1 This is Info file stabs.info, produced by Makeinfo version 1.68 from
2 the input file ./stabs.texinfo.
5 * Stabs: (stabs). The "stabs" debugging information format.
8 This document describes the stabs debugging symbol tables.
10 Copyright 1992, 93, 94, 95, 97, 1998 Free Software Foundation, Inc.
11 Contributed by Cygnus Support. Written by Julia Menapace, Jim Kingdon,
14 Permission is granted to make and distribute verbatim copies of this
15 manual provided the copyright notice and this permission notice are
16 preserved on all copies.
18 Permission is granted to copy or distribute modified versions of this
19 manual under the terms of the GPL (for which purpose this text may be
20 regarded as a program in the language TeX).
23 File: stabs.info, Node: Conformant Arrays, Prev: Reference Parameters, Up: Parameters
25 Passing Conformant Array Parameters
26 -----------------------------------
28 Conformant arrays are a feature of Modula-2, and perhaps other
29 languages, in which the size of an array parameter is not known to the
30 called function until run-time. Such parameters have two stabs: a `x'
31 for the array itself, and a `C', which represents the size of the
32 array. The value of the `x' stab is the offset in the argument list
33 where the address of the array is stored (it this right? it is a
34 guess); the value of the `C' stab is the offset in the argument list
35 where the size of the array (in elements? in bytes?) is stored.
38 File: stabs.info, Node: Types, Next: Symbol Tables, Prev: Variables, Up: Top
43 The examples so far have described types as references to previously
44 defined types, or defined in terms of subranges of or pointers to
45 previously defined types. This chapter describes the other type
46 descriptors that may follow the `=' in a type definition.
50 * Builtin Types:: Integers, floating point, void, etc.
51 * Miscellaneous Types:: Pointers, sets, files, etc.
52 * Cross-References:: Referring to a type not yet defined.
53 * Subranges:: A type with a specific range.
54 * Arrays:: An aggregate type of same-typed elements.
55 * Strings:: Like an array but also has a length.
56 * Enumerations:: Like an integer but the values have names.
57 * Structures:: An aggregate type of different-typed elements.
58 * Typedefs:: Giving a type a name.
59 * Unions:: Different types sharing storage.
63 File: stabs.info, Node: Builtin Types, Next: Miscellaneous Types, Up: Types
68 Certain types are built in (`int', `short', `void', `float', etc.);
69 the debugger recognizes these types and knows how to handle them.
70 Thus, don't be surprised if some of the following ways of specifying
71 builtin types do not specify everything that a debugger would need to
72 know about the type--in some cases they merely specify enough
73 information to distinguish the type from other types.
75 The traditional way to define builtin types is convolunted, so new
76 ways have been invented to describe them. Sun's `acc' uses special
77 builtin type descriptors (`b' and `R'), and IBM uses negative type
78 numbers. GDB accepts all three ways, as of version 4.8; dbx just
79 accepts the traditional builtin types and perhaps one of the other two
80 formats. The following sections describe each of these formats.
84 * Traditional Builtin Types:: Put on your seatbelts and prepare for kludgery
85 * Builtin Type Descriptors:: Builtin types with special type descriptors
86 * Negative Type Numbers:: Builtin types using negative type numbers
89 File: stabs.info, Node: Traditional Builtin Types, Next: Builtin Type Descriptors, Up: Builtin Types
91 Traditional Builtin Types
92 -------------------------
94 This is the traditional, convoluted method for defining builtin
95 types. There are several classes of such type definitions: integer,
96 floating point, and `void'.
100 * Traditional Integer Types::
101 * Traditional Other Types::
104 File: stabs.info, Node: Traditional Integer Types, Next: Traditional Other Types, Up: Traditional Builtin Types
106 Traditional Integer Types
107 .........................
109 Often types are defined as subranges of themselves. If the bounding
110 values fit within an `int', then they are given normally. For example:
112 .stabs "int:t1=r1;-2147483648;2147483647;",128,0,0,0 # 128 is N_LSYM
113 .stabs "char:t2=r2;0;127;",128,0,0,0
115 Builtin types can also be described as subranges of `int':
117 .stabs "unsigned short:t6=r1;0;65535;",128,0,0,0
119 If the lower bound of a subrange is 0 and the upper bound is -1, the
120 type is an unsigned integral type whose bounds are too big to describe
121 in an `int'. Traditionally this is only used for `unsigned int' and
124 .stabs "unsigned int:t4=r1;0;-1;",128,0,0,0
126 For larger types, GCC 2.4.5 puts out bounds in octal, with one or
127 more leading zeroes. In this case a negative bound consists of a number
128 which is a 1 bit (for the sign bit) followed by a 0 bit for each bit in
129 the number (except the sign bit), and a positive bound is one which is a
130 1 bit for each bit in the number (except possibly the sign bit). All
131 known versions of dbx and GDB version 4 accept this (at least in the
132 sense of not refusing to process the file), but GDB 3.5 refuses to read
133 the whole file containing such symbols. So GCC 2.3.3 did not output the
134 proper size for these types. As an example of octal bounds, the string
135 fields of the stabs for 64 bit integer types look like:
137 long int:t3=r1;001000000000000000000000;000777777777777777777777;
138 long unsigned int:t5=r1;000000000000000000000000;001777777777777777777777;
140 If the lower bound of a subrange is 0 and the upper bound is
141 negative, the type is an unsigned integral type whose size in bytes is
142 the absolute value of the upper bound. I believe this is a Convex
143 convention for `unsigned long long'.
145 If the lower bound of a subrange is negative and the upper bound is
146 0, the type is a signed integral type whose size in bytes is the
147 absolute value of the lower bound. I believe this is a Convex
148 convention for `long long'. To distinguish this from a legitimate
149 subrange, the type should be a subrange of itself. I'm not sure whether
150 this is the case for Convex.
153 File: stabs.info, Node: Traditional Other Types, Prev: Traditional Integer Types, Up: Traditional Builtin Types
155 Traditional Other Types
156 .......................
158 If the upper bound of a subrange is 0 and the lower bound is
159 positive, the type is a floating point type, and the lower bound of the
160 subrange indicates the number of bytes in the type:
162 .stabs "float:t12=r1;4;0;",128,0,0,0
163 .stabs "double:t13=r1;8;0;",128,0,0,0
165 However, GCC writes `long double' the same way it writes `double',
166 so there is no way to distinguish.
168 .stabs "long double:t14=r1;8;0;",128,0,0,0
170 Complex types are defined the same way as floating-point types;
171 there is no way to distinguish a single-precision complex from a
172 double-precision floating-point type.
174 The C `void' type is defined as itself:
176 .stabs "void:t15=15",128,0,0,0
178 I'm not sure how a boolean type is represented.
181 File: stabs.info, Node: Builtin Type Descriptors, Next: Negative Type Numbers, Prev: Traditional Builtin Types, Up: Builtin Types
183 Defining Builtin Types Using Builtin Type Descriptors
184 -----------------------------------------------------
186 This is the method used by Sun's `acc' for defining builtin types.
187 These are the type descriptors to define builtin types:
189 `b SIGNED CHAR-FLAG WIDTH ; OFFSET ; NBITS ;'
190 Define an integral type. SIGNED is `u' for unsigned or `s' for
191 signed. CHAR-FLAG is `c' which indicates this is a character
192 type, or is omitted. I assume this is to distinguish an integral
193 type from a character type of the same size, for example it might
194 make sense to set it for the C type `wchar_t' so the debugger can
195 print such variables differently (Solaris does not do this). Sun
196 sets it on the C types `signed char' and `unsigned char' which
197 arguably is wrong. WIDTH and OFFSET appear to be for small
198 objects stored in larger ones, for example a `short' in an `int'
199 register. WIDTH is normally the number of bytes in the type.
200 OFFSET seems to always be zero. NBITS is the number of bits in
203 Note that type descriptor `b' used for builtin types conflicts with
204 its use for Pascal space types (*note Miscellaneous Types::.);
205 they can be distinguished because the character following the type
206 descriptor will be a digit, `(', or `-' for a Pascal space type, or
207 `u' or `s' for a builtin type.
210 Documented by AIX to define a wide character type, but their
211 compiler actually uses negative type numbers (*note Negative Type
214 `R FP-TYPE ; BYTES ;'
215 Define a floating point type. FP-TYPE has one of the following
219 IEEE 32-bit (single precision) floating point format.
222 IEEE 64-bit (double precision) floating point format.
229 These are for complex numbers. A comment in the GDB source
230 describes them as Fortran `complex', `double complex', and
231 `complex*16', respectively, but what does that mean? (i.e.,
232 Single precision? Double precison?).
235 Long double. This should probably only be used for Sun format
236 `long double', and new codes should be used for other floating
237 point formats (`NF_DOUBLE' can be used if a `long double' is
238 really just an IEEE double, of course).
240 BYTES is the number of bytes occupied by the type. This allows a
241 debugger to perform some operations with the type even if it
242 doesn't understand FP-TYPE.
244 `g TYPE-INFORMATION ; NBITS'
245 Documented by AIX to define a floating type, but their compiler
246 actually uses negative type numbers (*note Negative Type
249 `c TYPE-INFORMATION ; NBITS'
250 Documented by AIX to define a complex type, but their compiler
251 actually uses negative type numbers (*note Negative Type
254 The C `void' type is defined as a signed integral type 0 bits long:
255 .stabs "void:t19=bs0;0;0",128,0,0,0
256 The Solaris compiler seems to omit the trailing semicolon in this
257 case. Getting sloppy in this way is not a swift move because if a type
258 is embedded in a more complex expression it is necessary to be able to
261 I'm not sure how a boolean type is represented.
264 File: stabs.info, Node: Negative Type Numbers, Prev: Builtin Type Descriptors, Up: Builtin Types
266 Negative Type Numbers
267 ---------------------
269 This is the method used in XCOFF for defining builtin types. Since
270 the debugger knows about the builtin types anyway, the idea of negative
271 type numbers is simply to give a special type number which indicates
272 the builtin type. There is no stab defining these types.
274 There are several subtle issues with negative type numbers.
276 One is the size of the type. A builtin type (for example the C types
277 `int' or `long') might have different sizes depending on compiler
278 options, the target architecture, the ABI, etc. This issue doesn't
279 come up for IBM tools since (so far) they just target the RS/6000; the
280 sizes indicated below for each size are what the IBM RS/6000 tools use.
281 To deal with differing sizes, either define separate negative type
282 numbers for each size (which works but requires changing the debugger,
283 and, unless you get both AIX dbx and GDB to accept the change,
284 introduces an incompatibility), or use a type attribute (*note String
285 Field::.) to define a new type with the appropriate size (which merely
286 requires a debugger which understands type attributes, like AIX dbx or
289 .stabs "boolean:t10=@s8;-16",128,0,0,0
291 defines an 8-bit boolean type, and
293 .stabs "boolean:t10=@s64;-16",128,0,0,0
295 defines a 64-bit boolean type.
297 A similar issue is the format of the type. This comes up most often
298 for floating-point types, which could have various formats (particularly
299 extended doubles, which vary quite a bit even among IEEE systems).
300 Again, it is best to define a new negative type number for each
301 different format; changing the format based on the target system has
302 various problems. One such problem is that the Alpha has both VAX and
303 IEEE floating types. One can easily imagine one library using the VAX
304 types and another library in the same executable using the IEEE types.
305 Another example is that the interpretation of whether a boolean is true
306 or false can be based on the least significant bit, most significant
307 bit, whether it is zero, etc., and different compilers (or different
308 options to the same compiler) might provide different kinds of boolean.
310 The last major issue is the names of the types. The name of a given
311 type depends *only* on the negative type number given; these do not
312 vary depending on the language, the target system, or anything else.
313 One can always define separate type numbers--in the following list you
314 will see for example separate `int' and `integer*4' types which are
315 identical except for the name. But compatibility can be maintained by
316 not inventing new negative type numbers and instead just defining a new
317 type with a new name. For example:
319 .stabs "CARDINAL:t10=-8",128,0,0,0
321 Here is the list of negative type numbers. The phrase "integral
322 type" is used to mean twos-complement (I strongly suspect that all
323 machines which use stabs use twos-complement; most machines use
324 twos-complement these days).
327 `int', 32 bit signed integral type.
330 `char', 8 bit type holding a character. Both GDB and dbx on AIX
331 treat this as signed. GCC uses this type whether `char' is signed
332 or not, which seems like a bad idea. The AIX compiler (`xlc')
333 seems to avoid this type; it uses -5 instead for `char'.
336 `short', 16 bit signed integral type.
339 `long', 32 bit signed integral type.
342 `unsigned char', 8 bit unsigned integral type.
345 `signed char', 8 bit signed integral type.
348 `unsigned short', 16 bit unsigned integral type.
351 `unsigned int', 32 bit unsigned integral type.
354 `unsigned', 32 bit unsigned integral type.
357 `unsigned long', 32 bit unsigned integral type.
360 `void', type indicating the lack of a value.
363 `float', IEEE single precision.
366 `double', IEEE double precision.
369 `long double', IEEE double precision. The compiler claims the size
370 will increase in a future release, and for binary compatibility
371 you have to avoid using `long double'. I hope when they increase
372 it they use a new negative type number.
375 `integer'. 32 bit signed integral type.
378 `boolean'. 32 bit type. GDB and GCC assume that zero is false,
379 one is true, and other values have unspecified meaning. I hope
380 this agrees with how the IBM tools use the type.
383 `short real'. IEEE single precision.
386 `real'. IEEE double precision.
389 `stringptr'. *Note Strings::.
392 `character', 8 bit unsigned character type.
395 `logical*1', 8 bit type. This Fortran type has a split
396 personality in that it is used for boolean variables, but can also
397 be used for unsigned integers. 0 is false, 1 is true, and other
398 values are non-boolean.
401 `logical*2', 16 bit type. This Fortran type has a split
402 personality in that it is used for boolean variables, but can also
403 be used for unsigned integers. 0 is false, 1 is true, and other
404 values are non-boolean.
407 `logical*4', 32 bit type. This Fortran type has a split
408 personality in that it is used for boolean variables, but can also
409 be used for unsigned integers. 0 is false, 1 is true, and other
410 values are non-boolean.
413 `logical', 32 bit type. This Fortran type has a split personality
414 in that it is used for boolean variables, but can also be used for
415 unsigned integers. 0 is false, 1 is true, and other values are
419 `complex'. A complex type consisting of two IEEE single-precision
420 floating point values.
423 `complex'. A complex type consisting of two IEEE double-precision
424 floating point values.
427 `integer*1', 8 bit signed integral type.
430 `integer*2', 16 bit signed integral type.
433 `integer*4', 32 bit signed integral type.
436 `wchar'. Wide character, 16 bits wide, unsigned (what format?
440 `long long', 64 bit signed integral type.
443 `unsigned long long', 64 bit unsigned integral type.
446 `logical*8', 64 bit unsigned integral type.
449 `integer*8', 64 bit signed integral type.
452 File: stabs.info, Node: Miscellaneous Types, Next: Cross-References, Prev: Builtin Types, Up: Types
457 `b TYPE-INFORMATION ; BYTES'
458 Pascal space type. This is documented by IBM; what does it mean?
460 This use of the `b' type descriptor can be distinguished from its
461 use for builtin integral types (*note Builtin Type Descriptors::.)
462 because the character following the type descriptor is always a
466 A volatile-qualified version of TYPE-INFORMATION. This is a Sun
467 extension. References and stores to a variable with a
468 volatile-qualified type must not be optimized or cached; they must
469 occur as the user specifies them.
472 File of type TYPE-INFORMATION. As far as I know this is only used
476 A const-qualified version of TYPE-INFORMATION. This is a Sun
477 extension. A variable with a const-qualified type cannot be
480 `M TYPE-INFORMATION ; LENGTH'
481 Multiple instance type. The type seems to composed of LENGTH
482 repetitions of TYPE-INFORMATION, for example `character*3' is
483 represented by `M-2;3', where `-2' is a reference to a character
484 type (*note Negative Type Numbers::.). I'm not sure how this
485 differs from an array. This appears to be a Fortran feature.
486 LENGTH is a bound, like those in range types; see *Note
490 Pascal set type. TYPE-INFORMATION must be a small type such as an
491 enumeration or a subrange, and the type is a bitmask whose length
492 is specified by the number of elements in TYPE-INFORMATION.
494 In CHILL, if it is a bitstring instead of a set, also use the `S'
495 type attribute (*note String Field::.).
498 Pointer to TYPE-INFORMATION.
501 File: stabs.info, Node: Cross-References, Next: Subranges, Prev: Miscellaneous Types, Up: Types
503 Cross-References to Other Types
504 ===============================
506 A type can be used before it is defined; one common way to deal with
507 that situation is just to use a type reference to a type which has not
510 Another way is with the `x' type descriptor, which is followed by
511 `s' for a structure tag, `u' for a union tag, or `e' for a enumerator
512 tag, followed by the name of the tag, followed by `:'. If the name
513 contains `::' between a `<' and `>' pair (for C++ templates), such a
514 `::' does not end the name--only a single `:' ends the name; see *Note
517 For example, the following C declarations:
524 .stabs "bar:G16=*17=xsfoo:",32,0,0,0
526 Not all debuggers support the `x' type descriptor, so on some
527 machines GCC does not use it. I believe that for the above example it
528 would just emit a reference to type 17 and never define it, but I
529 haven't verified that.
531 Modula-2 imported types, at least on AIX, use the `i' type
532 descriptor, which is followed by the name of the module from which the
533 type is imported, followed by `:', followed by the name of the type.
534 There is then optionally a comma followed by type information for the
535 type. This differs from merely naming the type (*note Typedefs::.) in
536 that it identifies the module; I don't understand whether the name of
537 the type given here is always just the same as the name we are giving
538 it, or whether this type descriptor is used with a nameless stab (*note
539 String Field::.), or what. The symbol ends with `;'.
542 File: stabs.info, Node: Subranges, Next: Arrays, Prev: Cross-References, Up: Types
547 The `r' type descriptor defines a type as a subrange of another
548 type. It is followed by type information for the type of which it is a
549 subrange, a semicolon, an integral lower bound, a semicolon, an
550 integral upper bound, and a semicolon. The AIX documentation does not
551 specify the trailing semicolon, in an effort to specify array indexes
552 more cleanly, but a subrange which is not an array index has always
553 included a trailing semicolon (*note Arrays::.).
555 Instead of an integer, either bound can be one of the following:
558 The bound is passed by reference on the stack at offset OFFSET
559 from the argument list. *Note Parameters::, for more information
563 The bound is passed by value on the stack at offset OFFSET from
567 The bound is pased by reference in register number REGISTER-NUMBER.
570 The bound is passed by value in register number REGISTER-NUMBER.
575 Subranges are also used for builtin types; see *Note Traditional
579 File: stabs.info, Node: Arrays, Next: Strings, Prev: Subranges, Up: Types
584 Arrays use the `a' type descriptor. Following the type descriptor
585 is the type of the index and the type of the array elements. If the
586 index type is a range type, it ends in a semicolon; otherwise (for
587 example, if it is a type reference), there does not appear to be any
588 way to tell where the types are separated. In an effort to clean up
589 this mess, IBM documents the two types as being separated by a
590 semicolon, and a range type as not ending in a semicolon (but this is
591 not right for range types which are not array indexes, *note
592 Subranges::.). I think probably the best solution is to specify that a
593 semicolon ends a range type, and that the index type and element type
594 of an array are separated by a semicolon, but that if the index type is
595 a range type, the extra semicolon can be omitted. GDB (at least
596 through version 4.9) doesn't support any kind of index type other than a
597 range anyway; I'm not sure about dbx.
599 It is well established, and widely used, that the type of the index,
600 unlike most types found in the stabs, is merely a type definition, not
601 type information (*note String Field::.) (that is, it need not start
602 with `TYPE-NUMBER=' if it is defining a new type). According to a
603 comment in GDB, this is also true of the type of the array elements; it
604 gives `ar1;1;10;ar1;1;10;4' as a legitimate way to express a two
605 dimensional array. According to AIX documentation, the element type
606 must be type information. GDB accepts either.
608 The type of the index is often a range type, expressed as the type
609 descriptor `r' and some parameters. It defines the size of the array.
610 In the example below, the range `r1;0;2;' defines an index type which
611 is a subrange of type 1 (integer), with a lower bound of 0 and an upper
612 bound of 2. This defines the valid range of subscripts of a
613 three-element C array.
615 For example, the definition:
617 char char_vec[3] = {'a','b','c'};
621 .stabs "char_vec:G19=ar1;0;2;2",32,0,0,0
629 If an array is "packed", the elements are spaced more closely than
630 normal, saving memory at the expense of speed. For example, an array
631 of 3-byte objects might, if unpacked, have each element aligned on a
632 4-byte boundary, but if packed, have no padding. One way to specify
633 that something is packed is with type attributes (*note String
634 Field::.). In the case of arrays, another is to use the `P' type
635 descriptor instead of `a'. Other than specifying a packed array, `P'
638 An open array is represented by the `A' type descriptor followed by
639 type information specifying the type of the array elements.
641 An N-dimensional dynamic array is represented by
643 D DIMENSIONS ; TYPE-INFORMATION
645 DIMENSIONS is the number of dimensions; TYPE-INFORMATION specifies
646 the type of the array elements.
648 A subarray of an N-dimensional array is represented by
650 E DIMENSIONS ; TYPE-INFORMATION
652 DIMENSIONS is the number of dimensions; TYPE-INFORMATION specifies
653 the type of the array elements.
656 File: stabs.info, Node: Strings, Next: Enumerations, Prev: Arrays, Up: Types
661 Some languages, like C or the original Pascal, do not have string
662 types, they just have related things like arrays of characters. But
663 most Pascals and various other languages have string types, which are
664 indicated as follows:
666 `n TYPE-INFORMATION ; BYTES'
667 BYTES is the maximum length. I'm not sure what TYPE-INFORMATION
668 is; I suspect that it means that this is a string of
669 TYPE-INFORMATION (thus allowing a string of integers, a string of
670 wide characters, etc., as well as a string of characters). Not
671 sure what the format of this type is. This is an AIX feature.
673 `z TYPE-INFORMATION ; BYTES'
674 Just like `n' except that this is a gstring, not an ordinary
675 string. I don't know the difference.
678 Pascal Stringptr. What is this? This is an AIX feature.
680 Languages, such as CHILL which have a string type which is basically
681 just an array of characters use the `S' type attribute (*note String
685 File: stabs.info, Node: Enumerations, Next: Structures, Prev: Strings, Up: Types
690 Enumerations are defined with the `e' type descriptor.
692 The source line below declares an enumeration type at file scope.
693 The type definition is located after the `N_RBRAC' that marks the end of
694 the previous procedure's block scope, and before the `N_FUN' that marks
695 the beginning of the next procedure's block scope. Therefore it does
696 not describe a block local symbol, but a file local one.
700 enum e_places {first,second=3,last};
702 generates the following stab:
704 .stabs "e_places:T22=efirst:0,second:3,last:4,;",128,0,0,0
706 The symbol descriptor (`T') says that the stab describes a
707 structure, enumeration, or union tag. The type descriptor `e',
708 following the `22=' of the type definition narrows it down to an
709 enumeration type. Following the `e' is a list of the elements of the
710 enumeration. The format is `NAME:VALUE,'. The list of elements ends
711 with `;'. The fact that VALUE is specified as an integer can cause
712 problems if the value is large. GCC 2.5.2 tries to output it in octal
713 in that case with a leading zero, which is probably a good thing,
714 although GDB 4.11 supports octal only in cases where decimal is
715 perfectly good. Negative decimal values are supported by both GDB and
718 There is no standard way to specify the size of an enumeration type;
719 it is determined by the architecture (normally all enumerations types
720 are 32 bits). Type attributes can be used to specify an enumeration
721 type of another size for debuggers which support them; see *Note String
724 Enumeration types are unusual in that they define symbols for the
725 enumeration values (`first', `second', and `third' in the above
726 example), and even though these symbols are visible in the file as a
727 whole (rather than being in a more local namespace like structure
728 member names), they are defined in the type definition for the
729 enumeration type rather than each having their own symbol. In order to
730 be fast, GDB will only get symbols from such types (in its initial scan
731 of the stabs) if the type is the first thing defined after a `T' or `t'
732 symbol descriptor (the above example fulfills this requirement). If
733 the type does not have a name, the compiler should emit it in a
734 nameless stab (*note String Field::.); GCC does this.
737 File: stabs.info, Node: Structures, Next: Typedefs, Prev: Enumerations, Up: Types
742 The encoding of structures in stabs can be shown with an example.
744 The following source code declares a structure tag and defines an
745 instance of the structure in global scope. Then a `typedef' equates the
746 structure tag with a new type. Seperate stabs are generated for the
747 structure tag, the structure `typedef', and the structure instance. The
748 stabs for the tag and the `typedef' are emited when the definitions are
749 encountered. Since the structure elements are not initialized, the
750 stab and code for the structure variable itself is located at the end
751 of the program in the bss section.
757 struct s_tag* s_next;
760 typedef struct s_tag s_typedef;
762 The structure tag has an `N_LSYM' stab type because, like the
763 enumeration, the symbol has file scope. Like the enumeration, the
764 symbol descriptor is `T', for enumeration, structure, or tag type. The
765 type descriptor `s' following the `16=' of the type definition narrows
766 the symbol type to structure.
768 Following the `s' type descriptor is the number of bytes the
769 structure occupies, followed by a description of each structure element.
770 The structure element descriptions are of the form NAME:TYPE, BIT
771 OFFSET FROM THE START OF THE STRUCT, NUMBER OF BITS IN THE ELEMENT.
774 .stabs "s_tag:T16=s20s_int:1,0,32;s_float:12,32,32;
775 s_char_vec:17=ar1;0;7;2,64,64;s_next:18=*16,128,32;;",128,0,0,0
777 In this example, the first two structure elements are previously
778 defined types. For these, the type following the `NAME:' part of the
779 element description is a simple type reference. The other two structure
780 elements are new types. In this case there is a type definition
781 embedded after the `NAME:'. The type definition for the array element
782 looks just like a type definition for a standalone array. The `s_next'
783 field is a pointer to the same kind of structure that the field is an
784 element of. So the definition of structure type 16 contains a type
785 definition for an element which is a pointer to type 16.
787 If a field is a static member (this is a C++ feature in which a
788 single variable appears to be a field of every structure of a given
789 type) it still starts out with the field name, a colon, and the type,
790 but then instead of a comma, bit position, comma, and bit size, there
791 is a colon followed by the name of the variable which each such field
794 If the structure has methods (a C++ feature), they follow the
795 non-method fields; see *Note Cplusplus::.
798 File: stabs.info, Node: Typedefs, Next: Unions, Prev: Structures, Up: Types
803 To give a type a name, use the `t' symbol descriptor. The type is
804 specified by the type information (*note String Field::.) for the stab.
807 .stabs "s_typedef:t16",128,0,0,0 # 128 is N_LSYM
809 specifies that `s_typedef' refers to type number 16. Such stabs
810 have symbol type `N_LSYM' (or `C_DECL' for XCOFF). (The Sun
811 documentation mentions using `N_GSYM' in some cases).
813 If you are specifying the tag name for a structure, union, or
814 enumeration, use the `T' symbol descriptor instead. I believe C is the
815 only language with this feature.
817 If the type is an opaque type (I believe this is a Modula-2 feature),
818 AIX provides a type descriptor to specify it. The type descriptor is
819 `o' and is followed by a name. I don't know what the name means--is it
820 always the same as the name of the type, or is this type descriptor
821 used with a nameless stab (*note String Field::.)? There optionally
822 follows a comma followed by type information which defines the type of
823 this type. If omitted, a semicolon is used in place of the comma and
824 the type information, and the type is much like a generic pointer
825 type--it has a known size but little else about it is specified.
828 File: stabs.info, Node: Unions, Next: Function Types, Prev: Typedefs, Up: Types
839 This code generates a stab for a union tag and a stab for a union
840 variable. Both use the `N_LSYM' stab type. If a union variable is
841 scoped locally to the procedure in which it is defined, its stab is
842 located immediately preceding the `N_LBRAC' for the procedure's block
845 The stab for the union tag, however, is located preceding the code
846 for the procedure in which it is defined. The stab type is `N_LSYM'.
847 This would seem to imply that the union type is file scope, like the
848 struct type `s_tag'. This is not true. The contents and position of
849 the stab for `u_type' do not convey any infomation about its procedure
853 .stabs "u_tag:T23=u4u_int:1,0,32;u_float:12,0,32;u_char:21,0,32;;",
856 The symbol descriptor `T', following the `name:' means that the stab
857 describes an enumeration, structure, or union tag. The type descriptor
858 `u', following the `23=' of the type definition, narrows it down to a
859 union type definition. Following the `u' is the number of bytes in the
860 union. After that is a list of union element descriptions. Their
861 format is NAME:TYPE, BIT OFFSET INTO THE UNION, NUMBER OF BYTES FOR THE
864 The stab for the union variable is:
866 .stabs "an_u:23",128,0,0,-20 # 128 is N_LSYM
868 `-20' specifies where the variable is stored (*note Stack
872 File: stabs.info, Node: Function Types, Prev: Unions, Up: Types
877 Various types can be defined for function variables. These types are
878 not used in defining functions (*note Procedures::.); they are used for
879 things like pointers to functions.
881 The simple, traditional, type is type descriptor `f' is followed by
882 type information for the return type of the function, followed by a
885 This does not deal with functions for which the number and types of
886 the parameters are part of the type, as in Modula-2 or ANSI C. AIX
887 provides extensions to specify these, using the `f', `F', `p', and `R'
890 First comes the type descriptor. If it is `f' or `F', this type
891 involves a function rather than a procedure, and the type information
892 for the return type of the function follows, followed by a comma. Then
893 comes the number of parameters to the function and a semicolon. Then,
894 for each parameter, there is the name of the parameter followed by a
895 colon (this is only present for type descriptors `R' and `F' which
896 represent Pascal function or procedure parameters), type information
897 for the parameter, a comma, 0 if passed by reference or 1 if passed by
898 value, and a semicolon. The type definition ends with a semicolon.
900 For example, this variable definition:
904 generates the following code:
906 .stabs "g_pf:G24=*25=f1",32,0,0,0
907 .common _g_pf,4,"bss"
909 The variable defines a new type, 24, which is a pointer to another
910 new type, 25, which is a function returning `int'.
913 File: stabs.info, Node: Symbol Tables, Next: Cplusplus, Prev: Types, Up: Top
915 Symbol Information in Symbol Tables
916 ***********************************
918 This chapter describes the format of symbol table entries and how
919 stab assembler directives map to them. It also describes the
920 transformations that the assembler and linker make on data from stabs.
924 * Symbol Table Format::
925 * Transformations On Symbol Tables::
928 File: stabs.info, Node: Symbol Table Format, Next: Transformations On Symbol Tables, Up: Symbol Tables
933 Each time the assembler encounters a stab directive, it puts each
934 field of the stab into a corresponding field in a symbol table entry of
935 its output file. If the stab contains a string field, the symbol table
936 entry for that stab points to a string table entry containing the
937 string data from the stab. Assembler labels become relocatable
938 addresses. Symbol table entries in a.out have the format:
940 struct internal_nlist {
941 unsigned long n_strx; /* index into string table of name */
942 unsigned char n_type; /* type of symbol */
943 unsigned char n_other; /* misc info (usually empty) */
944 unsigned short n_desc; /* description field */
945 bfd_vma n_value; /* value of symbol */
948 If the stab has a string, the `n_strx' field holds the offset in
949 bytes of the string within the string table. The string is terminated
950 by a NUL character. If the stab lacks a string (for example, it was
951 produced by a `.stabn' or `.stabd' directive), the `n_strx' field is
954 Symbol table entries with `n_type' field values greater than 0x1f
955 originated as stabs generated by the compiler (with one random
956 exception). The other entries were placed in the symbol table of the
957 executable by the assembler or the linker.
960 File: stabs.info, Node: Transformations On Symbol Tables, Prev: Symbol Table Format, Up: Symbol Tables
962 Transformations on Symbol Tables
963 ================================
965 The linker concatenates object files and does fixups of externally
968 You can see the transformations made on stab data by the assembler
969 and linker by examining the symbol table after each pass of the build.
970 To do this, use `nm -ap', which dumps the symbol table, including
971 debugging information, unsorted. For stab entries the columns are:
972 VALUE, OTHER, DESC, TYPE, STRING. For assembler and linker symbols,
973 the columns are: VALUE, TYPE, STRING.
975 The low 5 bits of the stab type tell the linker how to relocate the
976 value of the stab. Thus for stab types like `N_RSYM' and `N_LSYM',
977 where the value is an offset or a register number, the low 5 bits are
978 `N_ABS', which tells the linker not to relocate the value.
980 Where the value of a stab contains an assembly language label, it is
981 transformed by each build step. The assembler turns it into a
982 relocatable address and the linker turns it into an absolute address.
986 * Transformations On Static Variables::
987 * Transformations On Global Variables::
988 * Stab Section Transformations:: For some object file formats,
989 things are a bit different.
992 File: stabs.info, Node: Transformations On Static Variables, Next: Transformations On Global Variables, Up: Transformations On Symbol Tables
994 Transformations on Static Variables
995 -----------------------------------
997 This source line defines a static variable at file scope:
999 static int s_g_repeat
1001 The following stab describes the symbol:
1003 .stabs "s_g_repeat:S1",38,0,0,_s_g_repeat
1005 The assembler transforms the stab into this symbol table entry in the
1006 `.o' file. The location is expressed as a data segment offset.
1008 00000084 - 00 0000 STSYM s_g_repeat:S1
1010 In the symbol table entry from the executable, the linker has made the
1011 relocatable address absolute.
1013 0000e00c - 00 0000 STSYM s_g_repeat:S1
1016 File: stabs.info, Node: Transformations On Global Variables, Next: Stab Section Transformations, Prev: Transformations On Static Variables, Up: Transformations On Symbol Tables
1018 Transformations on Global Variables
1019 -----------------------------------
1021 Stabs for global variables do not contain location information. In
1022 this case, the debugger finds location information in the assembler or
1023 linker symbol table entry describing the variable. The source line:
1029 .stabs "g_foo:G2",32,0,0,0
1031 The variable is represented by two symbol table entries in the object
1032 file (see below). The first one originated as a stab. The second one
1033 is an external symbol. The upper case `D' signifies that the `n_type'
1034 field of the symbol table contains 7, `N_DATA' with local linkage. The
1035 stab's value is zero since the value is not used for `N_GSYM' stabs.
1036 The value of the linker symbol is the relocatable address corresponding
1039 00000000 - 00 0000 GSYM g_foo:G2
1042 These entries as transformed by the linker. The linker symbol table
1043 entry now holds an absolute address:
1045 00000000 - 00 0000 GSYM g_foo:G2
1050 File: stabs.info, Node: Stab Section Transformations, Prev: Transformations On Global Variables, Up: Transformations On Symbol Tables
1052 Transformations of Stabs in separate sections
1053 ---------------------------------------------
1055 For object file formats using stabs in separate sections (*note Stab
1056 Sections::.), use `objdump --stabs' instead of `nm' to show the stabs
1057 in an object or executable file. `objdump' is a GNU utility; Sun does
1058 not provide any equivalent.
1060 The following example is for a stab whose value is an address is
1061 relative to the compilation unit (*note ELF Linker Relocation::.). For
1062 example, if the source line
1066 appears within a function, then the assembly language output from the
1071 .stabs "ld:V(0,3)",0x26,0,4,.L18-Ddata.data # 0x26 is N_STSYM
1077 Because the value is formed by subtracting one symbol from another,
1078 the value is absolute, not relocatable, and so the object file contains
1080 Symnum n_type n_othr n_desc n_value n_strx String
1081 31 STSYM 0 4 00000004 680 ld:V(0,3)
1083 without any relocations, and the executable file also contains
1085 Symnum n_type n_othr n_desc n_value n_strx String
1086 31 STSYM 0 4 00000004 680 ld:V(0,3)
1089 File: stabs.info, Node: Cplusplus, Next: Stab Types, Prev: Symbol Tables, Up: Top
1096 * Class Names:: C++ class names are both tags and typedefs.
1097 * Nested Symbols:: C++ symbol names can be within other types.
1098 * Basic Cplusplus Types::
1101 * Methods:: Method definition
1102 * Method Type Descriptor:: The `#' type descriptor
1103 * Member Type Descriptor:: The `@' type descriptor
1105 * Method Modifiers::
1108 * Virtual Base Classes::
1112 File: stabs.info, Node: Class Names, Next: Nested Symbols, Up: Cplusplus
1117 In C++, a class name which is declared with `class', `struct', or
1118 `union', is not only a tag, as in C, but also a type name. Thus there
1119 should be stabs with both `t' and `T' symbol descriptors (*note
1122 To save space, there is a special abbreviation for this case. If the
1123 `T' symbol descriptor is followed by `t', then the stab defines both a
1124 type name and a tag.
1126 For example, the C++ code
1128 struct foo {int x;};
1130 can be represented as either
1132 .stabs "foo:T19=s4x:1,0,32;;",128,0,0,0 # 128 is N_LSYM
1133 .stabs "foo:t19",128,0,0,0
1137 .stabs "foo:Tt19=s4x:1,0,32;;",128,0,0,0
1140 File: stabs.info, Node: Nested Symbols, Next: Basic Cplusplus Types, Prev: Class Names, Up: Cplusplus
1142 Defining a Symbol Within Another Type
1143 =====================================
1145 In C++, a symbol (such as a type name) can be defined within another
1148 In stabs, this is sometimes represented by making the name of a
1149 symbol which contains `::'. Such a pair of colons does not end the name
1150 of the symbol, the way a single colon would (*note String Field::.).
1151 I'm not sure how consistently used or well thought out this mechanism
1152 is. So that a pair of colons in this position always has this meaning,
1153 `:' cannot be used as a symbol descriptor.
1155 For example, if the string for a stab is `foo::bar::baz:t5=*6', then
1156 `foo::bar::baz' is the name of the symbol, `t' is the symbol
1157 descriptor, and `5=*6' is the type information.
1160 File: stabs.info, Node: Basic Cplusplus Types, Next: Simple Classes, Prev: Nested Symbols, Up: Cplusplus
1165 << the examples that follow are based on a01.C >>
1167 C++ adds two more builtin types to the set defined for C. These are
1168 the unknown type and the vtable record type. The unknown type, type
1169 16, is defined in terms of itself like the void type.
1171 The vtable record type, type 17, is defined as a structure type and
1172 then as a structure tag. The structure has four fields: delta, index,
1173 pfn, and delta2. pfn is the function pointer.
1175 << In boilerplate $vtbl_ptr_type, what are the fields delta, index,
1176 and delta2 used for? >>
1178 This basic type is present in all C++ programs even if there are no
1179 virtual methods defined.
1181 .stabs "struct_name:sym_desc(type)type_def(17)=type_desc(struct)struct_bytes(8)
1182 elem_name(delta):type_ref(short int),bit_offset(0),field_bits(16);
1183 elem_name(index):type_ref(short int),bit_offset(16),field_bits(16);
1184 elem_name(pfn):type_def(18)=type_desc(ptr to)type_ref(void),
1185 bit_offset(32),field_bits(32);
1186 elem_name(delta2):type_def(short int);bit_offset(32),field_bits(16);;"
1189 .stabs "$vtbl_ptr_type:t17=s8
1190 delta:6,0,16;index:6,16,16;pfn:18=*15,32,32;delta2:6,32,16;;"
1193 .stabs "name:sym_dec(struct tag)type_ref($vtbl_ptr_type)",N_LSYM,NIL,NIL,NIL
1195 .stabs "$vtbl_ptr_type:T17",128,0,0,0
1198 File: stabs.info, Node: Simple Classes, Next: Class Instance, Prev: Basic Cplusplus Types, Up: Cplusplus
1200 Simple Class Definition
1201 =======================
1203 The stabs describing C++ language features are an extension of the
1204 stabs describing C. Stabs representing C++ class types elaborate
1205 extensively on the stab format used to describe structure types in C.
1206 Stabs representing class type variables look just like stabs
1207 representing C language variables.
1209 Consider the following very simple class definition.
1214 int Ameth(int in, char other);
1217 The class `baseA' is represented by two stabs. The first stab
1218 describes the class as a structure type. The second stab describes a
1219 structure tag of the class type. Both stabs are of stab type `N_LSYM'.
1220 Since the stab is not located between an `N_FUN' and an `N_LBRAC' stab
1221 this indicates that the class is defined at file scope. If it were,
1222 then the `N_LSYM' would signify a local variable.
1224 A stab describing a C++ class type is similar in format to a stab
1225 describing a C struct, with each class member shown as a field in the
1226 structure. The part of the struct format describing fields is expanded
1227 to include extra information relevent to C++ class members. In
1228 addition, if the class has multiple base classes or virtual functions
1229 the struct format outside of the field parts is also augmented.
1231 In this simple example the field part of the C++ class stab
1232 representing member data looks just like the field part of a C struct
1233 stab. The section on protections describes how its format is sometimes
1234 extended for member data.
1236 The field part of a C++ class stab representing a member function
1237 differs substantially from the field part of a C struct stab. It still
1238 begins with `name:' but then goes on to define a new type number for
1239 the member function, describe its return type, its argument types, its
1240 protection level, any qualifiers applied to the method definition, and
1241 whether the method is virtual or not. If the method is virtual then
1242 the method description goes on to give the vtable index of the method,
1243 and the type number of the first base class defining the method.
1245 When the field name is a method name it is followed by two colons
1246 rather than one. This is followed by a new type definition for the
1247 method. This is a number followed by an equal sign and the type of the
1248 method. Normally this will be a type declared using the `#' type
1249 descriptor; see *Note Method Type Descriptor::; static member functions
1250 are declared using the `f' type descriptor instead; see *Note Function
1253 The format of an overloaded operator method name differs from that of
1254 other methods. It is `op$::OPERATOR-NAME.' where OPERATOR-NAME is the
1255 operator name such as `+' or `+='. The name ends with a period, and
1256 any characters except the period can occur in the OPERATOR-NAME string.
1258 The next part of the method description represents the arguments to
1259 the method, preceeded by a colon and ending with a semi-colon. The
1260 types of the arguments are expressed in the same way argument types are
1261 expressed in C++ name mangling. In this example an `int' and a `char'
1264 This is followed by a number, a letter, and an asterisk or period,
1265 followed by another semicolon. The number indicates the protections
1266 that apply to the member function. Here the 2 means public. The
1267 letter encodes any qualifier applied to the method definition. In this
1268 case, `A' means that it is a normal function definition. The dot shows
1269 that the method is not virtual. The sections that follow elaborate
1270 further on these fields and describe the additional information present
1271 for virtual methods.
1273 .stabs "class_name:sym_desc(type)type_def(20)=type_desc(struct)struct_bytes(4)
1274 field_name(Adat):type(int),bit_offset(0),field_bits(32);
1276 method_name(Ameth)::type_def(21)=type_desc(method)return_type(int);
1277 :arg_types(int char);
1278 protection(public)qualifier(normal)virtual(no);;"
1281 .stabs "baseA:t20=s4Adat:1,0,32;Ameth::21=##1;:ic;2A.;;",128,0,0,0
1283 .stabs "class_name:sym_desc(struct tag)",N_LSYM,NIL,NIL,NIL
1285 .stabs "baseA:T20",128,0,0,0