2 @setfilename stabs.info
9 * Stabs: (stabs). The "stabs" debugging information format.
15 This document describes the stabs debugging symbol tables.
17 Copyright 1992, 93, 94, 95, 97, 1998 Free Software Foundation, Inc.
18 Contributed by Cygnus Support. Written by Julia Menapace, Jim Kingdon,
21 Permission is granted to make and distribute verbatim copies of
22 this manual provided the copyright notice and this permission notice
23 are preserved on all copies.
26 Permission is granted to process this file through Tex and print the
27 results, provided the printed document carries copying permission
28 notice identical to this one except for the removal of this paragraph
29 (this paragraph not being relevant to the printed manual).
32 Permission is granted to copy or distribute modified versions of this
33 manual under the terms of the GPL (for which purpose this text may be
34 regarded as a program in the language TeX).
37 @setchapternewpage odd
40 @title The ``stabs'' debug format
41 @author Julia Menapace, Jim Kingdon, David MacKenzie
42 @author Cygnus Support
45 \def\$#1${{#1}} % Kluge: collect RCS revision info without $...$
46 \xdef\manvers{\$Revision$} % For use in headers, footers too
48 \hfill Cygnus Support\par
50 \hfill \TeX{}info \texinfoversion\par
54 @vskip 0pt plus 1filll
55 Copyright @copyright{} 1992, 93, 94, 95, 97, 1998 Free Software Foundation, Inc.
56 Contributed by Cygnus Support.
58 Permission is granted to make and distribute verbatim copies of
59 this manual provided the copyright notice and this permission notice
60 are preserved on all copies.
66 @top The "stabs" representation of debugging information
68 This document describes the stabs debugging format.
71 * Overview:: Overview of stabs
72 * Program Structure:: Encoding of the structure of the program
73 * Constants:: Constants
75 * Types:: Type definitions
76 * Symbol Tables:: Symbol information in symbol tables
77 * Cplusplus:: Stabs specific to C++
78 * Stab Types:: Symbol types in a.out files
79 * Symbol Descriptors:: Table of symbol descriptors
80 * Type Descriptors:: Table of type descriptors
81 * Expanded Reference:: Reference information by stab type
82 * Questions:: Questions and anomalies
83 * Stab Sections:: In some object file formats, stabs are
85 * Symbol Types Index:: Index of symbolic stab symbol type names.
89 @c TeX can handle the contents at the start but makeinfo 3.12 can not
95 @chapter Overview of Stabs
97 @dfn{Stabs} refers to a format for information that describes a program
98 to a debugger. This format was apparently invented by
100 the University of California at Berkeley, for the @code{pdx} Pascal
101 debugger; the format has spread widely since then.
103 This document is one of the few published sources of documentation on
104 stabs. It is believed to be comprehensive for stabs used by C. The
105 lists of symbol descriptors (@pxref{Symbol Descriptors}) and type
106 descriptors (@pxref{Type Descriptors}) are believed to be completely
107 comprehensive. Stabs for COBOL-specific features and for variant
108 records (used by Pascal and Modula-2) are poorly documented here.
110 @c FIXME: Need to document all OS9000 stuff in GDB; see all references
111 @c to os9k_stabs in stabsread.c.
113 Other sources of information on stabs are @cite{Dbx and Dbxtool
114 Interfaces}, 2nd edition, by Sun, 1988, and @cite{AIX Version 3.2 Files
115 Reference}, Fourth Edition, September 1992, "dbx Stabstring Grammar" in
116 the a.out section, page 2-31. This document is believed to incorporate
117 the information from those two sources except where it explicitly directs
118 you to them for more information.
121 * Flow:: Overview of debugging information flow
122 * Stabs Format:: Overview of stab format
123 * String Field:: The string field
124 * C Example:: A simple example in C source
125 * Assembly Code:: The simple example at the assembly level
129 @section Overview of Debugging Information Flow
131 The GNU C compiler compiles C source in a @file{.c} file into assembly
132 language in a @file{.s} file, which the assembler translates into
133 a @file{.o} file, which the linker combines with other @file{.o} files and
134 libraries to produce an executable file.
136 With the @samp{-g} option, GCC puts in the @file{.s} file additional
137 debugging information, which is slightly transformed by the assembler
138 and linker, and carried through into the final executable. This
139 debugging information describes features of the source file like line
140 numbers, the types and scopes of variables, and function names,
141 parameters, and scopes.
143 For some object file formats, the debugging information is encapsulated
144 in assembler directives known collectively as @dfn{stab} (symbol table)
145 directives, which are interspersed with the generated code. Stabs are
146 the native format for debugging information in the a.out and XCOFF
147 object file formats. The GNU tools can also emit stabs in the COFF and
148 ECOFF object file formats.
150 The assembler adds the information from stabs to the symbol information
151 it places by default in the symbol table and the string table of the
152 @file{.o} file it is building. The linker consolidates the @file{.o}
153 files into one executable file, with one symbol table and one string
154 table. Debuggers use the symbol and string tables in the executable as
155 a source of debugging information about the program.
158 @section Overview of Stab Format
160 There are three overall formats for stab assembler directives,
161 differentiated by the first word of the stab. The name of the directive
162 describes which combination of four possible data fields follows. It is
163 either @code{.stabs} (string), @code{.stabn} (number), or @code{.stabd}
164 (dot). IBM's XCOFF assembler uses @code{.stabx} (and some other
165 directives such as @code{.file} and @code{.bi}) instead of
166 @code{.stabs}, @code{.stabn} or @code{.stabd}.
168 The overall format of each class of stab is:
171 .stabs "@var{string}",@var{type},@var{other},@var{desc},@var{value}
172 .stabn @var{type},@var{other},@var{desc},@var{value}
173 .stabd @var{type},@var{other},@var{desc}
174 .stabx "@var{string}",@var{value},@var{type},@var{sdb-type}
177 @c what is the correct term for "current file location"? My AIX
178 @c assembler manual calls it "the value of the current location counter".
179 For @code{.stabn} and @code{.stabd}, there is no @var{string} (the
180 @code{n_strx} field is zero; see @ref{Symbol Tables}). For
181 @code{.stabd}, the @var{value} field is implicit and has the value of
182 the current file location. For @code{.stabx}, the @var{sdb-type} field
183 is unused for stabs and can always be set to zero. The @var{other}
184 field is almost always unused and can be set to zero.
186 The number in the @var{type} field gives some basic information about
187 which type of stab this is (or whether it @emph{is} a stab, as opposed
188 to an ordinary symbol). Each valid type number defines a different stab
189 type; further, the stab type defines the exact interpretation of, and
190 possible values for, any remaining @var{string}, @var{desc}, or
191 @var{value} fields present in the stab. @xref{Stab Types}, for a list
192 in numeric order of the valid @var{type} field values for stab directives.
195 @section The String Field
197 For most stabs the string field holds the meat of the
198 debugging information. The flexible nature of this field
199 is what makes stabs extensible. For some stab types the string field
200 contains only a name. For other stab types the contents can be a great
203 The overall format of the string field for most stab types is:
206 "@var{name}:@var{symbol-descriptor} @var{type-information}"
209 @var{name} is the name of the symbol represented by the stab; it can
210 contain a pair of colons (@pxref{Nested Symbols}). @var{name} can be
211 omitted, which means the stab represents an unnamed object. For
212 example, @samp{:t10=*2} defines type 10 as a pointer to type 2, but does
213 not give the type a name. Omitting the @var{name} field is supported by
214 AIX dbx and GDB after about version 4.8, but not other debuggers. GCC
215 sometimes uses a single space as the name instead of omitting the name
216 altogether; apparently that is supported by most debuggers.
218 The @var{symbol-descriptor} following the @samp{:} is an alphabetic
219 character that tells more specifically what kind of symbol the stab
220 represents. If the @var{symbol-descriptor} is omitted, but type
221 information follows, then the stab represents a local variable. For a
222 list of symbol descriptors, see @ref{Symbol Descriptors}. The @samp{c}
223 symbol descriptor is an exception in that it is not followed by type
224 information. @xref{Constants}.
226 @var{type-information} is either a @var{type-number}, or
227 @samp{@var{type-number}=}. A @var{type-number} alone is a type
228 reference, referring directly to a type that has already been defined.
230 The @samp{@var{type-number}=} form is a type definition, where the
231 number represents a new type which is about to be defined. The type
232 definition may refer to other types by number, and those type numbers
233 may be followed by @samp{=} and nested definitions. Also, the Lucid
234 compiler will repeat @samp{@var{type-number}=} more than once if it
235 wants to define several type numbers at once.
237 In a type definition, if the character that follows the equals sign is
238 non-numeric then it is a @var{type-descriptor}, and tells what kind of
239 type is about to be defined. Any other values following the
240 @var{type-descriptor} vary, depending on the @var{type-descriptor}.
241 @xref{Type Descriptors}, for a list of @var{type-descriptor} values. If
242 a number follows the @samp{=} then the number is a @var{type-reference}.
243 For a full description of types, @ref{Types}.
245 A @var{type-number} is often a single number. The GNU and Sun tools
246 additionally permit a @var{type-number} to be a pair
247 (@var{file-number},@var{filetype-number}) (the parentheses appear in the
248 string, and serve to distinguish the two cases). The @var{file-number}
249 is 0 for the base source file, 1 for the first included file, 2 for the
250 next, and so on. The @var{filetype-number} is a number starting with
251 1 which is incremented for each new type defined in the file.
252 (Separating the file number and the type number permits the
253 @code{N_BINCL} optimization to succeed more often; see @ref{Include
256 There is an AIX extension for type attributes. Following the @samp{=}
257 are any number of type attributes. Each one starts with @samp{@@} and
258 ends with @samp{;}. Debuggers, including AIX's dbx and GDB 4.10, skip
259 any type attributes they do not recognize. GDB 4.9 and other versions
260 of dbx may not do this. Because of a conflict with C++
261 (@pxref{Cplusplus}), new attributes should not be defined which begin
262 with a digit, @samp{(}, or @samp{-}; GDB may be unable to distinguish
263 those from the C++ type descriptor @samp{@@}. The attributes are:
266 @item a@var{boundary}
267 @var{boundary} is an integer specifying the alignment. I assume it
268 applies to all variables of this type.
271 Pointer class (for checking). Not sure what this means, or how
272 @var{integer} is interpreted.
275 Indicate this is a packed type, meaning that structure fields or array
276 elements are placed more closely in memory, to save memory at the
280 Size in bits of a variable of this type. This is fully supported by GDB
284 Indicate that this type is a string instead of an array of characters,
285 or a bitstring instead of a set. It doesn't change the layout of the
286 data being represented, but does enable the debugger to know which type
290 All of this can make the string field quite long. All versions of GDB,
291 and some versions of dbx, can handle arbitrarily long strings. But many
292 versions of dbx (or assemblers or linkers, I'm not sure which)
293 cretinously limit the strings to about 80 characters, so compilers which
294 must work with such systems need to split the @code{.stabs} directive
295 into several @code{.stabs} directives. Each stab duplicates every field
296 except the string field. The string field of every stab except the last
297 is marked as continued with a backslash at the end (in the assembly code
298 this may be written as a double backslash, depending on the assembler).
299 Removing the backslashes and concatenating the string fields of each
300 stab produces the original, long string. Just to be incompatible (or so
301 they don't have to worry about what the assembler does with
302 backslashes), AIX can use @samp{?} instead of backslash.
305 @section A Simple Example in C Source
307 To get the flavor of how stabs describe source information for a C
308 program, let's look at the simple program:
313 printf("Hello world");
317 When compiled with @samp{-g}, the program above yields the following
318 @file{.s} file. Line numbers have been added to make it easier to refer
319 to parts of the @file{.s} file in the description of the stabs that
323 @section The Simple Example at the Assembly Level
325 This simple ``hello world'' example demonstrates several of the stab
326 types used to describe C language source files.
330 2 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0
331 3 .stabs "hello.c",100,0,0,Ltext0
334 6 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0
335 7 .stabs "char:t2=r2;0;127;",128,0,0,0
336 8 .stabs "long int:t3=r1;-2147483648;2147483647;",128,0,0,0
337 9 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
338 10 .stabs "long unsigned int:t5=r1;0;-1;",128,0,0,0
339 11 .stabs "short int:t6=r1;-32768;32767;",128,0,0,0
340 12 .stabs "long long int:t7=r1;0;-1;",128,0,0,0
341 13 .stabs "short unsigned int:t8=r1;0;65535;",128,0,0,0
342 14 .stabs "long long unsigned int:t9=r1;0;-1;",128,0,0,0
343 15 .stabs "signed char:t10=r1;-128;127;",128,0,0,0
344 16 .stabs "unsigned char:t11=r1;0;255;",128,0,0,0
345 17 .stabs "float:t12=r1;4;0;",128,0,0,0
346 18 .stabs "double:t13=r1;8;0;",128,0,0,0
347 19 .stabs "long double:t14=r1;8;0;",128,0,0,0
348 20 .stabs "void:t15=15",128,0,0,0
351 23 .ascii "Hello, world!\12\0"
366 38 sethi %hi(LC0),%o1
367 39 or %o1,%lo(LC0),%o0
378 50 .stabs "main:F1",36,0,0,_main
379 51 .stabn 192,0,0,LBB2
380 52 .stabn 224,0,0,LBE2
383 @node Program Structure
384 @chapter Encoding the Structure of the Program
386 The elements of the program structure that stabs encode include the name
387 of the main function, the names of the source and include files, the
388 line numbers, procedure names and types, and the beginnings and ends of
392 * Main Program:: Indicate what the main program is
393 * Source Files:: The path and name of the source file
394 * Include Files:: Names of include files
397 * Nested Procedures::
399 * Alternate Entry Points:: Entering procedures except at the beginning.
403 @section Main Program
406 Most languages allow the main program to have any name. The
407 @code{N_MAIN} stab type tells the debugger the name that is used in this
408 program. Only the string field is significant; it is the name of
409 a function which is the main program. Most C compilers do not use this
410 stab (they expect the debugger to assume that the name is @code{main}),
411 but some C compilers emit an @code{N_MAIN} stab for the @code{main}
412 function. I'm not sure how XCOFF handles this.
415 @section Paths and Names of the Source Files
418 Before any other stabs occur, there must be a stab specifying the source
419 file. This information is contained in a symbol of stab type
420 @code{N_SO}; the string field contains the name of the file. The
421 value of the symbol is the start address of the portion of the
422 text section corresponding to that file.
424 With the Sun Solaris2 compiler, the desc field contains a
425 source-language code.
426 @c Do the debuggers use it? What are the codes? -djm
428 Some compilers (for example, GCC2 and SunOS4 @file{/bin/cc}) also
429 include the directory in which the source was compiled, in a second
430 @code{N_SO} symbol preceding the one containing the file name. This
431 symbol can be distinguished by the fact that it ends in a slash. Code
432 from the @code{cfront} C++ compiler can have additional @code{N_SO} symbols for
433 nonexistent source files after the @code{N_SO} for the real source file;
434 these are believed to contain no useful information.
439 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0 # @r{100 is N_SO}
440 .stabs "hello.c",100,0,0,Ltext0
446 Instead of @code{N_SO} symbols, XCOFF uses a @code{.file} assembler
447 directive which assembles to a @code{C_FILE} symbol; explaining this in
448 detail is outside the scope of this document.
450 @c FIXME: Exactly when should the empty N_SO be used? Why?
451 If it is useful to indicate the end of a source file, this is done with
452 an @code{N_SO} symbol with an empty string for the name. The value is
453 the address of the end of the text section for the file. For some
454 systems, there is no indication of the end of a source file, and you
455 just need to figure it ended when you see an @code{N_SO} for a different
456 source file, or a symbol ending in @code{.o} (which at least some
457 linkers insert to mark the start of a new @code{.o} file).
460 @section Names of Include Files
462 There are several schemes for dealing with include files: the
463 traditional @code{N_SOL} approach, Sun's @code{N_BINCL} approach, and the
464 XCOFF @code{C_BINCL} approach (which despite the similar name has little in
465 common with @code{N_BINCL}).
468 An @code{N_SOL} symbol specifies which include file subsequent symbols
469 refer to. The string field is the name of the file and the value is the
470 text address corresponding to the end of the previous include file and
471 the start of this one. To specify the main source file again, use an
472 @code{N_SOL} symbol with the name of the main source file.
477 The @code{N_BINCL} approach works as follows. An @code{N_BINCL} symbol
478 specifies the start of an include file. In an object file, only the
479 string is significant; the linker puts data into some of the other
480 fields. The end of the include file is marked by an @code{N_EINCL}
481 symbol (which has no string field). In an object file, there is no
482 significant data in the @code{N_EINCL} symbol. @code{N_BINCL} and
483 @code{N_EINCL} can be nested.
485 If the linker detects that two source files have identical stabs between
486 an @code{N_BINCL} and @code{N_EINCL} pair (as will generally be the case
487 for a header file), then it only puts out the stabs once. Each
488 additional occurrence is replaced by an @code{N_EXCL} symbol. I believe
489 the GNU linker and the Sun (both SunOS4 and Solaris) linker are the only
490 ones which supports this feature.
492 A linker which supports this feature will set the value of a
493 @code{N_BINCL} symbol to the total of all the characters in the stabs
494 strings included in the header file, omitting any file numbers. The
495 value of an @code{N_EXCL} symbol is the same as the value of the
496 @code{N_BINCL} symbol it replaces. This information can be used to
497 match up @code{N_EXCL} and @code{N_BINCL} symbols which have the same
498 filename. The @code{N_EINCL} value, and the values of the other and
499 description fields for all three, appear to always be zero.
503 For the start of an include file in XCOFF, use the @file{.bi} assembler
504 directive, which generates a @code{C_BINCL} symbol. A @file{.ei}
505 directive, which generates a @code{C_EINCL} symbol, denotes the end of
506 the include file. Both directives are followed by the name of the
507 source file in quotes, which becomes the string for the symbol.
508 The value of each symbol, produced automatically by the assembler
509 and linker, is the offset into the executable of the beginning
510 (inclusive, as you'd expect) or end (inclusive, as you would not expect)
511 of the portion of the COFF line table that corresponds to this include
512 file. @code{C_BINCL} and @code{C_EINCL} do not nest.
515 @section Line Numbers
518 An @code{N_SLINE} symbol represents the start of a source line. The
519 desc field contains the line number and the value contains the code
520 address for the start of that source line. On most machines the address
521 is absolute; for stabs in sections (@pxref{Stab Sections}), it is
522 relative to the function in which the @code{N_SLINE} symbol occurs.
526 GNU documents @code{N_DSLINE} and @code{N_BSLINE} symbols for line
527 numbers in the data or bss segments, respectively. They are identical
528 to @code{N_SLINE} but are relocated differently by the linker. They
529 were intended to be used to describe the source location of a variable
530 declaration, but I believe that GCC2 actually puts the line number in
531 the desc field of the stab for the variable itself. GDB has been
532 ignoring these symbols (unless they contain a string field) since
535 For single source lines that generate discontiguous code, such as flow
536 of control statements, there may be more than one line number entry for
537 the same source line. In this case there is a line number entry at the
538 start of each code range, each with the same line number.
540 XCOFF does not use stabs for line numbers. Instead, it uses COFF line
541 numbers (which are outside the scope of this document). Standard COFF
542 line numbers cannot deal with include files, but in XCOFF this is fixed
543 with the @code{C_BINCL} method of marking include files (@pxref{Include
549 @findex N_FUN, for functions
551 @findex N_STSYM, for functions (Sun acc)
552 @findex N_GSYM, for functions (Sun acc)
553 All of the following stabs normally use the @code{N_FUN} symbol type.
554 However, Sun's @code{acc} compiler on SunOS4 uses @code{N_GSYM} and
555 @code{N_STSYM}, which means that the value of the stab for the function
556 is useless and the debugger must get the address of the function from
557 the non-stab symbols instead. On systems where non-stab symbols have
558 leading underscores, the stabs will lack underscores and the debugger
559 needs to know about the leading underscore to match up the stab and the
560 non-stab symbol. BSD Fortran is said to use @code{N_FNAME} with the
561 same restriction; the value of the symbol is not useful (I'm not sure it
562 really does use this, because GDB doesn't handle this and no one has
566 A function is represented by an @samp{F} symbol descriptor for a global
567 (extern) function, and @samp{f} for a static (local) function. For
568 a.out, the value of the symbol is the address of the start of the
569 function; it is already relocated. For stabs in ELF, the SunPRO
570 compiler version 2.0.1 and GCC put out an address which gets relocated
571 by the linker. In a future release SunPRO is planning to put out zero,
572 in which case the address can be found from the ELF (non-stab) symbol.
573 Because looking things up in the ELF symbols would probably be slow, I'm
574 not sure how to find which symbol of that name is the right one, and
575 this doesn't provide any way to deal with nested functions, it would
576 probably be better to make the value of the stab an address relative to
577 the start of the file, or just absolute. See @ref{ELF Linker
578 Relocation} for more information on linker relocation of stabs in ELF
579 files. For XCOFF, the stab uses the @code{C_FUN} storage class and the
580 value of the stab is meaningless; the address of the function can be
581 found from the csect symbol (XTY_LD/XMC_PR).
583 The type information of the stab represents the return type of the
584 function; thus @samp{foo:f5} means that foo is a function returning type
585 5. There is no need to try to get the line number of the start of the
586 function from the stab for the function; it is in the next
587 @code{N_SLINE} symbol.
589 @c FIXME: verify whether the "I suspect" below is true or not.
590 Some compilers (such as Sun's Solaris compiler) support an extension for
591 specifying the types of the arguments. I suspect this extension is not
592 used for old (non-prototyped) function definitions in C. If the
593 extension is in use, the type information of the stab for the function
594 is followed by type information for each argument, with each argument
595 preceded by @samp{;}. An argument type of 0 means that additional
596 arguments are being passed, whose types and number may vary (@samp{...}
597 in ANSI C). GDB has tolerated this extension (parsed the syntax, if not
598 necessarily used the information) since at least version 4.8; I don't
599 know whether all versions of dbx tolerate it. The argument types given
600 here are not redundant with the symbols for the formal parameters
601 (@pxref{Parameters}); they are the types of the arguments as they are
602 passed, before any conversions might take place. For example, if a C
603 function which is declared without a prototype takes a @code{float}
604 argument, the value is passed as a @code{double} but then converted to a
605 @code{float}. Debuggers need to use the types given in the arguments
606 when printing values, but when calling the function they need to use the
607 types given in the symbol defining the function.
609 If the return type and types of arguments of a function which is defined
610 in another source file are specified (i.e., a function prototype in ANSI
611 C), traditionally compilers emit no stab; the only way for the debugger
612 to find the information is if the source file where the function is
613 defined was also compiled with debugging symbols. As an extension the
614 Solaris compiler uses symbol descriptor @samp{P} followed by the return
615 type of the function, followed by the arguments, each preceded by
616 @samp{;}, as in a stab with symbol descriptor @samp{f} or @samp{F}.
617 This use of symbol descriptor @samp{P} can be distinguished from its use
618 for register parameters (@pxref{Register Parameters}) by the fact that it has
619 symbol type @code{N_FUN}.
621 The AIX documentation also defines symbol descriptor @samp{J} as an
622 internal function. I assume this means a function nested within another
623 function. It also says symbol descriptor @samp{m} is a module in
624 Modula-2 or extended Pascal.
626 Procedures (functions which do not return values) are represented as
627 functions returning the @code{void} type in C. I don't see why this couldn't
628 be used for all languages (inventing a @code{void} type for this purpose if
629 necessary), but the AIX documentation defines @samp{I}, @samp{P}, and
630 @samp{Q} for internal, global, and static procedures, respectively.
631 These symbol descriptors are unusual in that they are not followed by
634 The following example shows a stab for a function @code{main} which
635 returns type number @code{1}. The @code{_main} specified for the value
636 is a reference to an assembler label which is used to fill in the start
637 address of the function.
640 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
643 The stab representing a procedure is located immediately following the
644 code of the procedure. This stab is in turn directly followed by a
645 group of other stabs describing elements of the procedure. These other
646 stabs describe the procedure's parameters, its block local variables, and
649 If functions can appear in different sections, then the debugger may not
650 be able to find the end of a function. Recent versions of GCC will mark
651 the end of a function with an @code{N_FUN} symbol with an empty string
652 for the name. The value is the address of the end of the current
653 function. Without such a symbol, there is no indication of the address
654 of the end of a function, and you must assume that it ended at the
655 starting address of the next function or at the end of the text section
658 @node Nested Procedures
659 @section Nested Procedures
661 For any of the symbol descriptors representing procedures, after the
662 symbol descriptor and the type information is optionally a scope
663 specifier. This consists of a comma, the name of the procedure, another
664 comma, and the name of the enclosing procedure. The first name is local
665 to the scope specified, and seems to be redundant with the name of the
666 symbol (before the @samp{:}). This feature is used by GCC, and
667 presumably Pascal, Modula-2, etc., compilers, for nested functions.
669 If procedures are nested more than one level deep, only the immediately
670 containing scope is specified. For example, this code:
682 return baz (x + 2 * y);
684 return x + bar (3 * x);
692 .stabs "baz:f1,baz,bar",36,0,0,_baz.15 # @r{36 is N_FUN}
693 .stabs "bar:f1,bar,foo",36,0,0,_bar.12
694 .stabs "foo:F1",36,0,0,_foo
697 @node Block Structure
698 @section Block Structure
702 @c For GCC 2.5.8 or so stabs-in-coff, these are absolute instead of
703 @c function relative (as documented below). But GDB has never been able
704 @c to deal with that (it had wanted them to be relative to the file, but
705 @c I just fixed that (between GDB 4.12 and 4.13)), so it is function
706 @c relative just like ELF and SOM and the below documentation.
707 The program's block structure is represented by the @code{N_LBRAC} (left
708 brace) and the @code{N_RBRAC} (right brace) stab types. The variables
709 defined inside a block precede the @code{N_LBRAC} symbol for most
710 compilers, including GCC. Other compilers, such as the Convex, Acorn
711 RISC machine, and Sun @code{acc} compilers, put the variables after the
712 @code{N_LBRAC} symbol. The values of the @code{N_LBRAC} and
713 @code{N_RBRAC} symbols are the start and end addresses of the code of
714 the block, respectively. For most machines, they are relative to the
715 starting address of this source file. For the Gould NP1, they are
716 absolute. For stabs in sections (@pxref{Stab Sections}), they are
717 relative to the function in which they occur.
719 The @code{N_LBRAC} and @code{N_RBRAC} stabs that describe the block
720 scope of a procedure are located after the @code{N_FUN} stab that
721 represents the procedure itself.
723 Sun documents the desc field of @code{N_LBRAC} and
724 @code{N_RBRAC} symbols as containing the nesting level of the block.
725 However, dbx seems to not care, and GCC always sets desc to
731 For XCOFF, block scope is indicated with @code{C_BLOCK} symbols. If the
732 name of the symbol is @samp{.bb}, then it is the beginning of the block;
733 if the name of the symbol is @samp{.be}; it is the end of the block.
735 @node Alternate Entry Points
736 @section Alternate Entry Points
740 Some languages, like Fortran, have the ability to enter procedures at
741 some place other than the beginning. One can declare an alternate entry
742 point. The @code{N_ENTRY} stab is for this; however, the Sun FORTRAN
743 compiler doesn't use it. According to AIX documentation, only the name
744 of a @code{C_ENTRY} stab is significant; the address of the alternate
745 entry point comes from the corresponding external symbol. A previous
746 revision of this document said that the value of an @code{N_ENTRY} stab
747 was the address of the alternate entry point, but I don't know the
748 source for that information.
753 The @samp{c} symbol descriptor indicates that this stab represents a
754 constant. This symbol descriptor is an exception to the general rule
755 that symbol descriptors are followed by type information. Instead, it
756 is followed by @samp{=} and one of the following:
760 Boolean constant. @var{value} is a numeric value; I assume it is 0 for
764 Character constant. @var{value} is the numeric value of the constant.
766 @item e @var{type-information} , @var{value}
767 Constant whose value can be represented as integral.
768 @var{type-information} is the type of the constant, as it would appear
769 after a symbol descriptor (@pxref{String Field}). @var{value} is the
770 numeric value of the constant. GDB 4.9 does not actually get the right
771 value if @var{value} does not fit in a host @code{int}, but it does not
772 do anything violent, and future debuggers could be extended to accept
773 integers of any size (whether unsigned or not). This constant type is
774 usually documented as being only for enumeration constants, but GDB has
775 never imposed that restriction; I don't know about other debuggers.
778 Integer constant. @var{value} is the numeric value. The type is some
779 sort of generic integer type (for GDB, a host @code{int}); to specify
780 the type explicitly, use @samp{e} instead.
783 Real constant. @var{value} is the real value, which can be @samp{INF}
784 (optionally preceded by a sign) for infinity, @samp{QNAN} for a quiet
785 NaN (not-a-number), or @samp{SNAN} for a signalling NaN. If it is a
786 normal number the format is that accepted by the C library function
790 String constant. @var{string} is a string enclosed in either @samp{'}
791 (in which case @samp{'} characters within the string are represented as
792 @samp{\'} or @samp{"} (in which case @samp{"} characters within the
793 string are represented as @samp{\"}).
795 @item S @var{type-information} , @var{elements} , @var{bits} , @var{pattern}
796 Set constant. @var{type-information} is the type of the constant, as it
797 would appear after a symbol descriptor (@pxref{String Field}).
798 @var{elements} is the number of elements in the set (does this means
799 how many bits of @var{pattern} are actually used, which would be
800 redundant with the type, or perhaps the number of bits set in
801 @var{pattern}? I don't get it), @var{bits} is the number of bits in the
802 constant (meaning it specifies the length of @var{pattern}, I think),
803 and @var{pattern} is a hexadecimal representation of the set. AIX
804 documentation refers to a limit of 32 bytes, but I see no reason why
805 this limit should exist. This form could probably be used for arbitrary
806 constants, not just sets; the only catch is that @var{pattern} should be
807 understood to be target, not host, byte order and format.
810 The boolean, character, string, and set constants are not supported by
811 GDB 4.9, but it ignores them. GDB 4.8 and earlier gave an error
812 message and refused to read symbols from the file containing the
815 The above information is followed by @samp{;}.
820 Different types of stabs describe the various ways that variables can be
821 allocated: on the stack, globally, in registers, in common blocks,
822 statically, or as arguments to a function.
825 * Stack Variables:: Variables allocated on the stack.
826 * Global Variables:: Variables used by more than one source file.
827 * Register Variables:: Variables in registers.
828 * Common Blocks:: Variables statically allocated together.
829 * Statics:: Variables local to one source file.
830 * Based Variables:: Fortran pointer based variables.
831 * Parameters:: Variables for arguments to functions.
834 @node Stack Variables
835 @section Automatic Variables Allocated on the Stack
837 If a variable's scope is local to a function and its lifetime is only as
838 long as that function executes (C calls such variables
839 @dfn{automatic}), it can be allocated in a register (@pxref{Register
840 Variables}) or on the stack.
842 @findex N_LSYM, for stack variables
844 Each variable allocated on the stack has a stab with the symbol
845 descriptor omitted. Since type information should begin with a digit,
846 @samp{-}, or @samp{(}, only those characters precluded from being used
847 for symbol descriptors. However, the Acorn RISC machine (ARM) is said
848 to get this wrong: it puts out a mere type definition here, without the
849 preceding @samp{@var{type-number}=}. This is a bad idea; there is no
850 guarantee that type descriptors are distinct from symbol descriptors.
851 Stabs for stack variables use the @code{N_LSYM} stab type, or
852 @code{C_LSYM} for XCOFF.
854 The value of the stab is the offset of the variable within the
855 local variables. On most machines this is an offset from the frame
856 pointer and is negative. The location of the stab specifies which block
857 it is defined in; see @ref{Block Structure}.
859 For example, the following C code:
869 produces the following stabs:
872 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
873 .stabs "x:1",128,0,0,-12 # @r{128 is N_LSYM}
874 .stabn 192,0,0,LBB2 # @r{192 is N_LBRAC}
875 .stabn 224,0,0,LBE2 # @r{224 is N_RBRAC}
878 See @ref{Procedures} for more information on the @code{N_FUN} stab, and
879 @ref{Block Structure} for more information on the @code{N_LBRAC} and
880 @code{N_RBRAC} stabs.
882 @node Global Variables
883 @section Global Variables
887 @c FIXME: verify for sure that it really is C_GSYM on XCOFF
888 A variable whose scope is not specific to just one source file is
889 represented by the @samp{G} symbol descriptor. These stabs use the
890 @code{N_GSYM} stab type (C_GSYM for XCOFF). The type information for
891 the stab (@pxref{String Field}) gives the type of the variable.
893 For example, the following source code:
900 yields the following assembly code:
903 .stabs "g_foo:G2",32,0,0,0 # @r{32 is N_GSYM}
910 The address of the variable represented by the @code{N_GSYM} is not
911 contained in the @code{N_GSYM} stab. The debugger gets this information
912 from the external symbol for the global variable. In the example above,
913 the @code{.global _g_foo} and @code{_g_foo:} lines tell the assembler to
914 produce an external symbol.
916 Some compilers, like GCC, output @code{N_GSYM} stabs only once, where
917 the variable is defined. Other compilers, like SunOS4 /bin/cc, output a
918 @code{N_GSYM} stab for each compilation unit which references the
921 @node Register Variables
922 @section Register Variables
926 @c According to an old version of this manual, AIX uses C_RPSYM instead
927 @c of C_RSYM. I am skeptical; this should be verified.
928 Register variables have their own stab type, @code{N_RSYM}
929 (@code{C_RSYM} for XCOFF), and their own symbol descriptor, @samp{r}.
930 The stab's value is the number of the register where the variable data
932 @c .stabs "name:type",N_RSYM,0,RegSize,RegNumber (Sun doc)
934 AIX defines a separate symbol descriptor @samp{d} for floating point
935 registers. This seems unnecessary; why not just just give floating
936 point registers different register numbers? I have not verified whether
937 the compiler actually uses @samp{d}.
939 If the register is explicitly allocated to a global variable, but not
943 register int g_bar asm ("%g5");
947 then the stab may be emitted at the end of the object file, with
948 the other bss symbols.
951 @section Common Blocks
953 A common block is a statically allocated section of memory which can be
954 referred to by several source files. It may contain several variables.
955 I believe Fortran is the only language with this feature.
961 A @code{N_BCOMM} stab begins a common block and an @code{N_ECOMM} stab
962 ends it. The only field that is significant in these two stabs is the
963 string, which names a normal (non-debugging) symbol that gives the
964 address of the common block. According to IBM documentation, only the
965 @code{N_BCOMM} has the name of the common block (even though their
966 compiler actually puts it both places).
970 The stabs for the members of the common block are between the
971 @code{N_BCOMM} and the @code{N_ECOMM}; the value of each stab is the
972 offset within the common block of that variable. IBM uses the
973 @code{C_ECOML} stab type, and there is a corresponding @code{N_ECOML}
974 stab type, but Sun's Fortran compiler uses @code{N_GSYM} instead. The
975 variables within a common block use the @samp{V} symbol descriptor (I
976 believe this is true of all Fortran variables). Other stabs (at least
977 type declarations using @code{C_DECL}) can also be between the
978 @code{N_BCOMM} and the @code{N_ECOMM}.
981 @section Static Variables
983 Initialized static variables are represented by the @samp{S} and
984 @samp{V} symbol descriptors. @samp{S} means file scope static, and
985 @samp{V} means procedure scope static. One exception: in XCOFF, IBM's
986 xlc compiler always uses @samp{V}, and whether it is file scope or not
987 is distinguished by whether the stab is located within a function.
989 @c This is probably not worth mentioning; it is only true on the sparc
990 @c for `double' variables which although declared const are actually in
991 @c the data segment (the text segment can't guarantee 8 byte alignment).
993 @c 2.4.5 has a bug in that it uses @code{N_FUN}, so neither dbx nor GDB can
994 @c find the variables)
997 @findex N_FUN, for variables
999 In a.out files, @code{N_STSYM} means the data section, @code{N_FUN}
1000 means the text section, and @code{N_LCSYM} means the bss section. For
1001 those systems with a read-only data section separate from the text
1002 section (Solaris), @code{N_ROSYM} means the read-only data section.
1004 For example, the source lines:
1007 static const int var_const = 5;
1008 static int var_init = 2;
1009 static int var_noinit;
1013 yield the following stabs:
1016 .stabs "var_const:S1",36,0,0,_var_const # @r{36 is N_FUN}
1018 .stabs "var_init:S1",38,0,0,_var_init # @r{38 is N_STSYM}
1020 .stabs "var_noinit:S1",40,0,0,_var_noinit # @r{40 is N_LCSYM}
1026 In XCOFF files, the stab type need not indicate the section;
1027 @code{C_STSYM} can be used for all statics. Also, each static variable
1028 is enclosed in a static block. A @code{C_BSTAT} (emitted with a
1029 @samp{.bs} assembler directive) symbol begins the static block; its
1030 value is the symbol number of the csect symbol whose value is the
1031 address of the static block, its section is the section of the variables
1032 in that static block, and its name is @samp{.bs}. A @code{C_ESTAT}
1033 (emitted with a @samp{.es} assembler directive) symbol ends the static
1034 block; its name is @samp{.es} and its value and section are ignored.
1036 In ECOFF files, the storage class is used to specify the section, so the
1037 stab type need not indicate the section.
1039 In ELF files, for the SunPRO compiler version 2.0.1, symbol descriptor
1040 @samp{S} means that the address is absolute (the linker relocates it)
1041 and symbol descriptor @samp{V} means that the address is relative to the
1042 start of the relevant section for that compilation unit. SunPRO has
1043 plans to have the linker stop relocating stabs; I suspect that their the
1044 debugger gets the address from the corresponding ELF (not stab) symbol.
1045 I'm not sure how to find which symbol of that name is the right one.
1046 The clean way to do all this would be to have a the value of a symbol
1047 descriptor @samp{S} symbol be an offset relative to the start of the
1048 file, just like everything else, but that introduces obvious
1049 compatibility problems. For more information on linker stab relocation,
1050 @xref{ELF Linker Relocation}.
1052 @node Based Variables
1053 @section Fortran Based Variables
1055 Fortran (at least, the Sun and SGI dialects of FORTRAN-77) has a feature
1056 which allows allocating arrays with @code{malloc}, but which avoids
1057 blurring the line between arrays and pointers the way that C does. In
1058 stabs such a variable uses the @samp{b} symbol descriptor.
1060 For example, the Fortran declarations
1063 real foo, foo10(10), foo10_5(10,5)
1065 pointer (foo10p, foo10)
1066 pointer (foo105p, foo10_5)
1074 foo10_5:bar3;1;5;ar3;1;10;6
1077 In this example, @code{real} is type 6 and type 3 is an integral type
1078 which is the type of the subscripts of the array (probably
1081 The @samp{b} symbol descriptor is like @samp{V} in that it denotes a
1082 statically allocated symbol whose scope is local to a function; see
1083 @xref{Statics}. The value of the symbol, instead of being the address
1084 of the variable itself, is the address of a pointer to that variable.
1085 So in the above example, the value of the @code{foo} stab is the address
1086 of a pointer to a real, the value of the @code{foo10} stab is the
1087 address of a pointer to a 10-element array of reals, and the value of
1088 the @code{foo10_5} stab is the address of a pointer to a 5-element array
1089 of 10-element arrays of reals.
1094 Formal parameters to a function are represented by a stab (or sometimes
1095 two; see below) for each parameter. The stabs are in the order in which
1096 the debugger should print the parameters (i.e., the order in which the
1097 parameters are declared in the source file). The exact form of the stab
1098 depends on how the parameter is being passed.
1102 Parameters passed on the stack use the symbol descriptor @samp{p} and
1103 the @code{N_PSYM} symbol type (or @code{C_PSYM} for XCOFF). The value
1104 of the symbol is an offset used to locate the parameter on the stack;
1105 its exact meaning is machine-dependent, but on most machines it is an
1106 offset from the frame pointer.
1108 As a simple example, the code:
1119 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
1120 .stabs "argc:p1",160,0,0,68 # @r{160 is N_PSYM}
1121 .stabs "argv:p20=*21=*2",160,0,0,72
1124 The type definition of @code{argv} is interesting because it contains
1125 several type definitions. Type 21 is pointer to type 2 (char) and
1126 @code{argv} (type 20) is pointer to type 21.
1128 @c FIXME: figure out what these mean and describe them coherently.
1129 The following symbol descriptors are also said to go with @code{N_PSYM}.
1130 The value of the symbol is said to be an offset from the argument
1131 pointer (I'm not sure whether this is true or not).
1135 pF Fortran function parameter
1136 X (function result variable)
1140 * Register Parameters::
1141 * Local Variable Parameters::
1142 * Reference Parameters::
1143 * Conformant Arrays::
1146 @node Register Parameters
1147 @subsection Passing Parameters in Registers
1149 If the parameter is passed in a register, then traditionally there are
1150 two symbols for each argument:
1153 .stabs "arg:p1" . . . ; N_PSYM
1154 .stabs "arg:r1" . . . ; N_RSYM
1157 Debuggers use the second one to find the value, and the first one to
1158 know that it is an argument.
1161 @findex N_RSYM, for parameters
1162 Because that approach is kind of ugly, some compilers use symbol
1163 descriptor @samp{P} or @samp{R} to indicate an argument which is in a
1164 register. Symbol type @code{C_RPSYM} is used in XCOFF and @code{N_RSYM}
1165 is used otherwise. The symbol's value is the register number. @samp{P}
1166 and @samp{R} mean the same thing; the difference is that @samp{P} is a
1167 GNU invention and @samp{R} is an IBM (XCOFF) invention. As of version
1168 4.9, GDB should handle either one.
1170 There is at least one case where GCC uses a @samp{p} and @samp{r} pair
1171 rather than @samp{P}; this is where the argument is passed in the
1172 argument list and then loaded into a register.
1174 According to the AIX documentation, symbol descriptor @samp{D} is for a
1175 parameter passed in a floating point register. This seems
1176 unnecessary---why not just use @samp{R} with a register number which
1177 indicates that it's a floating point register? I haven't verified
1178 whether the system actually does what the documentation indicates.
1180 @c FIXME: On the hppa this is for any type > 8 bytes, I think, and not
1181 @c for small structures (investigate).
1182 On the sparc and hppa, for a @samp{P} symbol whose type is a structure
1183 or union, the register contains the address of the structure. On the
1184 sparc, this is also true of a @samp{p} and @samp{r} pair (using Sun
1185 @code{cc}) or a @samp{p} symbol. However, if a (small) structure is
1186 really in a register, @samp{r} is used. And, to top it all off, on the
1187 hppa it might be a structure which was passed on the stack and loaded
1188 into a register and for which there is a @samp{p} and @samp{r} pair! I
1189 believe that symbol descriptor @samp{i} is supposed to deal with this
1190 case (it is said to mean "value parameter by reference, indirect
1191 access"; I don't know the source for this information), but I don't know
1192 details or what compilers or debuggers use it, if any (not GDB or GCC).
1193 It is not clear to me whether this case needs to be dealt with
1194 differently than parameters passed by reference (@pxref{Reference Parameters}).
1196 @node Local Variable Parameters
1197 @subsection Storing Parameters as Local Variables
1199 There is a case similar to an argument in a register, which is an
1200 argument that is actually stored as a local variable. Sometimes this
1201 happens when the argument was passed in a register and then the compiler
1202 stores it as a local variable. If possible, the compiler should claim
1203 that it's in a register, but this isn't always done.
1205 If a parameter is passed as one type and converted to a smaller type by
1206 the prologue (for example, the parameter is declared as a @code{float},
1207 but the calling conventions specify that it is passed as a
1208 @code{double}), then GCC2 (sometimes) uses a pair of symbols. The first
1209 symbol uses symbol descriptor @samp{p} and the type which is passed.
1210 The second symbol has the type and location which the parameter actually
1211 has after the prologue. For example, suppose the following C code
1212 appears with no prototypes involved:
1221 if @code{f} is passed as a double at stack offset 8, and the prologue
1222 converts it to a float in register number 0, then the stabs look like:
1225 .stabs "f:p13",160,0,3,8 # @r{160 is @code{N_PSYM}, here 13 is @code{double}}
1226 .stabs "f:r12",64,0,3,0 # @r{64 is @code{N_RSYM}, here 12 is @code{float}}
1229 In both stabs 3 is the line number where @code{f} is declared
1230 (@pxref{Line Numbers}).
1232 @findex N_LSYM, for parameter
1233 GCC, at least on the 960, has another solution to the same problem. It
1234 uses a single @samp{p} symbol descriptor for an argument which is stored
1235 as a local variable but uses @code{N_LSYM} instead of @code{N_PSYM}. In
1236 this case, the value of the symbol is an offset relative to the local
1237 variables for that function, not relative to the arguments; on some
1238 machines those are the same thing, but not on all.
1240 @c This is mostly just background info; the part that logically belongs
1241 @c here is the last sentence.
1242 On the VAX or on other machines in which the calling convention includes
1243 the number of words of arguments actually passed, the debugger (GDB at
1244 least) uses the parameter symbols to keep track of whether it needs to
1245 print nameless arguments in addition to the formal parameters which it
1246 has printed because each one has a stab. For example, in
1249 extern int fprintf (FILE *stream, char *format, @dots{});
1251 fprintf (stdout, "%d\n", x);
1254 there are stabs for @code{stream} and @code{format}. On most machines,
1255 the debugger can only print those two arguments (because it has no way
1256 of knowing that additional arguments were passed), but on the VAX or
1257 other machines with a calling convention which indicates the number of
1258 words of arguments, the debugger can print all three arguments. To do
1259 so, the parameter symbol (symbol descriptor @samp{p}) (not necessarily
1260 @samp{r} or symbol descriptor omitted symbols) needs to contain the
1261 actual type as passed (for example, @code{double} not @code{float} if it
1262 is passed as a double and converted to a float).
1264 @node Reference Parameters
1265 @subsection Passing Parameters by Reference
1267 If the parameter is passed by reference (e.g., Pascal @code{VAR}
1268 parameters), then the symbol descriptor is @samp{v} if it is in the
1269 argument list, or @samp{a} if it in a register. Other than the fact
1270 that these contain the address of the parameter rather than the
1271 parameter itself, they are identical to @samp{p} and @samp{R},
1272 respectively. I believe @samp{a} is an AIX invention; @samp{v} is
1273 supported by all stabs-using systems as far as I know.
1275 @node Conformant Arrays
1276 @subsection Passing Conformant Array Parameters
1278 @c Is this paragraph correct? It is based on piecing together patchy
1279 @c information and some guesswork
1280 Conformant arrays are a feature of Modula-2, and perhaps other
1281 languages, in which the size of an array parameter is not known to the
1282 called function until run-time. Such parameters have two stabs: a
1283 @samp{x} for the array itself, and a @samp{C}, which represents the size
1284 of the array. The value of the @samp{x} stab is the offset in the
1285 argument list where the address of the array is stored (it this right?
1286 it is a guess); the value of the @samp{C} stab is the offset in the
1287 argument list where the size of the array (in elements? in bytes?) is
1291 @chapter Defining Types
1293 The examples so far have described types as references to previously
1294 defined types, or defined in terms of subranges of or pointers to
1295 previously defined types. This chapter describes the other type
1296 descriptors that may follow the @samp{=} in a type definition.
1299 * Builtin Types:: Integers, floating point, void, etc.
1300 * Miscellaneous Types:: Pointers, sets, files, etc.
1301 * Cross-References:: Referring to a type not yet defined.
1302 * Subranges:: A type with a specific range.
1303 * Arrays:: An aggregate type of same-typed elements.
1304 * Strings:: Like an array but also has a length.
1305 * Enumerations:: Like an integer but the values have names.
1306 * Structures:: An aggregate type of different-typed elements.
1307 * Typedefs:: Giving a type a name.
1308 * Unions:: Different types sharing storage.
1313 @section Builtin Types
1315 Certain types are built in (@code{int}, @code{short}, @code{void},
1316 @code{float}, etc.); the debugger recognizes these types and knows how
1317 to handle them. Thus, don't be surprised if some of the following ways
1318 of specifying builtin types do not specify everything that a debugger
1319 would need to know about the type---in some cases they merely specify
1320 enough information to distinguish the type from other types.
1322 The traditional way to define builtin types is convoluted, so new ways
1323 have been invented to describe them. Sun's @code{acc} uses special
1324 builtin type descriptors (@samp{b} and @samp{R}), and IBM uses negative
1325 type numbers. GDB accepts all three ways, as of version 4.8; dbx just
1326 accepts the traditional builtin types and perhaps one of the other two
1327 formats. The following sections describe each of these formats.
1330 * Traditional Builtin Types:: Put on your seat belts and prepare for kludgery
1331 * Builtin Type Descriptors:: Builtin types with special type descriptors
1332 * Negative Type Numbers:: Builtin types using negative type numbers
1335 @node Traditional Builtin Types
1336 @subsection Traditional Builtin Types
1338 This is the traditional, convoluted method for defining builtin types.
1339 There are several classes of such type definitions: integer, floating
1340 point, and @code{void}.
1343 * Traditional Integer Types::
1344 * Traditional Other Types::
1347 @node Traditional Integer Types
1348 @subsubsection Traditional Integer Types
1350 Often types are defined as subranges of themselves. If the bounding values
1351 fit within an @code{int}, then they are given normally. For example:
1354 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0 # @r{128 is N_LSYM}
1355 .stabs "char:t2=r2;0;127;",128,0,0,0
1358 Builtin types can also be described as subranges of @code{int}:
1361 .stabs "unsigned short:t6=r1;0;65535;",128,0,0,0
1364 If the lower bound of a subrange is 0 and the upper bound is -1,
1365 the type is an unsigned integral type whose bounds are too
1366 big to describe in an @code{int}. Traditionally this is only used for
1367 @code{unsigned int} and @code{unsigned long}:
1370 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
1373 For larger types, GCC 2.4.5 puts out bounds in octal, with one or more
1374 leading zeroes. In this case a negative bound consists of a number
1375 which is a 1 bit (for the sign bit) followed by a 0 bit for each bit in
1376 the number (except the sign bit), and a positive bound is one which is a
1377 1 bit for each bit in the number (except possibly the sign bit). All
1378 known versions of dbx and GDB version 4 accept this (at least in the
1379 sense of not refusing to process the file), but GDB 3.5 refuses to read
1380 the whole file containing such symbols. So GCC 2.3.3 did not output the
1381 proper size for these types. As an example of octal bounds, the string
1382 fields of the stabs for 64 bit integer types look like:
1384 @c .stabs directives, etc., omitted to make it fit on the page.
1386 long int:t3=r1;001000000000000000000000;000777777777777777777777;
1387 long unsigned int:t5=r1;000000000000000000000000;001777777777777777777777;
1390 If the lower bound of a subrange is 0 and the upper bound is negative,
1391 the type is an unsigned integral type whose size in bytes is the
1392 absolute value of the upper bound. I believe this is a Convex
1393 convention for @code{unsigned long long}.
1395 If the lower bound of a subrange is negative and the upper bound is 0,
1396 the type is a signed integral type whose size in bytes is
1397 the absolute value of the lower bound. I believe this is a Convex
1398 convention for @code{long long}. To distinguish this from a legitimate
1399 subrange, the type should be a subrange of itself. I'm not sure whether
1400 this is the case for Convex.
1402 @node Traditional Other Types
1403 @subsubsection Traditional Other Types
1405 If the upper bound of a subrange is 0 and the lower bound is positive,
1406 the type is a floating point type, and the lower bound of the subrange
1407 indicates the number of bytes in the type:
1410 .stabs "float:t12=r1;4;0;",128,0,0,0
1411 .stabs "double:t13=r1;8;0;",128,0,0,0
1414 However, GCC writes @code{long double} the same way it writes
1415 @code{double}, so there is no way to distinguish.
1418 .stabs "long double:t14=r1;8;0;",128,0,0,0
1421 Complex types are defined the same way as floating-point types; there is
1422 no way to distinguish a single-precision complex from a double-precision
1423 floating-point type.
1425 The C @code{void} type is defined as itself:
1428 .stabs "void:t15=15",128,0,0,0
1431 I'm not sure how a boolean type is represented.
1433 @node Builtin Type Descriptors
1434 @subsection Defining Builtin Types Using Builtin Type Descriptors
1436 This is the method used by Sun's @code{acc} for defining builtin types.
1437 These are the type descriptors to define builtin types:
1440 @c FIXME: clean up description of width and offset, once we figure out
1442 @item b @var{signed} @var{char-flag} @var{width} ; @var{offset} ; @var{nbits} ;
1443 Define an integral type. @var{signed} is @samp{u} for unsigned or
1444 @samp{s} for signed. @var{char-flag} is @samp{c} which indicates this
1445 is a character type, or is omitted. I assume this is to distinguish an
1446 integral type from a character type of the same size, for example it
1447 might make sense to set it for the C type @code{wchar_t} so the debugger
1448 can print such variables differently (Solaris does not do this). Sun
1449 sets it on the C types @code{signed char} and @code{unsigned char} which
1450 arguably is wrong. @var{width} and @var{offset} appear to be for small
1451 objects stored in larger ones, for example a @code{short} in an
1452 @code{int} register. @var{width} is normally the number of bytes in the
1453 type. @var{offset} seems to always be zero. @var{nbits} is the number
1454 of bits in the type.
1456 Note that type descriptor @samp{b} used for builtin types conflicts with
1457 its use for Pascal space types (@pxref{Miscellaneous Types}); they can
1458 be distinguished because the character following the type descriptor
1459 will be a digit, @samp{(}, or @samp{-} for a Pascal space type, or
1460 @samp{u} or @samp{s} for a builtin type.
1463 Documented by AIX to define a wide character type, but their compiler
1464 actually uses negative type numbers (@pxref{Negative Type Numbers}).
1466 @item R @var{fp-type} ; @var{bytes} ;
1467 Define a floating point type. @var{fp-type} has one of the following values:
1471 IEEE 32-bit (single precision) floating point format.
1474 IEEE 64-bit (double precision) floating point format.
1476 @item 3 (NF_COMPLEX)
1477 @item 4 (NF_COMPLEX16)
1478 @item 5 (NF_COMPLEX32)
1479 @c "GDB source" really means @file{include/aout/stab_gnu.h}, but trying
1480 @c to put that here got an overfull hbox.
1481 These are for complex numbers. A comment in the GDB source describes
1482 them as Fortran @code{complex}, @code{double complex}, and
1483 @code{complex*16}, respectively, but what does that mean? (i.e., Single
1484 precision? Double precision?).
1486 @item 6 (NF_LDOUBLE)
1487 Long double. This should probably only be used for Sun format
1488 @code{long double}, and new codes should be used for other floating
1489 point formats (@code{NF_DOUBLE} can be used if a @code{long double} is
1490 really just an IEEE double, of course).
1493 @var{bytes} is the number of bytes occupied by the type. This allows a
1494 debugger to perform some operations with the type even if it doesn't
1495 understand @var{fp-type}.
1497 @item g @var{type-information} ; @var{nbits}
1498 Documented by AIX to define a floating type, but their compiler actually
1499 uses negative type numbers (@pxref{Negative Type Numbers}).
1501 @item c @var{type-information} ; @var{nbits}
1502 Documented by AIX to define a complex type, but their compiler actually
1503 uses negative type numbers (@pxref{Negative Type Numbers}).
1506 The C @code{void} type is defined as a signed integral type 0 bits long:
1508 .stabs "void:t19=bs0;0;0",128,0,0,0
1510 The Solaris compiler seems to omit the trailing semicolon in this case.
1511 Getting sloppy in this way is not a swift move because if a type is
1512 embedded in a more complex expression it is necessary to be able to tell
1515 I'm not sure how a boolean type is represented.
1517 @node Negative Type Numbers
1518 @subsection Negative Type Numbers
1520 This is the method used in XCOFF for defining builtin types.
1521 Since the debugger knows about the builtin types anyway, the idea of
1522 negative type numbers is simply to give a special type number which
1523 indicates the builtin type. There is no stab defining these types.
1525 There are several subtle issues with negative type numbers.
1527 One is the size of the type. A builtin type (for example the C types
1528 @code{int} or @code{long}) might have different sizes depending on
1529 compiler options, the target architecture, the ABI, etc. This issue
1530 doesn't come up for IBM tools since (so far) they just target the
1531 RS/6000; the sizes indicated below for each size are what the IBM
1532 RS/6000 tools use. To deal with differing sizes, either define separate
1533 negative type numbers for each size (which works but requires changing
1534 the debugger, and, unless you get both AIX dbx and GDB to accept the
1535 change, introduces an incompatibility), or use a type attribute
1536 (@pxref{String Field}) to define a new type with the appropriate size
1537 (which merely requires a debugger which understands type attributes,
1538 like AIX dbx or GDB). For example,
1541 .stabs "boolean:t10=@@s8;-16",128,0,0,0
1544 defines an 8-bit boolean type, and
1547 .stabs "boolean:t10=@@s64;-16",128,0,0,0
1550 defines a 64-bit boolean type.
1552 A similar issue is the format of the type. This comes up most often for
1553 floating-point types, which could have various formats (particularly
1554 extended doubles, which vary quite a bit even among IEEE systems).
1555 Again, it is best to define a new negative type number for each
1556 different format; changing the format based on the target system has
1557 various problems. One such problem is that the Alpha has both VAX and
1558 IEEE floating types. One can easily imagine one library using the VAX
1559 types and another library in the same executable using the IEEE types.
1560 Another example is that the interpretation of whether a boolean is true
1561 or false can be based on the least significant bit, most significant
1562 bit, whether it is zero, etc., and different compilers (or different
1563 options to the same compiler) might provide different kinds of boolean.
1565 The last major issue is the names of the types. The name of a given
1566 type depends @emph{only} on the negative type number given; these do not
1567 vary depending on the language, the target system, or anything else.
1568 One can always define separate type numbers---in the following list you
1569 will see for example separate @code{int} and @code{integer*4} types
1570 which are identical except for the name. But compatibility can be
1571 maintained by not inventing new negative type numbers and instead just
1572 defining a new type with a new name. For example:
1575 .stabs "CARDINAL:t10=-8",128,0,0,0
1578 Here is the list of negative type numbers. The phrase @dfn{integral
1579 type} is used to mean twos-complement (I strongly suspect that all
1580 machines which use stabs use twos-complement; most machines use
1581 twos-complement these days).
1585 @code{int}, 32 bit signed integral type.
1588 @code{char}, 8 bit type holding a character. Both GDB and dbx on AIX
1589 treat this as signed. GCC uses this type whether @code{char} is signed
1590 or not, which seems like a bad idea. The AIX compiler (@code{xlc}) seems to
1591 avoid this type; it uses -5 instead for @code{char}.
1594 @code{short}, 16 bit signed integral type.
1597 @code{long}, 32 bit signed integral type.
1600 @code{unsigned char}, 8 bit unsigned integral type.
1603 @code{signed char}, 8 bit signed integral type.
1606 @code{unsigned short}, 16 bit unsigned integral type.
1609 @code{unsigned int}, 32 bit unsigned integral type.
1612 @code{unsigned}, 32 bit unsigned integral type.
1615 @code{unsigned long}, 32 bit unsigned integral type.
1618 @code{void}, type indicating the lack of a value.
1621 @code{float}, IEEE single precision.
1624 @code{double}, IEEE double precision.
1627 @code{long double}, IEEE double precision. The compiler claims the size
1628 will increase in a future release, and for binary compatibility you have
1629 to avoid using @code{long double}. I hope when they increase it they
1630 use a new negative type number.
1633 @code{integer}. 32 bit signed integral type.
1636 @code{boolean}. 32 bit type. GDB and GCC assume that zero is false,
1637 one is true, and other values have unspecified meaning. I hope this
1638 agrees with how the IBM tools use the type.
1641 @code{short real}. IEEE single precision.
1644 @code{real}. IEEE double precision.
1647 @code{stringptr}. @xref{Strings}.
1650 @code{character}, 8 bit unsigned character type.
1653 @code{logical*1}, 8 bit type. This Fortran type has a split
1654 personality in that it is used for boolean variables, but can also be
1655 used for unsigned integers. 0 is false, 1 is true, and other values are
1659 @code{logical*2}, 16 bit type. This Fortran type has a split
1660 personality in that it is used for boolean variables, but can also be
1661 used for unsigned integers. 0 is false, 1 is true, and other values are
1665 @code{logical*4}, 32 bit type. This Fortran type has a split
1666 personality in that it is used for boolean variables, but can also be
1667 used for unsigned integers. 0 is false, 1 is true, and other values are
1671 @code{logical}, 32 bit type. This Fortran type has a split
1672 personality in that it is used for boolean variables, but can also be
1673 used for unsigned integers. 0 is false, 1 is true, and other values are
1677 @code{complex}. A complex type consisting of two IEEE single-precision
1678 floating point values.
1681 @code{complex}. A complex type consisting of two IEEE double-precision
1682 floating point values.
1685 @code{integer*1}, 8 bit signed integral type.
1688 @code{integer*2}, 16 bit signed integral type.
1691 @code{integer*4}, 32 bit signed integral type.
1694 @code{wchar}. Wide character, 16 bits wide, unsigned (what format?
1698 @code{long long}, 64 bit signed integral type.
1701 @code{unsigned long long}, 64 bit unsigned integral type.
1704 @code{logical*8}, 64 bit unsigned integral type.
1707 @code{integer*8}, 64 bit signed integral type.
1710 @node Miscellaneous Types
1711 @section Miscellaneous Types
1714 @item b @var{type-information} ; @var{bytes}
1715 Pascal space type. This is documented by IBM; what does it mean?
1717 This use of the @samp{b} type descriptor can be distinguished
1718 from its use for builtin integral types (@pxref{Builtin Type
1719 Descriptors}) because the character following the type descriptor is
1720 always a digit, @samp{(}, or @samp{-}.
1722 @item B @var{type-information}
1723 A volatile-qualified version of @var{type-information}. This is
1724 a Sun extension. References and stores to a variable with a
1725 volatile-qualified type must not be optimized or cached; they
1726 must occur as the user specifies them.
1728 @item d @var{type-information}
1729 File of type @var{type-information}. As far as I know this is only used
1732 @item k @var{type-information}
1733 A const-qualified version of @var{type-information}. This is a Sun
1734 extension. A variable with a const-qualified type cannot be modified.
1736 @item M @var{type-information} ; @var{length}
1737 Multiple instance type. The type seems to composed of @var{length}
1738 repetitions of @var{type-information}, for example @code{character*3} is
1739 represented by @samp{M-2;3}, where @samp{-2} is a reference to a
1740 character type (@pxref{Negative Type Numbers}). I'm not sure how this
1741 differs from an array. This appears to be a Fortran feature.
1742 @var{length} is a bound, like those in range types; see @ref{Subranges}.
1744 @item S @var{type-information}
1745 Pascal set type. @var{type-information} must be a small type such as an
1746 enumeration or a subrange, and the type is a bitmask whose length is
1747 specified by the number of elements in @var{type-information}.
1749 In CHILL, if it is a bitstring instead of a set, also use the @samp{S}
1750 type attribute (@pxref{String Field}).
1752 @item * @var{type-information}
1753 Pointer to @var{type-information}.
1756 @node Cross-References
1757 @section Cross-References to Other Types
1759 A type can be used before it is defined; one common way to deal with
1760 that situation is just to use a type reference to a type which has not
1763 Another way is with the @samp{x} type descriptor, which is followed by
1764 @samp{s} for a structure tag, @samp{u} for a union tag, or @samp{e} for
1765 a enumerator tag, followed by the name of the tag, followed by @samp{:}.
1766 If the name contains @samp{::} between a @samp{<} and @samp{>} pair (for
1767 C++ templates), such a @samp{::} does not end the name---only a single
1768 @samp{:} ends the name; see @ref{Nested Symbols}.
1770 For example, the following C declarations:
1781 .stabs "bar:G16=*17=xsfoo:",32,0,0,0
1784 Not all debuggers support the @samp{x} type descriptor, so on some
1785 machines GCC does not use it. I believe that for the above example it
1786 would just emit a reference to type 17 and never define it, but I
1787 haven't verified that.
1789 Modula-2 imported types, at least on AIX, use the @samp{i} type
1790 descriptor, which is followed by the name of the module from which the
1791 type is imported, followed by @samp{:}, followed by the name of the
1792 type. There is then optionally a comma followed by type information for
1793 the type. This differs from merely naming the type (@pxref{Typedefs}) in
1794 that it identifies the module; I don't understand whether the name of
1795 the type given here is always just the same as the name we are giving
1796 it, or whether this type descriptor is used with a nameless stab
1797 (@pxref{String Field}), or what. The symbol ends with @samp{;}.
1800 @section Subrange Types
1802 The @samp{r} type descriptor defines a type as a subrange of another
1803 type. It is followed by type information for the type of which it is a
1804 subrange, a semicolon, an integral lower bound, a semicolon, an
1805 integral upper bound, and a semicolon. The AIX documentation does not
1806 specify the trailing semicolon, in an effort to specify array indexes
1807 more cleanly, but a subrange which is not an array index has always
1808 included a trailing semicolon (@pxref{Arrays}).
1810 Instead of an integer, either bound can be one of the following:
1813 @item A @var{offset}
1814 The bound is passed by reference on the stack at offset @var{offset}
1815 from the argument list. @xref{Parameters}, for more information on such
1818 @item T @var{offset}
1819 The bound is passed by value on the stack at offset @var{offset} from
1822 @item a @var{register-number}
1823 The bound is passed by reference in register number
1824 @var{register-number}.
1826 @item t @var{register-number}
1827 The bound is passed by value in register number @var{register-number}.
1833 Subranges are also used for builtin types; see @ref{Traditional Builtin Types}.
1836 @section Array Types
1838 Arrays use the @samp{a} type descriptor. Following the type descriptor
1839 is the type of the index and the type of the array elements. If the
1840 index type is a range type, it ends in a semicolon; otherwise
1841 (for example, if it is a type reference), there does not
1842 appear to be any way to tell where the types are separated. In an
1843 effort to clean up this mess, IBM documents the two types as being
1844 separated by a semicolon, and a range type as not ending in a semicolon
1845 (but this is not right for range types which are not array indexes,
1846 @pxref{Subranges}). I think probably the best solution is to specify
1847 that a semicolon ends a range type, and that the index type and element
1848 type of an array are separated by a semicolon, but that if the index
1849 type is a range type, the extra semicolon can be omitted. GDB (at least
1850 through version 4.9) doesn't support any kind of index type other than a
1851 range anyway; I'm not sure about dbx.
1853 It is well established, and widely used, that the type of the index,
1854 unlike most types found in the stabs, is merely a type definition, not
1855 type information (@pxref{String Field}) (that is, it need not start with
1856 @samp{@var{type-number}=} if it is defining a new type). According to a
1857 comment in GDB, this is also true of the type of the array elements; it
1858 gives @samp{ar1;1;10;ar1;1;10;4} as a legitimate way to express a two
1859 dimensional array. According to AIX documentation, the element type
1860 must be type information. GDB accepts either.
1862 The type of the index is often a range type, expressed as the type
1863 descriptor @samp{r} and some parameters. It defines the size of the
1864 array. In the example below, the range @samp{r1;0;2;} defines an index
1865 type which is a subrange of type 1 (integer), with a lower bound of 0
1866 and an upper bound of 2. This defines the valid range of subscripts of
1867 a three-element C array.
1869 For example, the definition:
1872 char char_vec[3] = @{'a','b','c'@};
1876 produces the output:
1879 .stabs "char_vec:G19=ar1;0;2;2",32,0,0,0
1888 If an array is @dfn{packed}, the elements are spaced more
1889 closely than normal, saving memory at the expense of speed. For
1890 example, an array of 3-byte objects might, if unpacked, have each
1891 element aligned on a 4-byte boundary, but if packed, have no padding.
1892 One way to specify that something is packed is with type attributes
1893 (@pxref{String Field}). In the case of arrays, another is to use the
1894 @samp{P} type descriptor instead of @samp{a}. Other than specifying a
1895 packed array, @samp{P} is identical to @samp{a}.
1897 @c FIXME-what is it? A pointer?
1898 An open array is represented by the @samp{A} type descriptor followed by
1899 type information specifying the type of the array elements.
1901 @c FIXME: what is the format of this type? A pointer to a vector of pointers?
1902 An N-dimensional dynamic array is represented by
1905 D @var{dimensions} ; @var{type-information}
1908 @c Does dimensions really have this meaning? The AIX documentation
1910 @var{dimensions} is the number of dimensions; @var{type-information}
1911 specifies the type of the array elements.
1913 @c FIXME: what is the format of this type? A pointer to some offsets in
1915 A subarray of an N-dimensional array is represented by
1918 E @var{dimensions} ; @var{type-information}
1921 @c Does dimensions really have this meaning? The AIX documentation
1923 @var{dimensions} is the number of dimensions; @var{type-information}
1924 specifies the type of the array elements.
1929 Some languages, like C or the original Pascal, do not have string types,
1930 they just have related things like arrays of characters. But most
1931 Pascals and various other languages have string types, which are
1932 indicated as follows:
1935 @item n @var{type-information} ; @var{bytes}
1936 @var{bytes} is the maximum length. I'm not sure what
1937 @var{type-information} is; I suspect that it means that this is a string
1938 of @var{type-information} (thus allowing a string of integers, a string
1939 of wide characters, etc., as well as a string of characters). Not sure
1940 what the format of this type is. This is an AIX feature.
1942 @item z @var{type-information} ; @var{bytes}
1943 Just like @samp{n} except that this is a gstring, not an ordinary
1944 string. I don't know the difference.
1947 Pascal Stringptr. What is this? This is an AIX feature.
1950 Languages, such as CHILL which have a string type which is basically
1951 just an array of characters use the @samp{S} type attribute
1952 (@pxref{String Field}).
1955 @section Enumerations
1957 Enumerations are defined with the @samp{e} type descriptor.
1959 @c FIXME: Where does this information properly go? Perhaps it is
1960 @c redundant with something we already explain.
1961 The source line below declares an enumeration type at file scope.
1962 The type definition is located after the @code{N_RBRAC} that marks the end of
1963 the previous procedure's block scope, and before the @code{N_FUN} that marks
1964 the beginning of the next procedure's block scope. Therefore it does not
1965 describe a block local symbol, but a file local one.
1970 enum e_places @{first,second=3,last@};
1974 generates the following stab:
1977 .stabs "e_places:T22=efirst:0,second:3,last:4,;",128,0,0,0
1980 The symbol descriptor (@samp{T}) says that the stab describes a
1981 structure, enumeration, or union tag. The type descriptor @samp{e},
1982 following the @samp{22=} of the type definition narrows it down to an
1983 enumeration type. Following the @samp{e} is a list of the elements of
1984 the enumeration. The format is @samp{@var{name}:@var{value},}. The
1985 list of elements ends with @samp{;}. The fact that @var{value} is
1986 specified as an integer can cause problems if the value is large. GCC
1987 2.5.2 tries to output it in octal in that case with a leading zero,
1988 which is probably a good thing, although GDB 4.11 supports octal only in
1989 cases where decimal is perfectly good. Negative decimal values are
1990 supported by both GDB and dbx.
1992 There is no standard way to specify the size of an enumeration type; it
1993 is determined by the architecture (normally all enumerations types are
1994 32 bits). Type attributes can be used to specify an enumeration type of
1995 another size for debuggers which support them; see @ref{String Field}.
1997 Enumeration types are unusual in that they define symbols for the
1998 enumeration values (@code{first}, @code{second}, and @code{third} in the
1999 above example), and even though these symbols are visible in the file as
2000 a whole (rather than being in a more local namespace like structure
2001 member names), they are defined in the type definition for the
2002 enumeration type rather than each having their own symbol. In order to
2003 be fast, GDB will only get symbols from such types (in its initial scan
2004 of the stabs) if the type is the first thing defined after a @samp{T} or
2005 @samp{t} symbol descriptor (the above example fulfills this
2006 requirement). If the type does not have a name, the compiler should
2007 emit it in a nameless stab (@pxref{String Field}); GCC does this.
2012 The encoding of structures in stabs can be shown with an example.
2014 The following source code declares a structure tag and defines an
2015 instance of the structure in global scope. Then a @code{typedef} equates the
2016 structure tag with a new type. Separate stabs are generated for the
2017 structure tag, the structure @code{typedef}, and the structure instance. The
2018 stabs for the tag and the @code{typedef} are emitted when the definitions are
2019 encountered. Since the structure elements are not initialized, the
2020 stab and code for the structure variable itself is located at the end
2021 of the program in the bss section.
2028 struct s_tag* s_next;
2031 typedef struct s_tag s_typedef;
2034 The structure tag has an @code{N_LSYM} stab type because, like the
2035 enumeration, the symbol has file scope. Like the enumeration, the
2036 symbol descriptor is @samp{T}, for enumeration, structure, or tag type.
2037 The type descriptor @samp{s} following the @samp{16=} of the type
2038 definition narrows the symbol type to structure.
2040 Following the @samp{s} type descriptor is the number of bytes the
2041 structure occupies, followed by a description of each structure element.
2042 The structure element descriptions are of the form @var{name:type, bit
2043 offset from the start of the struct, number of bits in the element}.
2045 @c FIXME: phony line break. Can probably be fixed by using an example
2046 @c with fewer fields.
2049 .stabs "s_tag:T16=s20s_int:1,0,32;s_float:12,32,32;
2050 s_char_vec:17=ar1;0;7;2,64,64;s_next:18=*16,128,32;;",128,0,0,0
2053 In this example, the first two structure elements are previously defined
2054 types. For these, the type following the @samp{@var{name}:} part of the
2055 element description is a simple type reference. The other two structure
2056 elements are new types. In this case there is a type definition
2057 embedded after the @samp{@var{name}:}. The type definition for the
2058 array element looks just like a type definition for a stand-alone array.
2059 The @code{s_next} field is a pointer to the same kind of structure that
2060 the field is an element of. So the definition of structure type 16
2061 contains a type definition for an element which is a pointer to type 16.
2063 If a field is a static member (this is a C++ feature in which a single
2064 variable appears to be a field of every structure of a given type) it
2065 still starts out with the field name, a colon, and the type, but then
2066 instead of a comma, bit position, comma, and bit size, there is a colon
2067 followed by the name of the variable which each such field refers to.
2069 If the structure has methods (a C++ feature), they follow the non-method
2070 fields; see @ref{Cplusplus}.
2073 @section Giving a Type a Name
2075 @findex N_LSYM, for types
2076 @findex C_DECL, for types
2077 To give a type a name, use the @samp{t} symbol descriptor. The type
2078 is specified by the type information (@pxref{String Field}) for the stab.
2082 .stabs "s_typedef:t16",128,0,0,0 # @r{128 is N_LSYM}
2085 specifies that @code{s_typedef} refers to type number 16. Such stabs
2086 have symbol type @code{N_LSYM} (or @code{C_DECL} for XCOFF). (The Sun
2087 documentation mentions using @code{N_GSYM} in some cases).
2089 If you are specifying the tag name for a structure, union, or
2090 enumeration, use the @samp{T} symbol descriptor instead. I believe C is
2091 the only language with this feature.
2093 If the type is an opaque type (I believe this is a Modula-2 feature),
2094 AIX provides a type descriptor to specify it. The type descriptor is
2095 @samp{o} and is followed by a name. I don't know what the name
2096 means---is it always the same as the name of the type, or is this type
2097 descriptor used with a nameless stab (@pxref{String Field})? There
2098 optionally follows a comma followed by type information which defines
2099 the type of this type. If omitted, a semicolon is used in place of the
2100 comma and the type information, and the type is much like a generic
2101 pointer type---it has a known size but little else about it is
2115 This code generates a stab for a union tag and a stab for a union
2116 variable. Both use the @code{N_LSYM} stab type. If a union variable is
2117 scoped locally to the procedure in which it is defined, its stab is
2118 located immediately preceding the @code{N_LBRAC} for the procedure's block
2121 The stab for the union tag, however, is located preceding the code for
2122 the procedure in which it is defined. The stab type is @code{N_LSYM}. This
2123 would seem to imply that the union type is file scope, like the struct
2124 type @code{s_tag}. This is not true. The contents and position of the stab
2125 for @code{u_type} do not convey any information about its procedure local
2128 @c FIXME: phony line break. Can probably be fixed by using an example
2129 @c with fewer fields.
2132 .stabs "u_tag:T23=u4u_int:1,0,32;u_float:12,0,32;u_char:21,0,32;;",
2136 The symbol descriptor @samp{T}, following the @samp{name:} means that
2137 the stab describes an enumeration, structure, or union tag. The type
2138 descriptor @samp{u}, following the @samp{23=} of the type definition,
2139 narrows it down to a union type definition. Following the @samp{u} is
2140 the number of bytes in the union. After that is a list of union element
2141 descriptions. Their format is @var{name:type, bit offset into the
2142 union, number of bytes for the element;}.
2144 The stab for the union variable is:
2147 .stabs "an_u:23",128,0,0,-20 # @r{128 is N_LSYM}
2150 @samp{-20} specifies where the variable is stored (@pxref{Stack
2153 @node Function Types
2154 @section Function Types
2156 Various types can be defined for function variables. These types are
2157 not used in defining functions (@pxref{Procedures}); they are used for
2158 things like pointers to functions.
2160 The simple, traditional, type is type descriptor @samp{f} is followed by
2161 type information for the return type of the function, followed by a
2164 This does not deal with functions for which the number and types of the
2165 parameters are part of the type, as in Modula-2 or ANSI C. AIX provides
2166 extensions to specify these, using the @samp{f}, @samp{F}, @samp{p}, and
2167 @samp{R} type descriptors.
2169 First comes the type descriptor. If it is @samp{f} or @samp{F}, this
2170 type involves a function rather than a procedure, and the type
2171 information for the return type of the function follows, followed by a
2172 comma. Then comes the number of parameters to the function and a
2173 semicolon. Then, for each parameter, there is the name of the parameter
2174 followed by a colon (this is only present for type descriptors @samp{R}
2175 and @samp{F} which represent Pascal function or procedure parameters),
2176 type information for the parameter, a comma, 0 if passed by reference or
2177 1 if passed by value, and a semicolon. The type definition ends with a
2180 For example, this variable definition:
2187 generates the following code:
2190 .stabs "g_pf:G24=*25=f1",32,0,0,0
2191 .common _g_pf,4,"bss"
2194 The variable defines a new type, 24, which is a pointer to another new
2195 type, 25, which is a function returning @code{int}.
2198 @chapter Symbol Information in Symbol Tables
2200 This chapter describes the format of symbol table entries
2201 and how stab assembler directives map to them. It also describes the
2202 transformations that the assembler and linker make on data from stabs.
2205 * Symbol Table Format::
2206 * Transformations On Symbol Tables::
2209 @node Symbol Table Format
2210 @section Symbol Table Format
2212 Each time the assembler encounters a stab directive, it puts
2213 each field of the stab into a corresponding field in a symbol table
2214 entry of its output file. If the stab contains a string field, the
2215 symbol table entry for that stab points to a string table entry
2216 containing the string data from the stab. Assembler labels become
2217 relocatable addresses. Symbol table entries in a.out have the format:
2219 @c FIXME: should refer to external, not internal.
2221 struct internal_nlist @{
2222 unsigned long n_strx; /* index into string table of name */
2223 unsigned char n_type; /* type of symbol */
2224 unsigned char n_other; /* misc info (usually empty) */
2225 unsigned short n_desc; /* description field */
2226 bfd_vma n_value; /* value of symbol */
2230 If the stab has a string, the @code{n_strx} field holds the offset in
2231 bytes of the string within the string table. The string is terminated
2232 by a NUL character. If the stab lacks a string (for example, it was
2233 produced by a @code{.stabn} or @code{.stabd} directive), the
2234 @code{n_strx} field is zero.
2236 Symbol table entries with @code{n_type} field values greater than 0x1f
2237 originated as stabs generated by the compiler (with one random
2238 exception). The other entries were placed in the symbol table of the
2239 executable by the assembler or the linker.
2241 @node Transformations On Symbol Tables
2242 @section Transformations on Symbol Tables
2244 The linker concatenates object files and does fixups of externally
2247 You can see the transformations made on stab data by the assembler and
2248 linker by examining the symbol table after each pass of the build. To
2249 do this, use @samp{nm -ap}, which dumps the symbol table, including
2250 debugging information, unsorted. For stab entries the columns are:
2251 @var{value}, @var{other}, @var{desc}, @var{type}, @var{string}. For
2252 assembler and linker symbols, the columns are: @var{value}, @var{type},
2255 The low 5 bits of the stab type tell the linker how to relocate the
2256 value of the stab. Thus for stab types like @code{N_RSYM} and
2257 @code{N_LSYM}, where the value is an offset or a register number, the
2258 low 5 bits are @code{N_ABS}, which tells the linker not to relocate the
2261 Where the value of a stab contains an assembly language label,
2262 it is transformed by each build step. The assembler turns it into a
2263 relocatable address and the linker turns it into an absolute address.
2266 * Transformations On Static Variables::
2267 * Transformations On Global Variables::
2268 * Stab Section Transformations:: For some object file formats,
2269 things are a bit different.
2272 @node Transformations On Static Variables
2273 @subsection Transformations on Static Variables
2275 This source line defines a static variable at file scope:
2278 static int s_g_repeat
2282 The following stab describes the symbol:
2285 .stabs "s_g_repeat:S1",38,0,0,_s_g_repeat
2289 The assembler transforms the stab into this symbol table entry in the
2290 @file{.o} file. The location is expressed as a data segment offset.
2293 00000084 - 00 0000 STSYM s_g_repeat:S1
2297 In the symbol table entry from the executable, the linker has made the
2298 relocatable address absolute.
2301 0000e00c - 00 0000 STSYM s_g_repeat:S1
2304 @node Transformations On Global Variables
2305 @subsection Transformations on Global Variables
2307 Stabs for global variables do not contain location information. In
2308 this case, the debugger finds location information in the assembler or
2309 linker symbol table entry describing the variable. The source line:
2319 .stabs "g_foo:G2",32,0,0,0
2322 The variable is represented by two symbol table entries in the object
2323 file (see below). The first one originated as a stab. The second one
2324 is an external symbol. The upper case @samp{D} signifies that the
2325 @code{n_type} field of the symbol table contains 7, @code{N_DATA} with
2326 local linkage. The stab's value is zero since the value is not used for
2327 @code{N_GSYM} stabs. The value of the linker symbol is the relocatable
2328 address corresponding to the variable.
2331 00000000 - 00 0000 GSYM g_foo:G2
2336 These entries as transformed by the linker. The linker symbol table
2337 entry now holds an absolute address:
2340 00000000 - 00 0000 GSYM g_foo:G2
2345 @node Stab Section Transformations
2346 @subsection Transformations of Stabs in separate sections
2348 For object file formats using stabs in separate sections (@pxref{Stab
2349 Sections}), use @code{objdump --stabs} instead of @code{nm} to show the
2350 stabs in an object or executable file. @code{objdump} is a GNU utility;
2351 Sun does not provide any equivalent.
2353 The following example is for a stab whose value is an address is
2354 relative to the compilation unit (@pxref{ELF Linker Relocation}). For
2355 example, if the source line
2361 appears within a function, then the assembly language output from the
2367 .stabs "ld:V(0,3)",0x26,0,4,.L18-Ddata.data # @r{0x26 is N_STSYM}
2374 Because the value is formed by subtracting one symbol from another, the
2375 value is absolute, not relocatable, and so the object file contains
2378 Symnum n_type n_othr n_desc n_value n_strx String
2379 31 STSYM 0 4 00000004 680 ld:V(0,3)
2382 without any relocations, and the executable file also contains
2385 Symnum n_type n_othr n_desc n_value n_strx String
2386 31 STSYM 0 4 00000004 680 ld:V(0,3)
2390 @chapter GNU C++ Stabs
2393 * Class Names:: C++ class names are both tags and typedefs.
2394 * Nested Symbols:: C++ symbol names can be within other types.
2395 * Basic Cplusplus Types::
2398 * Methods:: Method definition
2399 * Method Type Descriptor:: The @samp{#} type descriptor
2400 * Member Type Descriptor:: The @samp{@@} type descriptor
2402 * Method Modifiers::
2405 * Virtual Base Classes::
2410 @section C++ Class Names
2412 In C++, a class name which is declared with @code{class}, @code{struct},
2413 or @code{union}, is not only a tag, as in C, but also a type name. Thus
2414 there should be stabs with both @samp{t} and @samp{T} symbol descriptors
2417 To save space, there is a special abbreviation for this case. If the
2418 @samp{T} symbol descriptor is followed by @samp{t}, then the stab
2419 defines both a type name and a tag.
2421 For example, the C++ code
2424 struct foo @{int x;@};
2427 can be represented as either
2430 .stabs "foo:T19=s4x:1,0,32;;",128,0,0,0 # @r{128 is N_LSYM}
2431 .stabs "foo:t19",128,0,0,0
2437 .stabs "foo:Tt19=s4x:1,0,32;;",128,0,0,0
2440 @node Nested Symbols
2441 @section Defining a Symbol Within Another Type
2443 In C++, a symbol (such as a type name) can be defined within another type.
2444 @c FIXME: Needs example.
2446 In stabs, this is sometimes represented by making the name of a symbol
2447 which contains @samp{::}. Such a pair of colons does not end the name
2448 of the symbol, the way a single colon would (@pxref{String Field}). I'm
2449 not sure how consistently used or well thought out this mechanism is.
2450 So that a pair of colons in this position always has this meaning,
2451 @samp{:} cannot be used as a symbol descriptor.
2453 For example, if the string for a stab is @samp{foo::bar::baz:t5=*6},
2454 then @code{foo::bar::baz} is the name of the symbol, @samp{t} is the
2455 symbol descriptor, and @samp{5=*6} is the type information.
2457 @node Basic Cplusplus Types
2458 @section Basic Types For C++
2460 << the examples that follow are based on a01.C >>
2463 C++ adds two more builtin types to the set defined for C. These are
2464 the unknown type and the vtable record type. The unknown type, type
2465 16, is defined in terms of itself like the void type.
2467 The vtable record type, type 17, is defined as a structure type and
2468 then as a structure tag. The structure has four fields: delta, index,
2469 pfn, and delta2. pfn is the function pointer.
2471 << In boilerplate $vtbl_ptr_type, what are the fields delta,
2472 index, and delta2 used for? >>
2474 This basic type is present in all C++ programs even if there are no
2475 virtual methods defined.
2478 .stabs "struct_name:sym_desc(type)type_def(17)=type_desc(struct)struct_bytes(8)
2479 elem_name(delta):type_ref(short int),bit_offset(0),field_bits(16);
2480 elem_name(index):type_ref(short int),bit_offset(16),field_bits(16);
2481 elem_name(pfn):type_def(18)=type_desc(ptr to)type_ref(void),
2482 bit_offset(32),field_bits(32);
2483 elem_name(delta2):type_def(short int);bit_offset(32),field_bits(16);;"
2488 .stabs "$vtbl_ptr_type:t17=s8
2489 delta:6,0,16;index:6,16,16;pfn:18=*15,32,32;delta2:6,32,16;;"
2494 .stabs "name:sym_dec(struct tag)type_ref($vtbl_ptr_type)",N_LSYM,NIL,NIL,NIL
2498 .stabs "$vtbl_ptr_type:T17",128,0,0,0
2501 @node Simple Classes
2502 @section Simple Class Definition
2504 The stabs describing C++ language features are an extension of the
2505 stabs describing C. Stabs representing C++ class types elaborate
2506 extensively on the stab format used to describe structure types in C.
2507 Stabs representing class type variables look just like stabs
2508 representing C language variables.
2510 Consider the following very simple class definition.
2516 int Ameth(int in, char other);
2520 The class @code{baseA} is represented by two stabs. The first stab describes
2521 the class as a structure type. The second stab describes a structure
2522 tag of the class type. Both stabs are of stab type @code{N_LSYM}. Since the
2523 stab is not located between an @code{N_FUN} and an @code{N_LBRAC} stab this indicates
2524 that the class is defined at file scope. If it were, then the @code{N_LSYM}
2525 would signify a local variable.
2527 A stab describing a C++ class type is similar in format to a stab
2528 describing a C struct, with each class member shown as a field in the
2529 structure. The part of the struct format describing fields is
2530 expanded to include extra information relevant to C++ class members.
2531 In addition, if the class has multiple base classes or virtual
2532 functions the struct format outside of the field parts is also
2535 In this simple example the field part of the C++ class stab
2536 representing member data looks just like the field part of a C struct
2537 stab. The section on protections describes how its format is
2538 sometimes extended for member data.
2540 The field part of a C++ class stab representing a member function
2541 differs substantially from the field part of a C struct stab. It
2542 still begins with @samp{name:} but then goes on to define a new type number
2543 for the member function, describe its return type, its argument types,
2544 its protection level, any qualifiers applied to the method definition,
2545 and whether the method is virtual or not. If the method is virtual
2546 then the method description goes on to give the vtable index of the
2547 method, and the type number of the first base class defining the
2550 When the field name is a method name it is followed by two colons rather
2551 than one. This is followed by a new type definition for the method.
2552 This is a number followed by an equal sign and the type of the method.
2553 Normally this will be a type declared using the @samp{#} type
2554 descriptor; see @ref{Method Type Descriptor}; static member functions
2555 are declared using the @samp{f} type descriptor instead; see
2556 @ref{Function Types}.
2558 The format of an overloaded operator method name differs from that of
2559 other methods. It is @samp{op$::@var{operator-name}.} where
2560 @var{operator-name} is the operator name such as @samp{+} or @samp{+=}.
2561 The name ends with a period, and any characters except the period can
2562 occur in the @var{operator-name} string.
2564 The next part of the method description represents the arguments to the
2565 method, preceded by a colon and ending with a semi-colon. The types of
2566 the arguments are expressed in the same way argument types are expressed
2567 in C++ name mangling. In this example an @code{int} and a @code{char}
2570 This is followed by a number, a letter, and an asterisk or period,
2571 followed by another semicolon. The number indicates the protections
2572 that apply to the member function. Here the 2 means public. The
2573 letter encodes any qualifier applied to the method definition. In
2574 this case, @samp{A} means that it is a normal function definition. The dot
2575 shows that the method is not virtual. The sections that follow
2576 elaborate further on these fields and describe the additional
2577 information present for virtual methods.
2581 .stabs "class_name:sym_desc(type)type_def(20)=type_desc(struct)struct_bytes(4)
2582 field_name(Adat):type(int),bit_offset(0),field_bits(32);
2584 method_name(Ameth)::type_def(21)=type_desc(method)return_type(int);
2585 :arg_types(int char);
2586 protection(public)qualifier(normal)virtual(no);;"
2591 .stabs "baseA:t20=s4Adat:1,0,32;Ameth::21=##1;:ic;2A.;;",128,0,0,0
2593 .stabs "class_name:sym_desc(struct tag)",N_LSYM,NIL,NIL,NIL
2595 .stabs "baseA:T20",128,0,0,0
2598 @node Class Instance
2599 @section Class Instance
2601 As shown above, describing even a simple C++ class definition is
2602 accomplished by massively extending the stab format used in C to
2603 describe structure types. However, once the class is defined, C stabs
2604 with no modifications can be used to describe class instances. The
2614 yields the following stab describing the class instance. It looks no
2615 different from a standard C stab describing a local variable.
2618 .stabs "name:type_ref(baseA)", N_LSYM, NIL, NIL, frame_ptr_offset
2622 .stabs "AbaseA:20",128,0,0,-20
2626 @section Method Definition
2628 The class definition shown above declares Ameth. The C++ source below
2633 baseA::Ameth(int in, char other)
2640 This method definition yields three stabs following the code of the
2641 method. One stab describes the method itself and following two describe
2642 its parameters. Although there is only one formal argument all methods
2643 have an implicit argument which is the @code{this} pointer. The @code{this}
2644 pointer is a pointer to the object on which the method was called. Note
2645 that the method name is mangled to encode the class name and argument
2646 types. Name mangling is described in the @sc{arm} (@cite{The Annotated
2647 C++ Reference Manual}, by Ellis and Stroustrup, @sc{isbn}
2648 0-201-51459-1); @file{gpcompare.texi} in Cygnus GCC distributions
2649 describes the differences between GNU mangling and @sc{arm}
2651 @c FIXME: Use @xref, especially if this is generally installed in the
2653 @c FIXME: This information should be in a net release, either of GCC or
2654 @c GDB. But gpcompare.texi doesn't seem to be in the FSF GCC.
2657 .stabs "name:symbol_descriptor(global function)return_type(int)",
2658 N_FUN, NIL, NIL, code_addr_of_method_start
2660 .stabs "Ameth__5baseAic:F1",36,0,0,_Ameth__5baseAic
2663 Here is the stab for the @code{this} pointer implicit argument. The
2664 name of the @code{this} pointer is always @code{this}. Type 19, the
2665 @code{this} pointer is defined as a pointer to type 20, @code{baseA},
2666 but a stab defining @code{baseA} has not yet been emitted. Since the
2667 compiler knows it will be emitted shortly, here it just outputs a cross
2668 reference to the undefined symbol, by prefixing the symbol name with
2672 .stabs "name:sym_desc(register param)type_def(19)=
2673 type_desc(ptr to)type_ref(baseA)=
2674 type_desc(cross-reference to)baseA:",N_RSYM,NIL,NIL,register_number
2676 .stabs "this:P19=*20=xsbaseA:",64,0,0,8
2679 The stab for the explicit integer argument looks just like a parameter
2680 to a C function. The last field of the stab is the offset from the
2681 argument pointer, which in most systems is the same as the frame
2685 .stabs "name:sym_desc(value parameter)type_ref(int)",
2686 N_PSYM,NIL,NIL,offset_from_arg_ptr
2688 .stabs "in:p1",160,0,0,72
2691 << The examples that follow are based on A1.C >>
2693 @node Method Type Descriptor
2694 @section The @samp{#} Type Descriptor
2696 This is used to describe a class method. This is a function which takes
2697 an extra argument as its first argument, for the @code{this} pointer.
2699 If the @samp{#} is immediately followed by another @samp{#}, the second
2700 one will be followed by the return type and a semicolon. The class and
2701 argument types are not specified, and must be determined by demangling
2702 the name of the method if it is available.
2704 Otherwise, the single @samp{#} is followed by the class type, a comma,
2705 the return type, a comma, and zero or more parameter types separated by
2706 commas. The list of arguments is terminated by a semicolon. In the
2707 debugging output generated by gcc, a final argument type of @code{void}
2708 indicates a method which does not take a variable number of arguments.
2709 If the final argument type of @code{void} does not appear, the method
2710 was declared with an ellipsis.
2712 Note that although such a type will normally be used to describe fields
2713 in structures, unions, or classes, for at least some versions of the
2714 compiler it can also be used in other contexts.
2716 @node Member Type Descriptor
2717 @section The @samp{@@} Type Descriptor
2719 The @samp{@@} type descriptor is for a member (class and variable) type.
2720 It is followed by type information for the offset basetype, a comma, and
2721 type information for the type of the field being pointed to. (FIXME:
2722 this is acknowledged to be gibberish. Can anyone say what really goes
2725 Note that there is a conflict between this and type attributes
2726 (@pxref{String Field}); both use type descriptor @samp{@@}.
2727 Fortunately, the @samp{@@} type descriptor used in this C++ sense always
2728 will be followed by a digit, @samp{(}, or @samp{-}, and type attributes
2729 never start with those things.
2732 @section Protections
2734 In the simple class definition shown above all member data and
2735 functions were publicly accessible. The example that follows
2736 contrasts public, protected and privately accessible fields and shows
2737 how these protections are encoded in C++ stabs.
2739 If the character following the @samp{@var{field-name}:} part of the
2740 string is @samp{/}, then the next character is the visibility. @samp{0}
2741 means private, @samp{1} means protected, and @samp{2} means public.
2742 Debuggers should ignore visibility characters they do not recognize, and
2743 assume a reasonable default (such as public) (GDB 4.11 does not, but
2744 this should be fixed in the next GDB release). If no visibility is
2745 specified the field is public. The visibility @samp{9} means that the
2746 field has been optimized out and is public (there is no way to specify
2747 an optimized out field with a private or protected visibility).
2748 Visibility @samp{9} is not supported by GDB 4.11; this should be fixed
2749 in the next GDB release.
2751 The following C++ source:
2765 generates the following stab:
2769 .stabs "vis:T19=s12priv:/01,0,32;prot:/12,32,8;pub:12,64,32;;",128,0,0,0
2772 @samp{vis:T19=s12} indicates that type number 19 is a 12 byte structure
2773 named @code{vis} The @code{priv} field has public visibility
2774 (@samp{/0}), type int (@samp{1}), and offset and size @samp{,0,32;}.
2775 The @code{prot} field has protected visibility (@samp{/1}), type char
2776 (@samp{2}) and offset and size @samp{,32,8;}. The @code{pub} field has
2777 type float (@samp{12}), and offset and size @samp{,64,32;}.
2779 Protections for member functions are signified by one digit embedded in
2780 the field part of the stab describing the method. The digit is 0 if
2781 private, 1 if protected and 2 if public. Consider the C++ class
2785 class all_methods @{
2787 int priv_meth(int in)@{return in;@};
2789 char protMeth(char in)@{return in;@};
2791 float pubMeth(float in)@{return in;@};
2795 It generates the following stab. The digit in question is to the left
2796 of an @samp{A} in each case. Notice also that in this case two symbol
2797 descriptors apply to the class name struct tag and struct type.
2800 .stabs "class_name:sym_desc(struct tag&type)type_def(21)=
2801 sym_desc(struct)struct_bytes(1)
2802 meth_name::type_def(22)=sym_desc(method)returning(int);
2803 :args(int);protection(private)modifier(normal)virtual(no);
2804 meth_name::type_def(23)=sym_desc(method)returning(char);
2805 :args(char);protection(protected)modifier(normal)virtual(no);
2806 meth_name::type_def(24)=sym_desc(method)returning(float);
2807 :args(float);protection(public)modifier(normal)virtual(no);;",
2812 .stabs "all_methods:Tt21=s1priv_meth::22=##1;:i;0A.;protMeth::23=##2;:c;1A.;
2813 pubMeth::24=##12;:f;2A.;;",128,0,0,0
2816 @node Method Modifiers
2817 @section Method Modifiers (@code{const}, @code{volatile}, @code{const volatile})
2821 In the class example described above all the methods have the normal
2822 modifier. This method modifier information is located just after the
2823 protection information for the method. This field has four possible
2824 character values. Normal methods use @samp{A}, const methods use
2825 @samp{B}, volatile methods use @samp{C}, and const volatile methods use
2826 @samp{D}. Consider the class definition below:
2831 int ConstMeth (int arg) const @{ return arg; @};
2832 char VolatileMeth (char arg) volatile @{ return arg; @};
2833 float ConstVolMeth (float arg) const volatile @{return arg; @};
2837 This class is described by the following stab:
2840 .stabs "class(A):sym_desc(struct)type_def(20)=type_desc(struct)struct_bytes(1)
2841 meth_name(ConstMeth)::type_def(21)sym_desc(method)
2842 returning(int);:arg(int);protection(public)modifier(const)virtual(no);
2843 meth_name(VolatileMeth)::type_def(22)=sym_desc(method)
2844 returning(char);:arg(char);protection(public)modifier(volatile)virt(no)
2845 meth_name(ConstVolMeth)::type_def(23)=sym_desc(method)
2846 returning(float);:arg(float);protection(public)modifier(const volatile)
2847 virtual(no);;", @dots{}
2851 .stabs "A:T20=s1ConstMeth::21=##1;:i;2B.;VolatileMeth::22=##2;:c;2C.;
2852 ConstVolMeth::23=##12;:f;2D.;;",128,0,0,0
2855 @node Virtual Methods
2856 @section Virtual Methods
2858 << The following examples are based on a4.C >>
2860 The presence of virtual methods in a class definition adds additional
2861 data to the class description. The extra data is appended to the
2862 description of the virtual method and to the end of the class
2863 description. Consider the class definition below:
2869 virtual int A_virt (int arg) @{ return arg; @};
2873 This results in the stab below describing class A. It defines a new
2874 type (20) which is an 8 byte structure. The first field of the class
2875 struct is @samp{Adat}, an integer, starting at structure offset 0 and
2878 The second field in the class struct is not explicitly defined by the
2879 C++ class definition but is implied by the fact that the class
2880 contains a virtual method. This field is the vtable pointer. The
2881 name of the vtable pointer field starts with @samp{$vf} and continues with a
2882 type reference to the class it is part of. In this example the type
2883 reference for class A is 20 so the name of its vtable pointer field is
2884 @samp{$vf20}, followed by the usual colon.
2886 Next there is a type definition for the vtable pointer type (21).
2887 This is in turn defined as a pointer to another new type (22).
2889 Type 22 is the vtable itself, which is defined as an array, indexed by
2890 a range of integers between 0 and 1, and whose elements are of type
2891 17. Type 17 was the vtable record type defined by the boilerplate C++
2892 type definitions, as shown earlier.
2894 The bit offset of the vtable pointer field is 32. The number of bits
2895 in the field are not specified when the field is a vtable pointer.
2897 Next is the method definition for the virtual member function @code{A_virt}.
2898 Its description starts out using the same format as the non-virtual
2899 member functions described above, except instead of a dot after the
2900 @samp{A} there is an asterisk, indicating that the function is virtual.
2901 Since is is virtual some addition information is appended to the end
2902 of the method description.
2904 The first number represents the vtable index of the method. This is a
2905 32 bit unsigned number with the high bit set, followed by a
2908 The second number is a type reference to the first base class in the
2909 inheritance hierarchy defining the virtual member function. In this
2910 case the class stab describes a base class so the virtual function is
2911 not overriding any other definition of the method. Therefore the
2912 reference is to the type number of the class that the stab is
2915 This is followed by three semi-colons. One marks the end of the
2916 current sub-section, one marks the end of the method field, and the
2917 third marks the end of the struct definition.
2919 For classes containing virtual functions the very last section of the
2920 string part of the stab holds a type reference to the first base
2921 class. This is preceded by @samp{~%} and followed by a final semi-colon.
2924 .stabs "class_name(A):type_def(20)=sym_desc(struct)struct_bytes(8)
2925 field_name(Adat):type_ref(int),bit_offset(0),field_bits(32);
2926 field_name(A virt func ptr):type_def(21)=type_desc(ptr to)type_def(22)=
2927 sym_desc(array)index_type_ref(range of int from 0 to 1);
2928 elem_type_ref(vtbl elem type),
2930 meth_name(A_virt)::typedef(23)=sym_desc(method)returning(int);
2931 :arg_type(int),protection(public)normal(yes)virtual(yes)
2932 vtable_index(1);class_first_defining(A);;;~%first_base(A);",
2936 @c FIXME: bogus line break.
2938 .stabs "A:t20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
2939 A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
2943 @section Inheritance
2945 Stabs describing C++ derived classes include additional sections that
2946 describe the inheritance hierarchy of the class. A derived class stab
2947 also encodes the number of base classes. For each base class it tells
2948 if the base class is virtual or not, and if the inheritance is private
2949 or public. It also gives the offset into the object of the portion of
2950 the object corresponding to each base class.
2952 This additional information is embedded in the class stab following the
2953 number of bytes in the struct. First the number of base classes
2954 appears bracketed by an exclamation point and a comma.
2956 Then for each base type there repeats a series: a virtual character, a
2957 visibility character, a number, a comma, another number, and a
2960 The virtual character is @samp{1} if the base class is virtual and
2961 @samp{0} if not. The visibility character is @samp{2} if the derivation
2962 is public, @samp{1} if it is protected, and @samp{0} if it is private.
2963 Debuggers should ignore virtual or visibility characters they do not
2964 recognize, and assume a reasonable default (such as public and
2965 non-virtual) (GDB 4.11 does not, but this should be fixed in the next
2968 The number following the virtual and visibility characters is the offset
2969 from the start of the object to the part of the object pertaining to the
2972 After the comma, the second number is a type_descriptor for the base
2973 type. Finally a semi-colon ends the series, which repeats for each
2976 The source below defines three base classes @code{A}, @code{B}, and
2977 @code{C} and the derived class @code{D}.
2984 virtual int A_virt (int arg) @{ return arg; @};
2990 virtual int B_virt (int arg) @{return arg; @};
2996 virtual int C_virt (int arg) @{return arg; @};
2999 class D : A, virtual B, public C @{
3002 virtual int A_virt (int arg ) @{ return arg+1; @};
3003 virtual int B_virt (int arg) @{ return arg+2; @};
3004 virtual int C_virt (int arg) @{ return arg+3; @};
3005 virtual int D_virt (int arg) @{ return arg; @};
3009 Class stabs similar to the ones described earlier are generated for
3012 @c FIXME!!! the linebreaks in the following example probably make the
3013 @c examples literally unusable, but I don't know any other way to get
3014 @c them on the page.
3015 @c One solution would be to put some of the type definitions into
3016 @c separate stabs, even if that's not exactly what the compiler actually
3019 .stabs "A:T20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
3020 A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
3022 .stabs "B:Tt25=s8Bdat:1,0,32;$vf25:21,32;B_virt::26=##1;
3023 :i;2A*-2147483647;25;;;~%25;",128,0,0,0
3025 .stabs "C:Tt28=s8Cdat:1,0,32;$vf28:21,32;C_virt::29=##1;
3026 :i;2A*-2147483647;28;;;~%28;",128,0,0,0
3029 In the stab describing derived class @code{D} below, the information about
3030 the derivation of this class is encoded as follows.
3033 .stabs "derived_class_name:symbol_descriptors(struct tag&type)=
3034 type_descriptor(struct)struct_bytes(32)!num_bases(3),
3035 base_virtual(no)inheritance_public(no)base_offset(0),
3036 base_class_type_ref(A);
3037 base_virtual(yes)inheritance_public(no)base_offset(NIL),
3038 base_class_type_ref(B);
3039 base_virtual(no)inheritance_public(yes)base_offset(64),
3040 base_class_type_ref(C); @dots{}
3043 @c FIXME! fake linebreaks.
3045 .stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:
3046 1,160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt:
3047 :32:i;2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;
3048 28;;D_virt::32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
3051 @node Virtual Base Classes
3052 @section Virtual Base Classes
3054 A derived class object consists of a concatenation in memory of the data
3055 areas defined by each base class, starting with the leftmost and ending
3056 with the rightmost in the list of base classes. The exception to this
3057 rule is for virtual inheritance. In the example above, class @code{D}
3058 inherits virtually from base class @code{B}. This means that an
3059 instance of a @code{D} object will not contain its own @code{B} part but
3060 merely a pointer to a @code{B} part, known as a virtual base pointer.
3062 In a derived class stab, the base offset part of the derivation
3063 information, described above, shows how the base class parts are
3064 ordered. The base offset for a virtual base class is always given as 0.
3065 Notice that the base offset for @code{B} is given as 0 even though
3066 @code{B} is not the first base class. The first base class @code{A}
3069 The field information part of the stab for class @code{D} describes the field
3070 which is the pointer to the virtual base class @code{B}. The vbase pointer
3071 name is @samp{$vb} followed by a type reference to the virtual base class.
3072 Since the type id for @code{B} in this example is 25, the vbase pointer name
3075 @c FIXME!! fake linebreaks below
3077 .stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:1,
3078 160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt::32:i;
3079 2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;28;;D_virt:
3080 :32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
3083 Following the name and a semicolon is a type reference describing the
3084 type of the virtual base class pointer, in this case 24. Type 24 was
3085 defined earlier as the type of the @code{B} class @code{this} pointer. The
3086 @code{this} pointer for a class is a pointer to the class type.
3089 .stabs "this:P24=*25=xsB:",64,0,0,8
3092 Finally the field offset part of the vbase pointer field description
3093 shows that the vbase pointer is the first field in the @code{D} object,
3094 before any data fields defined by the class. The layout of a @code{D}
3095 class object is a follows, @code{Adat} at 0, the vtable pointer for
3096 @code{A} at 32, @code{Cdat} at 64, the vtable pointer for C at 96, the
3097 virtual base pointer for @code{B} at 128, and @code{Ddat} at 160.
3100 @node Static Members
3101 @section Static Members
3103 The data area for a class is a concatenation of the space used by the
3104 data members of the class. If the class has virtual methods, a vtable
3105 pointer follows the class data. The field offset part of each field
3106 description in the class stab shows this ordering.
3108 << How is this reflected in stabs? See Cygnus bug #677 for some info. >>
3111 @appendix Table of Stab Types
3113 The following are all the possible values for the stab type field, for
3114 a.out files, in numeric order. This does not apply to XCOFF, but
3115 it does apply to stabs in sections (@pxref{Stab Sections}). Stabs in
3116 ECOFF use these values but add 0x8f300 to distinguish them from non-stab
3119 The symbolic names are defined in the file @file{include/aout/stabs.def}.
3122 * Non-Stab Symbol Types:: Types from 0 to 0x1f
3123 * Stab Symbol Types:: Types from 0x20 to 0xff
3126 @node Non-Stab Symbol Types
3127 @appendixsec Non-Stab Symbol Types
3129 The following types are used by the linker and assembler, not by stab
3130 directives. Since this document does not attempt to describe aspects of
3131 object file format other than the debugging format, no details are
3134 @c Try to get most of these to fit on a single line.
3144 File scope absolute symbol
3146 @item 0x3 N_ABS | N_EXT
3147 External absolute symbol
3150 File scope text symbol
3152 @item 0x5 N_TEXT | N_EXT
3153 External text symbol
3156 File scope data symbol
3158 @item 0x7 N_DATA | N_EXT
3159 External data symbol
3162 File scope BSS symbol
3164 @item 0x9 N_BSS | N_EXT
3168 Same as @code{N_FN}, for Sequent compilers
3171 Symbol is indirected to another symbol
3174 Common---visible after shared library dynamic link
3177 @itemx 0x15 N_SETA | N_EXT
3178 Absolute set element
3181 @itemx 0x17 N_SETT | N_EXT
3182 Text segment set element
3185 @itemx 0x19 N_SETD | N_EXT
3186 Data segment set element
3189 @itemx 0x1b N_SETB | N_EXT
3190 BSS segment set element
3193 @itemx 0x1d N_SETV | N_EXT
3194 Pointer to set vector
3196 @item 0x1e N_WARNING
3197 Print a warning message during linking
3200 File name of a @file{.o} file
3203 @node Stab Symbol Types
3204 @appendixsec Stab Symbol Types
3206 The following symbol types indicate that this is a stab. This is the
3207 full list of stab numbers, including stab types that are used in
3208 languages other than C.
3212 Global symbol; see @ref{Global Variables}.
3215 Function name (for BSD Fortran); see @ref{Procedures}.
3218 Function name (@pxref{Procedures}) or text segment variable
3222 Data segment file-scope variable; see @ref{Statics}.
3225 BSS segment file-scope variable; see @ref{Statics}.
3228 Name of main routine; see @ref{Main Program}.
3231 Variable in @code{.rodata} section; see @ref{Statics}.
3234 Global symbol (for Pascal); see @ref{N_PC}.
3237 Number of symbols (according to Ultrix V4.0); see @ref{N_NSYMS}.
3240 No DST map; see @ref{N_NOMAP}.
3242 @c FIXME: describe this solaris feature in the body of the text (see
3243 @c comments in include/aout/stab.def).
3245 Object file (Solaris2).
3247 @c See include/aout/stab.def for (a little) more info.
3249 Debugger options (Solaris2).
3252 Register variable; see @ref{Register Variables}.
3255 Modula-2 compilation unit; see @ref{N_M2C}.
3258 Line number in text segment; see @ref{Line Numbers}.
3261 Line number in data segment; see @ref{Line Numbers}.
3264 Line number in bss segment; see @ref{Line Numbers}.
3267 Sun source code browser, path to @file{.cb} file; see @ref{N_BROWS}.
3270 GNU Modula2 definition module dependency; see @ref{N_DEFD}.
3273 Function start/body/end line numbers (Solaris2).
3276 GNU C++ exception variable; see @ref{N_EHDECL}.
3279 Modula2 info "for imc" (according to Ultrix V4.0); see @ref{N_MOD2}.
3282 GNU C++ @code{catch} clause; see @ref{N_CATCH}.
3285 Structure of union element; see @ref{N_SSYM}.
3288 Last stab for module (Solaris2).
3291 Path and name of source file; see @ref{Source Files}.
3294 Stack variable (@pxref{Stack Variables}) or type (@pxref{Typedefs}).
3297 Beginning of an include file (Sun only); see @ref{Include Files}.
3300 Name of include file; see @ref{Include Files}.
3303 Parameter variable; see @ref{Parameters}.
3306 End of an include file; see @ref{Include Files}.
3309 Alternate entry point; see @ref{Alternate Entry Points}.
3312 Beginning of a lexical block; see @ref{Block Structure}.
3315 Place holder for a deleted include file; see @ref{Include Files}.
3318 Modula2 scope information (Sun linker); see @ref{N_SCOPE}.
3321 End of a lexical block; see @ref{Block Structure}.
3324 Begin named common block; see @ref{Common Blocks}.
3327 End named common block; see @ref{Common Blocks}.
3330 Member of a common block; see @ref{Common Blocks}.
3332 @c FIXME: How does this really work? Move it to main body of document.
3334 Pascal @code{with} statement: type,,0,0,offset (Solaris2).
3337 Gould non-base registers; see @ref{Gould}.
3340 Gould non-base registers; see @ref{Gould}.
3343 Gould non-base registers; see @ref{Gould}.
3346 Gould non-base registers; see @ref{Gould}.
3349 Gould non-base registers; see @ref{Gould}.
3352 @c Restore the default table indent
3357 @node Symbol Descriptors
3358 @appendix Table of Symbol Descriptors
3360 The symbol descriptor is the character which follows the colon in many
3361 stabs, and which tells what kind of stab it is. @xref{String Field},
3362 for more information about their use.
3364 @c Please keep this alphabetical
3366 @c In TeX, this looks great, digit is in italics. But makeinfo insists
3367 @c on putting it in `', not realizing that @var should override @code.
3368 @c I don't know of any way to make makeinfo do the right thing. Seems
3369 @c like a makeinfo bug to me.
3373 Variable on the stack; see @ref{Stack Variables}.
3376 C++ nested symbol; see @xref{Nested Symbols}.
3379 Parameter passed by reference in register; see @ref{Reference Parameters}.
3382 Based variable; see @ref{Based Variables}.
3385 Constant; see @ref{Constants}.
3388 Conformant array bound (Pascal, maybe other languages); @ref{Conformant
3389 Arrays}. Name of a caught exception (GNU C++). These can be
3390 distinguished because the latter uses @code{N_CATCH} and the former uses
3391 another symbol type.
3394 Floating point register variable; see @ref{Register Variables}.
3397 Parameter in floating point register; see @ref{Register Parameters}.
3400 File scope function; see @ref{Procedures}.
3403 Global function; see @ref{Procedures}.
3406 Global variable; see @ref{Global Variables}.
3409 @xref{Register Parameters}.
3412 Internal (nested) procedure; see @ref{Nested Procedures}.
3415 Internal (nested) function; see @ref{Nested Procedures}.
3418 Label name (documented by AIX, no further information known).
3421 Module; see @ref{Procedures}.
3424 Argument list parameter; see @ref{Parameters}.
3430 Fortran Function parameter; see @ref{Parameters}.
3433 Unfortunately, three separate meanings have been independently invented
3434 for this symbol descriptor. At least the GNU and Sun uses can be
3435 distinguished by the symbol type. Global Procedure (AIX) (symbol type
3436 used unknown); see @ref{Procedures}. Register parameter (GNU) (symbol
3437 type @code{N_PSYM}); see @ref{Parameters}. Prototype of function
3438 referenced by this file (Sun @code{acc}) (symbol type @code{N_FUN}).
3441 Static Procedure; see @ref{Procedures}.
3444 Register parameter; see @ref{Register Parameters}.
3447 Register variable; see @ref{Register Variables}.
3450 File scope variable; see @ref{Statics}.
3453 Local variable (OS9000).
3456 Type name; see @ref{Typedefs}.
3459 Enumeration, structure, or union tag; see @ref{Typedefs}.
3462 Parameter passed by reference; see @ref{Reference Parameters}.
3465 Procedure scope static variable; see @ref{Statics}.
3468 Conformant array; see @ref{Conformant Arrays}.
3471 Function return variable; see @ref{Parameters}.
3474 @node Type Descriptors
3475 @appendix Table of Type Descriptors
3477 The type descriptor is the character which follows the type number and
3478 an equals sign. It specifies what kind of type is being defined.
3479 @xref{String Field}, for more information about their use.
3484 Type reference; see @ref{String Field}.
3487 Reference to builtin type; see @ref{Negative Type Numbers}.
3490 Method (C++); see @ref{Method Type Descriptor}.
3493 Pointer; see @ref{Miscellaneous Types}.
3499 Type Attributes (AIX); see @ref{String Field}. Member (class and variable)
3500 type (GNU C++); see @ref{Member Type Descriptor}.
3503 Array; see @ref{Arrays}.
3506 Open array; see @ref{Arrays}.
3509 Pascal space type (AIX); see @ref{Miscellaneous Types}. Builtin integer
3510 type (Sun); see @ref{Builtin Type Descriptors}. Const and volatile
3511 qualified type (OS9000).
3514 Volatile-qualified type; see @ref{Miscellaneous Types}.
3517 Complex builtin type (AIX); see @ref{Builtin Type Descriptors}.
3518 Const-qualified type (OS9000).
3521 COBOL Picture type. See AIX documentation for details.
3524 File type; see @ref{Miscellaneous Types}.
3527 N-dimensional dynamic array; see @ref{Arrays}.
3530 Enumeration type; see @ref{Enumerations}.
3533 N-dimensional subarray; see @ref{Arrays}.
3536 Function type; see @ref{Function Types}.
3539 Pascal function parameter; see @ref{Function Types}
3542 Builtin floating point type; see @ref{Builtin Type Descriptors}.
3545 COBOL Group. See AIX documentation for details.
3548 Imported type (AIX); see @ref{Cross-References}. Volatile-qualified
3552 Const-qualified type; see @ref{Miscellaneous Types}.
3555 COBOL File Descriptor. See AIX documentation for details.
3558 Multiple instance type; see @ref{Miscellaneous Types}.
3561 String type; see @ref{Strings}.
3564 Stringptr; see @ref{Strings}.
3567 Opaque type; see @ref{Typedefs}.
3570 Procedure; see @ref{Function Types}.
3573 Packed array; see @ref{Arrays}.
3576 Range type; see @ref{Subranges}.
3579 Builtin floating type; see @ref{Builtin Type Descriptors} (Sun). Pascal
3580 subroutine parameter; see @ref{Function Types} (AIX). Detecting this
3581 conflict is possible with careful parsing (hint: a Pascal subroutine
3582 parameter type will always contain a comma, and a builtin type
3583 descriptor never will).
3586 Structure type; see @ref{Structures}.
3589 Set type; see @ref{Miscellaneous Types}.
3592 Union; see @ref{Unions}.
3595 Variant record. This is a Pascal and Modula-2 feature which is like a
3596 union within a struct in C. See AIX documentation for details.
3599 Wide character; see @ref{Builtin Type Descriptors}.
3602 Cross-reference; see @ref{Cross-References}.
3605 Used by IBM's xlC C++ compiler (for structures, I think).
3608 gstring; see @ref{Strings}.
3611 @node Expanded Reference
3612 @appendix Expanded Reference by Stab Type
3614 @c FIXME: This appendix should go away; see N_PSYM or N_SO for an example.
3616 For a full list of stab types, and cross-references to where they are
3617 described, see @ref{Stab Types}. This appendix just covers certain
3618 stabs which are not yet described in the main body of this document;
3619 eventually the information will all be in one place.
3623 The first line is the symbol type (see @file{include/aout/stab.def}).
3625 The second line describes the language constructs the symbol type
3628 The third line is the stab format with the significant stab fields
3629 named and the rest NIL.
3631 Subsequent lines expand upon the meaning and possible values for each
3632 significant stab field.
3634 Finally, any further information.
3637 * N_PC:: Pascal global symbol
3638 * N_NSYMS:: Number of symbols
3639 * N_NOMAP:: No DST map
3640 * N_M2C:: Modula-2 compilation unit
3641 * N_BROWS:: Path to .cb file for Sun source code browser
3642 * N_DEFD:: GNU Modula2 definition module dependency
3643 * N_EHDECL:: GNU C++ exception variable
3644 * N_MOD2:: Modula2 information "for imc"
3645 * N_CATCH:: GNU C++ "catch" clause
3646 * N_SSYM:: Structure or union element
3647 * N_SCOPE:: Modula2 scope information (Sun only)
3648 * Gould:: non-base register symbols used on Gould systems
3649 * N_LENG:: Length of preceding entry
3655 @deffn @code{.stabs} N_PC
3657 Global symbol (for Pascal).
3660 "name" -> "symbol_name" <<?>>
3661 value -> supposedly the line number (stab.def is skeptical)
3665 @file{stabdump.c} says:
3667 global pascal symbol: name,,0,subtype,line
3675 @deffn @code{.stabn} N_NSYMS
3677 Number of symbols (according to Ultrix V4.0).
3680 0, files,,funcs,lines (stab.def)
3687 @deffn @code{.stabs} N_NOMAP
3689 No DST map for symbol (according to Ultrix V4.0). I think this means a
3690 variable has been optimized out.
3693 name, ,0,type,ignored (stab.def)
3700 @deffn @code{.stabs} N_M2C
3702 Modula-2 compilation unit.
3705 "string" -> "unit_name,unit_time_stamp[,code_time_stamp]"
3707 value -> 0 (main unit)
3711 See @cite{Dbx and Dbxtool Interfaces}, 2nd edition, by Sun, 1988, for
3719 @deffn @code{.stabs} N_BROWS
3721 Sun source code browser, path to @file{.cb} file
3724 "path to associated @file{.cb} file"
3726 Note: N_BROWS has the same value as N_BSLINE.
3732 @deffn @code{.stabn} N_DEFD
3734 GNU Modula2 definition module dependency.
3736 GNU Modula-2 definition module dependency. The value is the
3737 modification time of the definition file. The other field is non-zero
3738 if it is imported with the GNU M2 keyword @code{%INITIALIZE}. Perhaps
3739 @code{N_M2C} can be used if there are enough empty fields?
3745 @deffn @code{.stabs} N_EHDECL
3747 GNU C++ exception variable <<?>>.
3749 "@var{string} is variable name"
3751 Note: conflicts with @code{N_MOD2}.
3757 @deffn @code{.stab?} N_MOD2
3759 Modula2 info "for imc" (according to Ultrix V4.0)
3761 Note: conflicts with @code{N_EHDECL} <<?>>
3767 @deffn @code{.stabn} N_CATCH
3769 GNU C++ @code{catch} clause
3771 GNU C++ @code{catch} clause. The value is its address. The desc field
3772 is nonzero if this entry is immediately followed by a @code{CAUGHT} stab
3773 saying what exception was caught. Multiple @code{CAUGHT} stabs means
3774 that multiple exceptions can be caught here. If desc is 0, it means all
3775 exceptions are caught here.
3781 @deffn @code{.stabn} N_SSYM
3783 Structure or union element.
3785 The value is the offset in the structure.
3787 <<?looking at structs and unions in C I didn't see these>>
3793 @deffn @code{.stab?} N_SCOPE
3795 Modula2 scope information (Sun linker)
3800 @section Non-base registers on Gould systems
3802 @deffn @code{.stab?} N_NBTEXT
3803 @deffnx @code{.stab?} N_NBDATA
3804 @deffnx @code{.stab?} N_NBBSS
3805 @deffnx @code{.stab?} N_NBSTS
3806 @deffnx @code{.stab?} N_NBLCS
3812 These are used on Gould systems for non-base registers syms.
3814 However, the following values are not the values used by Gould; they are
3815 the values which GNU has been documenting for these values for a long
3816 time, without actually checking what Gould uses. I include these values
3817 only because perhaps some someone actually did something with the GNU
3818 information (I hope not, why GNU knowingly assigned wrong values to
3819 these in the header file is a complete mystery to me).
3822 240 0xf0 N_NBTEXT ??
3823 242 0xf2 N_NBDATA ??
3833 @deffn @code{.stabn} N_LENG
3835 Second symbol entry containing a length-value for the preceding entry.
3836 The value is the length.
3840 @appendix Questions and Anomalies
3844 @c I think this is changed in GCC 2.4.5 to put the line number there.
3845 For GNU C stabs defining local and global variables (@code{N_LSYM} and
3846 @code{N_GSYM}), the desc field is supposed to contain the source
3847 line number on which the variable is defined. In reality the desc
3848 field is always 0. (This behavior is defined in @file{dbxout.c} and
3849 putting a line number in desc is controlled by @samp{#ifdef
3850 WINNING_GDB}, which defaults to false). GDB supposedly uses this
3851 information if you say @samp{list @var{var}}. In reality, @var{var} can
3852 be a variable defined in the program and GDB says @samp{function
3853 @var{var} not defined}.
3856 In GNU C stabs, there seems to be no way to differentiate tag types:
3857 structures, unions, and enums (symbol descriptor @samp{T}) and typedefs
3858 (symbol descriptor @samp{t}) defined at file scope from types defined locally
3859 to a procedure or other more local scope. They all use the @code{N_LSYM}
3860 stab type. Types defined at procedure scope are emitted after the
3861 @code{N_RBRAC} of the preceding function and before the code of the
3862 procedure in which they are defined. This is exactly the same as
3863 types defined in the source file between the two procedure bodies.
3864 GDB over-compensates by placing all types in block #1, the block for
3865 symbols of file scope. This is true for default, @samp{-ansi} and
3866 @samp{-traditional} compiler options. (Bugs gcc/1063, gdb/1066.)
3869 What ends the procedure scope? Is it the proc block's @code{N_RBRAC} or the
3870 next @code{N_FUN}? (I believe its the first.)
3874 @appendix Using Stabs in Their Own Sections
3876 Many object file formats allow tools to create object files with custom
3877 sections containing any arbitrary data. For any such object file
3878 format, stabs can be embedded in special sections. This is how stabs
3879 are used with ELF and SOM, and aside from ECOFF and XCOFF, is how stabs
3883 * Stab Section Basics:: How to embed stabs in sections
3884 * ELF Linker Relocation:: Sun ELF hacks
3887 @node Stab Section Basics
3888 @appendixsec How to Embed Stabs in Sections
3890 The assembler creates two custom sections, a section named @code{.stab}
3891 which contains an array of fixed length structures, one struct per stab,
3892 and a section named @code{.stabstr} containing all the variable length
3893 strings that are referenced by stabs in the @code{.stab} section. The
3894 byte order of the stabs binary data depends on the object file format.
3895 For ELF, it matches the byte order of the ELF file itself, as determined
3896 from the @code{EI_DATA} field in the @code{e_ident} member of the ELF
3897 header. For SOM, it is always big-endian (is this true??? FIXME). For
3898 COFF, it matches the byte order of the COFF headers. The meaning of the
3899 fields is the same as for a.out (@pxref{Symbol Table Format}), except
3900 that the @code{n_strx} field is relative to the strings for the current
3901 compilation unit (which can be found using the synthetic N_UNDF stab
3902 described below), rather than the entire string table.
3904 The first stab in the @code{.stab} section for each compilation unit is
3905 synthetic, generated entirely by the assembler, with no corresponding
3906 @code{.stab} directive as input to the assembler. This stab contains
3907 the following fields:
3911 Offset in the @code{.stabstr} section to the source filename.
3917 Unused field, always zero.
3918 This may eventually be used to hold overflows from the count in
3919 the @code{n_desc} field.
3922 Count of upcoming symbols, i.e., the number of remaining stabs for this
3926 Size of the string table fragment associated with this source file, in
3930 The @code{.stabstr} section always starts with a null byte (so that string
3931 offsets of zero reference a null string), followed by random length strings,
3932 each of which is null byte terminated.
3934 The ELF section header for the @code{.stab} section has its
3935 @code{sh_link} member set to the section number of the @code{.stabstr}
3936 section, and the @code{.stabstr} section has its ELF section
3937 header @code{sh_type} member set to @code{SHT_STRTAB} to mark it as a
3938 string table. SOM and COFF have no way of linking the sections together
3939 or marking them as string tables.
3941 For COFF, the @code{.stab} and @code{.stabstr} sections may be simply
3942 concatenated by the linker. GDB then uses the @code{n_desc} fields to
3943 figure out the extent of the original sections. Similarly, the
3944 @code{n_value} fields of the header symbols are added together in order
3945 to get the actual position of the strings in a desired @code{.stabstr}
3946 section. Although this design obviates any need for the linker to
3947 relocate or otherwise manipulate @code{.stab} and @code{.stabstr}
3948 sections, it also requires some care to ensure that the offsets are
3949 calculated correctly. For instance, if the linker were to pad in
3950 between the @code{.stabstr} sections before concatenating, then the
3951 offsets to strings in the middle of the executable's @code{.stabstr}
3952 section would be wrong.
3954 The GNU linker is able to optimize stabs information by merging
3955 duplicate strings and removing duplicate header file information
3956 (@pxref{Include Files}). When some versions of the GNU linker optimize
3957 stabs in sections, they remove the leading @code{N_UNDF} symbol and
3958 arranges for all the @code{n_strx} fields to be relative to the start of
3959 the @code{.stabstr} section.
3961 @node ELF Linker Relocation
3962 @appendixsec Having the Linker Relocate Stabs in ELF
3964 This section describes some Sun hacks for Stabs in ELF; it does not
3965 apply to COFF or SOM.
3967 To keep linking fast, you don't want the linker to have to relocate very
3968 many stabs. Making sure this is done for @code{N_SLINE},
3969 @code{N_RBRAC}, and @code{N_LBRAC} stabs is the most important thing
3970 (see the descriptions of those stabs for more information). But Sun's
3971 stabs in ELF has taken this further, to make all addresses in the
3972 @code{n_value} field (functions and static variables) relative to the
3973 source file. For the @code{N_SO} symbol itself, Sun simply omits the
3974 address. To find the address of each section corresponding to a given
3975 source file, the compiler puts out symbols giving the address of each
3976 section for a given source file. Since these are ELF (not stab)
3977 symbols, the linker relocates them correctly without having to touch the
3978 stabs section. They are named @code{Bbss.bss} for the bss section,
3979 @code{Ddata.data} for the data section, and @code{Drodata.rodata} for
3980 the rodata section. For the text section, there is no such symbol (but
3981 there should be, see below). For an example of how these symbols work,
3982 @xref{Stab Section Transformations}. GCC does not provide these symbols;
3983 it instead relies on the stabs getting relocated. Thus addresses which
3984 would normally be relative to @code{Bbss.bss}, etc., are already
3985 relocated. The Sun linker provided with Solaris 2.2 and earlier
3986 relocates stabs using normal ELF relocation information, as it would do
3987 for any section. Sun has been threatening to kludge their linker to not
3988 do this (to speed up linking), even though the correct way to avoid
3989 having the linker do these relocations is to have the compiler no longer
3990 output relocatable values. Last I heard they had been talked out of the
3991 linker kludge. See Sun point patch 101052-01 and Sun bug 1142109. With
3992 the Sun compiler this affects @samp{S} symbol descriptor stabs
3993 (@pxref{Statics}) and functions (@pxref{Procedures}). In the latter
3994 case, to adopt the clean solution (making the value of the stab relative
3995 to the start of the compilation unit), it would be necessary to invent a
3996 @code{Ttext.text} symbol, analogous to the @code{Bbss.bss}, etc.,
3997 symbols. I recommend this rather than using a zero value and getting
3998 the address from the ELF symbols.
4000 Finding the correct @code{Bbss.bss}, etc., symbol is difficult, because
4001 the linker simply concatenates the @code{.stab} sections from each
4002 @file{.o} file without including any information about which part of a
4003 @code{.stab} section comes from which @file{.o} file. The way GDB does
4004 this is to look for an ELF @code{STT_FILE} symbol which has the same
4005 name as the last component of the file name from the @code{N_SO} symbol
4006 in the stabs (for example, if the file name is @file{../../gdb/main.c},
4007 it looks for an ELF @code{STT_FILE} symbol named @code{main.c}). This
4008 loses if different files have the same name (they could be in different
4009 directories, a library could have been copied from one system to
4010 another, etc.). It would be much cleaner to have the @code{Bbss.bss}
4011 symbols in the stabs themselves. Having the linker relocate them there
4012 is no more work than having the linker relocate ELF symbols, and it
4013 solves the problem of having to associate the ELF and stab symbols.
4014 However, no one has yet designed or implemented such a scheme.
4016 @node Symbol Types Index
4017 @unnumbered Symbol Types Index
4021 @c TeX can handle the contents at the start but makeinfo 3.12 can not