2 @setfilename stabs.info
7 * Stabs:: The "stabs" debugging information format.
13 This document describes the stabs debugging symbol tables.
15 Copyright 1992 Free Software Foundation, Inc.
16 Contributed by Cygnus Support. Written by Julia Menapace.
18 Permission is granted to make and distribute verbatim copies of
19 this manual provided the copyright notice and this permission notice
20 are preserved on all copies.
23 Permission is granted to process this file through Tex and print the
24 results, provided the printed document carries copying permission
25 notice identical to this one except for the removal of this paragraph
26 (this paragraph not being relevant to the printed manual).
29 Permission is granted to copy or distribute modified versions of this
30 manual under the terms of the GPL (for which purpose this text may be
31 regarded as a program in the language TeX).
34 @setchapternewpage odd
37 @title The ``stabs'' debug format
38 @author Julia Menapace
39 @author Cygnus Support
42 \def\$#1${{#1}} % Kluge: collect RCS revision info without $...$
43 \xdef\manvers{\$Revision$} % For use in headers, footers too
45 \hfill Cygnus Support\par
47 \hfill \TeX{}info \texinfoversion\par
51 @vskip 0pt plus 1filll
52 Copyright @copyright{} 1992 Free Software Foundation, Inc.
53 Contributed by Cygnus Support.
55 Permission is granted to make and distribute verbatim copies of
56 this manual provided the copyright notice and this permission notice
57 are preserved on all copies.
63 @top The "stabs" representation of debugging information
65 This document describes the stabs debugging format.
68 * Overview:: Overview of stabs
69 * Program structure:: Encoding of the structure of the program
70 * Constants:: Constants
71 * Example:: A comprehensive example in C
73 * Types:: Type definitions
74 * Symbol Tables:: Symbol information in symbol tables
75 * Cplusplus:: Appendixes:
76 * Example2.c:: Source code for extended example
77 * Example2.s:: Assembly code for extended example
78 * Stab Types:: Symbol types in a.out files
79 * Symbol Descriptors:: Table of Symbol Descriptors
80 * Type Descriptors:: Table of Symbol Descriptors
81 * Expanded reference:: Reference information by stab type
82 * Questions:: Questions and anomolies
83 * xcoff-differences:: Differences between GNU stabs in a.out
84 and GNU stabs in xcoff
85 * Sun-differences:: Differences between GNU stabs and Sun
87 * Stabs-in-elf:: Stabs in an ELF file.
93 @chapter Overview of stabs
95 @dfn{Stabs} refers to a format for information that describes a program
96 to a debugger. This format was apparently invented by
97 @c FIXME! <<name of inventor>> at
98 the University of California at Berkeley, for the @code{pdx} Pascal
99 debugger; the format has spread widely since then.
101 This document is one of the few published sources of documentation on
102 stabs. It is believed to be completely comprehensive for stabs used by
103 C. The lists of symbol descriptors (@pxref{Symbol Descriptors}) and
104 type descriptors (@pxref{Type Descriptors}) are believed to be completely
105 comprehensive. There are known to be stabs for C++ and COBOL which are
106 poorly documented here. Stabs specific to other languages (e.g. Pascal,
107 Modula-2) are probably not as well documented as they should be.
109 Other sources of information on stabs are @cite{dbx and dbxtool
110 interfaces}, 2nd edition, by Sun, circa 1988, and @cite{AIX Version 3.2
111 Files Reference}, Fourth Edition, September 1992, "dbx Stabstring
112 Grammar" in the a.out section, page 2-31. This document is believed to
113 incorporate the information from those two sources except where it
114 explictly directs you to them for more information.
117 * Flow:: Overview of debugging information flow
118 * Stabs Format:: Overview of stab format
119 * C example:: A simple example in C source
120 * Assembly code:: The simple example at the assembly level
124 @section Overview of debugging information flow
126 The GNU C compiler compiles C source in a @file{.c} file into assembly
127 language in a @file{.s} file, which is translated by the assembler into
128 a @file{.o} file, and then linked with other @file{.o} files and
129 libraries to produce an executable file.
131 With the @samp{-g} option, GCC puts additional debugging information in
132 the @file{.s} file, which is slightly transformed by the assembler and
133 linker, and carried through into the final executable. This debugging
134 information describes features of the source file like line numbers,
135 the types and scopes of variables, and functions, their parameters and
138 For some object file formats, the debugging information is
139 encapsulated in assembler directives known collectively as `stab' (symbol
140 table) directives, interspersed with the generated code. Stabs are
141 the native format for debugging information in the a.out and xcoff
142 object file formats. The GNU tools can also emit stabs in the coff
143 and ecoff object file formats.
145 The assembler adds the information from stabs to the symbol information
146 it places by default in the symbol table and the string table of the
147 @file{.o} file it is building. The linker consolidates the @file{.o}
148 files into one executable file, with one symbol table and one string
149 table. Debuggers use the symbol and string tables in the executable as
150 a source of debugging information about the program.
153 @section Overview of stab format
155 There are three overall formats for stab assembler directives
156 differentiated by the first word of the stab. The name of the directive
157 describes what combination of four possible data fields will follow. It
158 is either @code{.stabs} (string), @code{.stabn} (number), or
159 @code{.stabd} (dot). IBM's xcoff uses @code{.stabx} (and some other
160 directives such as @code{.file} and @code{.bi}) instead of
161 @code{.stabs}, @code{.stabn} or @code{.stabd}.
163 The overall format of each class of stab is:
166 .stabs "@var{string}",@var{type},0,@var{desc},@var{value}
167 .stabx "@var{string}",@var{value},@var{type},@var{sdb-type}
168 .stabn @var{type},0,@var{desc},@var{value}
169 .stabd @var{type},0,@var{desc}
172 @c what is the correct term for "current file location"? My AIX
173 @c assembler manual calls it "the value of the current location counter".
174 For @code{.stabn} and @code{.stabd}, there is no string (the
175 @code{n_strx} field is zero, @pxref{Symbol Tables}). For @code{.stabd}
176 the value field is implicit and has the value of the current file
177 location. The @var{sdb-type} field to @code{.stabx} is unused for stabs
178 and can always be set to 0.
180 The number in the type field gives some basic information about what
181 type of stab this is (or whether it @emph{is} a stab, as opposed to an
182 ordinary symbol). Each possible type number defines a different stab
183 type. The stab type further defines the exact interpretation of, and
184 possible values for, any remaining @code{"@var{string}"}, @var{desc}, or
185 @var{value} fields present in the stab. @xref{Stab Types}, for a list
186 in numeric order of the possible type field values for stab directives.
188 For @code{.stabs} the @code{"@var{string}"} field holds the meat of the
189 debugging information. The generally unstructured nature of this field
190 is what makes stabs extensible. For some stab types the string field
191 contains only a name. For other stab types the contents can be a great
194 The overall format is of the @code{"@var{string}"} field is:
197 "@var{name}:@var{symbol-descriptor} @var{type-information}"
200 @var{name} is the name of the symbol represented by the stab.
201 @var{name} can be omitted, which means the stab represents an unnamed
202 object. For example, @samp{:t10=*2} defines type 10 as a pointer to
203 type 2, but does not give the type a name. Omitting the @var{name}
204 field is supported by AIX dbx and GDB after about version 4.8, but not
205 other debuggers. GCC sometimes uses a single space as the name instead
206 of omitting the name altogether; apparently that is supported by most
209 The @var{symbol_descriptor} following the @samp{:} is an alphabetic
210 character that tells more specifically what kind of symbol the stab
211 represents. If the @var{symbol_descriptor} is omitted, but type
212 information follows, then the stab represents a local variable. For a
213 list of symbol descriptors, see @ref{Symbol Descriptors,,Table C: Symbol
216 The @samp{c} symbol descriptor is an exception in that it is not
217 followed by type information. @xref{Constants}.
219 Type information is either a @var{type_number}, or a
220 @samp{@var{type_number}=}. The @var{type_number} alone is a type
221 reference, referring directly to a type that has already been defined.
223 The @samp{@var{type_number}=} is a type definition, where the number
224 represents a new type which is about to be defined. The type definition
225 may refer to other types by number, and those type numbers may be
226 followed by @samp{=} and nested definitions.
228 In a type definition, if the character that follows the equals sign is
229 non-numeric then it is a @var{type_descriptor}, and tells what kind of
230 type is about to be defined. Any other values following the
231 @var{type_descriptor} vary, depending on the @var{type_descriptor}. If
232 a number follows the @samp{=} then the number is a @var{type_reference}.
233 This is described more thoroughly in the section on types. @xref{Type
234 Descriptors,,Table D: Type Descriptors}, for a list of
235 @var{type_descriptor} values.
237 There is an AIX extension for type attributes. Following the @samp{=}
238 is any number of type attributes. Each one starts with @samp{@@} and
239 ends with @samp{;}. Debuggers, including AIX's dbx, skip any type
240 attributes they do not recognize. GDB 4.9 does not do this---it will
241 ignore the entire symbol containing a type attribute. Hopefully this
242 will be fixed in the next GDB release. Because of a conflict with C++
243 (@pxref{Cplusplus}), new attributes should not be defined which begin
244 with a digit, @samp{(}, or @samp{-}; GDB may be unable to distinguish
245 those from the C++ type descriptor @samp{@@}. The attributes are:
248 @item a@var{boundary}
249 @var{boundary} is an integer specifying the alignment. I assume it
250 applies to all variables of this type.
253 Size in bits of a variable of this type.
256 Pointer class (for checking). Not sure what this means, or how
257 @var{integer} is interpreted.
260 Indicate this is a packed type, meaning that structure fields or array
261 elements are placed more closely in memory, to save memory at the
265 All this can make the @code{"@var{string}"} field quite long. All
266 versions of GDB, and some versions of DBX, can handle arbitrarily long
267 strings. But many versions of DBX cretinously limit the strings to
268 about 80 characters, so compilers which must work with such DBX's need
269 to split the @code{.stabs} directive into several @code{.stabs}
270 directives. Each stab duplicates exactly all but the
271 @code{"@var{string}"} field. The @code{"@var{string}"} field of
272 every stab except the last is marked as continued with a
273 double-backslash at the end. Removing the backslashes and concatenating
274 the @code{"@var{string}"} fields of each stab produces the original,
278 @section A simple example in C source
280 To get the flavor of how stabs describe source information for a C
281 program, let's look at the simple program:
286 printf("Hello world");
290 When compiled with @samp{-g}, the program above yields the following
291 @file{.s} file. Line numbers have been added to make it easier to refer
292 to parts of the @file{.s} file in the description of the stabs that
296 @section The simple example at the assembly level
300 2 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0
301 3 .stabs "hello.c",100,0,0,Ltext0
304 6 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0
305 7 .stabs "char:t2=r2;0;127;",128,0,0,0
306 8 .stabs "long int:t3=r1;-2147483648;2147483647;",128,0,0,0
307 9 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
308 10 .stabs "long unsigned int:t5=r1;0;-1;",128,0,0,0
309 11 .stabs "short int:t6=r1;-32768;32767;",128,0,0,0
310 12 .stabs "long long int:t7=r1;0;-1;",128,0,0,0
311 13 .stabs "short unsigned int:t8=r1;0;65535;",128,0,0,0
312 14 .stabs "long long unsigned int:t9=r1;0;-1;",128,0,0,0
313 15 .stabs "signed char:t10=r1;-128;127;",128,0,0,0
314 16 .stabs "unsigned char:t11=r1;0;255;",128,0,0,0
315 17 .stabs "float:t12=r1;4;0;",128,0,0,0
316 18 .stabs "double:t13=r1;8;0;",128,0,0,0
317 19 .stabs "long double:t14=r1;8;0;",128,0,0,0
318 20 .stabs "void:t15=15",128,0,0,0
321 23 .ascii "Hello, world!\12\0"
336 38 sethi %hi(LC0),%o1
337 39 or %o1,%lo(LC0),%o0
348 50 .stabs "main:F1",36,0,0,_main
349 51 .stabn 192,0,0,LBB2
350 52 .stabn 224,0,0,LBE2
353 This simple ``hello world'' example demonstrates several of the stab
354 types used to describe C language source files.
356 @node Program structure
357 @chapter Encoding for the structure of the program
360 * Main Program:: Indicate what the main program is
361 * Source Files:: The path and name of the source file
368 @section Main Program
370 Most languages allow the main program to have any name. The
371 @code{N_MAIN} stab type is used for a stab telling the debugger what
372 name is used in this program. Only the name is significant; it will be
373 the name of a function which is the main program. Most C compilers do
374 not use this stab; they expect the debugger to simply assume that the
375 name is @samp{main}, but some C compilers emit an @code{N_MAIN} stab for
376 the @samp{main} function.
379 @section The path and name of the source files
381 Before any other stabs occur, there must be a stab specifying the source
382 file. This information is contained in a symbol of stab type
383 @code{N_SO}; the string contains the name of the file. The value of the
384 symbol is the start address of portion of the text section corresponding
387 With the Sun Solaris2 compiler, the @code{desc} field contains a
388 source-language code.
390 Some compilers (for example, gcc2 and SunOS4 @file{/bin/cc}) also
391 include the directory in which the source was compiled, in a second
392 @code{N_SO} symbol preceding the one containing the file name. This
393 symbol can be distinguished by the fact that it ends in a slash. Code
394 from the cfront C++ compiler can have additional @code{N_SO} symbols for
395 nonexistent source files after the @code{N_SO} for the real source file;
396 these are believed to contain no useful information.
401 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0 ; 100 is N_SO
402 .stabs "hello.c",100,0,0,Ltext0
407 Instead of @code{N_SO} symbols, XCOFF uses a @code{.file} assembler
408 directive which assembles to a standard COFF @code{.file} symbol;
409 explaining this in detail is outside the scope of this document.
411 There are several different schemes for dealing with include files: the
412 traditional @code{N_SOL} approach, Sun's @code{N_BINCL} scheme, and the
413 XCOFF @code{C_BINCL} (which despite the similar name has little in
414 common with @code{N_BINCL}).
416 An @code{N_SOL} symbol specifies which include file subsequent symbols
417 refer to. The string field is the name of the file and the value is the
418 text address corresponding to the start of the previous include file and
419 the start of this one. To specify the main source file again, use an
420 @code{N_SOL} symbol with the name of the main source file.
422 A @code{N_BINCL} symbol specifies the start of an include file. In an
423 object file, only the name is significant. The Sun linker puts data
424 into some of the other fields. The end of the include file is marked by
425 a @code{N_EINCL} symbol (which has no name field). In an ojbect file,
426 there is no significant data in the @code{N_EINCL} symbol; the Sun
427 linker puts data into some of the fields. @code{N_BINCL} and
428 @code{N_EINCL} can be nested. If the linker detects that two source
429 files have identical stabs with a @code{N_BINCL} and @code{N_EINCL} pair
430 (as will generally be the case for a header file), then it only puts out
431 the stabs once. Each additional occurance is replaced by an
432 @code{N_EXCL} symbol. I believe the Sun (SunOS4, not sure about
433 Solaris) linker is the only one which supports this feature.
435 For the start of an include file in XCOFF, use the @file{.bi} assembler
436 directive which generates a @code{C_BINCL} symbol. A @file{.ei}
437 directive, which generates a @code{C_EINCL} symbol, denotes the end of
438 the include file. Both directives are followed by the name of the
439 source file in quotes, which becomes the string for the symbol. The
440 value of each symbol, produced automatically by the assembler and
441 linker, is an offset into the executable which points to the beginning
442 (inclusive, as you'd expect) and end (inclusive, as you would not
443 expect) of the portion of the COFF linetable which corresponds to this
444 include file. @code{C_BINCL} and @code{C_EINCL} do not nest.
447 @section Line Numbers
449 A @code{N_SLINE} symbol represents the start of a source line. The
450 @var{desc} field contains the line number and the @var{value} field
451 contains the code address for the start of that source line. On most
452 machines the address is absolute; for Sun's stabs-in-elf, it is relative
453 to the function in which the @code{N_SLINE} symbol occurs.
455 GNU documents @code{N_DSLINE} and @code{N_BSLINE} symbols for line
456 numbers in the data or bss segments, respectively. They are identical
457 to @code{N_SLINE} but are relocated differently by the linker. They
458 were intended to be used to describe the source location of a variable
459 declaration, but I believe that gcc2 actually puts the line number in
460 the desc field of the stab for the variable itself. GDB has been
461 ignoring these symbols (unless they contain a string field) at least
464 XCOFF uses COFF line numbers instead, which are outside the scope of
465 this document, ammeliorated by adequate marking of include files
466 (@pxref{Source Files}).
468 For single source lines that generate discontiguous code, such as flow
469 of control statements, there may be more than one line number entry for
470 the same source line. In this case there is a line number entry at the
471 start of each code range, each with the same line number.
476 All of the following stabs use the @samp{N_FUN} symbol type.
478 A function is represented by a @samp{F} symbol descriptor for a global
479 (extern) function, and @samp{f} for a static (local) function. The next
480 @samp{N_SLINE} symbol can be used to find the line number of the start
481 of the function. The value field is the address of the start of the
482 function. The type information of the stab represents the return type
483 of the function; thus @samp{foo:f5} means that foo is a function
486 The type information of the stab is optionally followed by type
487 information for each argument, with each argument preceded by @samp{;}.
488 An argument type of 0 means that additional arguments are being passed,
489 whose types and number may vary (@samp{...} in ANSI C). This extension
490 is used by Sun's Solaris compiler. GDB has tolerated it (i.e. at least
491 parsed the syntax, if not necessarily used the information) at least
492 since version 4.8; I don't know whether all versions of dbx will
493 tolerate it. The argument types given here are not merely redundant
494 with the symbols for the arguments themselves (@pxref{Parameters}), they
495 are the types of the arguments as they are passed, before any
496 conversions might take place. For example, if a C function which is
497 declared without a prototype takes a @code{float} argument, the value is
498 passed as a @code{double} but then converted to a @code{float}.
499 Debuggers need to use the types given in the arguments when printing
500 values, but if calling the function they need to use the types given in
501 the symbol defining the function.
503 If the return type and types of arguments of a function which is defined
504 in another source file are specified (i.e. a function prototype in ANSI
505 C), traditionally compilers emit no stab; the only way for the debugger
506 to find the information is if the source file where the function is
507 defined was also compiled with debugging symbols. As an extension the
508 Solaris compiler uses symbol descriptor @samp{P} followed by the return
509 type of the function, followed by the arguments, each preceded by
510 @samp{;}, as in a stab with symbol descriptor @samp{f} or @samp{F}.
511 This use of symbol descriptor @samp{P} can be distinguished from its use
512 for register parameters (@pxref{Parameters}) by the fact that it has
513 symbol type @code{N_FUN}.
515 The AIX documentation also defines symbol descriptor @samp{J} as an
516 internal function. I assume this means a function nested within another
517 function. It also says Symbol descriptor @samp{m} is a module in
518 Modula-2 or extended Pascal.
520 Procedures (functions which do not return values) are represented as
521 functions returning the void type in C. I don't see why this couldn't
522 be used for all languages (inventing a void type for this purpose if
523 necessary), but the AIX documentation defines @samp{I}, @samp{P}, and
524 @samp{Q} for internal, global, and static procedures, respectively.
525 These symbol descriptors are unusual in that they are not followed by
528 For any of the above symbol descriptors, after the symbol descriptor and
529 the type information, there is optionally a comma, followed by the name
530 of the procedure, followed by a comma, followed by a name specifying the
531 scope. The first name is local to the scope specified. I assume then
532 that the name of the symbol (before the @samp{:}), if specified, is some
533 sort of global name. I assume the name specifying the scope is the name
534 of a function specifying that scope. This feature is an AIX extension,
535 and this information is based on the manual; I haven't actually tried
538 The stab representing a procedure is located immediately following the
539 code of the procedure. This stab is in turn directly followed by a
540 group of other stabs describing elements of the procedure. These other
541 stabs describe the procedure's parameters, its block local variables and
549 The @code{.stabs} entry after this code fragment shows the @var{name} of
550 the procedure (@code{main}); the type descriptor @var{desc} (@code{F},
551 for a global procedure); a reference to the predefined type @code{int}
552 for the return type; and the starting @var{address} of the procedure.
554 Here is an exploded summary (with whitespace introduced for clarity),
555 followed by line 50 of our sample assembly output, which has this form:
559 @var{desc} @r{(global proc @samp{F})}
560 @var{return_type_ref} @r{(int)}
566 50 .stabs "main:F1",36,0,0,_main
569 @node Block Structure
570 @section Block Structure
572 The program's block structure is represented by the @code{N_LBRAC} (left
573 brace) and the @code{N_RBRAC} (right brace) stab types. The variables
574 defined inside a block preceded the @code{N_LBRAC} symbol for most
575 compilers, including GCC. Other compilers, such as the Convex, Acorn
576 RISC machine, and Sun acc compilers, put the variables after the
577 @code{N_LBRAC} symbol. The values of the @code{N_LBRAC} and
578 @code{N_RBRAC} symbols are the start and end addresses of the code of
579 the block, respectively. For most machines, they are relative to the
580 starting address of this source file. For the Gould NP1, they are
581 absolute. For Sun's stabs-in-elf, they are relative to the function in
584 The @code{N_LBRAC} and @code{N_RBRAC} stabs that describe the block
585 scope of a procedure are located after the @code{N_FUN} stab that
586 represents the procedure itself.
588 Sun documents the @code{desc} field of @code{N_LBRAC} and
589 @code{N_RBRAC} symbols as containing the nesting level of the block.
590 However, dbx seems not to care, and GCC just always set @code{desc} to
596 The @samp{c} symbol descriptor indicates that this stab represents a
597 constant. This symbol descriptor is an exception to the general rule
598 that symbol descriptors are followed by type information. Instead, it
599 is followed by @samp{=} and one of the following:
603 Boolean constant. @var{value} is a numeric value; I assume it is 0 for
607 Character constant. @var{value} is the numeric value of the constant.
609 @item e @var{type-information} , @var{value}
610 Constant whose value can be represented as integral.
611 @var{type-information} is the type of the constant, as it would appear
612 after a symbol descriptor (@pxref{Stabs Format}). @var{value} is the
613 numeric value of the constant. GDB 4.9 does not actually get the right
614 value if @var{value} does not fit in a host @code{int}, but it does not
615 do anything violent, and future debuggers could be extended to accept
616 integers of any size (whether unsigned or not). This constant type is
617 usually documented as being only for enumeration constants, but GDB has
618 never imposed that restriction; I don't know about other debuggers.
621 Integer constant. @var{value} is the numeric value. The type is some
622 sort of generic integer type (for GDB, a host @code{int}); to specify
623 the type explicitly, use @samp{e} instead.
626 Real constant. @var{value} is the real value, which can be @samp{INF}
627 (optionally preceded by a sign) for infinity, @samp{QNAN} for a quiet
628 NaN (not-a-number), or @samp{SNAN} for a signalling NaN. If it is a
629 normal number the format is that accepted by the C library function
633 String constant. @var{string} is a string enclosed in either @samp{'}
634 (in which case @samp{'} characters within the string are represented as
635 @samp{\'} or @samp{"} (in which case @samp{"} characters within the
636 string are represented as @samp{\"}).
638 @item S @var{type-information} , @var{elements} , @var{bits} , @var{pattern}
639 Set constant. @var{type-information} is the type of the constant, as it
640 would appear after a symbol descriptor (@pxref{Stabs Format}).
641 @var{elements} is the number of elements in the set (Does this means
642 how many bits of @var{pattern} are actually used, which would be
643 redundant with the type, or perhaps the number of bits set in
644 @var{pattern}? I don't get it), @var{bits} is the number of bits in the
645 constant (meaning it specifies the length of @var{pattern}, I think),
646 and @var{pattern} is a hexadecimal representation of the set. AIX
647 documentation refers to a limit of 32 bytes, but I see no reason why
648 this limit should exist. This form could probably be used for arbitrary
649 constants, not just sets; the only catch is that @var{pattern} should be
650 understood to be target, not host, byte order and format.
653 The boolean, character, string, and set constants are not supported by
654 GDB 4.9, but it will ignore them. GDB 4.8 and earlier gave an error
655 message and refused to read symbols from the file containing the
658 This information is followed by @samp{;}.
661 @chapter A Comprehensive Example in C
663 Now we'll examine a second program, @code{example2}, which builds on the
664 first example to introduce the rest of the stab types, symbol
665 descriptors, and type descriptors used in C.
666 @xref{Example2.c} for the complete @file{.c} source,
667 and @pxref{Example2.s} for the @file{.s} assembly code.
668 This description includes parts of those files.
670 @section Flow of control and nested scopes
676 @code{N_SLINE}, @code{N_LBRAC}, @code{N_RBRAC} (cont.)
679 Consider the body of @code{main}, from @file{example2.c}. It shows more
680 about how @code{N_SLINE}, @code{N_RBRAC}, and @code{N_LBRAC} stabs are used.
684 21 static float s_flap;
686 23 for (times=0; times < s_g_repeat; times++)@{
688 25 printf ("Hello world\n");
693 Here we have a single source line, the @samp{for} line, that generates
694 non-linear flow of control, and non-contiguous code. In this case, an
695 @code{N_SLINE} stab with the same line number proceeds each block of
696 non-contiguous code generated from the same source line.
698 The example also shows nested scopes. The @code{N_LBRAC} and
699 @code{N_LBRAC} stabs that describe block structure are nested in the
700 same order as the corresponding code blocks, those of the for loop
701 inside those for the body of main.
704 This is the label for the @code{N_LBRAC} (left brace) stab marking the
705 start of @code{main}.
712 In the first code range for C source line 23, the @code{for} loop
713 initialize and test, @code{N_SLINE} (68) records the line number:
720 58 .stabn 68,0,23,LM2
724 62 sethi %hi(_s_g_repeat),%o0
726 64 ld [%o0+%lo(_s_g_repeat)],%o0
731 @exdent label for the @code{N_LBRAC} (start block) marking the start of @code{for} loop
734 69 .stabn 68,0,25,LM3
736 71 sethi %hi(LC0),%o1
737 72 or %o1,%lo(LC0),%o0
740 75 .stabn 68,0,26,LM4
743 @exdent label for the @code{N_RBRAC} (end block) stab marking the end of the @code{for} loop
749 Now we come to the second code range for source line 23, the @code{for}
750 loop increment and return. Once again, @code{N_SLINE} (68) records the
754 .stabn, N_SLINE, NIL,
758 78 .stabn 68,0,23,LM5
766 86 .stabn 68,0,27,LM6
769 @exdent label for the @code{N_RBRAC} (end block) stab marking the end of the @code{for} loop
772 89 .stabn 68,0,27,LM7
777 94 .stabs "main:F1",36,0,0,_main
778 95 .stabs "argc:p1",160,0,0,68
779 96 .stabs "argv:p20=*21=*2",160,0,0,72
780 97 .stabs "s_flap:V12",40,0,0,_s_flap.0
781 98 .stabs "times:1",128,0,0,-20
785 Here is an illustration of stabs describing nested scopes. The scope
786 nesting is reflected in the nested bracketing stabs (@code{N_LBRAC},
790 .stabn N_LBRAC,NIL,NIL,
791 @var{block-start-address}
793 99 .stabn 192,0,0,LBB2 ## begin proc label
794 100 .stabs "inner:1",128,0,0,-24
795 101 .stabn 192,0,0,LBB3 ## begin for label
799 @code{N_RBRAC} (224), ``right brace'' ends a lexical block (scope).
802 .stabn N_RBRAC,NIL,NIL,
803 @var{block-end-address}
805 102 .stabn 224,0,0,LBE3 ## end for label
806 103 .stabn 224,0,0,LBE2 ## end proc label
813 * Automatic variables:: Variables allocated on the stack.
814 * Global Variables:: Variables used by more than one source file.
815 * Register variables:: Variables in registers.
816 * Common Blocks:: Variables statically allocated together.
817 * Statics:: Variables local to one source file.
818 * Parameters:: Variables for arguments to functions.
821 @node Automatic variables
822 @section Locally scoped automatic variables
829 @item Symbol Descriptor:
833 In addition to describing types, the @code{N_LSYM} stab type also
834 describes locally scoped automatic variables. Refer again to the body
835 of @code{main} in @file{example2.c}. It allocates two automatic
836 variables: @samp{times} is scoped to the body of @code{main}, and
837 @samp{inner} is scoped to the body of the @code{for} loop.
838 @samp{s_flap} is locally scoped but not automatic, and will be discussed
843 21 static float s_flap;
845 23 for (times=0; times < s_g_repeat; times++)@{
847 25 printf ("Hello world\n");
852 The @code{N_LSYM} stab for an automatic variable is located just before the
853 @code{N_LBRAC} stab describing the open brace of the block to which it is
857 @exdent @code{N_LSYM} (128): automatic variable, scoped locally to @code{main}
860 @var{type information}",
862 @var{frame-pointer-offset}
864 98 .stabs "times:1",128,0,0,-20
865 99 .stabn 192,0,0,LBB2 ## begin `main' N_LBRAC
867 @exdent @code{N_LSYM} (128): automatic variable, scoped locally to the @code{for} loop
870 @var{type information}",
872 @var{frame-pointer-offset}
874 100 .stabs "inner:1",128,0,0,-24
875 101 .stabn 192,0,0,LBB3 ## begin `for' loop N_LBRAC
878 The symbol descriptor is omitted for automatic variables. Since type
879 information should being with a digit, @samp{-}, or @samp{(}, only
880 digits, @samp{-}, and @samp{(} are precluded from being used for symbol
881 descriptors by this fact. However, the Acorn RISC machine (ARM) is said
882 to get this wrong: it puts out a mere type definition here, without the
883 preceding @code{@var{typenumber}=}. This is a bad idea; there is no
884 guarantee that type descriptors are distinct from symbol descriptors.
886 @node Global Variables
887 @section Global Variables
894 @item Symbol Descriptor:
898 Global variables are represented by the @code{N_GSYM} stab type. The symbol
899 descriptor, following the colon in the string field, is @samp{G}. Following
900 the @samp{G} is a type reference or type definition. In this example it is a
901 type reference to the basic C type, @code{char}. The first source line in
909 yields the following stab. The stab immediately precedes the code that
910 allocates storage for the variable it describes.
913 @exdent @code{N_GSYM} (32): global symbol
918 N_GSYM, NIL, NIL, NIL
920 21 .stabs "g_foo:G2",32,0,0,0
927 The address of the variable represented by the @code{N_GSYM} is not contained
928 in the @code{N_GSYM} stab. The debugger gets this information from the
929 external symbol for the global variable.
931 @node Register variables
932 @section Register variables
934 @c According to an old version of this manual, AIX uses C_RPSYM instead
935 @c of C_RSYM. I am skeptical; this should be verified.
936 Register variables have their own stab type, @code{N_RSYM}, and their
937 own symbol descriptor, @code{r}. The stab's value field contains the
938 number of the register where the variable data will be stored.
940 The value is the register number.
942 AIX defines a separate symbol descriptor @samp{d} for floating point
943 registers. This seems unnecessary---why not just just give floating
944 point registers different register numbers? I have not verified whether
945 the compiler actually uses @samp{d}.
947 If the register is explicitly allocated to a global variable, but not
951 register int g_bar asm ("%g5");
954 the stab may be emitted at the end of the object file, with
955 the other bss symbols.
958 @section Common Blocks
960 A common block is a statically allocated section of memory which can be
961 referred to by several source files. It may contain several variables.
962 I believe @sc{fortran} is the only language with this feature. A
963 @code{N_BCOMM} stab begins a common block and an @code{N_ECOMM} stab
964 ends it. The only thing which is significant about these two stabs is
965 their name, which can be used to look up a normal (non-debugging) symbol
966 which gives the address of the common block. Then each stab between the
967 @code{N_BCOMM} and the @code{N_ECOMM} specifies a member of that common
968 block; its value is the offset within the common block of that variable.
969 The @code{N_ECOML} stab type is documented for this purpose, but Sun's
970 @sc{fortran} compiler uses @code{N_GSYM} instead. The test case I
971 looked at had a common block local to a function and it used the
972 @samp{V} symbol descriptor; I assume one would use @samp{S} if not local
973 to a function (that is, if a common block @emph{can} be anything other
974 than local to a function).
977 @section Static Variables
979 Initialized static variables are represented by the @samp{S} and
980 @samp{V} symbol descriptors. @samp{S} means file scope static, and
981 @samp{V} means procedure scope static.
983 In a.out files, @code{N_STSYM} means the data segment (although gcc
984 2.4.5 has a bug in that it uses @code{N_FUN}, so neither dbx nor gdb can
985 find the variables), @code{N_FUN} means the text segment, and
986 @code{N_LCSYM} means the bss segment.
988 In xcoff files, each symbol has a section number, so the stab type
989 need not indicate the segment.
991 In ecoff files, the storage class is used to specify the section, so the
992 stab type need not indicate the segment.
994 @c In ELF files, it apparently is a big mess. See kludge in dbxread.c
995 @c in GDB. FIXME: Investigate where this kludge comes from.
997 @c This is the place to mention N_ROSYM; I'd rather do so once I can
998 @c coherently explain how this stuff works for stabs-in-elf.
1000 For example, the source lines
1003 static const int var_const = 5;
1004 static int var_init = 2;
1005 static int var_noinit;
1009 yield the following stabs:
1012 .stabs "var_const:S1",36,0,0,_var_const ; @r{36 = N_FUN}
1014 .stabs "var_init:S1",38,0,0,_var_init ; @r{38 = N_STSYM}
1016 .stabs "var_noinit:S1",40,0,0,_var_noinit ; @r{40 = N_LCSYM}
1022 Parameters to a function are represented by a stab (or sometimes two,
1023 see below) for each parameter. The stabs are in the order in which the
1024 debugger should print the parameters (i.e. the order in which the
1025 parameters are declared in the source file).
1027 The symbol descriptor @samp{p} is used to refer to parameters which are
1028 in the arglist. Symbols have symbol type @samp{N_PSYM}. The value of
1029 the symbol is the offset relative to the argument list.
1031 If the parameter is passed in a register, then the traditional way to do
1032 this is to provide two symbols for each argument:
1035 .stabs "arg:p1" . . . ; N_PSYM
1036 .stabs "arg:r1" . . . ; N_RSYM
1039 Debuggers are expected to use the second one to find the value, and the
1040 first one to know that it is an argument.
1042 Because this is kind of ugly, some compilers use symbol descriptor
1043 @samp{P} or @samp{R} to indicate an argument which is in a register.
1044 The symbol value is the register number. @samp{P} and @samp{R} mean the
1045 same thing, the difference is that @samp{P} is a GNU invention and
1046 @samp{R} is an IBM (xcoff) invention. As of version 4.9, GDB should
1047 handle either one. Symbol type @samp{C_RPSYM} is used with @samp{R} and
1048 @samp{N_RSYM} is used with @samp{P}.
1050 According to the AIX documentation symbol descriptor @samp{D} is for a
1051 parameter passed in a floating point register. This seems
1052 unnecessary---why not just use @samp{R} with a register number which
1053 indicates that it's a floating point register? I haven't verified
1054 whether the system actually does what the documentation indicates.
1056 There is at least one case where GCC uses a @samp{p}/@samp{r} pair
1057 rather than @samp{P}; this is where the argument is passed in the
1058 argument list and then loaded into a register.
1060 On the sparc and hppa, for a @samp{P} symbol whose type is a structure
1061 or union, the register contains the address of the structure. On the
1062 sparc, this is also true of a @samp{p}/@samp{r} pair (using Sun cc) or a
1063 @samp{p} symbol. However, if a (small) structure is really in a
1064 register, @samp{r} is used. And, to top it all off, on the hppa it
1065 might be a structure which was passed on the stack and loaded into a
1066 register and for which there is a @samp{p}/@samp{r} pair! I believe
1067 that symbol descriptor @samp{i} is supposed to deal with this case, (it
1068 is said to mean "value parameter by reference, indirect access", I don't
1069 know the source for this information) but I don't know details or what
1070 compilers or debuggers use it, if any (not GDB or GCC). It is not clear
1071 to me whether this case needs to be dealt with differently than
1072 parameters passed by reference (see below).
1074 There is another case similar to an argument in a register, which is an
1075 argument which is actually stored as a local variable. Sometimes this
1076 happens when the argument was passed in a register and then the compiler
1077 stores it as a local variable. If possible, the compiler should claim
1078 that it's in a register, but this isn't always done. Some compilers use
1079 the pair of symbols approach described above ("arg:p" followed by
1080 "arg:"); this includes gcc1 (not gcc2) on the sparc when passing a small
1081 structure and gcc2 (sometimes) when the argument type is float and it is
1082 passed as a double and converted to float by the prologue (in the latter
1083 case the type of the "arg:p" symbol is double and the type of the "arg:"
1084 symbol is float). GCC, at least on the 960, uses a single @samp{p}
1085 symbol descriptor for an argument which is stored as a local variable
1086 but uses @samp{N_LSYM} instead of @samp{N_PSYM}. In this case the value
1087 of the symbol is an offset relative to the local variables for that
1088 function, not relative to the arguments (on some machines those are the
1089 same thing, but not on all).
1091 If the parameter is passed by reference (e.g. Pascal VAR parameters),
1092 then type symbol descriptor is @samp{v} if it is in the argument list,
1093 or @samp{a} if it in a register. Other than the fact that these contain
1094 the address of the parameter other than the parameter itself, they are
1095 identical to @samp{p} and @samp{R}, respectively. I believe @samp{a} is
1096 an AIX invention; @samp{v} is supported by all stabs-using systems as
1099 @c Is this paragraph correct? It is based on piecing together patchy
1100 @c information and some guesswork
1101 Conformant arrays refer to a feature of Modula-2, and perhaps other
1102 languages, in which the size of an array parameter is not known to the
1103 called function until run-time. Such parameters have two stabs, a
1104 @samp{x} for the array itself, and a @samp{C}, which represents the size
1105 of the array. The value of the @samp{x} stab is the offset in the
1106 argument list where the address of the array is stored (it this right?
1107 it is a guess); the value of the @samp{C} stab is the offset in the
1108 argument list where the size of the array (in elements? in bytes?) is
1111 The following are also said to go with @samp{N_PSYM}:
1114 "name" -> "param_name:#type"
1116 -> pF FORTRAN function parameter
1117 -> X (function result variable)
1118 -> b (based variable)
1120 value -> offset from the argument pointer (positive).
1123 As a simple example, the code
1135 .stabs "main:F1",36,0,0,_main ; 36 is N_FUN
1136 .stabs "argc:p1",160,0,0,68 ; 160 is N_PSYM
1137 .stabs "argv:p20=*21=*2",160,0,0,72
1140 The type definition of argv is interesting because it contains several
1141 type definitions. Type 21 is pointer to type 2 (char) and argv (type 20) is
1145 @chapter Type Definitions
1147 Now let's look at some variable definitions involving complex types.
1148 This involves understanding better how types are described. In the
1149 examples so far types have been described as references to previously
1150 defined types or defined in terms of subranges of or pointers to
1151 previously defined types. The section that follows will talk about
1152 the various other type descriptors that may follow the = sign in a
1156 * Builtin types:: Integers, floating point, void, etc.
1157 * Miscellaneous Types:: Pointers, sets, files, etc.
1158 * Cross-references:: Referring to a type not yet defined.
1159 * Subranges:: A type with a specific range.
1160 * Arrays:: An aggregate type of same-typed elements.
1161 * Strings:: Like an array but also has a length.
1162 * Enumerations:: Like an integer but the values have names.
1163 * Structures:: An aggregate type of different-typed elements.
1164 * Typedefs:: Giving a type a name.
1165 * Unions:: Different types sharing storage.
1170 @section Builtin types
1172 Certain types are built in (@code{int}, @code{short}, @code{void},
1173 @code{float}, etc.); the debugger recognizes these types and knows how
1174 to handle them. Thus don't be surprised if some of the following ways
1175 of specifying builtin types do not specify everything that a debugger
1176 would need to know about the type---in some cases they merely specify
1177 enough information to distinguish the type from other types.
1179 The traditional way to define builtin types is convolunted, so new ways
1180 have been invented to describe them. Sun's ACC uses the @samp{b} and
1181 @samp{R} type descriptors, and IBM uses negative type numbers. GDB can
1182 accept all three, as of version 4.8; dbx just accepts the traditional
1183 builtin types and perhaps one of the other two formats.
1186 * Traditional Builtin Types:: Put on your seatbelts and prepare for kludgery
1187 * Builtin Type Descriptors:: Builtin types with special type descriptors
1188 * Negative Type Numbers:: Builtin types using negative type numbers
1191 @node Traditional Builtin Types
1192 @subsection Traditional Builtin types
1194 Often types are defined as subranges of themselves. If the array bounds
1195 can fit within an @code{int}, then they are given normally. For example:
1198 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0 ; 128 is N_LSYM
1199 .stabs "char:t2=r2;0;127;",128,0,0,0
1202 Builtin types can also be described as subranges of @code{int}:
1205 .stabs "unsigned short:t6=r1;0;65535;",128,0,0,0
1208 If the lower bound of a subrange is 0 and the upper bound is -1, it
1209 means that the type is an unsigned integral type whose bounds are too
1210 big to describe in an int. Traditionally this is only used for
1211 @code{unsigned int} and @code{unsigned long}; GCC also sometimes uses it
1212 for @code{long long} and @code{unsigned long long}, and the only way to
1213 tell those types apart is to look at their names. On other machines GCC
1214 puts out bounds in octal, with a leading 0. In this case a negative
1215 bound consists of a number which is a 1 bit followed by a bunch of 0
1216 bits, and a positive bound is one in which a bunch of bits are 1.
1219 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
1220 .stabs "long long int:t7=r1;0;-1;",128,0,0,0
1223 If the lower bound of a subrange is 0 and the upper bound is negative,
1224 it means that it is an unsigned integral type whose size in bytes is the
1225 absolute value of the upper bound. I believe this is a Convex
1226 convention for @code{unsigned long long}.
1228 If the lower bound of a subrange is negative and the upper bound is 0,
1229 it means that the type is a signed integral type whose size in bytes is
1230 the absolute value of the lower bound. I believe this is a Convex
1231 convention for @code{long long}. To distinguish this from a legitimate
1232 subrange, the type should be a subrange of itself. I'm not sure whether
1233 this is the case for Convex.
1235 If the upper bound of a subrange is 0, it means that this is a floating
1236 point type, and the lower bound of the subrange indicates the number of
1240 .stabs "float:t12=r1;4;0;",128,0,0,0
1241 .stabs "double:t13=r1;8;0;",128,0,0,0
1244 However, GCC writes @code{long double} the same way it writes
1245 @code{double}; the only way to distinguish them is by the name:
1248 .stabs "long double:t14=r1;8;0;",128,0,0,0
1251 Complex types are defined the same way as floating-point types; the only
1252 way to distinguish a single-precision complex from a double-precision
1253 floating-point type is by the name.
1255 The C @code{void} type is defined as itself:
1258 .stabs "void:t15=15",128,0,0,0
1261 I'm not sure how a boolean type is represented.
1263 @node Builtin Type Descriptors
1264 @subsection Defining Builtin Types using Builtin Type Descriptors
1266 There are various type descriptors to define builtin types:
1269 @c FIXME: clean up description of width and offset, once we figure out
1271 @item b @var{signed} @var{char-flag} @var{width} ; @var{offset} ; @var{nbits} ;
1272 Define an integral type. @var{signed} is @samp{u} for unsigned or
1273 @samp{s} for signed. @var{char-flag} is @samp{c} which indicates this
1274 is a character type, or is omitted. I assume this is to distinguish an
1275 integral type from a character type of the same size, for example it
1276 might make sense to set it for the C type @code{wchar_t} so the debugger
1277 can print such variables differently (Solaris does not do this). Sun
1278 sets it on the C types @code{signed char} and @code{unsigned char} which
1279 arguably is wrong. @var{width} and @var{offset} appear to be for small
1280 objects stored in larger ones, for example a @code{short} in an
1281 @code{int} register. @var{width} is normally the number of bytes in the
1282 type. @var{offset} seems to always be zero. @var{nbits} is the number
1283 of bits in the type.
1285 Note that type descriptor @samp{b} used for builtin types conflicts with
1286 its use for Pascal space types (@pxref{Miscellaneous Types}); they can
1287 be distinguished because the character following the type descriptor
1288 will be a digit, @samp{(}, or @samp{-} for a Pascal space type, or
1289 @samp{u} or @samp{s} for a builtin type.
1292 Documented by AIX to define a wide character type, but their compiler
1293 actually uses negative type numbers (@pxref{Negative Type Numbers}).
1295 @item R @var{fp_type} ; @var{bytes} ;
1296 Define a floating point type. @var{fp_type} has one of the following values:
1300 IEEE 32-bit (single precision) floating point format.
1303 IEEE 64-bit (double precision) floating point format.
1305 @item 3 (NF_COMPLEX)
1306 @item 4 (NF_COMPLEX16)
1307 @item 5 (NF_COMPLEX32)
1308 @c "GDB source" really means @file{include/aout/stab_gnu.h}, but trying
1309 @c to put that here got an overfull hbox.
1310 These are for complex numbers. A comment in the GDB source describes
1311 them as Fortran complex, double complex, and complex*16, respectively,
1312 but what does that mean? (i.e. Single precision? Double precison?).
1314 @item 6 (NF_LDOUBLE)
1315 Long double. This should probably only be used for Sun format long
1316 double, and new codes should be used for other floating point formats
1317 (NF_DOUBLE can be used if a long double is really just an IEEE double,
1321 @var{bytes} is the number of bytes occupied by the type. This allows a
1322 debugger to perform some operations with the type even if it doesn't
1323 understand @var{fp_code}.
1325 @item g @var{type-information} ; @var{nbits}
1326 Documented by AIX to define a floating type, but their compiler actually
1327 uses negative type numbers (@pxref{Negative Type Numbers}).
1329 @item c @var{type-information} ; @var{nbits}
1330 Documented by AIX to define a complex type, but their compiler actually
1331 uses negative type numbers (@pxref{Negative Type Numbers}).
1334 The C @code{void} type is defined as a signed integral type 0 bits long:
1336 .stabs "void:t19=bs0;0;0",128,0,0,0
1338 The Solaris compiler seems to omit the trailing semicolon in this case.
1339 Getting sloppy in this way is not a swift move because if a type is
1340 embedded in a more complex expression it is necessary to be able to tell
1343 I'm not sure how a boolean type is represented.
1345 @node Negative Type Numbers
1346 @subsection Negative Type numbers
1348 Since the debugger knows about the builtin types anyway, the idea of
1349 negative type numbers is simply to give a special type number which
1350 indicates the built in type. There is no stab defining these types.
1352 I'm not sure whether anyone has tried to define what this means if
1353 @code{int} can be other than 32 bits (or other types can be other than
1354 their customary size). If @code{int} has exactly one size for each
1355 architecture, then it can be handled easily enough, but if the size of
1356 @code{int} can vary according the compiler options, then it gets hairy.
1357 I guess the consistent way to do this would be to define separate
1358 negative type numbers for 16-bit @code{int} and 32-bit @code{int};
1359 therefore I have indicated below the customary size (and other format
1360 information) for each type. The information below is currently correct
1361 because AIX on the RS6000 is the only system which uses these type
1362 numbers. If these type numbers start to get used on other systems, I
1363 suspect the correct thing to do is to define a new number in cases where
1364 a type does not have the size and format indicated below.
1366 Also note that part of the definition of the negative type number is
1367 the name of the type. Types with identical size and format but
1368 different names have different negative type numbers.
1372 @code{int}, 32 bit signed integral type.
1375 @code{char}, 8 bit type holding a character. Both GDB and dbx on AIX
1376 treat this as signed. GCC uses this type whether @code{char} is signed
1377 or not, which seems like a bad idea. The AIX compiler (xlc) seems to
1378 avoid this type; it uses -5 instead for @code{char}.
1381 @code{short}, 16 bit signed integral type.
1384 @code{long}, 32 bit signed integral type.
1387 @code{unsigned char}, 8 bit unsigned integral type.
1390 @code{signed char}, 8 bit signed integral type.
1393 @code{unsigned short}, 16 bit unsigned integral type.
1396 @code{unsigned int}, 32 bit unsigned integral type.
1399 @code{unsigned}, 32 bit unsigned integral type.
1402 @code{unsigned long}, 32 bit unsigned integral type.
1405 @code{void}, type indicating the lack of a value.
1408 @code{float}, IEEE single precision.
1411 @code{double}, IEEE double precision.
1414 @code{long double}, IEEE double precision. The compiler claims the size
1415 will increase in a future release, and for binary compatibility you have
1416 to avoid using @code{long double}. I hope when they increase it they
1417 use a new negative type number.
1420 @code{integer}. 32 bit signed integral type.
1423 @code{boolean}. 32 bit type. How is the truth value encoded? Is it
1424 the least significant bit or is it a question of whether the whole value
1425 is zero or non-zero?
1428 @code{short real}. IEEE single precision.
1431 @code{real}. IEEE double precision.
1434 @code{stringptr}. @xref{Strings}.
1437 @code{character}, 8 bit unsigned character type.
1440 @code{logical*1}, 8 bit unsigned integral type.
1443 @code{logical*2}, 16 bit unsigned integral type.
1446 @code{logical*4}, 32 bit unsigned integral type.
1449 @code{logical}, 32 bit unsigned integral type.
1452 @code{complex}. A complex type consisting of two IEEE single-precision
1453 floating point values.
1456 @code{complex}. A complex type consisting of two IEEE double-precision
1457 floating point values.
1460 @code{integer*1}, 8 bit signed integral type.
1463 @code{integer*2}, 16 bit signed integral type.
1466 @code{integer*4}, 32 bit signed integral type.
1469 @code{wchar}. Wide character, 16 bits wide, unsigned (what format?
1473 @node Miscellaneous Types
1474 @section Miscellaneous Types
1477 @item b @var{type-information} ; @var{bytes}
1478 Pascal space type. This is documented by IBM; what does it mean?
1480 Note that this use of the @samp{b} type descriptor can be distinguished
1481 from its use for builtin integral types (@pxref{Builtin Type
1482 Descriptors}) because the character following the type descriptor is
1483 always a digit, @samp{(}, or @samp{-}.
1485 @item B @var{type-information}
1486 A volatile-qualified version of @var{type-information}. This is a Sun
1487 extension. A volatile-qualified type means that references and stores
1488 to a variable of that type must not be optimized or cached; they must
1489 occur as the user specifies them.
1491 @item d @var{type-information}
1492 File of type @var{type-information}. As far as I know this is only used
1495 @item k @var{type-information}
1496 A const-qualified version of @var{type-information}. This is a Sun
1497 extension. A const-qualified type means that a variable of this type
1500 @item M @var{type-information} ; @var{length}
1501 Multiple instance type. The type seems to composed of @var{length}
1502 repetitions of @var{type-information}, for example @code{character*3} is
1503 represented by @samp{M-2;3}, where @samp{-2} is a reference to a
1504 character type (@pxref{Negative Type Numbers}). I'm not sure how this
1505 differs from an array. This appears to be a FORTRAN feature.
1506 @var{length} is a bound, like those in range types, @xref{Subranges}.
1508 @item S @var{type-information}
1509 Pascal set type. @var{type-information} must be a small type such as an
1510 enumeration or a subrange, and the type is a bitmask whose length is
1511 specified by the number of elements in @var{type-information}.
1513 @item * @var{type-information}
1514 Pointer to @var{type-information}.
1517 @node Cross-references
1518 @section Cross-references to other types
1520 If a type is used before it is defined, one common way to deal with this
1521 is just to use a type reference to a type which has not yet been
1522 defined. The debugger is expected to be able to deal with this.
1524 Another way is with the @samp{x} type descriptor, which is followed by
1525 @samp{s} for a structure tag, @samp{u} for a union tag, or @samp{e} for
1526 a enumerator tag, followed by the name of the tag, followed by @samp{:}.
1527 for example the following C declarations:
1537 .stabs "bar:G16=*17=xsfoo:",32,0,0,0
1540 Not all debuggers support the @samp{x} type descriptor, so on some
1541 machines GCC does not use it. I believe that for the above example it
1542 would just emit a reference to type 17 and never define it, but I
1543 haven't verified that.
1545 Modula-2 imported types, at least on AIX, use the @samp{i} type
1546 descriptor, which is followed by the name of the module from which the
1547 type is imported, followed by @samp{:}, followed by the name of the
1548 type. There is then optionally a comma followed by type information for
1549 the type (This differs from merely naming the type (@pxref{Typedefs}) in
1550 that it identifies the module; I don't understand whether the name of
1551 the type given here is always just the same as the name we are giving
1552 it, or whether this type descriptor is used with a nameless stab
1553 (@pxref{Stabs Format}), or what). The symbol ends with @samp{;}.
1556 @section Subrange types
1558 The @samp{r} type descriptor defines a type as a subrange of another
1559 type. It is followed by type information for the type which it is a
1560 subrange of, a semicolon, an integral lower bound, a semicolon, an
1561 integral upper bound, and a semicolon. The AIX documentation does not
1562 specify the trailing semicolon, in an effort to specify array indexes
1563 more cleanly, but a subrange which is not an array index has always
1564 included a trailing semicolon (@pxref{Arrays}).
1566 Instead of an integer, either bound can be one of the following:
1569 @item A @var{offset}
1570 The bound is passed by reference on the stack at offset @var{offset}
1571 from the argument list. @xref{Parameters}, for more information on such
1574 @item T @var{offset}
1575 The bound is passed by value on the stack at offset @var{offset} from
1578 @item a @var{register-number}
1579 The bound is pased by reference in register number
1580 @var{register-number}.
1582 @item t @var{register-number}
1583 The bound is passed by value in register number @var{register-number}.
1589 Subranges are also used for builtin types, @xref{Traditional Builtin Types}.
1592 @section Array types
1594 Arrays use the @samp{a} type descriptor. Following the type descriptor
1595 is the type of the index and the type of the array elements. If the
1596 index type is a range type, it will end in a semicolon; if it is not a
1597 range type (for example, if it is a type reference), there does not
1598 appear to be any way to tell where the types are separated. In an
1599 effort to clean up this mess, IBM documents the two types as being
1600 separated by a semicolon, and a range type as not ending in a semicolon
1601 (but this is not right for range types which are not array indexes,
1602 @pxref{Subranges}). I think probably the best solution is to specify
1603 that a semicolon ends a range type, and that the index type and element
1604 type of an array are separated by a semicolon, but that if the index
1605 type is a range type, the extra semicolon can be omitted. GDB (at least
1606 through version 4.9) doesn't support any kind of index type other than a
1607 range anyway; I'm not sure about dbx.
1609 It is well established, and widely used, that the type of the index,
1610 unlike most types found in the stabs, is merely a type definition, not
1611 type information (@pxref{Stabs Format}) (that is, it need not start with
1612 @var{type-number}@code{=} if it is defining a new type). According to a
1613 comment in GDB, this is also true of the type of the array elements; it
1614 gives @samp{ar1;1;10;ar1;1;10;4} as a legitimate way to express a two
1615 dimensional array. According to AIX documentation, the element type
1616 must be type information. GDB accepts either.
1618 The type of the index is often a range type, expressed as the letter r
1619 and some parameters. It defines the size of the array. In the example
1620 below, the range @code{r1;0;2;} defines an index type which is a
1621 subrange of type 1 (integer), with a lower bound of 0 and an upper bound
1622 of 2. This defines the valid range of subscripts of a three-element C
1625 For example, the definition
1628 char char_vec[3] = @{'a','b','c'@};
1635 .stabs "char_vec:G19=ar1;0;2;2",32,0,0,0
1644 If an array is @dfn{packed}, it means that the elements are spaced more
1645 closely than normal, saving memory at the expense of speed. For
1646 example, an array of 3-byte objects might, if unpacked, have each
1647 element aligned on a 4-byte boundary, but if packed, have no padding.
1648 One way to specify that something is packed is with type attributes
1649 (@pxref{Stabs Format}), in the case of arrays another is to use the
1650 @samp{P} type descriptor instead of @samp{a}. Other than specifying a
1651 packed array, @samp{P} is identical to @samp{a}.
1653 @c FIXME-what is it? A pointer?
1654 An open array is represented by the @samp{A} type descriptor followed by
1655 type information specifying the type of the array elements.
1657 @c FIXME: what is the format of this type? A pointer to a vector of pointers?
1658 An N-dimensional dynamic array is represented by
1661 D @var{dimensions} ; @var{type-information}
1664 @c Does dimensions really have this meaning? The AIX documentation
1666 @var{dimensions} is the number of dimensions; @var{type-information}
1667 specifies the type of the array elements.
1669 @c FIXME: what is the format of this type? A pointer to some offsets in
1671 A subarray of an N-dimensional array is represented by
1674 E @var{dimensions} ; @var{type-information}
1677 @c Does dimensions really have this meaning? The AIX documentation
1679 @var{dimensions} is the number of dimensions; @var{type-information}
1680 specifies the type of the array elements.
1685 Some languages, like C or the original Pascal, do not have string types,
1686 they just have related things like arrays of characters. But most
1687 Pascals and various other languages have string types, which are
1688 indicated as follows:
1691 @item n @var{type-information} ; @var{bytes}
1692 @var{bytes} is the maximum length. I'm not sure what
1693 @var{type-information} is; I suspect that it means that this is a string
1694 of @var{type-information} (thus allowing a string of integers, a string
1695 of wide characters, etc., as well as a string of characters). Not sure
1696 what the format of this type is. This is an AIX feature.
1698 @item z @var{type-information} ; @var{bytes}
1699 Just like @samp{n} except that this is a gstring, not an ordinary
1700 string. I don't know the difference.
1703 Pascal Stringptr. What is this? This is an AIX feature.
1707 @section Enumerations
1709 Enumerations are defined with the @samp{e} type descriptor.
1711 @c FIXME: Where does this information properly go? Perhaps it is
1712 @c redundant with something we already explain.
1713 The source line below declares an enumeration type. It is defined at
1714 file scope between the bodies of main and s_proc in example2.c.
1715 The type definition is located after the N_RBRAC that marks the end of
1716 the previous procedure's block scope, and before the N_FUN that marks
1717 the beginning of the next procedure's block scope. Therefore it does not
1718 describe a block local symbol, but a file local one.
1723 enum e_places @{first,second=3,last@};
1727 generates the following stab
1730 .stabs "e_places:T22=efirst:0,second:3,last:4,;",128,0,0,0
1733 The symbol descriptor (T) says that the stab describes a structure,
1734 enumeration, or type tag. The type descriptor e, following the 22= of
1735 the type definition narrows it down to an enumeration type. Following
1736 the e is a list of the elements of the enumeration. The format is
1737 name:value,. The list of elements ends with a ;.
1739 There is no standard way to specify the size of an enumeration type; it
1740 is determined by the architecture (normally all enumerations types are
1741 32 bits). There should be a way to specify an enumeration type of
1742 another size; type attributes would be one way to do this @xref{Stabs
1752 @code{N_LSYM} or @code{C_DECL}
1753 @item Symbol Descriptor:
1755 @item Type Descriptor:
1759 The following source code declares a structure tag and defines an
1760 instance of the structure in global scope. Then a typedef equates the
1761 structure tag with a new type. A seperate stab is generated for the
1762 structure tag, the structure typedef, and the structure instance. The
1763 stabs for the tag and the typedef are emited when the definitions are
1764 encountered. Since the structure elements are not initialized, the
1765 stab and code for the structure variable itself is located at the end
1766 of the program in .common.
1772 9 char s_char_vec[8];
1773 10 struct s_tag* s_next;
1776 13 typedef struct s_tag s_typedef;
1779 The structure tag is an N_LSYM stab type because, like the enum, the
1780 symbol is file scope. Like the enum, the symbol descriptor is T, for
1781 enumeration, struct or tag type. The symbol descriptor s following
1782 the 16= of the type definition narrows the symbol type to struct.
1784 Following the struct symbol descriptor is the number of bytes the
1785 struct occupies, followed by a description of each structure element.
1786 The structure element descriptions are of the form name:type, bit
1787 offset from the start of the struct, and number of bits in the
1792 <128> N_LSYM - type definition
1793 .stabs "name:sym_desc(struct tag) Type_def(16)=type_desc(struct type)
1795 elem_name:type_ref(int),bit_offset,field_bits;
1796 elem_name:type_ref(float),bit_offset,field_bits;
1797 elem_name:type_def(17)=type_desc(array)
1798 index_type(range of int from 0 to 7);
1799 element_type(char),bit_offset,field_bits;;",
1802 30 .stabs "s_tag:T16=s20s_int:1,0,32;s_float:12,32,32;
1803 s_char_vec:17=ar1;0;7;2,64,64;s_next:18=*16,128,32;;",128,0,0,0
1806 In this example, two of the structure elements are previously defined
1807 types. For these, the type following the name: part of the element
1808 description is a simple type reference. The other two structure
1809 elements are new types. In this case there is a type definition
1810 embedded after the name:. The type definition for the array element
1811 looks just like a type definition for a standalone array. The s_next
1812 field is a pointer to the same kind of structure that the field is an
1813 element of. So the definition of structure type 16 contains an type
1814 definition for an element which is a pointer to type 16.
1817 @section Giving a type a name
1819 To give a type a name, use the @samp{t} symbol descriptor. For example,
1822 .stabs "s_typedef:t16",128,0,0,0
1825 specifies that @code{s_typedef} refers to type number 16. Such stabs
1826 have symbol type @code{N_LSYM} or @code{C_DECL}.
1828 If instead, you are specifying the tag name for a structure, union, or
1829 enumeration, use the @samp{T} symbol descriptor instead. I believe C is
1830 the only language with this feature.
1832 If the type is an opaque type (I believe this is a Modula-2 feature),
1833 AIX provides a type descriptor to specify it. The type descriptor is
1834 @samp{o} and is followed by a name. I don't know what the name
1835 means---is it always the same as the name of the type, or is this type
1836 descriptor used with a nameless stab (@pxref{Stabs Format})? There
1837 optionally follows a comma followed by type information which defines
1838 the type of this type. If omitted, a semicolon is used in place of the
1839 comma and the type information, and, the type is much like a generic
1840 pointer type---it has a known size but little else about it is
1846 Next let's look at unions. In example2 this union type is declared
1847 locally to a procedure and an instance of the union is defined.
1857 This code generates a stab for the union tag and a stab for the union
1858 variable. Both use the N_LSYM stab type. Since the union variable is
1859 scoped locally to the procedure in which it is defined, its stab is
1860 located immediately preceding the N_LBRAC for the procedure's block
1863 The stab for the union tag, however is located preceding the code for
1864 the procedure in which it is defined. The stab type is N_LSYM. This
1865 would seem to imply that the union type is file scope, like the struct
1866 type s_tag. This is not true. The contents and position of the stab
1867 for u_type do not convey any infomation about its procedure local
1872 .stabs "name:sym_desc(union tag)type_def(22)=type_desc(union)
1874 elem_name:type_ref(int),bit_offset(0),bit_size(32);
1875 elem_name:type_ref(float),bit_offset(0),bit_size(32);
1876 elem_name:type_ref(ptr to char),bit_offset(0),bit_size(32);;"
1877 N_LSYM, NIL, NIL, NIL
1881 105 .stabs "u_tag:T23=u4u_int:1,0,32;u_float:12,0,32;u_char:21,0,32;;",
1885 The symbol descriptor, T, following the name: means that the stab
1886 describes an enumeration, struct or type tag. The type descriptor u,
1887 following the 23= of the type definition, narrows it down to a union
1888 type definition. Following the u is the number of bytes in the union.
1889 After that is a list of union element descriptions. Their format is
1890 name:type, bit offset into the union, and number of bytes for the
1893 The stab for the union variable follows. Notice that the frame
1894 pointer offset for local variables is negative.
1897 <128> N_LSYM - local variable (with no symbol descriptor)
1898 .stabs "name:type_ref(u_tag)", N_LSYM, NIL, NIL, frame_ptr_offset
1902 130 .stabs "an_u:23",128,0,0,-20
1905 @node Function Types
1906 @section Function types
1908 There are various types for function variables. These types are not
1909 used in defining functions; see symbol descriptor @samp{f}; they are
1910 used for things like pointers to functions.
1912 The simple, traditional, type is type descriptor @samp{f} is followed by
1913 type information for the return type of the function, followed by a
1916 This does not deal with functions the number and type of whose
1917 parameters are part of their type, as found in Modula-2 or ANSI C. AIX
1918 provides extensions to specify these, using the @samp{f}, @samp{F},
1919 @samp{p}, and @samp{R} type descriptors.
1921 First comes the type descriptor. Then, if it is @samp{f} or @samp{F},
1922 this is a function, and the type information for the return type of the
1923 function follows, followed by a comma. Then comes the number of
1924 parameters to the function and a semicolon. Then, for each parameter,
1925 there is the name of the parameter followed by a colon (this is only
1926 present for type descriptors @samp{R} and @samp{F} which represent
1927 Pascal function or procedure parameters), type information for the
1928 parameter, a comma, @samp{0} if passed by reference or @samp{1} if
1929 passed by value, and a semicolon. The type definition ends with a
1939 generates the following code:
1942 .stabs "g_pf:G24=*25=f1",32,0,0,0
1943 .common _g_pf,4,"bss"
1946 The variable defines a new type, 24, which is a pointer to another new
1947 type, 25, which is defined as a function returning int.
1950 @chapter Symbol information in symbol tables
1952 This section examines more closely the format of symbol table entries
1953 and how stab assembler directives map to them. It also describes what
1954 transformations the assembler and linker make on data from stabs.
1956 Each time the assembler encounters a stab in its input file it puts
1957 each field of the stab into corresponding fields in a symbol table
1958 entry of its output file. If the stab contains a string field, the
1959 symbol table entry for that stab points to a string table entry
1960 containing the string data from the stab. Assembler labels become
1961 relocatable addresses. Symbol table entries in a.out have the format:
1964 struct internal_nlist @{
1965 unsigned long n_strx; /* index into string table of name */
1966 unsigned char n_type; /* type of symbol */
1967 unsigned char n_other; /* misc info (usually empty) */
1968 unsigned short n_desc; /* description field */
1969 bfd_vma n_value; /* value of symbol */
1973 For .stabs directives, the n_strx field holds the character offset
1974 from the start of the string table to the string table entry
1975 containing the "string" field. For other classes of stabs (.stabn and
1976 .stabd) this field is null.
1978 Symbol table entries with n_type fields containing a value greater or
1979 equal to 0x20 originated as stabs generated by the compiler (with one
1980 random exception). Those with n_type values less than 0x20 were
1981 placed in the symbol table of the executable by the assembler or the
1984 The linker concatenates object files and does fixups of externally
1985 defined symbols. You can see the transformations made on stab data by
1986 the assembler and linker by examining the symbol table after each pass
1987 of the build, first the assemble and then the link.
1989 To do this use nm with the -ap options. This dumps the symbol table,
1990 including debugging information, unsorted. For stab entries the
1991 columns are: value, other, desc, type, string. For assembler and
1992 linker symbols, the columns are: value, type, string.
1994 There are a few important things to notice about symbol tables. Where
1995 the value field of a stab contains a frame pointer offset, or a
1996 register number, that value is unchanged by the rest of the build.
1998 Where the value field of a stab contains an assembly language label,
1999 it is transformed by each build step. The assembler turns it into a
2000 relocatable address and the linker turns it into an absolute address.
2001 This source line defines a static variable at file scope:
2004 3 static int s_g_repeat
2008 The following stab describes the symbol.
2011 26 .stabs "s_g_repeat:S1",38,0,0,_s_g_repeat
2015 The assembler transforms the stab into this symbol table entry in the
2016 @file{.o} file. The location is expressed as a data segment offset.
2019 21 00000084 - 00 0000 STSYM s_g_repeat:S1
2023 in the symbol table entry from the executable, the linker has made the
2024 relocatable address absolute.
2027 22 0000e00c - 00 0000 STSYM s_g_repeat:S1
2030 Stabs for global variables do not contain location information. In
2031 this case the debugger finds location information in the assembler or
2032 linker symbol table entry describing the variable. The source line:
2042 21 .stabs "g_foo:G2",32,0,0,0
2045 The variable is represented by the following two symbol table entries
2046 in the object file. The first one originated as a stab. The second
2047 one is an external symbol. The upper case D signifies that the n_type
2048 field of the symbol table contains 7, N_DATA with local linkage (see
2049 Table B). The value field following the file's line number is empty
2050 for the stab entry. For the linker symbol it contains the
2051 rellocatable address corresponding to the variable.
2054 19 00000000 - 00 0000 GSYM g_foo:G2
2055 20 00000080 D _g_foo
2059 These entries as transformed by the linker. The linker symbol table
2060 entry now holds an absolute address.
2063 21 00000000 - 00 0000 GSYM g_foo:G2
2065 215 0000e008 D _g_foo
2069 @chapter GNU C++ stabs
2072 * Basic Cplusplus types::
2075 * Methods:: Method definition
2077 * Method Modifiers::
2080 * Virtual Base Classes::
2084 @subsection type descriptors added for C++ descriptions
2088 method type (two ## if minimal debug)
2091 Member (class and variable) type. It is followed by type information
2092 for the offset basetype, a comma, and type information for the type of
2093 the field being pointed to. (FIXME: this is acknowledged to be
2094 gibberish. Can anyone say what really goes here?).
2096 Note that there is a conflict between this and type attributes
2097 (@pxref{Stabs Format}); both use type descriptor @samp{@@}.
2098 Fortunately, the @samp{@@} type descriptor used in this C++ sense always
2099 will be followed by a digit, @samp{(}, or @samp{-}, and type attributes
2100 never start with those things.
2103 @node Basic Cplusplus types
2104 @section Basic types for C++
2106 << the examples that follow are based on a01.C >>
2109 C++ adds two more builtin types to the set defined for C. These are
2110 the unknown type and the vtable record type. The unknown type, type
2111 16, is defined in terms of itself like the void type.
2113 The vtable record type, type 17, is defined as a structure type and
2114 then as a structure tag. The structure has four fields, delta, index,
2115 pfn, and delta2. pfn is the function pointer.
2117 << In boilerplate $vtbl_ptr_type, what are the fields delta,
2118 index, and delta2 used for? >>
2120 This basic type is present in all C++ programs even if there are no
2121 virtual methods defined.
2124 .stabs "struct_name:sym_desc(type)type_def(17)=type_desc(struct)struct_bytes(8)
2125 elem_name(delta):type_ref(short int),bit_offset(0),field_bits(16);
2126 elem_name(index):type_ref(short int),bit_offset(16),field_bits(16);
2127 elem_name(pfn):type_def(18)=type_desc(ptr to)type_ref(void),
2128 bit_offset(32),field_bits(32);
2129 elem_name(delta2):type_def(short int);bit_offset(32),field_bits(16);;"
2134 .stabs "$vtbl_ptr_type:t17=s8
2135 delta:6,0,16;index:6,16,16;pfn:18=*15,32,32;delta2:6,32,16;;"
2140 .stabs "name:sym_dec(struct tag)type_ref($vtbl_ptr_type)",N_LSYM,NIL,NIL,NIL
2144 .stabs "$vtbl_ptr_type:T17",128,0,0,0
2147 @node Simple classes
2148 @section Simple class definition
2150 The stabs describing C++ language features are an extension of the
2151 stabs describing C. Stabs representing C++ class types elaborate
2152 extensively on the stab format used to describe structure types in C.
2153 Stabs representing class type variables look just like stabs
2154 representing C language variables.
2156 Consider the following very simple class definition.
2162 int Ameth(int in, char other);
2166 The class baseA is represented by two stabs. The first stab describes
2167 the class as a structure type. The second stab describes a structure
2168 tag of the class type. Both stabs are of stab type N_LSYM. Since the
2169 stab is not located between an N_FUN and a N_LBRAC stab this indicates
2170 that the class is defined at file scope. If it were, then the N_LSYM
2171 would signify a local variable.
2173 A stab describing a C++ class type is similar in format to a stab
2174 describing a C struct, with each class member shown as a field in the
2175 structure. The part of the struct format describing fields is
2176 expanded to include extra information relevent to C++ class members.
2177 In addition, if the class has multiple base classes or virtual
2178 functions the struct format outside of the field parts is also
2181 In this simple example the field part of the C++ class stab
2182 representing member data looks just like the field part of a C struct
2183 stab. The section on protections describes how its format is
2184 sometimes extended for member data.
2186 The field part of a C++ class stab representing a member function
2187 differs substantially from the field part of a C struct stab. It
2188 still begins with `name:' but then goes on to define a new type number
2189 for the member function, describe its return type, its argument types,
2190 its protection level, any qualifiers applied to the method definition,
2191 and whether the method is virtual or not. If the method is virtual
2192 then the method description goes on to give the vtable index of the
2193 method, and the type number of the first base class defining the
2196 When the field name is a method name it is followed by two colons
2197 rather than one. This is followed by a new type definition for the
2198 method. This is a number followed by an equal sign and then the
2199 symbol descriptor `##', indicating a method type. This is followed by
2200 a type reference showing the return type of the method and a
2203 The format of an overloaded operator method name differs from that
2204 of other methods. It is "op$::XXXX." where XXXX is the operator name
2205 such as + or +=. The name ends with a period, and any characters except
2206 the period can occur in the XXXX string.
2208 The next part of the method description represents the arguments to
2209 the method, preceeded by a colon and ending with a semi-colon. The
2210 types of the arguments are expressed in the same way argument types
2211 are expressed in C++ name mangling. In this example an int and a char
2214 This is followed by a number, a letter, and an asterisk or period,
2215 followed by another semicolon. The number indicates the protections
2216 that apply to the member function. Here the 2 means public. The
2217 letter encodes any qualifier applied to the method definition. In
2218 this case A means that it is a normal function definition. The dot
2219 shows that the method is not virtual. The sections that follow
2220 elaborate further on these fields and describe the additional
2221 information present for virtual methods.
2225 .stabs "class_name:sym_desc(type)type_def(20)=type_desc(struct)struct_bytes(4)
2226 field_name(Adat):type(int),bit_offset(0),field_bits(32);
2228 method_name(Ameth)::type_def(21)=type_desc(method)return_type(int);
2229 :arg_types(int char);
2230 protection(public)qualifier(normal)virtual(no);;"
2235 .stabs "baseA:t20=s4Adat:1,0,32;Ameth::21=##1;:ic;2A.;;",128,0,0,0
2237 .stabs "class_name:sym_desc(struct tag)",N_LSYM,NIL,NIL,NIL
2239 .stabs "baseA:T20",128,0,0,0
2242 @node Class instance
2243 @section Class instance
2245 As shown above, describing even a simple C++ class definition is
2246 accomplished by massively extending the stab format used in C to
2247 describe structure types. However, once the class is defined, C stabs
2248 with no modifications can be used to describe class instances. The
2258 yields the following stab describing the class instance. It looks no
2259 different from a standard C stab describing a local variable.
2262 .stabs "name:type_ref(baseA)", N_LSYM, NIL, NIL, frame_ptr_offset
2266 .stabs "AbaseA:20",128,0,0,-20
2270 @section Method defintion
2272 The class definition shown above declares Ameth. The C++ source below
2277 baseA::Ameth(int in, char other)
2284 This method definition yields three stabs following the code of the
2285 method. One stab describes the method itself and following two
2286 describe its parameters. Although there is only one formal argument
2287 all methods have an implicit argument which is the `this' pointer.
2288 The `this' pointer is a pointer to the object on which the method was
2289 called. Note that the method name is mangled to encode the class name
2290 and argument types. << Name mangling is not described by this
2291 document - Is there already such a doc? >>
2294 .stabs "name:symbol_desriptor(global function)return_type(int)",
2295 N_FUN, NIL, NIL, code_addr_of_method_start
2297 .stabs "Ameth__5baseAic:F1",36,0,0,_Ameth__5baseAic
2300 Here is the stab for the `this' pointer implicit argument. The name
2301 of the `this' pointer is always `this.' Type 19, the `this' pointer is
2302 defined as a pointer to type 20, baseA, but a stab defining baseA has
2303 not yet been emited. Since the compiler knows it will be emited
2304 shortly, here it just outputs a cross reference to the undefined
2305 symbol, by prefixing the symbol name with xs.
2308 .stabs "name:sym_desc(register param)type_def(19)=
2309 type_desc(ptr to)type_ref(baseA)=
2310 type_desc(cross-reference to)baseA:",N_RSYM,NIL,NIL,register_number
2312 .stabs "this:P19=*20=xsbaseA:",64,0,0,8
2315 The stab for the explicit integer argument looks just like a parameter
2316 to a C function. The last field of the stab is the offset from the
2317 argument pointer, which in most systems is the same as the frame
2321 .stabs "name:sym_desc(value parameter)type_ref(int)",
2322 N_PSYM,NIL,NIL,offset_from_arg_ptr
2324 .stabs "in:p1",160,0,0,72
2327 << The examples that follow are based on A1.C >>
2330 @section Protections
2333 In the simple class definition shown above all member data and
2334 functions were publicly accessable. The example that follows
2335 contrasts public, protected and privately accessable fields and shows
2336 how these protections are encoded in C++ stabs.
2338 Protections for class member data are signified by two characters
2339 embeded in the stab defining the class type. These characters are
2340 located after the name: part of the string. /0 means private, /1
2341 means protected, and /2 means public. If these characters are omited
2342 this means that the member is public. The following C++ source:
2356 generates the following stab to describe the class type all_data.
2359 .stabs "class_name:sym_desc(type)type_def(19)=type_desc(struct)struct_bytes
2360 data_name:/protection(private)type_ref(int),bit_offset,num_bits;
2361 data_name:/protection(protected)type_ref(char),bit_offset,num_bits;
2362 data_name:(/num omited, private)type_ref(float),bit_offset,num_bits;;"
2367 .stabs "all_data:t19=s12
2368 priv_dat:/01,0,32;prot_dat:/12,32,8;pub_dat:12,64,32;;",128,0,0,0
2371 Protections for member functions are signified by one digit embeded in
2372 the field part of the stab describing the method. The digit is 0 if
2373 private, 1 if protected and 2 if public. Consider the C++ class
2377 class all_methods @{
2379 int priv_meth(int in)@{return in;@};
2381 char protMeth(char in)@{return in;@};
2383 float pubMeth(float in)@{return in;@};
2387 It generates the following stab. The digit in question is to the left
2388 of an `A' in each case. Notice also that in this case two symbol
2389 descriptors apply to the class name struct tag and struct type.
2392 .stabs "class_name:sym_desc(struct tag&type)type_def(21)=
2393 sym_desc(struct)struct_bytes(1)
2394 meth_name::type_def(22)=sym_desc(method)returning(int);
2395 :args(int);protection(private)modifier(normal)virtual(no);
2396 meth_name::type_def(23)=sym_desc(method)returning(char);
2397 :args(char);protection(protected)modifier(normal)virual(no);
2398 meth_name::type_def(24)=sym_desc(method)returning(float);
2399 :args(float);protection(public)modifier(normal)virtual(no);;",
2404 .stabs "all_methods:Tt21=s1priv_meth::22=##1;:i;0A.;protMeth::23=##2;:c;1A.;
2405 pubMeth::24=##12;:f;2A.;;",128,0,0,0
2408 @node Method Modifiers
2409 @section Method Modifiers (const, volatile, const volatile)
2413 In the class example described above all the methods have the normal
2414 modifier. This method modifier information is located just after the
2415 protection information for the method. This field has four possible
2416 character values. Normal methods use A, const methods use B, volatile
2417 methods use C, and const volatile methods use D. Consider the class
2423 int ConstMeth (int arg) const @{ return arg; @};
2424 char VolatileMeth (char arg) volatile @{ return arg; @};
2425 float ConstVolMeth (float arg) const volatile @{return arg; @};
2429 This class is described by the following stab:
2432 .stabs "class(A):sym_desc(struct)type_def(20)=type_desc(struct)struct_bytes(1)
2433 meth_name(ConstMeth)::type_def(21)sym_desc(method)
2434 returning(int);:arg(int);protection(public)modifier(const)virtual(no);
2435 meth_name(VolatileMeth)::type_def(22)=sym_desc(method)
2436 returning(char);:arg(char);protection(public)modifier(volatile)virt(no)
2437 meth_name(ConstVolMeth)::type_def(23)=sym_desc(method)
2438 returning(float);:arg(float);protection(public)modifer(const volatile)
2439 virtual(no);;", @dots{}
2443 .stabs "A:T20=s1ConstMeth::21=##1;:i;2B.;VolatileMeth::22=##2;:c;2C.;
2444 ConstVolMeth::23=##12;:f;2D.;;",128,0,0,0
2447 @node Virtual Methods
2448 @section Virtual Methods
2450 << The following examples are based on a4.C >>
2452 The presence of virtual methods in a class definition adds additional
2453 data to the class description. The extra data is appended to the
2454 description of the virtual method and to the end of the class
2455 description. Consider the class definition below:
2461 virtual int A_virt (int arg) @{ return arg; @};
2465 This results in the stab below describing class A. It defines a new
2466 type (20) which is an 8 byte structure. The first field of the class
2467 struct is Adat, an integer, starting at structure offset 0 and
2470 The second field in the class struct is not explicitly defined by the
2471 C++ class definition but is implied by the fact that the class
2472 contains a virtual method. This field is the vtable pointer. The
2473 name of the vtable pointer field starts with $vf and continues with a
2474 type reference to the class it is part of. In this example the type
2475 reference for class A is 20 so the name of its vtable pointer field is
2476 $vf20, followed by the usual colon.
2478 Next there is a type definition for the vtable pointer type (21).
2479 This is in turn defined as a pointer to another new type (22).
2481 Type 22 is the vtable itself, which is defined as an array, indexed by
2482 a range of integers between 0 and 1, and whose elements are of type
2483 17. Type 17 was the vtable record type defined by the boilerplate C++
2484 type definitions, as shown earlier.
2486 The bit offset of the vtable pointer field is 32. The number of bits
2487 in the field are not specified when the field is a vtable pointer.
2489 Next is the method definition for the virtual member function A_virt.
2490 Its description starts out using the same format as the non-virtual
2491 member functions described above, except instead of a dot after the
2492 `A' there is an asterisk, indicating that the function is virtual.
2493 Since is is virtual some addition information is appended to the end
2494 of the method description.
2496 The first number represents the vtable index of the method. This is a
2497 32 bit unsigned number with the high bit set, followed by a
2500 The second number is a type reference to the first base class in the
2501 inheritence hierarchy defining the virtual member function. In this
2502 case the class stab describes a base class so the virtual function is
2503 not overriding any other definition of the method. Therefore the
2504 reference is to the type number of the class that the stab is
2507 This is followed by three semi-colons. One marks the end of the
2508 current sub-section, one marks the end of the method field, and the
2509 third marks the end of the struct definition.
2511 For classes containing virtual functions the very last section of the
2512 string part of the stab holds a type reference to the first base
2513 class. This is preceeded by `~%' and followed by a final semi-colon.
2516 .stabs "class_name(A):type_def(20)=sym_desc(struct)struct_bytes(8)
2517 field_name(Adat):type_ref(int),bit_offset(0),field_bits(32);
2518 field_name(A virt func ptr):type_def(21)=type_desc(ptr to)type_def(22)=
2519 sym_desc(array)index_type_ref(range of int from 0 to 1);
2520 elem_type_ref(vtbl elem type),
2522 meth_name(A_virt)::typedef(23)=sym_desc(method)returning(int);
2523 :arg_type(int),protection(public)normal(yes)virtual(yes)
2524 vtable_index(1);class_first_defining(A);;;~%first_base(A);",
2528 @c FIXME: bogus line break.
2530 .stabs "A:t20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
2531 A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
2535 @section Inheritence
2537 Stabs describing C++ derived classes include additional sections that
2538 describe the inheritence hierarchy of the class. A derived class stab
2539 also encodes the number of base classes. For each base class it tells
2540 if the base class is virtual or not, and if the inheritence is private
2541 or public. It also gives the offset into the object of the portion of
2542 the object corresponding to each base class.
2544 This additional information is embeded in the class stab following the
2545 number of bytes in the struct. First the number of base classes
2546 appears bracketed by an exclamation point and a comma.
2548 Then for each base type there repeats a series: two digits, a number,
2549 a comma, another number, and a semi-colon.
2551 The first of the two digits is 1 if the base class is virtual and 0 if
2552 not. The second digit is 2 if the derivation is public and 0 if not.
2554 The number following the first two digits is the offset from the start
2555 of the object to the part of the object pertaining to the base class.
2557 After the comma, the second number is a type_descriptor for the base
2558 type. Finally a semi-colon ends the series, which repeats for each
2561 The source below defines three base classes A, B, and C and the
2569 virtual int A_virt (int arg) @{ return arg; @};
2575 virtual int B_virt (int arg) @{return arg; @};
2581 virtual int C_virt (int arg) @{return arg; @};
2584 class D : A, virtual B, public C @{
2587 virtual int A_virt (int arg ) @{ return arg+1; @};
2588 virtual int B_virt (int arg) @{ return arg+2; @};
2589 virtual int C_virt (int arg) @{ return arg+3; @};
2590 virtual int D_virt (int arg) @{ return arg; @};
2594 Class stabs similar to the ones described earlier are generated for
2597 @c FIXME!!! the linebreaks in the following example probably make the
2598 @c examples literally unusable, but I don't know any other way to get
2599 @c them on the page.
2600 @c One solution would be to put some of the type definitions into
2601 @c separate stabs, even if that's not exactly what the compiler actually
2604 .stabs "A:T20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
2605 A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
2607 .stabs "B:Tt25=s8Bdat:1,0,32;$vf25:21,32;B_virt::26=##1;
2608 :i;2A*-2147483647;25;;;~%25;",128,0,0,0
2610 .stabs "C:Tt28=s8Cdat:1,0,32;$vf28:21,32;C_virt::29=##1;
2611 :i;2A*-2147483647;28;;;~%28;",128,0,0,0
2614 In the stab describing derived class D below, the information about
2615 the derivation of this class is encoded as follows.
2618 .stabs "derived_class_name:symbol_descriptors(struct tag&type)=
2619 type_descriptor(struct)struct_bytes(32)!num_bases(3),
2620 base_virtual(no)inheritence_public(no)base_offset(0),
2621 base_class_type_ref(A);
2622 base_virtual(yes)inheritence_public(no)base_offset(NIL),
2623 base_class_type_ref(B);
2624 base_virtual(no)inheritence_public(yes)base_offset(64),
2625 base_class_type_ref(C); @dots{}
2628 @c FIXME! fake linebreaks.
2630 .stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:
2631 1,160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt:
2632 :32:i;2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;
2633 28;;D_virt::32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
2636 @node Virtual Base Classes
2637 @section Virtual Base Classes
2639 A derived class object consists of a concatination in memory of the
2640 data areas defined by each base class, starting with the leftmost and
2641 ending with the rightmost in the list of base classes. The exception
2642 to this rule is for virtual inheritence. In the example above, class
2643 D inherits virtually from base class B. This means that an instance
2644 of a D object will not contain it's own B part but merely a pointer to
2645 a B part, known as a virtual base pointer.
2647 In a derived class stab, the base offset part of the derivation
2648 information, described above, shows how the base class parts are
2649 ordered. The base offset for a virtual base class is always given as
2650 0. Notice that the base offset for B is given as 0 even though B is
2651 not the first base class. The first base class A starts at offset 0.
2653 The field information part of the stab for class D describes the field
2654 which is the pointer to the virtual base class B. The vbase pointer
2655 name is $vb followed by a type reference to the virtual base class.
2656 Since the type id for B in this example is 25, the vbase pointer name
2659 @c FIXME!! fake linebreaks below
2661 .stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:1,
2662 160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt::32:i;
2663 2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;28;;D_virt:
2664 :32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
2667 Following the name and a semicolon is a type reference describing the
2668 type of the virtual base class pointer, in this case 24. Type 24 was
2669 defined earlier as the type of the B class `this` pointer. The
2670 `this' pointer for a class is a pointer to the class type.
2673 .stabs "this:P24=*25=xsB:",64,0,0,8
2676 Finally the field offset part of the vbase pointer field description
2677 shows that the vbase pointer is the first field in the D object,
2678 before any data fields defined by the class. The layout of a D class
2679 object is a follows, Adat at 0, the vtable pointer for A at 32, Cdat
2680 at 64, the vtable pointer for C at 96, the virtual ase pointer for B
2681 at 128, and Ddat at 160.
2684 @node Static Members
2685 @section Static Members
2687 The data area for a class is a concatenation of the space used by the
2688 data members of the class. If the class has virtual methods, a vtable
2689 pointer follows the class data. The field offset part of each field
2690 description in the class stab shows this ordering.
2692 << How is this reflected in stabs? See Cygnus bug #677 for some info. >>
2695 @appendix Example2.c - source code for extended example
2699 2 register int g_bar asm ("%g5");
2700 3 static int s_g_repeat = 2;
2706 9 char s_char_vec[8];
2707 10 struct s_tag* s_next;
2710 13 typedef struct s_tag s_typedef;
2712 15 char char_vec[3] = @{'a','b','c'@};
2714 17 main (argc, argv)
2718 21 static float s_flap;
2720 23 for (times=0; times < s_g_repeat; times++)@{
2722 25 printf ("Hello world\n");
2726 29 enum e_places @{first,second=3,last@};
2728 31 static s_proc (s_arg, s_ptr_arg, char_vec)
2730 33 s_typedef* s_ptr_arg;
2744 @appendix Example2.s - assembly code for extended example
2748 2 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0
2749 3 .stabs "example2.c",100,0,0,Ltext0
2752 6 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0
2753 7 .stabs "char:t2=r2;0;127;",128,0,0,0
2754 8 .stabs "long int:t3=r1;-2147483648;2147483647;",128,0,0,0
2755 9 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
2756 10 .stabs "long unsigned int:t5=r1;0;-1;",128,0,0,0
2757 11 .stabs "short int:t6=r1;-32768;32767;",128,0,0,0
2758 12 .stabs "long long int:t7=r1;0;-1;",128,0,0,0
2759 13 .stabs "short unsigned int:t8=r1;0;65535;",128,0,0,0
2760 14 .stabs "long long unsigned int:t9=r1;0;-1;",128,0,0,0
2761 15 .stabs "signed char:t10=r1;-128;127;",128,0,0,0
2762 16 .stabs "unsigned char:t11=r1;0;255;",128,0,0,0
2763 17 .stabs "float:t12=r1;4;0;",128,0,0,0
2764 18 .stabs "double:t13=r1;8;0;",128,0,0,0
2765 19 .stabs "long double:t14=r1;8;0;",128,0,0,0
2766 20 .stabs "void:t15=15",128,0,0,0
2767 21 .stabs "g_foo:G2",32,0,0,0
2772 26 .stabs "s_g_repeat:S1",38,0,0,_s_g_repeat
2776 @c FIXME! fake linebreak in line 30
2777 30 .stabs "s_tag:T16=s20s_int:1,0,32;s_float:12,32,32;s_char_vec:
2778 17=ar1;0;7;2,64,64;s_next:18=*16,128,32;;",128,0,0,0
2779 31 .stabs "s_typedef:t16",128,0,0,0
2780 32 .stabs "char_vec:G19=ar1;0;2;2",32,0,0,0
2781 33 .global _char_vec
2787 39 .reserve _s_flap.0,4,"bss",4
2791 43 .ascii "Hello world\12\0"
2796 48 .stabn 68,0,20,LM1
2799 51 save %sp,-144,%sp
2806 58 .stabn 68,0,23,LM2
2810 62 sethi %hi(_s_g_repeat),%o0
2812 64 ld [%o0+%lo(_s_g_repeat)],%o0
2817 69 .stabn 68,0,25,LM3
2819 71 sethi %hi(LC0),%o1
2820 72 or %o1,%lo(LC0),%o0
2823 75 .stabn 68,0,26,LM4
2826 78 .stabn 68,0,23,LM5
2834 86 .stabn 68,0,27,LM6
2837 89 .stabn 68,0,27,LM7
2842 94 .stabs "main:F1",36,0,0,_main
2843 95 .stabs "argc:p1",160,0,0,68
2844 96 .stabs "argv:p20=*21=*2",160,0,0,72
2845 97 .stabs "s_flap:V12",40,0,0,_s_flap.0
2846 98 .stabs "times:1",128,0,0,-20
2847 99 .stabn 192,0,0,LBB2
2848 100 .stabs "inner:1",128,0,0,-24
2849 101 .stabn 192,0,0,LBB3
2850 102 .stabn 224,0,0,LBE3
2851 103 .stabn 224,0,0,LBE2
2852 104 .stabs "e_places:T22=efirst:0,second:3,last:4,;",128,0,0,0
2853 @c FIXME: fake linebreak in line 105
2854 105 .stabs "u_tag:T23=u4u_int:1,0,32;u_float:12,0,32;u_char:21,0,32;;",
2859 109 .stabn 68,0,35,LM8
2862 112 save %sp,-120,%sp
2868 118 .stabn 68,0,41,LM9
2871 121 .stabn 68,0,41,LM10
2876 126 .stabs "s_proc:f1",36,0,0,_s_proc
2877 127 .stabs "s_arg:p16",160,0,0,0
2878 128 .stabs "s_ptr_arg:p18",160,0,0,72
2879 129 .stabs "char_vec:p21",160,0,0,76
2880 130 .stabs "an_u:23",128,0,0,-20
2881 131 .stabn 192,0,0,LBB4
2882 132 .stabn 224,0,0,LBE4
2883 133 .stabs "g_bar:r1",64,0,0,5
2884 134 .stabs "g_pf:G24=*25=f1",32,0,0,0
2885 135 .common _g_pf,4,"bss"
2886 136 .stabs "g_an_s:G16",32,0,0,0
2887 137 .common _g_an_s,20,"bss"
2891 @appendix Values for the Stab Type Field
2893 These are all the possible values for the stab type field, for
2894 @code{a.out} files. This does not apply to XCOFF.
2896 The following types are used by the linker and assembler; there is
2897 nothing stabs-specific about them. Since this document does not attempt
2898 to describe aspects of object file format other than the debugging
2899 format, no details are given.
2901 @c Try to get most of these to fit on a single line.
2911 File scope absolute symbol
2913 @item 0x3 N_ABS | N_EXT
2914 External absolute symbol
2917 File scope text symbol
2919 @item 0x5 N_TEXT | N_EXT
2920 External text symbol
2923 File scope data symbol
2925 @item 0x7 N_DATA | N_EXT
2926 External data symbol
2929 File scope BSS symbol
2931 @item 0x9 N_BSS | N_EXT
2935 Same as N_FN, for Sequent compilers
2938 Symbol is indirected to another symbol
2941 Common sym -- visable after shared lib dynamic link
2944 Absolute set element
2947 Text segment set element
2950 Data segment set element
2953 BSS segment set element
2956 Pointer to set vector
2958 @item 0x1e N_WARNING
2959 Print a warning message during linking
2962 File name of a .o file
2965 The following symbol types indicate that this is a stab. This is the
2966 full list of stab numbers, including stab types that are used in
2967 languages other than C.
2971 Global symbol, @xref{N_GSYM}.
2974 Function name (for BSD Fortran), @xref{N_FNAME}.
2977 Function name (@pxref{Procedures}) or text segment variable
2981 Data segment file-scope variable, @xref{Statics}.
2984 BSS segment file-scope variable, @xref{Statics}.
2987 Name of main routine, @xref{Main Program}.
2989 @c FIXME: discuss this in the main body of the text where we talk about
2990 @c using N_FUN for variables.
2992 Read-only data symbol (Solaris2). Most systems use N_FUN for this.
2995 Global symbol (for Pascal), @xref{N_PC}.
2998 Number of symbols (according to Ultrix V4.0), @xref{N_NSYMS}.
3001 No DST map for sym (according to Ultrix V4.0), @xref{N_NOMAP}.
3003 @c FIXME: describe this solaris feature in the body of the text (see
3004 @c comments in include/aout/stab.def).
3006 Object file (Solaris2).
3008 @c See include/aout/stab.def for (a little) more info.
3010 Debugger options (Solaris2).
3013 Register variable, @xref{N_RSYM}.
3016 Modula-2 compilation unit, @xref{N_M2C}.
3019 Line number in text segment, @xref{Line Numbers}.
3022 Line number in data segment, @xref{Line Numbers}.
3025 Line number in bss segment, @xref{Line Numbers}.
3028 Sun source code browser, path to .cb file, @xref{N_BROWS}.
3031 Gnu Modula2 definition module dependency, @xref{N_DEFD}.
3034 Function start/body/end line numbers (Solaris2).
3037 Gnu C++ exception variable, @xref{N_EHDECL}.
3040 Modula2 info "for imc" (according to Ultrix V4.0), @xref{N_MOD2}.
3043 Gnu C++ "catch" clause, @xref{N_CATCH}.
3046 Structure of union element, @xref{N_SSYM}.
3049 Last stab for module (Solaris2).
3052 Path and name of source file , @xref{Source Files}.
3055 Automatic var in the stack or type definition, @xref{N_LSYM}, @xref{Typedefs}.
3058 Beginning of an include file (Sun only), @xref{Source Files}.
3061 Name of include file, @xref{Source Files}.
3064 Parameter variable, @xref{Parameters}.
3067 End of an include file, @xref{Source Files}.
3070 Alternate entry point, @xref{N_ENTRY}.
3073 Beginning of a lexical block, @xref{Block Structure}.
3076 Place holder for a deleted include file, @xref{Source Files}.
3079 Modula2 scope information (Sun linker), @xref{N_SCOPE}.
3082 End of a lexical block, @xref{Block Structure}.
3085 Begin named common block, @xref{Common Blocks}.
3088 End named common block, @xref{Common Blocks}.
3091 Member of a common block, @xref{Common Blocks}.
3093 @c FIXME: How does this really work? Move it to main body of document.
3095 Pascal @code{with} statement: type,,0,0,offset (Solaris2).
3098 Gould non-base registers, @xref{Gould}.
3101 Gould non-base registers, @xref{Gould}.
3104 Gould non-base registers, @xref{Gould}.
3107 Gould non-base registers, @xref{Gould}.
3110 Gould non-base registers, @xref{Gould}.
3113 @c Restore the default table indent
3118 @node Symbol Descriptors
3119 @appendix Table of Symbol Descriptors
3121 @c Please keep this alphabetical
3123 @c In TeX, this looks great, digit is in italics. But makeinfo insists
3124 @c on putting it in `', not realizing that @var should override @code.
3125 @c I don't know of any way to make makeinfo do the right thing. Seems
3126 @c like a makeinfo bug to me.
3130 Local variable, @xref{Automatic variables}.
3133 Parameter passed by reference in register, @xref{Parameters}.
3136 Constant, @xref{Constants}.
3139 Conformant array bound (Pascal, maybe other languages),
3140 @xref{Parameters}. Name of a caught exception (GNU C++). These can be
3141 distinguished because the latter uses N_CATCH and the former uses
3142 another symbol type.
3145 Floating point register variable, @xref{Register variables}.
3148 Parameter in floating point register, @xref{Parameters}.
3151 File scope function, @xref{Procedures}.
3154 Global function, @xref{Procedures}.
3157 Global variable, @xref{Global Variables}.
3163 Internal (nested) procedure, @xref{Procedures}.
3166 Internal (nested) function, @xref{Procedures}.
3169 Label name (documented by AIX, no further information known).
3172 Module, @xref{Procedures}.
3175 Argument list parameter, @xref{Parameters}.
3181 FORTRAN Function parameter, @xref{Parameters}.
3184 Unfortunately, three separate meanings have been independently invented
3185 for this symbol descriptor. At least the GNU and Sun uses can be
3186 distinguished by the symbol type. Global Procedure (AIX) (symbol type
3187 used unknown), @xref{Procedures}. Register parameter (GNU) (symbol type
3188 N_PSYM), @xref{Parameters}. Prototype of function referenced by this
3189 file (Sun acc) (symbol type N_FUN).
3192 Static Procedure, @xref{Procedures}.
3195 Register parameter @xref{Parameters}.
3198 Register variable, @xref{Register variables}.
3201 File scope variable, @xref{Statics}.
3204 Type name, @xref{Typedefs}.
3207 enumeration, struct or union tag, @xref{Typedefs}.
3210 Parameter passed by reference, @xref{Parameters}.
3213 Procedure scope static variable, @xref{Statics}.
3216 Conformant array, @xref{Parameters}.
3219 Function return variable, @xref{Parameters}.
3222 @node Type Descriptors
3223 @appendix Table of Type Descriptors
3228 Type reference, @xref{Stabs Format}.
3231 Reference to builtin type, @xref{Negative Type Numbers}.
3234 Method (C++), @xref{Cplusplus}.
3237 Pointer, @xref{Miscellaneous Types}.
3243 Type Attributes (AIX), @xref{Stabs Format}. Member (class and variable)
3244 type (GNU C++), @xref{Cplusplus}.
3247 Array, @xref{Arrays}.
3250 Open array, @xref{Arrays}.
3253 Pascal space type (AIX), @xref{Miscellaneous Types}. Builtin integer
3254 type (Sun), @xref{Builtin Type Descriptors}.
3257 Volatile-qualified type, @xref{Miscellaneous Types}.
3260 Complex builtin type, @xref{Builtin Type Descriptors}.
3263 COBOL Picture type. See AIX documentation for details.
3266 File type, @xref{Miscellaneous Types}.
3269 N-dimensional dynamic array, @xref{Arrays}.
3272 Enumeration type, @xref{Enumerations}.
3275 N-dimensional subarray, @xref{Arrays}.
3278 Function type, @xref{Function Types}.
3281 Pascal function parameter, @xref{Function Types}
3284 Builtin floating point type, @xref{Builtin Type Descriptors}.
3287 COBOL Group. See AIX documentation for details.
3290 Imported type, @xref{Cross-references}.
3293 Const-qualified type, @xref{Miscellaneous Types}.
3296 COBOL File Descriptor. See AIX documentation for details.
3299 Multiple instance type, @xref{Miscellaneous Types}.
3302 String type, @xref{Strings}.
3305 Stringptr, @xref{Strings}.
3308 Opaque type, @xref{Typedefs}.
3311 Procedure, @xref{Function Types}.
3314 Packed array, @xref{Arrays}.
3317 Range type, @xref{Subranges}.
3320 Builtin floating type, @xref{Builtin Type Descriptors} (Sun). Pascal
3321 subroutine parameter, @xref{Function Types} (AIX). Detecting this
3322 conflict is possible with careful parsing (hint: a Pascal subroutine
3323 parameter type will always contain a comma, and a builtin type
3324 descriptor never will).
3327 Structure type, @xref{Structures}.
3330 Set type, @xref{Miscellaneous Types}.
3333 Union, @xref{Unions}.
3336 Variant record. This is a Pascal and Modula-2 feature which is like a
3337 union within a struct in C. See AIX documentation for details.
3340 Wide character, @xref{Builtin Type Descriptors}.
3343 Cross-reference, @xref{Cross-references}.
3346 gstring, @xref{Strings}.
3349 @node Expanded reference
3350 @appendix Expanded reference by stab type.
3352 @c FIXME: This appendix should go away, see N_PSYM or N_SO for an example.
3354 For a full list of stab types, and cross-references to where they are
3355 described, @xref{Stab Types}. This appendix just duplicates certain
3356 information from the main body of this document; eventually the
3357 information will all be in one place.
3361 The first line is the symbol type expressed in decimal, hexadecimal,
3362 and as a #define (see devo/include/aout/stab.def).
3364 The second line describes the language constructs the symbol type
3367 The third line is the stab format with the significant stab fields
3368 named and the rest NIL.
3370 Subsequent lines expand upon the meaning and possible values for each
3371 significant stab field. # stands in for the type descriptor.
3373 Finally, any further information.
3376 * N_GSYM:: Global variable
3377 * N_FNAME:: Function name (BSD Fortran)
3378 * N_PC:: Pascal global symbol
3379 * N_NSYMS:: Number of symbols
3380 * N_NOMAP:: No DST map
3381 * N_RSYM:: Register variable
3382 * N_M2C:: Modula-2 compilation unit
3383 * N_BROWS:: Path to .cb file for Sun source code browser
3384 * N_DEFD:: GNU Modula2 definition module dependency
3385 * N_EHDECL:: GNU C++ exception variable
3386 * N_MOD2:: Modula2 information "for imc"
3387 * N_CATCH:: GNU C++ "catch" clause
3388 * N_SSYM:: Structure or union element
3389 * N_LSYM:: Automatic variable
3390 * N_ENTRY:: Alternate entry point
3391 * N_SCOPE:: Modula2 scope information (Sun only)
3392 * Gould:: non-base register symbols used on Gould systems
3393 * N_LENG:: Length of preceding entry
3397 @section 32 - 0x20 - N_GYSM
3402 .stabs "name", N_GSYM, NIL, NIL, NIL
3406 "name" -> "symbol_name:#type"
3410 Only the "name" field is significant. The location of the variable is
3411 obtained from the corresponding external symbol.
3414 @section 34 - 0x22 - N_FNAME
3415 Function name (for BSD Fortran)
3418 .stabs "name", N_FNAME, NIL, NIL, NIL
3422 "name" -> "function_name"
3425 Only the "name" field is significant. The location of the symbol is
3426 obtained from the corresponding extern symbol.
3429 @section 48 - 0x30 - N_PC
3430 Global symbol (for Pascal)
3433 .stabs "name", N_PC, NIL, NIL, value
3437 "name" -> "symbol_name" <<?>>
3438 value -> supposedly the line number (stab.def is skeptical)
3444 global pascal symbol: name,,0,subtype,line
3449 @section 50 - 0x32 - N_NSYMS
3450 Number of symbols (according to Ultrix V4.0)
3453 0, files,,funcs,lines (stab.def)
3457 @section 52 - 0x34 - N_NOMAP
3458 no DST map for sym (according to Ultrix V4.0)
3461 name, ,0,type,ignored (stab.def)
3465 @section 64 - 0x40 - N_RSYM
3469 .stabs "name:type",N_RSYM,0,RegSize,RegNumber (Sun doc)
3473 @section 66 - 0x42 - N_M2C
3474 Modula-2 compilation unit
3477 .stabs "name", N_M2C, 0, desc, value
3481 "name" -> "unit_name,unit_time_stamp[,code_time_stamp]
3483 value -> 0 (main unit)
3488 @section 72 - 0x48 - N_BROWS
3489 Sun source code browser, path to .cb file
3492 "path to associated .cb file"
3494 Note: type field value overlaps with N_BSLINE
3497 @section 74 - 0x4a - N_DEFD
3498 GNU Modula2 definition module dependency
3500 GNU Modula-2 definition module dependency. Value is the modification
3501 time of the definition file. Other is non-zero if it is imported with
3502 the GNU M2 keyword %INITIALIZE. Perhaps N_M2C can be used if there
3503 are enough empty fields?
3506 @section 80 - 0x50 - N_EHDECL
3507 GNU C++ exception variable <<?>>
3509 "name is variable name"
3511 Note: conflicts with N_MOD2.
3514 @section 80 - 0x50 - N_MOD2
3515 Modula2 info "for imc" (according to Ultrix V4.0)
3517 Note: conflicts with N_EHDECL <<?>>
3520 @section 84 - 0x54 - N_CATCH
3521 GNU C++ "catch" clause
3523 GNU C++ `catch' clause. Value is its address. Desc is nonzero if
3524 this entry is immediately followed by a CAUGHT stab saying what
3525 exception was caught. Multiple CAUGHT stabs means that multiple
3526 exceptions can be caught here. If Desc is 0, it means all exceptions
3530 @section 96 - 0x60 - N_SSYM
3531 Structure or union element
3533 Value is offset in the structure.
3535 <<?looking at structs and unions in C I didn't see these>>
3538 @section 128 - 0x80 - N_LSYM
3539 Automatic var in the stack (also used for type descriptors.)
3542 .stabs "name" N_LSYM, NIL, NIL, value
3546 @exdent @emph{For stack based local variables:}
3548 "name" -> name of the variable
3549 value -> offset from frame pointer (negative)
3551 @exdent @emph{For type descriptors:}
3553 "name" -> "name_of_the_type:#type"
3556 type -> type_ref (or) type_def
3558 type_ref -> type_number
3559 type_def -> type_number=type_desc etc.
3562 Type may be either a type reference or a type definition. A type
3563 reference is a number that refers to a previously defined type. A
3564 type definition is the number that will refer to this type, followed
3565 by an equals sign, a type descriptor and the additional data that
3566 defines the type. See the Table D for type descriptors and the
3567 section on types for what data follows each type descriptor.
3570 @section 164 - 0xa4 - N_ENTRY
3572 Alternate entry point.
3573 Value is its address.
3577 @section 196 - 0xc4 - N_SCOPE
3579 Modula2 scope information (Sun linker)
3583 @section Non-base registers on Gould systems
3585 These are used on Gould systems for non-base registers syms.
3587 However, the following values are not the values used by Gould; they are
3588 the values which GNU has been documenting for these values for a long
3589 time, without actually checking what Gould uses. I include these values
3590 only because perhaps some someone actually did something with the GNU
3591 information (I hope not, why GNU knowingly assigned wrong values to
3592 these in the header file is a complete mystery to me).
3595 240 0xf0 N_NBTEXT ??
3596 242 0xf2 N_NBDATA ??
3603 @section - 0xfe - N_LENG
3605 Second symbol entry containing a length-value for the preceding entry.
3606 The value is the length.
3609 @appendix Questions and anomalies
3613 For GNU C stabs defining local and global variables (N_LSYM and
3614 N_GSYM), the desc field is supposed to contain the source line number
3615 on which the variable is defined. In reality the desc field is always
3616 0. (This behavour is defined in dbxout.c and putting a line number in
3617 desc is controlled by #ifdef WINNING_GDB which defaults to false). Gdb
3618 supposedly uses this information if you say 'list var'. In reality
3619 var can be a variable defined in the program and gdb says `function
3623 In GNU C stabs there seems to be no way to differentiate tag types:
3624 structures, unions, and enums (symbol descriptor T) and typedefs
3625 (symbol descriptor t) defined at file scope from types defined locally
3626 to a procedure or other more local scope. They all use the N_LSYM
3627 stab type. Types defined at procedure scope are emited after the
3628 N_RBRAC of the preceding function and before the code of the
3629 procedure in which they are defined. This is exactly the same as
3630 types defined in the source file between the two procedure bodies.
3631 GDB overcompensates by placing all types in block #1, the block for
3632 symbols of file scope. This is true for default, -ansi and
3633 -traditional compiler options. (Bugs gcc/1063, gdb/1066.)
3636 What ends the procedure scope? Is it the proc block's N_RBRAC or the
3637 next N_FUN? (I believe its the first.)
3640 @c FIXME: This should go with the other stuff about global variables.
3641 Global variable stabs don't have location information. This comes
3642 from the external symbol for the same variable. The external symbol
3643 has a leading underbar on the _name of the variable and the stab does
3644 not. How do we know these two symbol table entries are talking about
3645 the same symbol when their names are different? (Answer: the debugger
3646 knows that external symbols have leading underbars).
3648 @c FIXME: This is absurdly vague; there all kinds of differences, some
3649 @c of which are the same between gnu & sun, and some of which aren't.
3651 Can gcc be configured to output stabs the way the Sun compiler
3652 does, so that their native debugging tools work? <NO?> It doesn't by
3653 default. GDB reads either format of stab. (gcc or SunC). How about
3657 @node xcoff-differences
3658 @appendix Differences between GNU stabs in a.out and GNU stabs in xcoff
3660 @c FIXME: Merge *all* these into the main body of the document.
3661 (The AIX/RS6000 native object file format is xcoff with stabs). This
3662 appendix only covers those differences which are not covered in the main
3663 body of this document.
3667 BSD a.out stab types correspond to AIX xcoff storage classes. In general the
3668 mapping is N_STABTYPE becomes C_STABTYPE. Some stab types in a.out
3669 are not supported in xcoff. See Table E. for full mappings.
3671 @c FIXME: Get C_* types for the block, figure out whether it is always
3672 @c used (I suspect not), explain clearly, and move to node Statics.
3674 initialised static N_STSYM and un-initialized static N_LCSYM both map
3675 to the C_STSYM storage class. But the destinction is preserved
3676 because in xcoff N_STSYM and N_LCSYM must be emited in a named static
3677 block. Begin the block with .bs s[RW] data_section_name for N_STSYM
3678 or .bs s bss_section_name for N_LCSYM. End the block with .es
3680 @c FIXME: I think they are trying to say something about whether the
3681 @c assembler defaults the value to the location counter.
3683 If the xcoff stab is a N_FUN (C_FUN) then follow the string field with
3684 ,. instead of just ,
3687 (I think that's it for .s file differences. They could stand to be
3688 better presented. This is just a list of what I have noticed so far.
3689 There are a *lot* of differences in the information in the symbol
3690 tables of the executable and object files.)
3692 Table E: mapping a.out stab types to xcoff storage classes
3695 stab type storage class
3696 -------------------------------
3705 N_RPSYM (0x8e) C_RPSYM
3715 N_DECL (0x8c) C_DECL
3732 @node Sun-differences
3733 @appendix Differences between GNU stabs and Sun native stabs.
3735 @c FIXME: Merge all this stuff into the main body of the document.
3739 GNU C stabs define *all* types, file or procedure scope, as
3740 N_LSYM. Sun doc talks about using N_GSYM too.
3743 Sun C stabs use type number pairs in the format (a,b) where a is a
3744 number starting with 1 and incremented for each sub-source file in the
3745 compilation. b is a number starting with 1 and incremented for each
3746 new type defined in the compilation. GNU C stabs use the type number
3747 alone, with no source file number.
3751 @appendix Using stabs with the ELF object file format.
3753 The ELF object file format allows tools to create object files with custom
3754 sections containing any arbitrary data. To use stabs in ELF object files,
3755 the tools create two custom sections, a ".stab" section which contains
3756 an array of fixed length structures, one struct per stab, and a ".stabstr"
3757 section containing all the variable length strings that are referenced by
3758 stabs in the ".stab" section. The byte order of the stabs binary data
3759 matches the byte order of the ELF file itself, as determined from the
3760 EI_DATA field in the e_ident member of the ELF header.
3762 The first stab in the ".stab" section for each object file is a "synthetic
3763 stab", generated entirely by the assembler, with no corresponding ".stab"
3764 directive as input to the assembler. This stab contains the following
3769 Offset in the ".stabstr" section to the source filename.
3775 Unused field, always zero.
3778 Count of upcoming symbols. I.E. the number of remaining stabs for this
3782 Size of the string table fragment associated with this object module, in
3787 The ".stabstr" section always starts with a null byte (so that string
3788 offsets of zero reference a null string), followed by random length strings,
3789 each of which is null byte terminated.
3791 The ELF section header for the ".stab" section has it's sh_link member set
3792 to the section number of the ".stabstr" section, and the ".stabstr" section
3793 has it's ELF section header sh_type member set to SHT_STRTAB to mark it as