_BSTAT} (emitted with a
945 @samp{.bs} assembler directive) symbol begins the static block; its
946 value is the address of the static block, its section is the section of
947 the variables in that static block, and its name is @samp{.bs}. A
948 @code{C_ESTAT} (emitted with a @samp{.es} assembler directive) symbol
949 ends the static block; its name is @samp{.es} and its value and section
952 In ECOFF files, the storage class is used to specify the section, so the
953 stab type need not indicate the section.
955 In ELF files, for the SunPRO compiler version 2.0.1, symbol descriptor
956 @samp{S} means that the address is absolute (the linker relocates it)
957 and symbol descriptor @samp{V} means that the address is relative to the
958 start of the relevant section for that compilation unit. SunPRO has
959 plans to have the linker stop relocating stabs; I suspect that their the
960 debugger gets the address from the corresponding ELF (not stab) symbol.
961 I'm not sure how to find which symbol of that name is the right one.
962 The clean way to do all this would be to have a the value of a symbol
963 descriptor @samp{S} symbol be an offset relative to the start of the
964 file, just like everything else, but that introduces obvious
965 compatibility problems. For more information on linker stab relocation,
966 @xref{ELF Linker Relocation}.
968 @node Based Variables
969 @section Fortran Based Variables
971 Fortran (at least, the Sun and SGI dialects of FORTRAN-77) has a feature
972 which allows allocating arrays with @code{malloc}, but which avoids
973 blurring the line between arrays and pointers the way that C does. In
974 stabs such a variable uses the @samp{b} symbol descriptor.
976 For example, the Fortran declarations
979 real foo, foo10(10), foo10_5(10,5)
981 pointer (foo10p, foo10)
982 pointer (foo105p, foo10_5)
990 foo10_5:bar3;1;5;ar3;1;10;6
993 In this example, @code{real} is type 6 and type 3 is an integral type
994 which is the type of the subscripts of the array (probably
997 The @samp{b} symbol descriptor is like @samp{V} in that it denotes a
998 statically allocated symbol whose scope is local to a function; see
999 @xref{Statics}. The value of the symbol, instead of being the address
1000 of the variable itself, is the address of a pointer to that variable.
1001 So in the above example, the value of the @code{foo} stab is the address
1002 of a pointer to a real, the value of the @code{foo10} stab is the
1003 address of a pointer to a 10-element array of reals, and the value of
1004 the @code{foo10_5} stab is the address of a pointer to a 5-element array
1005 of 10-element arrays of reals.
1010 Formal parameters to a function are represented by a stab (or sometimes
1011 two; see below) for each parameter. The stabs are in the order in which
1012 the debugger should print the parameters (i.e., the order in which the
1013 parameters are declared in the source file). The exact form of the stab
1014 depends on how the parameter is being passed.
1017 Parameters passed on the stack use the symbol descriptor @samp{p} and
1018 the @code{N_PSYM} symbol type. The value of the symbol is an offset
1019 used to locate the parameter on the stack; its exact meaning is
1020 machine-dependent, but on most machines it is an offset from the frame
1023 As a simple example, the code:
1034 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
1035 .stabs "argc:p1",160,0,0,68 # @r{160 is N_PSYM}
1036 .stabs "argv:p20=*21=*2",160,0,0,72
1039 The type definition of @code{argv} is interesting because it contains
1040 several type definitions. Type 21 is pointer to type 2 (char) and
1041 @code{argv} (type 20) is pointer to type 21.
1043 @c FIXME: figure out what these mean and describe them coherently.
1044 The following symbol descriptors are also said to go with @code{N_PSYM}.
1045 The value of the symbol is said to be an offset from the argument
1046 pointer (I'm not sure whether this is true or not).
1050 pF Fortran function parameter
1051 X (function result variable)
1055 * Register Parameters::
1056 * Local Variable Parameters::
1057 * Reference Parameters::
1058 * Conformant Arrays::
1061 @node Register Parameters
1062 @subsection Passing Parameters in Registers
1064 If the parameter is passed in a register, then traditionally there are
1065 two symbols for each argument:
1068 .stabs "arg:p1" . . . ; N_PSYM
1069 .stabs "arg:r1" . . . ; N_RSYM
1072 Debuggers use the second one to find the value, and the first one to
1073 know that it is an argument.
1076 @findex N_RSYM, for parameters
1077 Because that approach is kind of ugly, some compilers use symbol
1078 descriptor @samp{P} or @samp{R} to indicate an argument which is in a
1079 register. Symbol type @code{C_RPSYM} is used with @samp{R} and
1080 @code{N_RSYM} is used with @samp{P}. The symbol's value is
1081 the register number. @samp{P} and @samp{R} mean the same thing; the
1082 difference is that @samp{P} is a GNU invention and @samp{R} is an IBM
1083 (XCOFF) invention. As of version 4.9, GDB should handle either one.
1085 There is at least one case where GCC uses a @samp{p} and @samp{r} pair
1086 rather than @samp{P}; this is where the argument is passed in the
1087 argument list and then loaded into a register.
1089 According to the AIX documentation, symbol descriptor @samp{D} is for a
1090 parameter passed in a floating point register. This seems
1091 unnecessary---why not just use @samp{R} with a register number which
1092 indicates that it's a floating point register? I haven't verified
1093 whether the system actually does what the documentation indicates.
1095 @c FIXME: On the hppa this is for any type > 8 bytes, I think, and not
1096 @c for small structures (investigate).
1097 On the sparc and hppa, for a @samp{P} symbol whose type is a structure
1098 or union, the register contains the address of the structure. On the
1099 sparc, this is also true of a @samp{p} and @samp{r} pair (using Sun
1100 @code{cc}) or a @samp{p} symbol. However, if a (small) structure is
1101 really in a register, @samp{r} is used. And, to top it all off, on the
1102 hppa it might be a structure which was passed on the stack and loaded
1103 into a register and for which there is a @samp{p} and @samp{r} pair! I
1104 believe that symbol descriptor @samp{i} is supposed to deal with this
1105 case (it is said to mean "value parameter by reference, indirect
1106 access"; I don't know the source for this information), but I don't know
1107 details or what compilers or debuggers use it, if any (not GDB or GCC).
1108 It is not clear to me whether this case needs to be dealt with
1109 differently than parameters passed by reference (@pxref{Reference Parameters}).
1111 @node Local Variable Parameters
1112 @subsection Storing Parameters as Local Variables
1114 There is a case similar to an argument in a register, which is an
1115 argument that is actually stored as a local variable. Sometimes this
1116 happens when the argument was passed in a register and then the compiler
1117 stores it as a local variable. If possible, the compiler should claim
1118 that it's in a register, but this isn't always done.
1120 If a parameter is passed as one type and converted to a smaller type by
1121 the prologue (for example, the parameter is declared as a @code{float},
1122 but the calling conventions specify that it is passed as a
1123 @code{double}), then GCC2 (sometimes) uses a pair of symbols. The first
1124 symbol uses symbol descriptor @samp{p} and the type which is passed.
1125 The second symbol has the type and location which the parameter actually
1126 has after the prologue. For example, suppose the following C code
1127 appears with no prototypes involved:
1136 if @code{f} is passed as a double at stack offset 8, and the prologue
1137 converts it to a float in register number 0, then the stabs look like:
1140 .stabs "f:p13",160,0,3,8 # @r{160 is @code{N_PSYM}, here 13 is @code{double}}
1141 .stabs "f:r12",64,0,3,0 # @r{64 is @code{N_RSYM}, here 12 is @code{float}}
1144 In both stabs 3 is the line number where @code{f} is declared
1145 (@pxref{Line Numbers}).
1147 @findex N_LSYM, for parameter
1148 GCC, at least on the 960, has another solution to the same problem. It
1149 uses a single @samp{p} symbol descriptor for an argument which is stored
1150 as a local variable but uses @code{N_LSYM} instead of @code{N_PSYM}. In
1151 this case, the value of the symbol is an offset relative to the local
1152 variables for that function, not relative to the arguments; on some
1153 machines those are the same thing, but not on all.
1155 @c This is mostly just background info; the part that logically belongs
1156 @c here is the last sentence.
1157 On the VAX or on other machines in which the calling convention includes
1158 the number of words of arguments actually passed, the debugger (GDB at
1159 least) uses the parameter symbols to keep track of whether it needs to
1160 print nameless arguments in addition to the formal parameters which it
1161 has printed because each one has a stab. For example, in
1164 extern int fprintf (FILE *stream, char *format, @dots{});
1166 fprintf (stdout, "%d\n", x);
1169 there are stabs for @code{stream} and @code{format}. On most machines,
1170 the debugger can only print those two arguments (because it has no way
1171 of knowing that additional arguments were passed), but on the VAX or
1172 other machines with a calling convention which indicates the number of
1173 words of arguments, the debugger can print all three arguments. To do
1174 so, the parameter symbol (symbol descriptor @samp{p}) (not necessarily
1175 @samp{r} or symbol descriptor omitted symbols) needs to contain the
1176 actual type as passed (for example, @code{double} not @code{float} if it
1177 is passed as a double and converted to a float).
1179 @node Reference Parameters
1180 @subsection Passing Parameters by Reference
1182 If the parameter is passed by reference (e.g., Pascal @code{VAR}
1183 parameters), then the symbol descriptor is @samp{v} if it is in the
1184 argument list, or @samp{a} if it in a register. Other than the fact
1185 that these contain the address of the parameter rather than the
1186 parameter itself, they are identical to @samp{p} and @samp{R},
1187 respectively. I believe @samp{a} is an AIX invention; @samp{v} is
1188 supported by all stabs-using systems as far as I know.
1190 @node Conformant Arrays
1191 @subsection Passing Conformant Array Parameters
1193 @c Is this paragraph correct? It is based on piecing together patchy
1194 @c information and some guesswork
1195 Conformant arrays are a feature of Modula-2, and perhaps other
1196 languages, in which the size of an array parameter is not known to the
1197 called function until run-time. Such parameters have two stabs: a
1198 @samp{x} for the array itself, and a @samp{C}, which represents the size
1199 of the array. The value of the @samp{x} stab is the offset in the
1200 argument list where the address of the array is stored (it this right?
1201 it is a guess); the value of the @samp{C} stab is the offset in the
1202 argument list where the size of the array (in elements? in bytes?) is
1206 @chapter Defining Types
1208 The examples so far have described types as references to previously
1209 defined types, or defined in terms of subranges of or pointers to
1210 previously defined types. This chapter describes the other type
1211 descriptors that may follow the @samp{=} in a type definition.
1214 * Builtin Types:: Integers, floating point, void, etc.
1215 * Miscellaneous Types:: Pointers, sets, files, etc.
1216 * Cross-References:: Referring to a type not yet defined.
1217 * Subranges:: A type with a specific range.
1218 * Arrays:: An aggregate type of same-typed elements.
1219 * Strings:: Like an array but also has a length.
1220 * Enumerations:: Like an integer but the values have names.
1221 * Structures:: An aggregate type of different-typed elements.
1222 * Typedefs:: Giving a type a name.
1223 * Unions:: Different types sharing storage.
1228 @section Builtin Types
1230 Certain types are built in (@code{int}, @code{short}, @code{void},
1231 @code{float}, etc.); the debugger recognizes these types and knows how
1232 to handle them. Thus, don't be surprised if some of the following ways
1233 of specifying builtin types do not specify everything that a debugger
1234 would need to know about the type---in some cases they merely specify
1235 enough information to distinguish the type from other types.
1237 The traditional way to define builtin types is convolunted, so new ways
1238 have been invented to describe them. Sun's @code{acc} uses special
1239 builtin type descriptors (@samp{b} and @samp{R}), and IBM uses negative
1240 type numbers. GDB accepts all three ways, as of version 4.8; dbx just
1241 accepts the traditional builtin types and perhaps one of the other two
1242 formats. The following sections describe each of these formats.
1245 * Traditional Builtin Types:: Put on your seatbelts and prepare for kludgery
1246 * Builtin Type Descriptors:: Builtin types with special type descriptors
1247 * Negative Type Numbers:: Builtin types using negative type numbers
1250 @node Traditional Builtin Types
1251 @subsection Traditional Builtin Types
1253 This is the traditional, convoluted method for defining builtin types.
1254 There are several classes of such type definitions: integer, floating
1255 point, and @code{void}.
1258 * Traditional Integer Types::
1259 * Traditional Other Types::
1262 @node Traditional Integer Types
1263 @subsubsection Traditional Integer Types
1265 Often types are defined as subranges of themselves. If the bounding values
1266 fit within an @code{int}, then they are given normally. For example:
1269 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0 # @r{128 is N_LSYM}
1270 .stabs "char:t2=r2;0;127;",128,0,0,0
1273 Builtin types can also be described as subranges of @code{int}:
1276 .stabs "unsigned short:t6=r1;0;65535;",128,0,0,0
1279 If the lower bound of a subrange is 0 and the upper bound is -1,
1280 the type is an unsigned integral type whose bounds are too
1281 big to describe in an @code{int}. Traditionally this is only used for
1282 @code{unsigned int} and @code{unsigned long}:
1285 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
1288 For larger types, GCC 2.4.5 puts out bounds in octal, with one or more
1289 leading zeroes. In this case a negative bound consists of a number
1290 which is a 1 bit (for the sign bit) followed by a 0 bit for each bit in
1291 the number (except the sign bit), and a positive bound is one which is a
1292 1 bit for each bit in the number (except possibly the sign bit). All
1293 known versions of dbx and GDB version 4 accept this (at least in the
1294 sense of not refusing to process the file), but GDB 3.5 refuses to read
1295 the whole file containing such symbols. So GCC 2.3.3 did not output the
1296 proper size for these types. As an example of octal bounds, the string
1297 fields of the stabs for 64 bit integer types look like:
1299 @c .stabs directives, etc., omitted to make it fit on the page.
1301 long int:t3=r1;001000000000000000000000;000777777777777777777777;
1302 long unsigned int:t5=r1;000000000000000000000000;001777777777777777777777;
1305 If the lower bound of a subrange is 0 and the upper bound is negative,
1306 the type is an unsigned integral type whose size in bytes is the
1307 absolute value of the upper bound. I believe this is a Convex
1308 convention for @code{unsigned long long}.
1310 If the lower bound of a subrange is negative and the upper bound is 0,
1311 the type is a signed integral type whose size in bytes is
1312 the absolute value of the lower bound. I believe this is a Convex
1313 convention for @code{long long}. To distinguish this from a legitimate
1314 subrange, the type should be a subrange of itself. I'm not sure whether
1315 this is the case for Convex.
1317 @node Traditional Other Types
1318 @subsubsection Traditional Other Types
1320 If the upper bound of a subrange is 0 and the lower bound is positive,
1321 the type is a floating point type, and the lower bound of the subrange
1322 indicates the number of bytes in the type:
1325 .stabs "float:t12=r1;4;0;",128,0,0,0
1326 .stabs "double:t13=r1;8;0;",128,0,0,0
1329 However, GCC writes @code{long double} the same way it writes
1330 @code{double}, so there is no way to distinguish.
1333 .stabs "long double:t14=r1;8;0;",128,0,0,0
1336 Complex types are defined the same way as floating-point types; there is
1337 no way to distinguish a single-precision complex from a double-precision
1338 floating-point type.
1340 The C @code{void} type is defined as itself:
1343 .stabs "void:t15=15",128,0,0,0
1346 I'm not sure how a boolean type is represented.
1348 @node Builtin Type Descriptors
1349 @subsection Defining Builtin Types Using Builtin Type Descriptors
1351 This is the method used by Sun's @code{acc} for defining builtin types.
1352 These are the type descriptors to define builtin types:
1355 @c FIXME: clean up description of width and offset, once we figure out
1357 @item b @var{signed} @var{char-flag} @var{width} ; @var{offset} ; @var{nbits} ;
1358 Define an integral type. @var{signed} is @samp{u} for unsigned or
1359 @samp{s} for signed. @var{char-flag} is @samp{c} which indicates this
1360 is a character type, or is omitted. I assume this is to distinguish an
1361 integral type from a character type of the same size, for example it
1362 might make sense to set it for the C type @code{wchar_t} so the debugger
1363 can print such variables differently (Solaris does not do this). Sun
1364 sets it on the C types @code{signed char} and @code{unsigned char} which
1365 arguably is wrong. @var{width} and @var{offset} appear to be for small
1366 objects stored in larger ones, for example a @code{short} in an
1367 @code{int} register. @var{width} is normally the number of bytes in the
1368 type. @var{offset} seems to always be zero. @var{nbits} is the number
1369 of bits in the type.
1371 Note that type descriptor @samp{b} used for builtin types conflicts with
1372 its use for Pascal space types (@pxref{Miscellaneous Types}); they can
1373 be distinguished because the character following the type descriptor
1374 will be a digit, @samp{(}, or @samp{-} for a Pascal space type, or
1375 @samp{u} or @samp{s} for a builtin type.
1378 Documented by AIX to define a wide character type, but their compiler
1379 actually uses negative type numbers (@pxref{Negative Type Numbers}).
1381 @item R @var{fp-type} ; @var{bytes} ;
1382 Define a floating point type. @var{fp-type} has one of the following values:
1386 IEEE 32-bit (single precision) floating point format.
1389 IEEE 64-bit (double precision) floating point format.
1391 @item 3 (NF_COMPLEX)
1392 @item 4 (NF_COMPLEX16)
1393 @item 5 (NF_COMPLEX32)
1394 @c "GDB source" really means @file{include/aout/stab_gnu.h}, but trying
1395 @c to put that here got an overfull hbox.
1396 These are for complex numbers. A comment in the GDB source describes
1397 them as Fortran @code{complex}, @code{double complex}, and
1398 @code{complex*16}, respectively, but what does that mean? (i.e., Single
1399 precision? Double precison?).
1401 @item 6 (NF_LDOUBLE)
1402 Long double. This should probably only be used for Sun format
1403 @code{long double}, and new codes should be used for other floating
1404 point formats (@code{NF_DOUBLE} can be used if a @code{long double} is
1405 really just an IEEE double, of course).
1408 @var{bytes} is the number of bytes occupied by the type. This allows a
1409 debugger to perform some operations with the type even if it doesn't
1410 understand @var{fp-type}.
1412 @item g @var{type-information} ; @var{nbits}
1413 Documented by AIX to define a floating type, but their compiler actually
1414 uses negative type numbers (@pxref{Negative Type Numbers}).
1416 @item c @var{type-information} ; @var{nbits}
1417 Documented by AIX to define a complex type, but their compiler actually
1418 uses negative type numbers (@pxref{Negative Type Numbers}).
1421 The C @code{void} type is defined as a signed integral type 0 bits long:
1423 .stabs "void:t19=bs0;0;0",128,0,0,0
1425 The Solaris compiler seems to omit the trailing semicolon in this case.
1426 Getting sloppy in this way is not a swift move because if a type is
1427 embedded in a more complex expression it is necessary to be able to tell
1430 I'm not sure how a boolean type is represented.
1432 @node Negative Type Numbers
1433 @subsection Negative Type Numbers
1435 This is the method used in XCOFF for defining builtin types.
1436 Since the debugger knows about the builtin types anyway, the idea of
1437 negative type numbers is simply to give a special type number which
1438 indicates the builtin type. There is no stab defining these types.
1440 There are several subtle issues with negative type numbers.
1442 One is the size of the type. A builtin type (for example the C types
1443 @code{int} or @code{long}) might have different sizes depending on
1444 compiler options, the target architecture, the ABI, etc. This issue
1445 doesn't come up for IBM tools since (so far) they just target the
1446 RS/6000; the sizes indicated below for each size are what the IBM
1447 RS/6000 tools use. To deal with differing sizes, either define separate
1448 negative type numbers for each size (which works but requires changing
1449 the debugger, and, unless you get both AIX dbx and GDB to accept the
1450 change, introduces an incompatibility), or use a type attribute
1451 (@pxref{String Field}) to define a new type with the appropriate size
1452 (which merely requires a debugger which understands type attributes,
1453 like AIX dbx). For example,
1456 .stabs "boolean:t10=@@s8;-16",128,0,0,0
1459 defines an 8-bit boolean type, and
1462 .stabs "boolean:t10=@@s64;-16",128,0,0,0
1465 defines a 64-bit boolean type.
1467 A similar issue is the format of the type. This comes up most often for
1468 floating-point types, which could have various formats (particularly
1469 extended doubles, which vary quite a bit even among IEEE systems).
1470 Again, it is best to define a new negative type number for each
1471 different format; changing the format based on the target system has
1472 various problems. One such problem is that the Alpha has both VAX and
1473 IEEE floating types. One can easily imagine one library using the VAX
1474 types and another library in the same executable using the IEEE types.
1475 Another example is that the interpretation of whether a boolean is true
1476 or false can be based on the least significant bit, most significant
1477 bit, whether it is zero, etc., and different compilers (or different
1478 options to the same compiler) might provide different kinds of boolean.
1480 The last major issue is the names of the types. The name of a given
1481 type depends @emph{only} on the negative type number given; these do not
1482 vary depending on the language, the target system, or anything else.
1483 One can always define separate type numbers---in the following list you
1484 will see for example separate @code{int} and @code{integer*4} types
1485 which are identical except for the name. But compatibility can be
1486 maintained by not inventing new negative type numbers and instead just
1487 defining a new type with a new name. For example:
1490 .stabs "CARDINAL:t10=-8",128,0,0,0
1493 Here is the list of negative type numbers. The phrase @dfn{integral
1494 type} is used to mean twos-complement (I strongly suspect that all
1495 machines which use stabs use twos-complement; most machines use
1496 twos-complement these days).
1500 @code{int}, 32 bit signed integral type.
1503 @code{char}, 8 bit type holding a character. Both GDB and dbx on AIX
1504 treat this as signed. GCC uses this type whether @code{char} is signed
1505 or not, which seems like a bad idea. The AIX compiler (@code{xlc}) seems to
1506 avoid this type; it uses -5 instead for @code{char}.
1509 @code{short}, 16 bit signed integral type.
1512 @code{long}, 32 bit signed integral type.
1515 @code{unsigned char}, 8 bit unsigned integral type.
1518 @code{signed char}, 8 bit signed integral type.
1521 @code{unsigned short}, 16 bit unsigned integral type.
1524 @code{unsigned int}, 32 bit unsigned integral type.
1527 @code{unsigned}, 32 bit unsigned integral type.
1530 @code{unsigned long}, 32 bit unsigned integral type.
1533 @code{void}, type indicating the lack of a value.
1536 @code{float}, IEEE single precision.
1539 @code{double}, IEEE double precision.
1542 @code{long double}, IEEE double precision. The compiler claims the size
1543 will increase in a future release, and for binary compatibility you have
1544 to avoid using @code{long double}. I hope when they increase it they
1545 use a new negative type number.
1548 @code{integer}. 32 bit signed integral type.
1551 @code{boolean}. 32 bit type. GDB and GCC assume that zero is false,
1552 one is true, and other values have unspecified meaning. I hope this
1553 agrees with how the IBM tools use the type.
1556 @code{short real}. IEEE single precision.
1559 @code{real}. IEEE double precision.
1562 @code{stringptr}. @xref{Strings}.
1565 @code{character}, 8 bit unsigned character type.
1568 @code{logical*1}, 8 bit type. This Fortran type has a split
1569 personality in that it is used for boolean variables, but can also be
1570 used for unsigned integers. 0 is false, 1 is true, and other values are
1574 @code{logical*2}, 16 bit type. This Fortran type has a split
1575 personality in that it is used for boolean variables, but can also be
1576 used for unsigned integers. 0 is false, 1 is true, and other values are
1580 @code{logical*4}, 32 bit type. This Fortran type has a split
1581 personality in that it is used for boolean variables, but can also be
1582 used for unsigned integers. 0 is false, 1 is true, and other values are
1586 @code{logical}, 32 bit type. This Fortran type has a split
1587 personality in that it is used for boolean variables, but can also be
1588 used for unsigned integers. 0 is false, 1 is true, and other values are
1592 @code{complex}. A complex type consisting of two IEEE single-precision
1593 floating point values.
1596 @code{complex}. A complex type consisting of two IEEE double-precision
1597 floating point values.
1600 @code{integer*1}, 8 bit signed integral type.
1603 @code{integer*2}, 16 bit signed integral type.
1606 @code{integer*4}, 32 bit signed integral type.
1609 @code{wchar}. Wide character, 16 bits wide, unsigned (what format?
1613 @node Miscellaneous Types
1614 @section Miscellaneous Types
1617 @item b @var{type-information} ; @var{bytes}
1618 Pascal space type. This is documented by IBM; what does it mean?
1620 This use of the @samp{b} type descriptor can be distinguished
1621 from its use for builtin integral types (@pxref{Builtin Type
1622 Descriptors}) because the character following the type descriptor is
1623 always a digit, @samp{(}, or @samp{-}.
1625 @item B @var{type-information}
1626 A volatile-qualified version of @var{type-information}. This is
1627 a Sun extension. References and stores to a variable with a
1628 volatile-qualified type must not be optimized or cached; they
1629 must occur as the user specifies them.
1631 @item d @var{type-information}
1632 File of type @var{type-information}. As far as I know this is only used
1635 @item k @var{type-information}
1636 A const-qualified version of @var{type-information}. This is a Sun
1637 extension. A variable with a const-qualified type cannot be modified.
1639 @item M @var{type-information} ; @var{length}
1640 Multiple instance type. The type seems to composed of @var{length}
1641 repetitions of @var{type-information}, for example @code{character*3} is
1642 represented by @samp{M-2;3}, where @samp{-2} is a reference to a
1643 character type (@pxref{Negative Type Numbers}). I'm not sure how this
1644 differs from an array. This appears to be a Fortran feature.
1645 @var{length} is a bound, like those in range types; see @ref{Subranges}.
1647 @item S @var{type-information}
1648 Pascal set type. @var{type-information} must be a small type such as an
1649 enumeration or a subrange, and the type is a bitmask whose length is
1650 specified by the number of elements in @var{type-information}.
1652 In CHILL, if it is a bitstring instead of a set, also use the @samp{S}
1653 type attribute (@pxref{String Field}).
1655 @item * @var{type-information}
1656 Pointer to @var{type-information}.
1659 @node Cross-References
1660 @section Cross-References to Other Types
1662 A type can be used before it is defined; one common way to deal with
1663 that situation is just to use a type reference to a type which has not
1666 Another way is with the @samp{x} type descriptor, which is followed by
1667 @samp{s} for a structure tag, @samp{u} for a union tag, or @samp{e} for
1668 a enumerator tag, followed by the name of the tag, followed by @samp{:}.
1669 If the name contains @samp{::} between a @samp{<} and @samp{>} pair (for
1670 C++ templates), such a @samp{::} does not end the name---only a single
1671 @samp{:} ends the name; see @ref{Nested Symbols}.
1673 For example, the following C declarations:
1684 .stabs "bar:G16=*17=xsfoo:",32,0,0,0
1687 Not all debuggers support the @samp{x} type descriptor, so on some
1688 machines GCC does not use it. I believe that for the above example it
1689 would just emit a reference to type 17 and never define it, but I
1690 haven't verified that.
1692 Modula-2 imported types, at least on AIX, use the @samp{i} type
1693 descriptor, which is followed by the name of the module from which the
1694 type is imported, followed by @samp{:}, followed by the name of the
1695 type. There is then optionally a comma followed by type information for
1696 the type. This differs from merely naming the type (@pxref{Typedefs}) in
1697 that it identifies the module; I don't understand whether the name of
1698 the type given here is always just the same as the name we are giving
1699 it, or whether this type descriptor is used with a nameless stab
1700 (@pxref{String Field}), or what. The symbol ends with @samp{;}.
1703 @section Subrange Types
1705 The @samp{r} type descriptor defines a type as a subrange of another
1706 type. It is followed by type information for the type of which it is a
1707 subrange, a semicolon, an integral lower bound, a semicolon, an
1708 integral upper bound, and a semicolon. The AIX documentation does not
1709 specify the trailing semicolon, in an effort to specify array indexes
1710 more cleanly, but a subrange which is not an array index has always
1711 included a trailing semicolon (@pxref{Arrays}).
1713 Instead of an integer, either bound can be one of the following:
1716 @item A @var{offset}
1717 The bound is passed by reference on the stack at offset @var{offset}
1718 from the argument list. @xref{Parameters}, for more information on such
1721 @item T @var{offset}
1722 The bound is passed by value on the stack at offset @var{offset} from
1725 @item a @var{register-number}
1726 The bound is pased by reference in register number
1727 @var{register-number}.
1729 @item t @var{register-number}
1730 The bound is passed by value in register number @var{register-number}.
1736 Subranges are also used for builtin types; see @ref{Traditional Builtin Types}.
1739 @section Array Types
1741 Arrays use the @samp{a} type descriptor. Following the type descriptor
1742 is the type of the index and the type of the array elements. If the
1743 index type is a range type, it ends in a semicolon; otherwise
1744 (for example, if it is a type reference), there does not
1745 appear to be any way to tell where the types are separated. In an
1746 effort to clean up this mess, IBM documents the two types as being
1747 separated by a semicolon, and a range type as not ending in a semicolon
1748 (but this is not right for range types which are not array indexes,
1749 @pxref{Subranges}). I think probably the best solution is to specify
1750 that a semicolon ends a range type, and that the index type and element
1751 type of an array are separated by a semicolon, but that if the index
1752 type is a range type, the extra semicolon can be omitted. GDB (at least
1753 through version 4.9) doesn't support any kind of index type other than a
1754 range anyway; I'm not sure about dbx.
1756 It is well established, and widely used, that the type of the index,
1757 unlike most types found in the stabs, is merely a type definition, not
1758 type information (@pxref{String Field}) (that is, it need not start with
1759 @samp{@var{type-number}=} if it is defining a new type). According to a
1760 comment in GDB, this is also true of the type of the array elements; it
1761 gives @samp{ar1;1;10;ar1;1;10;4} as a legitimate way to express a two
1762 dimensional array. According to AIX documentation, the element type
1763 must be type information. GDB accepts either.
1765 The type of the index is often a range type, expressed as the type
1766 descriptor @samp{r} and some parameters. It defines the size of the
1767 array. In the example below, the range @samp{r1;0;2;} defines an index
1768 type which is a subrange of type 1 (integer), with a lower bound of 0
1769 and an upper bound of 2. This defines the valid range of subscripts of
1770 a three-element C array.
1772 For example, the definition:
1775 char char_vec[3] = @{'a','b','c'@};
1779 produces the output:
1782 .stabs "char_vec:G19=ar1;0;2;2",32,0,0,0
1791 If an array is @dfn{packed}, the elements are spaced more
1792 closely than normal, saving memory at the expense of speed. For
1793 example, an array of 3-byte objects might, if unpacked, have each
1794 element aligned on a 4-byte boundary, but if packed, have no padding.
1795 One way to specify that something is packed is with type attributes
1796 (@pxref{String Field}). In the case of arrays, another is to use the
1797 @samp{P} type descriptor instead of @samp{a}. Other than specifying a
1798 packed array, @samp{P} is identical to @samp{a}.
1800 @c FIXME-what is it? A pointer?
1801 An open array is represented by the @samp{A} type descriptor followed by
1802 type information specifying the type of the array elements.
1804 @c FIXME: what is the format of this type? A pointer to a vector of pointers?
1805 An N-dimensional dynamic array is represented by
1808 D @var{dimensions} ; @var{type-information}
1811 @c Does dimensions really have this meaning? The AIX documentation
1813 @var{dimensions} is the number of dimensions; @var{type-information}
1814 specifies the type of the array elements.
1816 @c FIXME: what is the format of this type? A pointer to some offsets in
1818 A subarray of an N-dimensional array is represented by
1821 E @var{dimensions} ; @var{type-information}
1824 @c Does dimensions really have this meaning? The AIX documentation
1826 @var{dimensions} is the number of dimensions; @var{type-information}
1827 specifies the type of the array elements.
1832 Some languages, like C or the original Pascal, do not have string types,
1833 they just have related things like arrays of characters. But most
1834 Pascals and various other languages have string types, which are
1835 indicated as follows:
1838 @item n @var{type-information} ; @var{bytes}
1839 @var{bytes} is the maximum length. I'm not sure what
1840 @var{type-information} is; I suspect that it means that this is a string
1841 of @var{type-information} (thus allowing a string of integers, a string
1842 of wide characters, etc., as well as a string of characters). Not sure
1843 what the format of this type is. This is an AIX feature.
1845 @item z @var{type-information} ; @var{bytes}
1846 Just like @samp{n} except that this is a gstring, not an ordinary
1847 string. I don't know the difference.
1850 Pascal Stringptr. What is this? This is an AIX feature.
1853 Languages, such as CHILL which have a string type which is basically
1854 just an array of characters use the @samp{S} type attribute
1855 (@pxref{String Field}).
1858 @section Enumerations
1860 Enumerations are defined with the @samp{e} type descriptor.
1862 @c FIXME: Where does this information properly go? Perhaps it is
1863 @c redundant with something we already explain.
1864 The source line below declares an enumeration type at file scope.
1865 The type definition is located after the @code{N_RBRAC} that marks the end of
1866 the previous procedure's block scope, and before the @code{N_FUN} that marks
1867 the beginning of the next procedure's block scope. Therefore it does not
1868 describe a block local symbol, but a file local one.
1873 enum e_places @{first,second=3,last@};
1877 generates the following stab:
1880 .stabs "e_places:T22=efirst:0,second:3,last:4,;",128,0,0,0
1883 The symbol descriptor (@samp{T}) says that the stab describes a
1884 structure, enumeration, or union tag. The type descriptor @samp{e},
1885 following the @samp{22=} of the type definition narrows it down to an
1886 enumeration type. Following the @samp{e} is a list of the elements of
1887 the enumeration. The format is @samp{@var{name}:@var{value},}. The
1888 list of elements ends with @samp{;}. The fact that @var{value} is
1889 specified as an integer can cause problems if the value is large. GCC
1890 2.5.2 tries to output it in octal in that case with a leading zero,
1891 which is probably a good thing, although GDB 4.11 supports octal only in
1892 cases where decimal is perfectly good. Negative decimal values are
1893 supported by both GDB and dbx.
1895 There is no standard way to specify the size of an enumeration type; it
1896 is determined by the architecture (normally all enumerations types are
1897 32 bits). Type attributes can be used to specify an enumeration type of
1898 another size for debuggers which support them; see @ref{String Field}.
1900 Enumeration types are unusual in that they define symbols for the
1901 enumeration values (@code{first}, @code{second}, and @code{third} in the
1902 above example), and even though these symbols are visible in the file as
1903 a whole (rather than being in a more local namespace like structure
1904 member names), they are defined in the type definition for the
1905 enumeration type rather than each having their own symbol. In order to
1906 be fast, GDB will only get symbols from such types (in its initial scan
1907 of the stabs) if the type is the first thing defined after a @samp{T} or
1908 @samp{t} symbol descriptor (the above example fulfills this
1909 requirement). If the type does not have a name, the compiler should
1910 emit it in a nameless stab (@pxref{String Field}); GCC does this.
1915 The encoding of structures in stabs can be shown with an example.
1917 The following source code declares a structure tag and defines an
1918 instance of the structure in global scope. Then a @code{typedef} equates the
1919 structure tag with a new type. Seperate stabs are generated for the
1920 structure tag, the structure @code{typedef}, and the structure instance. The
1921 stabs for the tag and the @code{typedef} are emited when the definitions are
1922 encountered. Since the structure elements are not initialized, the
1923 stab and code for the structure variable itself is located at the end
1924 of the program in the bss section.
1931 struct s_tag* s_next;
1934 typedef struct s_tag s_typedef;
1937 The structure tag has an @code{N_LSYM} stab type because, like the
1938 enumeration, the symbol has file scope. Like the enumeration, the
1939 symbol descriptor is @samp{T}, for enumeration, structure, or tag type.
1940 The type descriptor @samp{s} following the @samp{16=} of the type
1941 definition narrows the symbol type to structure.
1943 Following the @samp{s} type descriptor is the number of bytes the
1944 structure occupies, followed by a description of each structure element.
1945 The structure element descriptions are of the form @var{name:type, bit
1946 offset from the start of the struct, number of bits in the element}.
1948 @c FIXME: phony line break. Can probably be fixed by using an example
1949 @c with fewer fields.
1952 .stabs "s_tag:T16=s20s_int:1,0,32;s_float:12,32,32;
1953 s_char_vec:17=ar1;0;7;2,64,64;s_next:18=*16,128,32;;",128,0,0,0
1956 In this example, the first two structure elements are previously defined
1957 types. For these, the type following the @samp{@var{name}:} part of the
1958 element description is a simple type reference. The other two structure
1959 elements are new types. In this case there is a type definition
1960 embedded after the @samp{@var{name}:}. The type definition for the
1961 array element looks just like a type definition for a standalone array.
1962 The @code{s_next} field is a pointer to the same kind of structure that
1963 the field is an element of. So the definition of structure type 16
1964 contains a type definition for an element which is a pointer to type 16.
1966 If a field is a static member (this is a C++ feature in which a single
1967 variable appears to be a field of every structure of a given type) it
1968 still starts out with the field name, a colon, and the type, but then
1969 instead of a comma, bit position, comma, and bit size, there is a colon
1970 followed by the name of the variable which each such field refers to.
1972 If the structure has methods (a C++ feature), they follow the non-method
1973 fields; see @ref{Cplusplus}.
1976 @section Giving a Type a Name
1978 To give a type a name, use the @samp{t} symbol descriptor. The type
1979 is specified by the type information (@pxref{String Field}) for the stab.
1983 .stabs "s_typedef:t16",128,0,0,0 # @r{128 is N_LSYM}
1986 specifies that @code{s_typedef} refers to type number 16. Such stabs
1987 have symbol type @code{N_LSYM} (or @code{C_DECL} for XCOFF).
1989 If you are specifying the tag name for a structure, union, or
1990 enumeration, use the @samp{T} symbol descriptor instead. I believe C is
1991 the only language with this feature.
1993 If the type is an opaque type (I believe this is a Modula-2 feature),
1994 AIX provides a type descriptor to specify it. The type descriptor is
1995 @samp{o} and is followed by a name. I don't know what the name
1996 means---is it always the same as the name of the type, or is this type
1997 descriptor used with a nameless stab (@pxref{String Field})? There
1998 optionally follows a comma followed by type information which defines
1999 the type of this type. If omitted, a semicolon is used in place of the
2000 comma and the type information, and the type is much like a generic
2001 pointer type---it has a known size but little else about it is
2015 This code generates a stab for a union tag and a stab for a union
2016 variable. Both use the @code{N_LSYM} stab type. If a union variable is
2017 scoped locally to the procedure in which it is defined, its stab is
2018 located immediately preceding the @code{N_LBRAC} for the procedure's block
2021 The stab for the union tag, however, is located preceding the code for
2022 the procedure in which it is defined. The stab type is @code{N_LSYM}. This
2023 would seem to imply that the union type is file scope, like the struct
2024 type @code{s_tag}. This is not true. The contents and position of the stab
2025 for @code{u_type} do not convey any infomation about its procedure local
2028 @c FIXME: phony line break. Can probably be fixed by using an example
2029 @c with fewer fields.
2032 .stabs "u_tag:T23=u4u_int:1,0,32;u_float:12,0,32;u_char:21,0,32;;",
2036 The symbol descriptor @samp{T}, following the @samp{name:} means that
2037 the stab describes an enumeration, structure, or union tag. The type
2038 descriptor @samp{u}, following the @samp{23=} of the type definition,
2039 narrows it down to a union type definition. Following the @samp{u} is
2040 the number of bytes in the union. After that is a list of union element
2041 descriptions. Their format is @var{name:type, bit offset into the
2042 union, number of bytes for the element;}.
2044 The stab for the union variable is:
2047 .stabs "an_u:23",128,0,0,-20 # @r{128 is N_LSYM}
2050 @samp{-20} specifies where the variable is stored (@pxref{Stack
2053 @node Function Types
2054 @section Function Types
2056 Various types can be defined for function variables. These types are
2057 not used in defining functions (@pxref{Procedures}); they are used for
2058 things like pointers to functions.
2060 The simple, traditional, type is type descriptor @samp{f} is followed by
2061 type information for the return type of the function, followed by a
2064 This does not deal with functions for which the number and types of the
2065 parameters are part of the type, as in Modula-2 or ANSI C. AIX provides
2066 extensions to specify these, using the @samp{f}, @samp{F}, @samp{p}, and
2067 @samp{R} type descriptors.
2069 First comes the type descriptor. If it is @samp{f} or @samp{F}, this
2070 type involves a function rather than a procedure, and the type
2071 information for the return type of the function follows, followed by a
2072 comma. Then comes the number of parameters to the function and a
2073 semicolon. Then, for each parameter, there is the name of the parameter
2074 followed by a colon (this is only present for type descriptors @samp{R}
2075 and @samp{F} which represent Pascal function or procedure parameters),
2076 type information for the parameter, a comma, 0 if passed by reference or
2077 1 if passed by value, and a semicolon. The type definition ends with a
2080 For example, this variable definition:
2087 generates the following code:
2090 .stabs "g_pf:G24=*25=f1",32,0,0,0
2091 .common _g_pf,4,"bss"
2094 The variable defines a new type, 24, which is a pointer to another new
2095 type, 25, which is a function returning @code{int}.
2098 @chapter Symbol Information in Symbol Tables
2100 This chapter describes the format of symbol table entries
2101 and how stab assembler directives map to them. It also describes the
2102 transformations that the assembler and linker make on data from stabs.
2105 * Symbol Table Format::
2106 * Transformations On Symbol Tables::
2109 @node Symbol Table Format
2110 @section Symbol Table Format
2112 Each time the assembler encounters a stab directive, it puts
2113 each field of the stab into a corresponding field in a symbol table
2114 entry of its output file. If the stab contains a string field, the
2115 symbol table entry for that stab points to a string table entry
2116 containing the string data from the stab. Assembler labels become
2117 relocatable addresses. Symbol table entries in a.out have the format:
2119 @c FIXME: should refer to external, not internal.
2121 struct internal_nlist @{
2122 unsigned long n_strx; /* index into string table of name */
2123 unsigned char n_type; /* type of symbol */
2124 unsigned char n_other; /* misc info (usually empty) */
2125 unsigned short n_desc; /* description field */
2126 bfd_vma n_value; /* value of symbol */
2130 If the stab has a string, the @code{n_strx} field holds the offset in
2131 bytes of the string within the string table. The string is terminated
2132 by a NUL character. If the stab lacks a string (for example, it was
2133 produced by a @code{.stabn} or @code{.stabd} directive), the
2134 @code{n_strx} field is zero.
2136 Symbol table entries with @code{n_type} field values greater than 0x1f
2137 originated as stabs generated by the compiler (with one random
2138 exception). The other entries were placed in the symbol table of the
2139 executable by the assembler or the linker.
2141 @node Transformations On Symbol Tables
2142 @section Transformations on Symbol Tables
2144 The linker concatenates object files and does fixups of externally
2147 You can see the transformations made on stab data by the assembler and
2148 linker by examining the symbol table after each pass of the build. To
2149 do this, use @samp{nm -ap}, which dumps the symbol table, including
2150 debugging information, unsorted. For stab entries the columns are:
2151 @var{value}, @var{other}, @var{desc}, @var{type}, @var{string}. For
2152 assembler and linker symbols, the columns are: @var{value}, @var{type},
2155 The low 5 bits of the stab type tell the linker how to relocate the
2156 value of the stab. Thus for stab types like @code{N_RSYM} and
2157 @code{N_LSYM}, where the value is an offset or a register number, the
2158 low 5 bits are @code{N_ABS}, which tells the linker not to relocate the
2161 Where the value of a stab contains an assembly language label,
2162 it is transformed by each build step. The assembler turns it into a
2163 relocatable address and the linker turns it into an absolute address.
2166 * Transformations On Static Variables::
2167 * Transformations On Global Variables::
2168 * Stab Section Transformations:: For some object file formats,
2169 things are a bit different.
2172 @node Transformations On Static Variables
2173 @subsection Transformations on Static Variables
2175 This source line defines a static variable at file scope:
2178 static int s_g_repeat
2182 The following stab describes the symbol:
2185 .stabs "s_g_repeat:S1",38,0,0,_s_g_repeat
2189 The assembler transforms the stab into this symbol table entry in the
2190 @file{.o} file. The location is expressed as a data segment offset.
2193 00000084 - 00 0000 STSYM s_g_repeat:S1
2197 In the symbol table entry from the executable, the linker has made the
2198 relocatable address absolute.
2201 0000e00c - 00 0000 STSYM s_g_repeat:S1
2204 @node Transformations On Global Variables
2205 @subsection Transformations on Global Variables
2207 Stabs for global variables do not contain location information. In
2208 this case, the debugger finds location information in the assembler or
2209 linker symbol table entry describing the variable. The source line:
2219 .stabs "g_foo:G2",32,0,0,0
2222 The variable is represented by two symbol table entries in the object
2223 file (see below). The first one originated as a stab. The second one
2224 is an external symbol. The upper case @samp{D} signifies that the
2225 @code{n_type} field of the symbol table contains 7, @code{N_DATA} with
2226 local linkage. The stab's value is zero since the value is not used for
2227 @code{N_GSYM} stabs. The value of the linker symbol is the relocatable
2228 address corresponding to the variable.
2231 00000000 - 00 0000 GSYM g_foo:G2
2236 These entries as transformed by the linker. The linker symbol table
2237 entry now holds an absolute address:
2240 00000000 - 00 0000 GSYM g_foo:G2
2245 @node Stab Section Transformations
2246 @subsection Transformations of Stabs in separate sections
2248 For object file formats using stabs in separate sections (@pxref{Stab
2249 Sections}), use @code{objdump --stabs} instead of @code{nm} to show the
2250 stabs in an object or executable file. @code{objdump} is a GNU utility;
2251 Sun does not provide any equivalent.
2253 The following example is for a stab whose value is an address is
2254 relative to the compilation unit (@pxref{ELF Linker Relocation}). For
2255 example, if the source line
2261 appears within a function, then the assembly language output from the
2267 .stabs "ld:V(0,3)",0x26,0,4,.L18-Ddata.data # @r{0x26 is N_STSYM}
2274 Because the value is formed by subtracting one symbol from another, the
2275 value is absolute, not relocatable, and so the object file contains
2278 Symnum n_type n_othr n_desc n_value n_strx String
2279 31 STSYM 0 4 00000004 680 ld:V(0,3)
2282 without any relocations, and the executable file also contains
2285 Symnum n_type n_othr n_desc n_value n_strx String
2286 31 STSYM 0 4 00000004 680 ld:V(0,3)
2290 @chapter GNU C++ Stabs
2293 * Class Names:: C++ class names are both tags and typedefs.
2294 * Nested Symbols:: C++ symbol names can be within other types.
2295 * Basic Cplusplus Types::
2298 * Methods:: Method definition
2300 * Method Modifiers::
2303 * Virtual Base Classes::
2307 Type descriptors added for C++ descriptions:
2311 method type (@code{##} if minimal debug)
2314 Member (class and variable) type. It is followed by type information
2315 for the offset basetype, a comma, and type information for the type of
2316 the field being pointed to. (FIXME: this is acknowledged to be
2317 gibberish. Can anyone say what really goes here?).
2319 Note that there is a conflict between this and type attributes
2320 (@pxref{String Field}); both use type descriptor @samp{@@}.
2321 Fortunately, the @samp{@@} type descriptor used in this C++ sense always
2322 will be followed by a digit, @samp{(}, or @samp{-}, and type attributes
2323 never start with those things.
2327 @section C++ Class Names
2329 In C++, a class name which is declared with @code{class}, @code{struct},
2330 or @code{union}, is not only a tag, as in C, but also a type name. Thus
2331 there should be stabs with both @samp{t} and @samp{T} symbol descriptors
2334 To save space, there is a special abbreviation for this case. If the
2335 @samp{T} symbol descriptor is followed by @samp{t}, then the stab
2336 defines both a type name and a tag.
2338 For example, the C++ code
2341 struct foo @{int x;@};
2344 can be represented as either
2347 .stabs "foo:T19=s4x:1,0,32;;",128,0,0,0 # @r{128 is N_LSYM}
2348 .stabs "foo:t19",128,0,0,0
2354 .stabs "foo:Tt19=s4x:1,0,32;;",128,0,0,0
2357 @node Nested Symbols
2358 @section Defining a Symbol Within Another Type
2360 In C++, a symbol (such as a type name) can be defined within another type.
2361 @c FIXME: Needs example.
2363 In stabs, this is sometimes represented by making the name of a symbol
2364 which contains @samp{::}. Such a pair of colons does not end the name
2365 of the symbol, the way a single colon would (@pxref{String Field}). I'm
2366 not sure how consistently used or well thought out this mechanism is.
2367 So that a pair of colons in this position always has this meaning,
2368 @samp{:} cannot be used as a symbol descriptor.
2370 For example, if the string for a stab is @samp{foo::bar::baz:t5=*6},
2371 then @code{foo::bar::baz} is the name of the symbol, @samp{t} is the
2372 symbol descriptor, and @samp{5=*6} is the type information.
2374 @node Basic Cplusplus Types
2375 @section Basic Types For C++
2377 << the examples that follow are based on a01.C >>
2380 C++ adds two more builtin types to the set defined for C. These are
2381 the unknown type and the vtable record type. The unknown type, type
2382 16, is defined in terms of itself like the void type.
2384 The vtable record type, type 17, is defined as a structure type and
2385 then as a structure tag. The structure has four fields: delta, index,
2386 pfn, and delta2. pfn is the function pointer.
2388 << In boilerplate $vtbl_ptr_type, what are the fields delta,
2389 index, and delta2 used for? >>
2391 This basic type is present in all C++ programs even if there are no
2392 virtual methods defined.
2395 .stabs "struct_name:sym_desc(type)type_def(17)=type_desc(struct)struct_bytes(8)
2396 elem_name(delta):type_ref(short int),bit_offset(0),field_bits(16);
2397 elem_name(index):type_ref(short int),bit_offset(16),field_bits(16);
2398 elem_name(pfn):type_def(18)=type_desc(ptr to)type_ref(void),
2399 bit_offset(32),field_bits(32);
2400 elem_name(delta2):type_def(short int);bit_offset(32),field_bits(16);;"
2405 .stabs "$vtbl_ptr_type:t17=s8
2406 delta:6,0,16;index:6,16,16;pfn:18=*15,32,32;delta2:6,32,16;;"
2411 .stabs "name:sym_dec(struct tag)type_ref($vtbl_ptr_type)",N_LSYM,NIL,NIL,NIL
2415 .stabs "$vtbl_ptr_type:T17",128,0,0,0
2418 @node Simple Classes
2419 @section Simple Class Definition
2421 The stabs describing C++ language features are an extension of the
2422 stabs describing C. Stabs representing C++ class types elaborate
2423 extensively on the stab format used to describe structure types in C.
2424 Stabs representing class type variables look just like stabs
2425 representing C language variables.
2427 Consider the following very simple class definition.
2433 int Ameth(int in, char other);
2437 The class @code{baseA} is represented by two stabs. The first stab describes
2438 the class as a structure type. The second stab describes a structure
2439 tag of the class type. Both stabs are of stab type @code{N_LSYM}. Since the
2440 stab is not located between an @code{N_FUN} and an @code{N_LBRAC} stab this indicates
2441 that the class is defined at file scope. If it were, then the @code{N_LSYM}
2442 would signify a local variable.
2444 A stab describing a C++ class type is similar in format to a stab
2445 describing a C struct, with each class member shown as a field in the
2446 structure. The part of the struct format describing fields is
2447 expanded to include extra information relevent to C++ class members.
2448 In addition, if the class has multiple base classes or virtual
2449 functions the struct format outside of the field parts is also
2452 In this simple example the field part of the C++ class stab
2453 representing member data looks just like the field part of a C struct
2454 stab. The section on protections describes how its format is
2455 sometimes extended for member data.
2457 The field part of a C++ class stab representing a member function
2458 differs substantially from the field part of a C struct stab. It
2459 still begins with @samp{name:} but then goes on to define a new type number
2460 for the member function, describe its return type, its argument types,
2461 its protection level, any qualifiers applied to the method definition,
2462 and whether the method is virtual or not. If the method is virtual
2463 then the method description goes on to give the vtable index of the
2464 method, and the type number of the first base class defining the
2467 When the field name is a method name it is followed by two colons rather
2468 than one. This is followed by a new type definition for the method.
2469 This is a number followed by an equal sign and the type descriptor
2470 @samp{#}, indicating a method type, and a second @samp{#}, indicating
2471 that this is the @dfn{minimal} type of method definition used by GCC2,
2472 not larger method definitions used by earlier versions of GCC. This is
2473 followed by a type reference showing the return type of the method and a
2476 The format of an overloaded operator method name differs from that of
2477 other methods. It is @samp{op$::@var{operator-name}.} where
2478 @var{operator-name} is the operator name such as @samp{+} or @samp{+=}.
2479 The name ends with a period, and any characters except the period can
2480 occur in the @var{operator-name} string.
2482 The next part of the method description represents the arguments to the
2483 method, preceeded by a colon and ending with a semi-colon. The types of
2484 the arguments are expressed in the same way argument types are expressed
2485 in C++ name mangling. In this example an @code{int} and a @code{char}
2488 This is followed by a number, a letter, and an asterisk or period,
2489 followed by another semicolon. The number indicates the protections
2490 that apply to the member function. Here the 2 means public. The
2491 letter encodes any qualifier applied to the method definition. In
2492 this case, @samp{A} means that it is a normal function definition. The dot
2493 shows that the method is not virtual. The sections that follow
2494 elaborate further on these fields and describe the additional
2495 information present for virtual methods.
2499 .stabs "class_name:sym_desc(type)type_def(20)=type_desc(struct)struct_bytes(4)
2500 field_name(Adat):type(int),bit_offset(0),field_bits(32);
2502 method_name(Ameth)::type_def(21)=type_desc(method)return_type(int);
2503 :arg_types(int char);
2504 protection(public)qualifier(normal)virtual(no);;"
2509 .stabs "baseA:t20=s4Adat:1,0,32;Ameth::21=##1;:ic;2A.;;",128,0,0,0
2511 .stabs "class_name:sym_desc(struct tag)",N_LSYM,NIL,NIL,NIL
2513 .stabs "baseA:T20",128,0,0,0
2516 @node Class Instance
2517 @section Class Instance
2519 As shown above, describing even a simple C++ class definition is
2520 accomplished by massively extending the stab format used in C to
2521 describe structure types. However, once the class is defined, C stabs
2522 with no modifications can be used to describe class instances. The
2532 yields the following stab describing the class instance. It looks no
2533 different from a standard C stab describing a local variable.
2536 .stabs "name:type_ref(baseA)", N_LSYM, NIL, NIL, frame_ptr_offset
2540 .stabs "AbaseA:20",128,0,0,-20
2544 @section Method Definition
2546 The class definition shown above declares Ameth. The C++ source below
2551 baseA::Ameth(int in, char other)
2558 This method definition yields three stabs following the code of the
2559 method. One stab describes the method itself and following two describe
2560 its parameters. Although there is only one formal argument all methods
2561 have an implicit argument which is the @code{this} pointer. The @code{this}
2562 pointer is a pointer to the object on which the method was called. Note
2563 that the method name is mangled to encode the class name and argument
2564 types. Name mangling is described in the @sc{arm} (@cite{The Annotated
2565 C++ Reference Manual}, by Ellis and Stroustrup, @sc{isbn}
2566 0-201-51459-1); @file{gpcompare.texi} in Cygnus GCC distributions
2567 describes the differences between GNU mangling and @sc{arm}
2569 @c FIXME: Use @xref, especially if this is generally installed in the
2571 @c FIXME: This information should be in a net release, either of GCC or
2572 @c GDB. But gpcompare.texi doesn't seem to be in the FSF GCC.
2575 .stabs "name:symbol_desriptor(global function)return_type(int)",
2576 N_FUN, NIL, NIL, code_addr_of_method_start
2578 .stabs "Ameth__5baseAic:F1",36,0,0,_Ameth__5baseAic
2581 Here is the stab for the @code{this} pointer implicit argument. The
2582 name of the @code{this} pointer is always @code{this}. Type 19, the
2583 @code{this} pointer is defined as a pointer to type 20, @code{baseA},
2584 but a stab defining @code{baseA} has not yet been emited. Since the
2585 compiler knows it will be emited shortly, here it just outputs a cross
2586 reference to the undefined symbol, by prefixing the symbol name with
2590 .stabs "name:sym_desc(register param)type_def(19)=
2591 type_desc(ptr to)type_ref(baseA)=
2592 type_desc(cross-reference to)baseA:",N_RSYM,NIL,NIL,register_number
2594 .stabs "this:P19=*20=xsbaseA:",64,0,0,8
2597 The stab for the explicit integer argument looks just like a parameter
2598 to a C function. The last field of the stab is the offset from the
2599 argument pointer, which in most systems is the same as the frame
2603 .stabs "name:sym_desc(value parameter)type_ref(int)",
2604 N_PSYM,NIL,NIL,offset_from_arg_ptr
2606 .stabs "in:p1",160,0,0,72
2609 << The examples that follow are based on A1.C >>
2612 @section Protections
2614 In the simple class definition shown above all member data and
2615 functions were publicly accessable. The example that follows
2616 contrasts public, protected and privately accessable fields and shows
2617 how these protections are encoded in C++ stabs.
2619 If the character following the @samp{@var{field-name}:} part of the
2620 string is @samp{/}, then the next character is the visibility. @samp{0}
2621 means private, @samp{1} means protected, and @samp{2} means public.
2622 Debuggers should ignore visibility characters they do not recognize, and
2623 assume a reasonable default (such as public) (GDB 4.11 does not, but
2624 this should be fixed in the next GDB release). If no visibility is
2625 specified the field is public. The visibility @samp{9} means that the
2626 field has been optimized out and is public (there is no way to specify
2627 an optimized out field with a private or protected visibility).
2628 Visibility @samp{9} is not supported by GDB 4.11; this should be fixed
2629 in the next GDB release.
2631 The following C++ source:
2645 generates the following stab:
2649 .stabs "vis:T19=s12priv:/01,0,32;prot:/12,32,8;pub:12,64,32;;",128,0,0,0
2652 @samp{vis:T19=s12} indicates that type number 19 is a 12 byte structure
2653 named @code{vis} The @code{priv} field has public visibility
2654 (@samp{/0}), type int (@samp{1}), and offset and size @samp{,0,32;}.
2655 The @code{prot} field has protected visibility (@samp{/1}), type char
2656 (@samp{2}) and offset and size @samp{,32,8;}. The @code{pub} field has
2657 type float (@samp{12}), and offset and size @samp{,64,32;}.
2659 Protections for member functions are signified by one digit embeded in
2660 the field part of the stab describing the method. The digit is 0 if
2661 private, 1 if protected and 2 if public. Consider the C++ class
2665 class all_methods @{
2667 int priv_meth(int in)@{return in;@};
2669 char protMeth(char in)@{return in;@};
2671 float pubMeth(float in)@{return in;@};
2675 It generates the following stab. The digit in question is to the left
2676 of an @samp{A} in each case. Notice also that in this case two symbol
2677 descriptors apply to the class name struct tag and struct type.
2680 .stabs "class_name:sym_desc(struct tag&type)type_def(21)=
2681 sym_desc(struct)struct_bytes(1)
2682 meth_name::type_def(22)=sym_desc(method)returning(int);
2683 :args(int);protection(private)modifier(normal)virtual(no);
2684 meth_name::type_def(23)=sym_desc(method)returning(char);
2685 :args(char);protection(protected)modifier(normal)virual(no);
2686 meth_name::type_def(24)=sym_desc(method)returning(float);
2687 :args(float);protection(public)modifier(normal)virtual(no);;",
2692 .stabs "all_methods:Tt21=s1priv_meth::22=##1;:i;0A.;protMeth::23=##2;:c;1A.;
2693 pubMeth::24=##12;:f;2A.;;",128,0,0,0
2696 @node Method Modifiers
2697 @section Method Modifiers (@code{const}, @code{volatile}, @code{const volatile})
2701 In the class example described above all the methods have the normal
2702 modifier. This method modifier information is located just after the
2703 protection information for the method. This field has four possible
2704 character values. Normal methods use @samp{A}, const methods use
2705 @samp{B}, volatile methods use @samp{C}, and const volatile methods use
2706 @samp{D}. Consider the class definition below:
2711 int ConstMeth (int arg) const @{ return arg; @};
2712 char VolatileMeth (char arg) volatile @{ return arg; @};
2713 float ConstVolMeth (float arg) const volatile @{return arg; @};
2717 This class is described by the following stab:
2720 .stabs "class(A):sym_desc(struct)type_def(20)=type_desc(struct)struct_bytes(1)
2721 meth_name(ConstMeth)::type_def(21)sym_desc(method)
2722 returning(int);:arg(int);protection(public)modifier(const)virtual(no);
2723 meth_name(VolatileMeth)::type_def(22)=sym_desc(method)
2724 returning(char);:arg(char);protection(public)modifier(volatile)virt(no)
2725 meth_name(ConstVolMeth)::type_def(23)=sym_desc(method)
2726 returning(float);:arg(float);protection(public)modifer(const volatile)
2727 virtual(no);;", @dots{}
2731 .stabs "A:T20=s1ConstMeth::21=##1;:i;2B.;VolatileMeth::22=##2;:c;2C.;
2732 ConstVolMeth::23=##12;:f;2D.;;",128,0,0,0
2735 @node Virtual Methods
2736 @section Virtual Methods
2738 << The following examples are based on a4.C >>
2740 The presence of virtual methods in a class definition adds additional
2741 data to the class description. The extra data is appended to the
2742 description of the virtual method and to the end of the class
2743 description. Consider the class definition below:
2749 virtual int A_virt (int arg) @{ return arg; @};
2753 This results in the stab below describing class A. It defines a new
2754 type (20) which is an 8 byte structure. The first field of the class
2755 struct is @samp{Adat}, an integer, starting at structure offset 0 and
2758 The second field in the class struct is not explicitly defined by the
2759 C++ class definition but is implied by the fact that the class
2760 contains a virtual method. This field is the vtable pointer. The
2761 name of the vtable pointer field starts with @samp{$vf} and continues with a
2762 type reference to the class it is part of. In this example the type
2763 reference for class A is 20 so the name of its vtable pointer field is
2764 @samp{$vf20}, followed by the usual colon.
2766 Next there is a type definition for the vtable pointer type (21).
2767 This is in turn defined as a pointer to another new type (22).
2769 Type 22 is the vtable itself, which is defined as an array, indexed by
2770 a range of integers between 0 and 1, and whose elements are of type
2771 17. Type 17 was the vtable record type defined by the boilerplate C++
2772 type definitions, as shown earlier.
2774 The bit offset of the vtable pointer field is 32. The number of bits
2775 in the field are not specified when the field is a vtable pointer.
2777 Next is the method definition for the virtual member function @code{A_virt}.
2778 Its description starts out using the same format as the non-virtual
2779 member functions described above, except instead of a dot after the
2780 @samp{A} there is an asterisk, indicating that the function is virtual.
2781 Since is is virtual some addition information is appended to the end
2782 of the method description.
2784 The first number represents the vtable index of the method. This is a
2785 32 bit unsigned number with the high bit set, followed by a
2788 The second number is a type reference to the first base class in the
2789 inheritence hierarchy defining the virtual member function. In this
2790 case the class stab describes a base class so the virtual function is
2791 not overriding any other definition of the method. Therefore the
2792 reference is to the type number of the class that the stab is
2795 This is followed by three semi-colons. One marks the end of the
2796 current sub-section, one marks the end of the method field, and the
2797 third marks the end of the struct definition.
2799 For classes containing virtual functions the very last section of the
2800 string part of the stab holds a type reference to the first base
2801 class. This is preceeded by @samp{~%} and followed by a final semi-colon.
2804 .stabs "class_name(A):type_def(20)=sym_desc(struct)struct_bytes(8)
2805 field_name(Adat):type_ref(int),bit_offset(0),field_bits(32);
2806 field_name(A virt func ptr):type_def(21)=type_desc(ptr to)type_def(22)=
2807 sym_desc(array)index_type_ref(range of int from 0 to 1);
2808 elem_type_ref(vtbl elem type),
2810 meth_name(A_virt)::typedef(23)=sym_desc(method)returning(int);
2811 :arg_type(int),protection(public)normal(yes)virtual(yes)
2812 vtable_index(1);class_first_defining(A);;;~%first_base(A);",
2816 @c FIXME: bogus line break.
2818 .stabs "A:t20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
2819 A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
2823 @section Inheritence
2825 Stabs describing C++ derived classes include additional sections that
2826 describe the inheritence hierarchy of the class. A derived class stab
2827 also encodes the number of base classes. For each base class it tells
2828 if the base class is virtual or not, and if the inheritence is private
2829 or public. It also gives the offset into the object of the portion of
2830 the object corresponding to each base class.
2832 This additional information is embeded in the class stab following the
2833 number of bytes in the struct. First the number of base classes
2834 appears bracketed by an exclamation point and a comma.
2836 Then for each base type there repeats a series: a virtual character, a
2837 visibilty character, a number, a comma, another number, and a
2840 The virtual character is @samp{1} if the base class is virtual and
2841 @samp{0} if not. The visibility character is @samp{2} if the derivation
2842 is public, @samp{1} if it is protected, and @samp{0} if it is private.
2843 Debuggers should ignore virtual or visibility characters they do not
2844 recognize, and assume a reasonable default (such as public and
2845 non-virtual) (GDB 4.11 does not, but this should be fixed in the next
2848 The number following the virtual and visibility characters is the offset
2849 from the start of the object to the part of the object pertaining to the
2852 After the comma, the second number is a type_descriptor for the base
2853 type. Finally a semi-colon ends the series, which repeats for each
2856 The source below defines three base classes @code{A}, @code{B}, and
2857 @code{C} and the derived class @code{D}.
2864 virtual int A_virt (int arg) @{ return arg; @};
2870 virtual int B_virt (int arg) @{return arg; @};
2876 virtual int C_virt (int arg) @{return arg; @};
2879 class D : A, virtual B, public C @{
2882 virtual int A_virt (int arg ) @{ return arg+1; @};
2883 virtual int B_virt (int arg) @{ return arg+2; @};
2884 virtual int C_virt (int arg) @{ return arg+3; @};
2885 virtual int D_virt (int arg) @{ return arg; @};
2889 Class stabs similar to the ones described earlier are generated for
2892 @c FIXME!!! the linebreaks in the following example probably make the
2893 @c examples literally unusable, but I don't know any other way to get
2894 @c them on the page.
2895 @c One solution would be to put some of the type definitions into
2896 @c separate stabs, even if that's not exactly what the compiler actually
2899 .stabs "A:T20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
2900 A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
2902 .stabs "B:Tt25=s8Bdat:1,0,32;$vf25:21,32;B_virt::26=##1;
2903 :i;2A*-2147483647;25;;;~%25;",128,0,0,0
2905 .stabs "C:Tt28=s8Cdat:1,0,32;$vf28:21,32;C_virt::29=##1;
2906 :i;2A*-2147483647;28;;;~%28;",128,0,0,0
2909 In the stab describing derived class @code{D} below, the information about
2910 the derivation of this class is encoded as follows.
2913 .stabs "derived_class_name:symbol_descriptors(struct tag&type)=
2914 type_descriptor(struct)struct_bytes(32)!num_bases(3),
2915 base_virtual(no)inheritence_public(no)base_offset(0),
2916 base_class_type_ref(A);
2917 base_virtual(yes)inheritence_public(no)base_offset(NIL),
2918 base_class_type_ref(B);
2919 base_virtual(no)inheritence_public(yes)base_offset(64),
2920 base_class_type_ref(C); @dots{}
2923 @c FIXME! fake linebreaks.
2925 .stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:
2926 1,160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt:
2927 :32:i;2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;
2928 28;;D_virt::32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
2931 @node Virtual Base Classes
2932 @section Virtual Base Classes
2934 A derived class object consists of a concatination in memory of the data
2935 areas defined by each base class, starting with the leftmost and ending
2936 with the rightmost in the list of base classes. The exception to this
2937 rule is for virtual inheritence. In the example above, class @code{D}
2938 inherits virtually from base class @code{B}. This means that an
2939 instance of a @code{D} object will not contain its own @code{B} part but
2940 merely a pointer to a @code{B} part, known as a virtual base pointer.
2942 In a derived class stab, the base offset part of the derivation
2943 information, described above, shows how the base class parts are
2944 ordered. The base offset for a virtual base class is always given as 0.
2945 Notice that the base offset for @code{B} is given as 0 even though
2946 @code{B} is not the first base class. The first base class @code{A}
2949 The field information part of the stab for class @code{D} describes the field
2950 which is the pointer to the virtual base class @code{B}. The vbase pointer
2951 name is @samp{$vb} followed by a type reference to the virtual base class.
2952 Since the type id for @code{B} in this example is 25, the vbase pointer name
2955 @c FIXME!! fake linebreaks below
2957 .stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:1,
2958 160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt::32:i;
2959 2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;28;;D_virt:
2960 :32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
2963 Following the name and a semicolon is a type reference describing the
2964 type of the virtual base class pointer, in this case 24. Type 24 was
2965 defined earlier as the type of the @code{B} class @code{this} pointer. The
2966 @code{this} pointer for a class is a pointer to the class type.
2969 .stabs "this:P24=*25=xsB:",64,0,0,8
2972 Finally the field offset part of the vbase pointer field description
2973 shows that the vbase pointer is the first field in the @code{D} object,
2974 before any data fields defined by the class. The layout of a @code{D}
2975 class object is a follows, @code{Adat} at 0, the vtable pointer for
2976 @code{A} at 32, @code{Cdat} at 64, the vtable pointer for C at 96, the
2977 virtual base pointer for @code{B} at 128, and @code{Ddat} at 160.
2980 @node Static Members
2981 @section Static Members
2983 The data area for a class is a concatenation of the space used by the
2984 data members of the class. If the class has virtual methods, a vtable
2985 pointer follows the class data. The field offset part of each field
2986 description in the class stab shows this ordering.
2988 << How is this reflected in stabs? See Cygnus bug #677 for some info. >>
2991 @appendix Table of Stab Types
2993 The following are all the possible values for the stab type field, for
2994 @code{a.out} files, in numeric order. This does not apply to XCOFF, but
2995 it does apply to stabs in sections (@pxref{Stab Sections}). Stabs in
2996 ECOFF use these values but add 0x8f300 to distinguish them from non-stab
2999 The symbolic names are defined in the file @file{include/aout/stabs.def}.
3002 * Non-Stab Symbol Types:: Types from 0 to 0x1f
3003 * Stab Symbol Types:: Types from 0x20 to 0xff
3006 @node Non-Stab Symbol Types
3007 @appendixsec Non-Stab Symbol Types
3009 The following types are used by the linker and assembler, not by stab
3010 directives. Since this document does not attempt to describe aspects of
3011 object file format other than the debugging format, no details are
3014 @c Try to get most of these to fit on a single line.
3024 File scope absolute symbol
3026 @item 0x3 N_ABS | N_EXT
3027 External absolute symbol
3030 File scope text symbol
3032 @item 0x5 N_TEXT | N_EXT
3033 External text symbol
3036 File scope data symbol
3038 @item 0x7 N_DATA | N_EXT
3039 External data symbol
3042 File scope BSS symbol
3044 @item 0x9 N_BSS | N_EXT
3048 Same as @code{N_FN}, for Sequent compilers
3051 Symbol is indirected to another symbol
3054 Common---visible after shared library dynamic link
3057 @itemx 0x15 N_SETA | N_EXT
3058 Absolute set element
3061 @itemx 0x17 N_SETT | N_EXT
3062 Text segment set element
3065 @itemx 0x19 N_SETD | N_EXT
3066 Data segment set element
3069 @itemx 0x1b N_SETB | N_EXT
3070 BSS segment set element
3073 @itemx 0x1d N_SETV | N_EXT
3074 Pointer to set vector
3076 @item 0x1e N_WARNING
3077 Print a warning message during linking
3080 File name of a @file{.o} file
3083 @node Stab Symbol Types
3084 @appendixsec Stab Symbol Types
3086 The following symbol types indicate that this is a stab. This is the
3087 full list of stab numbers, including stab types that are used in
3088 languages other than C.
3092 Global symbol; see @ref{Global Variables}.
3095 Function name (for BSD Fortran); see @ref{Procedures}.
3098 Function name (@pxref{Procedures}) or text segment variable
3102 Data segment file-scope variable; see @ref{Statics}.
3105 BSS segment file-scope variable; see @ref{Statics}.
3108 Name of main routine; see @ref{Main Program}.
3111 Variable in @code{.rodata} section; see @ref{Statics}.
3114 Global symbol (for Pascal); see @ref{N_PC}.
3117 Number of symbols (according to Ultrix V4.0); see @ref{N_NSYMS}.
3120 No DST map; see @ref{N_NOMAP}.
3122 @c FIXME: describe this solaris feature in the body of the text (see
3123 @c comments in include/aout/stab.def).
3125 Object file (Solaris2).
3127 @c See include/aout/stab.def for (a little) more info.
3129 Debugger options (Solaris2).
3132 Register variable; see @ref{Register Variables}.
3135 Modula-2 compilation unit; see @ref{N_M2C}.
3138 Line number in text segment; see @ref{Line Numbers}.
3141 Line number in data segment; see @ref{Line Numbers}.
3144 Line number in bss segment; see @ref{Line Numbers}.
3147 Sun source code browser, path to @file{.cb} file; see @ref{N_BROWS}.
3150 GNU Modula2 definition module dependency; see @ref{N_DEFD}.
3153 Function start/body/end line numbers (Solaris2).
3156 GNU C++ exception variable; see @ref{N_EHDECL}.
3159 Modula2 info "for imc" (according to Ultrix V4.0); see @ref{N_MOD2}.
3162 GNU C++ @code{catch} clause; see @ref{N_CATCH}.
3165 Structure of union element; see @ref{N_SSYM}.
3168 Last stab for module (Solaris2).
3171 Path and name of source file; see @ref{Source Files}.
3174 Stack variable (@pxref{Stack Variables}) or type (@pxref{Typedefs}).
3177 Beginning of an include file (Sun only); see @ref{Include Files}.
3180 Name of include file; see @ref{Include Files}.
3183 Parameter variable; see @ref{Parameters}.
3186 End of an include file; see @ref{Include Files}.
3189 Alternate entry point; see @ref{N_ENTRY}.
3192 Beginning of a lexical block; see @ref{Block Structure}.
3195 Place holder for a deleted include file; see @ref{Include Files}.
3198 Modula2 scope information (Sun linker); see @ref{N_SCOPE}.
3201 End of a lexical block; see @ref{Block Structure}.
3204 Begin named common block; see @ref{Common Blocks}.
3207 End named common block; see @ref{Common Blocks}.
3210 Member of a common block; see @ref{Common Blocks}.
3212 @c FIXME: How does this really work? Move it to main body of document.
3214 Pascal @code{with} statement: type,,0,0,offset (Solaris2).
3217 Gould non-base registers; see @ref{Gould}.
3220 Gould non-base registers; see @ref{Gould}.
3223 Gould non-base registers; see @ref{Gould}.
3226 Gould non-base registers; see @ref{Gould}.
3229 Gould non-base registers; see @ref{Gould}.
3232 @c Restore the default table indent
3237 @node Symbol Descriptors
3238 @appendix Table of Symbol Descriptors
3240 The symbol descriptor is the character which follows the colon in many
3241 stabs, and which tells what kind of stab it is. @xref{String Field},
3242 for more information about their use.
3244 @c Please keep this alphabetical
3246 @c In TeX, this looks great, digit is in italics. But makeinfo insists
3247 @c on putting it in `', not realizing that @var should override @code.
3248 @c I don't know of any way to make makeinfo do the right thing. Seems
3249 @c like a makeinfo bug to me.
3253 Variable on the stack; see @ref{Stack Variables}.
3256 C++ nested symbol; see @xref{Nested Symbols}
3259 Parameter passed by reference in register; see @ref{Reference Parameters}.
3262 Based variable; see @ref{Based Variables}.
3265 Constant; see @ref{Constants}.
3268 Conformant array bound (Pascal, maybe other languages); @ref{Conformant
3269 Arrays}. Name of a caught exception (GNU C++). These can be
3270 distinguished because the latter uses @code{N_CATCH} and the former uses
3271 another symbol type.
3274 Floating point register variable; see @ref{Register Variables}.
3277 Parameter in floating point register; see @ref{Register Parameters}.
3280 File scope function; see @ref{Procedures}.
3283 Global function; see @ref{Procedures}.
3286 Global variable; see @ref{Global Variables}.
3289 @xref{Register Parameters}.
3292 Internal (nested) procedure; see @ref{Nested Procedures}.
3295 Internal (nested) function; see @ref{Nested Procedures}.
3298 Label name (documented by AIX, no further information known).
3301 Module; see @ref{Procedures}.
3304 Argument list parameter; see @ref{Parameters}.
3310 Fortran Function parameter; see @ref{Parameters}.
3313 Unfortunately, three separate meanings have been independently invented
3314 for this symbol descriptor. At least the GNU and Sun uses can be
3315 distinguished by the symbol type. Global Procedure (AIX) (symbol type
3316 used unknown); see @ref{Procedures}. Register parameter (GNU) (symbol
3317 type @code{N_PSYM}); see @ref{Parameters}. Prototype of function
3318 referenced by this file (Sun @code{acc}) (symbol type @code{N_FUN}).
3321 Static Procedure; see @ref{Procedures}.
3324 Register parameter; see @ref{Register Parameters}.
3327 Register variable; see @ref{Register Variables}.
3330 File scope variable; see @ref{Statics}.
3333 Type name; see @ref{Typedefs}.
3336 Enumeration, structure, or union tag; see @ref{Typedefs}.
3339 Parameter passed by reference; see @ref{Reference Parameters}.
3342 Procedure scope static variable; see @ref{Statics}.
3345 Conformant array; see @ref{Conformant Arrays}.
3348 Function return variable; see @ref{Parameters}.
3351 @node Type Descriptors
3352 @appendix Table of Type Descriptors
3354 The type descriptor is the character which follows the type number and
3355 an equals sign. It specifies what kind of type is being defined.
3356 @xref{String Field}, for more information about their use.
3361 Type reference; see @ref{String Field}.
3364 Reference to builtin type; see @ref{Negative Type Numbers}.
3367 Method (C++); see @ref{Cplusplus}.
3370 Pointer; see @ref{Miscellaneous Types}.
3376 Type Attributes (AIX); see @ref{String Field}. Member (class and variable)
3377 type (GNU C++); see @ref{Cplusplus}.
3380 Array; see @ref{Arrays}.
3383 Open array; see @ref{Arrays}.
3386 Pascal space type (AIX); see @ref{Miscellaneous Types}. Builtin integer
3387 type (Sun); see @ref{Builtin Type Descriptors}.
3390 Volatile-qualified type; see @ref{Miscellaneous Types}.
3393 Complex builtin type; see @ref{Builtin Type Descriptors}.
3396 COBOL Picture type. See AIX documentation for details.
3399 File type; see @ref{Miscellaneous Types}.
3402 N-dimensional dynamic array; see @ref{Arrays}.
3405 Enumeration type; see @ref{Enumerations}.
3408 N-dimensional subarray; see @ref{Arrays}.
3411 Function type; see @ref{Function Types}.
3414 Pascal function parameter; see @ref{Function Types}
3417 Builtin floating point type; see @ref{Builtin Type Descriptors}.
3420 COBOL Group. See AIX documentation for details.
3423 Imported type; see @ref{Cross-References}.
3426 Const-qualified type; see @ref{Miscellaneous Types}.
3429 COBOL File Descriptor. See AIX documentation for details.
3432 Multiple instance type; see @ref{Miscellaneous Types}.
3435 String type; see @ref{Strings}.
3438 Stringptr; see @ref{Strings}.
3441 Opaque type; see @ref{Typedefs}.
3444 Procedure; see @ref{Function Types}.
3447 Packed array; see @ref{Arrays}.
3450 Range type; see @ref{Subranges}.
3453 Builtin floating type; see @ref{Builtin Type Descriptors} (Sun). Pascal
3454 subroutine parameter; see @ref{Function Types} (AIX). Detecting this
3455 conflict is possible with careful parsing (hint: a Pascal subroutine
3456 parameter type will always contain a comma, and a builtin type
3457 descriptor never will).
3460 Structure type; see @ref{Structures}.
3463 Set type; see @ref{Miscellaneous Types}.
3466 Union; see @ref{Unions}.
3469 Variant record. This is a Pascal and Modula-2 feature which is like a
3470 union within a struct in C. See AIX documentation for details.
3473 Wide character; see @ref{Builtin Type Descriptors}.
3476 Cross-reference; see @ref{Cross-References}.
3479 gstring; see @ref{Strings}.
3482 @node Expanded Reference
3483 @appendix Expanded Reference by Stab Type
3485 @c FIXME: This appendix should go away; see N_PSYM or N_SO for an example.
3487 For a full list of stab types, and cross-references to where they are
3488 described, see @ref{Stab Types}. This appendix just duplicates certain
3489 information from the main body of this document; eventually the
3490 information will all be in one place.
3494 The first line is the symbol type (see @file{include/aout/stab.def}).
3496 The second line describes the language constructs the symbol type
3499 The third line is the stab format with the significant stab fields
3500 named and the rest NIL.
3502 Subsequent lines expand upon the meaning and possible values for each
3503 significant stab field. @samp{#} stands in for the type descriptor.
3505 Finally, any further information.
3508 * N_PC:: Pascal global symbol
3509 * N_NSYMS:: Number of symbols
3510 * N_NOMAP:: No DST map
3511 * N_M2C:: Modula-2 compilation unit
3512 * N_BROWS:: Path to .cb file for Sun source code browser
3513 * N_DEFD:: GNU Modula2 definition module dependency
3514 * N_EHDECL:: GNU C++ exception variable
3515 * N_MOD2:: Modula2 information "for imc"
3516 * N_CATCH:: GNU C++ "catch" clause
3517 * N_SSYM:: Structure or union element
3518 * N_ENTRY:: Alternate entry point
3519 * N_SCOPE:: Modula2 scope information (Sun only)
3520 * Gould:: non-base register symbols used on Gould systems
3521 * N_LENG:: Length of preceding entry
3527 @deffn @code{.stabs} N_PC
3529 Global symbol (for Pascal).
3532 "name" -> "symbol_name" <<?>>
3533 value -> supposedly the line number (stab.def is skeptical)
3537 @file{stabdump.c} says:
3539 global pascal symbol: name,,0,subtype,line
3547 @deffn @code{.stabn} N_NSYMS
3549 Number of symbols (according to Ultrix V4.0).
3552 0, files,,funcs,lines (stab.def)
3559 @deffn @code{.stabs} N_NOMAP
3561 No DST map for symbol (according to Ultrix V4.0). I think this means a
3562 variable has been optimized out.
3565 name, ,0,type,ignored (stab.def)
3572 @deffn @code{.stabs} N_M2C
3574 Modula-2 compilation unit.
3577 "string" -> "unit_name,unit_time_stamp[,code_time_stamp]"
3579 value -> 0 (main unit)
3583 See @cite{Dbx and Dbxtool Interfaces}, 2nd edition, by Sun, 1988, for
3591 @deffn @code{.stabs} N_BROWS
3593 Sun source code browser, path to @file{.cb} file
3596 "path to associated @file{.cb} file"
3598 Note: N_BROWS has the same value as N_BSLINE.
3604 @deffn @code{.stabn} N_DEFD
3606 GNU Modula2 definition module dependency.
3608 GNU Modula-2 definition module dependency. The value is the
3609 modification time of the definition file. The other field is non-zero
3610 if it is imported with the GNU M2 keyword @code{%INITIALIZE}. Perhaps
3611 @code{N_M2C} can be used if there are enough empty fields?
3617 @deffn @code{.stabs} N_EHDECL
3619 GNU C++ exception variable <<?>>.
3621 "@var{string} is variable name"
3623 Note: conflicts with @code{N_MOD2}.
3629 @deffn @code{.stab?} N_MOD2
3631 Modula2 info "for imc" (according to Ultrix V4.0)
3633 Note: conflicts with @code{N_EHDECL} <<?>>
3639 @deffn @code{.stabn} N_CATCH
3641 GNU C++ @code{catch} clause
3643 GNU C++ @code{catch} clause. The value is its address. The desc field
3644 is nonzero if this entry is immediately followed by a @code{CAUGHT} stab
3645 saying what exception was caught. Multiple @code{CAUGHT} stabs means
3646 that multiple exceptions can be caught here. If desc is 0, it means all
3647 exceptions are caught here.
3653 @deffn @code{.stabn} N_SSYM
3655 Structure or union element.
3657 The value is the offset in the structure.
3659 <<?looking at structs and unions in C I didn't see these>>
3665 @deffn @code{.stabn} N_ENTRY
3667 Alternate entry point.
3668 The value is its address.
3675 @deffn @code{.stab?} N_SCOPE
3677 Modula2 scope information (Sun linker)
3682 @section Non-base registers on Gould systems
3684 @deffn @code{.stab?} N_NBTEXT
3685 @deffnx @code{.stab?} N_NBDATA
3686 @deffnx @code{.stab?} N_NBBSS
3687 @deffnx @code{.stab?} N_NBSTS
3688 @deffnx @code{.stab?} N_NBLCS
3694 These are used on Gould systems for non-base registers syms.
3696 However, the following values are not the values used by Gould; they are
3697 the values which GNU has been documenting for these values for a long
3698 time, without actually checking what Gould uses. I include these values
3699 only because perhaps some someone actually did something with the GNU
3700 information (I hope not, why GNU knowingly assigned wrong values to
3701 these in the header file is a complete mystery to me).
3704 240 0xf0 N_NBTEXT ??
3705 242 0xf2 N_NBDATA ??
3715 @deffn @code{.stabn} N_LENG
3717 Second symbol entry containing a length-value for the preceding entry.
3718 The value is the length.
3722 @appendix Questions and Anomalies
3726 @c I think this is changed in GCC 2.4.5 to put the line number there.
3727 For GNU C stabs defining local and global variables (@code{N_LSYM} and
3728 @code{N_GSYM}), the desc field is supposed to contain the source
3729 line number on which the variable is defined. In reality the desc
3730 field is always 0. (This behavior is defined in @file{dbxout.c} and
3731 putting a line number in desc is controlled by @samp{#ifdef
3732 WINNING_GDB}, which defaults to false). GDB supposedly uses this
3733 information if you say @samp{list @var{var}}. In reality, @var{var} can
3734 be a variable defined in the program and GDB says @samp{function
3735 @var{var} not defined}.
3738 In GNU C stabs, there seems to be no way to differentiate tag types:
3739 structures, unions, and enums (symbol descriptor @samp{T}) and typedefs
3740 (symbol descriptor @samp{t}) defined at file scope from types defined locally
3741 to a procedure or other more local scope. They all use the @code{N_LSYM}
3742 stab type. Types defined at procedure scope are emited after the
3743 @code{N_RBRAC} of the preceding function and before the code of the
3744 procedure in which they are defined. This is exactly the same as
3745 types defined in the source file between the two procedure bodies.
3746 GDB overcompensates by placing all types in block #1, the block for
3747 symbols of file scope. This is true for default, @samp{-ansi} and
3748 @samp{-traditional} compiler options. (Bugs gcc/1063, gdb/1066.)
3751 What ends the procedure scope? Is it the proc block's @code{N_RBRAC} or the
3752 next @code{N_FUN}? (I believe its the first.)
3755 @c FIXME: This should go with the other stuff about global variables.
3756 Global variable stabs don't have location information. This comes
3757 from the external symbol for the same variable. The external symbol
3758 has a leading underbar on the _name of the variable and the stab does
3759 not. How do we know these two symbol table entries are talking about
3760 the same symbol when their names are different? (Answer: the debugger
3761 knows that external symbols have leading underbars).
3763 @c FIXME: This is absurdly vague; there all kinds of differences, some
3764 @c of which are the same between gnu & sun, and some of which aren't.
3765 @c In particular, I'm pretty sure GCC works with Sun dbx by default.
3767 @c Can GCC be configured to output stabs the way the Sun compiler
3768 @c does, so that their native debugging tools work? <NO?> It doesn't by
3769 @c default. GDB reads either format of stab. (GCC or SunC). How about
3773 @node XCOFF Differences
3774 @appendix Differences Between GNU Stabs in a.out and GNU Stabs in XCOFF
3776 @c FIXME: Merge *all* these into the main body of the document.
3777 The AIX/RS6000 native object file format is XCOFF with stabs. This
3778 appendix only covers those differences which are not covered in the main
3779 body of this document.
3783 BSD a.out stab types correspond to AIX XCOFF storage classes. In general
3784 the mapping is @code{N_@var{stabtype}} becomes @code{C_@var{stabtype}}.
3785 Some stab types in a.out are not supported in XCOFF; most of these use
3788 @c FIXME: I think they are trying to say something about whether the
3789 @c assembler defaults the value to the location counter.
3791 If the XCOFF stab is an @code{N_FUN} (@code{C_FUN}) then follow the
3792 string field with @samp{,.} instead of just @samp{,}.
3795 I think that's it for @file{.s} file differences. They could stand to be
3796 better presented. This is just a list of what I have noticed so far.
3797 There are a @emph{lot} of differences in the information in the symbol
3798 tables of the executable and object files.
3800 Mapping of a.out stab types to XCOFF storage classes:
3803 stab type storage class
3804 -------------------------------
3840 @node Sun Differences
3841 @appendix Differences Between GNU Stabs and Sun Native Stabs
3843 @c FIXME: Merge all this stuff into the main body of the document.
3847 GNU C stabs define @emph{all} types, file or procedure scope, as
3848 @code{N_LSYM}. Sun doc talks about using @code{N_GSYM} too.
3851 Sun C stabs use type number pairs in the format
3852 (@var{file-number},@var{type-number}) where @var{file-number} is a
3853 number starting with 1 and incremented for each sub-source file in the
3854 compilation. @var{type-number} is a number starting with 1 and
3855 incremented for each new type defined in the compilation. GNU C stabs
3856 use the type number alone, with no source file number.
3860 @appendix Using Stabs in Their Own Sections
3862 Many object file formats allow tools to create object files with custom
3863 sections containing any arbitrary data. For any such object file
3864 format, stabs can be embedded in special sections. This is how stabs
3865 are used with ELF and SOM, and aside from ECOFF and XCOFF, is how stabs
3869 * Stab Section Basics:: How to embed stabs in sections
3870 * ELF Linker Relocation:: Sun ELF hacks
3873 @node Stab Section Basics
3874 @appendixsec How to Embed Stabs in Sections
3876 The assembler creates two custom sections, a section named @code{.stab}
3877 which contains an array of fixed length structures, one struct per stab,
3878 and a section named @code{.stabstr} containing all the variable length
3879 strings that are referenced by stabs in the @code{.stab} section. The
3880 byte order of the stabs binary data depends on the object file format.
3881 For ELF, it matches the byte order of the ELF file itself, as determined
3882 from the @code{EI_DATA} field in the @code{e_ident} member of the ELF
3883 header. For SOM, it is always big-endian (is this true??? FIXME). For
3884 COFF, it matches the byte order of the COFF headers.
3886 The first stab in the @code{.stab} section for each compilation unit is
3887 synthetic, generated entirely by the assembler, with no corresponding
3888 @code{.stab} directive as input to the assembler. This stab contains
3889 the following fields:
3893 Offset in the @code{.stabstr} section to the source filename.
3899 Unused field, always zero.
3902 Count of upcoming symbols, i.e., the number of remaining stabs for this
3906 Size of the string table fragment associated with this source file, in
3910 The @code{.stabstr} section always starts with a null byte (so that string
3911 offsets of zero reference a null string), followed by random length strings,
3912 each of which is null byte terminated.
3914 The ELF section header for the @code{.stab} section has its
3915 @code{sh_link} member set to the section number of the @code{.stabstr}
3916 section, and the @code{.stabstr} section has its ELF section
3917 header @code{sh_type} member set to @code{SHT_STRTAB} to mark it as a
3918 string table. SOM and COFF have no way of linking the sections together
3919 or marking them as string tables.
3921 @node ELF Linker Relocation
3922 @appendixsec Having the Linker Relocate Stabs in ELF
3924 This section describes some Sun hacks for Stabs in ELF; it does not
3925 apply to COFF or SOM.
3927 To keep linking fast, you don't want the linker to have to relocate very
3928 many stabs. Making sure this is done for @code{N_SLINE},
3929 @code{N_RBRAC}, and @code{N_LBRAC} stabs is the most important thing
3930 (see the descriptions of those stabs for more information). But Sun's
3931 stabs in ELF has taken this further, to make all addresses in the
3932 @code{n_value} field (functions and static variables) relative to the
3933 source file. For the @code{N_SO} symbol itself, Sun simply omits the
3934 address. To find the address of each section corresponding to a given
3935 source file, the compiler puts out symbols giving the address of each
3936 section for a given source file. Since these are ELF (not stab)
3937 symbols, the linker relocates them correctly without having to touch the
3938 stabs section. They are named @code{Bbss.bss} for the bss section,
3939 @code{Ddata.data} for the data section, and @code{Drodata.rodata} for
3940 the rodata section. For the text section, there is no such symbol (but
3941 there should be, see below). For an example of how these symbols work,
3942 @xref{Stab Section Transformations}. GCC does not provide these symbols;
3943 it instead relies on the stabs getting relocated. Thus addresses which
3944 would normally be relative to @code{Bbss.bss}, etc., are already
3945 relocated. The Sun linker provided with Solaris 2.2 and earlier
3946 relocates stabs using normal ELF relocation information, as it would do
3947 for any section. Sun has been threatening to kludge their linker to not
3948 do this (to speed up linking), even though the correct way to avoid
3949 having the linker do these relocations is to have the compiler no longer
3950 output relocatable values. Last I heard they had been talked out of the
3951 linker kludge. See Sun point patch 101052-01 and Sun bug 1142109. With
3952 the Sun compiler this affects @samp{S} symbol descriptor stabs
3953 (@pxref{Statics}) and functions (@pxref{Procedures}). In the latter
3954 case, to adopt the clean solution (making the value of the stab relative
3955 to the start of the compilation unit), it would be necessary to invent a
3956 @code{Ttext.text} symbol, analogous to the @code{Bbss.bss}, etc.,
3957 symbols. I recommend this rather than using a zero value and getting
3958 the address from the ELF symbols.
3960 Finding the correct @code{Bbss.bss}, etc., symbol is difficult, because
3961 the linker simply concatenates the @code{.stab} sections from each
3962 @file{.o} file without including any information about which part of a
3963 @code{.stab} section comes from which @file{.o} file. The way GDB does
3964 this is to look for an ELF @code{STT_FILE} symbol which has the same
3965 name as the last component of the file name from the @code{N_SO} symbol
3966 in the stabs (for example, if the file name is @file{../../gdb/main.c},
3967 it looks for an ELF @code{STT_FILE} symbol named @code{main.c}). This
3968 loses if different files have the same name (they could be in different
3969 directories, a library could have been copied from one system to
3970 another, etc.). It would be much cleaner to have the @code{Bbss.bss}
3971 symbols in the stabs themselves. Having the linker relocate them there
3972 is no more work than having the linker relocate ELF symbols, and it
3973 solves the problem of having to associate the ELF and stab symbols.
3974 However, no one has yet designed or implemented such a scheme.
3976 @node Symbol Types Index
3977 @unnumbered Symbol Types Index