2 \label{chap:typeentries}
3 This section presents the debugging information entries
4 that describe program types: base types, modified types and
5 user\dash defined types.
7 If the scope of the declaration of a named type begins after
8 \hypertarget{chap:DWATstartscopetypedeclaration}{}
9 the low pc value for the scope most closely enclosing the
10 declaration, the declaration may have a
12 attribute as described for objects in
13 Section \refersec{chap:dataobjectentries}.
15 \section{Base Type Entries}
16 \label{chap:basetypeentries}
18 \textit{A base type is a data type that is not defined in terms of
20 \addtoindexx{fundamental type|see{base type entry}}
21 Each programming language has a set of base
22 types that are considered to be built into that language.}
24 A base type is represented by a debugging information entry
28 A \addtoindex{base type entry}
29 has a \DWATname{} attribute
31 \addtoindexx{name attribute}
33 a null\dash terminated string containing the name of the base type
34 as recognized by the programming language of the compilation
35 unit containing the base type entry.
38 \addtoindexx{encoding attribute}
39 a \DWATencoding{} attribute describing
40 how the base type is encoded and is to be interpreted. The
41 value of this attribute is an
42 \livelink{chap:classconstant}{integer constant}. The set of
43 values and their meanings for the
44 \DWATencoding{} attribute
46 Table \refersec{tab:encodingattributevalues}
50 may have a \DWATendianity{} attribute
51 \addtoindexx{endianity attribute}
53 Section \refersec{chap:dataobjectentries}.
54 If omitted, the encoding assumes the representation that
55 is the default for the target architecture.
58 \hypertarget{chap:DWATbytesizedataobjectordatatypesize}{}
59 either a \DWATbytesize{} attribute
60 \hypertarget{chap:DWATbitsizebasetypebitsize}{}
61 or a \DWATbitsize{} attribute
62 \addtoindexx{bit size attribute}
63 whose \livelink{chap:classconstant}{integer constant} value
64 (see Section \refersec{chap:byteandbitsizes})
65 is the amount of storage needed to hold
69 \textit{For example, the
70 \addtoindex{C} type \texttt{int} on a machine that uses 32\dash bit
71 integers is represented by a base type entry with a name
72 attribute whose value is \doublequote{int}, an encoding attribute
73 whose value is \DWATEsigned{}
74 and a byte size attribute whose value is 4.}
76 If the value of an object of the given type does not fully
77 occupy the storage described by a byte size attribute,
78 \hypertarget{chap:DWATdatabitoffsetbasetypebitlocation}{}
79 the base type entry may also have
80 \addtoindexx{bit size attribute}
83 \DWATdatabitoffset{} attribute,
85 \addtoindexx{data bit offset attribute}
87 \livelink{chap:classconstant}{integer constant} values
88 (see Section \refersec{chap:staticanddynamicvaluesofattributes}).
90 attribute describes the actual size in bits used to represent
91 values of the given type. The data bit offset attribute is the
92 offset in bits from the beginning of the containing storage to
93 the beginning of the value. Bits that are part of the offset
94 are padding. The data bit offset uses the bit numbering and
95 direction conventions that are appropriate to the current
97 target system to locate the beginning of the storage and
98 value. If this attribute is omitted a default data bit offset
104 \addtoindexx{bit offset attribute (Version 3)}
105 \addtoindexx{bit offset attribute (Version 3)|see{\textit{also} data bit offset attribute}}
107 \addtoindexx{data bit offset attribute}
109 \addtoindex{DWARF Version 4}, unchanged in \addtoindex{DWARF Version 5}, and
110 is also used for bit field members
111 (see Section \refersec{chap:datamemberentries}).
113 \hypertarget{chap:DWATbitoffsetbasetypebitlocation}{}
114 replaces the attribute
117 \addtoindexx{bit offset attribute (Version 3)}
118 types as defined in \DWARFVersionIII{} and earlier.
120 is deprecated for use in base types in DWARF Version 4 and later.
121 See Section 5.1 in the \DWARFVersionIV{}
122 specification for a discussion of compatibility considerations.}
125 \caption{Encoding attribute values}
126 \label{tab:encodingattributevalues}
128 \begin{tabular}{l|p{8cm}}
130 Name&Meaning\\ \hline
131 \DWATEaddressTARG{} & linear machine address (for segmented\break
133 Section \refersec{chap:segmentedaddresses}) \\
134 \DWATEbooleanTARG& true or false \\
136 \DWATEcomplexfloatTARG& complex binary
137 floating\dash point number \\
138 \DWATEfloatTARG{} & binary floating\dash point number \\
139 \DWATEimaginaryfloatTARG& imaginary binary
140 floating\dash point number \\
141 \DWATEsignedTARG& signed binary integer \\
142 \DWATEsignedcharTARG& signed character \\
143 \DWATEunsignedTARG{} & unsigned binary integer \\
144 \DWATEunsignedcharTARG{} & unsigned character \\
145 \DWATEpackeddecimalTARG{} & packed decimal \\
146 \DWATEnumericstringTARG& numeric string \\
147 \DWATEeditedTARG{} & edited string \\
148 \DWATEsignedfixedTARG{} & signed fixed\dash point scaled integer \\
149 \DWATEunsignedfixedTARG& unsigned fixed\dash point scaled integer \\
150 \DWATEdecimalfloatTARG{} & decimal floating\dash point number \\
151 \DWATEUTFTARG{} & \addtoindexi{Unicode character}{Unicode character base type} \\
152 \DWATEASCIITARG{} & \addtoindexi{ASCII character}{ASCII character base type}\\
153 \DWATEUCSTARG{} & \addtoindexi{ISO 10646 character}{ISO 10646 character base type}
154 \addtoindexx{ISO 10646 character set standard} \\
159 \textit{The \DWATEdecimalfloat{} encoding is intended for
160 floating\dash point representations that have a power\dash of\dash ten
161 exponent, such as that specified in IEEE 754R.}
163 \textit{The \DWATEUTF{} encoding is intended for \addtoindex{Unicode}
164 string encodings (see the Universal Character Set standard,
165 ISO/IEC 10646\dash 1:1993).
166 \addtoindexx{ISO 10646 character set standard}
168 \addtoindex{C++} type char16\_t is
169 represented by a base type entry with a name attribute whose
170 value is \doublequote{char16\_t}, an encoding attribute whose value
171 is \DWATEUTF{} and a byte size attribute whose value is 2.}
173 \textit{The \DWATEASCII{} and \DWATEUCS{} encodings are intended for
174 the {Fortran 2003} string kinds
175 \texttt{ASCII}\index{ASCII@\texttt{ASCII} (Fortran string kind)} (ISO/IEC 646:1991) and
176 \texttt{ISO\_10646}\index{ISO\_10646@\texttt{ISO\_10646} (Fortran string kind)} (UCS-4 in ISO/IEC 10646:2000).}
177 \addtoindexx{ISO 10646 character set standard}
180 \DWATEpackeddecimal{}
182 \DWATEnumericstring{}
184 represent packed and unpacked decimal string numeric data
185 types, respectively, either of which may be
187 \addtoindexx{decimal scale attribute}
189 \addtoindexx{decimal sign attribute}
191 \addtoindexx{digit count attribute}
193 \hypertarget{chap:DWATdecimalsigndecimalsignrepresentation}{}
195 \hypertarget{chap:DWATdigitcountdigitcountforpackeddecimalornumericstringtype}{}
196 base types are used in combination with
198 \DWATdigitcount{} and
203 A \DWATdecimalsign{} attribute
204 \addtoindexx{decimal sign attribute}
205 is an \livelink{chap:classconstant}{integer constant} that
206 conveys the representation of the sign of the decimal type
207 (see Table \refersec{tab:decimalsignattributevalues}).
208 Its \livelink{chap:classconstant}{integer constant} value is interpreted to
209 mean that the type has a leading overpunch, trailing overpunch,
210 leading separate or trailing separate sign representation or,
211 alternatively, no sign at all.
214 \caption{Decimal sign attribute values}
215 \label{tab:decimalsignattributevalues}
217 \begin{tabular}{l|p{9cm}}
221 \DWDSunsignedTARG{} & Unsigned \\
222 \DWDSleadingoverpunchTARG{} & Sign
223 is encoded in the most significant digit in a target\dash dependent manner \\
224 \DWDStrailingoverpunchTARG{} & Sign
225 is encoded in the least significant digit in a target\dash dependent manner \\
226 \DWDSleadingseparateTARG{}
227 & Decimal type: Sign is a \doublequote{+} or \doublequote{-} character
228 to the left of the most significant digit. \\
229 \DWDStrailingseparateTARG{}
230 & Decimal type: Sign is a \doublequote{+} or \doublequote{-} character
231 to the right of the least significant digit. \\
232 &Packed decimal type: Least significant nibble contains
233 a target\dash dependent value
234 indicating positive or negative. \\
239 \hypertarget{chap:DWATdecimalscaledecimalscalefactor}{}
240 The \DWATdecimalscale{}
242 \addtoindexx{decimal scale attribute}
243 is an integer constant value
244 that represents the exponent of the base ten scale factor to
245 be applied to an instance of the type. A scale of zero puts the
246 decimal point immediately to the right of the least significant
247 digit. Positive scale moves the decimal point to the right
248 and implies that additional zero digits on the right are not
249 stored in an instance of the type. Negative scale moves the
250 decimal point to the left; if the absolute value of the scale
251 is larger than the digit count, this implies additional zero
252 digits on the left are not stored in an instance of the type.
257 \addtoindexx{digit count attribute}
258 is an \livelink{chap:classconstant}{integer constant}
259 value that represents the number of digits in an instance of
264 \hypertarget{chap:DWATpicturestringpicturestringfornumericstringtype}{}
265 type is used to represent an edited
266 numeric or alphanumeric data type. It is used in combination
267 with a \DWATpicturestring{} attribute whose value is a
268 null\dash terminated string containing the target\dash dependent picture
269 string associated with the type.
271 If the edited base type entry describes an edited numeric
272 data type, the edited type entry has a \DWATdigitcount{} and a
273 \DWATdecimalscale{} attribute.
274 \addtoindexx{decimal scale attribute}
275 These attributes have the same
276 interpretation as described for the
277 \DWATEpackeddecimal{} and
278 \DWATEnumericstring{} base
279 types. If the edited type entry
280 describes an edited alphanumeric data type, the edited type
281 entry does not have these attributes.
284 \textit{The presence or absence of the \DWATdigitcount{} and
285 \DWATdecimalscale{} attributes
286 \addtoindexx{decimal scale attribute}
287 allows a debugger to easily
288 distinguish edited numeric from edited alphanumeric, although
289 in principle the digit count and scale are derivable by
290 interpreting the picture string.}
292 The \DWATEsignedfixed{} and \DWATEunsignedfixed{} entries
293 describe signed and unsigned fixed\dash point binary data types,
296 The fixed binary type entries have
297 \addtoindexx{digit count attribute}
300 attribute with the same interpretation as described for the
301 \DWATEpackeddecimal{} and \DWATEnumericstring{} base types.
304 For a data type with a decimal scale factor, the fixed binary
306 \DWATdecimalscale{} attribute
307 \addtoindexx{decimal scale attribute}
309 interpretation as described for the
310 \DWATEpackeddecimal{}
311 and \DWATEnumericstring{} base types.
313 \hypertarget{chap:DWATbinaryscalebinaryscalefactorforfixedpointtype}{}
314 For a data type with a binary scale factor, the fixed
315 \addtoindexx{binary scale attribute}
316 binary type entry has a
317 \DWATbinaryscale{} attribute.
319 \DWATbinaryscale{} attribute
320 is an \livelink{chap:classconstant}{integer constant} value
321 that represents the exponent of the base two scale factor to
322 be applied to an instance of the type. Zero scale puts the
323 binary point immediately to the right of the least significant
324 bit. Positive scale moves the binary point to the right and
325 implies that additional zero bits on the right are not stored
326 in an instance of the type. Negative scale moves the binary
327 point to the left; if the absolute value of the scale is
328 larger than the number of bits, this implies additional zero
329 bits on the left are not stored in an instance of the type.
332 \hypertarget{chap:DWATsmallscalefactorforfixedpointtype}{}
333 a data type with a non\dash decimal and non\dash binary scale factor,
334 the fixed binary type entry has a
335 \DWATsmall{} attribute which
336 \addtoindexx{small attribute}
338 \DWTAGconstant{} entry. The scale factor value
339 is interpreted in accordance with the value defined by the
340 \DWTAGconstant{} entry. The value represented is the product
341 of the integer value in memory and the associated constant
344 \textit{The \DWATsmall{} attribute
345 is defined with the \addtoindex{Ada} \texttt{small}
348 \section{Unspecified Type Entries}
349 \label{chap:unspecifiedtypeentries}
350 \addtoindexx{unspecified type entry}
351 \addtoindexx{void type|see{unspecified type entry}}
352 Some languages have constructs in which a type
353 may be left unspecified or the absence of a type
354 may be explicitly indicated.
356 An unspecified (implicit, unknown, ambiguous or nonexistent)
357 type is represented by a debugging information entry with
358 the tag \DWTAGunspecifiedtypeTARG.
359 If a name has been given
360 to the type, then the corresponding unspecified type entry
361 has a \DWATname{} attribute
362 \addtoindexx{name attribute}
364 a null\dash terminated
365 string containing the name as it appears in the source program.
367 The interpretation of this debugging information entry is
368 intentionally left flexible to allow it to be interpreted
369 appropriately in different languages. For example, in
370 \addtoindex{C} and \addtoindex{C++}
371 the language implementation can provide an unspecified type
372 entry with the name \doublequote{void} which can be referenced by the
373 type attribute of pointer types and typedef declarations for
375 Sections \refersec{chap:typemodifierentries} and
376 %The following reference was valid, so the following is probably correct.
377 Section \refersec{chap:typedefentries},
378 respectively). As another
379 example, in \addtoindex{Ada} such an unspecified type entry can be referred
380 to by the type attribute of an access type where the denoted
381 \addtoindexx{incomplete type (Ada)}
382 type is incomplete (the name is declared as a type but the
383 definition is deferred to a separate compilation unit).
385 \addtoindex{C++} permits using the
386 \autoreturntype{} specifier for the return type of a member function declaration.
387 The actual return type is deduced based on the definition of the
388 function, so it may not be known when the function is declared. The language
389 implementation can provide an unspecified type entry with the name \texttt{auto} which
390 can be referenced by the return type attribute of a function declaration entry.
391 When the function is later defined, the \DWTAGsubprogram{} entry for the definition
392 includes a reference to the actual return type.
395 \section{Type Modifier Entries}
396 \label{chap:typemodifierentries}
397 \addtoindexx{type modifier entry}
398 \addtoindexx{type modifier|see{atomic type entry}}
399 \addtoindexx{type modifier|see{constant type entry}}
400 \addtoindexx{type modifier|see{reference type entry}}
401 \addtoindexx{type modifier|see{restricted type entry}}
402 \addtoindexx{type modifier|see{packed type entry}}
403 \addtoindexx{type modifier|see{pointer type entry}}
404 \addtoindexx{type modifier|see{shared type entry}}
405 \addtoindexx{type modifier|see{volatile type entry}}
406 A base or user\dash defined type may be modified in different ways
407 in different languages. A type modifier is represented in
408 DWARF by a debugging information entry with one of the tags
409 given in Table \refersec{tab:typemodifiertags}.
411 If a name has been given to the modified type in the source
412 program, then the corresponding modified type entry has
413 a \DWATname{} attribute
414 \addtoindexx{name attribute}
415 whose value is a null\dash terminated
416 string containing the modified type name as it appears in
419 Each of the type modifier entries has
420 \addtoindexx{type attribute}
422 \DWATtype{} attribute,
423 whose value is a \livelink{chap:classreference}{reference}
424 to a debugging information entry
425 describing a base type, a user-defined type or another type
428 A modified type entry describing a
429 \addtoindexx{pointer type entry}
430 pointer or \addtoindex{reference type}
431 (using \DWTAGpointertype,
432 \DWTAGreferencetype{} or
433 \DWTAGrvaluereferencetype)
434 % Another instance of no-good-place-to-put-index entry.
436 \addtoindexx{address class attribute}
438 \hypertarget{chap:DWATadressclasspointerorreferencetypes}{}
441 attribute to describe how objects having the given pointer
442 or reference type ought to be dereferenced.
444 A modified type entry describing a \addtoindex{UPC} shared qualified type
445 (using \DWTAGsharedtype) may have a
446 \DWATcount{} attribute
447 \addtoindexx{count attribute}
448 whose value is a constant expressing the (explicit or implied) blocksize specified for the
449 type in the source. If no count attribute is present, then the \doublequote{infinite}
450 blocksize is assumed.
452 When multiple type modifiers are chained together to modify
453 a base or user-defined type, the tree ordering reflects the
455 \addtoindexx{reference type entry, lvalue|see{reference type entry}}
457 \addtoindexx{reference type entry, rvalue|see{rvalue reference type entry}}
459 \addtoindexx{parameter|see{macro formal parameter list}}
461 \addtoindexx{parameter|see{\textit{this} parameter}}
463 \addtoindexx{parameter|see{variable parameter attribute}}
465 \addtoindexx{parameter|see{optional parameter attribute}}
467 \addtoindexx{parameter|see{unspecified parameters entry}}
469 \addtoindexx{parameter|see{template value parameter entry}}
471 \addtoindexx{parameter|see{template type parameter entry}}
473 \addtoindexx{parameter|see{formal parameter entry}}
477 \caption{Type modifier tags}
478 \label{tab:typemodifiertags}
480 \begin{tabular}{l|p{9cm}}
482 Name&Meaning\\ \hline
483 \DWTAGatomictypeTARG{} & C \addtoindex{\_Atomic} qualified type \\
484 \DWTAGconsttypeTARG{} & C or C++ const qualified type
485 \addtoindexx{const qualified type entry} \addtoindexx{C} \addtoindexx{C++} \\
486 \DWTAGpackedtypeTARG& \addtoindex{Pascal} or Ada packed type\addtoindexx{packed type entry}
487 \addtoindexx{packed qualified type entry} \addtoindexx{Ada} \addtoindexx{Pascal} \\
488 \DWTAGpointertypeTARG{} & Pointer to an object of
489 the type being modified \addtoindexx{pointer qualified type entry} \\
490 \DWTAGreferencetypeTARG& \addtoindex{C++} (lvalue) reference
491 to an object of the type
492 \addtoindexx{reference type entry}
493 \mbox{being} modified
494 \addtoindexx{reference qualified type entry} \\
495 \DWTAGrestricttypeTARG& \addtoindex{C}
497 \addtoindexx{restricted type entry}
499 \addtoindexx{restrict qualified type} \\
500 \DWTAGrvaluereferencetypeTARG{} & \addtoindex{C++}
501 \addtoindexx{rvalue reference type entry}
503 \addtoindexx{restricted type entry}
504 reference to an object of the type \mbox{being} modified
505 \addtoindexx{rvalue reference qualified type entry} \\
506 \DWTAGsharedtypeTARG&\addtoindex{UPC} shared qualified type
507 \addtoindexx{shared qualified type entry} \\
508 \DWTAGvolatiletypeTARG&\addtoindex{C} or \addtoindex{C++} volatile qualified type
509 \addtoindexx{volatile qualified type entry} \\
515 \textit{As examples of how type modifiers are ordered, consider the following
516 \addtoindex{C} declarations:}
517 \begin{lstlisting}[numbers=none]
518 const unsigned char * volatile p;
520 \textit{which represents a volatile pointer to a constant
521 character. This is encoded in DWARF as:}
525 \DWTAGvariable(p) -->
526 \DWTAGvolatiletype -->
527 \DWTAGpointertype -->
529 \DWTAGbasetype(unsigned char)
534 \textit{On the other hand}
535 \begin{lstlisting}[numbers=none]
536 volatile unsigned char * const restrict p;
538 \textit{represents a restricted constant
539 pointer to a volatile character. This is encoded as:}
543 \DWTAGvariable(p) -->
544 \DWTAGrestricttype -->
546 \DWTAGpointertype -->
547 \DWTAGvolatiletype -->
548 \DWTAGbasetype(unsigned char)
552 \section{Typedef Entries}
553 \label{chap:typedefentries}
554 A named type that is defined in terms of another type
555 definition is represented by a debugging information entry with
556 \addtoindexx{typedef entry}
557 the tag \DWTAGtypedefTARG.
558 The typedef entry has a \DWATname{} attribute
559 \addtoindexx{name attribute}
560 whose value is a null\dash terminated string containing
561 the name of the typedef as it appears in the source program.
563 The typedef entry may also contain
564 \addtoindexx{type attribute}
566 \DWATtype{} attribute whose
567 value is a \livelink{chap:classreference}{reference}
568 to the type named by the typedef. If
569 the debugging information entry for a typedef represents
570 a declaration of the type that is not also a definition,
571 it does not contain a type attribute.
573 \textit{Depending on the language, a named type that is defined in
574 terms of another type may be called a type alias, a subtype,
575 a constrained type and other terms. A type name declared with
576 no defining details may be termed an
577 \addtoindexx{incomplete type}
578 incomplete, forward or hidden type.
579 While the DWARF \DWTAGtypedef{} entry was
580 originally inspired by the like named construct in
581 \addtoindex{C} and \addtoindex{C++},
582 it is broadly suitable for similar constructs (by whatever
583 source syntax) in other languages.}
585 \section{Array Type Entries}
586 \label{chap:arraytypeentries}
587 \label{chap:DWTAGgenericsubrange}
589 \textit{Many languages share the concept of an \doublequote{array,} which is
590 \addtoindexx{array type entry}
591 a table of components of identical type.}
593 An array type is represented by a debugging information entry
594 with the tag \DWTAGarraytypeTARG.
595 If a name has been given to
596 \addtoindexx{array!declaration of type}
597 the array type in the source program, then the corresponding
598 array type entry has a \DWATname{} attribute
599 \addtoindexx{name attribute}
601 null\dash terminated string containing the array type name as it
602 appears in the source program.
605 \hypertarget{chap:DWATorderingarrayrowcolumnordering}{}
606 array type entry describing a multidimensional array may
607 \addtoindexx{array!element ordering}
608 have a \DWATordering{} attribute whose
609 \livelink{chap:classconstant}{integer constant} value is
610 interpreted to mean either row-major or column-major ordering
611 of array elements. The set of values and their meanings
612 for the ordering attribute are listed in
613 Table \refersec{tab:arrayordering}.
615 ordering attribute is present, the default ordering for the
616 source language (which is indicated by the
619 \addtoindexx{language attribute}
620 of the enclosing compilation unit entry) is assumed.
622 \begin{simplenametable}[1.8in]{Array ordering}{tab:arrayordering}
623 \DWORDcolmajorTARG{} \\
624 \DWORDrowmajorTARG{} \\
625 \end{simplenametable}
627 The ordering attribute may optionally appear on one-dimensional
628 arrays; it will be ignored.
630 An array type entry has
631 \addtoindexx{type attribute}
632 a \DWATtype{} attribute
634 \addtoindexx{array!element type}
635 the type of each element of the array.
637 If the amount of storage allocated to hold each element of an
638 object of the given array type is different from the amount
639 \addtoindexx{stride attribute|see{bit stride attribute or byte stride attribute}}
640 of storage that is normally allocated to hold an individual
641 \hypertarget{chap:DWATbitstridearrayelementstrideofarraytype}{}
643 \hypertarget{chap:DWATbytestridearrayelementstrideofarraytype}{}
644 indicated element type, then the array type
645 \addtoindexx{bit stride attribute}
649 \addtoindexx{byte stride attribute}
652 \addtoindexx{bit stride attribute}
654 (see Section \refersec{chap:staticanddynamicvaluesofattributes})
656 element of the array.
658 The array type entry may have either a \DWATbytesize{} or a
659 \DWATbitsize{} attribute
660 (see Section \refersec{chap:byteandbitsizes}),
662 amount of storage needed to hold an instance of the array type.
664 \textit{If the size of the array can be determined statically at
665 compile time, this value can usually be computed by multiplying
666 the number of array elements by the size of each element.}
669 Each array dimension is described by a debugging information
670 entry with either the
671 \addtoindexx{subrange type entry!as array dimension}
672 tag \DWTAGsubrangetype{} or the
673 \addtoindexx{enumeration type entry!as array dimension}
675 \DWTAGenumerationtype. These entries are
677 array type entry and are ordered to reflect the appearance of
678 the dimensions in the source program (that is, leftmost dimension
679 first, next to leftmost second, and so on).
681 \textit{In languages that have no concept of a
682 \doublequote{multidimensional array} (for example,
683 \addtoindex{C}), an array of arrays may
684 be represented by a debugging information entry for a
685 multidimensional array.}
687 Alternatively, for an array with dynamic rank the array dimensions
688 are described by a debugging information entry with the tag
689 \DWTAGgenericsubrangeTARG.
690 This entry has the same attributes as a
691 \DWTAGsubrangetype{} entry; however,
692 there is just one \DWTAGgenericsubrangeNAME{} entry and it describes all of the
693 dimensions of the array.
694 If \DWTAGgenericsubrangeNAME{}
695 is used, the number of dimensions must be specified using a
696 \DWATrank{} attribute. See also Section
697 \refersec{chap:DWATrank}.
701 Other attributes especially applicable to arrays are
703 \DWATassociated{} and
705 which are described in
706 Section \refersec{chap:dynamictypeproperties}.
707 For relevant examples, see also Appendix \refersec{app:fortranarrayexample}.
709 \section{Coarray Type Entries}
710 \label{chap:coarraytypeentries}
711 \addtoindexx{coarray}
712 \textit{In Fortran, a \doublequote{coarray} is an array whose
713 elements are located in different processes rather than in the
714 memory of one process. The individual elements
715 of a coarray can be scalars or arrays.
716 Similar to arrays, coarrays have \doublequote{codimensions} that are
717 indexed using a \doublequote{coindex} or multiple \doublequote{coindices}.
718 \addtoindexx{codimension|see{coarray}}
719 \addtoindexx{coindex|see{coarray}}
722 A coarray type is represented by a debugging information entry
723 with the tag \DWTAGcoarraytypeTARG.
724 If a name has been given to the
725 coarray type in the source, then the corresponding coarray type
726 entry has a \DWATname{} attribute whose value is a null-terminated
727 string containing the array type name as it appears in the source
730 A coarray entry has one or more \DWTAGsubrangetype{} child entries,
731 one for each codimension. It also has a \DWATtype{} attribute
732 describing the type of each element of the coarray.
734 \textit{In a coarray application, the run-time number of processes in the application
735 is part of the coindex calculation. It is represented in the Fortran source by
736 a coindex which is declared with a \doublequote{*} as the upper bound. To express this
737 concept in DWARF, the \DWTAGsubrangetype{} child entry for that index has
738 only a lower bound and no upper bound.}
740 \textit{How coarray elements are located and how coindices are
741 converted to process specifications is processor-dependent.}
744 \section{Structure, Union, Class and Interface Type Entries}
745 \label{chap:structureunionclassandinterfacetypeentries}
747 \textit{The languages
749 \addtoindex{C++}, and
750 \addtoindex{Pascal}, among others, allow the
751 programmer to define types that are collections of related
752 \addtoindexx{structure type entry}
754 In \addtoindex{C} and \addtoindex{C++}, these collections are called
755 \doublequote{structures.}
756 In \addtoindex{Pascal}, they are called \doublequote{records.}
757 The components may be of different types. The components are
758 called \doublequote{members} in \addtoindex{C} and
759 \addtoindex{C++}, and \doublequote{fields} in \addtoindex{Pascal}.}
761 \textit{The components of these collections each exist in their
762 own space in computer memory. The components of a \addtoindex{C} or \addtoindex{C++}
763 \doublequote{union} all coexist in the same memory.}
765 \textit{\addtoindex{Pascal} and
766 other languages have a \doublequote{discriminated union,}
767 \addtoindexx{discriminated union|see {variant entry}}
768 also called a \doublequote{variant record.} Here, selection of a
769 number of alternative substructures (\doublequote{variants}) is based
770 on the value of a component that is not part of any of those
771 substructures (the \doublequote{discriminant}).}
773 \textit{\addtoindex{C++} and
774 \addtoindex{Java} have the notion of \doublequote{class,} which is in some
775 ways similar to a structure. A class may have \doublequote{member
776 functions} which are subroutines that are within the scope
777 of a class or structure.}
779 \textit{The \addtoindex{C++} notion of
780 structure is more general than in \addtoindex{C}, being
781 equivalent to a class with minor differences. Accordingly,
782 in the following discussion statements about
783 \addtoindex{C++} classes may
784 be understood to apply to \addtoindex{C++} structures as well.}
786 \textit{\addtoindex{C++} has the notion of a "trivial" class,
787 whose objects can be bitwise copied. Trivial classes may have
788 different rules for passing objects of that type as parameters
791 \subsection{Structure, Union and Class Type Entries}
792 \label{chap:structureunionandclasstypeentries}
793 Structure, union, and class types are represented by debugging
794 \addtoindexx{structure type entry}
796 \addtoindexx{union type entry}
798 \addtoindexx{class type entry}
800 \DWTAGstructuretypeTARG,
802 and \DWTAGclasstypeTARG,
803 respectively. If a name has been given to the structure,
804 union, or class in the source program, then the corresponding
805 structure type, union type, or class type entry has a
806 \DWATname{} attribute
807 \addtoindexx{name attribute}
808 whose value is a null\dash terminated string
809 containing the type name as it appears in the source program.
811 The members of a structure, union, or class are represented
812 by debugging information entries that are owned by the
813 corresponding structure type, union type, or class type entry
814 and appear in the same order as the corresponding declarations
815 in the source program.
817 A structure, union, or class type may have a \DWATexportsymbolsNAME{}
819 \livetarg{chap:DWATexportsymbolsofstructunionclass}{}
820 which indicates that all member names defined within
821 the structure, union, or class may be referenced as if they were
822 defined within the containing structure, union, or class.
824 \textit{This may be used to describe anonymous structures, unions
825 and classes in \addtoindex{C} or \addtoindex{C++}}.
827 A structure type, union type or class type entry may have
828 either a \DWATbytesize{} or a
829 \DWATbitsize{} attribute
830 \hypertarget{chap:DWATbitsizedatamemberbitsize}{}
831 (see Section \refersec{chap:byteandbitsizes}),
832 whose value is the amount of storage needed
833 to hold an instance of the structure, union or class type,
834 including any padding.
836 An incomplete structure, union or class type
837 \addtoindexx{incomplete structure/union/class}
839 \addtoindexx{incomplete type}
840 represented by a structure, union or class
841 entry that does not have a byte size attribute and that has
842 \addtoindexx{declaration attribute}
843 a \DWATdeclaration{} attribute.
845 If the complete declaration of a type has been placed in
846 \hypertarget{chap:DWATsignaturetypesignature}{}
847 a separate \addtoindex{type unit}
848 (see Section \refersec{chap:separatetypeunitentries}),
849 an incomplete declaration
850 \addtoindexx{incomplete type}
851 of that type in the compilation unit may provide
852 the unique 64\dash bit signature of the type using
853 \addtoindexx{type signature}
857 If a structure, union or class entry represents the definition
858 of a structure, union or class member corresponding to a prior
859 incomplete structure, union or class, the entry may have a
860 \DWATspecification{} attribute
861 \addtoindexx{specification attribute}
862 whose value is a \livelink{chap:classreference}{reference} to
863 the debugging information entry representing that incomplete
866 Structure, union and class entries containing the
867 \DWATspecification{} attribute
868 \addtoindexx{specification attribute}
869 do not need to duplicate
870 information provided by the declaration entry referenced by the
871 specification attribute. In particular, such entries do not
872 need to contain an attribute for the name of the structure,
873 union or class they represent if such information is already
874 provided in the declaration.
876 \textit{For \addtoindex{C} and \addtoindex{C++},
878 \addtoindexx{data member|see {member entry (data)}}
879 member declarations occurring within
880 the declaration of a structure, union or class type are
881 considered to be \doublequote{definitions} of those members, with
882 the exception of \doublequote{static} data members, whose definitions
883 appear outside of the declaration of the enclosing structure,
884 union or class type. Function member declarations appearing
885 within a structure, union or class type declaration are
886 definitions only if the body of the function also appears
887 within the type declaration.}
889 If the definition for a given member of the structure, union
890 or class does not appear within the body of the declaration,
891 that member also has a debugging information entry describing
892 its definition. That latter entry has a
893 \DWATspecification{} attribute
894 \addtoindexx{specification attribute}
895 referencing the debugging information entry
896 owned by the body of the structure, union or class entry and
897 representing a non\dash defining declaration of the data, function
898 or type member. The referenced entry will not have information
899 about the location of that member (low and high pc attributes
900 for function members, location descriptions for data members)
901 and will have a \DWATdeclaration{} attribute.
904 \textit{Consider a nested class whose
905 definition occurs outside of the containing class definition, as in:}
907 \begin{lstlisting}[numbers=none]
914 \textit{The two different structs can be described in
915 different compilation units to
916 facilitate DWARF space compression
917 (see Appendix \refersec{app:usingcompilationunits}).}
920 A structure type, union type or class type entry may have a
921 \DWATcallingconvention{} attribute,
922 \addtoindexx{calling convention attribute}
923 whose value indicates whether a value of the type should be passed by reference
924 or passed by value. The set of calling convention codes for use with types
925 \addtoindexx{calling convention codes!for types}
926 \hypertarget{chap:DWATcallingconventionfortypes}{}
927 is given in Table \referfol{tab:callingconventioncodesfortypes}.
929 \begin{simplenametable}[2.2in]{Calling convention codes for types}{tab:callingconventioncodesfortypes}
931 \DWCCpassbyvalueTARG \\
932 \DWCCpassbyreferenceTARG \\
933 \end{simplenametable}
935 If this attribute is not present, or its value is
936 \DWCCnormalNAME, the convention to be used for an object of the
937 given type is assumed to be unspecified.
939 \textit{Note that \DWCCnormalNAME{} is also used as a calling convention
940 code for certain subprograms
941 (see Table \refersec{tab:callingconventioncodesforsubroutines}).}
943 \textit{If unspecified, a consumer may be able to deduce the calling
944 convention based on knowledge of the type and the ABI.}
947 \subsection{Interface Type Entries}
948 \label{chap:interfacetypeentries}
950 \textit{The \addtoindex{Java} language defines \doublequote{interface} types.
952 \addtoindexx{interface type entry}
953 in \addtoindex{Java} is similar to a \addtoindex{C++} or
954 \addtoindex{Java} class with only abstract
955 methods and constant data members.}
958 \addtoindexx{interface type entry}
959 are represented by debugging information
961 tag \DWTAGinterfacetypeTARG.
963 An interface type entry has
964 a \DWATname{} attribute,
965 \addtoindexx{name attribute}
967 value is a null\dash terminated string containing the type name
968 as it appears in the source program.
970 The members of an interface are represented by debugging
971 information entries that are owned by the interface type
972 entry and that appear in the same order as the corresponding
973 declarations in the source program.
975 \subsection{Derived or Extended Structs, Classes and Interfaces}
976 \label{chap:derivedorextendedstructsclasesandinterfaces}
978 \textit{In \addtoindex{C++}, a class (or struct)
980 \addtoindexx{derived type (C++)|see{inheritance entry}}
981 be \doublequote{derived from} or be a
982 \doublequote{subclass of} another class.
983 In \addtoindex{Java}, an interface may \doublequote{extend}
984 \addtoindexx{extended type (Java)|see{inheritance entry}}
986 \addtoindexx{implementing type (Java)|see{inheritance entry}}
987 or more other interfaces, and a class may \doublequote{extend} another
988 class and/or \doublequote{implement} one or more interfaces. All of these
989 relationships may be described using the following. Note that
990 in \addtoindex{Java},
991 the distinction between extends and implements is
992 implied by the entities at the two ends of the relationship.}
994 A class type or interface type entry that describes a
995 derived, extended or implementing class or interface owns
996 \addtoindexx{implementing type (Java)|see{inheritance entry}}
997 debugging information entries describing each of the classes
998 or interfaces it is derived from, extending or implementing,
999 respectively, ordered as they were in the source program. Each
1001 \addtoindexx{inheritance entry}
1003 tag \DWTAGinheritanceTARG.
1005 An inheritance entry
1006 \addtoindexx{type attribute}
1008 \addtoindexx{inheritance entry}
1010 \DWATtype{} attribute whose value is
1011 a reference to the debugging information entry describing the
1012 class or interface from which the parent class or structure
1013 of the inheritance entry is derived, extended or implementing.
1015 An inheritance entry
1016 \addtoindexx{inheritance entry}
1017 for a class that derives from or extends
1018 \hypertarget{chap:DWATdatamemberlocationinheritedmemberlocation}{}
1019 another class or struct also has
1020 \addtoindexx{data member location attribute}
1022 \DWATdatamemberlocation{}
1023 attribute, whose value describes the location of the beginning
1024 of the inherited type relative to the beginning address of the
1025 instance of the derived class. If that value is a constant, it is the offset
1026 in bytes from the beginning of the class to the beginning of
1027 the instance of the inherited type. Otherwise, the value must be a location
1028 description. In this latter case, the beginning address of
1029 the instance of the derived class is pushed on the expression stack before
1030 the \addtoindex{location description}
1031 is evaluated and the result of the
1032 evaluation is the location of the instance of the inherited type.
1034 \textit{The interpretation of the value of this attribute for
1035 inherited types is the same as the interpretation for data
1037 (see Section \referfol{chap:datamemberentries}). }
1040 \addtoindexx{inheritance entry}
1042 \hypertarget{chap:DWATaccessibilitycppinheritedmembers}{}
1044 \addtoindexx{accessibility attribute}
1046 \DWATaccessibility{}
1048 If no accessibility attribute
1049 is present, private access is assumed for an entry of a class
1050 and public access is assumed for an entry of an interface,
1054 \hypertarget{chap:DWATvirtualityvirtualityofbaseclass}{}
1055 the class referenced by the
1056 \addtoindexx{inheritance entry}
1057 inheritance entry serves
1058 as a \addtoindex{C++} virtual base class, the inheritance entry has a
1059 \DWATvirtuality{} attribute.
1061 \textit{For a \addtoindex{C++} virtual base, the
1062 \addtoindex{data member location attribute}
1063 will usually consist of a non-trivial
1064 \addtoindex{location description}.}
1066 \subsection{Access Declarations}
1067 \label{chap:accessdeclarations}
1069 \textit{In \addtoindex{C++}, a derived class may contain access declarations that
1070 \addtoindexx{access declaration entry}
1071 change the accessibility of individual class members from the
1072 overall accessibility specified by the inheritance declaration.
1073 A single access declaration may refer to a set of overloaded
1076 If a derived class or structure contains access declarations,
1077 each such declaration may be represented by a debugging
1078 information entry with the tag
1079 \DWTAGaccessdeclarationTARG.
1081 such entry is a child of the class or structure type entry.
1083 An access declaration entry has
1084 a \DWATname{} attribute,
1085 \addtoindexx{name attribute}
1087 value is a null\dash terminated string representing the name used
1088 in the declaration in the source program, including any class
1089 or structure qualifiers.
1091 An access declaration entry
1092 \hypertarget{chap:DWATaccessibilitycppbaseclasses}{}
1095 \DWATaccessibility{}
1096 attribute describing the declared accessibility of the named
1101 \subsection{Friends}
1102 \label{chap:friends}
1104 Each \doublequote{friend}
1105 \addtoindexx{friend entry}
1106 declared by a structure, union or class
1107 \hypertarget{chap:DWATfriendfriendrelationship}{}
1108 type may be represented by a debugging information entry
1109 that is a child of the structure, union or class type entry;
1110 the friend entry has the
1111 tag \DWTAGfriendTARG.
1114 \addtoindexx{friend attribute}
1115 a \DWATfriend{} attribute, whose value is
1116 a reference to the debugging information entry describing
1117 the declaration of the friend.
1120 \subsection{Data Member Entries}
1121 \label{chap:datamemberentries}
1123 A data member (as opposed to a member function) is
1124 represented by a debugging information entry with the
1125 tag \DWTAGmemberTARG.
1127 \addtoindexx{member entry (data)}
1128 member entry for a named member has
1129 a \DWATname{} attribute
1130 \addtoindexx{name attribute}
1131 whose value is a null\dash terminated
1132 string containing the member name as it appears in the source
1133 program. If the member entry describes an
1134 \addtoindex{anonymous union},
1135 the name attribute is omitted or the value of the attribute
1136 consists of a single zero byte.
1138 The data member entry has
1139 \addtoindexx{type attribute}
1141 \DWATtype{} attribute to denote
1142 \addtoindexx{member entry (data)}
1143 the type of that member.
1145 A data member entry may
1146 \addtoindexx{accessibility attribute}
1148 \DWATaccessibility{}
1149 attribute. If no accessibility attribute is present, private
1150 access is assumed for an entry of a class and public access
1151 is assumed for an entry of a structure, union, or interface.
1154 \hypertarget{chap:DWATmutablemutablepropertyofmemberdata}{}
1156 \addtoindexx{member entry (data)}
1158 \addtoindexx{mutable attribute}
1159 have a \DWATmutable{} attribute,
1160 which is a \livelink{chap:classflag}{flag}.
1161 This attribute indicates whether the data
1162 member was declared with the mutable storage class specifier.
1164 The beginning of a data member
1165 \addtoindexx{beginning of a data member}
1166 is described relative to
1167 \addtoindexx{beginning of an object}
1168 the beginning of the object in which it is immediately
1169 contained. In general, the beginning is characterized by
1170 both an address and a bit offset within the byte at that
1171 address. When the storage for an entity includes all of
1172 the bits in the beginning byte, the beginning bit offset is
1175 Bit offsets in DWARF use the bit numbering and direction
1176 conventions that are appropriate to the current language on
1180 \addtoindexx{member entry (data)}
1181 corresponding to a data member that is
1182 \hypertarget{chap:DWATdatabitoffsetdatamemberbitlocation}{}
1184 \hypertarget{chap:DWATdatamemberlocationdatamemberlocation}{}
1185 in a structure, union or class may have either
1186 \addtoindexx{data member location attribute}
1188 \DWATdatamemberlocation{} attribute or a
1189 \DWATdatabitoffset{}
1190 attribute. If the beginning of the data member is the same as
1191 the beginning of the containing entity then neither attribute
1195 For a \DWATdatamemberlocation{} attribute
1196 \addtoindexx{data member location attribute}
1197 there are two cases:
1198 \begin{enumerate}[1. ]
1199 \item If the value is an \livelink{chap:classconstant}{integer constant},
1201 in bytes from the beginning of the containing entity. If
1202 the beginning of the containing entity has a non-zero bit
1203 offset then the beginning of the member entry has that same
1206 \item Otherwise, the value must be a \addtoindex{location description}.
1208 this case, the beginning of the containing entity must be byte
1209 aligned. The beginning address is pushed on the DWARF stack
1210 before the \addtoindex{location} description is evaluated; the result of
1211 the evaluation is the base address of the member entry.
1213 \textit{The push on the DWARF expression stack of the base address of
1214 the containing construct is equivalent to execution of the
1215 \DWOPpushobjectaddress{} operation
1216 (see Section \refersec{chap:stackoperations});
1217 \DWOPpushobjectaddress{} therefore
1218 is not needed at the
1219 beginning of a \addtoindex{location description} for a data member.
1221 result of the evaluation is a location---either an address or
1222 the name of a register, not an offset to the member.}
1224 \textit{A \DWATdatamemberlocation{}
1226 \addtoindexx{data member location attribute}
1227 that has the form of a
1228 \addtoindex{location description} is not valid for a data member contained
1229 in an entity that is not byte aligned because DWARF operations
1230 do not allow for manipulating or computing bit offsets.}
1235 For a \DWATdatabitoffset{} attribute,
1236 the value is an \livelink{chap:classconstant}{integer constant}
1237 (see Section \refersec{chap:staticanddynamicvaluesofattributes})
1238 that specifies the number of bits
1239 from the beginning of the containing entity to the beginning
1240 of the data member. This value must be greater than or equal
1241 to zero, but is not limited to less than the number of bits
1244 If the size of a data member is not the same as the size
1245 of the type given for the data member, the data member has
1246 \addtoindexx{bit size attribute}
1247 either a \DWATbytesize{}
1248 or a \DWATbitsize{} attribute whose
1249 \livelink{chap:classconstant}{integer constant} value
1250 (see Section \refersec{chap:staticanddynamicvaluesofattributes})
1252 of storage needed to hold the value of the data member.
1254 \textit{Bit fields in \addtoindex{C} and \addtoindex{C++}
1256 \addtoindexx{bit fields}
1258 \addtoindexx{data bit offset}
1260 \addtoindexx{data bit size}
1262 \DWATdatabitoffset{} and
1263 \DWATbitsize{} attributes.}
1266 \textit{This Standard uses the following bit numbering and direction
1267 conventions in examples. These conventions are for illustrative
1268 purposes and other conventions may apply on particular
1271 \item \textit{For big\dash endian architectures, bit offsets are
1272 counted from high-order to low\dash order bits within a byte (or
1273 larger storage unit); in this case, the bit offset identifies
1274 the high\dash order bit of the object.}
1276 \item \textit{For little\dash endian architectures, bit offsets are
1277 counted from low\dash order to high\dash order bits within a byte (or
1278 larger storage unit); in this case, the bit offset identifies
1279 the low\dash order bit of the object.}
1283 \textit{In either case, the bit so identified is defined as the
1284 \addtoindexx{beginning of an object}
1285 beginning of the object.}
1288 \textit{For example, take one possible representation of the following
1289 \addtoindex{C} structure definition
1290 in both big\dash and little\dash endian byte orders:}
1301 \textit{Figures \referfol{fig:bigendiandatabitoffsets} and
1302 \refersec{fig:littleendiandatabitoffsets}
1303 show the structure layout
1304 and data bit offsets for example big\dash\ and little\dash endian
1305 architectures, respectively. Both diagrams show a structure
1306 that begins at address A and whose size is four bytes. Also,
1307 high order bits are to the left and low order bits are to
1319 Addresses increase ->
1320 | A | A + 1 | A + 2 | A + 3 |
1322 Data bit offsets increase ->
1323 +---------------+---------------+---------------+---------------+
1324 |0 4|5 10|11 15|16 23|24 31|
1325 | j | k | m | n | <pad> |
1327 +---------------------------------------------------------------+
1331 \caption{Big-endian data bit offsets}
1332 \label{fig:bigendiandatabitoffsets}
1343 <- Addresses increase
1344 | A + 3 | A + 2 | A + 1 | A |
1346 <- Data bit offsets increase
1347 +---------------+---------------+---------------+---------------+
1348 |31 24|23 16|15 11|10 5|4 0|
1349 | <pad> | n | m | k | j |
1351 +---------------------------------------------------------------+
1355 \caption{Little-endian data bit offsets}
1356 \label{fig:littleendiandatabitoffsets}
1360 \textit{Note that data member bit offsets in this example are the
1361 same for both big\dash\ and little\dash endian architectures even
1362 though the fields are allocated in different directions
1363 (high\dash order to low-order versus low\dash order to high\dash order);
1364 the bit naming conventions for memory and/or registers of
1365 the target architecture may or may not make this seem natural.}
1367 \textit{For a more extensive example showing nested and packed records
1369 Appendix \refersec{app:pascalexample}.}
1372 \textit{Attribute \DWATdatabitoffset{}
1374 \addtoindex{DWARF Version 4}, unchanged in \addtoindex{DWARF Version 5},
1375 and is also used for base types
1377 \refersec{chap:basetypeentries}).
1379 \livetarg{chap:DWATbitoffsetdatamemberbitlocation}{}
1380 attributes \DWATbitoffset{} and
1381 \DWATbytesize{} when used to
1382 identify the beginning of bit field data members as defined
1383 in DWARF V3 and earlier. The \DWATbytesize,
1386 attribute combination is deprecated for data members in DWARF
1387 Version 4 and later. See Section 5.6.6 in the DWARF Version 4
1388 specification for a discussion of compatibility considerations.}
1390 \subsection{Member Function Entries}
1391 \label{chap:memberfunctionentries}
1393 A member function is represented by a
1394 \addtoindexx{member function entry}
1395 debugging information entry
1397 \addtoindexx{subprogram entry!as member function}
1398 tag \DWTAGsubprogram.
1399 The member function entry
1400 may contain the same attributes and follows the same rules
1401 as non\dash member global subroutine entries
1402 (see Section \refersec{chap:subroutineandentrypointentries}).
1405 \textit{In particular, if the member function entry is an
1406 instantiation of a member function template, it follows the
1407 same rules as function template instantiations (see Section
1408 \refersec{chap:functiontemplateinstantiations}).
1412 \addtoindexx{accessibility attribute}
1413 member function entry may have a
1414 \DWATaccessibility{}
1415 attribute. If no accessibility attribute is present, private
1416 access is assumed for an entry of a class and public access
1417 is assumed for an entry of a structure, union or interface.
1420 \hypertarget{chap:DWATvirtualityvirtualityoffunction}{}
1421 the member function entry describes a virtual function,
1422 then that entry has a
1423 \DWATvirtuality{} attribute.
1426 \hypertarget{chap:DWATexplicitexplicitpropertyofmemberfunction}{}
1427 the member function entry describes an explicit member
1428 function, then that entry has
1429 \addtoindexx{explicit attribute}
1431 \DWATexplicit{} attribute.
1434 \hypertarget{chap:DWATvtableelemlocationvirtualfunctiontablevtableslot}{}
1435 entry for a virtual function also has a
1436 \DWATvtableelemlocation{}
1437 \addtoindexi{attribute}{vtable element location attribute} whose value contains
1438 a \addtoindex{location description}
1439 yielding the address of the slot
1440 for the function within the virtual function table for the
1441 enclosing class. The address of an object of the enclosing
1442 type is pushed onto the expression stack before the location
1443 description is evaluated.
1446 \hypertarget{chap:DWATobjectpointerobjectthisselfpointerofmemberfunction}{}
1447 the member function entry describes a non\dash static member
1448 \addtoindexx{this pointer attribute|see{object pointer attribute}}
1449 function, then that entry
1450 \addtoindexx{self pointer attribute|see{object pointer attribute}}
1452 \addtoindexx{object pointer attribute}
1453 a \DWATobjectpointer{}
1455 whose value is a \livelink{chap:classreference}{reference}
1456 to the formal parameter entry
1457 that corresponds to the object for which the function is
1458 called. The name attribute of that formal parameter is defined
1459 by the current language (for example,
1460 \texttt{this} for \addtoindex{C++} or \texttt{self}
1461 for \addtoindex{Objective C}
1462 and some other languages). That parameter
1463 also has a \DWATartificial{} attribute whose value is true.
1465 Conversely, if the member function entry describes a static
1466 member function, the entry does not have
1467 \addtoindexx{object pointer attribute}
1469 \DWATobjectpointer{}
1472 \textit{In \addtoindex{C++}, non-static member functions can have const-volatile
1473 qualifiers, which affect the type of the first formal parameter (the
1474 \doublequote{\texttt{this}}-pointer).}
1476 If the member function entry describes a non\dash static member
1477 function that has a const\dash volatile qualification, then
1478 the entry describes a non\dash static member function whose
1479 object formal parameter has a type that has an equivalent
1480 const\dash volatile qualification.
1482 \textit{In \addtoindex{C++:2011 (ISO)}, non-static member functions can also have one of the
1483 ref-qualifiers, \& and \&\&. They do not change the type of the
1484 \doublequote{\texttt{this}}-pointer, but they affect the types of object values the
1485 function can be invoked on.}
1487 The member function entry may have an \DWATreferenceNAME{} attribute
1488 \livetarg{chap:DWATreferenceofnonstaticmember}{}
1489 to indicate a non-static member function that can only be called on
1490 l-value objects, or the \DWATrvaluereferenceNAME{} attribute
1491 \livetarg{chap:DWATrvaluereferenceofnonstaticmember}{}
1492 to indicate that it can only be called on pr-values and x-values.
1494 If a subroutine entry represents the defining declaration
1495 of a member function and that definition appears outside of
1496 the body of the enclosing class declaration, the subroutine
1498 \DWATspecification{} attribute,
1499 \addtoindexx{specification attribute}
1501 a reference to the debugging information entry representing
1502 the declaration of this function member. The referenced entry
1503 will be a child of some class (or structure) type entry.
1505 Subroutine entries containing the
1506 \DWATspecification{} attribute
1507 \addtoindexx{specification attribute}
1508 do not need to duplicate information provided
1509 by the declaration entry referenced by the specification
1510 attribute. In particular, such entries do not need to contain
1511 a name attribute giving the name of the function member whose
1512 definition they represent.
1513 Similarly, such entries do not need to contain a return type
1514 attribute, unless the return type on the declaration was
1515 unspecified (for example, the declaration used the
1516 \addtoindex{C++} \autoreturntype{} specifier).
1518 \textit{In \addtoindex{C++}, a member function may be declared
1519 as deleted. This prevents the compiler from generating a default
1520 implementation of a special member function such as a
1521 constructor or destructor, and can affect overload resolution
1522 when used on other member functions.}
1524 If the member function entry has been declared as deleted,
1525 \hypertarget{chap:DWATdeleted}{}
1526 then that entry has a \DWATdeletedTARG{} attribute.\addtoindexx{deleted attribute}
1528 \textit{In \addtoindex{C++}, a special member function may be
1529 declared as defaulted, which explicitly declares a default
1530 compiler-generated implementation of the function. The
1531 declaration may have different effects on the calling
1532 convention used for objects of its class, depending on
1533 whether the default declaration is made inside or outside the
1536 If the member function has been declared as defaulted,
1537 then the entry has a \DWATdefaultedTARG{}
1538 attribute\addtoindexx{defaulted attribute}
1539 whose integer constant value indicates whether, and if so,
1540 how, that member is defaulted. The possible values and
1541 their meanings are shown in
1542 Table \referfol{tab:defaultedattributevaluenames}.
1545 \setlength{\extrarowheight}{0.1cm}
1546 \begin{longtable}{l|l}
1547 \caption{Defaulted attribute names} \label{tab:defaultedattributevaluenames} \\
1548 \hline \bfseries Defaulted attribute name & \bfseries Meaning \\ \hline
1550 \bfseries Defaulted attribute name & \bfseries Meaning \\ \hline
1552 \hline \emph{Continued on next page}
1555 \DWDEFAULTEDnoTARG & Not declared default \\
1556 \DWDEFAULTEDinclassTARG & Defaulted within the class \\
1557 \DWDEFAULTEDoutofclassTARG& Defaulted outside of the class \\
1562 \textit{An artificial member function (that is, a compiler-generated
1563 copy that does not appear in the source) does not have a
1564 \DWATdefaultedNAME{} attribute.}
1567 \subsection{Class Template Instantiations}
1568 \label{chap:classtemplateinstantiations}
1570 \textit{In \addtoindex{C++} a class template is a generic definition of a class
1571 type that may be instantiated when an instance of the class
1572 is declared or defined. The generic description of the class may include
1573 parameterized types, parameterized compile-time constant
1574 values, and/or parameterized run-time constant addresses.
1575 DWARF does not represent the generic template
1576 definition, but does represent each instantiation.}
1578 A class template instantiation is represented by a
1579 debugging information entry with the tag \DWTAGclasstype,
1580 \DWTAGstructuretype{} or
1581 \DWTAGuniontype. With the following
1582 exceptions, such an entry will contain the same attributes
1583 and have the same types of child entries as would an entry
1584 for a class type defined explicitly using the instantiation
1585 types and values. The exceptions are:
1587 \begin{enumerate}[1. ]
1588 \item Template parameters are described and referenced as
1589 specified in Section \refersec{chap:templateparameters}.
1592 \item If the compiler has generated a special compilation unit to
1594 \addtoindexx{template instantiation!and special compilation unit}
1595 template instantiation and that special compilation
1596 unit has a different name from the compilation unit containing
1597 the template definition, the name attribute for the debugging
1598 information entry representing the special compilation unit
1599 should be empty or omitted.
1602 \item If the class type entry representing the template
1603 instantiation or any of its child entries contains declaration
1604 coordinate attributes, those attributes should refer to
1605 the source for the template definition, not to any source
1606 generated artificially by the compiler.
1610 \subsection{Variant Entries}
1611 \label{chap:variantentries}
1613 A variant part of a structure is represented by a debugging
1614 information entry\addtoindexx{variant part entry} with the
1615 tag \DWTAGvariantpartTARG{} and is
1616 owned by the corresponding structure type entry.
1618 If the variant part has a discriminant, the discriminant is
1619 \hypertarget{chap:DWATdiscrdiscriminantofvariantpart}{}
1621 \addtoindexx{discriminant (entry)}
1622 separate debugging information entry which
1623 is a child of the variant part entry. This entry has the form
1625 \addtoindexx{member entry (data)!as discriminant}
1626 structure data member entry. The variant part entry will
1627 \addtoindexx{discriminant attribute}
1629 \DWATdiscr{} attribute
1630 whose value is a \livelink{chap:classreference}{reference} to
1631 the member entry for the discriminant.
1633 If the variant part does not have a discriminant (tag field),
1634 the variant part entry has
1635 \addtoindexx{type attribute}
1637 \DWATtype{} attribute to represent
1640 Each variant of a particular variant part is represented by
1641 \hypertarget{chap:DWATdiscrvaluediscriminantvalue}{}
1642 a debugging information entry\addtoindexx{variant entry} with the
1643 tag \DWTAGvariantTARG{}
1644 and is a child of the variant part entry. The value that
1645 selects a given variant may be represented in one of three
1646 ways. The variant entry may have a
1647 \DWATdiscrvalue{} attribute
1648 whose value represents a single case label. The value of this
1649 attribute is encoded as an LEB128 number. The number is signed
1650 if the tag type for the variant part containing this variant
1651 is a signed type. The number is unsigned if the tag type is
1656 \hypertarget{chap:DWATdiscrlistlistofdiscriminantvalues}{}
1657 the variant entry may contain
1658 \addtoindexx{discriminant list attribute}
1661 attribute, whose value represents a list of discriminant
1662 values. This list is represented by any of the
1663 \livelink{chap:classblock}{block} forms and
1664 may contain a mixture of case labels and label ranges. Each
1665 item on the list is prefixed with a discriminant value
1666 descriptor that determines whether the list item represents
1667 a single label or a label range. A single case label is
1668 represented as an LEB128 number as defined above for
1669 \addtoindexx{discriminant value attribute}
1672 attribute. A label range is represented by
1673 two LEB128 numbers, the low value of the range followed by the
1674 high value. Both values follow the rules for signedness just
1675 described. The discriminant value descriptor is an integer
1676 constant that may have one of the values given in
1677 Table \refersec{tab:discriminantdescriptorvalues}.
1679 \begin{simplenametable}[1.4in]{Discriminant descriptor values}{tab:discriminantdescriptorvalues}
1680 \DWDSClabelTARG{} \\
1681 \DWDSCrangeTARG{} \\
1682 \end{simplenametable}
1684 If a variant entry has neither a \DWATdiscrvalue{}
1685 attribute nor a \DWATdiscrlist{} attribute, or if it has
1686 a \DWATdiscrlist{} attribute with 0 size, the variant is a
1689 The components selected by a particular variant are represented
1690 by debugging information entries owned by the corresponding
1691 variant entry and appear in the same order as the corresponding
1692 declarations in the source program.
1695 \section{Condition Entries}
1696 \label{chap:conditionentries}
1698 \textit{COBOL has the notion of
1699 \addtoindexx{level-88 condition, COBOL}
1700 a \doublequote{level\dash 88 condition} that
1701 associates a data item, called the conditional variable, with
1702 a set of one or more constant values and/or value ranges.
1703 % Note: the {} after \textquoteright (twice) is necessary to assure a following space separator
1704 Semantically, the condition is \textquoteleft true\textquoteright{}
1706 variable's value matches any of the described constants,
1707 and the condition is \textquoteleft false\textquoteright{} otherwise.}
1709 The \DWTAGconditionTARG{}
1710 debugging information entry\addtoindexx{condition entry}
1712 logical condition that tests whether a given data item\textquoteright s
1713 value matches one of a set of constant values. If a name
1714 has been given to the condition, the condition entry has a
1715 \DWATname{} attribute
1716 \addtoindexx{name attribute}
1717 whose value is a null\dash terminated string
1718 giving the condition name as it appears in the source program.
1721 The condition entry's parent entry describes the conditional
1722 variable; normally this will be a \DWTAGvariable,
1724 \DWTAGformalparameter{} entry.
1726 \addtoindexx{formal parameter entry}
1728 entry has an array type, the condition can test any individual
1729 element, but not the array as a whole. The condition entry
1730 implicitly specifies a \doublequote{comparison type} that is the
1731 type of an array element if the parent has an array type;
1732 otherwise it is the type of the parent entry.
1735 The condition entry owns \DWTAGconstant{} and/or
1736 \DWTAGsubrangetype{} entries that describe the constant
1737 values associated with the condition. If any child entry
1738 \addtoindexx{type attribute}
1740 a \DWATtype{} attribute,
1741 that attribute should describe a type
1742 compatible with the comparison type (according to the source
1743 language); otherwise the child\textquoteright s type is the same as the
1746 \textit{For conditional variables with alphanumeric types, COBOL
1747 permits a source program to provide ranges of alphanumeric
1748 constants in the condition. Normally a subrange type entry
1749 does not describe ranges of strings; however, this can be
1750 represented using bounds attributes that are references to
1751 constant entries describing strings. A subrange type entry may
1752 refer to constant entries that are siblings of the subrange
1756 \section{Enumeration Type Entries}
1757 \label{chap:enumerationtypeentries}
1759 \textit{An \doublequote{enumeration type} is a scalar that can assume one of
1760 a fixed number of symbolic values.}
1762 An enumeration type is represented by a debugging information
1764 \DWTAGenumerationtypeTARG.
1766 If a name has been given to the enumeration type in the source
1767 program, then the corresponding enumeration type entry has
1768 a \DWATname{} attribute
1769 \addtoindexx{name attribute}
1770 whose value is a null\dash terminated
1771 string containing the enumeration type name as it appears
1772 in the source program.
1774 The \addtoindex{enumeration type entry}
1776 \addtoindexx{type attribute}
1777 a \DWATtype{} attribute
1778 which refers to the underlying data type used to implement
1779 the enumeration. The entry also may have a
1780 \DWATbytesize{} attribute whose
1781 \livelink{chap:classconstant}{integer constant} value is the number of bytes
1782 required to hold an instance of the enumeration. If no \DWATbytesize{} attribute
1783 is present, the size for holding an instance of the enumeration is given by the size
1784 of the underlying data type.
1787 If an enumeration type has type safe
1788 \addtoindexx{type safe enumeration types}
1791 \begin{enumerate}[1. ]
1792 \item Enumerators are contained in the scope of the enumeration type, and/or
1794 \item Enumerators are not implicitly converted to another type
1797 then the \addtoindex{enumeration type entry} may
1798 \addtoindexx{enum class|see{type-safe enumeration}}
1799 have a \DWATenumclass{}
1800 attribute, which is a \livelink{chap:classflag}{flag}.
1801 In a language that offers only
1802 one kind of enumeration declaration, this attribute is not
1805 \textit{In \addtoindex{C} or \addtoindex{C++},
1806 the underlying type will be the appropriate
1807 integral type determined by the compiler from the properties of
1808 \hypertarget{chap:DWATenumclasstypesafeenumerationdefinition}{}
1809 the enumeration literal values.
1810 A \addtoindex{C++} type declaration written
1811 using enum class declares a strongly typed enumeration and
1812 is represented using \DWTAGenumerationtype{}
1813 in combination with \DWATenumclass.}
1815 Each enumeration literal is represented by a debugging
1816 \addtoindexx{enumeration literal|see{enumeration entry}}
1817 information entry with the
1818 tag \DWTAGenumeratorTARG.
1820 such entry is a child of the
1821 \addtoindex{enumeration type entry}, and the
1822 enumerator entries appear in the same order as the declarations
1823 of the enumeration literals in the source program.
1825 Each \addtoindex{enumerator entry} has a
1826 \DWATname{} attribute, whose
1827 \addtoindexx{name attribute}
1828 value is a null\dash terminated string containing the name of the
1829 \hypertarget{chap:DWATconstvalueenumerationliteralvalue}{}
1830 enumeration literal as it appears in the source program.
1831 Each enumerator entry also has a
1832 \DWATconstvalue{} attribute,
1833 whose value is the actual numeric value of the enumerator as
1834 represented on the target system.
1837 If the enumeration type occurs as the description of a
1838 \addtoindexx{enumeration type entry!as array dimension}
1839 dimension of an array type, and the stride for that dimension
1840 \hypertarget{chap:DWATbytestrideenumerationstridedimensionofarraytype}{}
1841 is different than what would otherwise be determined, then
1842 \hypertarget{chap:DWATbitstrideenumerationstridedimensionofarraytype}{}
1843 the enumeration type entry has either a
1845 or \DWATbitstride{} attribute
1846 \addtoindexx{bit stride attribute}
1847 which specifies the separation
1848 between successive elements along the dimension as described
1850 Section \refersec{chap:staticanddynamicvaluesofattributes}.
1852 \DWATbitstride{} attribute
1853 \addtoindexx{bit stride attribute}
1854 is interpreted as bits and the value of
1855 \addtoindexx{byte stride attribute}
1858 attribute is interpreted as bytes.
1861 \section{Subroutine Type Entries}
1862 \label{chap:subroutinetypeentries}
1864 \textit{It is possible in \addtoindex{C}
1865 to declare pointers to subroutines
1866 that return a value of a specific type. In both
1867 \addtoindex{C} and \addtoindex{C++},
1868 it is possible to declare pointers to subroutines that not
1869 only return a value of a specific type, but accept only
1870 arguments of specific types. The type of such pointers would
1871 be described with a \doublequote{pointer to} modifier applied to a
1872 user\dash defined type.}
1875 A subroutine type is represented by a debugging information
1877 \addtoindexx{subroutine type entry}
1878 tag \DWTAGsubroutinetypeTARG.
1880 been given to the subroutine type in the source program,
1881 then the corresponding subroutine type entry has
1882 a \DWATname{} attribute
1883 \addtoindexx{name attribute}
1884 whose value is a null\dash terminated string containing
1885 the subroutine type name as it appears in the source program.
1887 If the subroutine type describes a function that returns
1888 a value, then the subroutine type entry has
1889 \addtoindexx{type attribute}
1891 attribute to denote the type returned by the subroutine. If
1892 the types of the arguments are necessary to describe the
1893 subroutine type, then the corresponding subroutine type
1894 entry owns debugging information entries that describe the
1895 arguments. These debugging information entries appear in the
1896 order that the corresponding argument types appear in the
1899 \textit{In \addtoindex{C} there
1900 is a difference between the types of functions
1901 declared using function prototype style declarations and
1902 those declared using non\dash prototype declarations.}
1905 \hypertarget{chap:DWATprototypedsubroutineprototype}{}
1906 subroutine entry declared with a function prototype style
1907 declaration may have
1908 \addtoindexx{prototyped attribute}
1910 \DWATprototyped{} attribute, which is
1911 a \livelink{chap:classflag}{flag}.
1913 Each debugging information entry owned by a subroutine
1914 type entry corresponds to either a formal parameter or the sequence of
1915 unspecified parameters of the subprogram type:
1917 \begin{enumerate}[1. ]
1918 \item A formal parameter of a parameter list (that has a
1919 specific type) is represented by a debugging information entry
1920 with the tag \DWTAGformalparameter.
1921 Each formal parameter
1923 \addtoindexx{type attribute}
1924 a \DWATtype{} attribute that refers to the type of
1925 the formal parameter.
1927 \item The unspecified parameters of a variable parameter list
1928 \addtoindexx{unspecified parameters entry}
1930 \addtoindexx{\texttt{...} parameters|see{unspecified parameters entry}}
1931 represented by a debugging information entry with the
1932 tag \DWTAGunspecifiedparameters.
1935 \textit{\addtoindex{C++} const-volatile qualifiers are encoded as
1936 part of the type of the
1937 \doublequote{\texttt{this}}-pointer.
1938 \addtoindex{C++:2011 (ISO)} reference and rvalue-reference qualifiers are encoded using
1939 the \DWATreference{} and \DWATrvaluereference{} attributes, respectively.
1940 See also Section \refersec{chap:memberfunctionentries}.}
1943 A subroutine type entry may have the \DWATreference{} or
1944 \DWATrvaluereference{} attribute to indicate that it describes the
1945 type of a member function with reference or rvalue-reference
1946 semantics, respectively.
1949 \section{String Type Entries}
1950 \label{chap:stringtypeentries}
1952 \textit{A \doublequote{string} is a sequence of characters that have specific
1953 \addtoindexx{string type entry}
1954 semantics and operations that distinguish them from arrays of
1956 \addtoindex{Fortran} is one of the languages that has a string
1957 type. Note that \doublequote{string} in this context refers to a target
1958 machine concept, not the class string as used in this document
1959 (except for the name attribute).}
1961 A string type is represented by a debugging information entry
1962 with the tag \DWTAGstringtypeTARG.
1963 If a name has been given to
1964 the string type in the source program, then the corresponding
1965 string type entry has a
1966 \DWATname{} attribute
1967 \addtoindexx{name attribute}
1969 a null\dash terminated string containing the string type name as
1970 it appears in the source program.
1973 \addtoindex{Fortran 2003} language standard allows string
1974 types that are composed of different types of (same sized) characters.
1975 While there is no standard list of character kinds, the kinds
1976 \texttt{ASCII}\index{ASCII@\texttt{ASCII} (Fortran string kind)} (see \DWATEASCII),
1977 \texttt{ISO\_10646}\index{ISO\_10646@\texttt{ISO\_10646} (Fortran string kind)}
1978 \addtoindexx{ISO 10646 character set standard}
1980 \texttt{DEFAULT}\index{DEFAULT@\texttt{DEFAULT} (Fortran string kind)}
1983 A string type entry may have a \DWATtype{}
1984 \livetargi{char:DWAATtypeofstringtype}{attribute}{type attribute!of string type entry}
1985 describing how each character is encoded and is to be interpreted.
1986 The value of this attribute is a \CLASSreference to a
1987 \DWTAGbasetype{} base type entry. If the attribute is absent,
1988 then the character is encoded using the system default.
1991 The string type entry may have a
1992 \DWATbytesize{} attribute or
1994 attribute, whose value
1995 (see Section \refersec{chap:byteandbitsizes})
1997 storage needed to hold a value of the string type.
2000 \hypertarget{chap:DWATstringlengthstringlengthofstringtype}{}
2001 string type entry may also have a
2002 \DWATstringlength{} attribute
2004 \addtoindexx{string length attribute}
2006 \addtoindex{location description} yielding the location
2007 where the length of the string is stored in the program.
2008 If the \DWATstringlength{} attribute is not present, the size
2009 of the string is assumed to be the amount of storage that is
2010 allocated for the string (as specified by the \DWATbytesize{}
2011 or \DWATbitsize{} attribute).
2013 The string type entry may also have a
2014 \DWATstringlengthbytesizeNAME{}
2016 \DWATstringlengthbitsizeNAME{} attribute,
2017 \addtoindexx{string length attribute!size of length data}
2018 whose value (see Section \refersec{chap:byteandbitsizes})
2019 is the size of the data to be retrieved from the location
2020 referenced by the string length attribute. If no (byte or bit)
2021 size attribute is present, the size of the data to be retrieved
2023 \addtoindex{size of an address} on the target machine.
2026 \addtoindexx{DWARF Version 5} % Avoid italics
2027 \textit{Prior to DWARF Version 5, the meaning of a
2028 \DWATbytesize{} attribute depends on the presence of the
2029 \DWATstringlength{} attribute:
2031 \item If \DWATstringlength{} is present, \DWATbytesize{}
2032 specifies the size of the length data to be retrieved
2033 from the location specified by the \DWATstringlength{} attribute.
2034 \item If \DWATstringlength{} is not present, \DWATbytesize{}
2035 specifies the amount of storage allocated for objects
2038 In DWARF Version 5, \DWATbytesize{} always specifies the amount of storage
2039 allocated for objects of the string type.}
2042 \section{Set Type Entries}
2043 \label{chap:settypeentries}
2045 \textit{\addtoindex{Pascal} provides the concept of a \doublequote{set,} which represents
2046 a group of values of ordinal type.}
2048 A set is represented by a debugging information entry with
2049 the tag \DWTAGsettypeTARG.
2050 \addtoindexx{set type entry}
2051 If a name has been given to the
2052 set type, then the set type entry has
2053 a \DWATname{} attribute
2054 \addtoindexx{name attribute}
2055 whose value is a null\dash terminated string containing the
2056 set type name as it appears in the source program.
2058 The set type entry has
2059 \addtoindexx{type attribute}
2060 a \DWATtype{} attribute to denote the
2061 type of an element of the set.
2064 If the amount of storage allocated to hold each element of an
2065 object of the given set type is different from the amount of
2066 storage that is normally allocated to hold an individual object
2067 of the indicated element type, then the set type entry has
2068 either a \DWATbytesize{} attribute, or
2069 \DWATbitsize{} attribute
2070 whose value (see Section \refersec{chap:byteandbitsizes}) is
2071 the amount of storage needed to hold a value of the set type.
2074 \section{Subrange Type Entries}
2075 \label{chap:subrangetypeentries}
2077 \textit{Several languages support the concept of a \doublequote{subrange}
2078 type object. These objects can represent a subset of the
2079 values that an object of the basis type for the subrange can
2081 Subrange type entries may also be used to represent
2082 the bounds of array dimensions.}
2084 A subrange type is represented by a debugging information
2086 \addtoindexx{subrange type entry}
2087 tag \DWTAGsubrangetypeTARG.
2089 given to the subrange type, then the subrange type entry
2090 has a \DWATname{} attribute
2091 \addtoindexx{name attribute}
2092 whose value is a null\dash terminated
2093 string containing the subrange type name as it appears in
2096 The tag \DWTAGgenericsubrange{} is
2097 used to describe arrays with a dynamic rank. See Section
2098 \refersec{chap:DWTAGgenericsubrange}.
2100 The subrange entry may have
2101 \addtoindexx{type attribute}
2102 a \DWATtype{} attribute to describe
2103 the type of object, called the basis type, of whose values
2104 this subrange is a subset.
2106 If the amount of storage allocated to hold each element of an
2107 object of the given subrange type is different from the amount
2108 of storage that is normally allocated to hold an individual
2109 object of the indicated element type, then the subrange
2111 \DWATbytesize{} attribute or
2113 attribute, whose value
2114 (see Section \refersec{chap:staticanddynamicvaluesofattributes})
2116 storage needed to hold a value of the subrange type.
2119 \hypertarget{chap:DWATthreadsscaledupcarrayboundthreadsscalfactor}{}
2120 subrange entry may have
2121 \addtoindexx{threads scaled attribute}
2123 \DWATthreadsscaled{} attribute,
2124 which is a \livelink{chap:classflag}{flag}.
2125 If present, this attribute indicates whether
2126 this subrange represents a \addtoindex{UPC} array bound which is scaled
2127 by the runtime THREADS value (the number of \addtoindex{UPC} threads in
2128 this execution of the program).
2130 \textit{This allows the representation of a \addtoindex{UPC} shared array such as}
2132 \begin{lstlisting}[numbers=none]
2133 int shared foo[34*THREADS][10][20];
2137 \hypertarget{chap:DWATlowerboundlowerboundofsubrange}{}
2139 \hypertarget{chap:DWATupperboundupperboundofsubrange}{}
2140 entry may have the attributes
2142 \addtoindexx{lower bound attribute}
2143 and \DWATupperbound{}
2144 \addtoindexx{upper bound attribute} to specify, respectively, the lower
2145 and upper bound values of the subrange. The
2148 \hypertarget{chap:DWATcountelementsofsubrangetype}{}
2150 % FIXME: The following matches DWARF4: odd as there is no default count.
2151 \addtoindexx{count attribute!default}
2153 \addtoindexx{count attribute}
2155 \DWATcount{} attribute,
2157 value describes the number of elements in the subrange rather
2158 than the value of the last element. The value of each of
2159 these attributes is determined as described in
2160 Section \refersec{chap:staticanddynamicvaluesofattributes}.
2162 If the lower bound value is missing, the value is assumed to
2163 be a language\dash dependent default constant as defined in
2164 Table \refersec{tab:languageencodings}.
2165 \addtoindexx{lower bound attribute!default}
2167 If the upper bound and count are missing, then the upper bound value is
2168 \textit{unknown}.\addtoindexx{upper bound attribute!default unknown}
2170 If the subrange entry has no type attribute describing the
2171 basis type, the basis type is determined as follows:
2172 \begin{enumerate}[1. ]
2174 If there is a lower bound attribute that references an object,
2175 the basis type is assumed to be the same as the type of that object.
2177 Otherwise, if there is an upper bound or count attribute that references
2178 an object, the basis type is assumed to be the same as the type of that object.
2180 Otherwise, the type is
2181 assumed to be the same type, in the source language of the
2182 compilation unit containing the subrange entry, as a signed
2183 integer with the same size as an address on the target machine.
2186 If the subrange type occurs as the description of a dimension
2187 of an array type, and the stride for that dimension is
2188 \hypertarget{chap:DWATbytestridesubrangestridedimensionofarraytype}{}
2189 different than what would otherwise be determined, then
2190 \hypertarget{chap:DWATbitstridesubrangestridedimensionofarraytype}{}
2191 the subrange type entry has either
2192 \addtoindexx{byte stride attribute}
2194 \DWATbytestride{} or
2195 \DWATbitstride{} attribute
2196 \addtoindexx{bit stride attribute}
2197 which specifies the separation
2198 between successive elements along the dimension as described
2200 Section \refersec{chap:byteandbitsizes}.
2202 \textit{Note that the stride can be negative.}
2205 \section{Pointer to Member Type Entries}
2206 \label{chap:pointertomembertypeentries}
2208 \textit{In \addtoindex{C++}, a
2209 pointer to a data or function member of a class or
2210 structure is a unique type.}
2212 A debugging information entry representing the type of an
2213 object that is a pointer to a structure or class member has
2214 the tag \DWTAGptrtomembertypeTARG.
2216 If the \addtoindex{pointer to member type} has a name, the
2217 \addtoindexx{pointer to member type entry}
2218 pointer to member entry has a
2219 \DWATname{} attribute,
2220 \addtoindexx{name attribute}
2222 null\dash terminated string containing the type name as it appears
2223 in the source program.
2225 The \addtoindex{pointer to member} entry
2227 \addtoindexx{type attribute}
2228 a \DWATtype{} attribute to
2229 describe the type of the class or structure member to which
2230 objects of this type may point.
2232 The \addtoindexx{pointer to member} entry also
2233 \hypertarget{chap:DWATcontainingtypecontainingtypeofpointertomembertype}{}
2235 \DWATcontainingtype{}
2236 attribute, whose value is a \livelink{chap:classreference}{reference} to a debugging
2237 information entry for the class or structure to whose members
2238 objects of this type may point.
2240 The \addtoindex{pointer to member entry}
2241 \hypertarget{chap:DWATuselocationmemberlocationforpointertomembertype}{}
2243 \DWATuselocation{} attribute
2244 \addtoindexx{use location attribute}
2246 \addtoindex{location description} that computes the
2247 address of the member of the class to which the pointer to
2248 member entry points.
2250 \textit{The method used to find the address of a given member of a
2251 class or structure is common to any instance of that class
2252 or structure and to any instance of the pointer or member
2253 type. The method is thus associated with the type entry,
2254 rather than with each instance of the type.}
2256 The \DWATuselocation{} description is used in conjunction
2257 with the location descriptions for a particular object of the
2258 given \addtoindex{pointer to member type} and for a particular structure or
2259 class instance. The \DWATuselocation{}
2260 attribute expects two values to be
2261 \addtoindexi{pushed}{address!implicit push for member operator}
2262 onto the DWARF expression stack before
2263 the \DWATuselocation{} description is evaluated.
2265 \addtoindexi{pushed}{address!implicit push for member operator}
2266 is the value of the \addtoindex{pointer to member} object
2267 itself. The second value
2268 \addtoindexi{pushed}{address!implicit push for member operator}
2269 is the base address of the
2270 entire structure or union instance containing the member
2271 whose address is being calculated.
2274 \textit{For an expression such as}
2276 \begin{lstlisting}[numbers=none]
2279 \textit{where \texttt{mbr\_ptr} has some \addtoindex{pointer to member type}, a debugger should:}
2280 \begin{enumerate}[1. ]
2281 \item \textit{Push the value of \texttt{mbr\_ptr} onto the DWARF expression stack.}
2282 \item \textit{Push the base address of \texttt{object} onto the DWARF expression stack.}
2283 \item \textit{Evaluate the \DWATuselocation{} description
2284 given in the type of \texttt{mbr\_ptr}.}
2288 \section{File Type Entries}
2289 \label{chap:filetypeentries}
2291 \textit{Some languages, such as \addtoindex{Pascal},
2292 provide a data type to represent
2295 A file type is represented by a debugging information entry
2297 \addtoindexx{file type entry}
2300 If the file type has a name,
2301 the file type entry has a \DWATname{} attribute,
2302 \addtoindexx{name attribute}
2304 is a null\dash terminated string containing the type name as it
2305 appears in the source program.
2307 The file type entry has
2308 \addtoindexx{type attribute}
2309 a \DWATtype{} attribute describing
2310 the type of the objects contained in the file.
2312 The file type entry also has a
2313 \DWATbytesize{}\addtoindexx{byte size attribute} or
2314 \DWATbitsize{}\addtoindexx{bit size attribute} attribute, whose value
2315 (see Section \refersec{chap:staticanddynamicvaluesofattributes})
2316 is the amount of storage need to hold a value of the file type.
2318 \section{Dynamic Type Entries and Properties}
2320 \subsection{Dynamic Type Entries}
2321 \textit{Some languages such as
2322 \addtoindex{Fortran 90}, provide types whose values
2323 may be dynamically allocated or associated with a variable
2324 under explicit program control. However, unlike the related
2325 pointer type in \addtoindex{C} or
2326 \addtoindex{C++}, the indirection involved in accessing
2327 the value of the variable is generally implicit, that is, not
2328 indicated as part of program source.}
2330 A dynamic type entry is used to declare a dynamic type that is
2331 \doublequote{just like} another non-dynamic type without needing to
2332 replicate the full description of that other type.
2334 A dynamic type is represented by a debugging information entry
2335 with the tag \DWTAGdynamictypeTARG. If a name has been given to the
2336 dynamic type, then the dynamic type has a \DWATname{} attribute
2337 whose value is a null-terminated string containing the dynamic
2338 type name as it appears in the source.
2340 A dynamic type entry has a \DWATtype{} attribute whose value is a
2341 reference to the type of the entities that are dynamically allocated.
2343 A dynamic type entry also has a \DWATdatalocation, and may also
2344 have \DWATallocated{} and/or \DWATassociated{} attributes as
2345 described in Section \referfol{chap:dynamictypeproperties}.
2346 A \DWATdatalocation, \DWATallocated{} or \DWATassociated{} attribute
2347 may not occur on a dynamic type entry if the same kind of attribute
2348 already occurs on the type referenced by the \DWATtype{} attribute.
2350 \subsection{Dynamic Type Properties}
2351 \label{chap:dynamictypeproperties}
2353 The \DWATdatalocation, \DWATallocated{} and \DWATassociated{}
2354 attributes described in this section can be used for any type, not
2355 just dynamic types.}
2358 \subsubsection{Data Location}
2359 \label{chap:datalocation}
2361 \textit{Some languages may represent objects using descriptors to hold
2362 information, including a location and/or run\dash time parameters,
2363 about the data that represents the value for that object.}
2365 \hypertarget{chap:DWATdatalocationindirectiontoactualdata}{}
2366 The \DWATdatalocation{}
2367 attribute may be used with any
2368 \addtoindexx{data location attribute}
2369 type that provides one or more levels of
2370 \addtoindexx{hidden indirection|see{data location attribute}}
2372 and/or run\dash time parameters in its representation. Its value
2373 is a \addtoindex{location description}.
2374 The result of evaluating this
2375 description yields the location of the data for an object.
2376 When this attribute is omitted, the address of the data is
2377 the same as the address of the object.
2380 \textit{This location description will typically begin with
2381 \DWOPpushobjectaddress{}
2382 which loads the address of the
2383 object which can then serve as a descriptor in subsequent
2384 calculation. For an example using
2386 for a \addtoindex{Fortran 90 array}, see
2387 Appendix \refersec{app:fortranarrayexample}.}
2389 \subsubsection{Allocation and Association Status}
2390 \label{chap:allocationandassociationstatus}
2392 \textit{Some languages, such as \addtoindex{Fortran 90},
2393 provide types whose values
2394 may be dynamically allocated or associated with a variable
2395 under explicit program control.}
2397 \hypertarget{chap:DWATallocatedallocationstatusoftypes}{}
2401 \addtoindexx{allocated attribute}
2402 may optionally be used with any
2403 type for which objects of the type can be explicitly allocated
2404 and deallocated. The presence of the attribute indicates that
2405 objects of the type are allocatable and deallocatable. The
2406 integer value of the attribute (see below) specifies whether
2407 an object of the type is
2408 currently allocated or not.
2411 \hypertarget{chap:DWATassociatedassociationstatusoftypes}{}
2413 \DWATassociated{} attribute
2415 \addtoindexx{associated attribute}
2416 optionally be used with
2417 any type for which objects of the type can be dynamically
2418 associated with other objects. The presence of the attribute
2419 indicates that objects of the type can be associated. The
2420 integer value of the attribute (see below) indicates whether
2421 an object of the type is currently associated or not.
2423 \textit{While these attributes are defined specifically with
2424 \addtoindex{Fortran 90} ALLOCATABLE and POINTER types
2425 in mind, usage is not limited
2426 to just that language.}
2428 The value of these attributes is determined as described in
2429 Section \refersec{chap:staticanddynamicvaluesofattributes}.
2431 A non\dash zero value is interpreted as allocated or associated,
2432 and zero is interpreted as not allocated or not associated.
2434 \textit{For \addtoindex{Fortran 90},
2435 if the \DWATassociated{}
2436 attribute is present,
2437 the type has the POINTER property where either the parent
2438 variable is never associated with a dynamic object or the
2439 implementation does not track whether the associated object
2440 is static or dynamic. If the \DWATallocated{} attribute is
2441 present and the \DWATassociated{} attribute is not, the type
2442 has the ALLOCATABLE property. If both attributes are present,
2443 then the type should be assumed to have the POINTER property
2444 (and not ALLOCATABLE); the \DWATallocated{} attribute may then
2445 be used to indicate that the association status of the object
2446 resulted from execution of an ALLOCATE statement rather than
2447 pointer assignment.}
2449 \textit{For examples using
2450 \DWATallocated{} for \addtoindex{Ada} and
2451 \addtoindex{Fortran 90}
2453 see Appendix \refersec{app:aggregateexamples}.}
2455 \subsubsection{Array Rank}
2456 \label{chap:DWATrank}
2457 \addtoindexx{array!assumed-rank}
2458 \addtoindexx{assumed-rank array|see{array, assumed-rank}}
2459 \textit{The Fortran language supports \doublequote{assumed-rank arrays}. The
2460 rank (the number of dimensions) of an assumed-rank array is unknown
2461 at compile time. The Fortran runtime stores the rank in the array
2462 descriptor metadata.}
2465 \hypertarget{chap:DWATrankofdynamicarray}{\DWATrankINDX}
2466 attribute indicates that an array's rank
2467 (number of dimensions) is dynamic, and therefore unknown at compile
2468 time. The value of the \DWATrankNAME{} attribute is either an integer constant
2469 or a location expression whose evaluation yields the dynamic rank.
2471 The bounds of an array with dynamic rank are described using a
2472 \DWTAGgenericsubrange{} entry, which
2473 is the dynamic rank array equivalent of
2474 \DWTAGsubrangetype. The
2475 difference is that a \DWTAGgenericsubrange{} entry contains generic
2476 lower/upper bound and stride expressions that need to be evaluated for
2477 each dimension. Before any expression contained in a
2478 \DWTAGgenericsubrange{} can be evaluated, the dimension for which the
2479 expression is to be evaluated needs to be pushed onto the stack. The
2480 expression will use it to find the offset of the respective field in
2481 the array descriptor metadata.
2483 \textit{The Fortran compiler is free to choose any layout for the
2484 array descriptor. In particular, the upper and lower bounds and
2485 stride values do not need to be bundled into a structure or record,
2486 but could be laid end to end in the containing descriptor, pointed
2487 to by the descriptor, or even allocated independently of the
2490 Dimensions are enumerated $0$ to $\mathit{rank}-1$ in a left-to-right
2493 \textit{For an example in Fortran 2008, see
2494 Section~\refersec{app:assumedrankexample}.}
2497 \section{Template Alias Entries}
2498 \label{chap:templatealiasentries}
2501 In \addtoindex{C++}, a template alias is a form of typedef that has template
2502 parameters. DWARF does not represent the template alias definition
2503 but does represent instantiations of the alias.
2506 A type named using a template alias is represented
2507 by a debugging information entry
2508 \addtoindexx{template alias entry}
2510 \DWTAGtemplatealiasTARG.
2511 The template alias entry has a
2512 \DWATname{} attribute
2513 \addtoindexx{name attribute}
2514 whose value is a null\dash terminated string
2515 containing the name of the template alias as it appears in
2517 The template alias entry has child entries describing the template
2518 actual parameters (see Section \refersec{chap:templateparameters}).