2 @setfilename stabs.info
9 * Stabs: (stabs). The "stabs" debugging information format.
15 This document describes the stabs debugging symbol tables.
17 Copyright 1992,1993,1994,1995,1997,1998,2000,2001
18 Free Software Foundation, Inc.
19 Contributed by Cygnus Support. Written by Julia Menapace, Jim Kingdon,
22 Permission is granted to copy, distribute and/or modify this document
23 under the terms of the GNU Free Documentation License, Version 1.1 or
24 any later version published by the Free Software Foundation; with no
25 Invariant Sections, with the Front-Cover Texts being ``A GNU Manual,''
26 and with the Back-Cover Texts as in (a) below.
28 (a) The FSF's Back-Cover Text is: ``You have freedom to copy and modify
29 this GNU Manual, like GNU software. Copies published by the Free
30 Software Foundation raise funds for GNU development.''
33 @setchapternewpage odd
36 @title The ``stabs'' debug format
37 @author Julia Menapace, Jim Kingdon, David MacKenzie
38 @author Cygnus Support
41 \def\$#1${{#1}} % Kluge: collect RCS revision info without $...$
42 \xdef\manvers{\$Revision$} % For use in headers, footers too
44 \hfill Cygnus Support\par
46 \hfill \TeX{}info \texinfoversion\par
50 @vskip 0pt plus 1filll
51 Copyright @copyright{} 1992,1993,1994,1995,1997,1998,2000,2001 Free Software Foundation, Inc.
52 Contributed by Cygnus Support.
54 Permission is granted to copy, distribute and/or modify this document
55 under the terms of the GNU Free Documentation License, Version 1.1 or
56 any later version published by the Free Software Foundation; with no
57 Invariant Sections, with the Front-Cover Texts being ``A GNU Manual,''
58 and with the Back-Cover Texts as in (a) below.
60 (a) The FSF's Back-Cover Text is: ``You have freedom to copy and modify
61 this GNU Manual, like GNU software. Copies published by the Free
62 Software Foundation raise funds for GNU development.''
68 @top The "stabs" representation of debugging information
70 This document describes the stabs debugging format.
73 * Overview:: Overview of stabs
74 * Program Structure:: Encoding of the structure of the program
75 * Constants:: Constants
77 * Types:: Type definitions
78 * Symbol Tables:: Symbol information in symbol tables
79 * Cplusplus:: Stabs specific to C++
80 * Stab Types:: Symbol types in a.out files
81 * Symbol Descriptors:: Table of symbol descriptors
82 * Type Descriptors:: Table of type descriptors
83 * Expanded Reference:: Reference information by stab type
84 * Questions:: Questions and anomalies
85 * Stab Sections:: In some object file formats, stabs are
87 * Symbol Types Index:: Index of symbolic stab symbol type names.
91 @c TeX can handle the contents at the start but makeinfo 3.12 can not
97 @chapter Overview of Stabs
99 @dfn{Stabs} refers to a format for information that describes a program
100 to a debugger. This format was apparently invented by
102 the University of California at Berkeley, for the @code{pdx} Pascal
103 debugger; the format has spread widely since then.
105 This document is one of the few published sources of documentation on
106 stabs. It is believed to be comprehensive for stabs used by C. The
107 lists of symbol descriptors (@pxref{Symbol Descriptors}) and type
108 descriptors (@pxref{Type Descriptors}) are believed to be completely
109 comprehensive. Stabs for COBOL-specific features and for variant
110 records (used by Pascal and Modula-2) are poorly documented here.
112 @c FIXME: Need to document all OS9000 stuff in GDB; see all references
113 @c to os9k_stabs in stabsread.c.
115 Other sources of information on stabs are @cite{Dbx and Dbxtool
116 Interfaces}, 2nd edition, by Sun, 1988, and @cite{AIX Version 3.2 Files
117 Reference}, Fourth Edition, September 1992, "dbx Stabstring Grammar" in
118 the a.out section, page 2-31. This document is believed to incorporate
119 the information from those two sources except where it explicitly directs
120 you to them for more information.
123 * Flow:: Overview of debugging information flow
124 * Stabs Format:: Overview of stab format
125 * String Field:: The string field
126 * C Example:: A simple example in C source
127 * Assembly Code:: The simple example at the assembly level
131 @section Overview of Debugging Information Flow
133 The GNU C compiler compiles C source in a @file{.c} file into assembly
134 language in a @file{.s} file, which the assembler translates into
135 a @file{.o} file, which the linker combines with other @file{.o} files and
136 libraries to produce an executable file.
138 With the @samp{-g} option, GCC puts in the @file{.s} file additional
139 debugging information, which is slightly transformed by the assembler
140 and linker, and carried through into the final executable. This
141 debugging information describes features of the source file like line
142 numbers, the types and scopes of variables, and function names,
143 parameters, and scopes.
145 For some object file formats, the debugging information is encapsulated
146 in assembler directives known collectively as @dfn{stab} (symbol table)
147 directives, which are interspersed with the generated code. Stabs are
148 the native format for debugging information in the a.out and XCOFF
149 object file formats. The GNU tools can also emit stabs in the COFF and
150 ECOFF object file formats.
152 The assembler adds the information from stabs to the symbol information
153 it places by default in the symbol table and the string table of the
154 @file{.o} file it is building. The linker consolidates the @file{.o}
155 files into one executable file, with one symbol table and one string
156 table. Debuggers use the symbol and string tables in the executable as
157 a source of debugging information about the program.
160 @section Overview of Stab Format
162 There are three overall formats for stab assembler directives,
163 differentiated by the first word of the stab. The name of the directive
164 describes which combination of four possible data fields follows. It is
165 either @code{.stabs} (string), @code{.stabn} (number), or @code{.stabd}
166 (dot). IBM's XCOFF assembler uses @code{.stabx} (and some other
167 directives such as @code{.file} and @code{.bi}) instead of
168 @code{.stabs}, @code{.stabn} or @code{.stabd}.
170 The overall format of each class of stab is:
173 .stabs "@var{string}",@var{type},@var{other},@var{desc},@var{value}
174 .stabn @var{type},@var{other},@var{desc},@var{value}
175 .stabd @var{type},@var{other},@var{desc}
176 .stabx "@var{string}",@var{value},@var{type},@var{sdb-type}
179 @c what is the correct term for "current file location"? My AIX
180 @c assembler manual calls it "the value of the current location counter".
181 For @code{.stabn} and @code{.stabd}, there is no @var{string} (the
182 @code{n_strx} field is zero; see @ref{Symbol Tables}). For
183 @code{.stabd}, the @var{value} field is implicit and has the value of
184 the current file location. For @code{.stabx}, the @var{sdb-type} field
185 is unused for stabs and can always be set to zero. The @var{other}
186 field is almost always unused and can be set to zero.
188 The number in the @var{type} field gives some basic information about
189 which type of stab this is (or whether it @emph{is} a stab, as opposed
190 to an ordinary symbol). Each valid type number defines a different stab
191 type; further, the stab type defines the exact interpretation of, and
192 possible values for, any remaining @var{string}, @var{desc}, or
193 @var{value} fields present in the stab. @xref{Stab Types}, for a list
194 in numeric order of the valid @var{type} field values for stab directives.
197 @section The String Field
199 For most stabs the string field holds the meat of the
200 debugging information. The flexible nature of this field
201 is what makes stabs extensible. For some stab types the string field
202 contains only a name. For other stab types the contents can be a great
205 The overall format of the string field for most stab types is:
208 "@var{name}:@var{symbol-descriptor} @var{type-information}"
211 @var{name} is the name of the symbol represented by the stab; it can
212 contain a pair of colons (@pxref{Nested Symbols}). @var{name} can be
213 omitted, which means the stab represents an unnamed object. For
214 example, @samp{:t10=*2} defines type 10 as a pointer to type 2, but does
215 not give the type a name. Omitting the @var{name} field is supported by
216 AIX dbx and GDB after about version 4.8, but not other debuggers. GCC
217 sometimes uses a single space as the name instead of omitting the name
218 altogether; apparently that is supported by most debuggers.
220 The @var{symbol-descriptor} following the @samp{:} is an alphabetic
221 character that tells more specifically what kind of symbol the stab
222 represents. If the @var{symbol-descriptor} is omitted, but type
223 information follows, then the stab represents a local variable. For a
224 list of symbol descriptors, see @ref{Symbol Descriptors}. The @samp{c}
225 symbol descriptor is an exception in that it is not followed by type
226 information. @xref{Constants}.
228 @var{type-information} is either a @var{type-number}, or
229 @samp{@var{type-number}=}. A @var{type-number} alone is a type
230 reference, referring directly to a type that has already been defined.
232 The @samp{@var{type-number}=} form is a type definition, where the
233 number represents a new type which is about to be defined. The type
234 definition may refer to other types by number, and those type numbers
235 may be followed by @samp{=} and nested definitions. Also, the Lucid
236 compiler will repeat @samp{@var{type-number}=} more than once if it
237 wants to define several type numbers at once.
239 In a type definition, if the character that follows the equals sign is
240 non-numeric then it is a @var{type-descriptor}, and tells what kind of
241 type is about to be defined. Any other values following the
242 @var{type-descriptor} vary, depending on the @var{type-descriptor}.
243 @xref{Type Descriptors}, for a list of @var{type-descriptor} values. If
244 a number follows the @samp{=} then the number is a @var{type-reference}.
245 For a full description of types, @ref{Types}.
247 A @var{type-number} is often a single number. The GNU and Sun tools
248 additionally permit a @var{type-number} to be a pair
249 (@var{file-number},@var{filetype-number}) (the parentheses appear in the
250 string, and serve to distinguish the two cases). The @var{file-number}
251 is 0 for the base source file, 1 for the first included file, 2 for the
252 next, and so on. The @var{filetype-number} is a number starting with
253 1 which is incremented for each new type defined in the file.
254 (Separating the file number and the type number permits the
255 @code{N_BINCL} optimization to succeed more often; see @ref{Include
258 There is an AIX extension for type attributes. Following the @samp{=}
259 are any number of type attributes. Each one starts with @samp{@@} and
260 ends with @samp{;}. Debuggers, including AIX's dbx and GDB 4.10, skip
261 any type attributes they do not recognize. GDB 4.9 and other versions
262 of dbx may not do this. Because of a conflict with C++
263 (@pxref{Cplusplus}), new attributes should not be defined which begin
264 with a digit, @samp{(}, or @samp{-}; GDB may be unable to distinguish
265 those from the C++ type descriptor @samp{@@}. The attributes are:
268 @item a@var{boundary}
269 @var{boundary} is an integer specifying the alignment. I assume it
270 applies to all variables of this type.
273 Pointer class (for checking). Not sure what this means, or how
274 @var{integer} is interpreted.
277 Indicate this is a packed type, meaning that structure fields or array
278 elements are placed more closely in memory, to save memory at the
282 Size in bits of a variable of this type. This is fully supported by GDB
286 Indicate that this type is a string instead of an array of characters,
287 or a bitstring instead of a set. It doesn't change the layout of the
288 data being represented, but does enable the debugger to know which type
292 Indicate that this type is a vector instead of an array. The only
293 major difference between vectors and arrays is that vectors are
294 passed by value instead of by reference (vector coprocessor extension).
298 All of this can make the string field quite long. All versions of GDB,
299 and some versions of dbx, can handle arbitrarily long strings. But many
300 versions of dbx (or assemblers or linkers, I'm not sure which)
301 cretinously limit the strings to about 80 characters, so compilers which
302 must work with such systems need to split the @code{.stabs} directive
303 into several @code{.stabs} directives. Each stab duplicates every field
304 except the string field. The string field of every stab except the last
305 is marked as continued with a backslash at the end (in the assembly code
306 this may be written as a double backslash, depending on the assembler).
307 Removing the backslashes and concatenating the string fields of each
308 stab produces the original, long string. Just to be incompatible (or so
309 they don't have to worry about what the assembler does with
310 backslashes), AIX can use @samp{?} instead of backslash.
313 @section A Simple Example in C Source
315 To get the flavor of how stabs describe source information for a C
316 program, let's look at the simple program:
321 printf("Hello world");
325 When compiled with @samp{-g}, the program above yields the following
326 @file{.s} file. Line numbers have been added to make it easier to refer
327 to parts of the @file{.s} file in the description of the stabs that
331 @section The Simple Example at the Assembly Level
333 This simple ``hello world'' example demonstrates several of the stab
334 types used to describe C language source files.
338 2 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0
339 3 .stabs "hello.c",100,0,0,Ltext0
342 6 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0
343 7 .stabs "char:t2=r2;0;127;",128,0,0,0
344 8 .stabs "long int:t3=r1;-2147483648;2147483647;",128,0,0,0
345 9 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
346 10 .stabs "long unsigned int:t5=r1;0;-1;",128,0,0,0
347 11 .stabs "short int:t6=r1;-32768;32767;",128,0,0,0
348 12 .stabs "long long int:t7=r1;0;-1;",128,0,0,0
349 13 .stabs "short unsigned int:t8=r1;0;65535;",128,0,0,0
350 14 .stabs "long long unsigned int:t9=r1;0;-1;",128,0,0,0
351 15 .stabs "signed char:t10=r1;-128;127;",128,0,0,0
352 16 .stabs "unsigned char:t11=r1;0;255;",128,0,0,0
353 17 .stabs "float:t12=r1;4;0;",128,0,0,0
354 18 .stabs "double:t13=r1;8;0;",128,0,0,0
355 19 .stabs "long double:t14=r1;8;0;",128,0,0,0
356 20 .stabs "void:t15=15",128,0,0,0
359 23 .ascii "Hello, world!\12\0"
374 38 sethi %hi(LC0),%o1
375 39 or %o1,%lo(LC0),%o0
386 50 .stabs "main:F1",36,0,0,_main
387 51 .stabn 192,0,0,LBB2
388 52 .stabn 224,0,0,LBE2
391 @node Program Structure
392 @chapter Encoding the Structure of the Program
394 The elements of the program structure that stabs encode include the name
395 of the main function, the names of the source and include files, the
396 line numbers, procedure names and types, and the beginnings and ends of
400 * Main Program:: Indicate what the main program is
401 * Source Files:: The path and name of the source file
402 * Include Files:: Names of include files
405 * Nested Procedures::
407 * Alternate Entry Points:: Entering procedures except at the beginning.
411 @section Main Program
414 Most languages allow the main program to have any name. The
415 @code{N_MAIN} stab type tells the debugger the name that is used in this
416 program. Only the string field is significant; it is the name of
417 a function which is the main program. Most C compilers do not use this
418 stab (they expect the debugger to assume that the name is @code{main}),
419 but some C compilers emit an @code{N_MAIN} stab for the @code{main}
420 function. I'm not sure how XCOFF handles this.
423 @section Paths and Names of the Source Files
426 Before any other stabs occur, there must be a stab specifying the source
427 file. This information is contained in a symbol of stab type
428 @code{N_SO}; the string field contains the name of the file. The
429 value of the symbol is the start address of the portion of the
430 text section corresponding to that file.
432 With the Sun Solaris2 compiler, the desc field contains a
433 source-language code.
434 @c Do the debuggers use it? What are the codes? -djm
436 Some compilers (for example, GCC2 and SunOS4 @file{/bin/cc}) also
437 include the directory in which the source was compiled, in a second
438 @code{N_SO} symbol preceding the one containing the file name. This
439 symbol can be distinguished by the fact that it ends in a slash. Code
440 from the @code{cfront} C++ compiler can have additional @code{N_SO} symbols for
441 nonexistent source files after the @code{N_SO} for the real source file;
442 these are believed to contain no useful information.
447 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0 # @r{100 is N_SO}
448 .stabs "hello.c",100,0,0,Ltext0
454 Instead of @code{N_SO} symbols, XCOFF uses a @code{.file} assembler
455 directive which assembles to a @code{C_FILE} symbol; explaining this in
456 detail is outside the scope of this document.
458 @c FIXME: Exactly when should the empty N_SO be used? Why?
459 If it is useful to indicate the end of a source file, this is done with
460 an @code{N_SO} symbol with an empty string for the name. The value is
461 the address of the end of the text section for the file. For some
462 systems, there is no indication of the end of a source file, and you
463 just need to figure it ended when you see an @code{N_SO} for a different
464 source file, or a symbol ending in @code{.o} (which at least some
465 linkers insert to mark the start of a new @code{.o} file).
468 @section Names of Include Files
470 There are several schemes for dealing with include files: the
471 traditional @code{N_SOL} approach, Sun's @code{N_BINCL} approach, and the
472 XCOFF @code{C_BINCL} approach (which despite the similar name has little in
473 common with @code{N_BINCL}).
476 An @code{N_SOL} symbol specifies which include file subsequent symbols
477 refer to. The string field is the name of the file and the value is the
478 text address corresponding to the end of the previous include file and
479 the start of this one. To specify the main source file again, use an
480 @code{N_SOL} symbol with the name of the main source file.
485 The @code{N_BINCL} approach works as follows. An @code{N_BINCL} symbol
486 specifies the start of an include file. In an object file, only the
487 string is significant; the linker puts data into some of the other
488 fields. The end of the include file is marked by an @code{N_EINCL}
489 symbol (which has no string field). In an object file, there is no
490 significant data in the @code{N_EINCL} symbol. @code{N_BINCL} and
491 @code{N_EINCL} can be nested.
493 If the linker detects that two source files have identical stabs between
494 an @code{N_BINCL} and @code{N_EINCL} pair (as will generally be the case
495 for a header file), then it only puts out the stabs once. Each
496 additional occurrence is replaced by an @code{N_EXCL} symbol. I believe
497 the GNU linker and the Sun (both SunOS4 and Solaris) linker are the only
498 ones which supports this feature.
500 A linker which supports this feature will set the value of a
501 @code{N_BINCL} symbol to the total of all the characters in the stabs
502 strings included in the header file, omitting any file numbers. The
503 value of an @code{N_EXCL} symbol is the same as the value of the
504 @code{N_BINCL} symbol it replaces. This information can be used to
505 match up @code{N_EXCL} and @code{N_BINCL} symbols which have the same
506 filename. The @code{N_EINCL} value, and the values of the other and
507 description fields for all three, appear to always be zero.
511 For the start of an include file in XCOFF, use the @file{.bi} assembler
512 directive, which generates a @code{C_BINCL} symbol. A @file{.ei}
513 directive, which generates a @code{C_EINCL} symbol, denotes the end of
514 the include file. Both directives are followed by the name of the
515 source file in quotes, which becomes the string for the symbol.
516 The value of each symbol, produced automatically by the assembler
517 and linker, is the offset into the executable of the beginning
518 (inclusive, as you'd expect) or end (inclusive, as you would not expect)
519 of the portion of the COFF line table that corresponds to this include
520 file. @code{C_BINCL} and @code{C_EINCL} do not nest.
523 @section Line Numbers
526 An @code{N_SLINE} symbol represents the start of a source line. The
527 desc field contains the line number and the value contains the code
528 address for the start of that source line. On most machines the address
529 is absolute; for stabs in sections (@pxref{Stab Sections}), it is
530 relative to the function in which the @code{N_SLINE} symbol occurs.
534 GNU documents @code{N_DSLINE} and @code{N_BSLINE} symbols for line
535 numbers in the data or bss segments, respectively. They are identical
536 to @code{N_SLINE} but are relocated differently by the linker. They
537 were intended to be used to describe the source location of a variable
538 declaration, but I believe that GCC2 actually puts the line number in
539 the desc field of the stab for the variable itself. GDB has been
540 ignoring these symbols (unless they contain a string field) since
543 For single source lines that generate discontiguous code, such as flow
544 of control statements, there may be more than one line number entry for
545 the same source line. In this case there is a line number entry at the
546 start of each code range, each with the same line number.
548 XCOFF does not use stabs for line numbers. Instead, it uses COFF line
549 numbers (which are outside the scope of this document). Standard COFF
550 line numbers cannot deal with include files, but in XCOFF this is fixed
551 with the @code{C_BINCL} method of marking include files (@pxref{Include
557 @findex N_FUN, for functions
559 @findex N_STSYM, for functions (Sun acc)
560 @findex N_GSYM, for functions (Sun acc)
561 All of the following stabs normally use the @code{N_FUN} symbol type.
562 However, Sun's @code{acc} compiler on SunOS4 uses @code{N_GSYM} and
563 @code{N_STSYM}, which means that the value of the stab for the function
564 is useless and the debugger must get the address of the function from
565 the non-stab symbols instead. On systems where non-stab symbols have
566 leading underscores, the stabs will lack underscores and the debugger
567 needs to know about the leading underscore to match up the stab and the
568 non-stab symbol. BSD Fortran is said to use @code{N_FNAME} with the
569 same restriction; the value of the symbol is not useful (I'm not sure it
570 really does use this, because GDB doesn't handle this and no one has
574 A function is represented by an @samp{F} symbol descriptor for a global
575 (extern) function, and @samp{f} for a static (local) function. For
576 a.out, the value of the symbol is the address of the start of the
577 function; it is already relocated. For stabs in ELF, the SunPRO
578 compiler version 2.0.1 and GCC put out an address which gets relocated
579 by the linker. In a future release SunPRO is planning to put out zero,
580 in which case the address can be found from the ELF (non-stab) symbol.
581 Because looking things up in the ELF symbols would probably be slow, I'm
582 not sure how to find which symbol of that name is the right one, and
583 this doesn't provide any way to deal with nested functions, it would
584 probably be better to make the value of the stab an address relative to
585 the start of the file, or just absolute. See @ref{ELF Linker
586 Relocation} for more information on linker relocation of stabs in ELF
587 files. For XCOFF, the stab uses the @code{C_FUN} storage class and the
588 value of the stab is meaningless; the address of the function can be
589 found from the csect symbol (XTY_LD/XMC_PR).
591 The type information of the stab represents the return type of the
592 function; thus @samp{foo:f5} means that foo is a function returning type
593 5. There is no need to try to get the line number of the start of the
594 function from the stab for the function; it is in the next
595 @code{N_SLINE} symbol.
597 @c FIXME: verify whether the "I suspect" below is true or not.
598 Some compilers (such as Sun's Solaris compiler) support an extension for
599 specifying the types of the arguments. I suspect this extension is not
600 used for old (non-prototyped) function definitions in C. If the
601 extension is in use, the type information of the stab for the function
602 is followed by type information for each argument, with each argument
603 preceded by @samp{;}. An argument type of 0 means that additional
604 arguments are being passed, whose types and number may vary (@samp{...}
605 in ANSI C). GDB has tolerated this extension (parsed the syntax, if not
606 necessarily used the information) since at least version 4.8; I don't
607 know whether all versions of dbx tolerate it. The argument types given
608 here are not redundant with the symbols for the formal parameters
609 (@pxref{Parameters}); they are the types of the arguments as they are
610 passed, before any conversions might take place. For example, if a C
611 function which is declared without a prototype takes a @code{float}
612 argument, the value is passed as a @code{double} but then converted to a
613 @code{float}. Debuggers need to use the types given in the arguments
614 when printing values, but when calling the function they need to use the
615 types given in the symbol defining the function.
617 If the return type and types of arguments of a function which is defined
618 in another source file are specified (i.e., a function prototype in ANSI
619 C), traditionally compilers emit no stab; the only way for the debugger
620 to find the information is if the source file where the function is
621 defined was also compiled with debugging symbols. As an extension the
622 Solaris compiler uses symbol descriptor @samp{P} followed by the return
623 type of the function, followed by the arguments, each preceded by
624 @samp{;}, as in a stab with symbol descriptor @samp{f} or @samp{F}.
625 This use of symbol descriptor @samp{P} can be distinguished from its use
626 for register parameters (@pxref{Register Parameters}) by the fact that it has
627 symbol type @code{N_FUN}.
629 The AIX documentation also defines symbol descriptor @samp{J} as an
630 internal function. I assume this means a function nested within another
631 function. It also says symbol descriptor @samp{m} is a module in
632 Modula-2 or extended Pascal.
634 Procedures (functions which do not return values) are represented as
635 functions returning the @code{void} type in C. I don't see why this couldn't
636 be used for all languages (inventing a @code{void} type for this purpose if
637 necessary), but the AIX documentation defines @samp{I}, @samp{P}, and
638 @samp{Q} for internal, global, and static procedures, respectively.
639 These symbol descriptors are unusual in that they are not followed by
642 The following example shows a stab for a function @code{main} which
643 returns type number @code{1}. The @code{_main} specified for the value
644 is a reference to an assembler label which is used to fill in the start
645 address of the function.
648 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
651 The stab representing a procedure is located immediately following the
652 code of the procedure. This stab is in turn directly followed by a
653 group of other stabs describing elements of the procedure. These other
654 stabs describe the procedure's parameters, its block local variables, and
657 If functions can appear in different sections, then the debugger may not
658 be able to find the end of a function. Recent versions of GCC will mark
659 the end of a function with an @code{N_FUN} symbol with an empty string
660 for the name. The value is the address of the end of the current
661 function. Without such a symbol, there is no indication of the address
662 of the end of a function, and you must assume that it ended at the
663 starting address of the next function or at the end of the text section
666 @node Nested Procedures
667 @section Nested Procedures
669 For any of the symbol descriptors representing procedures, after the
670 symbol descriptor and the type information is optionally a scope
671 specifier. This consists of a comma, the name of the procedure, another
672 comma, and the name of the enclosing procedure. The first name is local
673 to the scope specified, and seems to be redundant with the name of the
674 symbol (before the @samp{:}). This feature is used by GCC, and
675 presumably Pascal, Modula-2, etc., compilers, for nested functions.
677 If procedures are nested more than one level deep, only the immediately
678 containing scope is specified. For example, this code:
690 return baz (x + 2 * y);
692 return x + bar (3 * x);
700 .stabs "baz:f1,baz,bar",36,0,0,_baz.15 # @r{36 is N_FUN}
701 .stabs "bar:f1,bar,foo",36,0,0,_bar.12
702 .stabs "foo:F1",36,0,0,_foo
705 @node Block Structure
706 @section Block Structure
710 @c For GCC 2.5.8 or so stabs-in-coff, these are absolute instead of
711 @c function relative (as documented below). But GDB has never been able
712 @c to deal with that (it had wanted them to be relative to the file, but
713 @c I just fixed that (between GDB 4.12 and 4.13)), so it is function
714 @c relative just like ELF and SOM and the below documentation.
715 The program's block structure is represented by the @code{N_LBRAC} (left
716 brace) and the @code{N_RBRAC} (right brace) stab types. The variables
717 defined inside a block precede the @code{N_LBRAC} symbol for most
718 compilers, including GCC. Other compilers, such as the Convex, Acorn
719 RISC machine, and Sun @code{acc} compilers, put the variables after the
720 @code{N_LBRAC} symbol. The values of the @code{N_LBRAC} and
721 @code{N_RBRAC} symbols are the start and end addresses of the code of
722 the block, respectively. For most machines, they are relative to the
723 starting address of this source file. For the Gould NP1, they are
724 absolute. For stabs in sections (@pxref{Stab Sections}), they are
725 relative to the function in which they occur.
727 The @code{N_LBRAC} and @code{N_RBRAC} stabs that describe the block
728 scope of a procedure are located after the @code{N_FUN} stab that
729 represents the procedure itself.
731 Sun documents the desc field of @code{N_LBRAC} and
732 @code{N_RBRAC} symbols as containing the nesting level of the block.
733 However, dbx seems to not care, and GCC always sets desc to
739 For XCOFF, block scope is indicated with @code{C_BLOCK} symbols. If the
740 name of the symbol is @samp{.bb}, then it is the beginning of the block;
741 if the name of the symbol is @samp{.be}; it is the end of the block.
743 @node Alternate Entry Points
744 @section Alternate Entry Points
748 Some languages, like Fortran, have the ability to enter procedures at
749 some place other than the beginning. One can declare an alternate entry
750 point. The @code{N_ENTRY} stab is for this; however, the Sun FORTRAN
751 compiler doesn't use it. According to AIX documentation, only the name
752 of a @code{C_ENTRY} stab is significant; the address of the alternate
753 entry point comes from the corresponding external symbol. A previous
754 revision of this document said that the value of an @code{N_ENTRY} stab
755 was the address of the alternate entry point, but I don't know the
756 source for that information.
761 The @samp{c} symbol descriptor indicates that this stab represents a
762 constant. This symbol descriptor is an exception to the general rule
763 that symbol descriptors are followed by type information. Instead, it
764 is followed by @samp{=} and one of the following:
768 Boolean constant. @var{value} is a numeric value; I assume it is 0 for
772 Character constant. @var{value} is the numeric value of the constant.
774 @item e @var{type-information} , @var{value}
775 Constant whose value can be represented as integral.
776 @var{type-information} is the type of the constant, as it would appear
777 after a symbol descriptor (@pxref{String Field}). @var{value} is the
778 numeric value of the constant. GDB 4.9 does not actually get the right
779 value if @var{value} does not fit in a host @code{int}, but it does not
780 do anything violent, and future debuggers could be extended to accept
781 integers of any size (whether unsigned or not). This constant type is
782 usually documented as being only for enumeration constants, but GDB has
783 never imposed that restriction; I don't know about other debuggers.
786 Integer constant. @var{value} is the numeric value. The type is some
787 sort of generic integer type (for GDB, a host @code{int}); to specify
788 the type explicitly, use @samp{e} instead.
791 Real constant. @var{value} is the real value, which can be @samp{INF}
792 (optionally preceded by a sign) for infinity, @samp{QNAN} for a quiet
793 NaN (not-a-number), or @samp{SNAN} for a signalling NaN. If it is a
794 normal number the format is that accepted by the C library function
798 String constant. @var{string} is a string enclosed in either @samp{'}
799 (in which case @samp{'} characters within the string are represented as
800 @samp{\'} or @samp{"} (in which case @samp{"} characters within the
801 string are represented as @samp{\"}).
803 @item S @var{type-information} , @var{elements} , @var{bits} , @var{pattern}
804 Set constant. @var{type-information} is the type of the constant, as it
805 would appear after a symbol descriptor (@pxref{String Field}).
806 @var{elements} is the number of elements in the set (does this means
807 how many bits of @var{pattern} are actually used, which would be
808 redundant with the type, or perhaps the number of bits set in
809 @var{pattern}? I don't get it), @var{bits} is the number of bits in the
810 constant (meaning it specifies the length of @var{pattern}, I think),
811 and @var{pattern} is a hexadecimal representation of the set. AIX
812 documentation refers to a limit of 32 bytes, but I see no reason why
813 this limit should exist. This form could probably be used for arbitrary
814 constants, not just sets; the only catch is that @var{pattern} should be
815 understood to be target, not host, byte order and format.
818 The boolean, character, string, and set constants are not supported by
819 GDB 4.9, but it ignores them. GDB 4.8 and earlier gave an error
820 message and refused to read symbols from the file containing the
823 The above information is followed by @samp{;}.
828 Different types of stabs describe the various ways that variables can be
829 allocated: on the stack, globally, in registers, in common blocks,
830 statically, or as arguments to a function.
833 * Stack Variables:: Variables allocated on the stack.
834 * Global Variables:: Variables used by more than one source file.
835 * Register Variables:: Variables in registers.
836 * Common Blocks:: Variables statically allocated together.
837 * Statics:: Variables local to one source file.
838 * Based Variables:: Fortran pointer based variables.
839 * Parameters:: Variables for arguments to functions.
842 @node Stack Variables
843 @section Automatic Variables Allocated on the Stack
845 If a variable's scope is local to a function and its lifetime is only as
846 long as that function executes (C calls such variables
847 @dfn{automatic}), it can be allocated in a register (@pxref{Register
848 Variables}) or on the stack.
850 @findex N_LSYM, for stack variables
852 Each variable allocated on the stack has a stab with the symbol
853 descriptor omitted. Since type information should begin with a digit,
854 @samp{-}, or @samp{(}, only those characters precluded from being used
855 for symbol descriptors. However, the Acorn RISC machine (ARM) is said
856 to get this wrong: it puts out a mere type definition here, without the
857 preceding @samp{@var{type-number}=}. This is a bad idea; there is no
858 guarantee that type descriptors are distinct from symbol descriptors.
859 Stabs for stack variables use the @code{N_LSYM} stab type, or
860 @code{C_LSYM} for XCOFF.
862 The value of the stab is the offset of the variable within the
863 local variables. On most machines this is an offset from the frame
864 pointer and is negative. The location of the stab specifies which block
865 it is defined in; see @ref{Block Structure}.
867 For example, the following C code:
877 produces the following stabs:
880 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
881 .stabs "x:1",128,0,0,-12 # @r{128 is N_LSYM}
882 .stabn 192,0,0,LBB2 # @r{192 is N_LBRAC}
883 .stabn 224,0,0,LBE2 # @r{224 is N_RBRAC}
886 See @ref{Procedures} for more information on the @code{N_FUN} stab, and
887 @ref{Block Structure} for more information on the @code{N_LBRAC} and
888 @code{N_RBRAC} stabs.
890 @node Global Variables
891 @section Global Variables
895 @c FIXME: verify for sure that it really is C_GSYM on XCOFF
896 A variable whose scope is not specific to just one source file is
897 represented by the @samp{G} symbol descriptor. These stabs use the
898 @code{N_GSYM} stab type (C_GSYM for XCOFF). The type information for
899 the stab (@pxref{String Field}) gives the type of the variable.
901 For example, the following source code:
908 yields the following assembly code:
911 .stabs "g_foo:G2",32,0,0,0 # @r{32 is N_GSYM}
918 The address of the variable represented by the @code{N_GSYM} is not
919 contained in the @code{N_GSYM} stab. The debugger gets this information
920 from the external symbol for the global variable. In the example above,
921 the @code{.global _g_foo} and @code{_g_foo:} lines tell the assembler to
922 produce an external symbol.
924 Some compilers, like GCC, output @code{N_GSYM} stabs only once, where
925 the variable is defined. Other compilers, like SunOS4 /bin/cc, output a
926 @code{N_GSYM} stab for each compilation unit which references the
929 @node Register Variables
930 @section Register Variables
934 @c According to an old version of this manual, AIX uses C_RPSYM instead
935 @c of C_RSYM. I am skeptical; this should be verified.
936 Register variables have their own stab type, @code{N_RSYM}
937 (@code{C_RSYM} for XCOFF), and their own symbol descriptor, @samp{r}.
938 The stab's value is the number of the register where the variable data
940 @c .stabs "name:type",N_RSYM,0,RegSize,RegNumber (Sun doc)
942 AIX defines a separate symbol descriptor @samp{d} for floating point
943 registers. This seems unnecessary; why not just just give floating
944 point registers different register numbers? I have not verified whether
945 the compiler actually uses @samp{d}.
947 If the register is explicitly allocated to a global variable, but not
951 register int g_bar asm ("%g5");
955 then the stab may be emitted at the end of the object file, with
956 the other bss symbols.
959 @section Common Blocks
961 A common block is a statically allocated section of memory which can be
962 referred to by several source files. It may contain several variables.
963 I believe Fortran is the only language with this feature.
969 A @code{N_BCOMM} stab begins a common block and an @code{N_ECOMM} stab
970 ends it. The only field that is significant in these two stabs is the
971 string, which names a normal (non-debugging) symbol that gives the
972 address of the common block. According to IBM documentation, only the
973 @code{N_BCOMM} has the name of the common block (even though their
974 compiler actually puts it both places).
978 The stabs for the members of the common block are between the
979 @code{N_BCOMM} and the @code{N_ECOMM}; the value of each stab is the
980 offset within the common block of that variable. IBM uses the
981 @code{C_ECOML} stab type, and there is a corresponding @code{N_ECOML}
982 stab type, but Sun's Fortran compiler uses @code{N_GSYM} instead. The
983 variables within a common block use the @samp{V} symbol descriptor (I
984 believe this is true of all Fortran variables). Other stabs (at least
985 type declarations using @code{C_DECL}) can also be between the
986 @code{N_BCOMM} and the @code{N_ECOMM}.
989 @section Static Variables
991 Initialized static variables are represented by the @samp{S} and
992 @samp{V} symbol descriptors. @samp{S} means file scope static, and
993 @samp{V} means procedure scope static. One exception: in XCOFF, IBM's
994 xlc compiler always uses @samp{V}, and whether it is file scope or not
995 is distinguished by whether the stab is located within a function.
997 @c This is probably not worth mentioning; it is only true on the sparc
998 @c for `double' variables which although declared const are actually in
999 @c the data segment (the text segment can't guarantee 8 byte alignment).
1001 @c 2.4.5 has a bug in that it uses @code{N_FUN}, so neither dbx nor GDB can
1002 @c find the variables)
1005 @findex N_FUN, for variables
1007 In a.out files, @code{N_STSYM} means the data section, @code{N_FUN}
1008 means the text section, and @code{N_LCSYM} means the bss section. For
1009 those systems with a read-only data section separate from the text
1010 section (Solaris), @code{N_ROSYM} means the read-only data section.
1012 For example, the source lines:
1015 static const int var_const = 5;
1016 static int var_init = 2;
1017 static int var_noinit;
1021 yield the following stabs:
1024 .stabs "var_const:S1",36,0,0,_var_const # @r{36 is N_FUN}
1026 .stabs "var_init:S1",38,0,0,_var_init # @r{38 is N_STSYM}
1028 .stabs "var_noinit:S1",40,0,0,_var_noinit # @r{40 is N_LCSYM}
1034 In XCOFF files, the stab type need not indicate the section;
1035 @code{C_STSYM} can be used for all statics. Also, each static variable
1036 is enclosed in a static block. A @code{C_BSTAT} (emitted with a
1037 @samp{.bs} assembler directive) symbol begins the static block; its
1038 value is the symbol number of the csect symbol whose value is the
1039 address of the static block, its section is the section of the variables
1040 in that static block, and its name is @samp{.bs}. A @code{C_ESTAT}
1041 (emitted with a @samp{.es} assembler directive) symbol ends the static
1042 block; its name is @samp{.es} and its value and section are ignored.
1044 In ECOFF files, the storage class is used to specify the section, so the
1045 stab type need not indicate the section.
1047 In ELF files, for the SunPRO compiler version 2.0.1, symbol descriptor
1048 @samp{S} means that the address is absolute (the linker relocates it)
1049 and symbol descriptor @samp{V} means that the address is relative to the
1050 start of the relevant section for that compilation unit. SunPRO has
1051 plans to have the linker stop relocating stabs; I suspect that their the
1052 debugger gets the address from the corresponding ELF (not stab) symbol.
1053 I'm not sure how to find which symbol of that name is the right one.
1054 The clean way to do all this would be to have a the value of a symbol
1055 descriptor @samp{S} symbol be an offset relative to the start of the
1056 file, just like everything else, but that introduces obvious
1057 compatibility problems. For more information on linker stab relocation,
1058 @xref{ELF Linker Relocation}.
1060 @node Based Variables
1061 @section Fortran Based Variables
1063 Fortran (at least, the Sun and SGI dialects of FORTRAN-77) has a feature
1064 which allows allocating arrays with @code{malloc}, but which avoids
1065 blurring the line between arrays and pointers the way that C does. In
1066 stabs such a variable uses the @samp{b} symbol descriptor.
1068 For example, the Fortran declarations
1071 real foo, foo10(10), foo10_5(10,5)
1073 pointer (foo10p, foo10)
1074 pointer (foo105p, foo10_5)
1082 foo10_5:bar3;1;5;ar3;1;10;6
1085 In this example, @code{real} is type 6 and type 3 is an integral type
1086 which is the type of the subscripts of the array (probably
1089 The @samp{b} symbol descriptor is like @samp{V} in that it denotes a
1090 statically allocated symbol whose scope is local to a function; see
1091 @xref{Statics}. The value of the symbol, instead of being the address
1092 of the variable itself, is the address of a pointer to that variable.
1093 So in the above example, the value of the @code{foo} stab is the address
1094 of a pointer to a real, the value of the @code{foo10} stab is the
1095 address of a pointer to a 10-element array of reals, and the value of
1096 the @code{foo10_5} stab is the address of a pointer to a 5-element array
1097 of 10-element arrays of reals.
1102 Formal parameters to a function are represented by a stab (or sometimes
1103 two; see below) for each parameter. The stabs are in the order in which
1104 the debugger should print the parameters (i.e., the order in which the
1105 parameters are declared in the source file). The exact form of the stab
1106 depends on how the parameter is being passed.
1110 Parameters passed on the stack use the symbol descriptor @samp{p} and
1111 the @code{N_PSYM} symbol type (or @code{C_PSYM} for XCOFF). The value
1112 of the symbol is an offset used to locate the parameter on the stack;
1113 its exact meaning is machine-dependent, but on most machines it is an
1114 offset from the frame pointer.
1116 As a simple example, the code:
1127 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
1128 .stabs "argc:p1",160,0,0,68 # @r{160 is N_PSYM}
1129 .stabs "argv:p20=*21=*2",160,0,0,72
1132 The type definition of @code{argv} is interesting because it contains
1133 several type definitions. Type 21 is pointer to type 2 (char) and
1134 @code{argv} (type 20) is pointer to type 21.
1136 @c FIXME: figure out what these mean and describe them coherently.
1137 The following symbol descriptors are also said to go with @code{N_PSYM}.
1138 The value of the symbol is said to be an offset from the argument
1139 pointer (I'm not sure whether this is true or not).
1143 pF Fortran function parameter
1144 X (function result variable)
1148 * Register Parameters::
1149 * Local Variable Parameters::
1150 * Reference Parameters::
1151 * Conformant Arrays::
1154 @node Register Parameters
1155 @subsection Passing Parameters in Registers
1157 If the parameter is passed in a register, then traditionally there are
1158 two symbols for each argument:
1161 .stabs "arg:p1" . . . ; N_PSYM
1162 .stabs "arg:r1" . . . ; N_RSYM
1165 Debuggers use the second one to find the value, and the first one to
1166 know that it is an argument.
1169 @findex N_RSYM, for parameters
1170 Because that approach is kind of ugly, some compilers use symbol
1171 descriptor @samp{P} or @samp{R} to indicate an argument which is in a
1172 register. Symbol type @code{C_RPSYM} is used in XCOFF and @code{N_RSYM}
1173 is used otherwise. The symbol's value is the register number. @samp{P}
1174 and @samp{R} mean the same thing; the difference is that @samp{P} is a
1175 GNU invention and @samp{R} is an IBM (XCOFF) invention. As of version
1176 4.9, GDB should handle either one.
1178 There is at least one case where GCC uses a @samp{p} and @samp{r} pair
1179 rather than @samp{P}; this is where the argument is passed in the
1180 argument list and then loaded into a register.
1182 According to the AIX documentation, symbol descriptor @samp{D} is for a
1183 parameter passed in a floating point register. This seems
1184 unnecessary---why not just use @samp{R} with a register number which
1185 indicates that it's a floating point register? I haven't verified
1186 whether the system actually does what the documentation indicates.
1188 @c FIXME: On the hppa this is for any type > 8 bytes, I think, and not
1189 @c for small structures (investigate).
1190 On the sparc and hppa, for a @samp{P} symbol whose type is a structure
1191 or union, the register contains the address of the structure. On the
1192 sparc, this is also true of a @samp{p} and @samp{r} pair (using Sun
1193 @code{cc}) or a @samp{p} symbol. However, if a (small) structure is
1194 really in a register, @samp{r} is used. And, to top it all off, on the
1195 hppa it might be a structure which was passed on the stack and loaded
1196 into a register and for which there is a @samp{p} and @samp{r} pair! I
1197 believe that symbol descriptor @samp{i} is supposed to deal with this
1198 case (it is said to mean "value parameter by reference, indirect
1199 access"; I don't know the source for this information), but I don't know
1200 details or what compilers or debuggers use it, if any (not GDB or GCC).
1201 It is not clear to me whether this case needs to be dealt with
1202 differently than parameters passed by reference (@pxref{Reference Parameters}).
1204 @node Local Variable Parameters
1205 @subsection Storing Parameters as Local Variables
1207 There is a case similar to an argument in a register, which is an
1208 argument that is actually stored as a local variable. Sometimes this
1209 happens when the argument was passed in a register and then the compiler
1210 stores it as a local variable. If possible, the compiler should claim
1211 that it's in a register, but this isn't always done.
1213 If a parameter is passed as one type and converted to a smaller type by
1214 the prologue (for example, the parameter is declared as a @code{float},
1215 but the calling conventions specify that it is passed as a
1216 @code{double}), then GCC2 (sometimes) uses a pair of symbols. The first
1217 symbol uses symbol descriptor @samp{p} and the type which is passed.
1218 The second symbol has the type and location which the parameter actually
1219 has after the prologue. For example, suppose the following C code
1220 appears with no prototypes involved:
1229 if @code{f} is passed as a double at stack offset 8, and the prologue
1230 converts it to a float in register number 0, then the stabs look like:
1233 .stabs "f:p13",160,0,3,8 # @r{160 is @code{N_PSYM}, here 13 is @code{double}}
1234 .stabs "f:r12",64,0,3,0 # @r{64 is @code{N_RSYM}, here 12 is @code{float}}
1237 In both stabs 3 is the line number where @code{f} is declared
1238 (@pxref{Line Numbers}).
1240 @findex N_LSYM, for parameter
1241 GCC, at least on the 960, has another solution to the same problem. It
1242 uses a single @samp{p} symbol descriptor for an argument which is stored
1243 as a local variable but uses @code{N_LSYM} instead of @code{N_PSYM}. In
1244 this case, the value of the symbol is an offset relative to the local
1245 variables for that function, not relative to the arguments; on some
1246 machines those are the same thing, but not on all.
1248 @c This is mostly just background info; the part that logically belongs
1249 @c here is the last sentence.
1250 On the VAX or on other machines in which the calling convention includes
1251 the number of words of arguments actually passed, the debugger (GDB at
1252 least) uses the parameter symbols to keep track of whether it needs to
1253 print nameless arguments in addition to the formal parameters which it
1254 has printed because each one has a stab. For example, in
1257 extern int fprintf (FILE *stream, char *format, @dots{});
1259 fprintf (stdout, "%d\n", x);
1262 there are stabs for @code{stream} and @code{format}. On most machines,
1263 the debugger can only print those two arguments (because it has no way
1264 of knowing that additional arguments were passed), but on the VAX or
1265 other machines with a calling convention which indicates the number of
1266 words of arguments, the debugger can print all three arguments. To do
1267 so, the parameter symbol (symbol descriptor @samp{p}) (not necessarily
1268 @samp{r} or symbol descriptor omitted symbols) needs to contain the
1269 actual type as passed (for example, @code{double} not @code{float} if it
1270 is passed as a double and converted to a float).
1272 @node Reference Parameters
1273 @subsection Passing Parameters by Reference
1275 If the parameter is passed by reference (e.g., Pascal @code{VAR}
1276 parameters), then the symbol descriptor is @samp{v} if it is in the
1277 argument list, or @samp{a} if it in a register. Other than the fact
1278 that these contain the address of the parameter rather than the
1279 parameter itself, they are identical to @samp{p} and @samp{R},
1280 respectively. I believe @samp{a} is an AIX invention; @samp{v} is
1281 supported by all stabs-using systems as far as I know.
1283 @node Conformant Arrays
1284 @subsection Passing Conformant Array Parameters
1286 @c Is this paragraph correct? It is based on piecing together patchy
1287 @c information and some guesswork
1288 Conformant arrays are a feature of Modula-2, and perhaps other
1289 languages, in which the size of an array parameter is not known to the
1290 called function until run-time. Such parameters have two stabs: a
1291 @samp{x} for the array itself, and a @samp{C}, which represents the size
1292 of the array. The value of the @samp{x} stab is the offset in the
1293 argument list where the address of the array is stored (it this right?
1294 it is a guess); the value of the @samp{C} stab is the offset in the
1295 argument list where the size of the array (in elements? in bytes?) is
1299 @chapter Defining Types
1301 The examples so far have described types as references to previously
1302 defined types, or defined in terms of subranges of or pointers to
1303 previously defined types. This chapter describes the other type
1304 descriptors that may follow the @samp{=} in a type definition.
1307 * Builtin Types:: Integers, floating point, void, etc.
1308 * Miscellaneous Types:: Pointers, sets, files, etc.
1309 * Cross-References:: Referring to a type not yet defined.
1310 * Subranges:: A type with a specific range.
1311 * Arrays:: An aggregate type of same-typed elements.
1312 * Strings:: Like an array but also has a length.
1313 * Enumerations:: Like an integer but the values have names.
1314 * Structures:: An aggregate type of different-typed elements.
1315 * Typedefs:: Giving a type a name.
1316 * Unions:: Different types sharing storage.
1321 @section Builtin Types
1323 Certain types are built in (@code{int}, @code{short}, @code{void},
1324 @code{float}, etc.); the debugger recognizes these types and knows how
1325 to handle them. Thus, don't be surprised if some of the following ways
1326 of specifying builtin types do not specify everything that a debugger
1327 would need to know about the type---in some cases they merely specify
1328 enough information to distinguish the type from other types.
1330 The traditional way to define builtin types is convoluted, so new ways
1331 have been invented to describe them. Sun's @code{acc} uses special
1332 builtin type descriptors (@samp{b} and @samp{R}), and IBM uses negative
1333 type numbers. GDB accepts all three ways, as of version 4.8; dbx just
1334 accepts the traditional builtin types and perhaps one of the other two
1335 formats. The following sections describe each of these formats.
1338 * Traditional Builtin Types:: Put on your seat belts and prepare for kludgery
1339 * Builtin Type Descriptors:: Builtin types with special type descriptors
1340 * Negative Type Numbers:: Builtin types using negative type numbers
1343 @node Traditional Builtin Types
1344 @subsection Traditional Builtin Types
1346 This is the traditional, convoluted method for defining builtin types.
1347 There are several classes of such type definitions: integer, floating
1348 point, and @code{void}.
1351 * Traditional Integer Types::
1352 * Traditional Other Types::
1355 @node Traditional Integer Types
1356 @subsubsection Traditional Integer Types
1358 Often types are defined as subranges of themselves. If the bounding values
1359 fit within an @code{int}, then they are given normally. For example:
1362 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0 # @r{128 is N_LSYM}
1363 .stabs "char:t2=r2;0;127;",128,0,0,0
1366 Builtin types can also be described as subranges of @code{int}:
1369 .stabs "unsigned short:t6=r1;0;65535;",128,0,0,0
1372 If the lower bound of a subrange is 0 and the upper bound is -1,
1373 the type is an unsigned integral type whose bounds are too
1374 big to describe in an @code{int}. Traditionally this is only used for
1375 @code{unsigned int} and @code{unsigned long}:
1378 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
1381 For larger types, GCC 2.4.5 puts out bounds in octal, with one or more
1382 leading zeroes. In this case a negative bound consists of a number
1383 which is a 1 bit (for the sign bit) followed by a 0 bit for each bit in
1384 the number (except the sign bit), and a positive bound is one which is a
1385 1 bit for each bit in the number (except possibly the sign bit). All
1386 known versions of dbx and GDB version 4 accept this (at least in the
1387 sense of not refusing to process the file), but GDB 3.5 refuses to read
1388 the whole file containing such symbols. So GCC 2.3.3 did not output the
1389 proper size for these types. As an example of octal bounds, the string
1390 fields of the stabs for 64 bit integer types look like:
1392 @c .stabs directives, etc., omitted to make it fit on the page.
1394 long int:t3=r1;001000000000000000000000;000777777777777777777777;
1395 long unsigned int:t5=r1;000000000000000000000000;001777777777777777777777;
1398 If the lower bound of a subrange is 0 and the upper bound is negative,
1399 the type is an unsigned integral type whose size in bytes is the
1400 absolute value of the upper bound. I believe this is a Convex
1401 convention for @code{unsigned long long}.
1403 If the lower bound of a subrange is negative and the upper bound is 0,
1404 the type is a signed integral type whose size in bytes is
1405 the absolute value of the lower bound. I believe this is a Convex
1406 convention for @code{long long}. To distinguish this from a legitimate
1407 subrange, the type should be a subrange of itself. I'm not sure whether
1408 this is the case for Convex.
1410 @node Traditional Other Types
1411 @subsubsection Traditional Other Types
1413 If the upper bound of a subrange is 0 and the lower bound is positive,
1414 the type is a floating point type, and the lower bound of the subrange
1415 indicates the number of bytes in the type:
1418 .stabs "float:t12=r1;4;0;",128,0,0,0
1419 .stabs "double:t13=r1;8;0;",128,0,0,0
1422 However, GCC writes @code{long double} the same way it writes
1423 @code{double}, so there is no way to distinguish.
1426 .stabs "long double:t14=r1;8;0;",128,0,0,0
1429 Complex types are defined the same way as floating-point types; there is
1430 no way to distinguish a single-precision complex from a double-precision
1431 floating-point type.
1433 The C @code{void} type is defined as itself:
1436 .stabs "void:t15=15",128,0,0,0
1439 I'm not sure how a boolean type is represented.
1441 @node Builtin Type Descriptors
1442 @subsection Defining Builtin Types Using Builtin Type Descriptors
1444 This is the method used by Sun's @code{acc} for defining builtin types.
1445 These are the type descriptors to define builtin types:
1448 @c FIXME: clean up description of width and offset, once we figure out
1450 @item b @var{signed} @var{char-flag} @var{width} ; @var{offset} ; @var{nbits} ;
1451 Define an integral type. @var{signed} is @samp{u} for unsigned or
1452 @samp{s} for signed. @var{char-flag} is @samp{c} which indicates this
1453 is a character type, or is omitted. I assume this is to distinguish an
1454 integral type from a character type of the same size, for example it
1455 might make sense to set it for the C type @code{wchar_t} so the debugger
1456 can print such variables differently (Solaris does not do this). Sun
1457 sets it on the C types @code{signed char} and @code{unsigned char} which
1458 arguably is wrong. @var{width} and @var{offset} appear to be for small
1459 objects stored in larger ones, for example a @code{short} in an
1460 @code{int} register. @var{width} is normally the number of bytes in the
1461 type. @var{offset} seems to always be zero. @var{nbits} is the number
1462 of bits in the type.
1464 Note that type descriptor @samp{b} used for builtin types conflicts with
1465 its use for Pascal space types (@pxref{Miscellaneous Types}); they can
1466 be distinguished because the character following the type descriptor
1467 will be a digit, @samp{(}, or @samp{-} for a Pascal space type, or
1468 @samp{u} or @samp{s} for a builtin type.
1471 Documented by AIX to define a wide character type, but their compiler
1472 actually uses negative type numbers (@pxref{Negative Type Numbers}).
1474 @item R @var{fp-type} ; @var{bytes} ;
1475 Define a floating point type. @var{fp-type} has one of the following values:
1479 IEEE 32-bit (single precision) floating point format.
1482 IEEE 64-bit (double precision) floating point format.
1484 @item 3 (NF_COMPLEX)
1485 @item 4 (NF_COMPLEX16)
1486 @item 5 (NF_COMPLEX32)
1487 @c "GDB source" really means @file{include/aout/stab_gnu.h}, but trying
1488 @c to put that here got an overfull hbox.
1489 These are for complex numbers. A comment in the GDB source describes
1490 them as Fortran @code{complex}, @code{double complex}, and
1491 @code{complex*16}, respectively, but what does that mean? (i.e., Single
1492 precision? Double precision?).
1494 @item 6 (NF_LDOUBLE)
1495 Long double. This should probably only be used for Sun format
1496 @code{long double}, and new codes should be used for other floating
1497 point formats (@code{NF_DOUBLE} can be used if a @code{long double} is
1498 really just an IEEE double, of course).
1501 @var{bytes} is the number of bytes occupied by the type. This allows a
1502 debugger to perform some operations with the type even if it doesn't
1503 understand @var{fp-type}.
1505 @item g @var{type-information} ; @var{nbits}
1506 Documented by AIX to define a floating type, but their compiler actually
1507 uses negative type numbers (@pxref{Negative Type Numbers}).
1509 @item c @var{type-information} ; @var{nbits}
1510 Documented by AIX to define a complex type, but their compiler actually
1511 uses negative type numbers (@pxref{Negative Type Numbers}).
1514 The C @code{void} type is defined as a signed integral type 0 bits long:
1516 .stabs "void:t19=bs0;0;0",128,0,0,0
1518 The Solaris compiler seems to omit the trailing semicolon in this case.
1519 Getting sloppy in this way is not a swift move because if a type is
1520 embedded in a more complex expression it is necessary to be able to tell
1523 I'm not sure how a boolean type is represented.
1525 @node Negative Type Numbers
1526 @subsection Negative Type Numbers
1528 This is the method used in XCOFF for defining builtin types.
1529 Since the debugger knows about the builtin types anyway, the idea of
1530 negative type numbers is simply to give a special type number which
1531 indicates the builtin type. There is no stab defining these types.
1533 There are several subtle issues with negative type numbers.
1535 One is the size of the type. A builtin type (for example the C types
1536 @code{int} or @code{long}) might have different sizes depending on
1537 compiler options, the target architecture, the ABI, etc. This issue
1538 doesn't come up for IBM tools since (so far) they just target the
1539 RS/6000; the sizes indicated below for each size are what the IBM
1540 RS/6000 tools use. To deal with differing sizes, either define separate
1541 negative type numbers for each size (which works but requires changing
1542 the debugger, and, unless you get both AIX dbx and GDB to accept the
1543 change, introduces an incompatibility), or use a type attribute
1544 (@pxref{String Field}) to define a new type with the appropriate size
1545 (which merely requires a debugger which understands type attributes,
1546 like AIX dbx or GDB). For example,
1549 .stabs "boolean:t10=@@s8;-16",128,0,0,0
1552 defines an 8-bit boolean type, and
1555 .stabs "boolean:t10=@@s64;-16",128,0,0,0
1558 defines a 64-bit boolean type.
1560 A similar issue is the format of the type. This comes up most often for
1561 floating-point types, which could have various formats (particularly
1562 extended doubles, which vary quite a bit even among IEEE systems).
1563 Again, it is best to define a new negative type number for each
1564 different format; changing the format based on the target system has
1565 various problems. One such problem is that the Alpha has both VAX and
1566 IEEE floating types. One can easily imagine one library using the VAX
1567 types and another library in the same executable using the IEEE types.
1568 Another example is that the interpretation of whether a boolean is true
1569 or false can be based on the least significant bit, most significant
1570 bit, whether it is zero, etc., and different compilers (or different
1571 options to the same compiler) might provide different kinds of boolean.
1573 The last major issue is the names of the types. The name of a given
1574 type depends @emph{only} on the negative type number given; these do not
1575 vary depending on the language, the target system, or anything else.
1576 One can always define separate type numbers---in the following list you
1577 will see for example separate @code{int} and @code{integer*4} types
1578 which are identical except for the name. But compatibility can be
1579 maintained by not inventing new negative type numbers and instead just
1580 defining a new type with a new name. For example:
1583 .stabs "CARDINAL:t10=-8",128,0,0,0
1586 Here is the list of negative type numbers. The phrase @dfn{integral
1587 type} is used to mean twos-complement (I strongly suspect that all
1588 machines which use stabs use twos-complement; most machines use
1589 twos-complement these days).
1593 @code{int}, 32 bit signed integral type.
1596 @code{char}, 8 bit type holding a character. Both GDB and dbx on AIX
1597 treat this as signed. GCC uses this type whether @code{char} is signed
1598 or not, which seems like a bad idea. The AIX compiler (@code{xlc}) seems to
1599 avoid this type; it uses -5 instead for @code{char}.
1602 @code{short}, 16 bit signed integral type.
1605 @code{long}, 32 bit signed integral type.
1608 @code{unsigned char}, 8 bit unsigned integral type.
1611 @code{signed char}, 8 bit signed integral type.
1614 @code{unsigned short}, 16 bit unsigned integral type.
1617 @code{unsigned int}, 32 bit unsigned integral type.
1620 @code{unsigned}, 32 bit unsigned integral type.
1623 @code{unsigned long}, 32 bit unsigned integral type.
1626 @code{void}, type indicating the lack of a value.
1629 @code{float}, IEEE single precision.
1632 @code{double}, IEEE double precision.
1635 @code{long double}, IEEE double precision. The compiler claims the size
1636 will increase in a future release, and for binary compatibility you have
1637 to avoid using @code{long double}. I hope when they increase it they
1638 use a new negative type number.
1641 @code{integer}. 32 bit signed integral type.
1644 @code{boolean}. 32 bit type. GDB and GCC assume that zero is false,
1645 one is true, and other values have unspecified meaning. I hope this
1646 agrees with how the IBM tools use the type.
1649 @code{short real}. IEEE single precision.
1652 @code{real}. IEEE double precision.
1655 @code{stringptr}. @xref{Strings}.
1658 @code{character}, 8 bit unsigned character type.
1661 @code{logical*1}, 8 bit type. This Fortran type has a split
1662 personality in that it is used for boolean variables, but can also be
1663 used for unsigned integers. 0 is false, 1 is true, and other values are
1667 @code{logical*2}, 16 bit type. This Fortran type has a split
1668 personality in that it is used for boolean variables, but can also be
1669 used for unsigned integers. 0 is false, 1 is true, and other values are
1673 @code{logical*4}, 32 bit type. This Fortran type has a split
1674 personality in that it is used for boolean variables, but can also be
1675 used for unsigned integers. 0 is false, 1 is true, and other values are
1679 @code{logical}, 32 bit type. This Fortran type has a split
1680 personality in that it is used for boolean variables, but can also be
1681 used for unsigned integers. 0 is false, 1 is true, and other values are
1685 @code{complex}. A complex type consisting of two IEEE single-precision
1686 floating point values.
1689 @code{complex}. A complex type consisting of two IEEE double-precision
1690 floating point values.
1693 @code{integer*1}, 8 bit signed integral type.
1696 @code{integer*2}, 16 bit signed integral type.
1699 @code{integer*4}, 32 bit signed integral type.
1702 @code{wchar}. Wide character, 16 bits wide, unsigned (what format?
1706 @code{long long}, 64 bit signed integral type.
1709 @code{unsigned long long}, 64 bit unsigned integral type.
1712 @code{logical*8}, 64 bit unsigned integral type.
1715 @code{integer*8}, 64 bit signed integral type.
1718 @node Miscellaneous Types
1719 @section Miscellaneous Types
1722 @item b @var{type-information} ; @var{bytes}
1723 Pascal space type. This is documented by IBM; what does it mean?
1725 This use of the @samp{b} type descriptor can be distinguished
1726 from its use for builtin integral types (@pxref{Builtin Type
1727 Descriptors}) because the character following the type descriptor is
1728 always a digit, @samp{(}, or @samp{-}.
1730 @item B @var{type-information}
1731 A volatile-qualified version of @var{type-information}. This is
1732 a Sun extension. References and stores to a variable with a
1733 volatile-qualified type must not be optimized or cached; they
1734 must occur as the user specifies them.
1736 @item d @var{type-information}
1737 File of type @var{type-information}. As far as I know this is only used
1740 @item k @var{type-information}
1741 A const-qualified version of @var{type-information}. This is a Sun
1742 extension. A variable with a const-qualified type cannot be modified.
1744 @item M @var{type-information} ; @var{length}
1745 Multiple instance type. The type seems to composed of @var{length}
1746 repetitions of @var{type-information}, for example @code{character*3} is
1747 represented by @samp{M-2;3}, where @samp{-2} is a reference to a
1748 character type (@pxref{Negative Type Numbers}). I'm not sure how this
1749 differs from an array. This appears to be a Fortran feature.
1750 @var{length} is a bound, like those in range types; see @ref{Subranges}.
1752 @item S @var{type-information}
1753 Pascal set type. @var{type-information} must be a small type such as an
1754 enumeration or a subrange, and the type is a bitmask whose length is
1755 specified by the number of elements in @var{type-information}.
1757 In CHILL, @c OBSOLETE
1758 if it is a bitstring instead of a set, also use the @samp{S}
1759 type attribute (@pxref{String Field}).
1761 @item * @var{type-information}
1762 Pointer to @var{type-information}.
1765 @node Cross-References
1766 @section Cross-References to Other Types
1768 A type can be used before it is defined; one common way to deal with
1769 that situation is just to use a type reference to a type which has not
1772 Another way is with the @samp{x} type descriptor, which is followed by
1773 @samp{s} for a structure tag, @samp{u} for a union tag, or @samp{e} for
1774 a enumerator tag, followed by the name of the tag, followed by @samp{:}.
1775 If the name contains @samp{::} between a @samp{<} and @samp{>} pair (for
1776 C++ templates), such a @samp{::} does not end the name---only a single
1777 @samp{:} ends the name; see @ref{Nested Symbols}.
1779 For example, the following C declarations:
1790 .stabs "bar:G16=*17=xsfoo:",32,0,0,0
1793 Not all debuggers support the @samp{x} type descriptor, so on some
1794 machines GCC does not use it. I believe that for the above example it
1795 would just emit a reference to type 17 and never define it, but I
1796 haven't verified that.
1798 Modula-2 imported types, at least on AIX, use the @samp{i} type
1799 descriptor, which is followed by the name of the module from which the
1800 type is imported, followed by @samp{:}, followed by the name of the
1801 type. There is then optionally a comma followed by type information for
1802 the type. This differs from merely naming the type (@pxref{Typedefs}) in
1803 that it identifies the module; I don't understand whether the name of
1804 the type given here is always just the same as the name we are giving
1805 it, or whether this type descriptor is used with a nameless stab
1806 (@pxref{String Field}), or what. The symbol ends with @samp{;}.
1809 @section Subrange Types
1811 The @samp{r} type descriptor defines a type as a subrange of another
1812 type. It is followed by type information for the type of which it is a
1813 subrange, a semicolon, an integral lower bound, a semicolon, an
1814 integral upper bound, and a semicolon. The AIX documentation does not
1815 specify the trailing semicolon, in an effort to specify array indexes
1816 more cleanly, but a subrange which is not an array index has always
1817 included a trailing semicolon (@pxref{Arrays}).
1819 Instead of an integer, either bound can be one of the following:
1822 @item A @var{offset}
1823 The bound is passed by reference on the stack at offset @var{offset}
1824 from the argument list. @xref{Parameters}, for more information on such
1827 @item T @var{offset}
1828 The bound is passed by value on the stack at offset @var{offset} from
1831 @item a @var{register-number}
1832 The bound is passed by reference in register number
1833 @var{register-number}.
1835 @item t @var{register-number}
1836 The bound is passed by value in register number @var{register-number}.
1842 Subranges are also used for builtin types; see @ref{Traditional Builtin Types}.
1845 @section Array Types
1847 Arrays use the @samp{a} type descriptor. Following the type descriptor
1848 is the type of the index and the type of the array elements. If the
1849 index type is a range type, it ends in a semicolon; otherwise
1850 (for example, if it is a type reference), there does not
1851 appear to be any way to tell where the types are separated. In an
1852 effort to clean up this mess, IBM documents the two types as being
1853 separated by a semicolon, and a range type as not ending in a semicolon
1854 (but this is not right for range types which are not array indexes,
1855 @pxref{Subranges}). I think probably the best solution is to specify
1856 that a semicolon ends a range type, and that the index type and element
1857 type of an array are separated by a semicolon, but that if the index
1858 type is a range type, the extra semicolon can be omitted. GDB (at least
1859 through version 4.9) doesn't support any kind of index type other than a
1860 range anyway; I'm not sure about dbx.
1862 It is well established, and widely used, that the type of the index,
1863 unlike most types found in the stabs, is merely a type definition, not
1864 type information (@pxref{String Field}) (that is, it need not start with
1865 @samp{@var{type-number}=} if it is defining a new type). According to a
1866 comment in GDB, this is also true of the type of the array elements; it
1867 gives @samp{ar1;1;10;ar1;1;10;4} as a legitimate way to express a two
1868 dimensional array. According to AIX documentation, the element type
1869 must be type information. GDB accepts either.
1871 The type of the index is often a range type, expressed as the type
1872 descriptor @samp{r} and some parameters. It defines the size of the
1873 array. In the example below, the range @samp{r1;0;2;} defines an index
1874 type which is a subrange of type 1 (integer), with a lower bound of 0
1875 and an upper bound of 2. This defines the valid range of subscripts of
1876 a three-element C array.
1878 For example, the definition:
1881 char char_vec[3] = @{'a','b','c'@};
1885 produces the output:
1888 .stabs "char_vec:G19=ar1;0;2;2",32,0,0,0
1897 If an array is @dfn{packed}, the elements are spaced more
1898 closely than normal, saving memory at the expense of speed. For
1899 example, an array of 3-byte objects might, if unpacked, have each
1900 element aligned on a 4-byte boundary, but if packed, have no padding.
1901 One way to specify that something is packed is with type attributes
1902 (@pxref{String Field}). In the case of arrays, another is to use the
1903 @samp{P} type descriptor instead of @samp{a}. Other than specifying a
1904 packed array, @samp{P} is identical to @samp{a}.
1906 @c FIXME-what is it? A pointer?
1907 An open array is represented by the @samp{A} type descriptor followed by
1908 type information specifying the type of the array elements.
1910 @c FIXME: what is the format of this type? A pointer to a vector of pointers?
1911 An N-dimensional dynamic array is represented by
1914 D @var{dimensions} ; @var{type-information}
1917 @c Does dimensions really have this meaning? The AIX documentation
1919 @var{dimensions} is the number of dimensions; @var{type-information}
1920 specifies the type of the array elements.
1922 @c FIXME: what is the format of this type? A pointer to some offsets in
1924 A subarray of an N-dimensional array is represented by
1927 E @var{dimensions} ; @var{type-information}
1930 @c Does dimensions really have this meaning? The AIX documentation
1932 @var{dimensions} is the number of dimensions; @var{type-information}
1933 specifies the type of the array elements.
1938 Some languages, like C or the original Pascal, do not have string types,
1939 they just have related things like arrays of characters. But most
1940 Pascals and various other languages have string types, which are
1941 indicated as follows:
1944 @item n @var{type-information} ; @var{bytes}
1945 @var{bytes} is the maximum length. I'm not sure what
1946 @var{type-information} is; I suspect that it means that this is a string
1947 of @var{type-information} (thus allowing a string of integers, a string
1948 of wide characters, etc., as well as a string of characters). Not sure
1949 what the format of this type is. This is an AIX feature.
1951 @item z @var{type-information} ; @var{bytes}
1952 Just like @samp{n} except that this is a gstring, not an ordinary
1953 string. I don't know the difference.
1956 Pascal Stringptr. What is this? This is an AIX feature.
1959 Languages, such as CHILL @c OBSOLETE
1960 which have a string type which is basically
1961 just an array of characters use the @samp{S} type attribute
1962 (@pxref{String Field}).
1965 @section Enumerations
1967 Enumerations are defined with the @samp{e} type descriptor.
1969 @c FIXME: Where does this information properly go? Perhaps it is
1970 @c redundant with something we already explain.
1971 The source line below declares an enumeration type at file scope.
1972 The type definition is located after the @code{N_RBRAC} that marks the end of
1973 the previous procedure's block scope, and before the @code{N_FUN} that marks
1974 the beginning of the next procedure's block scope. Therefore it does not
1975 describe a block local symbol, but a file local one.
1980 enum e_places @{first,second=3,last@};
1984 generates the following stab:
1987 .stabs "e_places:T22=efirst:0,second:3,last:4,;",128,0,0,0
1990 The symbol descriptor (@samp{T}) says that the stab describes a
1991 structure, enumeration, or union tag. The type descriptor @samp{e},
1992 following the @samp{22=} of the type definition narrows it down to an
1993 enumeration type. Following the @samp{e} is a list of the elements of
1994 the enumeration. The format is @samp{@var{name}:@var{value},}. The
1995 list of elements ends with @samp{;}. The fact that @var{value} is
1996 specified as an integer can cause problems if the value is large. GCC
1997 2.5.2 tries to output it in octal in that case with a leading zero,
1998 which is probably a good thing, although GDB 4.11 supports octal only in
1999 cases where decimal is perfectly good. Negative decimal values are
2000 supported by both GDB and dbx.
2002 There is no standard way to specify the size of an enumeration type; it
2003 is determined by the architecture (normally all enumerations types are
2004 32 bits). Type attributes can be used to specify an enumeration type of
2005 another size for debuggers which support them; see @ref{String Field}.
2007 Enumeration types are unusual in that they define symbols for the
2008 enumeration values (@code{first}, @code{second}, and @code{third} in the
2009 above example), and even though these symbols are visible in the file as
2010 a whole (rather than being in a more local namespace like structure
2011 member names), they are defined in the type definition for the
2012 enumeration type rather than each having their own symbol. In order to
2013 be fast, GDB will only get symbols from such types (in its initial scan
2014 of the stabs) if the type is the first thing defined after a @samp{T} or
2015 @samp{t} symbol descriptor (the above example fulfills this
2016 requirement). If the type does not have a name, the compiler should
2017 emit it in a nameless stab (@pxref{String Field}); GCC does this.
2022 The encoding of structures in stabs can be shown with an example.
2024 The following source code declares a structure tag and defines an
2025 instance of the structure in global scope. Then a @code{typedef} equates the
2026 structure tag with a new type. Separate stabs are generated for the
2027 structure tag, the structure @code{typedef}, and the structure instance. The
2028 stabs for the tag and the @code{typedef} are emitted when the definitions are
2029 encountered. Since the structure elements are not initialized, the
2030 stab and code for the structure variable itself is located at the end
2031 of the program in the bss section.
2038 struct s_tag* s_next;
2041 typedef struct s_tag s_typedef;
2044 The structure tag has an @code{N_LSYM} stab type because, like the
2045 enumeration, the symbol has file scope. Like the enumeration, the
2046 symbol descriptor is @samp{T}, for enumeration, structure, or tag type.
2047 The type descriptor @samp{s} following the @samp{16=} of the type
2048 definition narrows the symbol type to structure.
2050 Following the @samp{s} type descriptor is the number of bytes the
2051 structure occupies, followed by a description of each structure element.
2052 The structure element descriptions are of the form @var{name:type, bit
2053 offset from the start of the struct, number of bits in the element}.
2055 @c FIXME: phony line break. Can probably be fixed by using an example
2056 @c with fewer fields.
2059 .stabs "s_tag:T16=s20s_int:1,0,32;s_float:12,32,32;
2060 s_char_vec:17=ar1;0;7;2,64,64;s_next:18=*16,128,32;;",128,0,0,0
2063 In this example, the first two structure elements are previously defined
2064 types. For these, the type following the @samp{@var{name}:} part of the
2065 element description is a simple type reference. The other two structure
2066 elements are new types. In this case there is a type definition
2067 embedded after the @samp{@var{name}:}. The type definition for the
2068 array element looks just like a type definition for a stand-alone array.
2069 The @code{s_next} field is a pointer to the same kind of structure that
2070 the field is an element of. So the definition of structure type 16
2071 contains a type definition for an element which is a pointer to type 16.
2073 If a field is a static member (this is a C++ feature in which a single
2074 variable appears to be a field of every structure of a given type) it
2075 still starts out with the field name, a colon, and the type, but then
2076 instead of a comma, bit position, comma, and bit size, there is a colon
2077 followed by the name of the variable which each such field refers to.
2079 If the structure has methods (a C++ feature), they follow the non-method
2080 fields; see @ref{Cplusplus}.
2083 @section Giving a Type a Name
2085 @findex N_LSYM, for types
2086 @findex C_DECL, for types
2087 To give a type a name, use the @samp{t} symbol descriptor. The type
2088 is specified by the type information (@pxref{String Field}) for the stab.
2092 .stabs "s_typedef:t16",128,0,0,0 # @r{128 is N_LSYM}
2095 specifies that @code{s_typedef} refers to type number 16. Such stabs
2096 have symbol type @code{N_LSYM} (or @code{C_DECL} for XCOFF). (The Sun
2097 documentation mentions using @code{N_GSYM} in some cases).
2099 If you are specifying the tag name for a structure, union, or
2100 enumeration, use the @samp{T} symbol descriptor instead. I believe C is
2101 the only language with this feature.
2103 If the type is an opaque type (I believe this is a Modula-2 feature),
2104 AIX provides a type descriptor to specify it. The type descriptor is
2105 @samp{o} and is followed by a name. I don't know what the name
2106 means---is it always the same as the name of the type, or is this type
2107 descriptor used with a nameless stab (@pxref{String Field})? There
2108 optionally follows a comma followed by type information which defines
2109 the type of this type. If omitted, a semicolon is used in place of the
2110 comma and the type information, and the type is much like a generic
2111 pointer type---it has a known size but little else about it is
2125 This code generates a stab for a union tag and a stab for a union
2126 variable. Both use the @code{N_LSYM} stab type. If a union variable is
2127 scoped locally to the procedure in which it is defined, its stab is
2128 located immediately preceding the @code{N_LBRAC} for the procedure's block
2131 The stab for the union tag, however, is located preceding the code for
2132 the procedure in which it is defined. The stab type is @code{N_LSYM}. This
2133 would seem to imply that the union type is file scope, like the struct
2134 type @code{s_tag}. This is not true. The contents and position of the stab
2135 for @code{u_type} do not convey any information about its procedure local
2138 @c FIXME: phony line break. Can probably be fixed by using an example
2139 @c with fewer fields.
2142 .stabs "u_tag:T23=u4u_int:1,0,32;u_float:12,0,32;u_char:21,0,32;;",
2146 The symbol descriptor @samp{T}, following the @samp{name:} means that
2147 the stab describes an enumeration, structure, or union tag. The type
2148 descriptor @samp{u}, following the @samp{23=} of the type definition,
2149 narrows it down to a union type definition. Following the @samp{u} is
2150 the number of bytes in the union. After that is a list of union element
2151 descriptions. Their format is @var{name:type, bit offset into the
2152 union, number of bytes for the element;}.
2154 The stab for the union variable is:
2157 .stabs "an_u:23",128,0,0,-20 # @r{128 is N_LSYM}
2160 @samp{-20} specifies where the variable is stored (@pxref{Stack
2163 @node Function Types
2164 @section Function Types
2166 Various types can be defined for function variables. These types are
2167 not used in defining functions (@pxref{Procedures}); they are used for
2168 things like pointers to functions.
2170 The simple, traditional, type is type descriptor @samp{f} is followed by
2171 type information for the return type of the function, followed by a
2174 This does not deal with functions for which the number and types of the
2175 parameters are part of the type, as in Modula-2 or ANSI C. AIX provides
2176 extensions to specify these, using the @samp{f}, @samp{F}, @samp{p}, and
2177 @samp{R} type descriptors.
2179 First comes the type descriptor. If it is @samp{f} or @samp{F}, this
2180 type involves a function rather than a procedure, and the type
2181 information for the return type of the function follows, followed by a
2182 comma. Then comes the number of parameters to the function and a
2183 semicolon. Then, for each parameter, there is the name of the parameter
2184 followed by a colon (this is only present for type descriptors @samp{R}
2185 and @samp{F} which represent Pascal function or procedure parameters),
2186 type information for the parameter, a comma, 0 if passed by reference or
2187 1 if passed by value, and a semicolon. The type definition ends with a
2190 For example, this variable definition:
2197 generates the following code:
2200 .stabs "g_pf:G24=*25=f1",32,0,0,0
2201 .common _g_pf,4,"bss"
2204 The variable defines a new type, 24, which is a pointer to another new
2205 type, 25, which is a function returning @code{int}.
2208 @chapter Symbol Information in Symbol Tables
2210 This chapter describes the format of symbol table entries
2211 and how stab assembler directives map to them. It also describes the
2212 transformations that the assembler and linker make on data from stabs.
2215 * Symbol Table Format::
2216 * Transformations On Symbol Tables::
2219 @node Symbol Table Format
2220 @section Symbol Table Format
2222 Each time the assembler encounters a stab directive, it puts
2223 each field of the stab into a corresponding field in a symbol table
2224 entry of its output file. If the stab contains a string field, the
2225 symbol table entry for that stab points to a string table entry
2226 containing the string data from the stab. Assembler labels become
2227 relocatable addresses. Symbol table entries in a.out have the format:
2229 @c FIXME: should refer to external, not internal.
2231 struct internal_nlist @{
2232 unsigned long n_strx; /* index into string table of name */
2233 unsigned char n_type; /* type of symbol */
2234 unsigned char n_other; /* misc info (usually empty) */
2235 unsigned short n_desc; /* description field */
2236 bfd_vma n_value; /* value of symbol */
2240 If the stab has a string, the @code{n_strx} field holds the offset in
2241 bytes of the string within the string table. The string is terminated
2242 by a NUL character. If the stab lacks a string (for example, it was
2243 produced by a @code{.stabn} or @code{.stabd} directive), the
2244 @code{n_strx} field is zero.
2246 Symbol table entries with @code{n_type} field values greater than 0x1f
2247 originated as stabs generated by the compiler (with one random
2248 exception). The other entries were placed in the symbol table of the
2249 executable by the assembler or the linker.
2251 @node Transformations On Symbol Tables
2252 @section Transformations on Symbol Tables
2254 The linker concatenates object files and does fixups of externally
2257 You can see the transformations made on stab data by the assembler and
2258 linker by examining the symbol table after each pass of the build. To
2259 do this, use @samp{nm -ap}, which dumps the symbol table, including
2260 debugging information, unsorted. For stab entries the columns are:
2261 @var{value}, @var{other}, @var{desc}, @var{type}, @var{string}. For
2262 assembler and linker symbols, the columns are: @var{value}, @var{type},
2265 The low 5 bits of the stab type tell the linker how to relocate the
2266 value of the stab. Thus for stab types like @code{N_RSYM} and
2267 @code{N_LSYM}, where the value is an offset or a register number, the
2268 low 5 bits are @code{N_ABS}, which tells the linker not to relocate the
2271 Where the value of a stab contains an assembly language label,
2272 it is transformed by each build step. The assembler turns it into a
2273 relocatable address and the linker turns it into an absolute address.
2276 * Transformations On Static Variables::
2277 * Transformations On Global Variables::
2278 * Stab Section Transformations:: For some object file formats,
2279 things are a bit different.
2282 @node Transformations On Static Variables
2283 @subsection Transformations on Static Variables
2285 This source line defines a static variable at file scope:
2288 static int s_g_repeat
2292 The following stab describes the symbol:
2295 .stabs "s_g_repeat:S1",38,0,0,_s_g_repeat
2299 The assembler transforms the stab into this symbol table entry in the
2300 @file{.o} file. The location is expressed as a data segment offset.
2303 00000084 - 00 0000 STSYM s_g_repeat:S1
2307 In the symbol table entry from the executable, the linker has made the
2308 relocatable address absolute.
2311 0000e00c - 00 0000 STSYM s_g_repeat:S1
2314 @node Transformations On Global Variables
2315 @subsection Transformations on Global Variables
2317 Stabs for global variables do not contain location information. In
2318 this case, the debugger finds location information in the assembler or
2319 linker symbol table entry describing the variable. The source line:
2329 .stabs "g_foo:G2",32,0,0,0
2332 The variable is represented by two symbol table entries in the object
2333 file (see below). The first one originated as a stab. The second one
2334 is an external symbol. The upper case @samp{D} signifies that the
2335 @code{n_type} field of the symbol table contains 7, @code{N_DATA} with
2336 local linkage. The stab's value is zero since the value is not used for
2337 @code{N_GSYM} stabs. The value of the linker symbol is the relocatable
2338 address corresponding to the variable.
2341 00000000 - 00 0000 GSYM g_foo:G2
2346 These entries as transformed by the linker. The linker symbol table
2347 entry now holds an absolute address:
2350 00000000 - 00 0000 GSYM g_foo:G2
2355 @node Stab Section Transformations
2356 @subsection Transformations of Stabs in separate sections
2358 For object file formats using stabs in separate sections (@pxref{Stab
2359 Sections}), use @code{objdump --stabs} instead of @code{nm} to show the
2360 stabs in an object or executable file. @code{objdump} is a GNU utility;
2361 Sun does not provide any equivalent.
2363 The following example is for a stab whose value is an address is
2364 relative to the compilation unit (@pxref{ELF Linker Relocation}). For
2365 example, if the source line
2371 appears within a function, then the assembly language output from the
2377 .stabs "ld:V(0,3)",0x26,0,4,.L18-Ddata.data # @r{0x26 is N_STSYM}
2384 Because the value is formed by subtracting one symbol from another, the
2385 value is absolute, not relocatable, and so the object file contains
2388 Symnum n_type n_othr n_desc n_value n_strx String
2389 31 STSYM 0 4 00000004 680 ld:V(0,3)
2392 without any relocations, and the executable file also contains
2395 Symnum n_type n_othr n_desc n_value n_strx String
2396 31 STSYM 0 4 00000004 680 ld:V(0,3)
2400 @chapter GNU C++ Stabs
2403 * Class Names:: C++ class names are both tags and typedefs.
2404 * Nested Symbols:: C++ symbol names can be within other types.
2405 * Basic Cplusplus Types::
2408 * Methods:: Method definition
2409 * Method Type Descriptor:: The @samp{#} type descriptor
2410 * Member Type Descriptor:: The @samp{@@} type descriptor
2412 * Method Modifiers::
2415 * Virtual Base Classes::
2420 @section C++ Class Names
2422 In C++, a class name which is declared with @code{class}, @code{struct},
2423 or @code{union}, is not only a tag, as in C, but also a type name. Thus
2424 there should be stabs with both @samp{t} and @samp{T} symbol descriptors
2427 To save space, there is a special abbreviation for this case. If the
2428 @samp{T} symbol descriptor is followed by @samp{t}, then the stab
2429 defines both a type name and a tag.
2431 For example, the C++ code
2434 struct foo @{int x;@};
2437 can be represented as either
2440 .stabs "foo:T19=s4x:1,0,32;;",128,0,0,0 # @r{128 is N_LSYM}
2441 .stabs "foo:t19",128,0,0,0
2447 .stabs "foo:Tt19=s4x:1,0,32;;",128,0,0,0
2450 @node Nested Symbols
2451 @section Defining a Symbol Within Another Type
2453 In C++, a symbol (such as a type name) can be defined within another type.
2454 @c FIXME: Needs example.
2456 In stabs, this is sometimes represented by making the name of a symbol
2457 which contains @samp{::}. Such a pair of colons does not end the name
2458 of the symbol, the way a single colon would (@pxref{String Field}). I'm
2459 not sure how consistently used or well thought out this mechanism is.
2460 So that a pair of colons in this position always has this meaning,
2461 @samp{:} cannot be used as a symbol descriptor.
2463 For example, if the string for a stab is @samp{foo::bar::baz:t5=*6},
2464 then @code{foo::bar::baz} is the name of the symbol, @samp{t} is the
2465 symbol descriptor, and @samp{5=*6} is the type information.
2467 @node Basic Cplusplus Types
2468 @section Basic Types For C++
2470 << the examples that follow are based on a01.C >>
2473 C++ adds two more builtin types to the set defined for C. These are
2474 the unknown type and the vtable record type. The unknown type, type
2475 16, is defined in terms of itself like the void type.
2477 The vtable record type, type 17, is defined as a structure type and
2478 then as a structure tag. The structure has four fields: delta, index,
2479 pfn, and delta2. pfn is the function pointer.
2481 << In boilerplate $vtbl_ptr_type, what are the fields delta,
2482 index, and delta2 used for? >>
2484 This basic type is present in all C++ programs even if there are no
2485 virtual methods defined.
2488 .stabs "struct_name:sym_desc(type)type_def(17)=type_desc(struct)struct_bytes(8)
2489 elem_name(delta):type_ref(short int),bit_offset(0),field_bits(16);
2490 elem_name(index):type_ref(short int),bit_offset(16),field_bits(16);
2491 elem_name(pfn):type_def(18)=type_desc(ptr to)type_ref(void),
2492 bit_offset(32),field_bits(32);
2493 elem_name(delta2):type_def(short int);bit_offset(32),field_bits(16);;"
2498 .stabs "$vtbl_ptr_type:t17=s8
2499 delta:6,0,16;index:6,16,16;pfn:18=*15,32,32;delta2:6,32,16;;"
2504 .stabs "name:sym_dec(struct tag)type_ref($vtbl_ptr_type)",N_LSYM,NIL,NIL,NIL
2508 .stabs "$vtbl_ptr_type:T17",128,0,0,0
2511 @node Simple Classes
2512 @section Simple Class Definition
2514 The stabs describing C++ language features are an extension of the
2515 stabs describing C. Stabs representing C++ class types elaborate
2516 extensively on the stab format used to describe structure types in C.
2517 Stabs representing class type variables look just like stabs
2518 representing C language variables.
2520 Consider the following very simple class definition.
2526 int Ameth(int in, char other);
2530 The class @code{baseA} is represented by two stabs. The first stab describes
2531 the class as a structure type. The second stab describes a structure
2532 tag of the class type. Both stabs are of stab type @code{N_LSYM}. Since the
2533 stab is not located between an @code{N_FUN} and an @code{N_LBRAC} stab this indicates
2534 that the class is defined at file scope. If it were, then the @code{N_LSYM}
2535 would signify a local variable.
2537 A stab describing a C++ class type is similar in format to a stab
2538 describing a C struct, with each class member shown as a field in the
2539 structure. The part of the struct format describing fields is
2540 expanded to include extra information relevant to C++ class members.
2541 In addition, if the class has multiple base classes or virtual
2542 functions the struct format outside of the field parts is also
2545 In this simple example the field part of the C++ class stab
2546 representing member data looks just like the field part of a C struct
2547 stab. The section on protections describes how its format is
2548 sometimes extended for member data.
2550 The field part of a C++ class stab representing a member function
2551 differs substantially from the field part of a C struct stab. It
2552 still begins with @samp{name:} but then goes on to define a new type number
2553 for the member function, describe its return type, its argument types,
2554 its protection level, any qualifiers applied to the method definition,
2555 and whether the method is virtual or not. If the method is virtual
2556 then the method description goes on to give the vtable index of the
2557 method, and the type number of the first base class defining the
2560 When the field name is a method name it is followed by two colons rather
2561 than one. This is followed by a new type definition for the method.
2562 This is a number followed by an equal sign and the type of the method.
2563 Normally this will be a type declared using the @samp{#} type
2564 descriptor; see @ref{Method Type Descriptor}; static member functions
2565 are declared using the @samp{f} type descriptor instead; see
2566 @ref{Function Types}.
2568 The format of an overloaded operator method name differs from that of
2569 other methods. It is @samp{op$::@var{operator-name}.} where
2570 @var{operator-name} is the operator name such as @samp{+} or @samp{+=}.
2571 The name ends with a period, and any characters except the period can
2572 occur in the @var{operator-name} string.
2574 The next part of the method description represents the arguments to the
2575 method, preceded by a colon and ending with a semi-colon. The types of
2576 the arguments are expressed in the same way argument types are expressed
2577 in C++ name mangling. In this example an @code{int} and a @code{char}
2580 This is followed by a number, a letter, and an asterisk or period,
2581 followed by another semicolon. The number indicates the protections
2582 that apply to the member function. Here the 2 means public. The
2583 letter encodes any qualifier applied to the method definition. In
2584 this case, @samp{A} means that it is a normal function definition. The dot
2585 shows that the method is not virtual. The sections that follow
2586 elaborate further on these fields and describe the additional
2587 information present for virtual methods.
2591 .stabs "class_name:sym_desc(type)type_def(20)=type_desc(struct)struct_bytes(4)
2592 field_name(Adat):type(int),bit_offset(0),field_bits(32);
2594 method_name(Ameth)::type_def(21)=type_desc(method)return_type(int);
2595 :arg_types(int char);
2596 protection(public)qualifier(normal)virtual(no);;"
2601 .stabs "baseA:t20=s4Adat:1,0,32;Ameth::21=##1;:ic;2A.;;",128,0,0,0
2603 .stabs "class_name:sym_desc(struct tag)",N_LSYM,NIL,NIL,NIL
2605 .stabs "baseA:T20",128,0,0,0
2608 @node Class Instance
2609 @section Class Instance
2611 As shown above, describing even a simple C++ class definition is
2612 accomplished by massively extending the stab format used in C to
2613 describe structure types. However, once the class is defined, C stabs
2614 with no modifications can be used to describe class instances. The
2624 yields the following stab describing the class instance. It looks no
2625 different from a standard C stab describing a local variable.
2628 .stabs "name:type_ref(baseA)", N_LSYM, NIL, NIL, frame_ptr_offset
2632 .stabs "AbaseA:20",128,0,0,-20
2636 @section Method Definition
2638 The class definition shown above declares Ameth. The C++ source below
2643 baseA::Ameth(int in, char other)
2650 This method definition yields three stabs following the code of the
2651 method. One stab describes the method itself and following two describe
2652 its parameters. Although there is only one formal argument all methods
2653 have an implicit argument which is the @code{this} pointer. The @code{this}
2654 pointer is a pointer to the object on which the method was called. Note
2655 that the method name is mangled to encode the class name and argument
2656 types. Name mangling is described in the @sc{arm} (@cite{The Annotated
2657 C++ Reference Manual}, by Ellis and Stroustrup, @sc{isbn}
2658 0-201-51459-1); @file{gpcompare.texi} in Cygnus GCC distributions
2659 describes the differences between GNU mangling and @sc{arm}
2661 @c FIXME: Use @xref, especially if this is generally installed in the
2663 @c FIXME: This information should be in a net release, either of GCC or
2664 @c GDB. But gpcompare.texi doesn't seem to be in the FSF GCC.
2667 .stabs "name:symbol_descriptor(global function)return_type(int)",
2668 N_FUN, NIL, NIL, code_addr_of_method_start
2670 .stabs "Ameth__5baseAic:F1",36,0,0,_Ameth__5baseAic
2673 Here is the stab for the @code{this} pointer implicit argument. The
2674 name of the @code{this} pointer is always @code{this}. Type 19, the
2675 @code{this} pointer is defined as a pointer to type 20, @code{baseA},
2676 but a stab defining @code{baseA} has not yet been emitted. Since the
2677 compiler knows it will be emitted shortly, here it just outputs a cross
2678 reference to the undefined symbol, by prefixing the symbol name with
2682 .stabs "name:sym_desc(register param)type_def(19)=
2683 type_desc(ptr to)type_ref(baseA)=
2684 type_desc(cross-reference to)baseA:",N_RSYM,NIL,NIL,register_number
2686 .stabs "this:P19=*20=xsbaseA:",64,0,0,8
2689 The stab for the explicit integer argument looks just like a parameter
2690 to a C function. The last field of the stab is the offset from the
2691 argument pointer, which in most systems is the same as the frame
2695 .stabs "name:sym_desc(value parameter)type_ref(int)",
2696 N_PSYM,NIL,NIL,offset_from_arg_ptr
2698 .stabs "in:p1",160,0,0,72
2701 << The examples that follow are based on A1.C >>
2703 @node Method Type Descriptor
2704 @section The @samp{#} Type Descriptor
2706 This is used to describe a class method. This is a function which takes
2707 an extra argument as its first argument, for the @code{this} pointer.
2709 If the @samp{#} is immediately followed by another @samp{#}, the second
2710 one will be followed by the return type and a semicolon. The class and
2711 argument types are not specified, and must be determined by demangling
2712 the name of the method if it is available.
2714 Otherwise, the single @samp{#} is followed by the class type, a comma,
2715 the return type, a comma, and zero or more parameter types separated by
2716 commas. The list of arguments is terminated by a semicolon. In the
2717 debugging output generated by gcc, a final argument type of @code{void}
2718 indicates a method which does not take a variable number of arguments.
2719 If the final argument type of @code{void} does not appear, the method
2720 was declared with an ellipsis.
2722 Note that although such a type will normally be used to describe fields
2723 in structures, unions, or classes, for at least some versions of the
2724 compiler it can also be used in other contexts.
2726 @node Member Type Descriptor
2727 @section The @samp{@@} Type Descriptor
2729 The @samp{@@} type descriptor is for a member (class and variable) type.
2730 It is followed by type information for the offset basetype, a comma, and
2731 type information for the type of the field being pointed to. (FIXME:
2732 this is acknowledged to be gibberish. Can anyone say what really goes
2735 Note that there is a conflict between this and type attributes
2736 (@pxref{String Field}); both use type descriptor @samp{@@}.
2737 Fortunately, the @samp{@@} type descriptor used in this C++ sense always
2738 will be followed by a digit, @samp{(}, or @samp{-}, and type attributes
2739 never start with those things.
2742 @section Protections
2744 In the simple class definition shown above all member data and
2745 functions were publicly accessible. The example that follows
2746 contrasts public, protected and privately accessible fields and shows
2747 how these protections are encoded in C++ stabs.
2749 If the character following the @samp{@var{field-name}:} part of the
2750 string is @samp{/}, then the next character is the visibility. @samp{0}
2751 means private, @samp{1} means protected, and @samp{2} means public.
2752 Debuggers should ignore visibility characters they do not recognize, and
2753 assume a reasonable default (such as public) (GDB 4.11 does not, but
2754 this should be fixed in the next GDB release). If no visibility is
2755 specified the field is public. The visibility @samp{9} means that the
2756 field has been optimized out and is public (there is no way to specify
2757 an optimized out field with a private or protected visibility).
2758 Visibility @samp{9} is not supported by GDB 4.11; this should be fixed
2759 in the next GDB release.
2761 The following C++ source:
2775 generates the following stab:
2779 .stabs "vis:T19=s12priv:/01,0,32;prot:/12,32,8;pub:12,64,32;;",128,0,0,0
2782 @samp{vis:T19=s12} indicates that type number 19 is a 12 byte structure
2783 named @code{vis} The @code{priv} field has public visibility
2784 (@samp{/0}), type int (@samp{1}), and offset and size @samp{,0,32;}.
2785 The @code{prot} field has protected visibility (@samp{/1}), type char
2786 (@samp{2}) and offset and size @samp{,32,8;}. The @code{pub} field has
2787 type float (@samp{12}), and offset and size @samp{,64,32;}.
2789 Protections for member functions are signified by one digit embedded in
2790 the field part of the stab describing the method. The digit is 0 if
2791 private, 1 if protected and 2 if public. Consider the C++ class
2795 class all_methods @{
2797 int priv_meth(int in)@{return in;@};
2799 char protMeth(char in)@{return in;@};
2801 float pubMeth(float in)@{return in;@};
2805 It generates the following stab. The digit in question is to the left
2806 of an @samp{A} in each case. Notice also that in this case two symbol
2807 descriptors apply to the class name struct tag and struct type.
2810 .stabs "class_name:sym_desc(struct tag&type)type_def(21)=
2811 sym_desc(struct)struct_bytes(1)
2812 meth_name::type_def(22)=sym_desc(method)returning(int);
2813 :args(int);protection(private)modifier(normal)virtual(no);
2814 meth_name::type_def(23)=sym_desc(method)returning(char);
2815 :args(char);protection(protected)modifier(normal)virtual(no);
2816 meth_name::type_def(24)=sym_desc(method)returning(float);
2817 :args(float);protection(public)modifier(normal)virtual(no);;",
2822 .stabs "all_methods:Tt21=s1priv_meth::22=##1;:i;0A.;protMeth::23=##2;:c;1A.;
2823 pubMeth::24=##12;:f;2A.;;",128,0,0,0
2826 @node Method Modifiers
2827 @section Method Modifiers (@code{const}, @code{volatile}, @code{const volatile})
2831 In the class example described above all the methods have the normal
2832 modifier. This method modifier information is located just after the
2833 protection information for the method. This field has four possible
2834 character values. Normal methods use @samp{A}, const methods use
2835 @samp{B}, volatile methods use @samp{C}, and const volatile methods use
2836 @samp{D}. Consider the class definition below:
2841 int ConstMeth (int arg) const @{ return arg; @};
2842 char VolatileMeth (char arg) volatile @{ return arg; @};
2843 float ConstVolMeth (float arg) const volatile @{return arg; @};
2847 This class is described by the following stab:
2850 .stabs "class(A):sym_desc(struct)type_def(20)=type_desc(struct)struct_bytes(1)
2851 meth_name(ConstMeth)::type_def(21)sym_desc(method)
2852 returning(int);:arg(int);protection(public)modifier(const)virtual(no);
2853 meth_name(VolatileMeth)::type_def(22)=sym_desc(method)
2854 returning(char);:arg(char);protection(public)modifier(volatile)virt(no)
2855 meth_name(ConstVolMeth)::type_def(23)=sym_desc(method)
2856 returning(float);:arg(float);protection(public)modifier(const volatile)
2857 virtual(no);;", @dots{}
2861 .stabs "A:T20=s1ConstMeth::21=##1;:i;2B.;VolatileMeth::22=##2;:c;2C.;
2862 ConstVolMeth::23=##12;:f;2D.;;",128,0,0,0
2865 @node Virtual Methods
2866 @section Virtual Methods
2868 << The following examples are based on a4.C >>
2870 The presence of virtual methods in a class definition adds additional
2871 data to the class description. The extra data is appended to the
2872 description of the virtual method and to the end of the class
2873 description. Consider the class definition below:
2879 virtual int A_virt (int arg) @{ return arg; @};
2883 This results in the stab below describing class A. It defines a new
2884 type (20) which is an 8 byte structure. The first field of the class
2885 struct is @samp{Adat}, an integer, starting at structure offset 0 and
2888 The second field in the class struct is not explicitly defined by the
2889 C++ class definition but is implied by the fact that the class
2890 contains a virtual method. This field is the vtable pointer. The
2891 name of the vtable pointer field starts with @samp{$vf} and continues with a
2892 type reference to the class it is part of. In this example the type
2893 reference for class A is 20 so the name of its vtable pointer field is
2894 @samp{$vf20}, followed by the usual colon.
2896 Next there is a type definition for the vtable pointer type (21).
2897 This is in turn defined as a pointer to another new type (22).
2899 Type 22 is the vtable itself, which is defined as an array, indexed by
2900 a range of integers between 0 and 1, and whose elements are of type
2901 17. Type 17 was the vtable record type defined by the boilerplate C++
2902 type definitions, as shown earlier.
2904 The bit offset of the vtable pointer field is 32. The number of bits
2905 in the field are not specified when the field is a vtable pointer.
2907 Next is the method definition for the virtual member function @code{A_virt}.
2908 Its description starts out using the same format as the non-virtual
2909 member functions described above, except instead of a dot after the
2910 @samp{A} there is an asterisk, indicating that the function is virtual.
2911 Since is is virtual some addition information is appended to the end
2912 of the method description.
2914 The first number represents the vtable index of the method. This is a
2915 32 bit unsigned number with the high bit set, followed by a
2918 The second number is a type reference to the first base class in the
2919 inheritance hierarchy defining the virtual member function. In this
2920 case the class stab describes a base class so the virtual function is
2921 not overriding any other definition of the method. Therefore the
2922 reference is to the type number of the class that the stab is
2925 This is followed by three semi-colons. One marks the end of the
2926 current sub-section, one marks the end of the method field, and the
2927 third marks the end of the struct definition.
2929 For classes containing virtual functions the very last section of the
2930 string part of the stab holds a type reference to the first base
2931 class. This is preceded by @samp{~%} and followed by a final semi-colon.
2934 .stabs "class_name(A):type_def(20)=sym_desc(struct)struct_bytes(8)
2935 field_name(Adat):type_ref(int),bit_offset(0),field_bits(32);
2936 field_name(A virt func ptr):type_def(21)=type_desc(ptr to)type_def(22)=
2937 sym_desc(array)index_type_ref(range of int from 0 to 1);
2938 elem_type_ref(vtbl elem type),
2940 meth_name(A_virt)::typedef(23)=sym_desc(method)returning(int);
2941 :arg_type(int),protection(public)normal(yes)virtual(yes)
2942 vtable_index(1);class_first_defining(A);;;~%first_base(A);",
2946 @c FIXME: bogus line break.
2948 .stabs "A:t20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
2949 A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
2953 @section Inheritance
2955 Stabs describing C++ derived classes include additional sections that
2956 describe the inheritance hierarchy of the class. A derived class stab
2957 also encodes the number of base classes. For each base class it tells
2958 if the base class is virtual or not, and if the inheritance is private
2959 or public. It also gives the offset into the object of the portion of
2960 the object corresponding to each base class.
2962 This additional information is embedded in the class stab following the
2963 number of bytes in the struct. First the number of base classes
2964 appears bracketed by an exclamation point and a comma.
2966 Then for each base type there repeats a series: a virtual character, a
2967 visibility character, a number, a comma, another number, and a
2970 The virtual character is @samp{1} if the base class is virtual and
2971 @samp{0} if not. The visibility character is @samp{2} if the derivation
2972 is public, @samp{1} if it is protected, and @samp{0} if it is private.
2973 Debuggers should ignore virtual or visibility characters they do not
2974 recognize, and assume a reasonable default (such as public and
2975 non-virtual) (GDB 4.11 does not, but this should be fixed in the next
2978 The number following the virtual and visibility characters is the offset
2979 from the start of the object to the part of the object pertaining to the
2982 After the comma, the second number is a type_descriptor for the base
2983 type. Finally a semi-colon ends the series, which repeats for each
2986 The source below defines three base classes @code{A}, @code{B}, and
2987 @code{C} and the derived class @code{D}.
2994 virtual int A_virt (int arg) @{ return arg; @};
3000 virtual int B_virt (int arg) @{return arg; @};
3006 virtual int C_virt (int arg) @{return arg; @};
3009 class D : A, virtual B, public C @{
3012 virtual int A_virt (int arg ) @{ return arg+1; @};
3013 virtual int B_virt (int arg) @{ return arg+2; @};
3014 virtual int C_virt (int arg) @{ return arg+3; @};
3015 virtual int D_virt (int arg) @{ return arg; @};
3019 Class stabs similar to the ones described earlier are generated for
3022 @c FIXME!!! the linebreaks in the following example probably make the
3023 @c examples literally unusable, but I don't know any other way to get
3024 @c them on the page.
3025 @c One solution would be to put some of the type definitions into
3026 @c separate stabs, even if that's not exactly what the compiler actually
3029 .stabs "A:T20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
3030 A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
3032 .stabs "B:Tt25=s8Bdat:1,0,32;$vf25:21,32;B_virt::26=##1;
3033 :i;2A*-2147483647;25;;;~%25;",128,0,0,0
3035 .stabs "C:Tt28=s8Cdat:1,0,32;$vf28:21,32;C_virt::29=##1;
3036 :i;2A*-2147483647;28;;;~%28;",128,0,0,0
3039 In the stab describing derived class @code{D} below, the information about
3040 the derivation of this class is encoded as follows.
3043 .stabs "derived_class_name:symbol_descriptors(struct tag&type)=
3044 type_descriptor(struct)struct_bytes(32)!num_bases(3),
3045 base_virtual(no)inheritance_public(no)base_offset(0),
3046 base_class_type_ref(A);
3047 base_virtual(yes)inheritance_public(no)base_offset(NIL),
3048 base_class_type_ref(B);
3049 base_virtual(no)inheritance_public(yes)base_offset(64),
3050 base_class_type_ref(C); @dots{}
3053 @c FIXME! fake linebreaks.
3055 .stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:
3056 1,160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt:
3057 :32:i;2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;
3058 28;;D_virt::32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
3061 @node Virtual Base Classes
3062 @section Virtual Base Classes
3064 A derived class object consists of a concatenation in memory of the data
3065 areas defined by each base class, starting with the leftmost and ending
3066 with the rightmost in the list of base classes. The exception to this
3067 rule is for virtual inheritance. In the example above, class @code{D}
3068 inherits virtually from base class @code{B}. This means that an
3069 instance of a @code{D} object will not contain its own @code{B} part but
3070 merely a pointer to a @code{B} part, known as a virtual base pointer.
3072 In a derived class stab, the base offset part of the derivation
3073 information, described above, shows how the base class parts are
3074 ordered. The base offset for a virtual base class is always given as 0.
3075 Notice that the base offset for @code{B} is given as 0 even though
3076 @code{B} is not the first base class. The first base class @code{A}
3079 The field information part of the stab for class @code{D} describes the field
3080 which is the pointer to the virtual base class @code{B}. The vbase pointer
3081 name is @samp{$vb} followed by a type reference to the virtual base class.
3082 Since the type id for @code{B} in this example is 25, the vbase pointer name
3085 @c FIXME!! fake linebreaks below
3087 .stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:1,
3088 160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt::32:i;
3089 2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;28;;D_virt:
3090 :32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
3093 Following the name and a semicolon is a type reference describing the
3094 type of the virtual base class pointer, in this case 24. Type 24 was
3095 defined earlier as the type of the @code{B} class @code{this} pointer. The
3096 @code{this} pointer for a class is a pointer to the class type.
3099 .stabs "this:P24=*25=xsB:",64,0,0,8
3102 Finally the field offset part of the vbase pointer field description
3103 shows that the vbase pointer is the first field in the @code{D} object,
3104 before any data fields defined by the class. The layout of a @code{D}
3105 class object is a follows, @code{Adat} at 0, the vtable pointer for
3106 @code{A} at 32, @code{Cdat} at 64, the vtable pointer for C at 96, the
3107 virtual base pointer for @code{B} at 128, and @code{Ddat} at 160.
3110 @node Static Members
3111 @section Static Members
3113 The data area for a class is a concatenation of the space used by the
3114 data members of the class. If the class has virtual methods, a vtable
3115 pointer follows the class data. The field offset part of each field
3116 description in the class stab shows this ordering.
3118 << How is this reflected in stabs? See Cygnus bug #677 for some info. >>
3121 @appendix Table of Stab Types
3123 The following are all the possible values for the stab type field, for
3124 a.out files, in numeric order. This does not apply to XCOFF, but
3125 it does apply to stabs in sections (@pxref{Stab Sections}). Stabs in
3126 ECOFF use these values but add 0x8f300 to distinguish them from non-stab
3129 The symbolic names are defined in the file @file{include/aout/stabs.def}.
3132 * Non-Stab Symbol Types:: Types from 0 to 0x1f
3133 * Stab Symbol Types:: Types from 0x20 to 0xff
3136 @node Non-Stab Symbol Types
3137 @appendixsec Non-Stab Symbol Types
3139 The following types are used by the linker and assembler, not by stab
3140 directives. Since this document does not attempt to describe aspects of
3141 object file format other than the debugging format, no details are
3144 @c Try to get most of these to fit on a single line.
3154 File scope absolute symbol
3156 @item 0x3 N_ABS | N_EXT
3157 External absolute symbol
3160 File scope text symbol
3162 @item 0x5 N_TEXT | N_EXT
3163 External text symbol
3166 File scope data symbol
3168 @item 0x7 N_DATA | N_EXT
3169 External data symbol
3172 File scope BSS symbol
3174 @item 0x9 N_BSS | N_EXT
3178 Same as @code{N_FN}, for Sequent compilers
3181 Symbol is indirected to another symbol
3184 Common---visible after shared library dynamic link
3187 @itemx 0x15 N_SETA | N_EXT
3188 Absolute set element
3191 @itemx 0x17 N_SETT | N_EXT
3192 Text segment set element
3195 @itemx 0x19 N_SETD | N_EXT
3196 Data segment set element
3199 @itemx 0x1b N_SETB | N_EXT
3200 BSS segment set element
3203 @itemx 0x1d N_SETV | N_EXT
3204 Pointer to set vector
3206 @item 0x1e N_WARNING
3207 Print a warning message during linking
3210 File name of a @file{.o} file
3213 @node Stab Symbol Types
3214 @appendixsec Stab Symbol Types
3216 The following symbol types indicate that this is a stab. This is the
3217 full list of stab numbers, including stab types that are used in
3218 languages other than C.
3222 Global symbol; see @ref{Global Variables}.
3225 Function name (for BSD Fortran); see @ref{Procedures}.
3228 Function name (@pxref{Procedures}) or text segment variable
3232 Data segment file-scope variable; see @ref{Statics}.
3235 BSS segment file-scope variable; see @ref{Statics}.
3238 Name of main routine; see @ref{Main Program}.
3241 Variable in @code{.rodata} section; see @ref{Statics}.
3244 Global symbol (for Pascal); see @ref{N_PC}.
3247 Number of symbols (according to Ultrix V4.0); see @ref{N_NSYMS}.
3250 No DST map; see @ref{N_NOMAP}.
3252 @c FIXME: describe this solaris feature in the body of the text (see
3253 @c comments in include/aout/stab.def).
3255 Object file (Solaris2).
3257 @c See include/aout/stab.def for (a little) more info.
3259 Debugger options (Solaris2).
3262 Register variable; see @ref{Register Variables}.
3265 Modula-2 compilation unit; see @ref{N_M2C}.
3268 Line number in text segment; see @ref{Line Numbers}.
3271 Line number in data segment; see @ref{Line Numbers}.
3274 Line number in bss segment; see @ref{Line Numbers}.
3277 Sun source code browser, path to @file{.cb} file; see @ref{N_BROWS}.
3280 GNU Modula2 definition module dependency; see @ref{N_DEFD}.
3283 Function start/body/end line numbers (Solaris2).
3286 GNU C++ exception variable; see @ref{N_EHDECL}.
3289 Modula2 info "for imc" (according to Ultrix V4.0); see @ref{N_MOD2}.
3292 GNU C++ @code{catch} clause; see @ref{N_CATCH}.
3295 Structure of union element; see @ref{N_SSYM}.
3298 Last stab for module (Solaris2).
3301 Path and name of source file; see @ref{Source Files}.
3304 Stack variable (@pxref{Stack Variables}) or type (@pxref{Typedefs}).
3307 Beginning of an include file (Sun only); see @ref{Include Files}.
3310 Name of include file; see @ref{Include Files}.
3313 Parameter variable; see @ref{Parameters}.
3316 End of an include file; see @ref{Include Files}.
3319 Alternate entry point; see @ref{Alternate Entry Points}.
3322 Beginning of a lexical block; see @ref{Block Structure}.
3325 Place holder for a deleted include file; see @ref{Include Files}.
3328 Modula2 scope information (Sun linker); see @ref{N_SCOPE}.
3331 End of a lexical block; see @ref{Block Structure}.
3334 Begin named common block; see @ref{Common Blocks}.
3337 End named common block; see @ref{Common Blocks}.
3340 Member of a common block; see @ref{Common Blocks}.
3342 @c FIXME: How does this really work? Move it to main body of document.
3344 Pascal @code{with} statement: type,,0,0,offset (Solaris2).
3347 Gould non-base registers; see @ref{Gould}.
3350 Gould non-base registers; see @ref{Gould}.
3353 Gould non-base registers; see @ref{Gould}.
3356 Gould non-base registers; see @ref{Gould}.
3359 Gould non-base registers; see @ref{Gould}.
3362 @c Restore the default table indent
3367 @node Symbol Descriptors
3368 @appendix Table of Symbol Descriptors
3370 The symbol descriptor is the character which follows the colon in many
3371 stabs, and which tells what kind of stab it is. @xref{String Field},
3372 for more information about their use.
3374 @c Please keep this alphabetical
3376 @c In TeX, this looks great, digit is in italics. But makeinfo insists
3377 @c on putting it in `', not realizing that @var should override @code.
3378 @c I don't know of any way to make makeinfo do the right thing. Seems
3379 @c like a makeinfo bug to me.
3383 Variable on the stack; see @ref{Stack Variables}.
3386 C++ nested symbol; see @xref{Nested Symbols}.
3389 Parameter passed by reference in register; see @ref{Reference Parameters}.
3392 Based variable; see @ref{Based Variables}.
3395 Constant; see @ref{Constants}.
3398 Conformant array bound (Pascal, maybe other languages); @ref{Conformant
3399 Arrays}. Name of a caught exception (GNU C++). These can be
3400 distinguished because the latter uses @code{N_CATCH} and the former uses
3401 another symbol type.
3404 Floating point register variable; see @ref{Register Variables}.
3407 Parameter in floating point register; see @ref{Register Parameters}.
3410 File scope function; see @ref{Procedures}.
3413 Global function; see @ref{Procedures}.
3416 Global variable; see @ref{Global Variables}.
3419 @xref{Register Parameters}.
3422 Internal (nested) procedure; see @ref{Nested Procedures}.
3425 Internal (nested) function; see @ref{Nested Procedures}.
3428 Label name (documented by AIX, no further information known).
3431 Module; see @ref{Procedures}.
3434 Argument list parameter; see @ref{Parameters}.
3440 Fortran Function parameter; see @ref{Parameters}.
3443 Unfortunately, three separate meanings have been independently invented
3444 for this symbol descriptor. At least the GNU and Sun uses can be
3445 distinguished by the symbol type. Global Procedure (AIX) (symbol type
3446 used unknown); see @ref{Procedures}. Register parameter (GNU) (symbol
3447 type @code{N_PSYM}); see @ref{Parameters}. Prototype of function
3448 referenced by this file (Sun @code{acc}) (symbol type @code{N_FUN}).
3451 Static Procedure; see @ref{Procedures}.
3454 Register parameter; see @ref{Register Parameters}.
3457 Register variable; see @ref{Register Variables}.
3460 File scope variable; see @ref{Statics}.
3463 Local variable (OS9000).
3466 Type name; see @ref{Typedefs}.
3469 Enumeration, structure, or union tag; see @ref{Typedefs}.
3472 Parameter passed by reference; see @ref{Reference Parameters}.
3475 Procedure scope static variable; see @ref{Statics}.
3478 Conformant array; see @ref{Conformant Arrays}.
3481 Function return variable; see @ref{Parameters}.
3484 @node Type Descriptors
3485 @appendix Table of Type Descriptors
3487 The type descriptor is the character which follows the type number and
3488 an equals sign. It specifies what kind of type is being defined.
3489 @xref{String Field}, for more information about their use.
3494 Type reference; see @ref{String Field}.
3497 Reference to builtin type; see @ref{Negative Type Numbers}.
3500 Method (C++); see @ref{Method Type Descriptor}.
3503 Pointer; see @ref{Miscellaneous Types}.
3509 Type Attributes (AIX); see @ref{String Field}. Member (class and variable)
3510 type (GNU C++); see @ref{Member Type Descriptor}.
3513 Array; see @ref{Arrays}.
3516 Open array; see @ref{Arrays}.
3519 Pascal space type (AIX); see @ref{Miscellaneous Types}. Builtin integer
3520 type (Sun); see @ref{Builtin Type Descriptors}. Const and volatile
3521 qualified type (OS9000).
3524 Volatile-qualified type; see @ref{Miscellaneous Types}.
3527 Complex builtin type (AIX); see @ref{Builtin Type Descriptors}.
3528 Const-qualified type (OS9000).
3531 COBOL Picture type. See AIX documentation for details.
3534 File type; see @ref{Miscellaneous Types}.
3537 N-dimensional dynamic array; see @ref{Arrays}.
3540 Enumeration type; see @ref{Enumerations}.
3543 N-dimensional subarray; see @ref{Arrays}.
3546 Function type; see @ref{Function Types}.
3549 Pascal function parameter; see @ref{Function Types}
3552 Builtin floating point type; see @ref{Builtin Type Descriptors}.
3555 COBOL Group. See AIX documentation for details.
3558 Imported type (AIX); see @ref{Cross-References}. Volatile-qualified
3562 Const-qualified type; see @ref{Miscellaneous Types}.
3565 COBOL File Descriptor. See AIX documentation for details.
3568 Multiple instance type; see @ref{Miscellaneous Types}.
3571 String type; see @ref{Strings}.
3574 Stringptr; see @ref{Strings}.
3577 Opaque type; see @ref{Typedefs}.
3580 Procedure; see @ref{Function Types}.
3583 Packed array; see @ref{Arrays}.
3586 Range type; see @ref{Subranges}.
3589 Builtin floating type; see @ref{Builtin Type Descriptors} (Sun). Pascal
3590 subroutine parameter; see @ref{Function Types} (AIX). Detecting this
3591 conflict is possible with careful parsing (hint: a Pascal subroutine
3592 parameter type will always contain a comma, and a builtin type
3593 descriptor never will).
3596 Structure type; see @ref{Structures}.
3599 Set type; see @ref{Miscellaneous Types}.
3602 Union; see @ref{Unions}.
3605 Variant record. This is a Pascal and Modula-2 feature which is like a
3606 union within a struct in C. See AIX documentation for details.
3609 Wide character; see @ref{Builtin Type Descriptors}.
3612 Cross-reference; see @ref{Cross-References}.
3615 Used by IBM's xlC C++ compiler (for structures, I think).
3618 gstring; see @ref{Strings}.
3621 @node Expanded Reference
3622 @appendix Expanded Reference by Stab Type
3624 @c FIXME: This appendix should go away; see N_PSYM or N_SO for an example.
3626 For a full list of stab types, and cross-references to where they are
3627 described, see @ref{Stab Types}. This appendix just covers certain
3628 stabs which are not yet described in the main body of this document;
3629 eventually the information will all be in one place.
3633 The first line is the symbol type (see @file{include/aout/stab.def}).
3635 The second line describes the language constructs the symbol type
3638 The third line is the stab format with the significant stab fields
3639 named and the rest NIL.
3641 Subsequent lines expand upon the meaning and possible values for each
3642 significant stab field.
3644 Finally, any further information.
3647 * N_PC:: Pascal global symbol
3648 * N_NSYMS:: Number of symbols
3649 * N_NOMAP:: No DST map
3650 * N_M2C:: Modula-2 compilation unit
3651 * N_BROWS:: Path to .cb file for Sun source code browser
3652 * N_DEFD:: GNU Modula2 definition module dependency
3653 * N_EHDECL:: GNU C++ exception variable
3654 * N_MOD2:: Modula2 information "for imc"
3655 * N_CATCH:: GNU C++ "catch" clause
3656 * N_SSYM:: Structure or union element
3657 * N_SCOPE:: Modula2 scope information (Sun only)
3658 * Gould:: non-base register symbols used on Gould systems
3659 * N_LENG:: Length of preceding entry
3665 @deffn @code{.stabs} N_PC
3667 Global symbol (for Pascal).
3670 "name" -> "symbol_name" <<?>>
3671 value -> supposedly the line number (stab.def is skeptical)
3675 @file{stabdump.c} says:
3677 global pascal symbol: name,,0,subtype,line
3685 @deffn @code{.stabn} N_NSYMS
3687 Number of symbols (according to Ultrix V4.0).
3690 0, files,,funcs,lines (stab.def)
3697 @deffn @code{.stabs} N_NOMAP
3699 No DST map for symbol (according to Ultrix V4.0). I think this means a
3700 variable has been optimized out.
3703 name, ,0,type,ignored (stab.def)
3710 @deffn @code{.stabs} N_M2C
3712 Modula-2 compilation unit.
3715 "string" -> "unit_name,unit_time_stamp[,code_time_stamp]"
3717 value -> 0 (main unit)
3721 See @cite{Dbx and Dbxtool Interfaces}, 2nd edition, by Sun, 1988, for
3729 @deffn @code{.stabs} N_BROWS
3731 Sun source code browser, path to @file{.cb} file
3734 "path to associated @file{.cb} file"
3736 Note: N_BROWS has the same value as N_BSLINE.
3742 @deffn @code{.stabn} N_DEFD
3744 GNU Modula2 definition module dependency.
3746 GNU Modula-2 definition module dependency. The value is the
3747 modification time of the definition file. The other field is non-zero
3748 if it is imported with the GNU M2 keyword @code{%INITIALIZE}. Perhaps
3749 @code{N_M2C} can be used if there are enough empty fields?
3755 @deffn @code{.stabs} N_EHDECL
3757 GNU C++ exception variable <<?>>.
3759 "@var{string} is variable name"
3761 Note: conflicts with @code{N_MOD2}.
3767 @deffn @code{.stab?} N_MOD2
3769 Modula2 info "for imc" (according to Ultrix V4.0)
3771 Note: conflicts with @code{N_EHDECL} <<?>>
3777 @deffn @code{.stabn} N_CATCH
3779 GNU C++ @code{catch} clause
3781 GNU C++ @code{catch} clause. The value is its address. The desc field
3782 is nonzero if this entry is immediately followed by a @code{CAUGHT} stab
3783 saying what exception was caught. Multiple @code{CAUGHT} stabs means
3784 that multiple exceptions can be caught here. If desc is 0, it means all
3785 exceptions are caught here.
3791 @deffn @code{.stabn} N_SSYM
3793 Structure or union element.
3795 The value is the offset in the structure.
3797 <<?looking at structs and unions in C I didn't see these>>
3803 @deffn @code{.stab?} N_SCOPE
3805 Modula2 scope information (Sun linker)
3810 @section Non-base registers on Gould systems
3812 @deffn @code{.stab?} N_NBTEXT
3813 @deffnx @code{.stab?} N_NBDATA
3814 @deffnx @code{.stab?} N_NBBSS
3815 @deffnx @code{.stab?} N_NBSTS
3816 @deffnx @code{.stab?} N_NBLCS
3822 These are used on Gould systems for non-base registers syms.
3824 However, the following values are not the values used by Gould; they are
3825 the values which GNU has been documenting for these values for a long
3826 time, without actually checking what Gould uses. I include these values
3827 only because perhaps some someone actually did something with the GNU
3828 information (I hope not, why GNU knowingly assigned wrong values to
3829 these in the header file is a complete mystery to me).
3832 240 0xf0 N_NBTEXT ??
3833 242 0xf2 N_NBDATA ??
3843 @deffn @code{.stabn} N_LENG
3845 Second symbol entry containing a length-value for the preceding entry.
3846 The value is the length.
3850 @appendix Questions and Anomalies
3854 @c I think this is changed in GCC 2.4.5 to put the line number there.
3855 For GNU C stabs defining local and global variables (@code{N_LSYM} and
3856 @code{N_GSYM}), the desc field is supposed to contain the source
3857 line number on which the variable is defined. In reality the desc
3858 field is always 0. (This behavior is defined in @file{dbxout.c} and
3859 putting a line number in desc is controlled by @samp{#ifdef
3860 WINNING_GDB}, which defaults to false). GDB supposedly uses this
3861 information if you say @samp{list @var{var}}. In reality, @var{var} can
3862 be a variable defined in the program and GDB says @samp{function
3863 @var{var} not defined}.
3866 In GNU C stabs, there seems to be no way to differentiate tag types:
3867 structures, unions, and enums (symbol descriptor @samp{T}) and typedefs
3868 (symbol descriptor @samp{t}) defined at file scope from types defined locally
3869 to a procedure or other more local scope. They all use the @code{N_LSYM}
3870 stab type. Types defined at procedure scope are emitted after the
3871 @code{N_RBRAC} of the preceding function and before the code of the
3872 procedure in which they are defined. This is exactly the same as
3873 types defined in the source file between the two procedure bodies.
3874 GDB over-compensates by placing all types in block #1, the block for
3875 symbols of file scope. This is true for default, @samp{-ansi} and
3876 @samp{-traditional} compiler options. (Bugs gcc/1063, gdb/1066.)
3879 What ends the procedure scope? Is it the proc block's @code{N_RBRAC} or the
3880 next @code{N_FUN}? (I believe its the first.)
3884 @appendix Using Stabs in Their Own Sections
3886 Many object file formats allow tools to create object files with custom
3887 sections containing any arbitrary data. For any such object file
3888 format, stabs can be embedded in special sections. This is how stabs
3889 are used with ELF and SOM, and aside from ECOFF and XCOFF, is how stabs
3893 * Stab Section Basics:: How to embed stabs in sections
3894 * ELF Linker Relocation:: Sun ELF hacks
3897 @node Stab Section Basics
3898 @appendixsec How to Embed Stabs in Sections
3900 The assembler creates two custom sections, a section named @code{.stab}
3901 which contains an array of fixed length structures, one struct per stab,
3902 and a section named @code{.stabstr} containing all the variable length
3903 strings that are referenced by stabs in the @code{.stab} section. The
3904 byte order of the stabs binary data depends on the object file format.
3905 For ELF, it matches the byte order of the ELF file itself, as determined
3906 from the @code{EI_DATA} field in the @code{e_ident} member of the ELF
3907 header. For SOM, it is always big-endian (is this true??? FIXME). For
3908 COFF, it matches the byte order of the COFF headers. The meaning of the
3909 fields is the same as for a.out (@pxref{Symbol Table Format}), except
3910 that the @code{n_strx} field is relative to the strings for the current
3911 compilation unit (which can be found using the synthetic N_UNDF stab
3912 described below), rather than the entire string table.
3914 The first stab in the @code{.stab} section for each compilation unit is
3915 synthetic, generated entirely by the assembler, with no corresponding
3916 @code{.stab} directive as input to the assembler. This stab contains
3917 the following fields:
3921 Offset in the @code{.stabstr} section to the source filename.
3927 Unused field, always zero.
3928 This may eventually be used to hold overflows from the count in
3929 the @code{n_desc} field.
3932 Count of upcoming symbols, i.e., the number of remaining stabs for this
3936 Size of the string table fragment associated with this source file, in
3940 The @code{.stabstr} section always starts with a null byte (so that string
3941 offsets of zero reference a null string), followed by random length strings,
3942 each of which is null byte terminated.
3944 The ELF section header for the @code{.stab} section has its
3945 @code{sh_link} member set to the section number of the @code{.stabstr}
3946 section, and the @code{.stabstr} section has its ELF section
3947 header @code{sh_type} member set to @code{SHT_STRTAB} to mark it as a
3948 string table. SOM and COFF have no way of linking the sections together
3949 or marking them as string tables.
3951 For COFF, the @code{.stab} and @code{.stabstr} sections may be simply
3952 concatenated by the linker. GDB then uses the @code{n_desc} fields to
3953 figure out the extent of the original sections. Similarly, the
3954 @code{n_value} fields of the header symbols are added together in order
3955 to get the actual position of the strings in a desired @code{.stabstr}
3956 section. Although this design obviates any need for the linker to
3957 relocate or otherwise manipulate @code{.stab} and @code{.stabstr}
3958 sections, it also requires some care to ensure that the offsets are
3959 calculated correctly. For instance, if the linker were to pad in
3960 between the @code{.stabstr} sections before concatenating, then the
3961 offsets to strings in the middle of the executable's @code{.stabstr}
3962 section would be wrong.
3964 The GNU linker is able to optimize stabs information by merging
3965 duplicate strings and removing duplicate header file information
3966 (@pxref{Include Files}). When some versions of the GNU linker optimize
3967 stabs in sections, they remove the leading @code{N_UNDF} symbol and
3968 arranges for all the @code{n_strx} fields to be relative to the start of
3969 the @code{.stabstr} section.
3971 @node ELF Linker Relocation
3972 @appendixsec Having the Linker Relocate Stabs in ELF
3974 This section describes some Sun hacks for Stabs in ELF; it does not
3975 apply to COFF or SOM.
3977 To keep linking fast, you don't want the linker to have to relocate very
3978 many stabs. Making sure this is done for @code{N_SLINE},
3979 @code{N_RBRAC}, and @code{N_LBRAC} stabs is the most important thing
3980 (see the descriptions of those stabs for more information). But Sun's
3981 stabs in ELF has taken this further, to make all addresses in the
3982 @code{n_value} field (functions and static variables) relative to the
3983 source file. For the @code{N_SO} symbol itself, Sun simply omits the
3984 address. To find the address of each section corresponding to a given
3985 source file, the compiler puts out symbols giving the address of each
3986 section for a given source file. Since these are ELF (not stab)
3987 symbols, the linker relocates them correctly without having to touch the
3988 stabs section. They are named @code{Bbss.bss} for the bss section,
3989 @code{Ddata.data} for the data section, and @code{Drodata.rodata} for
3990 the rodata section. For the text section, there is no such symbol (but
3991 there should be, see below). For an example of how these symbols work,
3992 @xref{Stab Section Transformations}. GCC does not provide these symbols;
3993 it instead relies on the stabs getting relocated. Thus addresses which
3994 would normally be relative to @code{Bbss.bss}, etc., are already
3995 relocated. The Sun linker provided with Solaris 2.2 and earlier
3996 relocates stabs using normal ELF relocation information, as it would do
3997 for any section. Sun has been threatening to kludge their linker to not
3998 do this (to speed up linking), even though the correct way to avoid
3999 having the linker do these relocations is to have the compiler no longer
4000 output relocatable values. Last I heard they had been talked out of the
4001 linker kludge. See Sun point patch 101052-01 and Sun bug 1142109. With
4002 the Sun compiler this affects @samp{S} symbol descriptor stabs
4003 (@pxref{Statics}) and functions (@pxref{Procedures}). In the latter
4004 case, to adopt the clean solution (making the value of the stab relative
4005 to the start of the compilation unit), it would be necessary to invent a
4006 @code{Ttext.text} symbol, analogous to the @code{Bbss.bss}, etc.,
4007 symbols. I recommend this rather than using a zero value and getting
4008 the address from the ELF symbols.
4010 Finding the correct @code{Bbss.bss}, etc., symbol is difficult, because
4011 the linker simply concatenates the @code{.stab} sections from each
4012 @file{.o} file without including any information about which part of a
4013 @code{.stab} section comes from which @file{.o} file. The way GDB does
4014 this is to look for an ELF @code{STT_FILE} symbol which has the same
4015 name as the last component of the file name from the @code{N_SO} symbol
4016 in the stabs (for example, if the file name is @file{../../gdb/main.c},
4017 it looks for an ELF @code{STT_FILE} symbol named @code{main.c}). This
4018 loses if different files have the same name (they could be in different
4019 directories, a library could have been copied from one system to
4020 another, etc.). It would be much cleaner to have the @code{Bbss.bss}
4021 symbols in the stabs themselves. Having the linker relocate them there
4022 is no more work than having the linker relocate ELF symbols, and it
4023 solves the problem of having to associate the ELF and stab symbols.
4024 However, no one has yet designed or implemented such a scheme.
4026 @node Symbol Types Index
4027 @unnumbered Symbol Types Index
4031 @c TeX can handle the contents at the start but makeinfo 3.12 can not