2 @setfilename stabs.info
9 * Stabs:: The "stabs" debugging information format.
15 This document describes the stabs debugging symbol tables.
17 Copyright 1992 Free Software Foundation, Inc.
18 Contributed by Cygnus Support. Written by Julia Menapace.
20 Permission is granted to make and distribute verbatim copies of
21 this manual provided the copyright notice and this permission notice
22 are preserved on all copies.
25 Permission is granted to process this file through Tex and print the
26 results, provided the printed document carries copying permission
27 notice identical to this one except for the removal of this paragraph
28 (this paragraph not being relevant to the printed manual).
31 Permission is granted to copy or distribute modified versions of this
32 manual under the terms of the GPL (for which purpose this text may be
33 regarded as a program in the language TeX).
36 @setchapternewpage odd
39 @title The ``stabs'' debug format
40 @author Julia Menapace
41 @author Cygnus Support
44 \def\$#1${{#1}} % Kluge: collect RCS revision info without $...$
45 \xdef\manvers{\$Revision$} % For use in headers, footers too
47 \hfill Cygnus Support\par
49 \hfill \TeX{}info \texinfoversion\par
53 @vskip 0pt plus 1filll
54 Copyright @copyright{} 1992 Free Software Foundation, Inc.
55 Contributed by Cygnus Support.
57 Permission is granted to make and distribute verbatim copies of
58 this manual provided the copyright notice and this permission notice
59 are preserved on all copies.
65 @top The "stabs" representation of debugging information
67 This document describes the stabs debugging format.
70 * Overview:: Overview of stabs
71 * Program structure:: Encoding of the structure of the program
72 * Constants:: Constants
73 * Example:: A comprehensive example in C
75 * Types:: Type definitions
76 * Symbol Tables:: Symbol information in symbol tables
77 * Cplusplus:: Appendixes:
78 * Example2.c:: Source code for extended example
79 * Example2.s:: Assembly code for extended example
80 * Stab Types:: Symbol types in a.out files
81 * Symbol Descriptors:: Table of Symbol Descriptors
82 * Type Descriptors:: Table of Symbol Descriptors
83 * Expanded reference:: Reference information by stab type
84 * Questions:: Questions and anomolies
85 * xcoff-differences:: Differences between GNU stabs in a.out
86 and GNU stabs in xcoff
87 * Sun-differences:: Differences between GNU stabs and Sun
89 * Stabs-in-elf:: Stabs in an ELF file.
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
99 @c FIXME! <<name of inventor>> at
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 completely comprehensive for stabs used by
105 C. The lists of symbol descriptors (@pxref{Symbol Descriptors}) and
106 type descriptors (@pxref{Type Descriptors}) are believed to be completely
107 comprehensive. There are known to be stabs for C++ and COBOL which are
108 poorly documented here. Stabs specific to other languages (e.g. Pascal,
109 Modula-2) are probably not as well documented as they should be.
111 Other sources of information on stabs are @cite{dbx and dbxtool
112 interfaces}, 2nd edition, by Sun, circa 1988, and @cite{AIX Version 3.2
113 Files Reference}, Fourth Edition, September 1992, "dbx Stabstring
114 Grammar" in the a.out section, page 2-31. This document is believed to
115 incorporate the information from those two sources except where it
116 explictly directs you to them for more information.
119 * Flow:: Overview of debugging information flow
120 * Stabs Format:: Overview of stab format
121 * C example:: A simple example in C source
122 * Assembly code:: The simple example at the assembly level
126 @section Overview of debugging information flow
128 The GNU C compiler compiles C source in a @file{.c} file into assembly
129 language in a @file{.s} file, which is translated by the assembler into
130 a @file{.o} file, and then linked with other @file{.o} files and
131 libraries to produce an executable file.
133 With the @samp{-g} option, GCC puts additional debugging information in
134 the @file{.s} file, which is slightly transformed by the assembler and
135 linker, and carried through into the final executable. This debugging
136 information describes features of the source file like line numbers,
137 the types and scopes of variables, and functions, their parameters and
140 For some object file formats, the debugging information is
141 encapsulated in assembler directives known collectively as `stab' (symbol
142 table) directives, interspersed with the generated code. Stabs are
143 the native format for debugging information in the a.out and xcoff
144 object file formats. The GNU tools can also emit stabs in the coff
145 and ecoff object file formats.
147 The assembler adds the information from stabs to the symbol information
148 it places by default in the symbol table and the string table of the
149 @file{.o} file it is building. The linker consolidates the @file{.o}
150 files into one executable file, with one symbol table and one string
151 table. Debuggers use the symbol and string tables in the executable as
152 a source of debugging information about the program.
155 @section Overview of stab format
157 There are three overall formats for stab assembler directives
158 differentiated by the first word of the stab. The name of the directive
159 describes what combination of four possible data fields will follow. It
160 is either @code{.stabs} (string), @code{.stabn} (number), or
161 @code{.stabd} (dot). IBM's xcoff uses @code{.stabx} (and some other
162 directives such as @code{.file} and @code{.bi}) instead of
163 @code{.stabs}, @code{.stabn} or @code{.stabd}.
165 The overall format of each class of stab is:
168 .stabs "@var{string}",@var{type},0,@var{desc},@var{value}
169 .stabx "@var{string}",@var{value},@var{type},@var{sdb-type}
170 .stabn @var{type},0,@var{desc},@var{value}
171 .stabd @var{type},0,@var{desc}
174 @c what is the correct term for "current file location"? My AIX
175 @c assembler manual calls it "the value of the current location counter".
176 For @code{.stabn} and @code{.stabd}, there is no string (the
177 @code{n_strx} field is zero, @pxref{Symbol Tables}). For @code{.stabd}
178 the value field is implicit and has the value of the current file
179 location. The @var{sdb-type} field to @code{.stabx} is unused for stabs
180 and can always be set to 0.
182 The number in the type field gives some basic information about what
183 type of stab this is (or whether it @emph{is} a stab, as opposed to an
184 ordinary symbol). Each possible type number defines a different stab
185 type. The stab type further defines the exact interpretation of, and
186 possible values for, any remaining @code{"@var{string}"}, @var{desc}, or
187 @var{value} fields present in the stab. @xref{Stab Types}, for a list
188 in numeric order of the possible type field values for stab directives.
190 For @code{.stabs} the @code{"@var{string}"} field holds the meat of the
191 debugging information. The generally unstructured nature of this field
192 is what makes stabs extensible. For some stab types the string field
193 contains only a name. For other stab types the contents can be a great
196 The overall format is of the @code{"@var{string}"} field is:
199 "@var{name}:@var{symbol-descriptor} @var{type-information}"
202 @var{name} is the name of the symbol represented by the stab.
203 @var{name} can be omitted, which means the stab represents an unnamed
204 object. For example, @samp{:t10=*2} defines type 10 as a pointer to
205 type 2, but does not give the type a name. Omitting the @var{name}
206 field is supported by AIX dbx and GDB after about version 4.8, but not
207 other debuggers. GCC sometimes uses a single space as the name instead
208 of omitting the name altogether; apparently that is supported by most
211 The @var{symbol_descriptor} following the @samp{:} is an alphabetic
212 character that tells more specifically what kind of symbol the stab
213 represents. If the @var{symbol_descriptor} is omitted, but type
214 information follows, then the stab represents a local variable. For a
215 list of symbol descriptors, see @ref{Symbol Descriptors,,Table C: Symbol
218 The @samp{c} symbol descriptor is an exception in that it is not
219 followed by type information. @xref{Constants}.
221 Type information is either a @var{type_number}, or a
222 @samp{@var{type_number}=}. The @var{type_number} alone is a type
223 reference, referring directly to a type that has already been defined.
225 The @samp{@var{type_number}=} is a type definition, where the number
226 represents a new type which is about to be defined. The type definition
227 may refer to other types by number, and those type numbers may be
228 followed by @samp{=} and nested definitions.
230 In a type definition, if the character that follows the equals sign is
231 non-numeric then it is a @var{type_descriptor}, and tells what kind of
232 type is about to be defined. Any other values following the
233 @var{type_descriptor} vary, depending on the @var{type_descriptor}. If
234 a number follows the @samp{=} then the number is a @var{type_reference}.
235 This is described more thoroughly in the section on types. @xref{Type
236 Descriptors,,Table D: Type Descriptors}, for a list of
237 @var{type_descriptor} values.
239 There is an AIX extension for type attributes. Following the @samp{=}
240 is any number of type attributes. Each one starts with @samp{@@} and
241 ends with @samp{;}. Debuggers, including AIX's dbx, skip any type
242 attributes they do not recognize. GDB 4.9 does not do this---it will
243 ignore the entire symbol containing a type attribute. Hopefully this
244 will be fixed in the next GDB release. Because of a conflict with C++
245 (@pxref{Cplusplus}), new attributes should not be defined which begin
246 with a digit, @samp{(}, or @samp{-}; GDB may be unable to distinguish
247 those from the C++ type descriptor @samp{@@}. The attributes are:
250 @item a@var{boundary}
251 @var{boundary} is an integer specifying the alignment. I assume it
252 applies to all variables of this type.
255 Size in bits of a variable of this type.
258 Pointer class (for checking). Not sure what this means, or how
259 @var{integer} is interpreted.
262 Indicate this is a packed type, meaning that structure fields or array
263 elements are placed more closely in memory, to save memory at the
267 All this can make the @code{"@var{string}"} field quite long. All
268 versions of GDB, and some versions of DBX, can handle arbitrarily long
269 strings. But many versions of DBX cretinously limit the strings to
270 about 80 characters, so compilers which must work with such DBX's need
271 to split the @code{.stabs} directive into several @code{.stabs}
272 directives. Each stab duplicates exactly all but the
273 @code{"@var{string}"} field. The @code{"@var{string}"} field of
274 every stab except the last is marked as continued with a
275 double-backslash at the end. Removing the backslashes and concatenating
276 the @code{"@var{string}"} fields of each stab produces the original,
280 @section A simple example in C source
282 To get the flavor of how stabs describe source information for a C
283 program, let's look at the simple program:
288 printf("Hello world");
292 When compiled with @samp{-g}, the program above yields the following
293 @file{.s} file. Line numbers have been added to make it easier to refer
294 to parts of the @file{.s} file in the description of the stabs that
298 @section The simple example at the assembly level
302 2 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0
303 3 .stabs "hello.c",100,0,0,Ltext0
306 6 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0
307 7 .stabs "char:t2=r2;0;127;",128,0,0,0
308 8 .stabs "long int:t3=r1;-2147483648;2147483647;",128,0,0,0
309 9 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
310 10 .stabs "long unsigned int:t5=r1;0;-1;",128,0,0,0
311 11 .stabs "short int:t6=r1;-32768;32767;",128,0,0,0
312 12 .stabs "long long int:t7=r1;0;-1;",128,0,0,0
313 13 .stabs "short unsigned int:t8=r1;0;65535;",128,0,0,0
314 14 .stabs "long long unsigned int:t9=r1;0;-1;",128,0,0,0
315 15 .stabs "signed char:t10=r1;-128;127;",128,0,0,0
316 16 .stabs "unsigned char:t11=r1;0;255;",128,0,0,0
317 17 .stabs "float:t12=r1;4;0;",128,0,0,0
318 18 .stabs "double:t13=r1;8;0;",128,0,0,0
319 19 .stabs "long double:t14=r1;8;0;",128,0,0,0
320 20 .stabs "void:t15=15",128,0,0,0
323 23 .ascii "Hello, world!\12\0"
338 38 sethi %hi(LC0),%o1
339 39 or %o1,%lo(LC0),%o0
350 50 .stabs "main:F1",36,0,0,_main
351 51 .stabn 192,0,0,LBB2
352 52 .stabn 224,0,0,LBE2
355 This simple ``hello world'' example demonstrates several of the stab
356 types used to describe C language source files.
358 @node Program structure
359 @chapter Encoding for the structure of the program
362 * Main Program:: Indicate what the main program is
363 * Source Files:: The path and name of the source file
370 @section Main Program
372 Most languages allow the main program to have any name. The
373 @code{N_MAIN} stab type is used for a stab telling the debugger what
374 name is used in this program. Only the name is significant; it will be
375 the name of a function which is the main program. Most C compilers do
376 not use this stab; they expect the debugger to simply assume that the
377 name is @samp{main}, but some C compilers emit an @code{N_MAIN} stab for
378 the @samp{main} function.
381 @section The path and name of the source files
383 Before any other stabs occur, there must be a stab specifying the source
384 file. This information is contained in a symbol of stab type
385 @code{N_SO}; the string contains the name of the file. The value of the
386 symbol is the start address of portion of the text section corresponding
389 With the Sun Solaris2 compiler, the @code{desc} field contains a
390 source-language code.
392 Some compilers (for example, gcc2 and SunOS4 @file{/bin/cc}) also
393 include the directory in which the source was compiled, in a second
394 @code{N_SO} symbol preceding the one containing the file name. This
395 symbol can be distinguished by the fact that it ends in a slash. Code
396 from the cfront C++ compiler can have additional @code{N_SO} symbols for
397 nonexistent source files after the @code{N_SO} for the real source file;
398 these are believed to contain no useful information.
403 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0 ; 100 is N_SO
404 .stabs "hello.c",100,0,0,Ltext0
409 Instead of @code{N_SO} symbols, XCOFF uses a @code{.file} assembler
410 directive which assembles to a standard COFF @code{.file} symbol;
411 explaining this in detail is outside the scope of this document.
413 There are several different schemes for dealing with include files: the
414 traditional @code{N_SOL} approach, Sun's @code{N_BINCL} scheme, and the
415 XCOFF @code{C_BINCL} (which despite the similar name has little in
416 common with @code{N_BINCL}).
418 An @code{N_SOL} symbol specifies which include file subsequent symbols
419 refer to. The string field is the name of the file and the value is the
420 text address corresponding to the start of the previous include file and
421 the start of this one. To specify the main source file again, use an
422 @code{N_SOL} symbol with the name of the main source file.
424 A @code{N_BINCL} symbol specifies the start of an include file. In an
425 object file, only the name is significant. The Sun linker puts data
426 into some of the other fields. The end of the include file is marked by
427 a @code{N_EINCL} symbol (which has no name field). In an ojbect file,
428 there is no significant data in the @code{N_EINCL} symbol; the Sun
429 linker puts data into some of the fields. @code{N_BINCL} and
430 @code{N_EINCL} can be nested. If the linker detects that two source
431 files have identical stabs with a @code{N_BINCL} and @code{N_EINCL} pair
432 (as will generally be the case for a header file), then it only puts out
433 the stabs once. Each additional occurance is replaced by an
434 @code{N_EXCL} symbol. I believe the Sun (SunOS4, not sure about
435 Solaris) linker is the only one which supports this feature.
437 For the start of an include file in XCOFF, use the @file{.bi} assembler
438 directive which generates a @code{C_BINCL} symbol. A @file{.ei}
439 directive, which generates a @code{C_EINCL} symbol, denotes the end of
440 the include file. Both directives are followed by the name of the
441 source file in quotes, which becomes the string for the symbol. The
442 value of each symbol, produced automatically by the assembler and
443 linker, is an offset into the executable which points to the beginning
444 (inclusive, as you'd expect) and end (inclusive, as you would not
445 expect) of the portion of the COFF linetable which corresponds to this
446 include file. @code{C_BINCL} and @code{C_EINCL} do not nest.
449 @section Line Numbers
451 A @code{N_SLINE} symbol represents the start of a source line. The
452 @var{desc} field contains the line number and the @var{value} field
453 contains the code address for the start of that source line. On most
454 machines the address is absolute; for Sun's stabs-in-elf, it is relative
455 to the function in which the @code{N_SLINE} symbol occurs.
457 GNU documents @code{N_DSLINE} and @code{N_BSLINE} symbols for line
458 numbers in the data or bss segments, respectively. They are identical
459 to @code{N_SLINE} but are relocated differently by the linker. They
460 were intended to be used to describe the source location of a variable
461 declaration, but I believe that gcc2 actually puts the line number in
462 the desc field of the stab for the variable itself. GDB has been
463 ignoring these symbols (unless they contain a string field) at least
466 XCOFF uses COFF line numbers instead, which are outside the scope of
467 this document, ammeliorated by adequate marking of include files
468 (@pxref{Source Files}).
470 For single source lines that generate discontiguous code, such as flow
471 of control statements, there may be more than one line number entry for
472 the same source line. In this case there is a line number entry at the
473 start of each code range, each with the same line number.
478 All of the following stabs use the @samp{N_FUN} symbol type.
480 A function is represented by a @samp{F} symbol descriptor for a global
481 (extern) function, and @samp{f} for a static (local) function. The next
482 @samp{N_SLINE} symbol can be used to find the line number of the start
483 of the function. The value field is the address of the start of the
484 function. The type information of the stab represents the return type
485 of the function; thus @samp{foo:f5} means that foo is a function
488 The type information of the stab is optionally followed by type
489 information for each argument, with each argument preceded by @samp{;}.
490 An argument type of 0 means that additional arguments are being passed,
491 whose types and number may vary (@samp{...} in ANSI C). This extension
492 is used by Sun's Solaris compiler. GDB has tolerated it (i.e. at least
493 parsed the syntax, if not necessarily used the information) at least
494 since version 4.8; I don't know whether all versions of dbx will
495 tolerate it. The argument types given here are not merely redundant
496 with the symbols for the arguments themselves (@pxref{Parameters}), they
497 are the types of the arguments as they are passed, before any
498 conversions might take place. For example, if a C function which is
499 declared without a prototype takes a @code{float} argument, the value is
500 passed as a @code{double} but then converted to a @code{float}.
501 Debuggers need to use the types given in the arguments when printing
502 values, but if calling the function they need to use the types given in
503 the symbol defining the function.
505 If the return type and types of arguments of a function which is defined
506 in another source file are specified (i.e. a function prototype in ANSI
507 C), traditionally compilers emit no stab; the only way for the debugger
508 to find the information is if the source file where the function is
509 defined was also compiled with debugging symbols. As an extension the
510 Solaris compiler uses symbol descriptor @samp{P} followed by the return
511 type of the function, followed by the arguments, each preceded by
512 @samp{;}, as in a stab with symbol descriptor @samp{f} or @samp{F}.
513 This use of symbol descriptor @samp{P} can be distinguished from its use
514 for register parameters (@pxref{Parameters}) by the fact that it has
515 symbol type @code{N_FUN}.
517 The AIX documentation also defines symbol descriptor @samp{J} as an
518 internal function. I assume this means a function nested within another
519 function. It also says Symbol descriptor @samp{m} is a module in
520 Modula-2 or extended Pascal.
522 Procedures (functions which do not return values) are represented as
523 functions returning the void type in C. I don't see why this couldn't
524 be used for all languages (inventing a void type for this purpose if
525 necessary), but the AIX documentation defines @samp{I}, @samp{P}, and
526 @samp{Q} for internal, global, and static procedures, respectively.
527 These symbol descriptors are unusual in that they are not followed by
530 For any of the above symbol descriptors, after the symbol descriptor and
531 the type information, there is optionally a comma, followed by the name
532 of the procedure, followed by a comma, followed by a name specifying the
533 scope. The first name is local to the scope specified, and seems to be
534 redundant with the name of the symbol (before the @samp{:}). The name
535 specifying the scope is the name of a procedure specifying that scope.
536 This feature is used by @sc{gcc}, and presumably Pascal, Modula-2, etc.,
537 compilers, for nested functions.
539 If procedures are nested more than one level deep, only the immediately
540 containing scope is specified, for example:
552 return baz (x + 2 * y);
554 return x + bar (3 * x);
562 .stabs "baz:f1,baz,bar",36,0,0,_baz.15 # 36 == N_FUN
563 .stabs "bar:f1,bar,foo",36,0,0,_bar.12
564 .stabs "foo:F1",36,0,0,_foo
567 The stab representing a procedure is located immediately following the
568 code of the procedure. This stab is in turn directly followed by a
569 group of other stabs describing elements of the procedure. These other
570 stabs describe the procedure's parameters, its block local variables and
578 The @code{.stabs} entry after this code fragment shows the @var{name} of
579 the procedure (@code{main}); the type descriptor @var{desc} (@code{F},
580 for a global procedure); a reference to the predefined type @code{int}
581 for the return type; and the starting @var{address} of the procedure.
583 Here is an exploded summary (with whitespace introduced for clarity),
584 followed by line 50 of our sample assembly output, which has this form:
588 @var{desc} @r{(global proc @samp{F})}
589 @var{return_type_ref} @r{(int)}
595 50 .stabs "main:F1",36,0,0,_main
598 @node Block Structure
599 @section Block Structure
601 The program's block structure is represented by the @code{N_LBRAC} (left
602 brace) and the @code{N_RBRAC} (right brace) stab types. The variables
603 defined inside a block preceded the @code{N_LBRAC} symbol for most
604 compilers, including GCC. Other compilers, such as the Convex, Acorn
605 RISC machine, and Sun acc compilers, put the variables after the
606 @code{N_LBRAC} symbol. The values of the @code{N_LBRAC} and
607 @code{N_RBRAC} symbols are the start and end addresses of the code of
608 the block, respectively. For most machines, they are relative to the
609 starting address of this source file. For the Gould NP1, they are
610 absolute. For Sun's stabs-in-elf, they are relative to the function in
613 The @code{N_LBRAC} and @code{N_RBRAC} stabs that describe the block
614 scope of a procedure are located after the @code{N_FUN} stab that
615 represents the procedure itself.
617 Sun documents the @code{desc} field of @code{N_LBRAC} and
618 @code{N_RBRAC} symbols as containing the nesting level of the block.
619 However, dbx seems not to care, and GCC just always set @code{desc} to
625 The @samp{c} symbol descriptor indicates that this stab represents a
626 constant. This symbol descriptor is an exception to the general rule
627 that symbol descriptors are followed by type information. Instead, it
628 is followed by @samp{=} and one of the following:
632 Boolean constant. @var{value} is a numeric value; I assume it is 0 for
636 Character constant. @var{value} is the numeric value of the constant.
638 @item e @var{type-information} , @var{value}
639 Constant whose value can be represented as integral.
640 @var{type-information} is the type of the constant, as it would appear
641 after a symbol descriptor (@pxref{Stabs Format}). @var{value} is the
642 numeric value of the constant. GDB 4.9 does not actually get the right
643 value if @var{value} does not fit in a host @code{int}, but it does not
644 do anything violent, and future debuggers could be extended to accept
645 integers of any size (whether unsigned or not). This constant type is
646 usually documented as being only for enumeration constants, but GDB has
647 never imposed that restriction; I don't know about other debuggers.
650 Integer constant. @var{value} is the numeric value. The type is some
651 sort of generic integer type (for GDB, a host @code{int}); to specify
652 the type explicitly, use @samp{e} instead.
655 Real constant. @var{value} is the real value, which can be @samp{INF}
656 (optionally preceded by a sign) for infinity, @samp{QNAN} for a quiet
657 NaN (not-a-number), or @samp{SNAN} for a signalling NaN. If it is a
658 normal number the format is that accepted by the C library function
662 String constant. @var{string} is a string enclosed in either @samp{'}
663 (in which case @samp{'} characters within the string are represented as
664 @samp{\'} or @samp{"} (in which case @samp{"} characters within the
665 string are represented as @samp{\"}).
667 @item S @var{type-information} , @var{elements} , @var{bits} , @var{pattern}
668 Set constant. @var{type-information} is the type of the constant, as it
669 would appear after a symbol descriptor (@pxref{Stabs Format}).
670 @var{elements} is the number of elements in the set (Does this means
671 how many bits of @var{pattern} are actually used, which would be
672 redundant with the type, or perhaps the number of bits set in
673 @var{pattern}? I don't get it), @var{bits} is the number of bits in the
674 constant (meaning it specifies the length of @var{pattern}, I think),
675 and @var{pattern} is a hexadecimal representation of the set. AIX
676 documentation refers to a limit of 32 bytes, but I see no reason why
677 this limit should exist. This form could probably be used for arbitrary
678 constants, not just sets; the only catch is that @var{pattern} should be
679 understood to be target, not host, byte order and format.
682 The boolean, character, string, and set constants are not supported by
683 GDB 4.9, but it will ignore them. GDB 4.8 and earlier gave an error
684 message and refused to read symbols from the file containing the
687 This information is followed by @samp{;}.
690 @chapter A Comprehensive Example in C
692 Now we'll examine a second program, @code{example2}, which builds on the
693 first example to introduce the rest of the stab types, symbol
694 descriptors, and type descriptors used in C.
695 @xref{Example2.c} for the complete @file{.c} source,
696 and @pxref{Example2.s} for the @file{.s} assembly code.
697 This description includes parts of those files.
699 @section Flow of control and nested scopes
705 @code{N_SLINE}, @code{N_LBRAC}, @code{N_RBRAC} (cont.)
708 Consider the body of @code{main}, from @file{example2.c}. It shows more
709 about how @code{N_SLINE}, @code{N_RBRAC}, and @code{N_LBRAC} stabs are used.
713 21 static float s_flap;
715 23 for (times=0; times < s_g_repeat; times++)@{
717 25 printf ("Hello world\n");
722 Here we have a single source line, the @samp{for} line, that generates
723 non-linear flow of control, and non-contiguous code. In this case, an
724 @code{N_SLINE} stab with the same line number proceeds each block of
725 non-contiguous code generated from the same source line.
727 The example also shows nested scopes. The @code{N_LBRAC} and
728 @code{N_LBRAC} stabs that describe block structure are nested in the
729 same order as the corresponding code blocks, those of the for loop
730 inside those for the body of main.
733 This is the label for the @code{N_LBRAC} (left brace) stab marking the
734 start of @code{main}.
741 In the first code range for C source line 23, the @code{for} loop
742 initialize and test, @code{N_SLINE} (68) records the line number:
749 58 .stabn 68,0,23,LM2
753 62 sethi %hi(_s_g_repeat),%o0
755 64 ld [%o0+%lo(_s_g_repeat)],%o0
760 @exdent label for the @code{N_LBRAC} (start block) marking the start of @code{for} loop
763 69 .stabn 68,0,25,LM3
765 71 sethi %hi(LC0),%o1
766 72 or %o1,%lo(LC0),%o0
769 75 .stabn 68,0,26,LM4
772 @exdent label for the @code{N_RBRAC} (end block) stab marking the end of the @code{for} loop
778 Now we come to the second code range for source line 23, the @code{for}
779 loop increment and return. Once again, @code{N_SLINE} (68) records the
783 .stabn, N_SLINE, NIL,
787 78 .stabn 68,0,23,LM5
795 86 .stabn 68,0,27,LM6
798 @exdent label for the @code{N_RBRAC} (end block) stab marking the end of the @code{for} loop
801 89 .stabn 68,0,27,LM7
806 94 .stabs "main:F1",36,0,0,_main
807 95 .stabs "argc:p1",160,0,0,68
808 96 .stabs "argv:p20=*21=*2",160,0,0,72
809 97 .stabs "s_flap:V12",40,0,0,_s_flap.0
810 98 .stabs "times:1",128,0,0,-20
814 Here is an illustration of stabs describing nested scopes. The scope
815 nesting is reflected in the nested bracketing stabs (@code{N_LBRAC},
819 .stabn N_LBRAC,NIL,NIL,
820 @var{block-start-address}
822 99 .stabn 192,0,0,LBB2 ## begin proc label
823 100 .stabs "inner:1",128,0,0,-24
824 101 .stabn 192,0,0,LBB3 ## begin for label
828 @code{N_RBRAC} (224), ``right brace'' ends a lexical block (scope).
831 .stabn N_RBRAC,NIL,NIL,
832 @var{block-end-address}
834 102 .stabn 224,0,0,LBE3 ## end for label
835 103 .stabn 224,0,0,LBE2 ## end proc label
842 * Automatic variables:: Variables allocated on the stack.
843 * Global Variables:: Variables used by more than one source file.
844 * Register variables:: Variables in registers.
845 * Common Blocks:: Variables statically allocated together.
846 * Statics:: Variables local to one source file.
847 * Parameters:: Variables for arguments to functions.
850 @node Automatic variables
851 @section Locally scoped automatic variables
858 @item Symbol Descriptor:
862 In addition to describing types, the @code{N_LSYM} stab type also
863 describes locally scoped automatic variables. Refer again to the body
864 of @code{main} in @file{example2.c}. It allocates two automatic
865 variables: @samp{times} is scoped to the body of @code{main}, and
866 @samp{inner} is scoped to the body of the @code{for} loop.
867 @samp{s_flap} is locally scoped but not automatic, and will be discussed
872 21 static float s_flap;
874 23 for (times=0; times < s_g_repeat; times++)@{
876 25 printf ("Hello world\n");
881 The @code{N_LSYM} stab for an automatic variable is located just before the
882 @code{N_LBRAC} stab describing the open brace of the block to which it is
886 @exdent @code{N_LSYM} (128): automatic variable, scoped locally to @code{main}
889 @var{type information}",
891 @var{frame-pointer-offset}
893 98 .stabs "times:1",128,0,0,-20
894 99 .stabn 192,0,0,LBB2 ## begin `main' N_LBRAC
896 @exdent @code{N_LSYM} (128): automatic variable, scoped locally to the @code{for} loop
899 @var{type information}",
901 @var{frame-pointer-offset}
903 100 .stabs "inner:1",128,0,0,-24
904 101 .stabn 192,0,0,LBB3 ## begin `for' loop N_LBRAC
907 The symbol descriptor is omitted for automatic variables. Since type
908 information should being with a digit, @samp{-}, or @samp{(}, only
909 digits, @samp{-}, and @samp{(} are precluded from being used for symbol
910 descriptors by this fact. However, the Acorn RISC machine (ARM) is said
911 to get this wrong: it puts out a mere type definition here, without the
912 preceding @code{@var{typenumber}=}. This is a bad idea; there is no
913 guarantee that type descriptors are distinct from symbol descriptors.
915 @node Global Variables
916 @section Global Variables
923 @item Symbol Descriptor:
927 Global variables are represented by the @code{N_GSYM} stab type. The symbol
928 descriptor, following the colon in the string field, is @samp{G}. Following
929 the @samp{G} is a type reference or type definition. In this example it is a
930 type reference to the basic C type, @code{char}. The first source line in
938 yields the following stab. The stab immediately precedes the code that
939 allocates storage for the variable it describes.
942 @exdent @code{N_GSYM} (32): global symbol
947 N_GSYM, NIL, NIL, NIL
949 21 .stabs "g_foo:G2",32,0,0,0
956 The address of the variable represented by the @code{N_GSYM} is not contained
957 in the @code{N_GSYM} stab. The debugger gets this information from the
958 external symbol for the global variable.
960 @node Register variables
961 @section Register variables
963 @c According to an old version of this manual, AIX uses C_RPSYM instead
964 @c of C_RSYM. I am skeptical; this should be verified.
965 Register variables have their own stab type, @code{N_RSYM}, and their
966 own symbol descriptor, @code{r}. The stab's value field contains the
967 number of the register where the variable data will be stored.
969 The value is the register number.
971 AIX defines a separate symbol descriptor @samp{d} for floating point
972 registers. This seems unnecessary---why not just just give floating
973 point registers different register numbers? I have not verified whether
974 the compiler actually uses @samp{d}.
976 If the register is explicitly allocated to a global variable, but not
980 register int g_bar asm ("%g5");
983 the stab may be emitted at the end of the object file, with
984 the other bss symbols.
987 @section Common Blocks
989 A common block is a statically allocated section of memory which can be
990 referred to by several source files. It may contain several variables.
991 I believe @sc{fortran} is the only language with this feature. A
992 @code{N_BCOMM} stab begins a common block and an @code{N_ECOMM} stab
993 ends it. The only thing which is significant about these two stabs is
994 their name, which can be used to look up a normal (non-debugging) symbol
995 which gives the address of the common block. Then each stab between the
996 @code{N_BCOMM} and the @code{N_ECOMM} specifies a member of that common
997 block; its value is the offset within the common block of that variable.
998 The @code{N_ECOML} stab type is documented for this purpose, but Sun's
999 @sc{fortran} compiler uses @code{N_GSYM} instead. The test case I
1000 looked at had a common block local to a function and it used the
1001 @samp{V} symbol descriptor; I assume one would use @samp{S} if not local
1002 to a function (that is, if a common block @emph{can} be anything other
1003 than local to a function).
1006 @section Static Variables
1008 Initialized static variables are represented by the @samp{S} and
1009 @samp{V} symbol descriptors. @samp{S} means file scope static, and
1010 @samp{V} means procedure scope static.
1012 In a.out files, @code{N_STSYM} means the data segment (although gcc
1013 2.4.5 has a bug in that it uses @code{N_FUN}, so neither dbx nor gdb can
1014 find the variables), @code{N_FUN} means the text segment, and
1015 @code{N_LCSYM} means the bss segment.
1017 In xcoff files, each symbol has a section number, so the stab type
1018 need not indicate the segment.
1020 In ecoff files, the storage class is used to specify the section, so the
1021 stab type need not indicate the segment.
1023 @c In ELF files, it apparently is a big mess. See kludge in dbxread.c
1024 @c in GDB. FIXME: Investigate where this kludge comes from.
1026 @c This is the place to mention N_ROSYM; I'd rather do so once I can
1027 @c coherently explain how this stuff works for stabs-in-elf.
1029 For example, the source lines
1032 static const int var_const = 5;
1033 static int var_init = 2;
1034 static int var_noinit;
1038 yield the following stabs:
1041 .stabs "var_const:S1",36,0,0,_var_const ; @r{36 = N_FUN}
1043 .stabs "var_init:S1",38,0,0,_var_init ; @r{38 = N_STSYM}
1045 .stabs "var_noinit:S1",40,0,0,_var_noinit ; @r{40 = N_LCSYM}
1051 Parameters to a function are represented by a stab (or sometimes two,
1052 see below) for each parameter. The stabs are in the order in which the
1053 debugger should print the parameters (i.e. the order in which the
1054 parameters are declared in the source file).
1056 The symbol descriptor @samp{p} is used to refer to parameters which are
1057 in the arglist. Symbols have symbol type @samp{N_PSYM}. The value of
1058 the symbol is the offset relative to the argument list.
1060 If the parameter is passed in a register, then the traditional way to do
1061 this is to provide two symbols for each argument:
1064 .stabs "arg:p1" . . . ; N_PSYM
1065 .stabs "arg:r1" . . . ; N_RSYM
1068 Debuggers are expected to use the second one to find the value, and the
1069 first one to know that it is an argument.
1071 Because this is kind of ugly, some compilers use symbol descriptor
1072 @samp{P} or @samp{R} to indicate an argument which is in a register.
1073 The symbol value is the register number. @samp{P} and @samp{R} mean the
1074 same thing, the difference is that @samp{P} is a GNU invention and
1075 @samp{R} is an IBM (xcoff) invention. As of version 4.9, GDB should
1076 handle either one. Symbol type @samp{C_RPSYM} is used with @samp{R} and
1077 @samp{N_RSYM} is used with @samp{P}.
1079 According to the AIX documentation symbol descriptor @samp{D} is for a
1080 parameter passed in a floating point register. This seems
1081 unnecessary---why not just use @samp{R} with a register number which
1082 indicates that it's a floating point register? I haven't verified
1083 whether the system actually does what the documentation indicates.
1085 There is at least one case where GCC uses a @samp{p}/@samp{r} pair
1086 rather than @samp{P}; this is where the argument is passed in the
1087 argument list and then loaded into a register.
1089 On the sparc and hppa, for a @samp{P} symbol whose type is a structure
1090 or union, the register contains the address of the structure. On the
1091 sparc, this is also true of a @samp{p}/@samp{r} pair (using Sun cc) or a
1092 @samp{p} symbol. However, if a (small) structure is really in a
1093 register, @samp{r} is used. And, to top it all off, on the hppa it
1094 might be a structure which was passed on the stack and loaded into a
1095 register and for which there is a @samp{p}/@samp{r} pair! I believe
1096 that symbol descriptor @samp{i} is supposed to deal with this case, (it
1097 is said to mean "value parameter by reference, indirect access", I don't
1098 know the source for this information) but I don't know details or what
1099 compilers or debuggers use it, if any (not GDB or GCC). It is not clear
1100 to me whether this case needs to be dealt with differently than
1101 parameters passed by reference (see below).
1103 There is another case similar to an argument in a register, which is an
1104 argument which is actually stored as a local variable. Sometimes this
1105 happens when the argument was passed in a register and then the compiler
1106 stores it as a local variable. If possible, the compiler should claim
1107 that it's in a register, but this isn't always done. Some compilers use
1108 the pair of symbols approach described above ("arg:p" followed by
1109 "arg:"); this includes gcc1 (not gcc2) on the sparc when passing a small
1110 structure and gcc2 (sometimes) when the argument type is float and it is
1111 passed as a double and converted to float by the prologue (in the latter
1112 case the type of the "arg:p" symbol is double and the type of the "arg:"
1113 symbol is float). GCC, at least on the 960, uses a single @samp{p}
1114 symbol descriptor for an argument which is stored as a local variable
1115 but uses @samp{N_LSYM} instead of @samp{N_PSYM}. In this case the value
1116 of the symbol is an offset relative to the local variables for that
1117 function, not relative to the arguments (on some machines those are the
1118 same thing, but not on all).
1120 If the parameter is passed by reference (e.g. Pascal VAR parameters),
1121 then type symbol descriptor is @samp{v} if it is in the argument list,
1122 or @samp{a} if it in a register. Other than the fact that these contain
1123 the address of the parameter other than the parameter itself, they are
1124 identical to @samp{p} and @samp{R}, respectively. I believe @samp{a} is
1125 an AIX invention; @samp{v} is supported by all stabs-using systems as
1128 @c Is this paragraph correct? It is based on piecing together patchy
1129 @c information and some guesswork
1130 Conformant arrays refer to a feature of Modula-2, and perhaps other
1131 languages, in which the size of an array parameter is not known to the
1132 called function until run-time. Such parameters have two stabs, a
1133 @samp{x} for the array itself, and a @samp{C}, which represents the size
1134 of the array. The value of the @samp{x} stab is the offset in the
1135 argument list where the address of the array is stored (it this right?
1136 it is a guess); the value of the @samp{C} stab is the offset in the
1137 argument list where the size of the array (in elements? in bytes?) is
1140 The following are also said to go with @samp{N_PSYM}:
1143 "name" -> "param_name:#type"
1145 -> pF FORTRAN function parameter
1146 -> X (function result variable)
1147 -> b (based variable)
1149 value -> offset from the argument pointer (positive).
1152 As a simple example, the code
1164 .stabs "main:F1",36,0,0,_main ; 36 is N_FUN
1165 .stabs "argc:p1",160,0,0,68 ; 160 is N_PSYM
1166 .stabs "argv:p20=*21=*2",160,0,0,72
1169 The type definition of argv is interesting because it contains several
1170 type definitions. Type 21 is pointer to type 2 (char) and argv (type 20) is
1174 @chapter Type Definitions
1176 Now let's look at some variable definitions involving complex types.
1177 This involves understanding better how types are described. In the
1178 examples so far types have been described as references to previously
1179 defined types or defined in terms of subranges of or pointers to
1180 previously defined types. The section that follows will talk about
1181 the various other type descriptors that may follow the = sign in a
1185 * Builtin types:: Integers, floating point, void, etc.
1186 * Miscellaneous Types:: Pointers, sets, files, etc.
1187 * Cross-references:: Referring to a type not yet defined.
1188 * Subranges:: A type with a specific range.
1189 * Arrays:: An aggregate type of same-typed elements.
1190 * Strings:: Like an array but also has a length.
1191 * Enumerations:: Like an integer but the values have names.
1192 * Structures:: An aggregate type of different-typed elements.
1193 * Typedefs:: Giving a type a name.
1194 * Unions:: Different types sharing storage.
1199 @section Builtin types
1201 Certain types are built in (@code{int}, @code{short}, @code{void},
1202 @code{float}, etc.); the debugger recognizes these types and knows how
1203 to handle them. Thus don't be surprised if some of the following ways
1204 of specifying builtin types do not specify everything that a debugger
1205 would need to know about the type---in some cases they merely specify
1206 enough information to distinguish the type from other types.
1208 The traditional way to define builtin types is convolunted, so new ways
1209 have been invented to describe them. Sun's ACC uses the @samp{b} and
1210 @samp{R} type descriptors, and IBM uses negative type numbers. GDB can
1211 accept all three, as of version 4.8; dbx just accepts the traditional
1212 builtin types and perhaps one of the other two formats.
1215 * Traditional Builtin Types:: Put on your seatbelts and prepare for kludgery
1216 * Builtin Type Descriptors:: Builtin types with special type descriptors
1217 * Negative Type Numbers:: Builtin types using negative type numbers
1220 @node Traditional Builtin Types
1221 @subsection Traditional Builtin types
1223 Often types are defined as subranges of themselves. If the array bounds
1224 can fit within an @code{int}, then they are given normally. For example:
1227 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0 ; 128 is N_LSYM
1228 .stabs "char:t2=r2;0;127;",128,0,0,0
1231 Builtin types can also be described as subranges of @code{int}:
1234 .stabs "unsigned short:t6=r1;0;65535;",128,0,0,0
1237 If the lower bound of a subrange is 0 and the upper bound is -1, it
1238 means that the type is an unsigned integral type whose bounds are too
1239 big to describe in an int. Traditionally this is only used for
1240 @code{unsigned int} and @code{unsigned long}; GCC also sometimes uses it
1241 for @code{long long} and @code{unsigned long long}, and the only way to
1242 tell those types apart is to look at their names. On other machines GCC
1243 puts out bounds in octal, with a leading 0. In this case a negative
1244 bound consists of a number which is a 1 bit followed by a bunch of 0
1245 bits, and a positive bound is one in which a bunch of bits are 1.
1248 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
1249 .stabs "long long int:t7=r1;0;-1;",128,0,0,0
1252 If the lower bound of a subrange is 0 and the upper bound is negative,
1253 it means that it is an unsigned integral type whose size in bytes is the
1254 absolute value of the upper bound. I believe this is a Convex
1255 convention for @code{unsigned long long}.
1257 If the lower bound of a subrange is negative and the upper bound is 0,
1258 it means that the type is a signed integral type whose size in bytes is
1259 the absolute value of the lower bound. I believe this is a Convex
1260 convention for @code{long long}. To distinguish this from a legitimate
1261 subrange, the type should be a subrange of itself. I'm not sure whether
1262 this is the case for Convex.
1264 If the upper bound of a subrange is 0, it means that this is a floating
1265 point type, and the lower bound of the subrange indicates the number of
1269 .stabs "float:t12=r1;4;0;",128,0,0,0
1270 .stabs "double:t13=r1;8;0;",128,0,0,0
1273 However, GCC writes @code{long double} the same way it writes
1274 @code{double}; the only way to distinguish them is by the name:
1277 .stabs "long double:t14=r1;8;0;",128,0,0,0
1280 Complex types are defined the same way as floating-point types; the only
1281 way to distinguish a single-precision complex from a double-precision
1282 floating-point type is by the name.
1284 The C @code{void} type is defined as itself:
1287 .stabs "void:t15=15",128,0,0,0
1290 I'm not sure how a boolean type is represented.
1292 @node Builtin Type Descriptors
1293 @subsection Defining Builtin Types using Builtin Type Descriptors
1295 There are various type descriptors to define builtin types:
1298 @c FIXME: clean up description of width and offset, once we figure out
1300 @item b @var{signed} @var{char-flag} @var{width} ; @var{offset} ; @var{nbits} ;
1301 Define an integral type. @var{signed} is @samp{u} for unsigned or
1302 @samp{s} for signed. @var{char-flag} is @samp{c} which indicates this
1303 is a character type, or is omitted. I assume this is to distinguish an
1304 integral type from a character type of the same size, for example it
1305 might make sense to set it for the C type @code{wchar_t} so the debugger
1306 can print such variables differently (Solaris does not do this). Sun
1307 sets it on the C types @code{signed char} and @code{unsigned char} which
1308 arguably is wrong. @var{width} and @var{offset} appear to be for small
1309 objects stored in larger ones, for example a @code{short} in an
1310 @code{int} register. @var{width} is normally the number of bytes in the
1311 type. @var{offset} seems to always be zero. @var{nbits} is the number
1312 of bits in the type.
1314 Note that type descriptor @samp{b} used for builtin types conflicts with
1315 its use for Pascal space types (@pxref{Miscellaneous Types}); they can
1316 be distinguished because the character following the type descriptor
1317 will be a digit, @samp{(}, or @samp{-} for a Pascal space type, or
1318 @samp{u} or @samp{s} for a builtin type.
1321 Documented by AIX to define a wide character type, but their compiler
1322 actually uses negative type numbers (@pxref{Negative Type Numbers}).
1324 @item R @var{fp_type} ; @var{bytes} ;
1325 Define a floating point type. @var{fp_type} has one of the following values:
1329 IEEE 32-bit (single precision) floating point format.
1332 IEEE 64-bit (double precision) floating point format.
1334 @item 3 (NF_COMPLEX)
1335 @item 4 (NF_COMPLEX16)
1336 @item 5 (NF_COMPLEX32)
1337 @c "GDB source" really means @file{include/aout/stab_gnu.h}, but trying
1338 @c to put that here got an overfull hbox.
1339 These are for complex numbers. A comment in the GDB source describes
1340 them as Fortran complex, double complex, and complex*16, respectively,
1341 but what does that mean? (i.e. Single precision? Double precison?).
1343 @item 6 (NF_LDOUBLE)
1344 Long double. This should probably only be used for Sun format long
1345 double, and new codes should be used for other floating point formats
1346 (NF_DOUBLE can be used if a long double is really just an IEEE double,
1350 @var{bytes} is the number of bytes occupied by the type. This allows a
1351 debugger to perform some operations with the type even if it doesn't
1352 understand @var{fp_code}.
1354 @item g @var{type-information} ; @var{nbits}
1355 Documented by AIX to define a floating type, but their compiler actually
1356 uses negative type numbers (@pxref{Negative Type Numbers}).
1358 @item c @var{type-information} ; @var{nbits}
1359 Documented by AIX to define a complex type, but their compiler actually
1360 uses negative type numbers (@pxref{Negative Type Numbers}).
1363 The C @code{void} type is defined as a signed integral type 0 bits long:
1365 .stabs "void:t19=bs0;0;0",128,0,0,0
1367 The Solaris compiler seems to omit the trailing semicolon in this case.
1368 Getting sloppy in this way is not a swift move because if a type is
1369 embedded in a more complex expression it is necessary to be able to tell
1372 I'm not sure how a boolean type is represented.
1374 @node Negative Type Numbers
1375 @subsection Negative Type numbers
1377 Since the debugger knows about the builtin types anyway, the idea of
1378 negative type numbers is simply to give a special type number which
1379 indicates the built in type. There is no stab defining these types.
1381 I'm not sure whether anyone has tried to define what this means if
1382 @code{int} can be other than 32 bits (or other types can be other than
1383 their customary size). If @code{int} has exactly one size for each
1384 architecture, then it can be handled easily enough, but if the size of
1385 @code{int} can vary according the compiler options, then it gets hairy.
1386 The best way to do this would be to define separate negative type
1387 numbers for 16-bit @code{int} and 32-bit @code{int}; therefore I have
1388 indicated below the customary size (and other format information) for
1389 each type. The information below is currently correct because AIX on
1390 the RS6000 is the only system which uses these type numbers. If these
1391 type numbers start to get used on other systems, I suspect the correct
1392 thing to do is to define a new number in cases where a type does not
1393 have the size and format indicated below (or avoid negative type numbers
1396 Also note that part of the definition of the negative type number is
1397 the name of the type. Types with identical size and format but
1398 different names have different negative type numbers.
1402 @code{int}, 32 bit signed integral type.
1405 @code{char}, 8 bit type holding a character. Both GDB and dbx on AIX
1406 treat this as signed. GCC uses this type whether @code{char} is signed
1407 or not, which seems like a bad idea. The AIX compiler (xlc) seems to
1408 avoid this type; it uses -5 instead for @code{char}.
1411 @code{short}, 16 bit signed integral type.
1414 @code{long}, 32 bit signed integral type.
1417 @code{unsigned char}, 8 bit unsigned integral type.
1420 @code{signed char}, 8 bit signed integral type.
1423 @code{unsigned short}, 16 bit unsigned integral type.
1426 @code{unsigned int}, 32 bit unsigned integral type.
1429 @code{unsigned}, 32 bit unsigned integral type.
1432 @code{unsigned long}, 32 bit unsigned integral type.
1435 @code{void}, type indicating the lack of a value.
1438 @code{float}, IEEE single precision.
1441 @code{double}, IEEE double precision.
1444 @code{long double}, IEEE double precision. The compiler claims the size
1445 will increase in a future release, and for binary compatibility you have
1446 to avoid using @code{long double}. I hope when they increase it they
1447 use a new negative type number.
1450 @code{integer}. 32 bit signed integral type.
1453 @code{boolean}. 32 bit type. How is the truth value encoded? Is it
1454 the least significant bit or is it a question of whether the whole value
1455 is zero or non-zero?
1458 @code{short real}. IEEE single precision.
1461 @code{real}. IEEE double precision.
1464 @code{stringptr}. @xref{Strings}.
1467 @code{character}, 8 bit unsigned character type.
1470 @code{logical*1}, 8 bit type. This @sc{fortran} type has a split
1471 personality in that it is used for boolean variables, but can also be
1472 used for unsigned integers. 0 is false, 1 is true, and other values are
1476 @code{logical*2}, 16 bit type. This @sc{fortran} type has a split
1477 personality in that it is used for boolean variables, but can also be
1478 used for unsigned integers. 0 is false, 1 is true, and other values are
1482 @code{logical*4}, 32 bit type. This @sc{fortran} type has a split
1483 personality in that it is used for boolean variables, but can also be
1484 used for unsigned integers. 0 is false, 1 is true, and other values are
1488 @code{logical}, 32 bit type. This @sc{fortran} type has a split
1489 personality in that it is used for boolean variables, but can also be
1490 used for unsigned integers. 0 is false, 1 is true, and other values are
1494 @code{complex}. A complex type consisting of two IEEE single-precision
1495 floating point values.
1498 @code{complex}. A complex type consisting of two IEEE double-precision
1499 floating point values.
1502 @code{integer*1}, 8 bit signed integral type.
1505 @code{integer*2}, 16 bit signed integral type.
1508 @code{integer*4}, 32 bit signed integral type.
1511 @code{wchar}. Wide character, 16 bits wide, unsigned (what format?
1515 @node Miscellaneous Types
1516 @section Miscellaneous Types
1519 @item b @var{type-information} ; @var{bytes}
1520 Pascal space type. This is documented by IBM; what does it mean?
1522 Note that this use of the @samp{b} type descriptor can be distinguished
1523 from its use for builtin integral types (@pxref{Builtin Type
1524 Descriptors}) because the character following the type descriptor is
1525 always a digit, @samp{(}, or @samp{-}.
1527 @item B @var{type-information}
1528 A volatile-qualified version of @var{type-information}. This is a Sun
1529 extension. A volatile-qualified type means that references and stores
1530 to a variable of that type must not be optimized or cached; they must
1531 occur as the user specifies them.
1533 @item d @var{type-information}
1534 File of type @var{type-information}. As far as I know this is only used
1537 @item k @var{type-information}
1538 A const-qualified version of @var{type-information}. This is a Sun
1539 extension. A const-qualified type means that a variable of this type
1542 @item M @var{type-information} ; @var{length}
1543 Multiple instance type. The type seems to composed of @var{length}
1544 repetitions of @var{type-information}, for example @code{character*3} is
1545 represented by @samp{M-2;3}, where @samp{-2} is a reference to a
1546 character type (@pxref{Negative Type Numbers}). I'm not sure how this
1547 differs from an array. This appears to be a FORTRAN feature.
1548 @var{length} is a bound, like those in range types, @xref{Subranges}.
1550 @item S @var{type-information}
1551 Pascal set type. @var{type-information} must be a small type such as an
1552 enumeration or a subrange, and the type is a bitmask whose length is
1553 specified by the number of elements in @var{type-information}.
1555 @item * @var{type-information}
1556 Pointer to @var{type-information}.
1559 @node Cross-references
1560 @section Cross-references to other types
1562 If a type is used before it is defined, one common way to deal with this
1563 is just to use a type reference to a type which has not yet been
1564 defined. The debugger is expected to be able to deal with this.
1566 Another way is with the @samp{x} type descriptor, which is followed by
1567 @samp{s} for a structure tag, @samp{u} for a union tag, or @samp{e} for
1568 a enumerator tag, followed by the name of the tag, followed by @samp{:}.
1569 for example the following C declarations:
1579 .stabs "bar:G16=*17=xsfoo:",32,0,0,0
1582 Not all debuggers support the @samp{x} type descriptor, so on some
1583 machines GCC does not use it. I believe that for the above example it
1584 would just emit a reference to type 17 and never define it, but I
1585 haven't verified that.
1587 Modula-2 imported types, at least on AIX, use the @samp{i} type
1588 descriptor, which is followed by the name of the module from which the
1589 type is imported, followed by @samp{:}, followed by the name of the
1590 type. There is then optionally a comma followed by type information for
1591 the type (This differs from merely naming the type (@pxref{Typedefs}) in
1592 that it identifies the module; I don't understand whether the name of
1593 the type given here is always just the same as the name we are giving
1594 it, or whether this type descriptor is used with a nameless stab
1595 (@pxref{Stabs Format}), or what). The symbol ends with @samp{;}.
1598 @section Subrange types
1600 The @samp{r} type descriptor defines a type as a subrange of another
1601 type. It is followed by type information for the type which it is a
1602 subrange of, a semicolon, an integral lower bound, a semicolon, an
1603 integral upper bound, and a semicolon. The AIX documentation does not
1604 specify the trailing semicolon, in an effort to specify array indexes
1605 more cleanly, but a subrange which is not an array index has always
1606 included a trailing semicolon (@pxref{Arrays}).
1608 Instead of an integer, either bound can be one of the following:
1611 @item A @var{offset}
1612 The bound is passed by reference on the stack at offset @var{offset}
1613 from the argument list. @xref{Parameters}, for more information on such
1616 @item T @var{offset}
1617 The bound is passed by value on the stack at offset @var{offset} from
1620 @item a @var{register-number}
1621 The bound is pased by reference in register number
1622 @var{register-number}.
1624 @item t @var{register-number}
1625 The bound is passed by value in register number @var{register-number}.
1631 Subranges are also used for builtin types, @xref{Traditional Builtin Types}.
1634 @section Array types
1636 Arrays use the @samp{a} type descriptor. Following the type descriptor
1637 is the type of the index and the type of the array elements. If the
1638 index type is a range type, it will end in a semicolon; if it is not a
1639 range type (for example, if it is a type reference), there does not
1640 appear to be any way to tell where the types are separated. In an
1641 effort to clean up this mess, IBM documents the two types as being
1642 separated by a semicolon, and a range type as not ending in a semicolon
1643 (but this is not right for range types which are not array indexes,
1644 @pxref{Subranges}). I think probably the best solution is to specify
1645 that a semicolon ends a range type, and that the index type and element
1646 type of an array are separated by a semicolon, but that if the index
1647 type is a range type, the extra semicolon can be omitted. GDB (at least
1648 through version 4.9) doesn't support any kind of index type other than a
1649 range anyway; I'm not sure about dbx.
1651 It is well established, and widely used, that the type of the index,
1652 unlike most types found in the stabs, is merely a type definition, not
1653 type information (@pxref{Stabs Format}) (that is, it need not start with
1654 @var{type-number}@code{=} if it is defining a new type). According to a
1655 comment in GDB, this is also true of the type of the array elements; it
1656 gives @samp{ar1;1;10;ar1;1;10;4} as a legitimate way to express a two
1657 dimensional array. According to AIX documentation, the element type
1658 must be type information. GDB accepts either.
1660 The type of the index is often a range type, expressed as the letter r
1661 and some parameters. It defines the size of the array. In the example
1662 below, the range @code{r1;0;2;} defines an index type which is a
1663 subrange of type 1 (integer), with a lower bound of 0 and an upper bound
1664 of 2. This defines the valid range of subscripts of a three-element C
1667 For example, the definition
1670 char char_vec[3] = @{'a','b','c'@};
1677 .stabs "char_vec:G19=ar1;0;2;2",32,0,0,0
1686 If an array is @dfn{packed}, it means that the elements are spaced more
1687 closely than normal, saving memory at the expense of speed. For
1688 example, an array of 3-byte objects might, if unpacked, have each
1689 element aligned on a 4-byte boundary, but if packed, have no padding.
1690 One way to specify that something is packed is with type attributes
1691 (@pxref{Stabs Format}), in the case of arrays another is to use the
1692 @samp{P} type descriptor instead of @samp{a}. Other than specifying a
1693 packed array, @samp{P} is identical to @samp{a}.
1695 @c FIXME-what is it? A pointer?
1696 An open array is represented by the @samp{A} type descriptor followed by
1697 type information specifying the type of the array elements.
1699 @c FIXME: what is the format of this type? A pointer to a vector of pointers?
1700 An N-dimensional dynamic array is represented by
1703 D @var{dimensions} ; @var{type-information}
1706 @c Does dimensions really have this meaning? The AIX documentation
1708 @var{dimensions} is the number of dimensions; @var{type-information}
1709 specifies the type of the array elements.
1711 @c FIXME: what is the format of this type? A pointer to some offsets in
1713 A subarray of an N-dimensional array is represented by
1716 E @var{dimensions} ; @var{type-information}
1719 @c Does dimensions really have this meaning? The AIX documentation
1721 @var{dimensions} is the number of dimensions; @var{type-information}
1722 specifies the type of the array elements.
1727 Some languages, like C or the original Pascal, do not have string types,
1728 they just have related things like arrays of characters. But most
1729 Pascals and various other languages have string types, which are
1730 indicated as follows:
1733 @item n @var{type-information} ; @var{bytes}
1734 @var{bytes} is the maximum length. I'm not sure what
1735 @var{type-information} is; I suspect that it means that this is a string
1736 of @var{type-information} (thus allowing a string of integers, a string
1737 of wide characters, etc., as well as a string of characters). Not sure
1738 what the format of this type is. This is an AIX feature.
1740 @item z @var{type-information} ; @var{bytes}
1741 Just like @samp{n} except that this is a gstring, not an ordinary
1742 string. I don't know the difference.
1745 Pascal Stringptr. What is this? This is an AIX feature.
1749 @section Enumerations
1751 Enumerations are defined with the @samp{e} type descriptor.
1753 @c FIXME: Where does this information properly go? Perhaps it is
1754 @c redundant with something we already explain.
1755 The source line below declares an enumeration type. It is defined at
1756 file scope between the bodies of main and s_proc in example2.c.
1757 The type definition is located after the N_RBRAC that marks the end of
1758 the previous procedure's block scope, and before the N_FUN that marks
1759 the beginning of the next procedure's block scope. Therefore it does not
1760 describe a block local symbol, but a file local one.
1765 enum e_places @{first,second=3,last@};
1769 generates the following stab
1772 .stabs "e_places:T22=efirst:0,second:3,last:4,;",128,0,0,0
1775 The symbol descriptor (T) says that the stab describes a structure,
1776 enumeration, or type tag. The type descriptor e, following the 22= of
1777 the type definition narrows it down to an enumeration type. Following
1778 the e is a list of the elements of the enumeration. The format is
1779 name:value,. The list of elements ends with a ;.
1781 There is no standard way to specify the size of an enumeration type; it
1782 is determined by the architecture (normally all enumerations types are
1783 32 bits). There should be a way to specify an enumeration type of
1784 another size; type attributes would be one way to do this @xref{Stabs
1794 @code{N_LSYM} or @code{C_DECL}
1795 @item Symbol Descriptor:
1797 @item Type Descriptor:
1801 The following source code declares a structure tag and defines an
1802 instance of the structure in global scope. Then a typedef equates the
1803 structure tag with a new type. A seperate stab is generated for the
1804 structure tag, the structure typedef, and the structure instance. The
1805 stabs for the tag and the typedef are emited when the definitions are
1806 encountered. Since the structure elements are not initialized, the
1807 stab and code for the structure variable itself is located at the end
1808 of the program in .common.
1814 9 char s_char_vec[8];
1815 10 struct s_tag* s_next;
1818 13 typedef struct s_tag s_typedef;
1821 The structure tag is an N_LSYM stab type because, like the enum, the
1822 symbol is file scope. Like the enum, the symbol descriptor is T, for
1823 enumeration, struct or tag type. The symbol descriptor s following
1824 the 16= of the type definition narrows the symbol type to struct.
1826 Following the struct symbol descriptor is the number of bytes the
1827 struct occupies, followed by a description of each structure element.
1828 The structure element descriptions are of the form name:type, bit
1829 offset from the start of the struct, and number of bits in the
1834 <128> N_LSYM - type definition
1835 .stabs "name:sym_desc(struct tag) Type_def(16)=type_desc(struct type)
1837 elem_name:type_ref(int),bit_offset,field_bits;
1838 elem_name:type_ref(float),bit_offset,field_bits;
1839 elem_name:type_def(17)=type_desc(array)
1840 index_type(range of int from 0 to 7);
1841 element_type(char),bit_offset,field_bits;;",
1844 30 .stabs "s_tag:T16=s20s_int:1,0,32;s_float:12,32,32;
1845 s_char_vec:17=ar1;0;7;2,64,64;s_next:18=*16,128,32;;",128,0,0,0
1848 In this example, two of the structure elements are previously defined
1849 types. For these, the type following the name: part of the element
1850 description is a simple type reference. The other two structure
1851 elements are new types. In this case there is a type definition
1852 embedded after the name:. The type definition for the array element
1853 looks just like a type definition for a standalone array. The s_next
1854 field is a pointer to the same kind of structure that the field is an
1855 element of. So the definition of structure type 16 contains an type
1856 definition for an element which is a pointer to type 16.
1859 @section Giving a type a name
1861 To give a type a name, use the @samp{t} symbol descriptor. For example,
1864 .stabs "s_typedef:t16",128,0,0,0
1867 specifies that @code{s_typedef} refers to type number 16. Such stabs
1868 have symbol type @code{N_LSYM} or @code{C_DECL}.
1870 If instead, you are specifying the tag name for a structure, union, or
1871 enumeration, use the @samp{T} symbol descriptor instead. I believe C is
1872 the only language with this feature.
1874 If the type is an opaque type (I believe this is a Modula-2 feature),
1875 AIX provides a type descriptor to specify it. The type descriptor is
1876 @samp{o} and is followed by a name. I don't know what the name
1877 means---is it always the same as the name of the type, or is this type
1878 descriptor used with a nameless stab (@pxref{Stabs Format})? There
1879 optionally follows a comma followed by type information which defines
1880 the type of this type. If omitted, a semicolon is used in place of the
1881 comma and the type information, and, the type is much like a generic
1882 pointer type---it has a known size but little else about it is
1888 Next let's look at unions. In example2 this union type is declared
1889 locally to a procedure and an instance of the union is defined.
1899 This code generates a stab for the union tag and a stab for the union
1900 variable. Both use the N_LSYM stab type. Since the union variable is
1901 scoped locally to the procedure in which it is defined, its stab is
1902 located immediately preceding the N_LBRAC for the procedure's block
1905 The stab for the union tag, however is located preceding the code for
1906 the procedure in which it is defined. The stab type is N_LSYM. This
1907 would seem to imply that the union type is file scope, like the struct
1908 type s_tag. This is not true. The contents and position of the stab
1909 for u_type do not convey any infomation about its procedure local
1914 .stabs "name:sym_desc(union tag)type_def(22)=type_desc(union)
1916 elem_name:type_ref(int),bit_offset(0),bit_size(32);
1917 elem_name:type_ref(float),bit_offset(0),bit_size(32);
1918 elem_name:type_ref(ptr to char),bit_offset(0),bit_size(32);;"
1919 N_LSYM, NIL, NIL, NIL
1923 105 .stabs "u_tag:T23=u4u_int:1,0,32;u_float:12,0,32;u_char:21,0,32;;",
1927 The symbol descriptor, T, following the name: means that the stab
1928 describes an enumeration, struct or type tag. The type descriptor u,
1929 following the 23= of the type definition, narrows it down to a union
1930 type definition. Following the u is the number of bytes in the union.
1931 After that is a list of union element descriptions. Their format is
1932 name:type, bit offset into the union, and number of bytes for the
1935 The stab for the union variable follows. Notice that the frame
1936 pointer offset for local variables is negative.
1939 <128> N_LSYM - local variable (with no symbol descriptor)
1940 .stabs "name:type_ref(u_tag)", N_LSYM, NIL, NIL, frame_ptr_offset
1944 130 .stabs "an_u:23",128,0,0,-20
1947 @node Function Types
1948 @section Function types
1950 There are various types for function variables. These types are not
1951 used in defining functions; see symbol descriptor @samp{f}; they are
1952 used for things like pointers to functions.
1954 The simple, traditional, type is type descriptor @samp{f} is followed by
1955 type information for the return type of the function, followed by a
1958 This does not deal with functions the number and type of whose
1959 parameters are part of their type, as found in Modula-2 or ANSI C. AIX
1960 provides extensions to specify these, using the @samp{f}, @samp{F},
1961 @samp{p}, and @samp{R} type descriptors.
1963 First comes the type descriptor. Then, if it is @samp{f} or @samp{F},
1964 this is a function, and the type information for the return type of the
1965 function follows, followed by a comma. Then comes the number of
1966 parameters to the function and a semicolon. Then, for each parameter,
1967 there is the name of the parameter followed by a colon (this is only
1968 present for type descriptors @samp{R} and @samp{F} which represent
1969 Pascal function or procedure parameters), type information for the
1970 parameter, a comma, @samp{0} if passed by reference or @samp{1} if
1971 passed by value, and a semicolon. The type definition ends with a
1981 generates the following code:
1984 .stabs "g_pf:G24=*25=f1",32,0,0,0
1985 .common _g_pf,4,"bss"
1988 The variable defines a new type, 24, which is a pointer to another new
1989 type, 25, which is defined as a function returning int.
1992 @chapter Symbol information in symbol tables
1994 This section examines more closely the format of symbol table entries
1995 and how stab assembler directives map to them. It also describes what
1996 transformations the assembler and linker make on data from stabs.
1998 Each time the assembler encounters a stab in its input file it puts
1999 each field of the stab into corresponding fields in a symbol table
2000 entry of its output file. If the stab contains a string field, the
2001 symbol table entry for that stab points to a string table entry
2002 containing the string data from the stab. Assembler labels become
2003 relocatable addresses. Symbol table entries in a.out have the format:
2006 struct internal_nlist @{
2007 unsigned long n_strx; /* index into string table of name */
2008 unsigned char n_type; /* type of symbol */
2009 unsigned char n_other; /* misc info (usually empty) */
2010 unsigned short n_desc; /* description field */
2011 bfd_vma n_value; /* value of symbol */
2015 For .stabs directives, the n_strx field holds the character offset
2016 from the start of the string table to the string table entry
2017 containing the "string" field. For other classes of stabs (.stabn and
2018 .stabd) this field is null.
2020 Symbol table entries with n_type fields containing a value greater or
2021 equal to 0x20 originated as stabs generated by the compiler (with one
2022 random exception). Those with n_type values less than 0x20 were
2023 placed in the symbol table of the executable by the assembler or the
2026 The linker concatenates object files and does fixups of externally
2027 defined symbols. You can see the transformations made on stab data by
2028 the assembler and linker by examining the symbol table after each pass
2029 of the build, first the assemble and then the link.
2031 To do this use nm with the -ap options. This dumps the symbol table,
2032 including debugging information, unsorted. For stab entries the
2033 columns are: value, other, desc, type, string. For assembler and
2034 linker symbols, the columns are: value, type, string.
2036 There are a few important things to notice about symbol tables. Where
2037 the value field of a stab contains a frame pointer offset, or a
2038 register number, that value is unchanged by the rest of the build.
2040 Where the value field of a stab contains an assembly language label,
2041 it is transformed by each build step. The assembler turns it into a
2042 relocatable address and the linker turns it into an absolute address.
2043 This source line defines a static variable at file scope:
2046 3 static int s_g_repeat
2050 The following stab describes the symbol.
2053 26 .stabs "s_g_repeat:S1",38,0,0,_s_g_repeat
2057 The assembler transforms the stab into this symbol table entry in the
2058 @file{.o} file. The location is expressed as a data segment offset.
2061 21 00000084 - 00 0000 STSYM s_g_repeat:S1
2065 in the symbol table entry from the executable, the linker has made the
2066 relocatable address absolute.
2069 22 0000e00c - 00 0000 STSYM s_g_repeat:S1
2072 Stabs for global variables do not contain location information. In
2073 this case the debugger finds location information in the assembler or
2074 linker symbol table entry describing the variable. The source line:
2084 21 .stabs "g_foo:G2",32,0,0,0
2087 The variable is represented by the following two symbol table entries
2088 in the object file. The first one originated as a stab. The second
2089 one is an external symbol. The upper case D signifies that the n_type
2090 field of the symbol table contains 7, N_DATA with local linkage (see
2091 Table B). The value field following the file's line number is empty
2092 for the stab entry. For the linker symbol it contains the
2093 rellocatable address corresponding to the variable.
2096 19 00000000 - 00 0000 GSYM g_foo:G2
2097 20 00000080 D _g_foo
2101 These entries as transformed by the linker. The linker symbol table
2102 entry now holds an absolute address.
2105 21 00000000 - 00 0000 GSYM g_foo:G2
2107 215 0000e008 D _g_foo
2111 @chapter GNU C++ stabs
2114 * Basic Cplusplus types::
2117 * Methods:: Method definition
2119 * Method Modifiers::
2122 * Virtual Base Classes::
2126 @subsection type descriptors added for C++ descriptions
2130 method type (two ## if minimal debug)
2133 Member (class and variable) type. It is followed by type information
2134 for the offset basetype, a comma, and type information for the type of
2135 the field being pointed to. (FIXME: this is acknowledged to be
2136 gibberish. Can anyone say what really goes here?).
2138 Note that there is a conflict between this and type attributes
2139 (@pxref{Stabs Format}); both use type descriptor @samp{@@}.
2140 Fortunately, the @samp{@@} type descriptor used in this C++ sense always
2141 will be followed by a digit, @samp{(}, or @samp{-}, and type attributes
2142 never start with those things.
2145 @node Basic Cplusplus types
2146 @section Basic types for C++
2148 << the examples that follow are based on a01.C >>
2151 C++ adds two more builtin types to the set defined for C. These are
2152 the unknown type and the vtable record type. The unknown type, type
2153 16, is defined in terms of itself like the void type.
2155 The vtable record type, type 17, is defined as a structure type and
2156 then as a structure tag. The structure has four fields, delta, index,
2157 pfn, and delta2. pfn is the function pointer.
2159 << In boilerplate $vtbl_ptr_type, what are the fields delta,
2160 index, and delta2 used for? >>
2162 This basic type is present in all C++ programs even if there are no
2163 virtual methods defined.
2166 .stabs "struct_name:sym_desc(type)type_def(17)=type_desc(struct)struct_bytes(8)
2167 elem_name(delta):type_ref(short int),bit_offset(0),field_bits(16);
2168 elem_name(index):type_ref(short int),bit_offset(16),field_bits(16);
2169 elem_name(pfn):type_def(18)=type_desc(ptr to)type_ref(void),
2170 bit_offset(32),field_bits(32);
2171 elem_name(delta2):type_def(short int);bit_offset(32),field_bits(16);;"
2176 .stabs "$vtbl_ptr_type:t17=s8
2177 delta:6,0,16;index:6,16,16;pfn:18=*15,32,32;delta2:6,32,16;;"
2182 .stabs "name:sym_dec(struct tag)type_ref($vtbl_ptr_type)",N_LSYM,NIL,NIL,NIL
2186 .stabs "$vtbl_ptr_type:T17",128,0,0,0
2189 @node Simple classes
2190 @section Simple class definition
2192 The stabs describing C++ language features are an extension of the
2193 stabs describing C. Stabs representing C++ class types elaborate
2194 extensively on the stab format used to describe structure types in C.
2195 Stabs representing class type variables look just like stabs
2196 representing C language variables.
2198 Consider the following very simple class definition.
2204 int Ameth(int in, char other);
2208 The class baseA is represented by two stabs. The first stab describes
2209 the class as a structure type. The second stab describes a structure
2210 tag of the class type. Both stabs are of stab type N_LSYM. Since the
2211 stab is not located between an N_FUN and a N_LBRAC stab this indicates
2212 that the class is defined at file scope. If it were, then the N_LSYM
2213 would signify a local variable.
2215 A stab describing a C++ class type is similar in format to a stab
2216 describing a C struct, with each class member shown as a field in the
2217 structure. The part of the struct format describing fields is
2218 expanded to include extra information relevent to C++ class members.
2219 In addition, if the class has multiple base classes or virtual
2220 functions the struct format outside of the field parts is also
2223 In this simple example the field part of the C++ class stab
2224 representing member data looks just like the field part of a C struct
2225 stab. The section on protections describes how its format is
2226 sometimes extended for member data.
2228 The field part of a C++ class stab representing a member function
2229 differs substantially from the field part of a C struct stab. It
2230 still begins with `name:' but then goes on to define a new type number
2231 for the member function, describe its return type, its argument types,
2232 its protection level, any qualifiers applied to the method definition,
2233 and whether the method is virtual or not. If the method is virtual
2234 then the method description goes on to give the vtable index of the
2235 method, and the type number of the first base class defining the
2238 When the field name is a method name it is followed by two colons
2239 rather than one. This is followed by a new type definition for the
2240 method. This is a number followed by an equal sign and then the
2241 symbol descriptor `##', indicating a method type. This is followed by
2242 a type reference showing the return type of the method and a
2245 The format of an overloaded operator method name differs from that
2246 of other methods. It is "op$::XXXX." where XXXX is the operator name
2247 such as + or +=. The name ends with a period, and any characters except
2248 the period can occur in the XXXX string.
2250 The next part of the method description represents the arguments to
2251 the method, preceeded by a colon and ending with a semi-colon. The
2252 types of the arguments are expressed in the same way argument types
2253 are expressed in C++ name mangling. In this example an int and a char
2256 This is followed by a number, a letter, and an asterisk or period,
2257 followed by another semicolon. The number indicates the protections
2258 that apply to the member function. Here the 2 means public. The
2259 letter encodes any qualifier applied to the method definition. In
2260 this case A means that it is a normal function definition. The dot
2261 shows that the method is not virtual. The sections that follow
2262 elaborate further on these fields and describe the additional
2263 information present for virtual methods.
2267 .stabs "class_name:sym_desc(type)type_def(20)=type_desc(struct)struct_bytes(4)
2268 field_name(Adat):type(int),bit_offset(0),field_bits(32);
2270 method_name(Ameth)::type_def(21)=type_desc(method)return_type(int);
2271 :arg_types(int char);
2272 protection(public)qualifier(normal)virtual(no);;"
2277 .stabs "baseA:t20=s4Adat:1,0,32;Ameth::21=##1;:ic;2A.;;",128,0,0,0
2279 .stabs "class_name:sym_desc(struct tag)",N_LSYM,NIL,NIL,NIL
2281 .stabs "baseA:T20",128,0,0,0
2284 @node Class instance
2285 @section Class instance
2287 As shown above, describing even a simple C++ class definition is
2288 accomplished by massively extending the stab format used in C to
2289 describe structure types. However, once the class is defined, C stabs
2290 with no modifications can be used to describe class instances. The
2300 yields the following stab describing the class instance. It looks no
2301 different from a standard C stab describing a local variable.
2304 .stabs "name:type_ref(baseA)", N_LSYM, NIL, NIL, frame_ptr_offset
2308 .stabs "AbaseA:20",128,0,0,-20
2312 @section Method defintion
2314 The class definition shown above declares Ameth. The C++ source below
2319 baseA::Ameth(int in, char other)
2326 This method definition yields three stabs following the code of the
2327 method. One stab describes the method itself and following two describe
2328 its parameters. Although there is only one formal argument all methods
2329 have an implicit argument which is the `this' pointer. The `this'
2330 pointer is a pointer to the object on which the method was called. Note
2331 that the method name is mangled to encode the class name and argument
2332 types. Name mangling is described in the @sc{arm} (@cite{The Annotated
2333 C++ Reference Manual}, by Ellis and Stroustrup, @sc{isbn}
2334 0-201-51459-1); @file{gpcompare.texi} in Cygnus GCC distributions
2335 describes the differences between @sc{gnu} mangling and @sc{arm}
2337 @c FIXME: Use @xref, especially if this is generally installed in the
2339 @c FIXME: This information should be in a net release, either of GCC or
2340 @c GDB. But gpcompare.texi doesn't seem to be in the FSF GCC.
2343 .stabs "name:symbol_desriptor(global function)return_type(int)",
2344 N_FUN, NIL, NIL, code_addr_of_method_start
2346 .stabs "Ameth__5baseAic:F1",36,0,0,_Ameth__5baseAic
2349 Here is the stab for the `this' pointer implicit argument. The name
2350 of the `this' pointer is always `this.' Type 19, the `this' pointer is
2351 defined as a pointer to type 20, baseA, but a stab defining baseA has
2352 not yet been emited. Since the compiler knows it will be emited
2353 shortly, here it just outputs a cross reference to the undefined
2354 symbol, by prefixing the symbol name with xs.
2357 .stabs "name:sym_desc(register param)type_def(19)=
2358 type_desc(ptr to)type_ref(baseA)=
2359 type_desc(cross-reference to)baseA:",N_RSYM,NIL,NIL,register_number
2361 .stabs "this:P19=*20=xsbaseA:",64,0,0,8
2364 The stab for the explicit integer argument looks just like a parameter
2365 to a C function. The last field of the stab is the offset from the
2366 argument pointer, which in most systems is the same as the frame
2370 .stabs "name:sym_desc(value parameter)type_ref(int)",
2371 N_PSYM,NIL,NIL,offset_from_arg_ptr
2373 .stabs "in:p1",160,0,0,72
2376 << The examples that follow are based on A1.C >>
2379 @section Protections
2382 In the simple class definition shown above all member data and
2383 functions were publicly accessable. The example that follows
2384 contrasts public, protected and privately accessable fields and shows
2385 how these protections are encoded in C++ stabs.
2387 Protections for class member data are signified by two characters
2388 embeded in the stab defining the class type. These characters are
2389 located after the name: part of the string. /0 means private, /1
2390 means protected, and /2 means public. If these characters are omited
2391 this means that the member is public. The following C++ source:
2405 generates the following stab to describe the class type all_data.
2408 .stabs "class_name:sym_desc(type)type_def(19)=type_desc(struct)struct_bytes
2409 data_name:/protection(private)type_ref(int),bit_offset,num_bits;
2410 data_name:/protection(protected)type_ref(char),bit_offset,num_bits;
2411 data_name:(/num omited, private)type_ref(float),bit_offset,num_bits;;"
2416 .stabs "all_data:t19=s12
2417 priv_dat:/01,0,32;prot_dat:/12,32,8;pub_dat:12,64,32;;",128,0,0,0
2420 Protections for member functions are signified by one digit embeded in
2421 the field part of the stab describing the method. The digit is 0 if
2422 private, 1 if protected and 2 if public. Consider the C++ class
2426 class all_methods @{
2428 int priv_meth(int in)@{return in;@};
2430 char protMeth(char in)@{return in;@};
2432 float pubMeth(float in)@{return in;@};
2436 It generates the following stab. The digit in question is to the left
2437 of an `A' in each case. Notice also that in this case two symbol
2438 descriptors apply to the class name struct tag and struct type.
2441 .stabs "class_name:sym_desc(struct tag&type)type_def(21)=
2442 sym_desc(struct)struct_bytes(1)
2443 meth_name::type_def(22)=sym_desc(method)returning(int);
2444 :args(int);protection(private)modifier(normal)virtual(no);
2445 meth_name::type_def(23)=sym_desc(method)returning(char);
2446 :args(char);protection(protected)modifier(normal)virual(no);
2447 meth_name::type_def(24)=sym_desc(method)returning(float);
2448 :args(float);protection(public)modifier(normal)virtual(no);;",
2453 .stabs "all_methods:Tt21=s1priv_meth::22=##1;:i;0A.;protMeth::23=##2;:c;1A.;
2454 pubMeth::24=##12;:f;2A.;;",128,0,0,0
2457 @node Method Modifiers
2458 @section Method Modifiers (const, volatile, const volatile)
2462 In the class example described above all the methods have the normal
2463 modifier. This method modifier information is located just after the
2464 protection information for the method. This field has four possible
2465 character values. Normal methods use A, const methods use B, volatile
2466 methods use C, and const volatile methods use D. Consider the class
2472 int ConstMeth (int arg) const @{ return arg; @};
2473 char VolatileMeth (char arg) volatile @{ return arg; @};
2474 float ConstVolMeth (float arg) const volatile @{return arg; @};
2478 This class is described by the following stab:
2481 .stabs "class(A):sym_desc(struct)type_def(20)=type_desc(struct)struct_bytes(1)
2482 meth_name(ConstMeth)::type_def(21)sym_desc(method)
2483 returning(int);:arg(int);protection(public)modifier(const)virtual(no);
2484 meth_name(VolatileMeth)::type_def(22)=sym_desc(method)
2485 returning(char);:arg(char);protection(public)modifier(volatile)virt(no)
2486 meth_name(ConstVolMeth)::type_def(23)=sym_desc(method)
2487 returning(float);:arg(float);protection(public)modifer(const volatile)
2488 virtual(no);;", @dots{}
2492 .stabs "A:T20=s1ConstMeth::21=##1;:i;2B.;VolatileMeth::22=##2;:c;2C.;
2493 ConstVolMeth::23=##12;:f;2D.;;",128,0,0,0
2496 @node Virtual Methods
2497 @section Virtual Methods
2499 << The following examples are based on a4.C >>
2501 The presence of virtual methods in a class definition adds additional
2502 data to the class description. The extra data is appended to the
2503 description of the virtual method and to the end of the class
2504 description. Consider the class definition below:
2510 virtual int A_virt (int arg) @{ return arg; @};
2514 This results in the stab below describing class A. It defines a new
2515 type (20) which is an 8 byte structure. The first field of the class
2516 struct is Adat, an integer, starting at structure offset 0 and
2519 The second field in the class struct is not explicitly defined by the
2520 C++ class definition but is implied by the fact that the class
2521 contains a virtual method. This field is the vtable pointer. The
2522 name of the vtable pointer field starts with $vf and continues with a
2523 type reference to the class it is part of. In this example the type
2524 reference for class A is 20 so the name of its vtable pointer field is
2525 $vf20, followed by the usual colon.
2527 Next there is a type definition for the vtable pointer type (21).
2528 This is in turn defined as a pointer to another new type (22).
2530 Type 22 is the vtable itself, which is defined as an array, indexed by
2531 a range of integers between 0 and 1, and whose elements are of type
2532 17. Type 17 was the vtable record type defined by the boilerplate C++
2533 type definitions, as shown earlier.
2535 The bit offset of the vtable pointer field is 32. The number of bits
2536 in the field are not specified when the field is a vtable pointer.
2538 Next is the method definition for the virtual member function A_virt.
2539 Its description starts out using the same format as the non-virtual
2540 member functions described above, except instead of a dot after the
2541 `A' there is an asterisk, indicating that the function is virtual.
2542 Since is is virtual some addition information is appended to the end
2543 of the method description.
2545 The first number represents the vtable index of the method. This is a
2546 32 bit unsigned number with the high bit set, followed by a
2549 The second number is a type reference to the first base class in the
2550 inheritence hierarchy defining the virtual member function. In this
2551 case the class stab describes a base class so the virtual function is
2552 not overriding any other definition of the method. Therefore the
2553 reference is to the type number of the class that the stab is
2556 This is followed by three semi-colons. One marks the end of the
2557 current sub-section, one marks the end of the method field, and the
2558 third marks the end of the struct definition.
2560 For classes containing virtual functions the very last section of the
2561 string part of the stab holds a type reference to the first base
2562 class. This is preceeded by `~%' and followed by a final semi-colon.
2565 .stabs "class_name(A):type_def(20)=sym_desc(struct)struct_bytes(8)
2566 field_name(Adat):type_ref(int),bit_offset(0),field_bits(32);
2567 field_name(A virt func ptr):type_def(21)=type_desc(ptr to)type_def(22)=
2568 sym_desc(array)index_type_ref(range of int from 0 to 1);
2569 elem_type_ref(vtbl elem type),
2571 meth_name(A_virt)::typedef(23)=sym_desc(method)returning(int);
2572 :arg_type(int),protection(public)normal(yes)virtual(yes)
2573 vtable_index(1);class_first_defining(A);;;~%first_base(A);",
2577 @c FIXME: bogus line break.
2579 .stabs "A:t20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
2580 A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
2584 @section Inheritence
2586 Stabs describing C++ derived classes include additional sections that
2587 describe the inheritence hierarchy of the class. A derived class stab
2588 also encodes the number of base classes. For each base class it tells
2589 if the base class is virtual or not, and if the inheritence is private
2590 or public. It also gives the offset into the object of the portion of
2591 the object corresponding to each base class.
2593 This additional information is embeded in the class stab following the
2594 number of bytes in the struct. First the number of base classes
2595 appears bracketed by an exclamation point and a comma.
2597 Then for each base type there repeats a series: two digits, a number,
2598 a comma, another number, and a semi-colon.
2600 The first of the two digits is 1 if the base class is virtual and 0 if
2601 not. The second digit is 2 if the derivation is public and 0 if not.
2603 The number following the first two digits is the offset from the start
2604 of the object to the part of the object pertaining to the base class.
2606 After the comma, the second number is a type_descriptor for the base
2607 type. Finally a semi-colon ends the series, which repeats for each
2610 The source below defines three base classes A, B, and C and the
2618 virtual int A_virt (int arg) @{ return arg; @};
2624 virtual int B_virt (int arg) @{return arg; @};
2630 virtual int C_virt (int arg) @{return arg; @};
2633 class D : A, virtual B, public C @{
2636 virtual int A_virt (int arg ) @{ return arg+1; @};
2637 virtual int B_virt (int arg) @{ return arg+2; @};
2638 virtual int C_virt (int arg) @{ return arg+3; @};
2639 virtual int D_virt (int arg) @{ return arg; @};
2643 Class stabs similar to the ones described earlier are generated for
2646 @c FIXME!!! the linebreaks in the following example probably make the
2647 @c examples literally unusable, but I don't know any other way to get
2648 @c them on the page.
2649 @c One solution would be to put some of the type definitions into
2650 @c separate stabs, even if that's not exactly what the compiler actually
2653 .stabs "A:T20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
2654 A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
2656 .stabs "B:Tt25=s8Bdat:1,0,32;$vf25:21,32;B_virt::26=##1;
2657 :i;2A*-2147483647;25;;;~%25;",128,0,0,0
2659 .stabs "C:Tt28=s8Cdat:1,0,32;$vf28:21,32;C_virt::29=##1;
2660 :i;2A*-2147483647;28;;;~%28;",128,0,0,0
2663 In the stab describing derived class D below, the information about
2664 the derivation of this class is encoded as follows.
2667 .stabs "derived_class_name:symbol_descriptors(struct tag&type)=
2668 type_descriptor(struct)struct_bytes(32)!num_bases(3),
2669 base_virtual(no)inheritence_public(no)base_offset(0),
2670 base_class_type_ref(A);
2671 base_virtual(yes)inheritence_public(no)base_offset(NIL),
2672 base_class_type_ref(B);
2673 base_virtual(no)inheritence_public(yes)base_offset(64),
2674 base_class_type_ref(C); @dots{}
2677 @c FIXME! fake linebreaks.
2679 .stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:
2680 1,160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt:
2681 :32:i;2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;
2682 28;;D_virt::32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
2685 @node Virtual Base Classes
2686 @section Virtual Base Classes
2688 A derived class object consists of a concatination in memory of the
2689 data areas defined by each base class, starting with the leftmost and
2690 ending with the rightmost in the list of base classes. The exception
2691 to this rule is for virtual inheritence. In the example above, class
2692 D inherits virtually from base class B. This means that an instance
2693 of a D object will not contain it's own B part but merely a pointer to
2694 a B part, known as a virtual base pointer.
2696 In a derived class stab, the base offset part of the derivation
2697 information, described above, shows how the base class parts are
2698 ordered. The base offset for a virtual base class is always given as
2699 0. Notice that the base offset for B is given as 0 even though B is
2700 not the first base class. The first base class A starts at offset 0.
2702 The field information part of the stab for class D describes the field
2703 which is the pointer to the virtual base class B. The vbase pointer
2704 name is $vb followed by a type reference to the virtual base class.
2705 Since the type id for B in this example is 25, the vbase pointer name
2708 @c FIXME!! fake linebreaks below
2710 .stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:1,
2711 160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt::32:i;
2712 2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;28;;D_virt:
2713 :32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
2716 Following the name and a semicolon is a type reference describing the
2717 type of the virtual base class pointer, in this case 24. Type 24 was
2718 defined earlier as the type of the B class `this` pointer. The
2719 `this' pointer for a class is a pointer to the class type.
2722 .stabs "this:P24=*25=xsB:",64,0,0,8
2725 Finally the field offset part of the vbase pointer field description
2726 shows that the vbase pointer is the first field in the D object,
2727 before any data fields defined by the class. The layout of a D class
2728 object is a follows, Adat at 0, the vtable pointer for A at 32, Cdat
2729 at 64, the vtable pointer for C at 96, the virtual ase pointer for B
2730 at 128, and Ddat at 160.
2733 @node Static Members
2734 @section Static Members
2736 The data area for a class is a concatenation of the space used by the
2737 data members of the class. If the class has virtual methods, a vtable
2738 pointer follows the class data. The field offset part of each field
2739 description in the class stab shows this ordering.
2741 << How is this reflected in stabs? See Cygnus bug #677 for some info. >>
2744 @appendix Example2.c - source code for extended example
2748 2 register int g_bar asm ("%g5");
2749 3 static int s_g_repeat = 2;
2755 9 char s_char_vec[8];
2756 10 struct s_tag* s_next;
2759 13 typedef struct s_tag s_typedef;
2761 15 char char_vec[3] = @{'a','b','c'@};
2763 17 main (argc, argv)
2767 21 static float s_flap;
2769 23 for (times=0; times < s_g_repeat; times++)@{
2771 25 printf ("Hello world\n");
2775 29 enum e_places @{first,second=3,last@};
2777 31 static s_proc (s_arg, s_ptr_arg, char_vec)
2779 33 s_typedef* s_ptr_arg;
2793 @appendix Example2.s - assembly code for extended example
2797 2 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0
2798 3 .stabs "example2.c",100,0,0,Ltext0
2801 6 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0
2802 7 .stabs "char:t2=r2;0;127;",128,0,0,0
2803 8 .stabs "long int:t3=r1;-2147483648;2147483647;",128,0,0,0
2804 9 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
2805 10 .stabs "long unsigned int:t5=r1;0;-1;",128,0,0,0
2806 11 .stabs "short int:t6=r1;-32768;32767;",128,0,0,0
2807 12 .stabs "long long int:t7=r1;0;-1;",128,0,0,0
2808 13 .stabs "short unsigned int:t8=r1;0;65535;",128,0,0,0
2809 14 .stabs "long long unsigned int:t9=r1;0;-1;",128,0,0,0
2810 15 .stabs "signed char:t10=r1;-128;127;",128,0,0,0
2811 16 .stabs "unsigned char:t11=r1;0;255;",128,0,0,0
2812 17 .stabs "float:t12=r1;4;0;",128,0,0,0
2813 18 .stabs "double:t13=r1;8;0;",128,0,0,0
2814 19 .stabs "long double:t14=r1;8;0;",128,0,0,0
2815 20 .stabs "void:t15=15",128,0,0,0
2816 21 .stabs "g_foo:G2",32,0,0,0
2821 26 .stabs "s_g_repeat:S1",38,0,0,_s_g_repeat
2825 @c FIXME! fake linebreak in line 30
2826 30 .stabs "s_tag:T16=s20s_int:1,0,32;s_float:12,32,32;s_char_vec:
2827 17=ar1;0;7;2,64,64;s_next:18=*16,128,32;;",128,0,0,0
2828 31 .stabs "s_typedef:t16",128,0,0,0
2829 32 .stabs "char_vec:G19=ar1;0;2;2",32,0,0,0
2830 33 .global _char_vec
2836 39 .reserve _s_flap.0,4,"bss",4
2840 43 .ascii "Hello world\12\0"
2845 48 .stabn 68,0,20,LM1
2848 51 save %sp,-144,%sp
2855 58 .stabn 68,0,23,LM2
2859 62 sethi %hi(_s_g_repeat),%o0
2861 64 ld [%o0+%lo(_s_g_repeat)],%o0
2866 69 .stabn 68,0,25,LM3
2868 71 sethi %hi(LC0),%o1
2869 72 or %o1,%lo(LC0),%o0
2872 75 .stabn 68,0,26,LM4
2875 78 .stabn 68,0,23,LM5
2883 86 .stabn 68,0,27,LM6
2886 89 .stabn 68,0,27,LM7
2891 94 .stabs "main:F1",36,0,0,_main
2892 95 .stabs "argc:p1",160,0,0,68
2893 96 .stabs "argv:p20=*21=*2",160,0,0,72
2894 97 .stabs "s_flap:V12",40,0,0,_s_flap.0
2895 98 .stabs "times:1",128,0,0,-20
2896 99 .stabn 192,0,0,LBB2
2897 100 .stabs "inner:1",128,0,0,-24
2898 101 .stabn 192,0,0,LBB3
2899 102 .stabn 224,0,0,LBE3
2900 103 .stabn 224,0,0,LBE2
2901 104 .stabs "e_places:T22=efirst:0,second:3,last:4,;",128,0,0,0
2902 @c FIXME: fake linebreak in line 105
2903 105 .stabs "u_tag:T23=u4u_int:1,0,32;u_float:12,0,32;u_char:21,0,32;;",
2908 109 .stabn 68,0,35,LM8
2911 112 save %sp,-120,%sp
2917 118 .stabn 68,0,41,LM9
2920 121 .stabn 68,0,41,LM10
2925 126 .stabs "s_proc:f1",36,0,0,_s_proc
2926 127 .stabs "s_arg:p16",160,0,0,0
2927 128 .stabs "s_ptr_arg:p18",160,0,0,72
2928 129 .stabs "char_vec:p21",160,0,0,76
2929 130 .stabs "an_u:23",128,0,0,-20
2930 131 .stabn 192,0,0,LBB4
2931 132 .stabn 224,0,0,LBE4
2932 133 .stabs "g_bar:r1",64,0,0,5
2933 134 .stabs "g_pf:G24=*25=f1",32,0,0,0
2934 135 .common _g_pf,4,"bss"
2935 136 .stabs "g_an_s:G16",32,0,0,0
2936 137 .common _g_an_s,20,"bss"
2940 @appendix Values for the Stab Type Field
2942 These are all the possible values for the stab type field, for
2943 @code{a.out} files. This does not apply to XCOFF.
2945 The following types are used by the linker and assembler; there is
2946 nothing stabs-specific about them. Since this document does not attempt
2947 to describe aspects of object file format other than the debugging
2948 format, no details are given.
2950 @c Try to get most of these to fit on a single line.
2960 File scope absolute symbol
2962 @item 0x3 N_ABS | N_EXT
2963 External absolute symbol
2966 File scope text symbol
2968 @item 0x5 N_TEXT | N_EXT
2969 External text symbol
2972 File scope data symbol
2974 @item 0x7 N_DATA | N_EXT
2975 External data symbol
2978 File scope BSS symbol
2980 @item 0x9 N_BSS | N_EXT
2984 Same as N_FN, for Sequent compilers
2987 Symbol is indirected to another symbol
2990 Common sym -- visable after shared lib dynamic link
2993 Absolute set element
2996 Text segment set element
2999 Data segment set element
3002 BSS segment set element
3005 Pointer to set vector
3007 @item 0x1e N_WARNING
3008 Print a warning message during linking
3011 File name of a .o file
3014 The following symbol types indicate that this is a stab. This is the
3015 full list of stab numbers, including stab types that are used in
3016 languages other than C.
3020 Global symbol, @xref{N_GSYM}.
3023 Function name (for BSD Fortran), @xref{N_FNAME}.
3026 Function name (@pxref{Procedures}) or text segment variable
3030 Data segment file-scope variable, @xref{Statics}.
3033 BSS segment file-scope variable, @xref{Statics}.
3036 Name of main routine, @xref{Main Program}.
3038 @c FIXME: discuss this in the main body of the text where we talk about
3039 @c using N_FUN for variables.
3041 Read-only data symbol (Solaris2). Most systems use N_FUN for this.
3044 Global symbol (for Pascal), @xref{N_PC}.
3047 Number of symbols (according to Ultrix V4.0), @xref{N_NSYMS}.
3050 No DST map for sym (according to Ultrix V4.0), @xref{N_NOMAP}.
3052 @c FIXME: describe this solaris feature in the body of the text (see
3053 @c comments in include/aout/stab.def).
3055 Object file (Solaris2).
3057 @c See include/aout/stab.def for (a little) more info.
3059 Debugger options (Solaris2).
3062 Register variable, @xref{N_RSYM}.
3065 Modula-2 compilation unit, @xref{N_M2C}.
3068 Line number in text segment, @xref{Line Numbers}.
3071 Line number in data segment, @xref{Line Numbers}.
3074 Line number in bss segment, @xref{Line Numbers}.
3077 Sun source code browser, path to .cb file, @xref{N_BROWS}.
3080 Gnu Modula2 definition module dependency, @xref{N_DEFD}.
3083 Function start/body/end line numbers (Solaris2).
3086 Gnu C++ exception variable, @xref{N_EHDECL}.
3089 Modula2 info "for imc" (according to Ultrix V4.0), @xref{N_MOD2}.
3092 Gnu C++ "catch" clause, @xref{N_CATCH}.
3095 Structure of union element, @xref{N_SSYM}.
3098 Last stab for module (Solaris2).
3101 Path and name of source file , @xref{Source Files}.
3104 Automatic var in the stack or type definition, @xref{N_LSYM}, @xref{Typedefs}.
3107 Beginning of an include file (Sun only), @xref{Source Files}.
3110 Name of include file, @xref{Source Files}.
3113 Parameter variable, @xref{Parameters}.
3116 End of an include file, @xref{Source Files}.
3119 Alternate entry point, @xref{N_ENTRY}.
3122 Beginning of a lexical block, @xref{Block Structure}.
3125 Place holder for a deleted include file, @xref{Source Files}.
3128 Modula2 scope information (Sun linker), @xref{N_SCOPE}.
3131 End of a lexical block, @xref{Block Structure}.
3134 Begin named common block, @xref{Common Blocks}.
3137 End named common block, @xref{Common Blocks}.
3140 Member of a common block, @xref{Common Blocks}.
3142 @c FIXME: How does this really work? Move it to main body of document.
3144 Pascal @code{with} statement: type,,0,0,offset (Solaris2).
3147 Gould non-base registers, @xref{Gould}.
3150 Gould non-base registers, @xref{Gould}.
3153 Gould non-base registers, @xref{Gould}.
3156 Gould non-base registers, @xref{Gould}.
3159 Gould non-base registers, @xref{Gould}.
3162 @c Restore the default table indent
3167 @node Symbol Descriptors
3168 @appendix Table of Symbol Descriptors
3170 @c Please keep this alphabetical
3172 @c In TeX, this looks great, digit is in italics. But makeinfo insists
3173 @c on putting it in `', not realizing that @var should override @code.
3174 @c I don't know of any way to make makeinfo do the right thing. Seems
3175 @c like a makeinfo bug to me.
3179 Local variable, @xref{Automatic variables}.
3182 Parameter passed by reference in register, @xref{Parameters}.
3185 Constant, @xref{Constants}.
3188 Conformant array bound (Pascal, maybe other languages),
3189 @xref{Parameters}. Name of a caught exception (GNU C++). These can be
3190 distinguished because the latter uses N_CATCH and the former uses
3191 another symbol type.
3194 Floating point register variable, @xref{Register variables}.
3197 Parameter in floating point register, @xref{Parameters}.
3200 File scope function, @xref{Procedures}.
3203 Global function, @xref{Procedures}.
3206 Global variable, @xref{Global Variables}.
3212 Internal (nested) procedure, @xref{Procedures}.
3215 Internal (nested) function, @xref{Procedures}.
3218 Label name (documented by AIX, no further information known).
3221 Module, @xref{Procedures}.
3224 Argument list parameter, @xref{Parameters}.
3230 FORTRAN Function parameter, @xref{Parameters}.
3233 Unfortunately, three separate meanings have been independently invented
3234 for this symbol descriptor. At least the GNU and Sun uses can be
3235 distinguished by the symbol type. Global Procedure (AIX) (symbol type
3236 used unknown), @xref{Procedures}. Register parameter (GNU) (symbol type
3237 N_PSYM), @xref{Parameters}. Prototype of function referenced by this
3238 file (Sun acc) (symbol type N_FUN).
3241 Static Procedure, @xref{Procedures}.
3244 Register parameter @xref{Parameters}.
3247 Register variable, @xref{Register variables}.
3250 File scope variable, @xref{Statics}.
3253 Type name, @xref{Typedefs}.
3256 enumeration, struct or union tag, @xref{Typedefs}.
3259 Parameter passed by reference, @xref{Parameters}.
3262 Procedure scope static variable, @xref{Statics}.
3265 Conformant array, @xref{Parameters}.
3268 Function return variable, @xref{Parameters}.
3271 @node Type Descriptors
3272 @appendix Table of Type Descriptors
3277 Type reference, @xref{Stabs Format}.
3280 Reference to builtin type, @xref{Negative Type Numbers}.
3283 Method (C++), @xref{Cplusplus}.
3286 Pointer, @xref{Miscellaneous Types}.
3292 Type Attributes (AIX), @xref{Stabs Format}. Member (class and variable)
3293 type (GNU C++), @xref{Cplusplus}.
3296 Array, @xref{Arrays}.
3299 Open array, @xref{Arrays}.
3302 Pascal space type (AIX), @xref{Miscellaneous Types}. Builtin integer
3303 type (Sun), @xref{Builtin Type Descriptors}.
3306 Volatile-qualified type, @xref{Miscellaneous Types}.
3309 Complex builtin type, @xref{Builtin Type Descriptors}.
3312 COBOL Picture type. See AIX documentation for details.
3315 File type, @xref{Miscellaneous Types}.
3318 N-dimensional dynamic array, @xref{Arrays}.
3321 Enumeration type, @xref{Enumerations}.
3324 N-dimensional subarray, @xref{Arrays}.
3327 Function type, @xref{Function Types}.
3330 Pascal function parameter, @xref{Function Types}
3333 Builtin floating point type, @xref{Builtin Type Descriptors}.
3336 COBOL Group. See AIX documentation for details.
3339 Imported type, @xref{Cross-references}.
3342 Const-qualified type, @xref{Miscellaneous Types}.
3345 COBOL File Descriptor. See AIX documentation for details.
3348 Multiple instance type, @xref{Miscellaneous Types}.
3351 String type, @xref{Strings}.
3354 Stringptr, @xref{Strings}.
3357 Opaque type, @xref{Typedefs}.
3360 Procedure, @xref{Function Types}.
3363 Packed array, @xref{Arrays}.
3366 Range type, @xref{Subranges}.
3369 Builtin floating type, @xref{Builtin Type Descriptors} (Sun). Pascal
3370 subroutine parameter, @xref{Function Types} (AIX). Detecting this
3371 conflict is possible with careful parsing (hint: a Pascal subroutine
3372 parameter type will always contain a comma, and a builtin type
3373 descriptor never will).
3376 Structure type, @xref{Structures}.
3379 Set type, @xref{Miscellaneous Types}.
3382 Union, @xref{Unions}.
3385 Variant record. This is a Pascal and Modula-2 feature which is like a
3386 union within a struct in C. See AIX documentation for details.
3389 Wide character, @xref{Builtin Type Descriptors}.
3392 Cross-reference, @xref{Cross-references}.
3395 gstring, @xref{Strings}.
3398 @node Expanded reference
3399 @appendix Expanded reference by stab type.
3401 @c FIXME: This appendix should go away, see N_PSYM or N_SO for an example.
3403 For a full list of stab types, and cross-references to where they are
3404 described, @xref{Stab Types}. This appendix just duplicates certain
3405 information from the main body of this document; eventually the
3406 information will all be in one place.
3410 The first line is the symbol type expressed in decimal, hexadecimal,
3411 and as a #define (see devo/include/aout/stab.def).
3413 The second line describes the language constructs the symbol type
3416 The third line is the stab format with the significant stab fields
3417 named and the rest NIL.
3419 Subsequent lines expand upon the meaning and possible values for each
3420 significant stab field. # stands in for the type descriptor.
3422 Finally, any further information.
3425 * N_GSYM:: Global variable
3426 * N_FNAME:: Function name (BSD Fortran)
3427 * N_PC:: Pascal global symbol
3428 * N_NSYMS:: Number of symbols
3429 * N_NOMAP:: No DST map
3430 * N_RSYM:: Register variable
3431 * N_M2C:: Modula-2 compilation unit
3432 * N_BROWS:: Path to .cb file for Sun source code browser
3433 * N_DEFD:: GNU Modula2 definition module dependency
3434 * N_EHDECL:: GNU C++ exception variable
3435 * N_MOD2:: Modula2 information "for imc"
3436 * N_CATCH:: GNU C++ "catch" clause
3437 * N_SSYM:: Structure or union element
3438 * N_LSYM:: Automatic variable
3439 * N_ENTRY:: Alternate entry point
3440 * N_SCOPE:: Modula2 scope information (Sun only)
3441 * Gould:: non-base register symbols used on Gould systems
3442 * N_LENG:: Length of preceding entry
3446 @section 32 - 0x20 - N_GYSM
3451 .stabs "name", N_GSYM, NIL, NIL, NIL
3455 "name" -> "symbol_name:#type"
3459 Only the "name" field is significant. The location of the variable is
3460 obtained from the corresponding external symbol.
3463 @section 34 - 0x22 - N_FNAME
3464 Function name (for BSD Fortran)
3467 .stabs "name", N_FNAME, NIL, NIL, NIL
3471 "name" -> "function_name"
3474 Only the "name" field is significant. The location of the symbol is
3475 obtained from the corresponding extern symbol.
3478 @section 48 - 0x30 - N_PC
3479 Global symbol (for Pascal)
3482 .stabs "name", N_PC, NIL, NIL, value
3486 "name" -> "symbol_name" <<?>>
3487 value -> supposedly the line number (stab.def is skeptical)
3493 global pascal symbol: name,,0,subtype,line
3498 @section 50 - 0x32 - N_NSYMS
3499 Number of symbols (according to Ultrix V4.0)
3502 0, files,,funcs,lines (stab.def)
3506 @section 52 - 0x34 - N_NOMAP
3507 no DST map for sym (according to Ultrix V4.0)
3510 name, ,0,type,ignored (stab.def)
3514 @section 64 - 0x40 - N_RSYM
3518 .stabs "name:type",N_RSYM,0,RegSize,RegNumber (Sun doc)
3522 @section 66 - 0x42 - N_M2C
3523 Modula-2 compilation unit
3526 .stabs "name", N_M2C, 0, desc, value
3530 "name" -> "unit_name,unit_time_stamp[,code_time_stamp]
3532 value -> 0 (main unit)
3537 @section 72 - 0x48 - N_BROWS
3538 Sun source code browser, path to .cb file
3541 "path to associated .cb file"
3543 Note: type field value overlaps with N_BSLINE
3546 @section 74 - 0x4a - N_DEFD
3547 GNU Modula2 definition module dependency
3549 GNU Modula-2 definition module dependency. Value is the modification
3550 time of the definition file. Other is non-zero if it is imported with
3551 the GNU M2 keyword %INITIALIZE. Perhaps N_M2C can be used if there
3552 are enough empty fields?
3555 @section 80 - 0x50 - N_EHDECL
3556 GNU C++ exception variable <<?>>
3558 "name is variable name"
3560 Note: conflicts with N_MOD2.
3563 @section 80 - 0x50 - N_MOD2
3564 Modula2 info "for imc" (according to Ultrix V4.0)
3566 Note: conflicts with N_EHDECL <<?>>
3569 @section 84 - 0x54 - N_CATCH
3570 GNU C++ "catch" clause
3572 GNU C++ `catch' clause. Value is its address. Desc is nonzero if
3573 this entry is immediately followed by a CAUGHT stab saying what
3574 exception was caught. Multiple CAUGHT stabs means that multiple
3575 exceptions can be caught here. If Desc is 0, it means all exceptions
3579 @section 96 - 0x60 - N_SSYM
3580 Structure or union element
3582 Value is offset in the structure.
3584 <<?looking at structs and unions in C I didn't see these>>
3587 @section 128 - 0x80 - N_LSYM
3588 Automatic var in the stack (also used for type descriptors.)
3591 .stabs "name" N_LSYM, NIL, NIL, value
3595 @exdent @emph{For stack based local variables:}
3597 "name" -> name of the variable
3598 value -> offset from frame pointer (negative)
3600 @exdent @emph{For type descriptors:}
3602 "name" -> "name_of_the_type:#type"
3605 type -> type_ref (or) type_def
3607 type_ref -> type_number
3608 type_def -> type_number=type_desc etc.
3611 Type may be either a type reference or a type definition. A type
3612 reference is a number that refers to a previously defined type. A
3613 type definition is the number that will refer to this type, followed
3614 by an equals sign, a type descriptor and the additional data that
3615 defines the type. See the Table D for type descriptors and the
3616 section on types for what data follows each type descriptor.
3619 @section 164 - 0xa4 - N_ENTRY
3621 Alternate entry point.
3622 Value is its address.
3626 @section 196 - 0xc4 - N_SCOPE
3628 Modula2 scope information (Sun linker)
3632 @section Non-base registers on Gould systems
3634 These are used on Gould systems for non-base registers syms.
3636 However, the following values are not the values used by Gould; they are
3637 the values which GNU has been documenting for these values for a long
3638 time, without actually checking what Gould uses. I include these values
3639 only because perhaps some someone actually did something with the GNU
3640 information (I hope not, why GNU knowingly assigned wrong values to
3641 these in the header file is a complete mystery to me).
3644 240 0xf0 N_NBTEXT ??
3645 242 0xf2 N_NBDATA ??
3652 @section - 0xfe - N_LENG
3654 Second symbol entry containing a length-value for the preceding entry.
3655 The value is the length.
3658 @appendix Questions and anomalies
3662 For GNU C stabs defining local and global variables (N_LSYM and
3663 N_GSYM), the desc field is supposed to contain the source line number
3664 on which the variable is defined. In reality the desc field is always
3665 0. (This behavour is defined in dbxout.c and putting a line number in
3666 desc is controlled by #ifdef WINNING_GDB which defaults to false). Gdb
3667 supposedly uses this information if you say 'list var'. In reality
3668 var can be a variable defined in the program and gdb says `function
3672 In GNU C stabs there seems to be no way to differentiate tag types:
3673 structures, unions, and enums (symbol descriptor T) and typedefs
3674 (symbol descriptor t) defined at file scope from types defined locally
3675 to a procedure or other more local scope. They all use the N_LSYM
3676 stab type. Types defined at procedure scope are emited after the
3677 N_RBRAC of the preceding function and before the code of the
3678 procedure in which they are defined. This is exactly the same as
3679 types defined in the source file between the two procedure bodies.
3680 GDB overcompensates by placing all types in block #1, the block for
3681 symbols of file scope. This is true for default, -ansi and
3682 -traditional compiler options. (Bugs gcc/1063, gdb/1066.)
3685 What ends the procedure scope? Is it the proc block's N_RBRAC or the
3686 next N_FUN? (I believe its the first.)
3689 @c FIXME: This should go with the other stuff about global variables.
3690 Global variable stabs don't have location information. This comes
3691 from the external symbol for the same variable. The external symbol
3692 has a leading underbar on the _name of the variable and the stab does
3693 not. How do we know these two symbol table entries are talking about
3694 the same symbol when their names are different? (Answer: the debugger
3695 knows that external symbols have leading underbars).
3697 @c FIXME: This is absurdly vague; there all kinds of differences, some
3698 @c of which are the same between gnu & sun, and some of which aren't.
3700 Can gcc be configured to output stabs the way the Sun compiler
3701 does, so that their native debugging tools work? <NO?> It doesn't by
3702 default. GDB reads either format of stab. (gcc or SunC). How about
3706 @node xcoff-differences
3707 @appendix Differences between GNU stabs in a.out and GNU stabs in xcoff
3709 @c FIXME: Merge *all* these into the main body of the document.
3710 (The AIX/RS6000 native object file format is xcoff with stabs). This
3711 appendix only covers those differences which are not covered in the main
3712 body of this document.
3716 BSD a.out stab types correspond to AIX xcoff storage classes. In general the
3717 mapping is N_STABTYPE becomes C_STABTYPE. Some stab types in a.out
3718 are not supported in xcoff. See Table E. for full mappings.
3720 @c FIXME: Get C_* types for the block, figure out whether it is always
3721 @c used (I suspect not), explain clearly, and move to node Statics.
3723 initialised static N_STSYM and un-initialized static N_LCSYM both map
3724 to the C_STSYM storage class. But the destinction is preserved
3725 because in xcoff N_STSYM and N_LCSYM must be emited in a named static
3726 block. Begin the block with .bs s[RW] data_section_name for N_STSYM
3727 or .bs s bss_section_name for N_LCSYM. End the block with .es
3729 @c FIXME: I think they are trying to say something about whether the
3730 @c assembler defaults the value to the location counter.
3732 If the xcoff stab is a N_FUN (C_FUN) then follow the string field with
3733 ,. instead of just ,
3736 (I think that's it for .s file differences. They could stand to be
3737 better presented. This is just a list of what I have noticed so far.
3738 There are a *lot* of differences in the information in the symbol
3739 tables of the executable and object files.)
3741 Table E: mapping a.out stab types to xcoff storage classes
3744 stab type storage class
3745 -------------------------------
3754 N_RPSYM (0x8e) C_RPSYM
3764 N_DECL (0x8c) C_DECL
3781 @node Sun-differences
3782 @appendix Differences between GNU stabs and Sun native stabs.
3784 @c FIXME: Merge all this stuff into the main body of the document.
3788 GNU C stabs define *all* types, file or procedure scope, as
3789 N_LSYM. Sun doc talks about using N_GSYM too.
3792 Sun C stabs use type number pairs in the format (a,b) where a is a
3793 number starting with 1 and incremented for each sub-source file in the
3794 compilation. b is a number starting with 1 and incremented for each
3795 new type defined in the compilation. GNU C stabs use the type number
3796 alone, with no source file number.
3800 @appendix Using stabs with the ELF object file format.
3802 The ELF object file format allows tools to create object files with custom
3803 sections containing any arbitrary data. To use stabs in ELF object files,
3804 the tools create two custom sections, a ".stab" section which contains
3805 an array of fixed length structures, one struct per stab, and a ".stabstr"
3806 section containing all the variable length strings that are referenced by
3807 stabs in the ".stab" section. The byte order of the stabs binary data
3808 matches the byte order of the ELF file itself, as determined from the
3809 EI_DATA field in the e_ident member of the ELF header.
3811 The first stab in the ".stab" section for each object file is a "synthetic
3812 stab", generated entirely by the assembler, with no corresponding ".stab"
3813 directive as input to the assembler. This stab contains the following
3818 Offset in the ".stabstr" section to the source filename.
3824 Unused field, always zero.
3827 Count of upcoming symbols. I.E. the number of remaining stabs for this
3831 Size of the string table fragment associated with this object module, in
3836 The ".stabstr" section always starts with a null byte (so that string
3837 offsets of zero reference a null string), followed by random length strings,
3838 each of which is null byte terminated.
3840 The ELF section header for the ".stab" section has it's sh_link member set
3841 to the section number of the ".stabstr" section, and the ".stabstr" section
3842 has it's ELF section header sh_type member set to SHT_STRTAB to mark it as