2 @setfilename stabs.info
9 * Stabs: (stabs). The "stabs" debugging information format.
15 This document describes the stabs debugging symbol tables.
17 Copyright 1992,1993,1994,1995,1997,1998,2000,2001
18 Free Software Foundation, Inc.
19 Contributed by Cygnus Support. Written by Julia Menapace, Jim Kingdon,
22 Permission is granted to copy, distribute and/or modify this document
23 under the terms of the GNU Free Documentation License, Version 1.1 or
24 any later version published by the Free Software Foundation; with no
25 Invariant Sections, with no Front-Cover Texts, and with no Back-Cover
26 Texts. A copy of the license is included in the section entitled ``GNU
27 Free Documentation License''.
30 @setchapternewpage odd
33 @title The ``stabs'' debug format
34 @author Julia Menapace, Jim Kingdon, David MacKenzie
35 @author Cygnus Support
38 \def\$#1${{#1}} % Kluge: collect RCS revision info without $...$
39 \xdef\manvers{\$Revision$} % For use in headers, footers too
41 \hfill Cygnus Support\par
43 \hfill \TeX{}info \texinfoversion\par
47 @vskip 0pt plus 1filll
48 Copyright @copyright{} 1992,1993,1994,1995,1997,1998,2000,2001 Free Software Foundation, Inc.
49 Contributed by Cygnus Support.
51 Permission is granted to copy, distribute and/or modify this document
52 under the terms of the GNU Free Documentation License, Version 1.1 or
53 any later version published by the Free Software Foundation; with no
54 Invariant Sections, with no Front-Cover Texts, and with no Back-Cover
55 Texts. A copy of the license is included in the section entitled ``GNU
56 Free Documentation License''.
61 @top The "stabs" representation of debugging information
63 This document describes the stabs debugging format.
66 * Overview:: Overview of stabs
67 * Program Structure:: Encoding of the structure of the program
68 * Constants:: Constants
70 * Types:: Type definitions
71 * Symbol Tables:: Symbol information in symbol tables
72 * Cplusplus:: Stabs specific to C++
73 * Stab Types:: Symbol types in a.out files
74 * Symbol Descriptors:: Table of symbol descriptors
75 * Type Descriptors:: Table of type descriptors
76 * Expanded Reference:: Reference information by stab type
77 * Questions:: Questions and anomalies
78 * Stab Sections:: In some object file formats, stabs are
80 * Symbol Types Index:: Index of symbolic stab symbol type names.
81 * GNU Free Documentation License:: The license for this documentation
85 @c TeX can handle the contents at the start but makeinfo 3.12 can not
91 @chapter Overview of Stabs
93 @dfn{Stabs} refers to a format for information that describes a program
94 to a debugger. This format was apparently invented by
96 the University of California at Berkeley, for the @code{pdx} Pascal
97 debugger; the format has spread widely since then.
99 This document is one of the few published sources of documentation on
100 stabs. It is believed to be comprehensive for stabs used by C. The
101 lists of symbol descriptors (@pxref{Symbol Descriptors}) and type
102 descriptors (@pxref{Type Descriptors}) are believed to be completely
103 comprehensive. Stabs for COBOL-specific features and for variant
104 records (used by Pascal and Modula-2) are poorly documented here.
106 @c FIXME: Need to document all OS9000 stuff in GDB; see all references
107 @c to os9k_stabs in stabsread.c.
109 Other sources of information on stabs are @cite{Dbx and Dbxtool
110 Interfaces}, 2nd edition, by Sun, 1988, and @cite{AIX Version 3.2 Files
111 Reference}, Fourth Edition, September 1992, "dbx Stabstring Grammar" in
112 the a.out section, page 2-31. This document is believed to incorporate
113 the information from those two sources except where it explicitly directs
114 you to them for more information.
117 * Flow:: Overview of debugging information flow
118 * Stabs Format:: Overview of stab format
119 * String Field:: The string field
120 * C Example:: A simple example in C source
121 * Assembly Code:: The simple example at the assembly level
125 @section Overview of Debugging Information Flow
127 The GNU C compiler compiles C source in a @file{.c} file into assembly
128 language in a @file{.s} file, which the assembler translates into
129 a @file{.o} file, which the linker combines with other @file{.o} files and
130 libraries to produce an executable file.
132 With the @samp{-g} option, GCC puts in the @file{.s} file additional
133 debugging information, which is slightly transformed by the assembler
134 and linker, and carried through into the final executable. This
135 debugging information describes features of the source file like line
136 numbers, the types and scopes of variables, and function names,
137 parameters, and scopes.
139 For some object file formats, the debugging information is encapsulated
140 in assembler directives known collectively as @dfn{stab} (symbol table)
141 directives, which are interspersed with the generated code. Stabs are
142 the native format for debugging information in the a.out and XCOFF
143 object file formats. The GNU tools can also emit stabs in the COFF and
144 ECOFF object file formats.
146 The assembler adds the information from stabs to the symbol information
147 it places by default in the symbol table and the string table of the
148 @file{.o} file it is building. The linker consolidates the @file{.o}
149 files into one executable file, with one symbol table and one string
150 table. Debuggers use the symbol and string tables in the executable as
151 a source of debugging information about the program.
154 @section Overview of Stab Format
156 There are three overall formats for stab assembler directives,
157 differentiated by the first word of the stab. The name of the directive
158 describes which combination of four possible data fields follows. It is
159 either @code{.stabs} (string), @code{.stabn} (number), or @code{.stabd}
160 (dot). IBM's XCOFF assembler uses @code{.stabx} (and some other
161 directives such as @code{.file} and @code{.bi}) instead of
162 @code{.stabs}, @code{.stabn} or @code{.stabd}.
164 The overall format of each class of stab is:
167 .stabs "@var{string}",@var{type},@var{other},@var{desc},@var{value}
168 .stabn @var{type},@var{other},@var{desc},@var{value}
169 .stabd @var{type},@var{other},@var{desc}
170 .stabx "@var{string}",@var{value},@var{type},@var{sdb-type}
173 @c what is the correct term for "current file location"? My AIX
174 @c assembler manual calls it "the value of the current location counter".
175 For @code{.stabn} and @code{.stabd}, there is no @var{string} (the
176 @code{n_strx} field is zero; see @ref{Symbol Tables}). For
177 @code{.stabd}, the @var{value} field is implicit and has the value of
178 the current file location. For @code{.stabx}, the @var{sdb-type} field
179 is unused for stabs and can always be set to zero. The @var{other}
180 field is almost always unused and can be set to zero.
182 The number in the @var{type} field gives some basic information about
183 which type of stab this is (or whether it @emph{is} a stab, as opposed
184 to an ordinary symbol). Each valid type number defines a different stab
185 type; further, the stab type defines the exact interpretation of, and
186 possible values for, any remaining @var{string}, @var{desc}, or
187 @var{value} fields present in the stab. @xref{Stab Types}, for a list
188 in numeric order of the valid @var{type} field values for stab directives.
191 @section The String Field
193 For most stabs the string field holds the meat of the
194 debugging information. The flexible nature of this field
195 is what makes stabs extensible. For some stab types the string field
196 contains only a name. For other stab types the contents can be a great
199 The overall format of the string field for most stab types is:
202 "@var{name}:@var{symbol-descriptor} @var{type-information}"
205 @var{name} is the name of the symbol represented by the stab; it can
206 contain a pair of colons (@pxref{Nested Symbols}). @var{name} can be
207 omitted, which means the stab represents an unnamed object. For
208 example, @samp{:t10=*2} defines type 10 as a pointer to type 2, but does
209 not give the type a name. Omitting the @var{name} field is supported by
210 AIX dbx and GDB after about version 4.8, but not other debuggers. GCC
211 sometimes uses a single space as the name instead of omitting the name
212 altogether; apparently that is supported by most debuggers.
214 The @var{symbol-descriptor} following the @samp{:} is an alphabetic
215 character that tells more specifically what kind of symbol the stab
216 represents. If the @var{symbol-descriptor} is omitted, but type
217 information follows, then the stab represents a local variable. For a
218 list of symbol descriptors, see @ref{Symbol Descriptors}. The @samp{c}
219 symbol descriptor is an exception in that it is not followed by type
220 information. @xref{Constants}.
222 @var{type-information} is either a @var{type-number}, or
223 @samp{@var{type-number}=}. A @var{type-number} alone is a type
224 reference, referring directly to a type that has already been defined.
226 The @samp{@var{type-number}=} form is a type definition, where the
227 number represents a new type which is about to be defined. The type
228 definition may refer to other types by number, and those type numbers
229 may be followed by @samp{=} and nested definitions. Also, the Lucid
230 compiler will repeat @samp{@var{type-number}=} more than once if it
231 wants to define several type numbers at once.
233 In a type definition, if the character that follows the equals sign is
234 non-numeric then it is a @var{type-descriptor}, and tells what kind of
235 type is about to be defined. Any other values following the
236 @var{type-descriptor} vary, depending on the @var{type-descriptor}.
237 @xref{Type Descriptors}, for a list of @var{type-descriptor} values. If
238 a number follows the @samp{=} then the number is a @var{type-reference}.
239 For a full description of types, @ref{Types}.
241 A @var{type-number} is often a single number. The GNU and Sun tools
242 additionally permit a @var{type-number} to be a pair
243 (@var{file-number},@var{filetype-number}) (the parentheses appear in the
244 string, and serve to distinguish the two cases). The @var{file-number}
245 is 0 for the base source file, 1 for the first included file, 2 for the
246 next, and so on. The @var{filetype-number} is a number starting with
247 1 which is incremented for each new type defined in the file.
248 (Separating the file number and the type number permits the
249 @code{N_BINCL} optimization to succeed more often; see @ref{Include
252 There is an AIX extension for type attributes. Following the @samp{=}
253 are any number of type attributes. Each one starts with @samp{@@} and
254 ends with @samp{;}. Debuggers, including AIX's dbx and GDB 4.10, skip
255 any type attributes they do not recognize. GDB 4.9 and other versions
256 of dbx may not do this. Because of a conflict with C++
257 (@pxref{Cplusplus}), new attributes should not be defined which begin
258 with a digit, @samp{(}, or @samp{-}; GDB may be unable to distinguish
259 those from the C++ type descriptor @samp{@@}. The attributes are:
262 @item a@var{boundary}
263 @var{boundary} is an integer specifying the alignment. I assume it
264 applies to all variables of this type.
267 Pointer class (for checking). Not sure what this means, or how
268 @var{integer} is interpreted.
271 Indicate this is a packed type, meaning that structure fields or array
272 elements are placed more closely in memory, to save memory at the
276 Size in bits of a variable of this type. This is fully supported by GDB
280 Indicate that this type is a string instead of an array of characters,
281 or a bitstring instead of a set. It doesn't change the layout of the
282 data being represented, but does enable the debugger to know which type
286 Indicate that this type is a vector instead of an array. The only
287 major difference between vectors and arrays is that vectors are
288 passed by value instead of by reference (vector coprocessor extension).
292 All of this can make the string field quite long. All versions of GDB,
293 and some versions of dbx, can handle arbitrarily long strings. But many
294 versions of dbx (or assemblers or linkers, I'm not sure which)
295 cretinously limit the strings to about 80 characters, so compilers which
296 must work with such systems need to split the @code{.stabs} directive
297 into several @code{.stabs} directives. Each stab duplicates every field
298 except the string field. The string field of every stab except the last
299 is marked as continued with a backslash at the end (in the assembly code
300 this may be written as a double backslash, depending on the assembler).
301 Removing the backslashes and concatenating the string fields of each
302 stab produces the original, long string. Just to be incompatible (or so
303 they don't have to worry about what the assembler does with
304 backslashes), AIX can use @samp{?} instead of backslash.
307 @section A Simple Example in C Source
309 To get the flavor of how stabs describe source information for a C
310 program, let's look at the simple program:
315 printf("Hello world");
319 When compiled with @samp{-g}, the program above yields the following
320 @file{.s} file. Line numbers have been added to make it easier to refer
321 to parts of the @file{.s} file in the description of the stabs that
325 @section The Simple Example at the Assembly Level
327 This simple ``hello world'' example demonstrates several of the stab
328 types used to describe C language source files.
332 2 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0
333 3 .stabs "hello.c",100,0,0,Ltext0
336 6 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0
337 7 .stabs "char:t2=r2;0;127;",128,0,0,0
338 8 .stabs "long int:t3=r1;-2147483648;2147483647;",128,0,0,0
339 9 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
340 10 .stabs "long unsigned int:t5=r1;0;-1;",128,0,0,0
341 11 .stabs "short int:t6=r1;-32768;32767;",128,0,0,0
342 12 .stabs "long long int:t7=r1;0;-1;",128,0,0,0
343 13 .stabs "short unsigned int:t8=r1;0;65535;",128,0,0,0
344 14 .stabs "long long unsigned int:t9=r1;0;-1;",128,0,0,0
345 15 .stabs "signed char:t10=r1;-128;127;",128,0,0,0
346 16 .stabs "unsigned char:t11=r1;0;255;",128,0,0,0
347 17 .stabs "float:t12=r1;4;0;",128,0,0,0
348 18 .stabs "double:t13=r1;8;0;",128,0,0,0
349 19 .stabs "long double:t14=r1;8;0;",128,0,0,0
350 20 .stabs "void:t15=15",128,0,0,0
353 23 .ascii "Hello, world!\12\0"
368 38 sethi %hi(LC0),%o1
369 39 or %o1,%lo(LC0),%o0
380 50 .stabs "main:F1",36,0,0,_main
381 51 .stabn 192,0,0,LBB2
382 52 .stabn 224,0,0,LBE2
385 @node Program Structure
386 @chapter Encoding the Structure of the Program
388 The elements of the program structure that stabs encode include the name
389 of the main function, the names of the source and include files, the
390 line numbers, procedure names and types, and the beginnings and ends of
394 * Main Program:: Indicate what the main program is
395 * Source Files:: The path and name of the source file
396 * Include Files:: Names of include files
399 * Nested Procedures::
401 * Alternate Entry Points:: Entering procedures except at the beginning.
405 @section Main Program
408 Most languages allow the main program to have any name. The
409 @code{N_MAIN} stab type tells the debugger the name that is used in this
410 program. Only the string field is significant; it is the name of
411 a function which is the main program. Most C compilers do not use this
412 stab (they expect the debugger to assume that the name is @code{main}),
413 but some C compilers emit an @code{N_MAIN} stab for the @code{main}
414 function. I'm not sure how XCOFF handles this.
417 @section Paths and Names of the Source Files
420 Before any other stabs occur, there must be a stab specifying the source
421 file. This information is contained in a symbol of stab type
422 @code{N_SO}; the string field contains the name of the file. The
423 value of the symbol is the start address of the portion of the
424 text section corresponding to that file.
426 With the Sun Solaris2 compiler, the desc field contains a
427 source-language code.
428 @c Do the debuggers use it? What are the codes? -djm
430 Some compilers (for example, GCC2 and SunOS4 @file{/bin/cc}) also
431 include the directory in which the source was compiled, in a second
432 @code{N_SO} symbol preceding the one containing the file name. This
433 symbol can be distinguished by the fact that it ends in a slash. Code
434 from the @code{cfront} C++ compiler can have additional @code{N_SO} symbols for
435 nonexistent source files after the @code{N_SO} for the real source file;
436 these are believed to contain no useful information.
441 .stabs "/cygint/s1/users/jcm/play/",100,0,0,Ltext0 # @r{100 is N_SO}
442 .stabs "hello.c",100,0,0,Ltext0
448 Instead of @code{N_SO} symbols, XCOFF uses a @code{.file} assembler
449 directive which assembles to a @code{C_FILE} symbol; explaining this in
450 detail is outside the scope of this document.
452 @c FIXME: Exactly when should the empty N_SO be used? Why?
453 If it is useful to indicate the end of a source file, this is done with
454 an @code{N_SO} symbol with an empty string for the name. The value is
455 the address of the end of the text section for the file. For some
456 systems, there is no indication of the end of a source file, and you
457 just need to figure it ended when you see an @code{N_SO} for a different
458 source file, or a symbol ending in @code{.o} (which at least some
459 linkers insert to mark the start of a new @code{.o} file).
462 @section Names of Include Files
464 There are several schemes for dealing with include files: the
465 traditional @code{N_SOL} approach, Sun's @code{N_BINCL} approach, and the
466 XCOFF @code{C_BINCL} approach (which despite the similar name has little in
467 common with @code{N_BINCL}).
470 An @code{N_SOL} symbol specifies which include file subsequent symbols
471 refer to. The string field is the name of the file and the value is the
472 text address corresponding to the end of the previous include file and
473 the start of this one. To specify the main source file again, use an
474 @code{N_SOL} symbol with the name of the main source file.
479 The @code{N_BINCL} approach works as follows. An @code{N_BINCL} symbol
480 specifies the start of an include file. In an object file, only the
481 string is significant; the linker puts data into some of the other
482 fields. The end of the include file is marked by an @code{N_EINCL}
483 symbol (which has no string field). In an object file, there is no
484 significant data in the @code{N_EINCL} symbol. @code{N_BINCL} and
485 @code{N_EINCL} can be nested.
487 If the linker detects that two source files have identical stabs between
488 an @code{N_BINCL} and @code{N_EINCL} pair (as will generally be the case
489 for a header file), then it only puts out the stabs once. Each
490 additional occurrence is replaced by an @code{N_EXCL} symbol. I believe
491 the GNU linker and the Sun (both SunOS4 and Solaris) linker are the only
492 ones which supports this feature.
494 A linker which supports this feature will set the value of a
495 @code{N_BINCL} symbol to the total of all the characters in the stabs
496 strings included in the header file, omitting any file numbers. The
497 value of an @code{N_EXCL} symbol is the same as the value of the
498 @code{N_BINCL} symbol it replaces. This information can be used to
499 match up @code{N_EXCL} and @code{N_BINCL} symbols which have the same
500 filename. The @code{N_EINCL} value, and the values of the other and
501 description fields for all three, appear to always be zero.
505 For the start of an include file in XCOFF, use the @file{.bi} assembler
506 directive, which generates a @code{C_BINCL} symbol. A @file{.ei}
507 directive, which generates a @code{C_EINCL} symbol, denotes the end of
508 the include file. Both directives are followed by the name of the
509 source file in quotes, which becomes the string for the symbol.
510 The value of each symbol, produced automatically by the assembler
511 and linker, is the offset into the executable of the beginning
512 (inclusive, as you'd expect) or end (inclusive, as you would not expect)
513 of the portion of the COFF line table that corresponds to this include
514 file. @code{C_BINCL} and @code{C_EINCL} do not nest.
517 @section Line Numbers
520 An @code{N_SLINE} symbol represents the start of a source line. The
521 desc field contains the line number and the value contains the code
522 address for the start of that source line. On most machines the address
523 is absolute; for stabs in sections (@pxref{Stab Sections}), it is
524 relative to the function in which the @code{N_SLINE} symbol occurs.
528 GNU documents @code{N_DSLINE} and @code{N_BSLINE} symbols for line
529 numbers in the data or bss segments, respectively. They are identical
530 to @code{N_SLINE} but are relocated differently by the linker. They
531 were intended to be used to describe the source location of a variable
532 declaration, but I believe that GCC2 actually puts the line number in
533 the desc field of the stab for the variable itself. GDB has been
534 ignoring these symbols (unless they contain a string field) since
537 For single source lines that generate discontiguous code, such as flow
538 of control statements, there may be more than one line number entry for
539 the same source line. In this case there is a line number entry at the
540 start of each code range, each with the same line number.
542 XCOFF does not use stabs for line numbers. Instead, it uses COFF line
543 numbers (which are outside the scope of this document). Standard COFF
544 line numbers cannot deal with include files, but in XCOFF this is fixed
545 with the @code{C_BINCL} method of marking include files (@pxref{Include
551 @findex N_FUN, for functions
553 @findex N_STSYM, for functions (Sun acc)
554 @findex N_GSYM, for functions (Sun acc)
555 All of the following stabs normally use the @code{N_FUN} symbol type.
556 However, Sun's @code{acc} compiler on SunOS4 uses @code{N_GSYM} and
557 @code{N_STSYM}, which means that the value of the stab for the function
558 is useless and the debugger must get the address of the function from
559 the non-stab symbols instead. On systems where non-stab symbols have
560 leading underscores, the stabs will lack underscores and the debugger
561 needs to know about the leading underscore to match up the stab and the
562 non-stab symbol. BSD Fortran is said to use @code{N_FNAME} with the
563 same restriction; the value of the symbol is not useful (I'm not sure it
564 really does use this, because GDB doesn't handle this and no one has
568 A function is represented by an @samp{F} symbol descriptor for a global
569 (extern) function, and @samp{f} for a static (local) function. For
570 a.out, the value of the symbol is the address of the start of the
571 function; it is already relocated. For stabs in ELF, the SunPRO
572 compiler version 2.0.1 and GCC put out an address which gets relocated
573 by the linker. In a future release SunPRO is planning to put out zero,
574 in which case the address can be found from the ELF (non-stab) symbol.
575 Because looking things up in the ELF symbols would probably be slow, I'm
576 not sure how to find which symbol of that name is the right one, and
577 this doesn't provide any way to deal with nested functions, it would
578 probably be better to make the value of the stab an address relative to
579 the start of the file, or just absolute. See @ref{ELF Linker
580 Relocation} for more information on linker relocation of stabs in ELF
581 files. For XCOFF, the stab uses the @code{C_FUN} storage class and the
582 value of the stab is meaningless; the address of the function can be
583 found from the csect symbol (XTY_LD/XMC_PR).
585 The type information of the stab represents the return type of the
586 function; thus @samp{foo:f5} means that foo is a function returning type
587 5. There is no need to try to get the line number of the start of the
588 function from the stab for the function; it is in the next
589 @code{N_SLINE} symbol.
591 @c FIXME: verify whether the "I suspect" below is true or not.
592 Some compilers (such as Sun's Solaris compiler) support an extension for
593 specifying the types of the arguments. I suspect this extension is not
594 used for old (non-prototyped) function definitions in C. If the
595 extension is in use, the type information of the stab for the function
596 is followed by type information for each argument, with each argument
597 preceded by @samp{;}. An argument type of 0 means that additional
598 arguments are being passed, whose types and number may vary (@samp{...}
599 in ANSI C). GDB has tolerated this extension (parsed the syntax, if not
600 necessarily used the information) since at least version 4.8; I don't
601 know whether all versions of dbx tolerate it. The argument types given
602 here are not redundant with the symbols for the formal parameters
603 (@pxref{Parameters}); they are the types of the arguments as they are
604 passed, before any conversions might take place. For example, if a C
605 function which is declared without a prototype takes a @code{float}
606 argument, the value is passed as a @code{double} but then converted to a
607 @code{float}. Debuggers need to use the types given in the arguments
608 when printing values, but when calling the function they need to use the
609 types given in the symbol defining the function.
611 If the return type and types of arguments of a function which is defined
612 in another source file are specified (i.e., a function prototype in ANSI
613 C), traditionally compilers emit no stab; the only way for the debugger
614 to find the information is if the source file where the function is
615 defined was also compiled with debugging symbols. As an extension the
616 Solaris compiler uses symbol descriptor @samp{P} followed by the return
617 type of the function, followed by the arguments, each preceded by
618 @samp{;}, as in a stab with symbol descriptor @samp{f} or @samp{F}.
619 This use of symbol descriptor @samp{P} can be distinguished from its use
620 for register parameters (@pxref{Register Parameters}) by the fact that it has
621 symbol type @code{N_FUN}.
623 The AIX documentation also defines symbol descriptor @samp{J} as an
624 internal function. I assume this means a function nested within another
625 function. It also says symbol descriptor @samp{m} is a module in
626 Modula-2 or extended Pascal.
628 Procedures (functions which do not return values) are represented as
629 functions returning the @code{void} type in C. I don't see why this couldn't
630 be used for all languages (inventing a @code{void} type for this purpose if
631 necessary), but the AIX documentation defines @samp{I}, @samp{P}, and
632 @samp{Q} for internal, global, and static procedures, respectively.
633 These symbol descriptors are unusual in that they are not followed by
636 The following example shows a stab for a function @code{main} which
637 returns type number @code{1}. The @code{_main} specified for the value
638 is a reference to an assembler label which is used to fill in the start
639 address of the function.
642 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
645 The stab representing a procedure is located immediately following the
646 code of the procedure. This stab is in turn directly followed by a
647 group of other stabs describing elements of the procedure. These other
648 stabs describe the procedure's parameters, its block local variables, and
651 If functions can appear in different sections, then the debugger may not
652 be able to find the end of a function. Recent versions of GCC will mark
653 the end of a function with an @code{N_FUN} symbol with an empty string
654 for the name. The value is the address of the end of the current
655 function. Without such a symbol, there is no indication of the address
656 of the end of a function, and you must assume that it ended at the
657 starting address of the next function or at the end of the text section
660 @node Nested Procedures
661 @section Nested Procedures
663 For any of the symbol descriptors representing procedures, after the
664 symbol descriptor and the type information is optionally a scope
665 specifier. This consists of a comma, the name of the procedure, another
666 comma, and the name of the enclosing procedure. The first name is local
667 to the scope specified, and seems to be redundant with the name of the
668 symbol (before the @samp{:}). This feature is used by GCC, and
669 presumably Pascal, Modula-2, etc., compilers, for nested functions.
671 If procedures are nested more than one level deep, only the immediately
672 containing scope is specified. For example, this code:
684 return baz (x + 2 * y);
686 return x + bar (3 * x);
694 .stabs "baz:f1,baz,bar",36,0,0,_baz.15 # @r{36 is N_FUN}
695 .stabs "bar:f1,bar,foo",36,0,0,_bar.12
696 .stabs "foo:F1",36,0,0,_foo
699 @node Block Structure
700 @section Block Structure
704 @c For GCC 2.5.8 or so stabs-in-coff, these are absolute instead of
705 @c function relative (as documented below). But GDB has never been able
706 @c to deal with that (it had wanted them to be relative to the file, but
707 @c I just fixed that (between GDB 4.12 and 4.13)), so it is function
708 @c relative just like ELF and SOM and the below documentation.
709 The program's block structure is represented by the @code{N_LBRAC} (left
710 brace) and the @code{N_RBRAC} (right brace) stab types. The variables
711 defined inside a block precede the @code{N_LBRAC} symbol for most
712 compilers, including GCC. Other compilers, such as the Convex, Acorn
713 RISC machine, and Sun @code{acc} compilers, put the variables after the
714 @code{N_LBRAC} symbol. The values of the @code{N_LBRAC} and
715 @code{N_RBRAC} symbols are the start and end addresses of the code of
716 the block, respectively. For most machines, they are relative to the
717 starting address of this source file. For the Gould NP1, they are
718 absolute. For stabs in sections (@pxref{Stab Sections}), they are
719 relative to the function in which they occur.
721 The @code{N_LBRAC} and @code{N_RBRAC} stabs that describe the block
722 scope of a procedure are located after the @code{N_FUN} stab that
723 represents the procedure itself.
725 Sun documents the desc field of @code{N_LBRAC} and
726 @code{N_RBRAC} symbols as containing the nesting level of the block.
727 However, dbx seems to not care, and GCC always sets desc to
733 For XCOFF, block scope is indicated with @code{C_BLOCK} symbols. If the
734 name of the symbol is @samp{.bb}, then it is the beginning of the block;
735 if the name of the symbol is @samp{.be}; it is the end of the block.
737 @node Alternate Entry Points
738 @section Alternate Entry Points
742 Some languages, like Fortran, have the ability to enter procedures at
743 some place other than the beginning. One can declare an alternate entry
744 point. The @code{N_ENTRY} stab is for this; however, the Sun FORTRAN
745 compiler doesn't use it. According to AIX documentation, only the name
746 of a @code{C_ENTRY} stab is significant; the address of the alternate
747 entry point comes from the corresponding external symbol. A previous
748 revision of this document said that the value of an @code{N_ENTRY} stab
749 was the address of the alternate entry point, but I don't know the
750 source for that information.
755 The @samp{c} symbol descriptor indicates that this stab represents a
756 constant. This symbol descriptor is an exception to the general rule
757 that symbol descriptors are followed by type information. Instead, it
758 is followed by @samp{=} and one of the following:
762 Boolean constant. @var{value} is a numeric value; I assume it is 0 for
766 Character constant. @var{value} is the numeric value of the constant.
768 @item e @var{type-information} , @var{value}
769 Constant whose value can be represented as integral.
770 @var{type-information} is the type of the constant, as it would appear
771 after a symbol descriptor (@pxref{String Field}). @var{value} is the
772 numeric value of the constant. GDB 4.9 does not actually get the right
773 value if @var{value} does not fit in a host @code{int}, but it does not
774 do anything violent, and future debuggers could be extended to accept
775 integers of any size (whether unsigned or not). This constant type is
776 usually documented as being only for enumeration constants, but GDB has
777 never imposed that restriction; I don't know about other debuggers.
780 Integer constant. @var{value} is the numeric value. The type is some
781 sort of generic integer type (for GDB, a host @code{int}); to specify
782 the type explicitly, use @samp{e} instead.
785 Real constant. @var{value} is the real value, which can be @samp{INF}
786 (optionally preceded by a sign) for infinity, @samp{QNAN} for a quiet
787 NaN (not-a-number), or @samp{SNAN} for a signalling NaN. If it is a
788 normal number the format is that accepted by the C library function
792 String constant. @var{string} is a string enclosed in either @samp{'}
793 (in which case @samp{'} characters within the string are represented as
794 @samp{\'} or @samp{"} (in which case @samp{"} characters within the
795 string are represented as @samp{\"}).
797 @item S @var{type-information} , @var{elements} , @var{bits} , @var{pattern}
798 Set constant. @var{type-information} is the type of the constant, as it
799 would appear after a symbol descriptor (@pxref{String Field}).
800 @var{elements} is the number of elements in the set (does this means
801 how many bits of @var{pattern} are actually used, which would be
802 redundant with the type, or perhaps the number of bits set in
803 @var{pattern}? I don't get it), @var{bits} is the number of bits in the
804 constant (meaning it specifies the length of @var{pattern}, I think),
805 and @var{pattern} is a hexadecimal representation of the set. AIX
806 documentation refers to a limit of 32 bytes, but I see no reason why
807 this limit should exist. This form could probably be used for arbitrary
808 constants, not just sets; the only catch is that @var{pattern} should be
809 understood to be target, not host, byte order and format.
812 The boolean, character, string, and set constants are not supported by
813 GDB 4.9, but it ignores them. GDB 4.8 and earlier gave an error
814 message and refused to read symbols from the file containing the
817 The above information is followed by @samp{;}.
822 Different types of stabs describe the various ways that variables can be
823 allocated: on the stack, globally, in registers, in common blocks,
824 statically, or as arguments to a function.
827 * Stack Variables:: Variables allocated on the stack.
828 * Global Variables:: Variables used by more than one source file.
829 * Register Variables:: Variables in registers.
830 * Common Blocks:: Variables statically allocated together.
831 * Statics:: Variables local to one source file.
832 * Based Variables:: Fortran pointer based variables.
833 * Parameters:: Variables for arguments to functions.
836 @node Stack Variables
837 @section Automatic Variables Allocated on the Stack
839 If a variable's scope is local to a function and its lifetime is only as
840 long as that function executes (C calls such variables
841 @dfn{automatic}), it can be allocated in a register (@pxref{Register
842 Variables}) or on the stack.
844 @findex N_LSYM, for stack variables
846 Each variable allocated on the stack has a stab with the symbol
847 descriptor omitted. Since type information should begin with a digit,
848 @samp{-}, or @samp{(}, only those characters precluded from being used
849 for symbol descriptors. However, the Acorn RISC machine (ARM) is said
850 to get this wrong: it puts out a mere type definition here, without the
851 preceding @samp{@var{type-number}=}. This is a bad idea; there is no
852 guarantee that type descriptors are distinct from symbol descriptors.
853 Stabs for stack variables use the @code{N_LSYM} stab type, or
854 @code{C_LSYM} for XCOFF.
856 The value of the stab is the offset of the variable within the
857 local variables. On most machines this is an offset from the frame
858 pointer and is negative. The location of the stab specifies which block
859 it is defined in; see @ref{Block Structure}.
861 For example, the following C code:
871 produces the following stabs:
874 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
875 .stabs "x:1",128,0,0,-12 # @r{128 is N_LSYM}
876 .stabn 192,0,0,LBB2 # @r{192 is N_LBRAC}
877 .stabn 224,0,0,LBE2 # @r{224 is N_RBRAC}
880 See @ref{Procedures} for more information on the @code{N_FUN} stab, and
881 @ref{Block Structure} for more information on the @code{N_LBRAC} and
882 @code{N_RBRAC} stabs.
884 @node Global Variables
885 @section Global Variables
889 @c FIXME: verify for sure that it really is C_GSYM on XCOFF
890 A variable whose scope is not specific to just one source file is
891 represented by the @samp{G} symbol descriptor. These stabs use the
892 @code{N_GSYM} stab type (C_GSYM for XCOFF). The type information for
893 the stab (@pxref{String Field}) gives the type of the variable.
895 For example, the following source code:
902 yields the following assembly code:
905 .stabs "g_foo:G2",32,0,0,0 # @r{32 is N_GSYM}
912 The address of the variable represented by the @code{N_GSYM} is not
913 contained in the @code{N_GSYM} stab. The debugger gets this information
914 from the external symbol for the global variable. In the example above,
915 the @code{.global _g_foo} and @code{_g_foo:} lines tell the assembler to
916 produce an external symbol.
918 Some compilers, like GCC, output @code{N_GSYM} stabs only once, where
919 the variable is defined. Other compilers, like SunOS4 /bin/cc, output a
920 @code{N_GSYM} stab for each compilation unit which references the
923 @node Register Variables
924 @section Register Variables
928 @c According to an old version of this manual, AIX uses C_RPSYM instead
929 @c of C_RSYM. I am skeptical; this should be verified.
930 Register variables have their own stab type, @code{N_RSYM}
931 (@code{C_RSYM} for XCOFF), and their own symbol descriptor, @samp{r}.
932 The stab's value is the number of the register where the variable data
934 @c .stabs "name:type",N_RSYM,0,RegSize,RegNumber (Sun doc)
936 AIX defines a separate symbol descriptor @samp{d} for floating point
937 registers. This seems unnecessary; why not just just give floating
938 point registers different register numbers? I have not verified whether
939 the compiler actually uses @samp{d}.
941 If the register is explicitly allocated to a global variable, but not
945 register int g_bar asm ("%g5");
949 then the stab may be emitted at the end of the object file, with
950 the other bss symbols.
953 @section Common Blocks
955 A common block is a statically allocated section of memory which can be
956 referred to by several source files. It may contain several variables.
957 I believe Fortran is the only language with this feature.
963 A @code{N_BCOMM} stab begins a common block and an @code{N_ECOMM} stab
964 ends it. The only field that is significant in these two stabs is the
965 string, which names a normal (non-debugging) symbol that gives the
966 address of the common block. According to IBM documentation, only the
967 @code{N_BCOMM} has the name of the common block (even though their
968 compiler actually puts it both places).
972 The stabs for the members of the common block are between the
973 @code{N_BCOMM} and the @code{N_ECOMM}; the value of each stab is the
974 offset within the common block of that variable. IBM uses the
975 @code{C_ECOML} stab type, and there is a corresponding @code{N_ECOML}
976 stab type, but Sun's Fortran compiler uses @code{N_GSYM} instead. The
977 variables within a common block use the @samp{V} symbol descriptor (I
978 believe this is true of all Fortran variables). Other stabs (at least
979 type declarations using @code{C_DECL}) can also be between the
980 @code{N_BCOMM} and the @code{N_ECOMM}.
983 @section Static Variables
985 Initialized static variables are represented by the @samp{S} and
986 @samp{V} symbol descriptors. @samp{S} means file scope static, and
987 @samp{V} means procedure scope static. One exception: in XCOFF, IBM's
988 xlc compiler always uses @samp{V}, and whether it is file scope or not
989 is distinguished by whether the stab is located within a function.
991 @c This is probably not worth mentioning; it is only true on the sparc
992 @c for `double' variables which although declared const are actually in
993 @c the data segment (the text segment can't guarantee 8 byte alignment).
995 @c 2.4.5 has a bug in that it uses @code{N_FUN}, so neither dbx nor GDB can
996 @c find the variables)
999 @findex N_FUN, for variables
1001 In a.out files, @code{N_STSYM} means the data section, @code{N_FUN}
1002 means the text section, and @code{N_LCSYM} means the bss section. For
1003 those systems with a read-only data section separate from the text
1004 section (Solaris), @code{N_ROSYM} means the read-only data section.
1006 For example, the source lines:
1009 static const int var_const = 5;
1010 static int var_init = 2;
1011 static int var_noinit;
1015 yield the following stabs:
1018 .stabs "var_const:S1",36,0,0,_var_const # @r{36 is N_FUN}
1020 .stabs "var_init:S1",38,0,0,_var_init # @r{38 is N_STSYM}
1022 .stabs "var_noinit:S1",40,0,0,_var_noinit # @r{40 is N_LCSYM}
1028 In XCOFF files, the stab type need not indicate the section;
1029 @code{C_STSYM} can be used for all statics. Also, each static variable
1030 is enclosed in a static block. A @code{C_BSTAT} (emitted with a
1031 @samp{.bs} assembler directive) symbol begins the static block; its
1032 value is the symbol number of the csect symbol whose value is the
1033 address of the static block, its section is the section of the variables
1034 in that static block, and its name is @samp{.bs}. A @code{C_ESTAT}
1035 (emitted with a @samp{.es} assembler directive) symbol ends the static
1036 block; its name is @samp{.es} and its value and section are ignored.
1038 In ECOFF files, the storage class is used to specify the section, so the
1039 stab type need not indicate the section.
1041 In ELF files, for the SunPRO compiler version 2.0.1, symbol descriptor
1042 @samp{S} means that the address is absolute (the linker relocates it)
1043 and symbol descriptor @samp{V} means that the address is relative to the
1044 start of the relevant section for that compilation unit. SunPRO has
1045 plans to have the linker stop relocating stabs; I suspect that their the
1046 debugger gets the address from the corresponding ELF (not stab) symbol.
1047 I'm not sure how to find which symbol of that name is the right one.
1048 The clean way to do all this would be to have a the value of a symbol
1049 descriptor @samp{S} symbol be an offset relative to the start of the
1050 file, just like everything else, but that introduces obvious
1051 compatibility problems. For more information on linker stab relocation,
1052 @xref{ELF Linker Relocation}.
1054 @node Based Variables
1055 @section Fortran Based Variables
1057 Fortran (at least, the Sun and SGI dialects of FORTRAN-77) has a feature
1058 which allows allocating arrays with @code{malloc}, but which avoids
1059 blurring the line between arrays and pointers the way that C does. In
1060 stabs such a variable uses the @samp{b} symbol descriptor.
1062 For example, the Fortran declarations
1065 real foo, foo10(10), foo10_5(10,5)
1067 pointer (foo10p, foo10)
1068 pointer (foo105p, foo10_5)
1076 foo10_5:bar3;1;5;ar3;1;10;6
1079 In this example, @code{real} is type 6 and type 3 is an integral type
1080 which is the type of the subscripts of the array (probably
1083 The @samp{b} symbol descriptor is like @samp{V} in that it denotes a
1084 statically allocated symbol whose scope is local to a function; see
1085 @xref{Statics}. The value of the symbol, instead of being the address
1086 of the variable itself, is the address of a pointer to that variable.
1087 So in the above example, the value of the @code{foo} stab is the address
1088 of a pointer to a real, the value of the @code{foo10} stab is the
1089 address of a pointer to a 10-element array of reals, and the value of
1090 the @code{foo10_5} stab is the address of a pointer to a 5-element array
1091 of 10-element arrays of reals.
1096 Formal parameters to a function are represented by a stab (or sometimes
1097 two; see below) for each parameter. The stabs are in the order in which
1098 the debugger should print the parameters (i.e., the order in which the
1099 parameters are declared in the source file). The exact form of the stab
1100 depends on how the parameter is being passed.
1104 Parameters passed on the stack use the symbol descriptor @samp{p} and
1105 the @code{N_PSYM} symbol type (or @code{C_PSYM} for XCOFF). The value
1106 of the symbol is an offset used to locate the parameter on the stack;
1107 its exact meaning is machine-dependent, but on most machines it is an
1108 offset from the frame pointer.
1110 As a simple example, the code:
1121 .stabs "main:F1",36,0,0,_main # @r{36 is N_FUN}
1122 .stabs "argc:p1",160,0,0,68 # @r{160 is N_PSYM}
1123 .stabs "argv:p20=*21=*2",160,0,0,72
1126 The type definition of @code{argv} is interesting because it contains
1127 several type definitions. Type 21 is pointer to type 2 (char) and
1128 @code{argv} (type 20) is pointer to type 21.
1130 @c FIXME: figure out what these mean and describe them coherently.
1131 The following symbol descriptors are also said to go with @code{N_PSYM}.
1132 The value of the symbol is said to be an offset from the argument
1133 pointer (I'm not sure whether this is true or not).
1137 pF Fortran function parameter
1138 X (function result variable)
1142 * Register Parameters::
1143 * Local Variable Parameters::
1144 * Reference Parameters::
1145 * Conformant Arrays::
1148 @node Register Parameters
1149 @subsection Passing Parameters in Registers
1151 If the parameter is passed in a register, then traditionally there are
1152 two symbols for each argument:
1155 .stabs "arg:p1" . . . ; N_PSYM
1156 .stabs "arg:r1" . . . ; N_RSYM
1159 Debuggers use the second one to find the value, and the first one to
1160 know that it is an argument.
1163 @findex N_RSYM, for parameters
1164 Because that approach is kind of ugly, some compilers use symbol
1165 descriptor @samp{P} or @samp{R} to indicate an argument which is in a
1166 register. Symbol type @code{C_RPSYM} is used in XCOFF and @code{N_RSYM}
1167 is used otherwise. The symbol's value is the register number. @samp{P}
1168 and @samp{R} mean the same thing; the difference is that @samp{P} is a
1169 GNU invention and @samp{R} is an IBM (XCOFF) invention. As of version
1170 4.9, GDB should handle either one.
1172 There is at least one case where GCC uses a @samp{p} and @samp{r} pair
1173 rather than @samp{P}; this is where the argument is passed in the
1174 argument list and then loaded into a register.
1176 According to the AIX documentation, symbol descriptor @samp{D} is for a
1177 parameter passed in a floating point register. This seems
1178 unnecessary---why not just use @samp{R} with a register number which
1179 indicates that it's a floating point register? I haven't verified
1180 whether the system actually does what the documentation indicates.
1182 @c FIXME: On the hppa this is for any type > 8 bytes, I think, and not
1183 @c for small structures (investigate).
1184 On the sparc and hppa, for a @samp{P} symbol whose type is a structure
1185 or union, the register contains the address of the structure. On the
1186 sparc, this is also true of a @samp{p} and @samp{r} pair (using Sun
1187 @code{cc}) or a @samp{p} symbol. However, if a (small) structure is
1188 really in a register, @samp{r} is used. And, to top it all off, on the
1189 hppa it might be a structure which was passed on the stack and loaded
1190 into a register and for which there is a @samp{p} and @samp{r} pair! I
1191 believe that symbol descriptor @samp{i} is supposed to deal with this
1192 case (it is said to mean "value parameter by reference, indirect
1193 access"; I don't know the source for this information), but I don't know
1194 details or what compilers or debuggers use it, if any (not GDB or GCC).
1195 It is not clear to me whether this case needs to be dealt with
1196 differently than parameters passed by reference (@pxref{Reference Parameters}).
1198 @node Local Variable Parameters
1199 @subsection Storing Parameters as Local Variables
1201 There is a case similar to an argument in a register, which is an
1202 argument that is actually stored as a local variable. Sometimes this
1203 happens when the argument was passed in a register and then the compiler
1204 stores it as a local variable. If possible, the compiler should claim
1205 that it's in a register, but this isn't always done.
1207 If a parameter is passed as one type and converted to a smaller type by
1208 the prologue (for example, the parameter is declared as a @code{float},
1209 but the calling conventions specify that it is passed as a
1210 @code{double}), then GCC2 (sometimes) uses a pair of symbols. The first
1211 symbol uses symbol descriptor @samp{p} and the type which is passed.
1212 The second symbol has the type and location which the parameter actually
1213 has after the prologue. For example, suppose the following C code
1214 appears with no prototypes involved:
1223 if @code{f} is passed as a double at stack offset 8, and the prologue
1224 converts it to a float in register number 0, then the stabs look like:
1227 .stabs "f:p13",160,0,3,8 # @r{160 is @code{N_PSYM}, here 13 is @code{double}}
1228 .stabs "f:r12",64,0,3,0 # @r{64 is @code{N_RSYM}, here 12 is @code{float}}
1231 In both stabs 3 is the line number where @code{f} is declared
1232 (@pxref{Line Numbers}).
1234 @findex N_LSYM, for parameter
1235 GCC, at least on the 960, has another solution to the same problem. It
1236 uses a single @samp{p} symbol descriptor for an argument which is stored
1237 as a local variable but uses @code{N_LSYM} instead of @code{N_PSYM}. In
1238 this case, the value of the symbol is an offset relative to the local
1239 variables for that function, not relative to the arguments; on some
1240 machines those are the same thing, but not on all.
1242 @c This is mostly just background info; the part that logically belongs
1243 @c here is the last sentence.
1244 On the VAX or on other machines in which the calling convention includes
1245 the number of words of arguments actually passed, the debugger (GDB at
1246 least) uses the parameter symbols to keep track of whether it needs to
1247 print nameless arguments in addition to the formal parameters which it
1248 has printed because each one has a stab. For example, in
1251 extern int fprintf (FILE *stream, char *format, @dots{});
1253 fprintf (stdout, "%d\n", x);
1256 there are stabs for @code{stream} and @code{format}. On most machines,
1257 the debugger can only print those two arguments (because it has no way
1258 of knowing that additional arguments were passed), but on the VAX or
1259 other machines with a calling convention which indicates the number of
1260 words of arguments, the debugger can print all three arguments. To do
1261 so, the parameter symbol (symbol descriptor @samp{p}) (not necessarily
1262 @samp{r} or symbol descriptor omitted symbols) needs to contain the
1263 actual type as passed (for example, @code{double} not @code{float} if it
1264 is passed as a double and converted to a float).
1266 @node Reference Parameters
1267 @subsection Passing Parameters by Reference
1269 If the parameter is passed by reference (e.g., Pascal @code{VAR}
1270 parameters), then the symbol descriptor is @samp{v} if it is in the
1271 argument list, or @samp{a} if it in a register. Other than the fact
1272 that these contain the address of the parameter rather than the
1273 parameter itself, they are identical to @samp{p} and @samp{R},
1274 respectively. I believe @samp{a} is an AIX invention; @samp{v} is
1275 supported by all stabs-using systems as far as I know.
1277 @node Conformant Arrays
1278 @subsection Passing Conformant Array Parameters
1280 @c Is this paragraph correct? It is based on piecing together patchy
1281 @c information and some guesswork
1282 Conformant arrays are a feature of Modula-2, and perhaps other
1283 languages, in which the size of an array parameter is not known to the
1284 called function until run-time. Such parameters have two stabs: a
1285 @samp{x} for the array itself, and a @samp{C}, which represents the size
1286 of the array. The value of the @samp{x} stab is the offset in the
1287 argument list where the address of the array is stored (it this right?
1288 it is a guess); the value of the @samp{C} stab is the offset in the
1289 argument list where the size of the array (in elements? in bytes?) is
1293 @chapter Defining Types
1295 The examples so far have described types as references to previously
1296 defined types, or defined in terms of subranges of or pointers to
1297 previously defined types. This chapter describes the other type
1298 descriptors that may follow the @samp{=} in a type definition.
1301 * Builtin Types:: Integers, floating point, void, etc.
1302 * Miscellaneous Types:: Pointers, sets, files, etc.
1303 * Cross-References:: Referring to a type not yet defined.
1304 * Subranges:: A type with a specific range.
1305 * Arrays:: An aggregate type of same-typed elements.
1306 * Strings:: Like an array but also has a length.
1307 * Enumerations:: Like an integer but the values have names.
1308 * Structures:: An aggregate type of different-typed elements.
1309 * Typedefs:: Giving a type a name.
1310 * Unions:: Different types sharing storage.
1315 @section Builtin Types
1317 Certain types are built in (@code{int}, @code{short}, @code{void},
1318 @code{float}, etc.); the debugger recognizes these types and knows how
1319 to handle them. Thus, don't be surprised if some of the following ways
1320 of specifying builtin types do not specify everything that a debugger
1321 would need to know about the type---in some cases they merely specify
1322 enough information to distinguish the type from other types.
1324 The traditional way to define builtin types is convoluted, so new ways
1325 have been invented to describe them. Sun's @code{acc} uses special
1326 builtin type descriptors (@samp{b} and @samp{R}), and IBM uses negative
1327 type numbers. GDB accepts all three ways, as of version 4.8; dbx just
1328 accepts the traditional builtin types and perhaps one of the other two
1329 formats. The following sections describe each of these formats.
1332 * Traditional Builtin Types:: Put on your seat belts and prepare for kludgery
1333 * Builtin Type Descriptors:: Builtin types with special type descriptors
1334 * Negative Type Numbers:: Builtin types using negative type numbers
1337 @node Traditional Builtin Types
1338 @subsection Traditional Builtin Types
1340 This is the traditional, convoluted method for defining builtin types.
1341 There are several classes of such type definitions: integer, floating
1342 point, and @code{void}.
1345 * Traditional Integer Types::
1346 * Traditional Other Types::
1349 @node Traditional Integer Types
1350 @subsubsection Traditional Integer Types
1352 Often types are defined as subranges of themselves. If the bounding values
1353 fit within an @code{int}, then they are given normally. For example:
1356 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0 # @r{128 is N_LSYM}
1357 .stabs "char:t2=r2;0;127;",128,0,0,0
1360 Builtin types can also be described as subranges of @code{int}:
1363 .stabs "unsigned short:t6=r1;0;65535;",128,0,0,0
1366 If the lower bound of a subrange is 0 and the upper bound is -1,
1367 the type is an unsigned integral type whose bounds are too
1368 big to describe in an @code{int}. Traditionally this is only used for
1369 @code{unsigned int} and @code{unsigned long}:
1372 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
1375 For larger types, GCC 2.4.5 puts out bounds in octal, with one or more
1376 leading zeroes. In this case a negative bound consists of a number
1377 which is a 1 bit (for the sign bit) followed by a 0 bit for each bit in
1378 the number (except the sign bit), and a positive bound is one which is a
1379 1 bit for each bit in the number (except possibly the sign bit). All
1380 known versions of dbx and GDB version 4 accept this (at least in the
1381 sense of not refusing to process the file), but GDB 3.5 refuses to read
1382 the whole file containing such symbols. So GCC 2.3.3 did not output the
1383 proper size for these types. As an example of octal bounds, the string
1384 fields of the stabs for 64 bit integer types look like:
1386 @c .stabs directives, etc., omitted to make it fit on the page.
1388 long int:t3=r1;001000000000000000000000;000777777777777777777777;
1389 long unsigned int:t5=r1;000000000000000000000000;001777777777777777777777;
1392 If the lower bound of a subrange is 0 and the upper bound is negative,
1393 the type is an unsigned integral type whose size in bytes is the
1394 absolute value of the upper bound. I believe this is a Convex
1395 convention for @code{unsigned long long}.
1397 If the lower bound of a subrange is negative and the upper bound is 0,
1398 the type is a signed integral type whose size in bytes is
1399 the absolute value of the lower bound. I believe this is a Convex
1400 convention for @code{long long}. To distinguish this from a legitimate
1401 subrange, the type should be a subrange of itself. I'm not sure whether
1402 this is the case for Convex.
1404 @node Traditional Other Types
1405 @subsubsection Traditional Other Types
1407 If the upper bound of a subrange is 0 and the lower bound is positive,
1408 the type is a floating point type, and the lower bound of the subrange
1409 indicates the number of bytes in the type:
1412 .stabs "float:t12=r1;4;0;",128,0,0,0
1413 .stabs "double:t13=r1;8;0;",128,0,0,0
1416 However, GCC writes @code{long double} the same way it writes
1417 @code{double}, so there is no way to distinguish.
1420 .stabs "long double:t14=r1;8;0;",128,0,0,0
1423 Complex types are defined the same way as floating-point types; there is
1424 no way to distinguish a single-precision complex from a double-precision
1425 floating-point type.
1427 The C @code{void} type is defined as itself:
1430 .stabs "void:t15=15",128,0,0,0
1433 I'm not sure how a boolean type is represented.
1435 @node Builtin Type Descriptors
1436 @subsection Defining Builtin Types Using Builtin Type Descriptors
1438 This is the method used by Sun's @code{acc} for defining builtin types.
1439 These are the type descriptors to define builtin types:
1442 @c FIXME: clean up description of width and offset, once we figure out
1444 @item b @var{signed} @var{char-flag} @var{width} ; @var{offset} ; @var{nbits} ;
1445 Define an integral type. @var{signed} is @samp{u} for unsigned or
1446 @samp{s} for signed. @var{char-flag} is @samp{c} which indicates this
1447 is a character type, or is omitted. I assume this is to distinguish an
1448 integral type from a character type of the same size, for example it
1449 might make sense to set it for the C type @code{wchar_t} so the debugger
1450 can print such variables differently (Solaris does not do this). Sun
1451 sets it on the C types @code{signed char} and @code{unsigned char} which
1452 arguably is wrong. @var{width} and @var{offset} appear to be for small
1453 objects stored in larger ones, for example a @code{short} in an
1454 @code{int} register. @var{width} is normally the number of bytes in the
1455 type. @var{offset} seems to always be zero. @var{nbits} is the number
1456 of bits in the type.
1458 Note that type descriptor @samp{b} used for builtin types conflicts with
1459 its use for Pascal space types (@pxref{Miscellaneous Types}); they can
1460 be distinguished because the character following the type descriptor
1461 will be a digit, @samp{(}, or @samp{-} for a Pascal space type, or
1462 @samp{u} or @samp{s} for a builtin type.
1465 Documented by AIX to define a wide character type, but their compiler
1466 actually uses negative type numbers (@pxref{Negative Type Numbers}).
1468 @item R @var{fp-type} ; @var{bytes} ;
1469 Define a floating point type. @var{fp-type} has one of the following values:
1473 IEEE 32-bit (single precision) floating point format.
1476 IEEE 64-bit (double precision) floating point format.
1478 @item 3 (NF_COMPLEX)
1479 @item 4 (NF_COMPLEX16)
1480 @item 5 (NF_COMPLEX32)
1481 @c "GDB source" really means @file{include/aout/stab_gnu.h}, but trying
1482 @c to put that here got an overfull hbox.
1483 These are for complex numbers. A comment in the GDB source describes
1484 them as Fortran @code{complex}, @code{double complex}, and
1485 @code{complex*16}, respectively, but what does that mean? (i.e., Single
1486 precision? Double precision?).
1488 @item 6 (NF_LDOUBLE)
1489 Long double. This should probably only be used for Sun format
1490 @code{long double}, and new codes should be used for other floating
1491 point formats (@code{NF_DOUBLE} can be used if a @code{long double} is
1492 really just an IEEE double, of course).
1495 @var{bytes} is the number of bytes occupied by the type. This allows a
1496 debugger to perform some operations with the type even if it doesn't
1497 understand @var{fp-type}.
1499 @item g @var{type-information} ; @var{nbits}
1500 Documented by AIX to define a floating type, but their compiler actually
1501 uses negative type numbers (@pxref{Negative Type Numbers}).
1503 @item c @var{type-information} ; @var{nbits}
1504 Documented by AIX to define a complex type, but their compiler actually
1505 uses negative type numbers (@pxref{Negative Type Numbers}).
1508 The C @code{void} type is defined as a signed integral type 0 bits long:
1510 .stabs "void:t19=bs0;0;0",128,0,0,0
1512 The Solaris compiler seems to omit the trailing semicolon in this case.
1513 Getting sloppy in this way is not a swift move because if a type is
1514 embedded in a more complex expression it is necessary to be able to tell
1517 I'm not sure how a boolean type is represented.
1519 @node Negative Type Numbers
1520 @subsection Negative Type Numbers
1522 This is the method used in XCOFF for defining builtin types.
1523 Since the debugger knows about the builtin types anyway, the idea of
1524 negative type numbers is simply to give a special type number which
1525 indicates the builtin type. There is no stab defining these types.
1527 There are several subtle issues with negative type numbers.
1529 One is the size of the type. A builtin type (for example the C types
1530 @code{int} or @code{long}) might have different sizes depending on
1531 compiler options, the target architecture, the ABI, etc. This issue
1532 doesn't come up for IBM tools since (so far) they just target the
1533 RS/6000; the sizes indicated below for each size are what the IBM
1534 RS/6000 tools use. To deal with differing sizes, either define separate
1535 negative type numbers for each size (which works but requires changing
1536 the debugger, and, unless you get both AIX dbx and GDB to accept the
1537 change, introduces an incompatibility), or use a type attribute
1538 (@pxref{String Field}) to define a new type with the appropriate size
1539 (which merely requires a debugger which understands type attributes,
1540 like AIX dbx or GDB). For example,
1543 .stabs "boolean:t10=@@s8;-16",128,0,0,0
1546 defines an 8-bit boolean type, and
1549 .stabs "boolean:t10=@@s64;-16",128,0,0,0
1552 defines a 64-bit boolean type.
1554 A similar issue is the format of the type. This comes up most often for
1555 floating-point types, which could have various formats (particularly
1556 extended doubles, which vary quite a bit even among IEEE systems).
1557 Again, it is best to define a new negative type number for each
1558 different format; changing the format based on the target system has
1559 various problems. One such problem is that the Alpha has both VAX and
1560 IEEE floating types. One can easily imagine one library using the VAX
1561 types and another library in the same executable using the IEEE types.
1562 Another example is that the interpretation of whether a boolean is true
1563 or false can be based on the least significant bit, most significant
1564 bit, whether it is zero, etc., and different compilers (or different
1565 options to the same compiler) might provide different kinds of boolean.
1567 The last major issue is the names of the types. The name of a given
1568 type depends @emph{only} on the negative type number given; these do not
1569 vary depending on the language, the target system, or anything else.
1570 One can always define separate type numbers---in the following list you
1571 will see for example separate @code{int} and @code{integer*4} types
1572 which are identical except for the name. But compatibility can be
1573 maintained by not inventing new negative type numbers and instead just
1574 defining a new type with a new name. For example:
1577 .stabs "CARDINAL:t10=-8",128,0,0,0
1580 Here is the list of negative type numbers. The phrase @dfn{integral
1581 type} is used to mean twos-complement (I strongly suspect that all
1582 machines which use stabs use twos-complement; most machines use
1583 twos-complement these days).
1587 @code{int}, 32 bit signed integral type.
1590 @code{char}, 8 bit type holding a character. Both GDB and dbx on AIX
1591 treat this as signed. GCC uses this type whether @code{char} is signed
1592 or not, which seems like a bad idea. The AIX compiler (@code{xlc}) seems to
1593 avoid this type; it uses -5 instead for @code{char}.
1596 @code{short}, 16 bit signed integral type.
1599 @code{long}, 32 bit signed integral type.
1602 @code{unsigned char}, 8 bit unsigned integral type.
1605 @code{signed char}, 8 bit signed integral type.
1608 @code{unsigned short}, 16 bit unsigned integral type.
1611 @code{unsigned int}, 32 bit unsigned integral type.
1614 @code{unsigned}, 32 bit unsigned integral type.
1617 @code{unsigned long}, 32 bit unsigned integral type.
1620 @code{void}, type indicating the lack of a value.
1623 @code{float}, IEEE single precision.
1626 @code{double}, IEEE double precision.
1629 @code{long double}, IEEE double precision. The compiler claims the size
1630 will increase in a future release, and for binary compatibility you have
1631 to avoid using @code{long double}. I hope when they increase it they
1632 use a new negative type number.
1635 @code{integer}. 32 bit signed integral type.
1638 @code{boolean}. 32 bit type. GDB and GCC assume that zero is false,
1639 one is true, and other values have unspecified meaning. I hope this
1640 agrees with how the IBM tools use the type.
1643 @code{short real}. IEEE single precision.
1646 @code{real}. IEEE double precision.
1649 @code{stringptr}. @xref{Strings}.
1652 @code{character}, 8 bit unsigned character type.
1655 @code{logical*1}, 8 bit type. This Fortran type has a split
1656 personality in that it is used for boolean variables, but can also be
1657 used for unsigned integers. 0 is false, 1 is true, and other values are
1661 @code{logical*2}, 16 bit type. This Fortran type has a split
1662 personality in that it is used for boolean variables, but can also be
1663 used for unsigned integers. 0 is false, 1 is true, and other values are
1667 @code{logical*4}, 32 bit type. This Fortran type has a split
1668 personality in that it is used for boolean variables, but can also be
1669 used for unsigned integers. 0 is false, 1 is true, and other values are
1673 @code{logical}, 32 bit type. This Fortran type has a split
1674 personality in that it is used for boolean variables, but can also be
1675 used for unsigned integers. 0 is false, 1 is true, and other values are
1679 @code{complex}. A complex type consisting of two IEEE single-precision
1680 floating point values.
1683 @code{complex}. A complex type consisting of two IEEE double-precision
1684 floating point values.
1687 @code{integer*1}, 8 bit signed integral type.
1690 @code{integer*2}, 16 bit signed integral type.
1693 @code{integer*4}, 32 bit signed integral type.
1696 @code{wchar}. Wide character, 16 bits wide, unsigned (what format?
1700 @code{long long}, 64 bit signed integral type.
1703 @code{unsigned long long}, 64 bit unsigned integral type.
1706 @code{logical*8}, 64 bit unsigned integral type.
1709 @code{integer*8}, 64 bit signed integral type.
1712 @node Miscellaneous Types
1713 @section Miscellaneous Types
1716 @item b @var{type-information} ; @var{bytes}
1717 Pascal space type. This is documented by IBM; what does it mean?
1719 This use of the @samp{b} type descriptor can be distinguished
1720 from its use for builtin integral types (@pxref{Builtin Type
1721 Descriptors}) because the character following the type descriptor is
1722 always a digit, @samp{(}, or @samp{-}.
1724 @item B @var{type-information}
1725 A volatile-qualified version of @var{type-information}. This is
1726 a Sun extension. References and stores to a variable with a
1727 volatile-qualified type must not be optimized or cached; they
1728 must occur as the user specifies them.
1730 @item d @var{type-information}
1731 File of type @var{type-information}. As far as I know this is only used
1734 @item k @var{type-information}
1735 A const-qualified version of @var{type-information}. This is a Sun
1736 extension. A variable with a const-qualified type cannot be modified.
1738 @item M @var{type-information} ; @var{length}
1739 Multiple instance type. The type seems to composed of @var{length}
1740 repetitions of @var{type-information}, for example @code{character*3} is
1741 represented by @samp{M-2;3}, where @samp{-2} is a reference to a
1742 character type (@pxref{Negative Type Numbers}). I'm not sure how this
1743 differs from an array. This appears to be a Fortran feature.
1744 @var{length} is a bound, like those in range types; see @ref{Subranges}.
1746 @item S @var{type-information}
1747 Pascal set type. @var{type-information} must be a small type such as an
1748 enumeration or a subrange, and the type is a bitmask whose length is
1749 specified by the number of elements in @var{type-information}.
1751 In CHILL, if it is a bitstring instead of a set, also use the @samp{S}
1752 type attribute (@pxref{String Field}).
1754 @item * @var{type-information}
1755 Pointer to @var{type-information}.
1758 @node Cross-References
1759 @section Cross-References to Other Types
1761 A type can be used before it is defined; one common way to deal with
1762 that situation is just to use a type reference to a type which has not
1765 Another way is with the @samp{x} type descriptor, which is followed by
1766 @samp{s} for a structure tag, @samp{u} for a union tag, or @samp{e} for
1767 a enumerator tag, followed by the name of the tag, followed by @samp{:}.
1768 If the name contains @samp{::} between a @samp{<} and @samp{>} pair (for
1769 C++ templates), such a @samp{::} does not end the name---only a single
1770 @samp{:} ends the name; see @ref{Nested Symbols}.
1772 For example, the following C declarations:
1783 .stabs "bar:G16=*17=xsfoo:",32,0,0,0
1786 Not all debuggers support the @samp{x} type descriptor, so on some
1787 machines GCC does not use it. I believe that for the above example it
1788 would just emit a reference to type 17 and never define it, but I
1789 haven't verified that.
1791 Modula-2 imported types, at least on AIX, use the @samp{i} type
1792 descriptor, which is followed by the name of the module from which the
1793 type is imported, followed by @samp{:}, followed by the name of the
1794 type. There is then optionally a comma followed by type information for
1795 the type. This differs from merely naming the type (@pxref{Typedefs}) in
1796 that it identifies the module; I don't understand whether the name of
1797 the type given here is always just the same as the name we are giving
1798 it, or whether this type descriptor is used with a nameless stab
1799 (@pxref{String Field}), or what. The symbol ends with @samp{;}.
1802 @section Subrange Types
1804 The @samp{r} type descriptor defines a type as a subrange of another
1805 type. It is followed by type information for the type of which it is a
1806 subrange, a semicolon, an integral lower bound, a semicolon, an
1807 integral upper bound, and a semicolon. The AIX documentation does not
1808 specify the trailing semicolon, in an effort to specify array indexes
1809 more cleanly, but a subrange which is not an array index has always
1810 included a trailing semicolon (@pxref{Arrays}).
1812 Instead of an integer, either bound can be one of the following:
1815 @item A @var{offset}
1816 The bound is passed by reference on the stack at offset @var{offset}
1817 from the argument list. @xref{Parameters}, for more information on such
1820 @item T @var{offset}
1821 The bound is passed by value on the stack at offset @var{offset} from
1824 @item a @var{register-number}
1825 The bound is passed by reference in register number
1826 @var{register-number}.
1828 @item t @var{register-number}
1829 The bound is passed by value in register number @var{register-number}.
1835 Subranges are also used for builtin types; see @ref{Traditional Builtin Types}.
1838 @section Array Types
1840 Arrays use the @samp{a} type descriptor. Following the type descriptor
1841 is the type of the index and the type of the array elements. If the
1842 index type is a range type, it ends in a semicolon; otherwise
1843 (for example, if it is a type reference), there does not
1844 appear to be any way to tell where the types are separated. In an
1845 effort to clean up this mess, IBM documents the two types as being
1846 separated by a semicolon, and a range type as not ending in a semicolon
1847 (but this is not right for range types which are not array indexes,
1848 @pxref{Subranges}). I think probably the best solution is to specify
1849 that a semicolon ends a range type, and that the index type and element
1850 type of an array are separated by a semicolon, but that if the index
1851 type is a range type, the extra semicolon can be omitted. GDB (at least
1852 through version 4.9) doesn't support any kind of index type other than a
1853 range anyway; I'm not sure about dbx.
1855 It is well established, and widely used, that the type of the index,
1856 unlike most types found in the stabs, is merely a type definition, not
1857 type information (@pxref{String Field}) (that is, it need not start with
1858 @samp{@var{type-number}=} if it is defining a new type). According to a
1859 comment in GDB, this is also true of the type of the array elements; it
1860 gives @samp{ar1;1;10;ar1;1;10;4} as a legitimate way to express a two
1861 dimensional array. According to AIX documentation, the element type
1862 must be type information. GDB accepts either.
1864 The type of the index is often a range type, expressed as the type
1865 descriptor @samp{r} and some parameters. It defines the size of the
1866 array. In the example below, the range @samp{r1;0;2;} defines an index
1867 type which is a subrange of type 1 (integer), with a lower bound of 0
1868 and an upper bound of 2. This defines the valid range of subscripts of
1869 a three-element C array.
1871 For example, the definition:
1874 char char_vec[3] = @{'a','b','c'@};
1878 produces the output:
1881 .stabs "char_vec:G19=ar1;0;2;2",32,0,0,0
1890 If an array is @dfn{packed}, the elements are spaced more
1891 closely than normal, saving memory at the expense of speed. For
1892 example, an array of 3-byte objects might, if unpacked, have each
1893 element aligned on a 4-byte boundary, but if packed, have no padding.
1894 One way to specify that something is packed is with type attributes
1895 (@pxref{String Field}). In the case of arrays, another is to use the
1896 @samp{P} type descriptor instead of @samp{a}. Other than specifying a
1897 packed array, @samp{P} is identical to @samp{a}.
1899 @c FIXME-what is it? A pointer?
1900 An open array is represented by the @samp{A} type descriptor followed by
1901 type information specifying the type of the array elements.
1903 @c FIXME: what is the format of this type? A pointer to a vector of pointers?
1904 An N-dimensional dynamic array is represented by
1907 D @var{dimensions} ; @var{type-information}
1910 @c Does dimensions really have this meaning? The AIX documentation
1912 @var{dimensions} is the number of dimensions; @var{type-information}
1913 specifies the type of the array elements.
1915 @c FIXME: what is the format of this type? A pointer to some offsets in
1917 A subarray of an N-dimensional array is represented by
1920 E @var{dimensions} ; @var{type-information}
1923 @c Does dimensions really have this meaning? The AIX documentation
1925 @var{dimensions} is the number of dimensions; @var{type-information}
1926 specifies the type of the array elements.
1931 Some languages, like C or the original Pascal, do not have string types,
1932 they just have related things like arrays of characters. But most
1933 Pascals and various other languages have string types, which are
1934 indicated as follows:
1937 @item n @var{type-information} ; @var{bytes}
1938 @var{bytes} is the maximum length. I'm not sure what
1939 @var{type-information} is; I suspect that it means that this is a string
1940 of @var{type-information} (thus allowing a string of integers, a string
1941 of wide characters, etc., as well as a string of characters). Not sure
1942 what the format of this type is. This is an AIX feature.
1944 @item z @var{type-information} ; @var{bytes}
1945 Just like @samp{n} except that this is a gstring, not an ordinary
1946 string. I don't know the difference.
1949 Pascal Stringptr. What is this? This is an AIX feature.
1952 Languages, such as CHILL which have a string type which is basically
1953 just an array of characters use the @samp{S} type attribute
1954 (@pxref{String Field}).
1957 @section Enumerations
1959 Enumerations are defined with the @samp{e} type descriptor.
1961 @c FIXME: Where does this information properly go? Perhaps it is
1962 @c redundant with something we already explain.
1963 The source line below declares an enumeration type at file scope.
1964 The type definition is located after the @code{N_RBRAC} that marks the end of
1965 the previous procedure's block scope, and before the @code{N_FUN} that marks
1966 the beginning of the next procedure's block scope. Therefore it does not
1967 describe a block local symbol, but a file local one.
1972 enum e_places @{first,second=3,last@};
1976 generates the following stab:
1979 .stabs "e_places:T22=efirst:0,second:3,last:4,;",128,0,0,0
1982 The symbol descriptor (@samp{T}) says that the stab describes a
1983 structure, enumeration, or union tag. The type descriptor @samp{e},
1984 following the @samp{22=} of the type definition narrows it down to an
1985 enumeration type. Following the @samp{e} is a list of the elements of
1986 the enumeration. The format is @samp{@var{name}:@var{value},}. The
1987 list of elements ends with @samp{;}. The fact that @var{value} is
1988 specified as an integer can cause problems if the value is large. GCC
1989 2.5.2 tries to output it in octal in that case with a leading zero,
1990 which is probably a good thing, although GDB 4.11 supports octal only in
1991 cases where decimal is perfectly good. Negative decimal values are
1992 supported by both GDB and dbx.
1994 There is no standard way to specify the size of an enumeration type; it
1995 is determined by the architecture (normally all enumerations types are
1996 32 bits). Type attributes can be used to specify an enumeration type of
1997 another size for debuggers which support them; see @ref{String Field}.
1999 Enumeration types are unusual in that they define symbols for the
2000 enumeration values (@code{first}, @code{second}, and @code{third} in the
2001 above example), and even though these symbols are visible in the file as
2002 a whole (rather than being in a more local namespace like structure
2003 member names), they are defined in the type definition for the
2004 enumeration type rather than each having their own symbol. In order to
2005 be fast, GDB will only get symbols from such types (in its initial scan
2006 of the stabs) if the type is the first thing defined after a @samp{T} or
2007 @samp{t} symbol descriptor (the above example fulfills this
2008 requirement). If the type does not have a name, the compiler should
2009 emit it in a nameless stab (@pxref{String Field}); GCC does this.
2014 The encoding of structures in stabs can be shown with an example.
2016 The following source code declares a structure tag and defines an
2017 instance of the structure in global scope. Then a @code{typedef} equates the
2018 structure tag with a new type. Separate stabs are generated for the
2019 structure tag, the structure @code{typedef}, and the structure instance. The
2020 stabs for the tag and the @code{typedef} are emitted when the definitions are
2021 encountered. Since the structure elements are not initialized, the
2022 stab and code for the structure variable itself is located at the end
2023 of the program in the bss section.
2030 struct s_tag* s_next;
2033 typedef struct s_tag s_typedef;
2036 The structure tag has an @code{N_LSYM} stab type because, like the
2037 enumeration, the symbol has file scope. Like the enumeration, the
2038 symbol descriptor is @samp{T}, for enumeration, structure, or tag type.
2039 The type descriptor @samp{s} following the @samp{16=} of the type
2040 definition narrows the symbol type to structure.
2042 Following the @samp{s} type descriptor is the number of bytes the
2043 structure occupies, followed by a description of each structure element.
2044 The structure element descriptions are of the form
2045 @samp{@var{name}:@var{type}, @var{bit offset from the start of the
2046 struct}, @var{number of bits in the element}}.
2048 @c FIXME: phony line break. Can probably be fixed by using an example
2049 @c with fewer fields.
2052 .stabs "s_tag:T16=s20s_int:1,0,32;s_float:12,32,32;
2053 s_char_vec:17=ar1;0;7;2,64,64;s_next:18=*16,128,32;;",128,0,0,0
2056 In this example, the first two structure elements are previously defined
2057 types. For these, the type following the @samp{@var{name}:} part of the
2058 element description is a simple type reference. The other two structure
2059 elements are new types. In this case there is a type definition
2060 embedded after the @samp{@var{name}:}. The type definition for the
2061 array element looks just like a type definition for a stand-alone array.
2062 The @code{s_next} field is a pointer to the same kind of structure that
2063 the field is an element of. So the definition of structure type 16
2064 contains a type definition for an element which is a pointer to type 16.
2066 If a field is a static member (this is a C++ feature in which a single
2067 variable appears to be a field of every structure of a given type) it
2068 still starts out with the field name, a colon, and the type, but then
2069 instead of a comma, bit position, comma, and bit size, there is a colon
2070 followed by the name of the variable which each such field refers to.
2072 If the structure has methods (a C++ feature), they follow the non-method
2073 fields; see @ref{Cplusplus}.
2076 @section Giving a Type a Name
2078 @findex N_LSYM, for types
2079 @findex C_DECL, for types
2080 To give a type a name, use the @samp{t} symbol descriptor. The type
2081 is specified by the type information (@pxref{String Field}) for the stab.
2085 .stabs "s_typedef:t16",128,0,0,0 # @r{128 is N_LSYM}
2088 specifies that @code{s_typedef} refers to type number 16. Such stabs
2089 have symbol type @code{N_LSYM} (or @code{C_DECL} for XCOFF). (The Sun
2090 documentation mentions using @code{N_GSYM} in some cases).
2092 If you are specifying the tag name for a structure, union, or
2093 enumeration, use the @samp{T} symbol descriptor instead. I believe C is
2094 the only language with this feature.
2096 If the type is an opaque type (I believe this is a Modula-2 feature),
2097 AIX provides a type descriptor to specify it. The type descriptor is
2098 @samp{o} and is followed by a name. I don't know what the name
2099 means---is it always the same as the name of the type, or is this type
2100 descriptor used with a nameless stab (@pxref{String Field})? There
2101 optionally follows a comma followed by type information which defines
2102 the type of this type. If omitted, a semicolon is used in place of the
2103 comma and the type information, and the type is much like a generic
2104 pointer type---it has a known size but little else about it is
2118 This code generates a stab for a union tag and a stab for a union
2119 variable. Both use the @code{N_LSYM} stab type. If a union variable is
2120 scoped locally to the procedure in which it is defined, its stab is
2121 located immediately preceding the @code{N_LBRAC} for the procedure's block
2124 The stab for the union tag, however, is located preceding the code for
2125 the procedure in which it is defined. The stab type is @code{N_LSYM}. This
2126 would seem to imply that the union type is file scope, like the struct
2127 type @code{s_tag}. This is not true. The contents and position of the stab
2128 for @code{u_type} do not convey any information about its procedure local
2131 @c FIXME: phony line break. Can probably be fixed by using an example
2132 @c with fewer fields.
2135 .stabs "u_tag:T23=u4u_int:1,0,32;u_float:12,0,32;u_char:21,0,32;;",
2139 The symbol descriptor @samp{T}, following the @samp{name:} means that
2140 the stab describes an enumeration, structure, or union tag. The type
2141 descriptor @samp{u}, following the @samp{23=} of the type definition,
2142 narrows it down to a union type definition. Following the @samp{u} is
2143 the number of bytes in the union. After that is a list of union element
2144 descriptions. Their format is @samp{@var{name}:@var{type}, @var{bit
2145 offset into the union}, @var{number of bytes for the element};}.
2147 The stab for the union variable is:
2150 .stabs "an_u:23",128,0,0,-20 # @r{128 is N_LSYM}
2153 @samp{-20} specifies where the variable is stored (@pxref{Stack
2156 @node Function Types
2157 @section Function Types
2159 Various types can be defined for function variables. These types are
2160 not used in defining functions (@pxref{Procedures}); they are used for
2161 things like pointers to functions.
2163 The simple, traditional, type is type descriptor @samp{f} is followed by
2164 type information for the return type of the function, followed by a
2167 This does not deal with functions for which the number and types of the
2168 parameters are part of the type, as in Modula-2 or ANSI C. AIX provides
2169 extensions to specify these, using the @samp{f}, @samp{F}, @samp{p}, and
2170 @samp{R} type descriptors.
2172 First comes the type descriptor. If it is @samp{f} or @samp{F}, this
2173 type involves a function rather than a procedure, and the type
2174 information for the return type of the function follows, followed by a
2175 comma. Then comes the number of parameters to the function and a
2176 semicolon. Then, for each parameter, there is the name of the parameter
2177 followed by a colon (this is only present for type descriptors @samp{R}
2178 and @samp{F} which represent Pascal function or procedure parameters),
2179 type information for the parameter, a comma, 0 if passed by reference or
2180 1 if passed by value, and a semicolon. The type definition ends with a
2183 For example, this variable definition:
2190 generates the following code:
2193 .stabs "g_pf:G24=*25=f1",32,0,0,0
2194 .common _g_pf,4,"bss"
2197 The variable defines a new type, 24, which is a pointer to another new
2198 type, 25, which is a function returning @code{int}.
2201 @chapter Symbol Information in Symbol Tables
2203 This chapter describes the format of symbol table entries
2204 and how stab assembler directives map to them. It also describes the
2205 transformations that the assembler and linker make on data from stabs.
2208 * Symbol Table Format::
2209 * Transformations On Symbol Tables::
2212 @node Symbol Table Format
2213 @section Symbol Table Format
2215 Each time the assembler encounters a stab directive, it puts
2216 each field of the stab into a corresponding field in a symbol table
2217 entry of its output file. If the stab contains a string field, the
2218 symbol table entry for that stab points to a string table entry
2219 containing the string data from the stab. Assembler labels become
2220 relocatable addresses. Symbol table entries in a.out have the format:
2222 @c FIXME: should refer to external, not internal.
2224 struct internal_nlist @{
2225 unsigned long n_strx; /* index into string table of name */
2226 unsigned char n_type; /* type of symbol */
2227 unsigned char n_other; /* misc info (usually empty) */
2228 unsigned short n_desc; /* description field */
2229 bfd_vma n_value; /* value of symbol */
2233 If the stab has a string, the @code{n_strx} field holds the offset in
2234 bytes of the string within the string table. The string is terminated
2235 by a NUL character. If the stab lacks a string (for example, it was
2236 produced by a @code{.stabn} or @code{.stabd} directive), the
2237 @code{n_strx} field is zero.
2239 Symbol table entries with @code{n_type} field values greater than 0x1f
2240 originated as stabs generated by the compiler (with one random
2241 exception). The other entries were placed in the symbol table of the
2242 executable by the assembler or the linker.
2244 @node Transformations On Symbol Tables
2245 @section Transformations on Symbol Tables
2247 The linker concatenates object files and does fixups of externally
2250 You can see the transformations made on stab data by the assembler and
2251 linker by examining the symbol table after each pass of the build. To
2252 do this, use @samp{nm -ap}, which dumps the symbol table, including
2253 debugging information, unsorted. For stab entries the columns are:
2254 @var{value}, @var{other}, @var{desc}, @var{type}, @var{string}. For
2255 assembler and linker symbols, the columns are: @var{value}, @var{type},
2258 The low 5 bits of the stab type tell the linker how to relocate the
2259 value of the stab. Thus for stab types like @code{N_RSYM} and
2260 @code{N_LSYM}, where the value is an offset or a register number, the
2261 low 5 bits are @code{N_ABS}, which tells the linker not to relocate the
2264 Where the value of a stab contains an assembly language label,
2265 it is transformed by each build step. The assembler turns it into a
2266 relocatable address and the linker turns it into an absolute address.
2269 * Transformations On Static Variables::
2270 * Transformations On Global Variables::
2271 * Stab Section Transformations:: For some object file formats,
2272 things are a bit different.
2275 @node Transformations On Static Variables
2276 @subsection Transformations on Static Variables
2278 This source line defines a static variable at file scope:
2281 static int s_g_repeat
2285 The following stab describes the symbol:
2288 .stabs "s_g_repeat:S1",38,0,0,_s_g_repeat
2292 The assembler transforms the stab into this symbol table entry in the
2293 @file{.o} file. The location is expressed as a data segment offset.
2296 00000084 - 00 0000 STSYM s_g_repeat:S1
2300 In the symbol table entry from the executable, the linker has made the
2301 relocatable address absolute.
2304 0000e00c - 00 0000 STSYM s_g_repeat:S1
2307 @node Transformations On Global Variables
2308 @subsection Transformations on Global Variables
2310 Stabs for global variables do not contain location information. In
2311 this case, the debugger finds location information in the assembler or
2312 linker symbol table entry describing the variable. The source line:
2322 .stabs "g_foo:G2",32,0,0,0
2325 The variable is represented by two symbol table entries in the object
2326 file (see below). The first one originated as a stab. The second one
2327 is an external symbol. The upper case @samp{D} signifies that the
2328 @code{n_type} field of the symbol table contains 7, @code{N_DATA} with
2329 local linkage. The stab's value is zero since the value is not used for
2330 @code{N_GSYM} stabs. The value of the linker symbol is the relocatable
2331 address corresponding to the variable.
2334 00000000 - 00 0000 GSYM g_foo:G2
2339 These entries as transformed by the linker. The linker symbol table
2340 entry now holds an absolute address:
2343 00000000 - 00 0000 GSYM g_foo:G2
2348 @node Stab Section Transformations
2349 @subsection Transformations of Stabs in separate sections
2351 For object file formats using stabs in separate sections (@pxref{Stab
2352 Sections}), use @code{objdump --stabs} instead of @code{nm} to show the
2353 stabs in an object or executable file. @code{objdump} is a GNU utility;
2354 Sun does not provide any equivalent.
2356 The following example is for a stab whose value is an address is
2357 relative to the compilation unit (@pxref{ELF Linker Relocation}). For
2358 example, if the source line
2364 appears within a function, then the assembly language output from the
2370 .stabs "ld:V(0,3)",0x26,0,4,.L18-Ddata.data # @r{0x26 is N_STSYM}
2377 Because the value is formed by subtracting one symbol from another, the
2378 value is absolute, not relocatable, and so the object file contains
2381 Symnum n_type n_othr n_desc n_value n_strx String
2382 31 STSYM 0 4 00000004 680 ld:V(0,3)
2385 without any relocations, and the executable file also contains
2388 Symnum n_type n_othr n_desc n_value n_strx String
2389 31 STSYM 0 4 00000004 680 ld:V(0,3)
2393 @chapter GNU C++ Stabs
2396 * Class Names:: C++ class names are both tags and typedefs.
2397 * Nested Symbols:: C++ symbol names can be within other types.
2398 * Basic Cplusplus Types::
2401 * Methods:: Method definition
2402 * Method Type Descriptor:: The @samp{#} type descriptor
2403 * Member Type Descriptor:: The @samp{@@} type descriptor
2405 * Method Modifiers::
2408 * Virtual Base Classes::
2413 @section C++ Class Names
2415 In C++, a class name which is declared with @code{class}, @code{struct},
2416 or @code{union}, is not only a tag, as in C, but also a type name. Thus
2417 there should be stabs with both @samp{t} and @samp{T} symbol descriptors
2420 To save space, there is a special abbreviation for this case. If the
2421 @samp{T} symbol descriptor is followed by @samp{t}, then the stab
2422 defines both a type name and a tag.
2424 For example, the C++ code
2427 struct foo @{int x;@};
2430 can be represented as either
2433 .stabs "foo:T19=s4x:1,0,32;;",128,0,0,0 # @r{128 is N_LSYM}
2434 .stabs "foo:t19",128,0,0,0
2440 .stabs "foo:Tt19=s4x:1,0,32;;",128,0,0,0
2443 @node Nested Symbols
2444 @section Defining a Symbol Within Another Type
2446 In C++, a symbol (such as a type name) can be defined within another type.
2447 @c FIXME: Needs example.
2449 In stabs, this is sometimes represented by making the name of a symbol
2450 which contains @samp{::}. Such a pair of colons does not end the name
2451 of the symbol, the way a single colon would (@pxref{String Field}). I'm
2452 not sure how consistently used or well thought out this mechanism is.
2453 So that a pair of colons in this position always has this meaning,
2454 @samp{:} cannot be used as a symbol descriptor.
2456 For example, if the string for a stab is @samp{foo::bar::baz:t5=*6},
2457 then @code{foo::bar::baz} is the name of the symbol, @samp{t} is the
2458 symbol descriptor, and @samp{5=*6} is the type information.
2460 @node Basic Cplusplus Types
2461 @section Basic Types For C++
2463 << the examples that follow are based on a01.C >>
2466 C++ adds two more builtin types to the set defined for C. These are
2467 the unknown type and the vtable record type. The unknown type, type
2468 16, is defined in terms of itself like the void type.
2470 The vtable record type, type 17, is defined as a structure type and
2471 then as a structure tag. The structure has four fields: delta, index,
2472 pfn, and delta2. pfn is the function pointer.
2474 << In boilerplate $vtbl_ptr_type, what are the fields delta,
2475 index, and delta2 used for? >>
2477 This basic type is present in all C++ programs even if there are no
2478 virtual methods defined.
2481 .stabs "struct_name:sym_desc(type)type_def(17)=type_desc(struct)struct_bytes(8)
2482 elem_name(delta):type_ref(short int),bit_offset(0),field_bits(16);
2483 elem_name(index):type_ref(short int),bit_offset(16),field_bits(16);
2484 elem_name(pfn):type_def(18)=type_desc(ptr to)type_ref(void),
2485 bit_offset(32),field_bits(32);
2486 elem_name(delta2):type_def(short int);bit_offset(32),field_bits(16);;"
2491 .stabs "$vtbl_ptr_type:t17=s8
2492 delta:6,0,16;index:6,16,16;pfn:18=*15,32,32;delta2:6,32,16;;"
2497 .stabs "name:sym_dec(struct tag)type_ref($vtbl_ptr_type)",N_LSYM,NIL,NIL,NIL
2501 .stabs "$vtbl_ptr_type:T17",128,0,0,0
2504 @node Simple Classes
2505 @section Simple Class Definition
2507 The stabs describing C++ language features are an extension of the
2508 stabs describing C. Stabs representing C++ class types elaborate
2509 extensively on the stab format used to describe structure types in C.
2510 Stabs representing class type variables look just like stabs
2511 representing C language variables.
2513 Consider the following very simple class definition.
2519 int Ameth(int in, char other);
2523 The class @code{baseA} is represented by two stabs. The first stab describes
2524 the class as a structure type. The second stab describes a structure
2525 tag of the class type. Both stabs are of stab type @code{N_LSYM}. Since the
2526 stab is not located between an @code{N_FUN} and an @code{N_LBRAC} stab this indicates
2527 that the class is defined at file scope. If it were, then the @code{N_LSYM}
2528 would signify a local variable.
2530 A stab describing a C++ class type is similar in format to a stab
2531 describing a C struct, with each class member shown as a field in the
2532 structure. The part of the struct format describing fields is
2533 expanded to include extra information relevant to C++ class members.
2534 In addition, if the class has multiple base classes or virtual
2535 functions the struct format outside of the field parts is also
2538 In this simple example the field part of the C++ class stab
2539 representing member data looks just like the field part of a C struct
2540 stab. The section on protections describes how its format is
2541 sometimes extended for member data.
2543 The field part of a C++ class stab representing a member function
2544 differs substantially from the field part of a C struct stab. It
2545 still begins with @samp{name:} but then goes on to define a new type number
2546 for the member function, describe its return type, its argument types,
2547 its protection level, any qualifiers applied to the method definition,
2548 and whether the method is virtual or not. If the method is virtual
2549 then the method description goes on to give the vtable index of the
2550 method, and the type number of the first base class defining the
2553 When the field name is a method name it is followed by two colons rather
2554 than one. This is followed by a new type definition for the method.
2555 This is a number followed by an equal sign and the type of the method.
2556 Normally this will be a type declared using the @samp{#} type
2557 descriptor; see @ref{Method Type Descriptor}; static member functions
2558 are declared using the @samp{f} type descriptor instead; see
2559 @ref{Function Types}.
2561 The format of an overloaded operator method name differs from that of
2562 other methods. It is @samp{op$::@var{operator-name}.} where
2563 @var{operator-name} is the operator name such as @samp{+} or @samp{+=}.
2564 The name ends with a period, and any characters except the period can
2565 occur in the @var{operator-name} string.
2567 The next part of the method description represents the arguments to the
2568 method, preceded by a colon and ending with a semi-colon. The types of
2569 the arguments are expressed in the same way argument types are expressed
2570 in C++ name mangling. In this example an @code{int} and a @code{char}
2573 This is followed by a number, a letter, and an asterisk or period,
2574 followed by another semicolon. The number indicates the protections
2575 that apply to the member function. Here the 2 means public. The
2576 letter encodes any qualifier applied to the method definition. In
2577 this case, @samp{A} means that it is a normal function definition. The dot
2578 shows that the method is not virtual. The sections that follow
2579 elaborate further on these fields and describe the additional
2580 information present for virtual methods.
2584 .stabs "class_name:sym_desc(type)type_def(20)=type_desc(struct)struct_bytes(4)
2585 field_name(Adat):type(int),bit_offset(0),field_bits(32);
2587 method_name(Ameth)::type_def(21)=type_desc(method)return_type(int);
2588 :arg_types(int char);
2589 protection(public)qualifier(normal)virtual(no);;"
2594 .stabs "baseA:t20=s4Adat:1,0,32;Ameth::21=##1;:ic;2A.;;",128,0,0,0
2596 .stabs "class_name:sym_desc(struct tag)",N_LSYM,NIL,NIL,NIL
2598 .stabs "baseA:T20",128,0,0,0
2601 @node Class Instance
2602 @section Class Instance
2604 As shown above, describing even a simple C++ class definition is
2605 accomplished by massively extending the stab format used in C to
2606 describe structure types. However, once the class is defined, C stabs
2607 with no modifications can be used to describe class instances. The
2617 yields the following stab describing the class instance. It looks no
2618 different from a standard C stab describing a local variable.
2621 .stabs "name:type_ref(baseA)", N_LSYM, NIL, NIL, frame_ptr_offset
2625 .stabs "AbaseA:20",128,0,0,-20
2629 @section Method Definition
2631 The class definition shown above declares Ameth. The C++ source below
2636 baseA::Ameth(int in, char other)
2643 This method definition yields three stabs following the code of the
2644 method. One stab describes the method itself and following two describe
2645 its parameters. Although there is only one formal argument all methods
2646 have an implicit argument which is the @code{this} pointer. The @code{this}
2647 pointer is a pointer to the object on which the method was called. Note
2648 that the method name is mangled to encode the class name and argument
2649 types. Name mangling is described in the @sc{arm} (@cite{The Annotated
2650 C++ Reference Manual}, by Ellis and Stroustrup, @sc{isbn}
2651 0-201-51459-1); @file{gpcompare.texi} in Cygnus GCC distributions
2652 describes the differences between GNU mangling and @sc{arm}
2654 @c FIXME: Use @xref, especially if this is generally installed in the
2656 @c FIXME: This information should be in a net release, either of GCC or
2657 @c GDB. But gpcompare.texi doesn't seem to be in the FSF GCC.
2660 .stabs "name:symbol_descriptor(global function)return_type(int)",
2661 N_FUN, NIL, NIL, code_addr_of_method_start
2663 .stabs "Ameth__5baseAic:F1",36,0,0,_Ameth__5baseAic
2666 Here is the stab for the @code{this} pointer implicit argument. The
2667 name of the @code{this} pointer is always @code{this}. Type 19, the
2668 @code{this} pointer is defined as a pointer to type 20, @code{baseA},
2669 but a stab defining @code{baseA} has not yet been emitted. Since the
2670 compiler knows it will be emitted shortly, here it just outputs a cross
2671 reference to the undefined symbol, by prefixing the symbol name with
2675 .stabs "name:sym_desc(register param)type_def(19)=
2676 type_desc(ptr to)type_ref(baseA)=
2677 type_desc(cross-reference to)baseA:",N_RSYM,NIL,NIL,register_number
2679 .stabs "this:P19=*20=xsbaseA:",64,0,0,8
2682 The stab for the explicit integer argument looks just like a parameter
2683 to a C function. The last field of the stab is the offset from the
2684 argument pointer, which in most systems is the same as the frame
2688 .stabs "name:sym_desc(value parameter)type_ref(int)",
2689 N_PSYM,NIL,NIL,offset_from_arg_ptr
2691 .stabs "in:p1",160,0,0,72
2694 << The examples that follow are based on A1.C >>
2696 @node Method Type Descriptor
2697 @section The @samp{#} Type Descriptor
2699 This is used to describe a class method. This is a function which takes
2700 an extra argument as its first argument, for the @code{this} pointer.
2702 If the @samp{#} is immediately followed by another @samp{#}, the second
2703 one will be followed by the return type and a semicolon. The class and
2704 argument types are not specified, and must be determined by demangling
2705 the name of the method if it is available.
2707 Otherwise, the single @samp{#} is followed by the class type, a comma,
2708 the return type, a comma, and zero or more parameter types separated by
2709 commas. The list of arguments is terminated by a semicolon. In the
2710 debugging output generated by gcc, a final argument type of @code{void}
2711 indicates a method which does not take a variable number of arguments.
2712 If the final argument type of @code{void} does not appear, the method
2713 was declared with an ellipsis.
2715 Note that although such a type will normally be used to describe fields
2716 in structures, unions, or classes, for at least some versions of the
2717 compiler it can also be used in other contexts.
2719 @node Member Type Descriptor
2720 @section The @samp{@@} Type Descriptor
2722 The @samp{@@} type descriptor is used together with the @samp{*} type
2723 descriptor for a pointer-to-non-static-member-data type. It is followed
2724 by type information for the class (or union), a comma, and type
2725 information for the member data.
2727 The following C++ source:
2730 typedef int A::*int_in_a;
2733 generates the following stab:
2736 .stabs "int_in_a:t20=*21=@@19,1",128,0,0,0
2739 Note that there is a conflict between this and type attributes
2740 (@pxref{String Field}); both use type descriptor @samp{@@}.
2741 Fortunately, the @samp{@@} type descriptor used in this C++ sense always
2742 will be followed by a digit, @samp{(}, or @samp{-}, and type attributes
2743 never start with those things.
2746 @section Protections
2748 In the simple class definition shown above all member data and
2749 functions were publicly accessible. The example that follows
2750 contrasts public, protected and privately accessible fields and shows
2751 how these protections are encoded in C++ stabs.
2753 If the character following the @samp{@var{field-name}:} part of the
2754 string is @samp{/}, then the next character is the visibility. @samp{0}
2755 means private, @samp{1} means protected, and @samp{2} means public.
2756 Debuggers should ignore visibility characters they do not recognize, and
2757 assume a reasonable default (such as public) (GDB 4.11 does not, but
2758 this should be fixed in the next GDB release). If no visibility is
2759 specified the field is public. The visibility @samp{9} means that the
2760 field has been optimized out and is public (there is no way to specify
2761 an optimized out field with a private or protected visibility).
2762 Visibility @samp{9} is not supported by GDB 4.11; this should be fixed
2763 in the next GDB release.
2765 The following C++ source:
2779 generates the following stab:
2783 .stabs "vis:T19=s12priv:/01,0,32;prot:/12,32,8;pub:12,64,32;;",128,0,0,0
2786 @samp{vis:T19=s12} indicates that type number 19 is a 12 byte structure
2787 named @code{vis} The @code{priv} field has public visibility
2788 (@samp{/0}), type int (@samp{1}), and offset and size @samp{,0,32;}.
2789 The @code{prot} field has protected visibility (@samp{/1}), type char
2790 (@samp{2}) and offset and size @samp{,32,8;}. The @code{pub} field has
2791 type float (@samp{12}), and offset and size @samp{,64,32;}.
2793 Protections for member functions are signified by one digit embedded in
2794 the field part of the stab describing the method. The digit is 0 if
2795 private, 1 if protected and 2 if public. Consider the C++ class
2799 class all_methods @{
2801 int priv_meth(int in)@{return in;@};
2803 char protMeth(char in)@{return in;@};
2805 float pubMeth(float in)@{return in;@};
2809 It generates the following stab. The digit in question is to the left
2810 of an @samp{A} in each case. Notice also that in this case two symbol
2811 descriptors apply to the class name struct tag and struct type.
2814 .stabs "class_name:sym_desc(struct tag&type)type_def(21)=
2815 sym_desc(struct)struct_bytes(1)
2816 meth_name::type_def(22)=sym_desc(method)returning(int);
2817 :args(int);protection(private)modifier(normal)virtual(no);
2818 meth_name::type_def(23)=sym_desc(method)returning(char);
2819 :args(char);protection(protected)modifier(normal)virtual(no);
2820 meth_name::type_def(24)=sym_desc(method)returning(float);
2821 :args(float);protection(public)modifier(normal)virtual(no);;",
2826 .stabs "all_methods:Tt21=s1priv_meth::22=##1;:i;0A.;protMeth::23=##2;:c;1A.;
2827 pubMeth::24=##12;:f;2A.;;",128,0,0,0
2830 @node Method Modifiers
2831 @section Method Modifiers (@code{const}, @code{volatile}, @code{const volatile})
2835 In the class example described above all the methods have the normal
2836 modifier. This method modifier information is located just after the
2837 protection information for the method. This field has four possible
2838 character values. Normal methods use @samp{A}, const methods use
2839 @samp{B}, volatile methods use @samp{C}, and const volatile methods use
2840 @samp{D}. Consider the class definition below:
2845 int ConstMeth (int arg) const @{ return arg; @};
2846 char VolatileMeth (char arg) volatile @{ return arg; @};
2847 float ConstVolMeth (float arg) const volatile @{return arg; @};
2851 This class is described by the following stab:
2854 .stabs "class(A):sym_desc(struct)type_def(20)=type_desc(struct)struct_bytes(1)
2855 meth_name(ConstMeth)::type_def(21)sym_desc(method)
2856 returning(int);:arg(int);protection(public)modifier(const)virtual(no);
2857 meth_name(VolatileMeth)::type_def(22)=sym_desc(method)
2858 returning(char);:arg(char);protection(public)modifier(volatile)virt(no)
2859 meth_name(ConstVolMeth)::type_def(23)=sym_desc(method)
2860 returning(float);:arg(float);protection(public)modifier(const volatile)
2861 virtual(no);;", @dots{}
2865 .stabs "A:T20=s1ConstMeth::21=##1;:i;2B.;VolatileMeth::22=##2;:c;2C.;
2866 ConstVolMeth::23=##12;:f;2D.;;",128,0,0,0
2869 @node Virtual Methods
2870 @section Virtual Methods
2872 << The following examples are based on a4.C >>
2874 The presence of virtual methods in a class definition adds additional
2875 data to the class description. The extra data is appended to the
2876 description of the virtual method and to the end of the class
2877 description. Consider the class definition below:
2883 virtual int A_virt (int arg) @{ return arg; @};
2887 This results in the stab below describing class A. It defines a new
2888 type (20) which is an 8 byte structure. The first field of the class
2889 struct is @samp{Adat}, an integer, starting at structure offset 0 and
2892 The second field in the class struct is not explicitly defined by the
2893 C++ class definition but is implied by the fact that the class
2894 contains a virtual method. This field is the vtable pointer. The
2895 name of the vtable pointer field starts with @samp{$vf} and continues with a
2896 type reference to the class it is part of. In this example the type
2897 reference for class A is 20 so the name of its vtable pointer field is
2898 @samp{$vf20}, followed by the usual colon.
2900 Next there is a type definition for the vtable pointer type (21).
2901 This is in turn defined as a pointer to another new type (22).
2903 Type 22 is the vtable itself, which is defined as an array, indexed by
2904 a range of integers between 0 and 1, and whose elements are of type
2905 17. Type 17 was the vtable record type defined by the boilerplate C++
2906 type definitions, as shown earlier.
2908 The bit offset of the vtable pointer field is 32. The number of bits
2909 in the field are not specified when the field is a vtable pointer.
2911 Next is the method definition for the virtual member function @code{A_virt}.
2912 Its description starts out using the same format as the non-virtual
2913 member functions described above, except instead of a dot after the
2914 @samp{A} there is an asterisk, indicating that the function is virtual.
2915 Since is is virtual some addition information is appended to the end
2916 of the method description.
2918 The first number represents the vtable index of the method. This is a
2919 32 bit unsigned number with the high bit set, followed by a
2922 The second number is a type reference to the first base class in the
2923 inheritance hierarchy defining the virtual member function. In this
2924 case the class stab describes a base class so the virtual function is
2925 not overriding any other definition of the method. Therefore the
2926 reference is to the type number of the class that the stab is
2929 This is followed by three semi-colons. One marks the end of the
2930 current sub-section, one marks the end of the method field, and the
2931 third marks the end of the struct definition.
2933 For classes containing virtual functions the very last section of the
2934 string part of the stab holds a type reference to the first base
2935 class. This is preceded by @samp{~%} and followed by a final semi-colon.
2938 .stabs "class_name(A):type_def(20)=sym_desc(struct)struct_bytes(8)
2939 field_name(Adat):type_ref(int),bit_offset(0),field_bits(32);
2940 field_name(A virt func ptr):type_def(21)=type_desc(ptr to)type_def(22)=
2941 sym_desc(array)index_type_ref(range of int from 0 to 1);
2942 elem_type_ref(vtbl elem type),
2944 meth_name(A_virt)::typedef(23)=sym_desc(method)returning(int);
2945 :arg_type(int),protection(public)normal(yes)virtual(yes)
2946 vtable_index(1);class_first_defining(A);;;~%first_base(A);",
2950 @c FIXME: bogus line break.
2952 .stabs "A:t20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
2953 A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
2957 @section Inheritance
2959 Stabs describing C++ derived classes include additional sections that
2960 describe the inheritance hierarchy of the class. A derived class stab
2961 also encodes the number of base classes. For each base class it tells
2962 if the base class is virtual or not, and if the inheritance is private
2963 or public. It also gives the offset into the object of the portion of
2964 the object corresponding to each base class.
2966 This additional information is embedded in the class stab following the
2967 number of bytes in the struct. First the number of base classes
2968 appears bracketed by an exclamation point and a comma.
2970 Then for each base type there repeats a series: a virtual character, a
2971 visibility character, a number, a comma, another number, and a
2974 The virtual character is @samp{1} if the base class is virtual and
2975 @samp{0} if not. The visibility character is @samp{2} if the derivation
2976 is public, @samp{1} if it is protected, and @samp{0} if it is private.
2977 Debuggers should ignore virtual or visibility characters they do not
2978 recognize, and assume a reasonable default (such as public and
2979 non-virtual) (GDB 4.11 does not, but this should be fixed in the next
2982 The number following the virtual and visibility characters is the offset
2983 from the start of the object to the part of the object pertaining to the
2986 After the comma, the second number is a type_descriptor for the base
2987 type. Finally a semi-colon ends the series, which repeats for each
2990 The source below defines three base classes @code{A}, @code{B}, and
2991 @code{C} and the derived class @code{D}.
2998 virtual int A_virt (int arg) @{ return arg; @};
3004 virtual int B_virt (int arg) @{return arg; @};
3010 virtual int C_virt (int arg) @{return arg; @};
3013 class D : A, virtual B, public C @{
3016 virtual int A_virt (int arg ) @{ return arg+1; @};
3017 virtual int B_virt (int arg) @{ return arg+2; @};
3018 virtual int C_virt (int arg) @{ return arg+3; @};
3019 virtual int D_virt (int arg) @{ return arg; @};
3023 Class stabs similar to the ones described earlier are generated for
3026 @c FIXME!!! the linebreaks in the following example probably make the
3027 @c examples literally unusable, but I don't know any other way to get
3028 @c them on the page.
3029 @c One solution would be to put some of the type definitions into
3030 @c separate stabs, even if that's not exactly what the compiler actually
3033 .stabs "A:T20=s8Adat:1,0,32;$vf20:21=*22=ar1;0;1;17,32;
3034 A_virt::23=##1;:i;2A*-2147483647;20;;;~%20;",128,0,0,0
3036 .stabs "B:Tt25=s8Bdat:1,0,32;$vf25:21,32;B_virt::26=##1;
3037 :i;2A*-2147483647;25;;;~%25;",128,0,0,0
3039 .stabs "C:Tt28=s8Cdat:1,0,32;$vf28:21,32;C_virt::29=##1;
3040 :i;2A*-2147483647;28;;;~%28;",128,0,0,0
3043 In the stab describing derived class @code{D} below, the information about
3044 the derivation of this class is encoded as follows.
3047 .stabs "derived_class_name:symbol_descriptors(struct tag&type)=
3048 type_descriptor(struct)struct_bytes(32)!num_bases(3),
3049 base_virtual(no)inheritance_public(no)base_offset(0),
3050 base_class_type_ref(A);
3051 base_virtual(yes)inheritance_public(no)base_offset(NIL),
3052 base_class_type_ref(B);
3053 base_virtual(no)inheritance_public(yes)base_offset(64),
3054 base_class_type_ref(C); @dots{}
3057 @c FIXME! fake linebreaks.
3059 .stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:
3060 1,160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt:
3061 :32:i;2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;
3062 28;;D_virt::32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
3065 @node Virtual Base Classes
3066 @section Virtual Base Classes
3068 A derived class object consists of a concatenation in memory of the data
3069 areas defined by each base class, starting with the leftmost and ending
3070 with the rightmost in the list of base classes. The exception to this
3071 rule is for virtual inheritance. In the example above, class @code{D}
3072 inherits virtually from base class @code{B}. This means that an
3073 instance of a @code{D} object will not contain its own @code{B} part but
3074 merely a pointer to a @code{B} part, known as a virtual base pointer.
3076 In a derived class stab, the base offset part of the derivation
3077 information, described above, shows how the base class parts are
3078 ordered. The base offset for a virtual base class is always given as 0.
3079 Notice that the base offset for @code{B} is given as 0 even though
3080 @code{B} is not the first base class. The first base class @code{A}
3083 The field information part of the stab for class @code{D} describes the field
3084 which is the pointer to the virtual base class @code{B}. The vbase pointer
3085 name is @samp{$vb} followed by a type reference to the virtual base class.
3086 Since the type id for @code{B} in this example is 25, the vbase pointer name
3089 @c FIXME!! fake linebreaks below
3091 .stabs "D:Tt31=s32!3,000,20;100,25;0264,28;$vb25:24,128;Ddat:1,
3092 160,32;A_virt::32=##1;:i;2A*-2147483647;20;;B_virt::32:i;
3093 2A*-2147483647;25;;C_virt::32:i;2A*-2147483647;28;;D_virt:
3094 :32:i;2A*-2147483646;31;;;~%20;",128,0,0,0
3097 Following the name and a semicolon is a type reference describing the
3098 type of the virtual base class pointer, in this case 24. Type 24 was
3099 defined earlier as the type of the @code{B} class @code{this} pointer. The
3100 @code{this} pointer for a class is a pointer to the class type.
3103 .stabs "this:P24=*25=xsB:",64,0,0,8
3106 Finally the field offset part of the vbase pointer field description
3107 shows that the vbase pointer is the first field in the @code{D} object,
3108 before any data fields defined by the class. The layout of a @code{D}
3109 class object is a follows, @code{Adat} at 0, the vtable pointer for
3110 @code{A} at 32, @code{Cdat} at 64, the vtable pointer for C at 96, the
3111 virtual base pointer for @code{B} at 128, and @code{Ddat} at 160.
3114 @node Static Members
3115 @section Static Members
3117 The data area for a class is a concatenation of the space used by the
3118 data members of the class. If the class has virtual methods, a vtable
3119 pointer follows the class data. The field offset part of each field
3120 description in the class stab shows this ordering.
3122 << How is this reflected in stabs? See Cygnus bug #677 for some info. >>
3125 @appendix Table of Stab Types
3127 The following are all the possible values for the stab type field, for
3128 a.out files, in numeric order. This does not apply to XCOFF, but
3129 it does apply to stabs in sections (@pxref{Stab Sections}). Stabs in
3130 ECOFF use these values but add 0x8f300 to distinguish them from non-stab
3133 The symbolic names are defined in the file @file{include/aout/stabs.def}.
3136 * Non-Stab Symbol Types:: Types from 0 to 0x1f
3137 * Stab Symbol Types:: Types from 0x20 to 0xff
3140 @node Non-Stab Symbol Types
3141 @appendixsec Non-Stab Symbol Types
3143 The following types are used by the linker and assembler, not by stab
3144 directives. Since this document does not attempt to describe aspects of
3145 object file format other than the debugging format, no details are
3148 @c Try to get most of these to fit on a single line.
3158 File scope absolute symbol
3160 @item 0x3 N_ABS | N_EXT
3161 External absolute symbol
3164 File scope text symbol
3166 @item 0x5 N_TEXT | N_EXT
3167 External text symbol
3170 File scope data symbol
3172 @item 0x7 N_DATA | N_EXT
3173 External data symbol
3176 File scope BSS symbol
3178 @item 0x9 N_BSS | N_EXT
3182 Same as @code{N_FN}, for Sequent compilers
3185 Symbol is indirected to another symbol
3188 Common---visible after shared library dynamic link
3191 @itemx 0x15 N_SETA | N_EXT
3192 Absolute set element
3195 @itemx 0x17 N_SETT | N_EXT
3196 Text segment set element
3199 @itemx 0x19 N_SETD | N_EXT
3200 Data segment set element
3203 @itemx 0x1b N_SETB | N_EXT
3204 BSS segment set element
3207 @itemx 0x1d N_SETV | N_EXT
3208 Pointer to set vector
3210 @item 0x1e N_WARNING
3211 Print a warning message during linking
3214 File name of a @file{.o} file
3217 @node Stab Symbol Types
3218 @appendixsec Stab Symbol Types
3220 The following symbol types indicate that this is a stab. This is the
3221 full list of stab numbers, including stab types that are used in
3222 languages other than C.
3226 Global symbol; see @ref{Global Variables}.
3229 Function name (for BSD Fortran); see @ref{Procedures}.
3232 Function name (@pxref{Procedures}) or text segment variable
3236 Data segment file-scope variable; see @ref{Statics}.
3239 BSS segment file-scope variable; see @ref{Statics}.
3242 Name of main routine; see @ref{Main Program}.
3245 Variable in @code{.rodata} section; see @ref{Statics}.
3248 Global symbol (for Pascal); see @ref{N_PC}.
3251 Number of symbols (according to Ultrix V4.0); see @ref{N_NSYMS}.
3254 No DST map; see @ref{N_NOMAP}.
3256 @c FIXME: describe this solaris feature in the body of the text (see
3257 @c comments in include/aout/stab.def).
3259 Object file (Solaris2).
3261 @c See include/aout/stab.def for (a little) more info.
3263 Debugger options (Solaris2).
3266 Register variable; see @ref{Register Variables}.
3269 Modula-2 compilation unit; see @ref{N_M2C}.
3272 Line number in text segment; see @ref{Line Numbers}.
3275 Line number in data segment; see @ref{Line Numbers}.
3278 Line number in bss segment; see @ref{Line Numbers}.
3281 Sun source code browser, path to @file{.cb} file; see @ref{N_BROWS}.
3284 GNU Modula2 definition module dependency; see @ref{N_DEFD}.
3287 Function start/body/end line numbers (Solaris2).
3290 GNU C++ exception variable; see @ref{N_EHDECL}.
3293 Modula2 info "for imc" (according to Ultrix V4.0); see @ref{N_MOD2}.
3296 GNU C++ @code{catch} clause; see @ref{N_CATCH}.
3299 Structure of union element; see @ref{N_SSYM}.
3302 Last stab for module (Solaris2).
3305 Path and name of source file; see @ref{Source Files}.
3308 Stack variable (@pxref{Stack Variables}) or type (@pxref{Typedefs}).
3311 Beginning of an include file (Sun only); see @ref{Include Files}.
3314 Name of include file; see @ref{Include Files}.
3317 Parameter variable; see @ref{Parameters}.
3320 End of an include file; see @ref{Include Files}.
3323 Alternate entry point; see @ref{Alternate Entry Points}.
3326 Beginning of a lexical block; see @ref{Block Structure}.
3329 Place holder for a deleted include file; see @ref{Include Files}.
3332 Modula2 scope information (Sun linker); see @ref{N_SCOPE}.
3335 End of a lexical block; see @ref{Block Structure}.
3338 Begin named common block; see @ref{Common Blocks}.
3341 End named common block; see @ref{Common Blocks}.
3344 Member of a common block; see @ref{Common Blocks}.
3346 @c FIXME: How does this really work? Move it to main body of document.
3348 Pascal @code{with} statement: type,,0,0,offset (Solaris2).
3351 Gould non-base registers; see @ref{Gould}.
3354 Gould non-base registers; see @ref{Gould}.
3357 Gould non-base registers; see @ref{Gould}.
3360 Gould non-base registers; see @ref{Gould}.
3363 Gould non-base registers; see @ref{Gould}.
3366 @c Restore the default table indent
3371 @node Symbol Descriptors
3372 @appendix Table of Symbol Descriptors
3374 The symbol descriptor is the character which follows the colon in many
3375 stabs, and which tells what kind of stab it is. @xref{String Field},
3376 for more information about their use.
3378 @c Please keep this alphabetical
3380 @c In TeX, this looks great, digit is in italics. But makeinfo insists
3381 @c on putting it in `', not realizing that @var should override @code.
3382 @c I don't know of any way to make makeinfo do the right thing. Seems
3383 @c like a makeinfo bug to me.
3387 Variable on the stack; see @ref{Stack Variables}.
3390 C++ nested symbol; see @xref{Nested Symbols}.
3393 Parameter passed by reference in register; see @ref{Reference Parameters}.
3396 Based variable; see @ref{Based Variables}.
3399 Constant; see @ref{Constants}.
3402 Conformant array bound (Pascal, maybe other languages); @ref{Conformant
3403 Arrays}. Name of a caught exception (GNU C++). These can be
3404 distinguished because the latter uses @code{N_CATCH} and the former uses
3405 another symbol type.
3408 Floating point register variable; see @ref{Register Variables}.
3411 Parameter in floating point register; see @ref{Register Parameters}.
3414 File scope function; see @ref{Procedures}.
3417 Global function; see @ref{Procedures}.
3420 Global variable; see @ref{Global Variables}.
3423 @xref{Register Parameters}.
3426 Internal (nested) procedure; see @ref{Nested Procedures}.
3429 Internal (nested) function; see @ref{Nested Procedures}.
3432 Label name (documented by AIX, no further information known).
3435 Module; see @ref{Procedures}.
3438 Argument list parameter; see @ref{Parameters}.
3444 Fortran Function parameter; see @ref{Parameters}.
3447 Unfortunately, three separate meanings have been independently invented
3448 for this symbol descriptor. At least the GNU and Sun uses can be
3449 distinguished by the symbol type. Global Procedure (AIX) (symbol type
3450 used unknown); see @ref{Procedures}. Register parameter (GNU) (symbol
3451 type @code{N_PSYM}); see @ref{Parameters}. Prototype of function
3452 referenced by this file (Sun @code{acc}) (symbol type @code{N_FUN}).
3455 Static Procedure; see @ref{Procedures}.
3458 Register parameter; see @ref{Register Parameters}.
3461 Register variable; see @ref{Register Variables}.
3464 File scope variable; see @ref{Statics}.
3467 Local variable (OS9000).
3470 Type name; see @ref{Typedefs}.
3473 Enumeration, structure, or union tag; see @ref{Typedefs}.
3476 Parameter passed by reference; see @ref{Reference Parameters}.
3479 Procedure scope static variable; see @ref{Statics}.
3482 Conformant array; see @ref{Conformant Arrays}.
3485 Function return variable; see @ref{Parameters}.
3488 @node Type Descriptors
3489 @appendix Table of Type Descriptors
3491 The type descriptor is the character which follows the type number and
3492 an equals sign. It specifies what kind of type is being defined.
3493 @xref{String Field}, for more information about their use.
3498 Type reference; see @ref{String Field}.
3501 Reference to builtin type; see @ref{Negative Type Numbers}.
3504 Method (C++); see @ref{Method Type Descriptor}.
3507 Pointer; see @ref{Miscellaneous Types}.
3513 Type Attributes (AIX); see @ref{String Field}. Member (class and variable)
3514 type (GNU C++); see @ref{Member Type Descriptor}.
3517 Array; see @ref{Arrays}.
3520 Open array; see @ref{Arrays}.
3523 Pascal space type (AIX); see @ref{Miscellaneous Types}. Builtin integer
3524 type (Sun); see @ref{Builtin Type Descriptors}. Const and volatile
3525 qualified type (OS9000).
3528 Volatile-qualified type; see @ref{Miscellaneous Types}.
3531 Complex builtin type (AIX); see @ref{Builtin Type Descriptors}.
3532 Const-qualified type (OS9000).
3535 COBOL Picture type. See AIX documentation for details.
3538 File type; see @ref{Miscellaneous Types}.
3541 N-dimensional dynamic array; see @ref{Arrays}.
3544 Enumeration type; see @ref{Enumerations}.
3547 N-dimensional subarray; see @ref{Arrays}.
3550 Function type; see @ref{Function Types}.
3553 Pascal function parameter; see @ref{Function Types}
3556 Builtin floating point type; see @ref{Builtin Type Descriptors}.
3559 COBOL Group. See AIX documentation for details.
3562 Imported type (AIX); see @ref{Cross-References}. Volatile-qualified
3566 Const-qualified type; see @ref{Miscellaneous Types}.
3569 COBOL File Descriptor. See AIX documentation for details.
3572 Multiple instance type; see @ref{Miscellaneous Types}.
3575 String type; see @ref{Strings}.
3578 Stringptr; see @ref{Strings}.
3581 Opaque type; see @ref{Typedefs}.
3584 Procedure; see @ref{Function Types}.
3587 Packed array; see @ref{Arrays}.
3590 Range type; see @ref{Subranges}.
3593 Builtin floating type; see @ref{Builtin Type Descriptors} (Sun). Pascal
3594 subroutine parameter; see @ref{Function Types} (AIX). Detecting this
3595 conflict is possible with careful parsing (hint: a Pascal subroutine
3596 parameter type will always contain a comma, and a builtin type
3597 descriptor never will).
3600 Structure type; see @ref{Structures}.
3603 Set type; see @ref{Miscellaneous Types}.
3606 Union; see @ref{Unions}.
3609 Variant record. This is a Pascal and Modula-2 feature which is like a
3610 union within a struct in C. See AIX documentation for details.
3613 Wide character; see @ref{Builtin Type Descriptors}.
3616 Cross-reference; see @ref{Cross-References}.
3619 Used by IBM's xlC C++ compiler (for structures, I think).
3622 gstring; see @ref{Strings}.
3625 @node Expanded Reference
3626 @appendix Expanded Reference by Stab Type
3628 @c FIXME: This appendix should go away; see N_PSYM or N_SO for an example.
3630 For a full list of stab types, and cross-references to where they are
3631 described, see @ref{Stab Types}. This appendix just covers certain
3632 stabs which are not yet described in the main body of this document;
3633 eventually the information will all be in one place.
3637 The first line is the symbol type (see @file{include/aout/stab.def}).
3639 The second line describes the language constructs the symbol type
3642 The third line is the stab format with the significant stab fields
3643 named and the rest NIL.
3645 Subsequent lines expand upon the meaning and possible values for each
3646 significant stab field.
3648 Finally, any further information.
3651 * N_PC:: Pascal global symbol
3652 * N_NSYMS:: Number of symbols
3653 * N_NOMAP:: No DST map
3654 * N_M2C:: Modula-2 compilation unit
3655 * N_BROWS:: Path to .cb file for Sun source code browser
3656 * N_DEFD:: GNU Modula2 definition module dependency
3657 * N_EHDECL:: GNU C++ exception variable
3658 * N_MOD2:: Modula2 information "for imc"
3659 * N_CATCH:: GNU C++ "catch" clause
3660 * N_SSYM:: Structure or union element
3661 * N_SCOPE:: Modula2 scope information (Sun only)
3662 * Gould:: non-base register symbols used on Gould systems
3663 * N_LENG:: Length of preceding entry
3669 @deffn @code{.stabs} N_PC
3671 Global symbol (for Pascal).
3674 "name" -> "symbol_name" <<?>>
3675 value -> supposedly the line number (stab.def is skeptical)
3679 @file{stabdump.c} says:
3681 global pascal symbol: name,,0,subtype,line
3689 @deffn @code{.stabn} N_NSYMS
3691 Number of symbols (according to Ultrix V4.0).
3694 0, files,,funcs,lines (stab.def)
3701 @deffn @code{.stabs} N_NOMAP
3703 No DST map for symbol (according to Ultrix V4.0). I think this means a
3704 variable has been optimized out.
3707 name, ,0,type,ignored (stab.def)
3714 @deffn @code{.stabs} N_M2C
3716 Modula-2 compilation unit.
3719 "string" -> "unit_name,unit_time_stamp[,code_time_stamp]"
3721 value -> 0 (main unit)
3725 See @cite{Dbx and Dbxtool Interfaces}, 2nd edition, by Sun, 1988, for
3733 @deffn @code{.stabs} N_BROWS
3735 Sun source code browser, path to @file{.cb} file
3738 "path to associated @file{.cb} file"
3740 Note: N_BROWS has the same value as N_BSLINE.
3746 @deffn @code{.stabn} N_DEFD
3748 GNU Modula2 definition module dependency.
3750 GNU Modula-2 definition module dependency. The value is the
3751 modification time of the definition file. The other field is non-zero
3752 if it is imported with the GNU M2 keyword @code{%INITIALIZE}. Perhaps
3753 @code{N_M2C} can be used if there are enough empty fields?
3759 @deffn @code{.stabs} N_EHDECL
3761 GNU C++ exception variable <<?>>.
3763 "@var{string} is variable name"
3765 Note: conflicts with @code{N_MOD2}.
3771 @deffn @code{.stab?} N_MOD2
3773 Modula2 info "for imc" (according to Ultrix V4.0)
3775 Note: conflicts with @code{N_EHDECL} <<?>>
3781 @deffn @code{.stabn} N_CATCH
3783 GNU C++ @code{catch} clause
3785 GNU C++ @code{catch} clause. The value is its address. The desc field
3786 is nonzero if this entry is immediately followed by a @code{CAUGHT} stab
3787 saying what exception was caught. Multiple @code{CAUGHT} stabs means
3788 that multiple exceptions can be caught here. If desc is 0, it means all
3789 exceptions are caught here.
3795 @deffn @code{.stabn} N_SSYM
3797 Structure or union element.
3799 The value is the offset in the structure.
3801 <<?looking at structs and unions in C I didn't see these>>
3807 @deffn @code{.stab?} N_SCOPE
3809 Modula2 scope information (Sun linker)
3814 @section Non-base registers on Gould systems
3816 @deffn @code{.stab?} N_NBTEXT
3817 @deffnx @code{.stab?} N_NBDATA
3818 @deffnx @code{.stab?} N_NBBSS
3819 @deffnx @code{.stab?} N_NBSTS
3820 @deffnx @code{.stab?} N_NBLCS
3826 These are used on Gould systems for non-base registers syms.
3828 However, the following values are not the values used by Gould; they are
3829 the values which GNU has been documenting for these values for a long
3830 time, without actually checking what Gould uses. I include these values
3831 only because perhaps some someone actually did something with the GNU
3832 information (I hope not, why GNU knowingly assigned wrong values to
3833 these in the header file is a complete mystery to me).
3836 240 0xf0 N_NBTEXT ??
3837 242 0xf2 N_NBDATA ??
3847 @deffn @code{.stabn} N_LENG
3849 Second symbol entry containing a length-value for the preceding entry.
3850 The value is the length.
3854 @appendix Questions and Anomalies
3858 @c I think this is changed in GCC 2.4.5 to put the line number there.
3859 For GNU C stabs defining local and global variables (@code{N_LSYM} and
3860 @code{N_GSYM}), the desc field is supposed to contain the source
3861 line number on which the variable is defined. In reality the desc
3862 field is always 0. (This behavior is defined in @file{dbxout.c} and
3863 putting a line number in desc is controlled by @samp{#ifdef
3864 WINNING_GDB}, which defaults to false). GDB supposedly uses this
3865 information if you say @samp{list @var{var}}. In reality, @var{var} can
3866 be a variable defined in the program and GDB says @samp{function
3867 @var{var} not defined}.
3870 In GNU C stabs, there seems to be no way to differentiate tag types:
3871 structures, unions, and enums (symbol descriptor @samp{T}) and typedefs
3872 (symbol descriptor @samp{t}) defined at file scope from types defined locally
3873 to a procedure or other more local scope. They all use the @code{N_LSYM}
3874 stab type. Types defined at procedure scope are emitted after the
3875 @code{N_RBRAC} of the preceding function and before the code of the
3876 procedure in which they are defined. This is exactly the same as
3877 types defined in the source file between the two procedure bodies.
3878 GDB over-compensates by placing all types in block #1, the block for
3879 symbols of file scope. This is true for default, @samp{-ansi} and
3880 @samp{-traditional} compiler options. (Bugs gcc/1063, gdb/1066.)
3883 What ends the procedure scope? Is it the proc block's @code{N_RBRAC} or the
3884 next @code{N_FUN}? (I believe its the first.)
3888 @appendix Using Stabs in Their Own Sections
3890 Many object file formats allow tools to create object files with custom
3891 sections containing any arbitrary data. For any such object file
3892 format, stabs can be embedded in special sections. This is how stabs
3893 are used with ELF and SOM, and aside from ECOFF and XCOFF, is how stabs
3897 * Stab Section Basics:: How to embed stabs in sections
3898 * ELF Linker Relocation:: Sun ELF hacks
3901 @node Stab Section Basics
3902 @appendixsec How to Embed Stabs in Sections
3904 The assembler creates two custom sections, a section named @code{.stab}
3905 which contains an array of fixed length structures, one struct per stab,
3906 and a section named @code{.stabstr} containing all the variable length
3907 strings that are referenced by stabs in the @code{.stab} section. The
3908 byte order of the stabs binary data depends on the object file format.
3909 For ELF, it matches the byte order of the ELF file itself, as determined
3910 from the @code{EI_DATA} field in the @code{e_ident} member of the ELF
3911 header. For SOM, it is always big-endian (is this true??? FIXME). For
3912 COFF, it matches the byte order of the COFF headers. The meaning of the
3913 fields is the same as for a.out (@pxref{Symbol Table Format}), except
3914 that the @code{n_strx} field is relative to the strings for the current
3915 compilation unit (which can be found using the synthetic N_UNDF stab
3916 described below), rather than the entire string table.
3918 The first stab in the @code{.stab} section for each compilation unit is
3919 synthetic, generated entirely by the assembler, with no corresponding
3920 @code{.stab} directive as input to the assembler. This stab contains
3921 the following fields:
3925 Offset in the @code{.stabstr} section to the source filename.
3931 Unused field, always zero.
3932 This may eventually be used to hold overflows from the count in
3933 the @code{n_desc} field.
3936 Count of upcoming symbols, i.e., the number of remaining stabs for this
3940 Size of the string table fragment associated with this source file, in
3944 The @code{.stabstr} section always starts with a null byte (so that string
3945 offsets of zero reference a null string), followed by random length strings,
3946 each of which is null byte terminated.
3948 The ELF section header for the @code{.stab} section has its
3949 @code{sh_link} member set to the section number of the @code{.stabstr}
3950 section, and the @code{.stabstr} section has its ELF section
3951 header @code{sh_type} member set to @code{SHT_STRTAB} to mark it as a
3952 string table. SOM and COFF have no way of linking the sections together
3953 or marking them as string tables.
3955 For COFF, the @code{.stab} and @code{.stabstr} sections may be simply
3956 concatenated by the linker. GDB then uses the @code{n_desc} fields to
3957 figure out the extent of the original sections. Similarly, the
3958 @code{n_value} fields of the header symbols are added together in order
3959 to get the actual position of the strings in a desired @code{.stabstr}
3960 section. Although this design obviates any need for the linker to
3961 relocate or otherwise manipulate @code{.stab} and @code{.stabstr}
3962 sections, it also requires some care to ensure that the offsets are
3963 calculated correctly. For instance, if the linker were to pad in
3964 between the @code{.stabstr} sections before concatenating, then the
3965 offsets to strings in the middle of the executable's @code{.stabstr}
3966 section would be wrong.
3968 The GNU linker is able to optimize stabs information by merging
3969 duplicate strings and removing duplicate header file information
3970 (@pxref{Include Files}). When some versions of the GNU linker optimize
3971 stabs in sections, they remove the leading @code{N_UNDF} symbol and
3972 arranges for all the @code{n_strx} fields to be relative to the start of
3973 the @code{.stabstr} section.
3975 @node ELF Linker Relocation
3976 @appendixsec Having the Linker Relocate Stabs in ELF
3978 This section describes some Sun hacks for Stabs in ELF; it does not
3979 apply to COFF or SOM.
3981 To keep linking fast, you don't want the linker to have to relocate very
3982 many stabs. Making sure this is done for @code{N_SLINE},
3983 @code{N_RBRAC}, and @code{N_LBRAC} stabs is the most important thing
3984 (see the descriptions of those stabs for more information). But Sun's
3985 stabs in ELF has taken this further, to make all addresses in the
3986 @code{n_value} field (functions and static variables) relative to the
3987 source file. For the @code{N_SO} symbol itself, Sun simply omits the
3988 address. To find the address of each section corresponding to a given
3989 source file, the compiler puts out symbols giving the address of each
3990 section for a given source file. Since these are ELF (not stab)
3991 symbols, the linker relocates them correctly without having to touch the
3992 stabs section. They are named @code{Bbss.bss} for the bss section,
3993 @code{Ddata.data} for the data section, and @code{Drodata.rodata} for
3994 the rodata section. For the text section, there is no such symbol (but
3995 there should be, see below). For an example of how these symbols work,
3996 @xref{Stab Section Transformations}. GCC does not provide these symbols;
3997 it instead relies on the stabs getting relocated. Thus addresses which
3998 would normally be relative to @code{Bbss.bss}, etc., are already
3999 relocated. The Sun linker provided with Solaris 2.2 and earlier
4000 relocates stabs using normal ELF relocation information, as it would do
4001 for any section. Sun has been threatening to kludge their linker to not
4002 do this (to speed up linking), even though the correct way to avoid
4003 having the linker do these relocations is to have the compiler no longer
4004 output relocatable values. Last I heard they had been talked out of the
4005 linker kludge. See Sun point patch 101052-01 and Sun bug 1142109. With
4006 the Sun compiler this affects @samp{S} symbol descriptor stabs
4007 (@pxref{Statics}) and functions (@pxref{Procedures}). In the latter
4008 case, to adopt the clean solution (making the value of the stab relative
4009 to the start of the compilation unit), it would be necessary to invent a
4010 @code{Ttext.text} symbol, analogous to the @code{Bbss.bss}, etc.,
4011 symbols. I recommend this rather than using a zero value and getting
4012 the address from the ELF symbols.
4014 Finding the correct @code{Bbss.bss}, etc., symbol is difficult, because
4015 the linker simply concatenates the @code{.stab} sections from each
4016 @file{.o} file without including any information about which part of a
4017 @code{.stab} section comes from which @file{.o} file. The way GDB does
4018 this is to look for an ELF @code{STT_FILE} symbol which has the same
4019 name as the last component of the file name from the @code{N_SO} symbol
4020 in the stabs (for example, if the file name is @file{../../gdb/main.c},
4021 it looks for an ELF @code{STT_FILE} symbol named @code{main.c}). This
4022 loses if different files have the same name (they could be in different
4023 directories, a library could have been copied from one system to
4024 another, etc.). It would be much cleaner to have the @code{Bbss.bss}
4025 symbols in the stabs themselves. Having the linker relocate them there
4026 is no more work than having the linker relocate ELF symbols, and it
4027 solves the problem of having to associate the ELF and stab symbols.
4028 However, no one has yet designed or implemented such a scheme.
4032 @node Symbol Types Index
4033 @unnumbered Symbol Types Index
4037 @c TeX can handle the contents at the start but makeinfo 3.12 can not