1 \chapter{Data Representation}
2 \label{datarep:datarepresentation}
4 This section describes the binary representation of the
5 debugging information entry itself, of the attribute types
6 and of other fundamental elements described above.
9 \section{Vendor Extensibility}
10 \label{datarep:vendorextensibility}
11 \addtoindexx{vendor extensibility}
12 \addtoindexx{vendor specific extensions|see{vendor extensibility}}
15 \addtoindexx{extensibility|see{vendor extensibility}}
16 reserve a portion of the DWARF name space and ranges of
17 enumeration values for use for vendor specific extensions,
18 special labels are reserved for tag names, attribute names,
19 base type encodings, location operations, language names,
20 calling conventions and call frame instructions.
22 The labels denoting the beginning and end of the reserved
23 \hypertarget{chap:DWXXXlohiuser}{}
24 value range for vendor specific extensions consist of the
26 (\DWATlouserMARK{}\DWAThiuserMARK{} DW\_AT,
27 \DWATElouserMARK{}\DWATEhiuserMARK{} DW\_ATE,
28 \DWCClouserMARK{}\DWCChiuserMARK{} DW\_CC,
29 \DWCFAlouserMARK{}\DWCFAhiuserMARK{} DW\_CFA
30 \DWENDlouserMARK{}\DWENDhiuserMARK{} DW\_END,
31 \DWLANGlouserMARK{}\DWLANGhiuserMARK{} DW\_LANG,
32 \DWLNElouserMARK{}\DWLNEhiuserMARK{} DW\_LNE,
33 \DWMACROlouserMARK{}\DWMACROhiuserMARK{}DW\_MACRO,
34 \DWOPlouserMARK{}\DWOPhiuserMARK{} DW\_OP or
35 \DWTAGlouserMARK{}\DWTAGhiuserMARK{} DW\_TAG,
36 respectively) followed by
37 \_lo\_user or \_hi\_user.
38 Values in the range between \textit{prefix}\_lo\_user
39 and \textit{prefix}\_hi\_user inclusive,
40 are reserved for vendor specific extensions. Vendors may
41 use values in this range without conflicting with current or
42 future system\dash defined values. All other values are reserved
43 for use by the system.
45 \textit{For example, for DIE tags, the special
46 labels are \DWTAGlouserNAME{} and \DWTAGhiuserNAME.}
48 \textit{There may also be codes for vendor specific extensions
49 between the number of standard line number opcodes and
50 the first special line number opcode. However, since the
51 number of standard opcodes varies with the DWARF version,
52 the range for extensions is also version dependent. Thus,
53 \DWLNSlouserTARG{} and
54 \DWLNShiuserTARG{} symbols are not defined.
57 Vendor defined tags, attributes, base type encodings, location
58 atoms, language names, line number actions, calling conventions
59 and call frame instructions, conventionally use the form
60 \text{prefix\_vendor\_id\_name}, where
61 \textit{vendor\_id}\addtoindexx{vendor id} is some identifying
62 character sequence chosen so as to avoid conflicts with
65 To ensure that extensions added by one vendor may be safely
66 ignored by consumers that do not understand those extensions,
67 the following rules should be followed:
68 \begin{enumerate}[1. ]
70 \item New attributes should be added in such a way that a
71 debugger may recognize the format of a new attribute value
72 without knowing the content of that attribute value.
74 \item The semantics of any new attributes should not alter
75 the semantics of previously existing attributes.
77 \item The semantics of any new tags should not conflict with
78 the semantics of previously existing tags.
80 \item Do not add any new forms of attribute value.
85 \section{Reserved Values}
86 \label{datarep:reservedvalues}
87 \subsection{Error Values}
88 \label{datarep:errorvalues}
89 \addtoindexx{reserved values!error}
92 \addtoindexx{error value}
93 a convenience for consumers of DWARF information, the value
94 0 is reserved in the encodings for attribute names, attribute
95 forms, base type encodings, location operations, languages,
96 line number program opcodes, macro information entries and tag
97 names to represent an error condition or unknown value. DWARF
98 does not specify names for these reserved values, since they
99 do not represent valid encodings for the given type and should
100 not appear in DWARF debugging information.
103 \subsection{Initial Length Values}
104 \label{datarep:initiallengthvalues}
105 \addtoindexx{reserved values!initial length}
107 An \livetarg{datarep:initiallengthvalues}{initial length field} is one of the length fields that occur
109 of those DWARF sections that
114 \dotdebugpubnames{}, and
115 \dotdebugpubtypes{}) or the length field
116 that occurs at the beginning of the CIE and FDE structures
117 in the \dotdebugframe{} section.
119 In an \addtoindex{initial length field}, the values \wfffffffzero through
120 \wffffffff are reserved by DWARF to indicate some form of
121 extension relative to \addtoindex{DWARF Version 2}; such values must not
122 be interpreted as a length field. The use of one such value,
123 \xffffffff, is defined below
124 (see Section \refersec{datarep:32bitand64bitdwarfformats});
126 the other values is reserved for possible future extensions.
130 \section{Relocatable, Split, Executable and Shared Objects}
131 \label{datarep:executableobjectsandsharedobjects}
133 \subsection{Relocatable Objects}
135 \subsection{Split DWARF Objects}
136 \label{datarep:splitdwarfobjects}
137 A DWARF producer may partition the debugging
138 information such that the majority of the debugging
139 information can remain in individual object files without
140 being processed by the linker. The first partition contains
141 debugging information that must still be processed by the linker,
142 and includes the following:
145 The line number tables, range tables, frame tables, and
146 accelerated access tables, in the usual sections:
147 \dotdebugline, \dotdebugranges, \dotdebugframe,
148 \dotdebugpubnames, \dotdebugpubtypes{} and \dotdebugaranges,
151 An address table, in the \dotdebugaddr{} section. This table
152 contains all addresses and constants that require
153 link-time relocation, and items in the table can be
154 referenced indirectly from the debugging information via
155 the \DWFORMaddrx{} form, and by the \DWOPaddrx{} and
156 \DWOPconstx{} operators.
158 A skeleton compilation unit, as described in Section
159 \refersec{chap:skeletoncompilationunitentries},
160 in the \dotdebuginfo{} section.
162 An abbreviations table for the skeleton compilation unit,
163 in the \dotdebugabbrev{} section.
165 A string table, in the \dotdebugstr{} section. The string
166 table is necessary only if the skeleton compilation unit
167 uses either indirect string form, \DWFORMstrp{} or
170 A string offsets table, in the \dotdebugstroffsets{}
171 section. The string offsets table is necessary only if
172 the skeleton compilation unit uses the \DWFORMstrx{} form.
174 The attributes contained in the skeleton compilation
175 unit can be used by a DWARF consumer to find the object file
176 or DWARF object file that contains the second partition.
178 The second partition contains the debugging information that
179 does not need to be processed by the linker. These sections
180 may be left in the object files and ignored by the linker
181 (that is, not combined and copied to the executable object), or
182 they may be placed by the producer in a separate DWARF object
183 file. This partition includes the following:
186 The full compilation unit, in the \dotdebuginfodwo{} section.
187 Attributes in debugging information entries may refer to
188 machine addresses indirectly using the \DWFORMaddrx{} form,
189 and location expressions may do so using the \DWOPaddrx{} and
190 \DWOPconstx{} forms. Attributes may refer to range table
191 entries with an offset relative to a base offset in the
192 range table for the compilation unit.
194 \item Separate type units, in the \dotdebuginfodwo{} section.
197 Abbreviations table(s) for the compilation unit and type
198 units, in the \dotdebugabbrevdwo{} section.
200 \item Location lists, in the \dotdebuglocdwo{} section.
203 A skeleton line table (for the type units), in the
204 \dotdebuglinedwo{} section (see
205 Section \refersec{chap:skeletoncompilationunitentries}).
207 \item Macro information, in the \dotdebugmacrodwo{} section.
209 \item A string table, in the \dotdebugstrdwo{} section.
211 \item A string offsets table, in the \dotdebugstroffsetsdwo{}
215 Except where noted otherwise, all references in this document
216 to a debugging information section (for example, \dotdebuginfo),
217 applies also to the corresponding split DWARF section (for example,
220 \subsection{Executable Objects}
221 \label{chap:executableobjects}
222 The relocated addresses in the debugging information for an
223 executable object are virtual addresses.
225 \subsection{Shared Objects}
226 \label{datarep:sharedobjects}
228 addresses in the debugging information for a shared object
229 are offsets relative to the start of the lowest region of
230 memory loaded from that shared object.
232 \textit{This requirement makes the debugging information for
233 shared objects position independent. Virtual addresses in a
234 shared object may be calculated by adding the offset to the
235 base address at which the object was attached. This offset
236 is available in the run\dash time linker\textquoteright s data structures.}
240 \section{32-Bit and 64-Bit DWARF Formats}
241 \label{datarep:32bitand64bitdwarfformats}
242 \hypertarget{datarep:xxbitdwffmt}{}
243 \addtoindexx{32-bit DWARF format}
244 \addtoindexx{64-bit DWARF format}
245 There are two closely related file formats. In the 32\dash bit DWARF
246 format, all values that represent lengths of DWARF sections
247 and offsets relative to the beginning of DWARF sections are
248 represented using 32\dash bits. In the 64\dash bit DWARF format, all
249 values that represent lengths of DWARF sections and offsets
250 relative to the beginning of DWARF sections are represented
251 using 64\dash bits. A special convention applies to the initial
252 length field of certain DWARF sections, as well as the CIE and
253 FDE structures, so that the 32\dash bit and 64\dash bit DWARF formats
254 can coexist and be distinguished within a single linked object.
256 The differences between the 32\dash\ and 64\dash bit
258 detailed in the following:
259 \begin{enumerate}[1. ]
261 \item In the 32\dash bit DWARF format, an
262 \addtoindex{initial length field}
264 \addtoindexx{initial length field!encoding}
265 Section \refersec{datarep:initiallengthvalues})
266 is an unsigned 32\dash bit integer (which
267 must be less than \xfffffffzero); in the 64\dash bit DWARF format,
268 an \addtoindex{initial length field} is 96 bits in size,
271 \item The first 32\dash bits have the value \xffffffff.
273 \item The following 64\dash bits contain the actual length
274 represented as an unsigned 64\dash bit integer.
277 \textit{This representation allows a DWARF consumer to dynamically
278 detect that a DWARF section contribution is using the 64\dash bit
279 format and to adapt its processing accordingly.}
281 \item Section offset and section length
282 \hypertarget{datarep:sectionoffsetlength}{}
283 \addtoindexx{section length!use in headers}
285 \addtoindexx{section offset!use in headers}
286 in the headers of DWARF sections (other
287 \addtoindexx{initial length field}
289 \addtoindex{initial length}
290 fields) are listed following. In the 32\dash bit DWARF format these
291 are 32\dash bit unsigned integer values; in the 64\dash bit DWARF format,
293 \addtoindexx{section length!in .debug\_aranges header}
295 \addtoindexx{section length!in .debug\_pubnames header}
297 \addtoindexx{section length!in .debug\_pubtypes header}
298 unsigned integer values.
302 Section &Name & Role \\ \hline
303 \dotdebugaranges{} & \addtoindex{debug\_info\_offset} & offset in \dotdebuginfo{} \\
304 \dotdebugframe{}/CIE & \addtoindex{CIE\_id} & CIE distinguished value \\
305 \dotdebugframe{}/FDE & \addtoindex{CIE\_pointer} & offset in \dotdebugframe{} \\
306 \dotdebuginfo{} & \addtoindex{debug\_abbrev\_offset} & offset in \dotdebugabbrev{} \\
307 \dotdebugline{} & \addtoindex{header\_length} & length of header itself \\
308 \dotdebugpubnames{} & \addtoindex{debug\_info\_offset} & offset in \dotdebuginfo{} \\
309 & \addtoindex{debug\_info\_length} & length of \dotdebuginfo{} \\
311 \dotdebugpubtypes{} & \addtoindex{debug\_info\_offset} & offset in \dotdebuginfo{} \\
312 & \addtoindex{debug\_info\_length} & length of \dotdebuginfo{} \\
317 The \texttt{CIE\_id} field in a CIE structure must be 64 bits because
318 it overlays the \texttt{CIE\_pointer} in a FDE structure; this implicit
319 union must be accessed to distinguish whether a CIE or FDE is
320 present, consequently, these two fields must exactly overlay
321 each other (both offset and size).
323 \item Within the body of the \dotdebuginfo{}
324 section, certain forms of attribute value depend on the choice
325 of DWARF format as follows. For the 32\dash bit DWARF format,
326 the value is a 32\dash bit unsigned integer; for the 64\dash bit DWARF
327 format, the value is a 64\dash bit unsigned integer.
330 Form & Role \\ \hline
331 \DWFORMrefaddr& offset in \dotdebuginfo{} \\
332 \DWFORMsecoffset& offset in a section other than \\
333 &\dotdebuginfo{} or \dotdebugstr{} \\
334 \DWFORMstrp&offset in \dotdebugstr{} \\
335 \DWOPcallref&offset in \dotdebuginfo{} \\
339 \item Within the body of the \dotdebugpubnames{} and
341 sections, the representation of the first field
342 of each tuple (which represents an offset in the
344 section) depends on the DWARF format as follows: in the
345 32\dash bit DWARF format, this field is a 32\dash bit unsigned integer;
346 in the 64\dash bit DWARF format, it is a 64\dash bit unsigned integer.
349 \item In the body of the \dotdebugstroffsets{} and \dotdebugstroffsetsdwo{}
350 sections, the size of entries in the body depend on the DWARF
351 format as follows: in the 32-bit DWARF format, entries are 32-bit
352 unsigned integer values; in the 64-bit DWARF format, they are
353 64-bit unsigned integers.
355 \item In the body of the \dotdebugaddr{}, \dotdebugloc{} and \dotdebugranges{}
356 sections, the contents of the address size fields depends on the
357 DWARF format as follows: in the 32-bit DWARF format, these fields
358 contain 4; in the 64-bit DWARF format these fields contain 8.
362 The 32\dash bit and 64\dash bit DWARF format conventions must \emph{not} be
363 intermixed within a single compilation unit.
365 \textit{Attribute values and section header fields that represent
366 addresses in the target program are not affected by these
369 A DWARF consumer that supports the 64\dash bit DWARF format must
370 support executables in which some compilation units use the
371 32\dash bit format and others use the 64\dash bit format provided that
372 the combination links correctly (that is, provided that there
373 are no link\dash time errors due to truncation or overflow). (An
374 implementation is not required to guarantee detection and
375 reporting of all such errors.)
377 \textit{It is expected that DWARF producing compilers will \emph{not} use
378 the 64\dash bit format \emph{by default}. In most cases, the division of
379 even very large applications into a number of executable and
380 shared objects will suffice to assure that the DWARF sections
381 within each individual linked object are less than 4 GBytes
382 in size. However, for those cases where needed, the 64\dash bit
383 format allows the unusual case to be handled as well. Even
384 in this case, it is expected that only application supplied
385 objects will need to be compiled using the 64\dash bit format;
386 separate 32\dash bit format versions of system supplied shared
387 executable libraries can still be used.}
391 \section{Format of Debugging Information}
392 \label{datarep:formatofdebugginginformation}
394 For each compilation unit compiled with a DWARF producer,
395 a contribution is made to the \dotdebuginfo{} section of
396 the object file. Each such contribution consists of a
397 compilation unit header
398 (see Section \refersec{datarep:compilationunitheader})
400 single \DWTAGcompileunit{} or
401 \DWTAGpartialunit{} debugging
402 information entry, together with its children.
404 For each type defined in a compilation unit, a separate
405 contribution may also be made to the
407 section of the object file. Each
408 such contribution consists of a
409 \addtoindex{type unit} header
410 (see Section \refersec{datarep:typeunitheader})
411 followed by a \DWTAGtypeunit{} entry, together with
414 Each debugging information entry begins with a code that
415 represents an entry in a separate
416 \addtoindex{abbreviations table}. This
417 code is followed directly by a series of attribute values.
419 The appropriate entry in the
420 \addtoindex{abbreviations table} guides the
421 interpretation of the information contained directly in the
422 \dotdebuginfo{} section.
425 Multiple debugging information entries may share the same
426 abbreviation table entry. Each compilation unit is associated
427 with a particular abbreviation table, but multiple compilation
428 units may share the same table.
430 \subsection{Unit Headers}
431 \label{datarep:unitheaders}
432 Unit headers contain a field, \texttt{unit\_type}, whose value indicates the kind of
433 compilation unit that follows. The encodings for the unit type
434 enumeration are shown in Table \refersec{tab:unitheaderunitkindencodings}.
438 \setlength{\extrarowheight}{0.1cm}
439 \begin{longtable}{l|l}
440 \caption{Unit header unit kind encodings}
441 \label{tab:unitheaderunitkindencodings}
442 \addtoindexx{Unit header unit kind encodings} \\
443 \hline \bfseries Unit header unit kind encodings&\bfseries Value \\ \hline
445 \bfseries Unit header unit kind encodings&\bfseries Value \\ \hline
447 \hline \emph{Continued on next page}
451 \DWUTcompileTARG &0x01 \\
452 \DWUTtypeTARG &0x02 \\
453 \DWUTpartialTARG &0x03 \\ \hline
457 \subsubsection{Compilation Unit Header}
458 \label{datarep:compilationunitheader}
459 \begin{enumerate}[1. ]
461 \item \texttt{unit\_length} (\livelink{datarep:initiallengthvalues}{initial length}) \\
462 \addttindexx{unit\_length}
463 A 4\dash byte or 12\dash byte
464 \addtoindexx{initial length}
465 unsigned integer representing the length
466 of the \dotdebuginfo{}
467 contribution for that compilation unit,
468 not including the length field itself. In the \thirtytwobitdwarfformat,
469 this is a 4\dash byte unsigned integer (which must be less
470 than \xfffffffzero); in the \sixtyfourbitdwarfformat, this consists
471 of the 4\dash byte value \wffffffff followed by an 8\dash byte unsigned
472 integer that gives the actual length
473 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
475 \item \texttt{version} (\addtoindex{uhalf}) \\
476 A 2\dash byte unsigned integer representing the version of the
477 DWARF information for the compilation unit \addtoindexx{version number!compilation unit}
478 (see Appendix \refersec{app:dwarfsectionversionnumbersinformative}).
479 The value in this field is \versiondotdebuginfo.
482 \item \texttt{unit\_type} (\addtoindex{ubyte}) \\
483 A 1-byte unsigned integer identifying this unit as a compilation unit.
484 The value of this field is
485 \DWUTcompile{} for a {normal compilation} unit or
486 \DWUTpartial{} for a {partial compilation} unit
487 (see Section \refersec{chap:normalandpartialcompilationunitentries}).
489 \textit{This field is new in \DWARFVersionV.}
492 \item \addttindex{debug\_abbrev\_offset} (\livelink{datarep:sectionoffsetlength}{section offset}) \\
494 \addtoindexx{section offset!in .debug\_info header}
495 4\dash byte or 8\dash byte unsigned offset into the
497 section. This offset associates the compilation unit with a
498 particular set of debugging information entry abbreviations. In
499 the \thirtytwobitdwarfformat, this is a 4\dash byte unsigned length;
500 in the \sixtyfourbitdwarfformat, this is an 8\dash byte unsigned length
501 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
503 \item \texttt{address\_size} (\addtoindex{ubyte}) \\
504 A 1\dash byte unsigned integer representing the size in bytes of
505 \addttindexx{address\_size}
506 an address on the target architecture. If the system uses
507 \addtoindexx{address space!segmented}
508 segmented addressing, this value represents the size of the
509 offset portion of an address.
514 \subsubsection{Type Unit Header}
515 \label{datarep:typeunitheader}
517 The header for the series of debugging information entries
518 contributing to the description of a type that has been
519 placed in its own \addtoindex{type unit}, within the
520 \dotdebuginfo{} section,
521 consists of the following information:
522 \begin{enumerate}[1. ]
524 \item \texttt{unit\_length} (\livelink{datarep:initiallengthvalues}{initial length}) \\
525 \addttindexx{unit\_length}
526 A 4\dash byte or 12\dash byte unsigned integer
527 \addtoindexx{initial length}
528 representing the length
529 of the \dotdebuginfo{} contribution for that type unit,
530 not including the length field itself. In the \thirtytwobitdwarfformat,
531 this is a 4\dash byte unsigned integer (which must be
532 less than \xfffffffzero); in the \sixtyfourbitdwarfformat, this
533 consists of the 4\dash byte value \wffffffff followed by an
534 8\dash byte unsigned integer that gives the actual length
535 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
538 \item \texttt{version} (\addtoindex{uhalf}) \\
539 A 2\dash byte unsigned integer representing the version of the
540 DWARF information for the
541 type unit\addtoindexx{version number!type unit}
542 (see Appendix \refersec{app:dwarfsectionversionnumbersinformative}).
543 The value in this field is \versiondotdebuginfo.
545 \item \texttt{unit\_type} (\addtoindex{ubyte}) \\
546 A 1-byte unsigned integer identifying this unit as a type unit.
547 The value of this field is \DWUTtype{} for a type unit
548 (see Section \refersec{chap:separatetypeunitentries}).
550 \textit{This field is new in \DWARFVersionV.}
553 \item \addttindex{debug\_abbrev\_offset} (\livelink{datarep:sectionoffsetlength}{section offset}) \\
555 \addtoindexx{section offset!in .debug\_types header}
556 4\dash byte or 8\dash byte unsigned offset into the
558 section. This offset associates the type unit with a
559 particular set of debugging information entry abbreviations. In
560 the \thirtytwobitdwarfformat, this is a 4\dash byte unsigned length;
561 in the \sixtyfourbitdwarfformat, this is an 8\dash byte unsigned length
562 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
565 \item \texttt{address\_size} (ubyte) \\
566 A 1\dash byte unsigned integer representing the size
567 \addtoindexx{size of an address}
569 \addttindexx{address\_size}
570 an address on the target architecture. If the system uses
571 \addtoindexx{address space!segmented}
572 segmented addressing, this value represents the size of the
573 offset portion of an address.
575 \item \texttt{type\_signature} (8\dash byte unsigned integer) \\
576 \addtoindexx{type signature}
578 \addttindexx{type\_signature}
579 64\dash bit unique signature (see Section
580 \refersec{datarep:typesignaturecomputation})
581 of the type described in this type
584 \textit{An attribute that refers (using
585 \DWFORMrefsigeight{}) to
586 the primary type contained in this
587 \addtoindex{type unit} uses this value.}
589 \item \texttt{type\_offset} (\livelink{datarep:sectionoffsetlength}{section offset}) \\
590 \addttindexx{type\_offset}
591 A 4\dash byte or 8\dash byte unsigned offset
592 \addtoindexx{section offset!in .debug\_types header}
593 relative to the beginning
594 of the \addtoindex{type unit} header.
595 This offset refers to the debugging
596 information entry that describes the type. Because the type
597 may be nested inside a namespace or other structures, and may
598 contain references to other types that have not been placed in
599 separate type units, it is not necessarily either the first or
600 the only entry in the type unit. In the \thirtytwobitdwarfformat,
601 this is a 4\dash byte unsigned length; in the \sixtyfourbitdwarfformat,
602 this is an 8\dash byte unsigned length
603 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
607 \subsection{Debugging Information Entry}
608 \label{datarep:debugginginformationentry}
610 Each debugging information entry begins with an unsigned LEB128
611 number containing the abbreviation code for the entry. This
612 code represents an entry within the abbreviations table
613 associated with the compilation unit containing this entry. The
614 abbreviation code is followed by a series of attribute values.
616 On some architectures, there are alignment constraints on
617 section boundaries. To make it easier to pad debugging
618 information sections to satisfy such constraints, the
619 abbreviation code 0 is reserved. Debugging information entries
620 consisting of only the abbreviation code 0 are considered
623 \subsection{Abbreviations Tables}
624 \label{datarep:abbreviationstables}
626 The abbreviations tables for all compilation units
627 are contained in a separate object file section called
629 As mentioned before, multiple compilation
630 units may share the same abbreviations table.
632 The abbreviations table for a single compilation unit consists
633 of a series of abbreviation declarations. Each declaration
634 specifies the tag and attributes for a particular form of
635 debugging information entry. Each declaration begins with
636 an unsigned LEB128 number representing the abbreviation
637 code itself. It is this code that appears at the beginning
638 of a debugging information entry in the
640 section. As described above, the abbreviation
641 code 0 is reserved for null debugging information entries. The
642 abbreviation code is followed by another unsigned LEB128
643 number that encodes the entry\textquoteright s tag. The encodings for the
644 tag names are given in
645 Table \refersec{tab:tagencodings}.
648 \setlength{\extrarowheight}{0.1cm}
649 \begin{longtable}{l|l}
651 \caption{Tag encodings} \label{tab:tagencodings} \\
652 \hline \bfseries Tag name&\bfseries Value\\ \hline
654 \bfseries Tag name&\bfseries Value \\ \hline
656 \hline \emph{Continued on next page}
658 \hline \ddag\ \textit{New in DWARF Version 5}
660 \DWTAGarraytype{} &0x01 \\
661 \DWTAGclasstype&0x02 \\
662 \DWTAGentrypoint&0x03 \\
663 \DWTAGenumerationtype&0x04 \\
664 \DWTAGformalparameter&0x05 \\
665 \DWTAGimporteddeclaration&0x08 \\
667 \DWTAGlexicalblock&0x0b \\
669 \DWTAGpointertype&0x0f \\
670 \DWTAGreferencetype&0x10 \\
671 \DWTAGcompileunit&0x11 \\
672 \DWTAGstringtype&0x12 \\
673 \DWTAGstructuretype&0x13 \\
674 \DWTAGsubroutinetype&0x15 \\
675 \DWTAGtypedef&0x16 \\
676 \DWTAGuniontype&0x17 \\
677 \DWTAGunspecifiedparameters&0x18 \\
678 \DWTAGvariant&0x19 \\
679 \DWTAGcommonblock&0x1a \\
680 \DWTAGcommoninclusion&0x1b \\
681 \DWTAGinheritance&0x1c \\
682 \DWTAGinlinedsubroutine&0x1d \\
684 \DWTAGptrtomembertype&0x1f \\
685 \DWTAGsettype&0x20 \\
686 \DWTAGsubrangetype&0x21 \\
687 \DWTAGwithstmt&0x22 \\
688 \DWTAGaccessdeclaration&0x23 \\
689 \DWTAGbasetype&0x24 \\
690 \DWTAGcatchblock&0x25 \\
691 \DWTAGconsttype&0x26 \\
692 \DWTAGconstant&0x27 \\
693 \DWTAGenumerator&0x28 \\
694 \DWTAGfiletype&0x29 \\
696 \DWTAGnamelist&0x2b \\
697 \DWTAGnamelistitem&0x2c \\
698 \DWTAGpackedtype&0x2d \\
699 \DWTAGsubprogram&0x2e \\
700 \DWTAGtemplatetypeparameter&0x2f \\
701 \DWTAGtemplatevalueparameter&0x30 \\
702 \DWTAGthrowntype&0x31 \\
703 \DWTAGtryblock&0x32 \\
704 \DWTAGvariantpart&0x33 \\
705 \DWTAGvariable&0x34 \\
706 \DWTAGvolatiletype&0x35 \\
707 \DWTAGdwarfprocedure&0x36 \\
708 \DWTAGrestricttype&0x37 \\
709 \DWTAGinterfacetype&0x38 \\
710 \DWTAGnamespace&0x39 \\
711 \DWTAGimportedmodule&0x3a \\
712 \DWTAGunspecifiedtype&0x3b \\
713 \DWTAGpartialunit&0x3c \\
714 \DWTAGimportedunit&0x3d \\
715 \DWTAGcondition&\xiiif \\
716 \DWTAGsharedtype&0x40 \\
717 \DWTAGtypeunit{} &0x41 \\
718 \DWTAGrvaluereferencetype{} &0x42 \\
719 \DWTAGtemplatealias{} &0x43 \\
720 \DWTAGcoarraytype~\ddag &0x44 \\
721 \DWTAGgenericsubrange~\ddag &0x45 \\
722 \DWTAGdynamictype~\ddag & 0x46 \\
723 \DWTAGatomictype~\ddag & 0x47 \\
724 \DWTAGlouser&0x4080 \\
725 \DWTAGhiuser&\xffff \\
729 Following the tag encoding is a 1\dash byte value that determines
730 whether a debugging information entry using this abbreviation
731 has child entries or not. If the value is
733 the next physically succeeding entry of any debugging
734 information entry using this abbreviation is the first
735 child of that entry. If the 1\dash byte value following the
736 abbreviation\textquoteright s tag encoding is
737 \DWCHILDRENnoTARG, the next
738 physically succeeding entry of any debugging information entry
739 using this abbreviation is a sibling of that entry. (Either
740 the first child or sibling entries may be null entries). The
741 encodings for the child determination byte are given in
742 Table \refersec{tab:childdeterminationencodings}
744 Section \refersec{chap:relationshipofdebugginginformationentries},
745 each chain of sibling entries is terminated by a null entry.)
749 \setlength{\extrarowheight}{0.1cm}
750 \begin{longtable}{l|l}
751 \caption{Child determination encodings}
752 \label{tab:childdeterminationencodings}
753 \addtoindexx{Child determination encodings} \\
754 \hline \bfseries Children determination name&\bfseries Value \\ \hline
756 \bfseries Children determination name&\bfseries Value \\ \hline
758 \hline \emph{Continued on next page}
762 \DWCHILDRENno&0x00 \\
763 \DWCHILDRENyes&0x01 \\ \hline
768 Finally, the child encoding is followed by a series of
769 attribute specifications. Each attribute specification
770 consists of two parts. The first part is an unsigned LEB128
771 number representing the attribute\textquoteright s name. The second part
772 is an unsigned LEB128 number representing the attribute\textquoteright s
773 form. The series of attribute specifications ends with an
774 entry containing 0 for the name and 0 for the form.
777 \DWFORMindirectTARG{} is a special case. For
778 attributes with this form, the attribute value itself in the
780 section begins with an unsigned
781 LEB128 number that represents its form. This allows producers
782 to choose forms for particular attributes
783 \addtoindexx{abbreviations table!dynamic forms in}
785 without having to add a new entry to the abbreviations table.
787 The abbreviations for a given compilation unit end with an
788 entry consisting of a 0 byte for the abbreviation code.
791 Appendix \refersec{app:compilationunitsandabbreviationstableexample}
792 for a depiction of the organization of the
793 debugging information.}
796 \subsection{Attribute Encodings}
797 \label{datarep:attributeencodings}
799 The encodings for the attribute names are given in
800 Table \refersec{tab:attributeencodings}.
802 The attribute form governs how the value of the attribute is
803 encoded. There are nine classes of form, listed below. Each
804 class is a set of forms which have related representations
805 and which are given a common interpretation according to the
806 attribute in which the form is used.
808 Form \DWFORMsecoffsetTARG{}
810 \addtoindexx{rangelistptr class}
812 \addtoindexx{macptr class}
814 \addtoindexx{loclistptr class}
816 \addtoindexx{lineptr class}
817 namely \livelink{chap:classlineptr}{lineptr},
818 \livelink{chap:classloclistptr}{loclistptr},
819 \livelink{chap:classmacptr}{macptr} or
820 \livelink{chap:classrangelistptr}{rangelistptr}; the list
821 of classes allowed by the applicable attribute in
822 Table \refersec{tab:attributeencodings}
823 determines the class of the form.
825 \textit{In DWARF V3 the forms \DWFORMdatafour{} and
826 \DWFORMdataeight{} were
827 \addtoindexx{lineptr class}
829 \addtoindexx{rangelistptr class}
831 \addtoindexx{macptr class}
833 \addtoindexx{loclistptr class}
834 class constant \addtoindexx{constant class}
835 or one of the classes
836 \livelink{chap:classlineptr}{lineptr},
837 \livelink{chap:classloclistptr}{loclistptr},
838 \livelink{chap:classmacptr}{macptr} or
839 \livelink{chap:classrangelistptr}{rangelistptr}, depending on context. In
841 \DWFORMdatafour{} and
842 \DWFORMdataeight{} are members of class
843 constant in all cases.
845 \DWFORMsecoffset{} replaces
846 their usage for the other classes.}
849 Each possible form belongs to one or more of the following classes:
852 \item \livelinki{chap:classaddress}{address}{address class} \\
853 \livetarg{datarep:classaddress}{}
854 Represented as either:
856 \item An object of appropriate size to hold an
857 address on the target machine
859 The size is encoded in the compilation unit header
860 (see Section \refersec{datarep:compilationunitheader}).
861 This address is relocatable in a relocatable object file and
862 is relocated in an executable file or shared object.
864 \item An indirect index into a table of addresses (as
865 described in the previous bullet) in the
866 \dotdebugaddr{} section (\DWFORMaddrxTARG).
867 The representation of a \DWFORMaddrxNAME{} value is an unsigned
868 \addtoindex{LEB128} value, which is interpreted as a zero-based
869 index into an array of addresses in the \dotdebugaddr{} section.
870 The index is relative to the value of the \DWATaddrbase{} attribute
871 of the associated compilation unit.
875 \item \livelink{chap:classaddrptr}{addrptr} \\
876 \livetarg{datarep:classaddrptr}{}
877 This is an offset into the \dotdebugaddr{} section (\DWFORMsecoffset). It
878 consists of an offset from the beginning of the \dotdebugaddr{} section to the
879 beginning of the list of machine addresses information for the
880 referencing entity. It is relocatable in
881 a relocatable object file, and relocated in an executable or
882 shared object. In the \thirtytwobitdwarfformat, this offset
883 is a 4\dash byte unsigned value; in the 64\dash bit DWARF
884 format, it is an 8\dash byte unsigned value (see Section
885 \refersec{datarep:32bitand64bitdwarfformats}).
887 \textit{This class is new in \DWARFVersionV.}
890 \item \livelink{chap:classblock}{block} \\
891 \livetarg{datarep:classblock}{}
892 Blocks come in four forms:
894 \begin{myindentpara}{1cm}
895 A 1\dash byte length followed by 0 to 255 contiguous information
896 bytes (\DWFORMblockoneTARG).
899 \begin{myindentpara}{1cm}
900 A 2\dash byte length followed by 0 to 65,535 contiguous information
901 bytes (\DWFORMblocktwoTARG).
904 \begin{myindentpara}{1cm}
905 A 4\dash byte length followed by 0 to 4,294,967,295 contiguous
906 information bytes (\DWFORMblockfourTARG).
909 \begin{myindentpara}{1cm}
910 An unsigned LEB128 length followed by the number of bytes
911 specified by the length (\DWFORMblockTARG).
914 In all forms, the length is the number of information bytes
915 that follow. The information bytes may contain any mixture
916 of relocated (or relocatable) addresses, references to other
917 debugging information entries or data bytes.
919 \item \livelinki{chap:classconstant}{constant}{constant class} \\
920 \livetarg{datarep:classconstant}{}
921 There are six forms of constants. There are fixed length
922 constant data forms for one, two, four and eight byte values
927 and \DWFORMdataeightTARG).
928 There are also variable length constant
929 data forms encoded using LEB128 numbers (see below). Both
930 signed (\DWFORMsdataTARG) and unsigned
931 (\DWFORMudataTARG) variable
932 length constants are available
935 The data in \DWFORMdataone,
937 \DWFORMdatafour{} and
939 can be anything. Depending on context, it may
940 be a signed integer, an unsigned integer, a floating\dash point
941 constant, or anything else. A consumer must use context to
942 know how to interpret the bits, which if they are target
943 machine data (such as an integer or floating point constant)
944 will be in target machine byte\dash order.
946 \textit{If one of the \DWFORMdataTARG\textless n\textgreater
947 forms is used to represent a
948 signed or unsigned integer, it can be hard for a consumer
949 to discover the context necessary to determine which
950 interpretation is intended. Producers are therefore strongly
951 encouraged to use \DWFORMsdata{} or
952 \DWFORMudata{} for signed and
953 unsigned integers respectively, rather than
954 \DWFORMdata\textless n\textgreater.}
957 \item \livelinki{chap:classexprloc}{exprloc}{exprloc class} \\
958 \livetarg{datarep:classexprloc}{}
959 This is an unsigned LEB128 length followed by the
960 number of information bytes specified by the length
961 (\DWFORMexprlocTARG).
962 The information bytes contain a DWARF expression
963 (see Section \refersec{chap:dwarfexpressions})
964 or location description
965 (see Section \refersec{chap:locationdescriptions}).
967 \item \livelinki{chap:classflag}{flag}{flag class} \\
968 \livetarg{datarep:classflag}{}
969 A flag \addtoindexx{flag class}
970 is represented explicitly as a single byte of data
972 implicitly (\DWFORMflagpresentTARG).
974 first case, if the \nolink{flag} has value zero, it indicates the
975 absence of the attribute; if the \nolink{flag} has a non\dash zero value,
976 it indicates the presence of the attribute. In the second
977 case, the attribute is implicitly indicated as present, and
978 no value is encoded in the debugging information entry itself.
980 \item \livelinki{chap:classlineptr}{lineptr}{lineptr class} \\
981 \livetarg{datarep:classlineptr}{}
982 This is an offset into
983 \addtoindexx{section offset!in class lineptr value}
985 \dotdebugline{} or \dotdebuglinedwo{} section
987 It consists of an offset from the beginning of the
989 section to the first byte of
990 the data making up the line number list for the compilation
992 It is relocatable in a relocatable object file, and
993 relocated in an executable or shared object. In the
994 \thirtytwobitdwarfformat, this offset is a 4\dash byte unsigned value;
995 in the \sixtyfourbitdwarfformat, it is an 8\dash byte unsigned value
996 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
999 \item \livelinki{chap:classloclistptr}{loclistptr}{loclistptr class} \\
1000 \livetarg{datarep:classloclistptr}{}
1001 This is an offset into the
1005 It consists of an offset from the
1006 \addtoindexx{section offset!in class loclistptr value}
1009 section to the first byte of
1010 the data making up the
1011 \addtoindex{location list} for the compilation unit.
1012 It is relocatable in a relocatable object file, and
1013 relocated in an executable or shared object. In the
1014 \thirtytwobitdwarfformat, this offset is a 4\dash byte unsigned value;
1015 in the \sixtyfourbitdwarfformat, it is an 8\dash byte unsigned value
1016 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
1019 \item \livelinki{chap:classmacptr}{macptr}{macptr class} \\
1020 \livetarg{datarep:classmacptr}{}
1022 \addtoindexx{section offset!in class macptr value}
1024 \dotdebugmacro{} or \dotdebugmacrodwo{} section
1026 It consists of an offset from the beginning of the
1027 \dotdebugmacro{} or \dotdebugmacrodwo{}
1028 section to the the header making up the
1029 macro information list for the compilation unit.
1030 It is relocatable in a relocatable object file, and
1031 relocated in an executable or shared object. In the
1032 \thirtytwobitdwarfformat, this offset is a 4\dash byte unsigned value;
1033 in the \sixtyfourbitdwarfformat, it is an 8\dash byte unsigned value
1034 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
1037 \item \livelinki{chap:classrangelistptr}{rangelistptr}{rangelistptr class} \\
1038 \livetarg{datarep:classrangelistptr}{}
1040 \addtoindexx{section offset!in class rangelistptr value}
1041 offset into the \dotdebugranges{} section
1044 offset from the beginning of the
1045 \dotdebugranges{} section
1046 to the beginning of the non\dash contiguous address ranges
1047 information for the referencing entity.
1048 It is relocatable in
1049 a relocatable object file, and relocated in an executable or
1050 shared object. In the \thirtytwobitdwarfformat, this offset
1051 is a 4\dash byte unsigned value; in the 64\dash bit DWARF
1052 format, it is an 8\dash byte unsigned value (see Section
1053 \refersec{datarep:32bitand64bitdwarfformats}).
1056 \textit{Because classes \livelink{chap:classlineptr}{lineptr},
1057 \livelink{chap:classloclistptr}{loclistptr},
1058 \livelink{chap:classmacptr}{macptr} and
1059 \livelink{chap:classrangelistptr}{rangelistptr}
1060 share a common representation, it is not possible for an
1061 attribute to allow more than one of these classes}
1065 \item \livelinki{chap:classreference}{reference}{reference class} \\
1066 \livetarg{datarep:classreference}{}
1067 There are three types of reference.
1070 \addtoindexx{reference class}
1071 first type of reference can identify any debugging
1072 information entry within the containing unit.
1075 \addtoindexx{section offset!in class reference value}
1076 offset from the first byte of the compilation
1077 header for the compilation unit containing the reference. There
1078 are five forms for this type of reference. There are fixed
1079 length forms for one, two, four and eight byte offsets
1085 and \DWFORMrefeightTARG).
1086 There is also an unsigned variable
1087 length offset encoded form that uses unsigned LEB128 numbers
1088 (\DWFORMrefudataTARG).
1089 Because this type of reference is within
1090 the containing compilation unit no relocation of the value
1093 The second type of reference can identify any debugging
1094 information entry within a
1095 \dotdebuginfo{} section; in particular,
1096 it may refer to an entry in a different compilation unit
1097 from the unit containing the reference, and may refer to an
1098 entry in a different shared object. This type of reference
1099 (\DWFORMrefaddrTARG)
1100 is an offset from the beginning of the
1102 section of the target executable or shared object;
1103 it is relocatable in a relocatable object file and frequently
1104 relocated in an executable file or shared object. For
1105 references from one shared object or static executable file
1106 to another, the relocation and identification of the target
1107 object must be performed by the consumer. In the
1108 \thirtytwobitdwarfformat, this offset is a 4\dash byte unsigned value;
1109 in the \sixtyfourbitdwarfformat, it is an 8\dash byte
1111 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
1113 \textit{A debugging information entry that may be referenced by
1114 another compilation unit using
1115 \DWFORMrefaddr{} must have a
1116 global symbolic name.}
1118 \textit{For a reference from one executable or shared object to
1119 another, the reference is resolved by the debugger to identify
1120 the shared object or executable and the offset into that
1121 object\textquoteright s \dotdebuginfo{}
1122 section in the same fashion as the run
1123 time loader, either when the debug information is first read,
1124 or when the reference is used.}
1126 The third type of reference can identify any debugging
1127 information type entry that has been placed in its own
1128 \addtoindex{type unit}. This type of
1129 reference (\DWFORMrefsigeightTARG) is the
1130 \addtoindexx{type signature}
1131 64\dash bit type signature
1132 (see Section \refersec{datarep:typesignaturecomputation})
1136 \textit{The use of compilation unit relative references will reduce the
1137 number of link\dash time relocations and so speed up linking. The
1138 use of the second and third type of reference allows for the
1139 sharing of information, such as types, across compilation
1142 \textit{A reference to any kind of compilation unit identifies the
1143 debugging information entry for that unit, not the preceding
1146 \item \livelinki{chap:classstring}{string}{string class} \\
1147 \livetarg{datarep:classstring}{}
1148 A string is a sequence of contiguous non\dash null bytes followed by
1150 \addtoindexx{string class}
1151 A string may be represented:
1153 \item immediately in the debugging information entry itself
1154 (\DWFORMstringTARG),
1156 \addtoindexx{section offset!in class string value}
1157 offset into a string table contained in
1158 the \dotdebugstr{} section of the object file
1160 In the \thirtytwobitdwarfformat, the representation of a
1162 value is a 4\dash byte unsigned offset; in the \sixtyfourbitdwarfformat,
1163 it is an 8\dash byte unsigned offset
1164 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
1165 \item as an indirect offset into the string table using an
1166 index into a table of offsets contained in the
1167 \dotdebugstroffsets{} section of the object file (\DWFORMstrxTARG).
1168 The representation of a \DWFORMstrxNAME{} value is an unsigned
1169 \addtoindex{LEB128} value, which is interpreted as a zero-based
1170 index into an array of offsets in the \dotdebugstroffsets{} section.
1171 The offset entries in the \dotdebugstroffsets{} section have the
1172 same representation as \DWFORMstrp{} values.
1174 Any combination of these three forms may be used within a single compilation.
1176 If the \DWATuseUTFeight{}
1177 \addtoindexx{use UTF8 attribute}\addtoindexx{UTF-8} attribute is specified for the
1178 compilation, partial, skeleton or type unit entry, string values are encoded using the
1179 UTF\dash 8 (\addtoindex{Unicode} Transformation Format\dash 8) from the Universal
1180 Character Set standard (ISO/IEC 10646\dash 1:1993). Otherwise,
1181 the string representation is unspecified.
1183 \textit{The \addtoindex{Unicode} Standard Version 3 is fully compatible with
1184 ISO/IEC 10646\dash 1:1993. It contains all the same characters
1185 and encoding points as ISO/IEC 10646, as well as additional
1186 information about the characters and their use.}
1188 \textit{Earlier versions of DWARF did not specify the representation
1189 of strings; for compatibility, this version also does
1190 not. However, the UTF\dash 8 representation is strongly recommended.}
1192 \item \livelinki{chap:classstroffsetsptr}{stroffsetsptr}{stroffsetsptr class} \\
1193 \livetarg{datarep:classstroffsetsptr}{}
1194 This is an offset into the \dotdebugstroffsets{} section
1195 (\DWFORMsecoffset). It consists of an offset from the beginning of the
1196 \dotdebugstroffsets{} section to the
1197 beginning of the string offsets information for the
1198 referencing entity. It is relocatable in
1199 a relocatable object file, and relocated in an executable or
1200 shared object. In the \thirtytwobitdwarfformat, this offset
1201 is a 4\dash byte unsigned value; in the 64\dash bit DWARF
1202 format, it is an 8\dash byte unsigned value (see Section
1203 \refersec{datarep:32bitand64bitdwarfformats}).
1205 \textit{This class is new in \DWARFVersionV.}
1209 In no case does an attribute use
1210 \addtoindexx{rangelistptr class}
1212 \addtoindexx{loclistptr class}
1214 \addtoindexx{lineptr class}
1216 \addtoindexx{macptr class}
1217 classes \livelink{chap:classlineptr}{lineptr},
1218 \livelink{chap:classloclistptr}{loclistptr}, \livelink{chap:classmacptr}{macptr} or
1219 \livelink{chap:classrangelistptr}{rangelistptr} to point into either the
1220 \dotdebuginfo{} or \dotdebugstr{} section.
1222 The form encodings are listed in
1223 Table \refersec{tab:attributeformencodings}.
1227 \setlength{\extrarowheight}{0.1cm}
1228 \begin{longtable}{l|l|l}
1229 \caption{Attribute encodings}
1230 \label{tab:attributeencodings}
1231 \addtoindexx{attribute encodings} \\
1232 \hline \bfseries Attribute name&\bfseries Value &\bfseries Classes \\ \hline
1234 \bfseries Attribute name&\bfseries Value &\bfseries Classes\\ \hline
1236 \hline \emph{Continued on next page}
1238 \hline \ddag\ \textit{New in DWARF Version 5}
1240 \DWATsibling&0x01&\livelink{chap:classreference}{reference}
1241 \addtoindexx{sibling attribute!encoding} \\
1242 \DWATlocation&0x02&\livelink{chap:classexprloc}{exprloc},
1243 \livelink{chap:classloclistptr}{loclistptr}
1244 \addtoindexx{location attribute!encoding} \\
1245 \DWATname&0x03&\livelink{chap:classstring}{string}
1246 \addtoindexx{name attribute!encoding} \\
1247 \DWATordering&0x09&\livelink{chap:classconstant}{constant}
1248 \addtoindexx{ordering attribute!encoding} \\
1249 \DWATbytesize&0x0b&\livelink{chap:classconstant}{constant},
1250 \livelink{chap:classexprloc}{exprloc},
1251 \livelink{chap:classreference}{reference}
1252 \addtoindexx{byte size attribute!encoding} \\
1253 \DWATbitoffset&0x0c&\livelink{chap:classconstant}{constant},
1254 \livelink{chap:classexprloc}{exprloc},
1255 \livelink{chap:classreference}{reference}
1256 \addtoindexx{bit offset attribute!encoding} \\
1257 \DWATbitsize&0x0d&\livelink{chap:classconstant}{constant},
1258 \livelink{chap:classexprloc}{exprloc},
1259 \livelink{chap:classreference}{reference}
1260 \addtoindexx{bit size attribute!encoding} \\
1261 \DWATstmtlist&0x10&\livelink{chap:classlineptr}{lineptr}
1262 \addtoindexx{statement list attribute!encoding} \\
1263 \DWATlowpc&0x11&\livelink{chap:classaddress}{address}
1264 \addtoindexx{low PC attribute!encoding} \\
1265 \DWAThighpc&0x12&\livelink{chap:classaddress}{address},
1266 \livelink{chap:classconstant}{constant}
1267 \addtoindexx{high PC attribute!encoding} \\
1268 \DWATlanguage&0x13&\livelink{chap:classconstant}{constant}
1269 \addtoindexx{language attribute!encoding} \\
1270 \DWATdiscr&0x15&\livelink{chap:classreference}{reference}
1271 \addtoindexx{discriminant attribute!encoding} \\
1272 \DWATdiscrvalue&0x16&\livelink{chap:classconstant}{constant}
1273 \addtoindexx{discriminant value attribute!encoding} \\
1274 \DWATvisibility&0x17&\livelink{chap:classconstant}{constant}
1275 \addtoindexx{visibility attribute!encoding} \\
1276 \DWATimport&0x18&\livelink{chap:classreference}{reference}
1277 \addtoindexx{import attribute!encoding} \\
1278 \DWATstringlength&0x19&\livelink{chap:classexprloc}{exprloc},
1279 \livelink{chap:classloclistptr}{loclistptr}
1280 \addtoindexx{string length attribute!encoding} \\
1281 \DWATcommonreference&0x1a&\livelink{chap:classreference}{reference}
1282 \addtoindexx{common reference attribute!encoding} \\
1283 \DWATcompdir&0x1b&\livelink{chap:classstring}{string}
1284 \addtoindexx{compilation directory attribute!encoding} \\
1285 \DWATconstvalue&0x1c&\livelink{chap:classblock}{block},
1286 \livelink{chap:classconstant}{constant},
1287 \livelink{chap:classstring}{string}
1288 \addtoindexx{constant value attribute!encoding} \\
1289 \DWATcontainingtype&0x1d&\livelink{chap:classreference}{reference}
1290 \addtoindexx{containing type attribute!encoding} \\
1291 \DWATdefaultvalue&0x1e&\livelink{chap:classconstant}{constant},
1292 \livelink{chap:classreference}{reference},
1293 \livelink{chap:classflag}{flag}
1294 \addtoindexx{default value attribute!encoding} \\
1295 \DWATinline&0x20&\livelink{chap:classconstant}{constant}
1296 \addtoindexx{inline attribute!encoding} \\
1297 \DWATisoptional&0x21&\livelink{chap:classflag}{flag}
1298 \addtoindexx{is optional attribute!encoding} \\
1299 \DWATlowerbound&0x22&\livelink{chap:classconstant}{constant},
1300 \livelink{chap:classexprloc}{exprloc},
1301 \livelink{chap:classreference}{reference}
1302 \addtoindexx{lower bound attribute!encoding} \\
1303 \DWATproducer&0x25&\livelink{chap:classstring}{string}
1304 \addtoindexx{producer attribute!encoding} \\
1305 \DWATprototyped&0x27&\livelink{chap:classflag}{flag}
1306 \addtoindexx{prototyped attribute!encoding} \\
1307 \DWATreturnaddr&0x2a&\livelink{chap:classexprloc}{exprloc},
1308 \livelink{chap:classloclistptr}{loclistptr}
1309 \addtoindexx{return address attribute!encoding} \\
1310 \DWATstartscope&0x2c&\livelink{chap:classconstant}{constant},
1311 \livelink{chap:classrangelistptr}{rangelistptr}
1312 \addtoindexx{start scope attribute!encoding} \\
1313 \DWATbitstride&0x2e&\livelink{chap:classconstant}{constant},
1314 \livelink{chap:classexprloc}{exprloc},
1315 \livelink{chap:classreference}{reference}
1316 \addtoindexx{bit stride attribute!encoding} \\
1317 \DWATupperbound&0x2f&\livelink{chap:classconstant}{constant},
1318 \livelink{chap:classexprloc}{exprloc},
1319 \livelink{chap:classreference}{reference}
1320 \addtoindexx{upper bound attribute!encoding} \\
1321 \DWATabstractorigin&0x31&\livelink{chap:classreference}{reference}
1322 \addtoindexx{abstract origin attribute!encoding} \\
1323 \DWATaccessibility&0x32&\livelink{chap:classconstant}{constant}
1324 \addtoindexx{accessibility attribute!encoding} \\
1325 \DWATaddressclass&0x33&\livelink{chap:classconstant}{constant}
1326 \addtoindexx{address class attribute!encoding} \\
1327 \DWATartificial&0x34&\livelink{chap:classflag}{flag}
1328 \addtoindexx{artificial attribute!encoding} \\
1329 \DWATbasetypes&0x35&\livelink{chap:classreference}{reference}
1330 \addtoindexx{base types attribute!encoding} \\
1331 \DWATcallingconvention&0x36&\livelink{chap:classconstant}{constant}
1332 \addtoindexx{calling convention attribute!encoding} \\
1333 \DWATcount&0x37&\livelink{chap:classconstant}{constant},
1334 \livelink{chap:classexprloc}{exprloc},
1335 \livelink{chap:classreference}{reference}
1336 \addtoindexx{count attribute!encoding} \\
1337 \DWATdatamemberlocation&0x38&\livelink{chap:classconstant}{constant},
1338 \livelink{chap:classexprloc}{exprloc},
1339 \livelink{chap:classloclistptr}{loclistptr}
1340 \addtoindexx{data member attribute!encoding} \\
1341 \DWATdeclcolumn&0x39&\livelink{chap:classconstant}{constant}
1342 \addtoindexx{declaration column attribute!encoding} \\
1343 \DWATdeclfile&0x3a&\livelink{chap:classconstant}{constant}
1344 \addtoindexx{declaration file attribute!encoding} \\
1345 \DWATdeclline&0x3b&\livelink{chap:classconstant}{constant}
1346 \addtoindexx{declaration line attribute!encoding} \\
1347 \DWATdeclaration&0x3c&\livelink{chap:classflag}{flag}
1348 \addtoindexx{declaration attribute!encoding} \\
1349 \DWATdiscrlist&0x3d&\livelink{chap:classblock}{block}
1350 \addtoindexx{discriminant list attribute!encoding} \\
1351 \DWATencoding&0x3e&\livelink{chap:classconstant}{constant}
1352 \addtoindexx{encoding attribute!encoding} \\
1353 \DWATexternal&\xiiif&\livelink{chap:classflag}{flag}
1354 \addtoindexx{external attribute!encoding} \\
1355 \DWATframebase&0x40&\livelink{chap:classexprloc}{exprloc},
1356 \livelink{chap:classloclistptr}{loclistptr}
1357 \addtoindexx{frame base attribute!encoding} \\
1358 \DWATfriend&0x41&\livelink{chap:classreference}{reference}
1359 \addtoindexx{friend attribute!encoding} \\
1360 \DWATidentifiercase&0x42&\livelink{chap:classconstant}{constant}
1361 \addtoindexx{identifier case attribute!encoding} \\
1362 \DWATmacroinfo\footnote{\raggedright Not used in \DWARFVersionV.
1363 Reserved for compatibility and coexistence
1364 with prior DWARF versions.}
1365 &0x43&\livelink{chap:classmacptr}{macptr}
1366 \addtoindexx{macro information attribute (legacy)!encoding} \\
1367 \DWATnamelistitem&0x44&\livelink{chap:classreference}{reference}
1368 \addtoindexx{name list item attribute!encoding} \\
1369 \DWATpriority&0x45&\livelink{chap:classreference}{reference}
1370 \addtoindexx{priority attribute!encoding} \\
1371 \DWATsegment&0x46&\livelink{chap:classexprloc}{exprloc},
1372 \livelink{chap:classloclistptr}{loclistptr}
1373 \addtoindexx{segment attribute!encoding} \\
1374 \DWATspecification&0x47&\livelink{chap:classreference}{reference}
1375 \addtoindexx{specification attribute!encoding} \\
1376 \DWATstaticlink&0x48&\livelink{chap:classexprloc}{exprloc},
1377 \livelink{chap:classloclistptr}{loclistptr}
1378 \addtoindexx{static link attribute!encoding} \\
1379 \DWATtype&0x49&\livelink{chap:classreference}{reference}
1380 \addtoindexx{type attribute!encoding} \\
1381 \DWATuselocation&0x4a&\livelink{chap:classexprloc}{exprloc},
1382 \livelink{chap:classloclistptr}{loclistptr}
1383 \addtoindexx{location list attribute!encoding} \\
1384 \DWATvariableparameter&0x4b&\livelink{chap:classflag}{flag}
1385 \addtoindexx{variable parameter attribute!encoding} \\
1386 \DWATvirtuality&0x4c&\livelink{chap:classconstant}{constant}
1387 \addtoindexx{virtuality attribute!encoding} \\
1388 \DWATvtableelemlocation&0x4d&\livelink{chap:classexprloc}{exprloc},
1389 \livelink{chap:classloclistptr}{loclistptr}
1390 \addtoindexx{vtable element location attribute!encoding} \\
1391 \DWATallocated&0x4e&\livelink{chap:classconstant}{constant},
1392 \livelink{chap:classexprloc}{exprloc},
1393 \livelink{chap:classreference}{reference}
1394 \addtoindexx{allocated attribute!encoding} \\
1395 \DWATassociated&0x4f&\livelink{chap:classconstant}{constant},
1396 \livelink{chap:classexprloc}{exprloc},
1397 \livelink{chap:classreference}{reference}
1398 \addtoindexx{associated attribute!encoding} \\
1399 \DWATdatalocation&0x50&\livelink{chap:classexprloc}{exprloc}
1400 \addtoindexx{data location attribute!encoding} \\
1401 \DWATbytestride&0x51&\livelink{chap:classconstant}{constant},
1402 \livelink{chap:classexprloc}{exprloc},
1403 \livelink{chap:classreference}{reference}
1404 \addtoindexx{byte stride attribute!encoding} \\
1405 \DWATentrypc&0x52&\livelink{chap:classaddress}{address},
1406 \livelink{chap:classconstant}{constant}
1407 \addtoindexx{entry pc attribute!encoding} \\
1408 \DWATuseUTFeight&0x53&\livelink{chap:classflag}{flag}
1409 \addtoindexx{use UTF8 attribute!encoding}\addtoindexx{UTF-8} \\
1410 \DWATextension&0x54&\livelink{chap:classreference}{reference}
1411 \addtoindexx{extension attribute!encoding} \\
1412 \DWATranges&0x55&\livelink{chap:classrangelistptr}{rangelistptr}
1413 \addtoindexx{ranges attribute!encoding} \\
1414 \DWATtrampoline&0x56&\livelink{chap:classaddress}{address},
1415 \livelink{chap:classflag}{flag},
1416 \livelink{chap:classreference}{reference},
1417 \livelink{chap:classstring}{string}
1418 \addtoindexx{trampoline attribute!encoding} \\
1419 \DWATcallcolumn&0x57&\livelink{chap:classconstant}{constant}
1420 \addtoindexx{call column attribute!encoding} \\
1421 \DWATcallfile&0x58&\livelink{chap:classconstant}{constant}
1422 \addtoindexx{call file attribute!encoding} \\
1423 \DWATcallline&0x59&\livelink{chap:classconstant}{constant}
1424 \addtoindexx{call line attribute!encoding} \\
1425 \DWATdescription&0x5a&\livelink{chap:classstring}{string}
1426 \addtoindexx{description attribute!encoding} \\
1427 \DWATbinaryscale&0x5b&\livelink{chap:classconstant}{constant}
1428 \addtoindexx{binary scale attribute!encoding} \\
1429 \DWATdecimalscale&0x5c&\livelink{chap:classconstant}{constant}
1430 \addtoindexx{decimal scale attribute!encoding} \\
1431 \DWATsmall{} &0x5d&\livelink{chap:classreference}{reference}
1432 \addtoindexx{small attribute!encoding} \\
1433 \DWATdecimalsign&0x5e&\livelink{chap:classconstant}{constant}
1434 \addtoindexx{decimal scale attribute!encoding} \\
1435 \DWATdigitcount&0x5f&\livelink{chap:classconstant}{constant}
1436 \addtoindexx{digit count attribute!encoding} \\
1437 \DWATpicturestring&0x60&\livelink{chap:classstring}{string}
1438 \addtoindexx{picture string attribute!encoding} \\
1439 \DWATmutable&0x61&\livelink{chap:classflag}{flag}
1440 \addtoindexx{mutable attribute!encoding} \\
1441 \DWATthreadsscaled&0x62&\livelink{chap:classflag}{flag}
1442 \addtoindexx{thread scaled attribute!encoding} \\
1443 \DWATexplicit&0x63&\livelink{chap:classflag}{flag}
1444 \addtoindexx{explicit attribute!encoding} \\
1445 \DWATobjectpointer&0x64&\livelink{chap:classreference}{reference}
1446 \addtoindexx{object pointer attribute!encoding} \\
1447 \DWATendianity&0x65&\livelink{chap:classconstant}{constant}
1448 \addtoindexx{endianity attribute!encoding} \\
1449 \DWATelemental&0x66&\livelink{chap:classflag}{flag}
1450 \addtoindexx{elemental attribute!encoding} \\
1451 \DWATpure&0x67&\livelink{chap:classflag}{flag}
1452 \addtoindexx{pure attribute!encoding} \\
1453 \DWATrecursive&0x68&\livelink{chap:classflag}{flag}
1454 \addtoindexx{recursive attribute!encoding} \\
1455 \DWATsignature{} &0x69&\livelink{chap:classreference}{reference}
1456 \addtoindexx{signature attribute!encoding} \\
1457 \DWATmainsubprogram{} &0x6a&\livelink{chap:classflag}{flag}
1458 \addtoindexx{main subprogram attribute!encoding} \\
1459 \DWATdatabitoffset{} &0x6b&\livelink{chap:classconstant}{constant}
1460 \addtoindexx{data bit offset attribute!encoding} \\
1461 \DWATconstexpr{} &0x6c&\livelink{chap:classflag}{flag}
1462 \addtoindexx{constant expression attribute!encoding} \\
1463 \DWATenumclass{} &0x6d&\livelink{chap:classflag}{flag}
1464 \addtoindexx{enumeration class attribute!encoding} \\
1465 \DWATlinkagename{} &0x6e&\livelink{chap:classstring}{string}
1466 \addtoindexx{linkage name attribute!encoding} \\
1467 \DWATstringlengthbitsize{} \ddag&0x6f&
1468 \livelink{chap:classconstant}{constant}
1469 \addtoindexx{string length attribute!size of length} \\
1470 \DWATstringlengthbytesize{} \ddag&0x70&
1471 \livelink{chap:classconstant}{constant}
1472 \addtoindexx{string length attribute!size of length} \\
1473 \DWATrank~\ddag&0x71&
1474 \livelink{chap:classconstant}{constant},
1475 \livelink{chap:classexprloc}{exprloc}
1476 \addtoindexx{rank attribute!encoding} \\
1477 \DWATstroffsetsbase~\ddag&0x72&
1478 \livelinki{chap:classstring}{stroffsetsptr}{stroffsetsptr class}
1479 \addtoindexx{string offsets base!encoding} \\
1480 \DWATaddrbase~\ddag &0x73&
1481 \livelinki{chap:DWATaddrbase}{addrptr}{addrptr class}
1482 \addtoindexx{address table base!encoding} \\
1483 \DWATrangesbase~\ddag&0x74&
1484 \livelinki{chap:DWATrangesbase}{rangelistptr}{rangelistptr class}
1485 \addtoindexx{ranges base!encoding} \\
1486 \DWATdwoid~\ddag &0x75&
1487 \livelink{chap:DWATdwoid}{constant}
1488 \addtoindexx{split DWARF object id!encoding} \\
1489 \DWATdwoname~\ddag &0x76&
1490 \livelink{chap:DWATdwoname}{string}
1491 \addtoindexx{split DWARF object file name!encoding} \\
1492 \DWATreference~\ddag &0x77&
1493 \livelink{chap:DWATreference}{flag} \\
1494 \DWATrvaluereference~\ddag &0x78&
1495 \livelink{chap:DWATrvaluereference}{flag} \\
1496 \DWATmacros~\ddag &0x79&\livelink{chap:classmacptr}{macptr}
1497 \addtoindexx{macro information attribute!encoding} \\
1499 \DWATlouser&0x2000 & --- \addtoindexx{low user attribute encoding} \\
1500 \DWAThiuser&\xiiifff& --- \addtoindexx{high user attribute encoding} \\
1507 \setlength{\extrarowheight}{0.1cm}
1508 \begin{longtable}{l|l|l}
1509 \caption{Attribute form encodings} \label{tab:attributeformencodings} \\
1510 \hline \bfseries Form name&\bfseries Value &\bfseries Classes \\ \hline
1512 \bfseries Form name&\bfseries Value &\bfseries Classes\\ \hline
1514 \hline \emph{Continued on next page}
1516 \hline \ddag\ \textit{New in DWARF Version 5}
1519 \DWFORMaddr &0x01&\livelink{chap:classaddress}{address} \\
1520 \textit{Reserved} &0x02& \\
1521 \DWFORMblocktwo &0x03&\livelink{chap:classblock}{block} \\
1522 \DWFORMblockfour &0x04&\livelink{chap:classblock}{block} \\
1523 \DWFORMdatatwo &0x05&\livelink{chap:classconstant}{constant} \\
1524 \DWFORMdatafour &0x06&\livelink{chap:classconstant}{constant} \\
1525 \DWFORMdataeight &0x07&\livelink{chap:classconstant}{constant} \\
1526 \DWFORMstring&0x08&\livelink{chap:classstring}{string} \\
1527 \DWFORMblock&0x09&\livelink{chap:classblock}{block} \\
1528 \DWFORMblockone &0x0a&\livelink{chap:classblock}{block} \\
1529 \DWFORMdataone &0x0b&\livelink{chap:classconstant}{constant} \\
1530 \DWFORMflag&0x0c&\livelink{chap:classflag}{flag} \\
1531 \DWFORMsdata&0x0d&\livelink{chap:classconstant}{constant} \\
1532 \DWFORMstrp&0x0e&\livelink{chap:classstring}{string} \\
1533 \DWFORMudata&0x0f&\livelink{chap:classconstant}{constant} \\
1534 \DWFORMrefaddr&0x10&\livelink{chap:classreference}{reference} \\
1535 \DWFORMrefone&0x11&\livelink{chap:classreference}{reference} \\
1536 \DWFORMreftwo&0x12&\livelink{chap:classreference}{reference} \\
1537 \DWFORMreffour&0x13&\livelink{chap:classreference}{reference} \\
1538 \DWFORMrefeight&0x14&\livelink{chap:classreference}{reference} \\
1539 \DWFORMrefudata&0x15&\livelink{chap:classreference}{reference} \\
1540 \DWFORMindirect&0x16&(see Section \refersec{datarep:abbreviationstables}) \\
1541 \DWFORMsecoffset{} &0x17&\livelink{chap:classlineptr}{lineptr}, \livelink{chap:classloclistptr}{loclistptr}, \livelink{chap:classmacptr}{macptr}, \livelink{chap:classrangelistptr}{rangelistptr} \\
1542 \DWFORMexprloc{} &0x18&\livelink{chap:classexprloc}{exprloc} \\
1543 \DWFORMflagpresent{} &0x19&\livelink{chap:classflag}{flag} \\
1544 \DWFORMstrx{} \ddag &0x1a&\livelink{chap:classstring}{string} \\
1545 \DWFORMaddrx{} \ddag &0x1b&\livelink{chap:classaddress}{address} \\
1546 \DWFORMrefsigeight &0x20&\livelink{chap:classreference}{reference} \\
1553 \section{Variable Length Data}
1554 \label{datarep:variablelengthdata}
1555 \addtoindexx{variable length data|see {LEB128}}
1557 \addtoindexx{Little Endian Base 128|see{LEB128}}
1558 encoded using \doublequote{Little Endian Base 128}
1559 \addtoindexx{little-endian encoding|see{endian attribute}}
1561 \addtoindexx{LEB128}
1562 LEB128 is a scheme for encoding integers
1563 densely that exploits the assumption that most integers are
1566 \textit{This encoding is equally suitable whether the target machine
1567 architecture represents data in big\dash\ endian or little\dash endian
1568 order. It is \doublequote{little\dash endian} only in the sense that it
1569 avoids using space to represent the \doublequote{big} end of an
1570 unsigned integer, when the big end is all zeroes or sign
1573 Unsigned LEB128 (ULEB128) numbers are encoded as follows:
1574 \addtoindexx{LEB128!unsigned, encoding as}
1575 start at the low order end of an unsigned integer and chop
1576 it into 7\dash bit chunks. Place each chunk into the low order 7
1577 bits of a byte. Typically, several of the high order bytes
1578 will be zero; discard them. Emit the remaining bytes in a
1579 stream, starting with the low order byte; set the high order
1580 bit on each byte except the last emitted byte. The high bit
1581 of zero on the last byte indicates to the decoder that it
1582 has encountered the last byte.
1584 The integer zero is a special case, consisting of a single
1587 Table \refersec{tab:examplesofunsignedleb128encodings}
1588 gives some examples of unsigned LEB128 numbers. The
1589 0x80 in each case is the high order bit of the byte, indicating
1590 that an additional byte follows.
1593 The encoding for signed, two\textquoteright s complement LEB128 (SLEB128)
1594 \addtoindexx{LEB128!signed, encoding as}
1595 numbers is similar, except that the criterion for discarding
1596 high order bytes is not whether they are zero, but whether
1597 they consist entirely of sign extension bits. Consider the
1598 32\dash bit integer -2. The three high level bytes of the number
1599 are sign extension, thus LEB128 would represent it as a single
1600 byte containing the low order 7 bits, with the high order
1601 bit cleared to indicate the end of the byte stream. Note
1602 that there is nothing within the LEB128 representation that
1603 indicates whether an encoded number is signed or unsigned. The
1604 decoder must know what type of number to expect.
1605 Table \refersec{tab:examplesofunsignedleb128encodings}
1606 gives some examples of unsigned LEB128 numbers and
1607 Table \refersec{tab:examplesofsignedleb128encodings}
1608 gives some examples of signed LEB128
1611 \textit{Appendix \refersec{app:variablelengthdataencodingdecodinginformative}
1612 \addtoindexx{LEB128!examples}
1613 gives algorithms for encoding and decoding these forms.}
1617 \setlength{\extrarowheight}{0.1cm}
1618 \begin{longtable}{l|l|l}
1619 \caption{Examples of unsigned LEB128 encodings}
1620 \label{tab:examplesofunsignedleb128encodings}
1621 \addtoindexx{LEB128 encoding!examples} \\
1622 \hline \bfseries Number&\bfseries First byte &\bfseries Second byte \\ \hline
1624 \bfseries Number&\bfseries First Byte &\bfseries Second byte\\ \hline
1626 \hline \emph{Continued on next page}
1632 128& 0 + 0x80 & 1 \\
1633 129& 1 + 0x80 & 1 \\
1634 130& 2 + 0x80 & 1 \\
1635 12857& 57 + 0x80 & 100 \\
1642 \setlength{\extrarowheight}{0.1cm}
1643 \begin{longtable}{l|l|l}
1644 \caption{Examples of signed LEB128 encodings} \label{tab:examplesofsignedleb128encodings} \\
1645 \hline \bfseries Number&\bfseries First byte &\bfseries Second byte \\ \hline
1647 \bfseries Number&\bfseries First Byte &\bfseries Second byte\\ \hline
1649 \hline \emph{Continued on next page}
1655 127& 127 + 0x80 & 0 \\
1656 -127& 1 + 0x80 & 0x7f \\
1657 128& 0 + 0x80 & 1 \\
1658 -128& 0 + 0x80 & 0x7f \\
1659 129& 1 + 0x80 & 1 \\
1660 -129& 0x7f + 0x80 & 0x7e \\
1667 \section{DWARF Expressions and Location Descriptions}
1668 \label{datarep:dwarfexpressionsandlocationdescriptions}
1669 \subsection{DWARF Expressions}
1670 \label{datarep:dwarfexpressions}
1673 \addtoindexx{DWARF Expression!operator encoding}
1674 DWARF expression is stored in a \nolink{block} of contiguous
1675 bytes. The bytes form a sequence of operations. Each operation
1676 is a 1\dash byte code that identifies that operation, followed by
1677 zero or more bytes of additional data. The encodings for the
1678 operations are described in
1679 Table \refersec{tab:dwarfoperationencodings}.
1682 \setlength{\extrarowheight}{0.1cm}
1683 \begin{longtable}{l|c|c|l}
1684 \caption{DWARF operation encodings} \label{tab:dwarfoperationencodings} \\
1685 \hline & &\bfseries No. of &\\
1686 \bfseries Operation&\bfseries Code &\bfseries Operands &\bfseries Notes\\ \hline
1688 & &\bfseries No. of &\\
1689 \bfseries Operation&\bfseries Code &\bfseries Operands &\bfseries Notes\\ \hline
1691 \hline \emph{Continued on next page}
1693 \hline \ddag\ \textit{New in DWARF Version 5}
1696 \DWOPaddr&0x03&1 & constant address \\
1697 & & &(size is target specific) \\
1699 \DWOPderef&0x06&0 & \\
1701 \DWOPconstoneu&0x08&1&1\dash byte constant \\
1702 \DWOPconstones&0x09&1&1\dash byte constant \\
1703 \DWOPconsttwou&0x0a&1&2\dash byte constant \\
1704 \DWOPconsttwos&0x0b&1&2\dash byte constant \\
1705 \DWOPconstfouru&0x0c&1&4\dash byte constant \\
1706 \DWOPconstfours&0x0d&1&4\dash byte constant \\
1707 \DWOPconsteightu&0x0e&1&8\dash byte constant \\
1708 \DWOPconsteights&0x0f&1&8\dash byte constant \\
1709 \DWOPconstu&0x10&1&ULEB128 constant \\
1710 \DWOPconsts&0x11&1&SLEB128 constant \\
1711 \DWOPdup&0x12&0 & \\
1712 \DWOPdrop&0x13&0 & \\
1713 \DWOPover&0x14&0 & \\
1714 \DWOPpick&0x15&1&1\dash byte stack index \\
1715 \DWOPswap&0x16&0 & \\
1716 \DWOProt&0x17&0 & \\
1717 \DWOPxderef&0x18&0 & \\
1718 \DWOPabs&0x19&0 & \\
1719 \DWOPand&0x1a&0 & \\
1720 \DWOPdiv&0x1b&0 & \\
1721 \DWOPminus&0x1c&0 & \\
1722 \DWOPmod&0x1d&0 & \\
1723 \DWOPmul&0x1e&0 & \\
1724 \DWOPneg&0x1f&0 & \\
1725 \DWOPnot&0x20&0 & \\
1727 \DWOPplus&0x22&0 & \\
1728 \DWOPplusuconst&0x23&1&ULEB128 addend \\
1729 \DWOPshl&0x24&0 & \\
1730 \DWOPshr&0x25&0 & \\
1731 \DWOPshra&0x26&0 & \\
1732 \DWOPxor&0x27&0 & \\
1734 \DWOPbra&0x28&1 & signed 2\dash byte constant \\
1741 \DWOPskip&0x2f&1&signed 2\dash byte constant \\ \hline
1743 \DWOPlitzero & 0x30 & 0 & \\
1744 \DWOPlitone & 0x31 & 0& literals 0 .. 31 = \\
1745 \ldots & & &\hspace{0.3cm}(\DWOPlitzero{} + literal) \\
1746 \DWOPlitthirtyone & 0x4f & 0 & \\ \hline
1748 \DWOPregzero & 0x50 & 0 & \\*
1749 \DWOPregone & 0x51 & 0® 0 .. 31 = \\*
1750 \ldots & & &\hspace{0.3cm}(\DWOPregzero{} + regnum) \\*
1751 \DWOPregthirtyone & 0x6f & 0 & \\ \hline
1753 \DWOPbregzero & 0x70 &1 & SLEB128 offset \\*
1754 \DWOPbregone & 0x71 & 1 &base register 0 .. 31 = \\*
1755 ... & & &\hspace{0.3cm}(\DWOPbregzero{} + regnum) \\*
1756 \DWOPbregthirtyone & 0x8f & 1 & \\ \hline
1758 \DWOPregx{} & 0x90 &1&ULEB128 register \\
1759 \DWOPfbreg{} & 0x91&1&SLEB128 offset \\
1760 \DWOPbregx{} & 0x92&2 &ULEB128 register, \\*
1761 & & &SLEB128 offset \\
1762 \DWOPpiece{} & 0x93 &1& ULEB128 size of piece \\
1763 \DWOPderefsize{} & 0x94 &1& 1-byte size of data retrieved \\
1764 \DWOPxderefsize{} & 0x95&1&1-byte size of data retrieved \\
1765 \DWOPnop{} & 0x96 &0& \\
1767 \DWOPpushobjectaddress&0x97&0 & \\
1768 \DWOPcalltwo&0x98&1& 2\dash byte offset of DIE \\
1769 \DWOPcallfour&0x99&1& 4\dash byte offset of DIE \\
1770 \DWOPcallref&0x9a&1& 4\dash\ or 8\dash byte offset of DIE \\
1771 \DWOPformtlsaddress&0x9b &0& \\
1772 \DWOPcallframecfa{} &0x9c &0& \\
1773 \DWOPbitpiece&0x9d &2&ULEB128 size, \\*
1775 \DWOPimplicitvalue{} &0x9e &2&ULEB128 size, \\*
1776 &&&\nolink{block} of that size\\
1777 \DWOPstackvalue{} &0x9f &0& \\
1778 \DWOPimplicitpointer{}~\ddag &0xa0& 2 &4- or 8-byte offset of DIE, \\*
1779 &&&SLEB128 constant offset \\
1780 \DWOPaddrx~\ddag&0xa1&1&ULEB128 indirect address \\
1781 \DWOPconstx~\ddag&0xa2&1&ULEB128 indirect constant \\
1782 \DWOPlouser{} &0xe0 && \\
1783 \DWOPhiuser{} &\xff && \\
1789 \subsection{Location Descriptions}
1790 \label{datarep:locationdescriptions}
1792 A location description is used to compute the
1793 location of a variable or other entity.
1795 \subsection{Location Lists}
1796 \label{datarep:locationlists}
1798 Each entry in a \addtoindex{location list} is either a location list entry,
1799 a base address selection entry, or an
1800 \addtoindexx{end of list entry!in location list}
1804 \subsubsection{Location List Entries in Non-Split Objects}
1805 A \addtoindex{location list} entry consists of two address offsets followed
1806 by a 2\dash byte length, followed by a block of contiguous bytes
1807 that contains a DWARF location description. The length
1808 specifies the number of bytes in that block. The two offsets
1809 are the same size as an address on the target machine.
1812 A base address selection entry and an
1813 \addtoindexx{end of list entry!in location list}
1814 end of list entry each
1815 consist of two (constant or relocated) address offsets. The two
1816 offsets are the same size as an address on the target machine.
1818 For a \addtoindex{location list} to be specified, the base address of
1819 \addtoindexx{base address selection entry!in location list}
1820 the corresponding compilation unit must be defined
1821 (see Section \refersec{chap:normalandpartialcompilationunitentries}).
1823 \subsubsection{Location List Entries in Split Objects}
1824 An alternate form for location list entries is used in split objects.
1825 Each entry begins with a one-byte code that indicates the kind of entry
1826 that follows. The encodings for these constants are given in
1827 Table \refersec{tab:locationlistentryencodingvalues}.
1830 \setlength{\extrarowheight}{0.1cm}
1831 \begin{longtable}{l|c}
1832 \caption{Location list entry encoding values} \label{tab:locationlistentryencodingvalues} \\
1833 \hline \bfseries Location list entry encoding name&\bfseries Value \\ \hline
1835 \bfseries Location list entry encoding name&\bfseries Value\\ \hline
1837 \hline \emph{Continued on next page}
1841 \DWLLEendoflistentry & 0x0 \\
1842 \DWLLEbaseaddressselectionentry & 0x01 \\
1843 \DWLLEstartendentry & 0x02 \\
1844 \DWLLEstartlengthentry & 0x03 \\
1845 \DWLLEoffsetpairentry & 0x04 \\
1849 \section{Base Type Attribute Encodings}
1850 \label{datarep:basetypeattributeencodings}
1852 The encodings of the
1853 \hypertarget{chap:DWATencodingencodingofbasetype}{}
1855 \addtoindexx{encoding attribute!encoding}
1858 attribute are given in
1859 Table \refersec{tab:basetypeencodingvalues}
1862 \setlength{\extrarowheight}{0.1cm}
1863 \begin{longtable}{l|c}
1864 \caption{Base type encoding values} \label{tab:basetypeencodingvalues} \\
1865 \hline \bfseries Base type encoding name&\bfseries Value \\ \hline
1867 \bfseries Base type encoding name&\bfseries Value\\ \hline
1869 \hline \emph{Continued on next page}
1873 \DWATEaddress&0x01 \\
1874 \DWATEboolean&0x02 \\
1875 \DWATEcomplexfloat&0x03 \\
1877 \DWATEsigned&0x05 \\
1878 \DWATEsignedchar&0x06 \\
1879 \DWATEunsigned&0x07 \\
1880 \DWATEunsignedchar&0x08 \\
1881 \DWATEimaginaryfloat&0x09 \\
1882 \DWATEpackeddecimal&0x0a \\
1883 \DWATEnumericstring&0x0b \\
1884 \DWATEedited&0x0c \\
1885 \DWATEsignedfixed&0x0d \\
1886 \DWATEunsignedfixed&0x0e \\
1887 \DWATEdecimalfloat{} & 0x0f \\
1888 \DWATEUTF{} & 0x10 \\
1889 \DWATElouser{} & 0x80 \\
1890 \DWATEhiuser{} & \xff \\
1895 The encodings of the constants used in the
1896 \DWATdecimalsign{} attribute
1898 Table \refersec{tab:decimalsignencodings}.
1901 \setlength{\extrarowheight}{0.1cm}
1902 \begin{longtable}{l|c}
1903 \caption{Decimal sign encodings} \label{tab:decimalsignencodings} \\
1904 \hline \bfseries Decimal sign code name&\bfseries Value \\ \hline
1906 \bfseries Decimal sign code name&\bfseries Value\\ \hline
1908 \hline \emph{Continued on next page}
1913 \DWDSunsigned{} & 0x01 \\
1914 \DWDSleadingoverpunch{} & 0x02 \\
1915 \DWDStrailingoverpunch{} & 0x03 \\
1916 \DWDSleadingseparate{} & 0x04 \\
1917 \DWDStrailingseparate{} & 0x05 \\
1923 The encodings of the constants used in the
1924 \DWATendianity{} attribute are given in
1925 Table \refersec{tab:endianityencodings}.
1928 \setlength{\extrarowheight}{0.1cm}
1929 \begin{longtable}{l|c}
1930 \caption{Endianity encodings} \label{tab:endianityencodings}\\
1931 \hline \bfseries Endian code name&\bfseries Value \\ \hline
1933 \bfseries Endian code name&\bfseries Value\\ \hline
1935 \hline \emph{Continued on next page}
1940 \DWENDdefault{} & 0x00 \\
1941 \DWENDbig{} & 0x01 \\
1942 \DWENDlittle{} & 0x02 \\
1943 \DWENDlouser{} & 0x40 \\
1944 \DWENDhiuser{} & \xff \\
1949 \section{Accessibility Codes}
1950 \label{datarep:accessibilitycodes}
1951 The encodings of the constants used in the
1952 \DWATaccessibility{}
1954 \addtoindexx{accessibility attribute!encoding}
1956 Table \refersec{tab:accessibilityencodings}.
1959 \setlength{\extrarowheight}{0.1cm}
1960 \begin{longtable}{l|c}
1961 \caption{Accessibility encodings} \label{tab:accessibilityencodings}\\
1962 \hline \bfseries Accessibility code name&\bfseries Value \\ \hline
1964 \bfseries Accessibility code name&\bfseries Value\\ \hline
1966 \hline \emph{Continued on next page}
1971 \DWACCESSpublic&0x01 \\
1972 \DWACCESSprotected&0x02 \\
1973 \DWACCESSprivate&0x03 \\
1979 \section{Visibility Codes}
1980 \label{datarep:visibilitycodes}
1981 The encodings of the constants used in the
1982 \DWATvisibility{} attribute are given in
1983 Table \refersec{tab:visibilityencodings}.
1986 \setlength{\extrarowheight}{0.1cm}
1987 \begin{longtable}{l|c}
1988 \caption{Visibility encodings} \label{tab:visibilityencodings}\\
1989 \hline \bfseries Visibility code name&\bfseries Value \\ \hline
1991 \bfseries Visibility code name&\bfseries Value\\ \hline
1993 \hline \emph{Continued on next page}
1999 \DWVISexported&0x02 \\
2000 \DWVISqualified&0x03 \\
2005 \section{Virtuality Codes}
2006 \label{datarep:vitualitycodes}
2008 The encodings of the constants used in the
2009 \DWATvirtuality{} attribute are given in
2010 Table \refersec{tab:virtualityencodings}.
2013 \setlength{\extrarowheight}{0.1cm}
2014 \begin{longtable}{l|c}
2015 \caption{Virtuality encodings} \label{tab:virtualityencodings}\\
2016 \hline \bfseries Virtuality code name&\bfseries Value \\ \hline
2018 \bfseries Virtuality code name&\bfseries Value\\ \hline
2020 \hline \emph{Continued on next page}
2025 \DWVIRTUALITYnone&0x00 \\
2026 \DWVIRTUALITYvirtual&0x01 \\
2027 \DWVIRTUALITYpurevirtual&0x02 \\
2035 \DWVIRTUALITYnone{} is equivalent to the absence of the
2039 \section{Source Languages}
2040 \label{datarep:sourcelanguages}
2042 The encodings of the constants used
2043 \addtoindexx{language attribute, encoding}
2045 \addtoindexx{language name encoding}
2048 attribute are given in
2049 Table \refersec{tab:languageencodings}.
2051 % If we don't force a following space it looks odd
2053 and their associated values are reserved, but the
2054 languages they represent are not well supported.
2055 Table \refersec{tab:languageencodings}
2057 \addtoindexx{lower bound attribute!default}
2058 default lower bound, if any, assumed for
2059 an omitted \DWATlowerbound{} attribute in the context of a
2060 \DWTAGsubrangetype{} debugging information entry for each
2064 \setlength{\extrarowheight}{0.1cm}
2065 \begin{longtable}{l|c|c}
2066 \caption{Language encodings} \label{tab:languageencodings}\\
2067 \hline \bfseries Language name&\bfseries Value &\bfseries Default Lower Bound \\ \hline
2069 \bfseries Language name&\bfseries Value &\bfseries Default Lower Bound\\ \hline
2071 \hline \emph{Continued on next page}
2074 \dag \ \textit{See text} \\ \ddag \ \textit{New in \addtoindex{DWARF Version 5}}
2077 \DWLANGCeightynine &0x0001 &0 \\
2078 \DWLANGC{} &0x0002 &0 \\
2079 \DWLANGAdaeightythree{} \dag &0x0003 &1 \addtoindexx{Ada} \\
2080 \DWLANGCplusplus{} &0x0004 &0 \\
2081 \DWLANGCobolseventyfour{} \dag &0x0005 &1 \\
2082 \DWLANGCoboleightyfive{} \dag &0x0006 &1 \\
2083 \DWLANGFortranseventyseven &0x0007 &1 \\
2084 \DWLANGFortranninety &0x0008 &1 \\
2085 \DWLANGPascaleightythree &0x0009 &1 \\
2086 \DWLANGModulatwo &0x000a &1 \\
2087 \DWLANGJava &0x000b &0 \\
2088 \DWLANGCninetynine &0x000c &0 \\
2089 \DWLANGAdaninetyfive{} \dag &0x000d &1 \addtoindexx{Ada} \\
2090 \DWLANGFortranninetyfive &0x000e &1 \\
2091 \DWLANGPLI{} \dag &0x000f &1 \\
2092 \DWLANGObjC{} &0x0010 &0 \\
2093 \DWLANGObjCplusplus{} &0x0011 &0 \\
2094 \DWLANGUPC{} &0x0012 &0 \\
2095 \DWLANGD{} &0x0013 &0 \\
2096 \DWLANGPython{} \dag &0x0014 &0 \\
2097 \DWLANGOpenCL{} \dag \ddag &0x0015 &0 \\
2098 \DWLANGGo{} \dag \ddag &0x0016 &0 \\
2099 \DWLANGModulathree{} \dag \ddag &0x0017 &1 \\
2100 \DWLANGHaskell{} \dag \ddag &0x0018 &0 \\
2101 \DWLANGCpluspluszerothree{} \ddag &0x0019 &0 \\
2102 \DWLANGCpluspluseleven{} \ddag &0x001a &0 \\
2103 \DWLANGOCaml{} \ddag &0x001b &0 \\
2104 \DWLANGRust{} \ddag &0x001c &0 \\
2105 \DWLANGlouser{} &0x8000 & \\
2106 \DWLANGhiuser{} &\xffff & \\
2111 \section{Address Class Encodings}
2112 \label{datarep:addressclassencodings}
2114 The value of the common
2115 \addtoindexi{address}{address class!attribute encoding}
2120 \section{Identifier Case}
2121 \label{datarep:identifiercase}
2123 The encodings of the constants used in the
2124 \DWATidentifiercase{} attribute are given in
2125 Table \refersec{tab:identifiercaseencodings}.
2128 \setlength{\extrarowheight}{0.1cm}
2129 \begin{longtable}{l|c}
2130 \caption{Identifier case encodings} \label{tab:identifiercaseencodings}\\
2131 \hline \bfseries Identifier case name&\bfseries Value \\ \hline
2133 \bfseries Identifier case name&\bfseries Value\\ \hline
2135 \hline \emph{Continued on next page}
2139 \DWIDcasesensitive&0x00 \\
2141 \DWIDdowncase&0x02 \\
2142 \DWIDcaseinsensitive&0x03 \\
2146 \section{Calling Convention Encodings}
2147 \label{datarep:callingconventionencodings}
2148 The encodings of the constants used in the
2149 \DWATcallingconvention{} attribute are given in
2150 Table \refersec{tab:callingconventionencodings}.
2153 \setlength{\extrarowheight}{0.1cm}
2154 \begin{longtable}{l|c}
2155 \caption{Calling convention encodings} \label{tab:callingconventionencodings}\\
2156 \hline \bfseries Calling convention name&\bfseries Value \\ \hline
2158 \bfseries Calling convention name&\bfseries Value\\ \hline
2160 \hline \emph{Continued on next page}
2166 \DWCCprogram&0x02 \\
2174 \section{Inline Codes}
2175 \label{datarep:inlinecodes}
2177 The encodings of the constants used in
2178 \addtoindexx{inline attribute!encoding}
2180 \DWATinline{} attribute are given in
2181 Table \refersec{tab:inlineencodings}.
2185 \setlength{\extrarowheight}{0.1cm}
2186 \begin{longtable}{l|c}
2187 \caption{Inline encodings} \label{tab:inlineencodings}\\
2188 \hline \bfseries Inline code name&\bfseries Value \\ \hline
2190 \bfseries Inline Code name&\bfseries Value\\ \hline
2192 \hline \emph{Continued on next page}
2197 \DWINLnotinlined&0x00 \\
2198 \DWINLinlined&0x01 \\
2199 \DWINLdeclarednotinlined&0x02 \\
2200 \DWINLdeclaredinlined&0x03 \\
2205 % this clearpage is ugly, but the following table came
2206 % out oddly without it.
2208 \section{Array Ordering}
2209 \label{datarep:arrayordering}
2211 The encodings of the constants used in the
2212 \DWATordering{} attribute are given in
2213 Table \refersec{tab:orderingencodings}.
2217 \setlength{\extrarowheight}{0.1cm}
2218 \begin{longtable}{l|c}
2219 \caption{Ordering encodings} \label{tab:orderingencodings}\\
2220 \hline \bfseries Ordering name&\bfseries Value \\ \hline
2222 \bfseries Ordering name&\bfseries Value\\ \hline
2224 \hline \emph{Continued on next page}
2229 \DWORDrowmajor&0x00 \\
2230 \DWORDcolmajor&0x01 \\
2236 \section{Discriminant Lists}
2237 \label{datarep:discriminantlists}
2239 The descriptors used in
2240 \addtoindexx{discriminant list attribute!encoding}
2242 \DWATdiscrlist{} attribute are
2243 encoded as 1\dash byte constants. The
2244 defined values are given in
2245 Table \refersec{tab:discriminantdescriptorencodings}.
2247 % Odd that the 'Name' field capitalized here, it is not caps elsewhere.
2249 \setlength{\extrarowheight}{0.1cm}
2250 \begin{longtable}{l|c}
2251 \caption{Discriminant descriptor encodings} \label{tab:discriminantdescriptorencodings}\\
2252 \hline \bfseries Descriptor name&\bfseries Value \\ \hline
2254 \bfseries Descriptor name&\bfseries Value\\ \hline
2256 \hline \emph{Continued on next page}
2268 \section{Name Lookup Tables}
2269 \label{datarep:namelookuptables}
2271 Each set of entries in the table of global names contained
2272 in the \dotdebugpubnames{} and
2273 \dotdebugpubtypes{} sections begins
2274 with a header consisting of:
2275 \begin{enumerate}[1. ]
2277 \item \texttt{unit\_length} (\livelink{datarep:initiallengthvalues}{initial length}) \\
2278 \addttindexx{unit\_length}
2279 A 4\dash byte or 12\dash byte unsigned integer
2280 \addtoindexx{initial length}
2281 representing the length
2282 of the \dotdebuginfo{}
2283 contribution for that compilation unit,
2284 not including the length field itself. In the
2285 \thirtytwobitdwarfformat, this is a 4\dash byte unsigned integer (which must be less
2286 than \xfffffffzero); in the \sixtyfourbitdwarfformat, this consists
2287 of the 4\dash byte value \wffffffff followed by an 8\dash byte unsigned
2288 integer that gives the actual length
2289 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
2291 \item version (\addtoindex{uhalf}) \\
2292 A 2\dash byte unsigned integer representing the version of the
2293 DWARF information for the name lookup table
2294 \addtoindexx{version number!name lookup table}
2295 (see Appendix \refersec{app:dwarfsectionversionnumbersinformative}).
2296 The value in this field is 2.
2299 \item \addtoindex{debug\_info\_offset} (section offset) \\
2301 \addtoindexx{section offset!in name lookup table set of entries}
2302 4\dash byte or 8\dash byte
2304 \dotdebuginfo{} or \dotdebuginfodwo{}
2305 section of the compilation unit header.
2306 In the \thirtytwobitdwarfformat, this is a 4\dash byte unsigned offset;
2307 in the \sixtyfourbitdwarfformat, this is an 8\dash byte unsigned offsets
2308 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
2310 \item \addtoindex{debug\_info\_length} (\livelink{datarep:sectionoffsetlength}{section length}) \\
2311 \addtoindexx{section length!in .debug\_pubnames header}
2313 \addtoindexx{section length!in .debug\_pubtypes header}
2314 4\dash byte or 8\dash byte length containing the size in bytes of the
2315 contents of the \dotdebuginfo{}
2316 section generated to represent
2317 this compilation unit. In the \thirtytwobitdwarfformat, this is
2318 a 4\dash byte unsigned length; in the \sixtyfourbitdwarfformat, this
2319 is an 8-byte unsigned length
2320 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
2325 This header is followed by a series of tuples. Each tuple
2326 consists of a 4\dash byte or 8\dash byte offset followed by a string
2327 of non\dash null bytes terminated by one null byte.
2329 DWARF format, this is a 4\dash byte offset; in the 64\dash bit DWARF
2330 format, it is an 8\dash byte offset.
2331 Each set is terminated by an
2332 offset containing the value 0.
2336 \section{Address Range Table}
2337 \label{datarep:addrssrangetable}
2339 Each set of entries in the table of address ranges contained
2340 in the \dotdebugaranges{}
2341 section begins with a header containing:
2342 \begin{enumerate}[1. ]
2343 % FIXME The unit length text is not fully consistent across
2346 \item \texttt{unit\_length} (\livelink{datarep:initiallengthvalues}{initial length}) \\
2347 \addttindexx{unit\_length}
2348 A 4-byte or 12-byte length containing the length of the
2349 \addtoindexx{initial length}
2350 set of entries for this compilation unit, not including the
2351 length field itself. In the \thirtytwobitdwarfformat, this is a
2352 4-byte unsigned integer (which must be less than \xfffffffzero);
2353 in the \sixtyfourbitdwarfformat, this consists of the 4-byte value
2354 \wffffffff followed by an 8-byte unsigned integer that gives
2356 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
2358 \item version (\addtoindex{uhalf}) \\
2359 A 2\dash byte version identifier representing the version of the
2360 DWARF information for the address range table
2361 (see Appendix \refersec{app:dwarfsectionversionnumbersinformative}).
2362 This value in this field \addtoindexx{version number!address range table} is 2.
2365 \item debug\_info\_offset (\livelink{datarep:sectionoffsetlength}{section offset}) \\
2367 \addtoindexx{section offset!in .debug\_aranges header}
2368 4\dash byte or 8\dash byte offset into the
2369 \dotdebuginfo{} section of
2370 the compilation unit header. In the \thirtytwobitdwarfformat,
2371 this is a 4\dash byte unsigned offset; in the \sixtyfourbitdwarfformat,
2372 this is an 8\dash byte unsigned offset
2373 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
2375 \item address\_size (ubyte) \\
2376 A 1\dash byte unsigned integer containing the size in bytes of an
2377 \addtoindexx{address\_size}
2379 \addtoindexx{size of an address}
2380 (or the offset portion of an address for segmented
2381 \addtoindexx{address space!segmented}
2382 addressing) on the target system.
2384 \item segment\_size (ubyte) \\
2386 \addtoindexx{segment\_size}
2387 1\dash byte unsigned integer containing the size in bytes of a
2388 segment selector on the target system.
2392 This header is followed by a series of tuples. Each tuple
2393 consists of a segment, an address and a length.
2395 size is given by the \addtoindex{segment\_size} field of the header; the
2396 address and length size are each given by the address\_size
2397 field of the header.
2398 The first tuple following the header in
2399 each set begins at an offset that is a multiple of the size
2400 of a single tuple (that is, the size of a segment selector
2401 plus twice the \addtoindex{size of an address}).
2402 The header is padded, if
2403 necessary, to that boundary. Each set of tuples is terminated
2404 by a 0 for the segment, a 0 for the address and 0 for the
2405 length. If the \addtoindex{segment\_size} field in the header is zero,
2406 the segment selectors are omitted from all tuples, including
2407 the terminating tuple.
2410 \section{Line Number Information}
2411 \label{datarep:linenumberinformation}
2413 The \addtoindexi{version number}{version number!line number information}
2414 in the line number program header is \versiondotdebugline{}
2415 (see Appendix \refersec{app:dwarfsectionversionnumbersinformative}).
2417 The boolean values \doublequote{true} and \doublequote{false}
2418 used by the line number information program are encoded
2419 as a single byte containing the value 0
2420 for \doublequote{false,} and a non-zero value for \doublequote{true.}
2422 The encodings for the standard opcodes are given in
2423 \addtoindexx{line number opcodes!standard opcode encoding}
2424 Table \refersec{tab:linenumberstandardopcodeencodings}.
2426 % Odd that the 'Name' field capitalized here, it is not caps elsewhere.
2428 \setlength{\extrarowheight}{0.1cm}
2429 \begin{longtable}{l|c}
2430 \caption{Line number standard opcode encodings} \label{tab:linenumberstandardopcodeencodings}\\
2431 \hline \bfseries Opcode name&\bfseries Value \\ \hline
2433 \bfseries Opcode name&\bfseries Value\\ \hline
2435 \hline \emph{Continued on next page}
2441 \DWLNSadvancepc&0x02 \\
2442 \DWLNSadvanceline&0x03 \\
2443 \DWLNSsetfile&0x04 \\
2444 \DWLNSsetcolumn&0x05 \\
2445 \DWLNSnegatestmt&0x06 \\
2446 \DWLNSsetbasicblock&0x07 \\
2447 \DWLNSconstaddpc&0x08 \\
2448 \DWLNSfixedadvancepc&0x09 \\
2449 \DWLNSsetprologueend&0x0a \\*
2450 \DWLNSsetepiloguebegin&0x0b \\*
2451 \DWLNSsetisa&0x0c \\*
2458 The encodings for the extended opcodes are given in
2459 \addtoindexx{line number opcodes!extended opcode encoding}
2460 Table \refersec{tab:linenumberextendedopcodeencodings}.
2463 \setlength{\extrarowheight}{0.1cm}
2464 \begin{longtable}{l|c}
2465 \caption{Line number extended opcode encodings} \label{tab:linenumberextendedopcodeencodings}\\
2466 \hline \bfseries Opcode name&\bfseries Value \\ \hline
2468 \bfseries Opcode name&\bfseries Value\\ \hline
2470 \hline \emph{Continued on next page}
2472 \hline \ddag~\textit{New in DWARF Version 5}
2475 \DWLNEendsequence &0x01 \\
2476 \DWLNEsetaddress &0x02 \\
2477 \DWLNEdefinefile &0x03 \\
2478 \DWLNEsetdiscriminator &0x04 \\
2479 \DWLNEdefinefileMDfive~\ddag &0x05 \\
2480 \DWLNElouser &0x80 \\
2481 \DWLNEhiuser &\xff \\
2487 The encodings for the file entry format are given in
2488 \addtoindexx{line number opcodes!file entry format encoding}
2489 Table \refersec{tab:linenumberfileentryformatencodings}.
2492 \setlength{\extrarowheight}{0.1cm}
2493 \begin{longtable}{l|c}
2494 \caption{Line number file entry format \mbox{encodings}} \label{tab:linenumberfileentryformatencodings}\\
2495 \hline \bfseries File entry format name&\bfseries Value \\ \hline
2497 \bfseries File entry format name&\bfseries Value\\ \hline
2499 \hline \emph{Continued on next page}
2504 \DWLNFtimestampsize & 0x01 \\
2505 \DWLNFMDfive & 0x02 \\
2510 \section{Macro Information}
2511 \label{datarep:macroinformation}
2513 The source line numbers and source file indices encoded in the
2514 macro information section are represented as unsigned LEB128
2517 The macro information entry type is encoded as a single byte.
2519 \addtoindexx{macro information entry types!encoding}
2521 Table \refersec{tab:macroinfoentrytypeencodings}.
2525 \setlength{\extrarowheight}{0.1cm}
2526 \begin{longtable}{l|c}
2527 \caption{Macro information entry type encodings} \label{tab:macroinfoentrytypeencodings}\\
2528 \hline \bfseries Macro information entry type name&\bfseries Value \\ \hline
2530 \bfseries Macro information entry type name&\bfseries Value\\ \hline
2532 \hline \emph{Continued on next page}
2537 \DWMACROdefine &0x01 \\
2538 \DWMACROundef &0x02 \\
2539 \DWMACROstartfile &0x03 \\
2540 \DWMACROendfile &0x04 \\
2541 \DWMACROdefineindirect &0x05 \\
2542 \DWMACROundefindirect &0x06 \\
2543 \DWMACROtransparentinclude &0x07 \\
2544 % what about 0x08 thru 0x0a??
2545 \DWMACROdefineindirectx &0x0b \\
2546 \DWMACROundefindirectx &0x0c \\
2547 \DWMACROlouser &0xe0 \\
2548 \DWMACROhiuser &\xff \\
2554 \section{Call Frame Information}
2555 \label{datarep:callframeinformation}
2557 In the \thirtytwobitdwarfformat, the value of the CIE id in the
2558 CIE header is \xffffffff; in the \sixtyfourbitdwarfformat, the
2559 value is \xffffffffffffffff.
2561 The value of the CIE \addtoindexi{version number}{version number!call frame information}
2562 is 4 (see Appendix \refersec{app:dwarfsectionversionnumbersinformative}).
2564 Call frame instructions are encoded in one or more bytes. The
2565 primary opcode is encoded in the high order two bits of
2566 the first byte (that is, opcode = byte $\gg$ 6). An operand
2567 or extended opcode may be encoded in the low order 6
2568 bits. Additional operands are encoded in subsequent bytes.
2569 The instructions and their encodings are presented in
2570 Table \refersec{tab:callframeinstructionencodings}.
2573 \setlength{\extrarowheight}{0.1cm}
2574 \begin{longtable}{l|c|c|l|l}
2575 \caption{Call frame instruction encodings} \label{tab:callframeinstructionencodings} \\
2576 \hline &\bfseries High 2 &\bfseries Low 6 & & \\
2577 \bfseries Instruction&\bfseries Bits &\bfseries Bits &\bfseries Operand 1 &\bfseries Operand 2\\ \hline
2579 & \bfseries High 2 &\bfseries Low 6 & &\\
2580 \bfseries Instruction&\bfseries Bits &\bfseries Bits &\bfseries Operand 1 &\bfseries Operand 2\\ \hline
2582 \hline \emph{Continued on next page}
2587 \DWCFAadvanceloc&0x1&delta & \\
2588 \DWCFAoffset&0x2®ister&ULEB128 offset \\
2589 \DWCFArestore&0x3®ister & & \\
2590 \DWCFAnop&0&0 & & \\
2591 \DWCFAsetloc&0&0x01&address & \\
2592 \DWCFAadvancelocone&0&0x02&1\dash byte delta & \\
2593 \DWCFAadvanceloctwo&0&0x03&2\dash byte delta & \\
2594 \DWCFAadvancelocfour&0&0x04&4\dash byte delta & \\
2595 \DWCFAoffsetextended&0&0x05&ULEB128 register&ULEB128 offset \\
2596 \DWCFArestoreextended&0&0x06&ULEB128 register & \\
2597 \DWCFAundefined&0&0x07&ULEB128 register & \\
2598 \DWCFAsamevalue&0&0x08 &ULEB128 register & \\
2599 \DWCFAregister&0&0x09&ULEB128 register &ULEB128 offset \\
2600 \DWCFArememberstate&0&0x0a & & \\
2601 \DWCFArestorestate&0&0x0b & & \\
2602 \DWCFAdefcfa&0&0x0c &ULEB128 register&ULEB128 offset \\
2603 \DWCFAdefcfaregister&0&0x0d&ULEB128 register & \\
2604 \DWCFAdefcfaoffset&0&0x0e &ULEB128 offset & \\
2605 \DWCFAdefcfaexpression&0&0x0f &BLOCK \\
2606 \DWCFAexpression&0&0x10&ULEB128 register & BLOCK \\
2608 \DWCFAoffsetextendedsf&0&0x11&ULEB128 register&SLEB128 offset \\
2609 \DWCFAdefcfasf&0&0x12&ULEB128 register&SLEB128 offset \\
2610 \DWCFAdefcfaoffsetsf&0&0x13&SLEB128 offset & \\
2611 \DWCFAvaloffset&0&0x14&ULEB128&ULEB128 \\
2612 \DWCFAvaloffsetsf&0&0x15&ULEB128&SLEB128 \\
2613 \DWCFAvalexpression&0&0x16&ULEB128&BLOCK \\
2614 \DWCFAlouser&0&0x1c & & \\
2615 \DWCFAhiuser&0&\xiiif & & \\
2619 \section{Non-contiguous Address Ranges}
2620 \label{datarep:noncontiguousaddressranges}
2622 Each entry in a \addtoindex{range list}
2623 (see Section \refersec{chap:noncontiguousaddressranges})
2625 \addtoindexx{base address selection entry!in range list}
2627 \addtoindexx{range list}
2628 a base address selection entry, or an end
2631 A \addtoindex{range list} entry consists of two relative addresses. The
2632 addresses are the same size as addresses on the target machine.
2634 A base address selection entry and an
2635 \addtoindexx{end of list entry!in range list}
2636 end of list entry each
2637 \addtoindexx{base address selection entry!in range list}
2638 consist of two (constant or relocated) addresses. The two
2639 addresses are the same size as addresses on the target machine.
2641 For a \addtoindex{range list} to be specified, the base address of the
2642 \addtoindexx{base address selection entry!in range list}
2643 corresponding compilation unit must be defined
2644 (see Section \refersec{chap:normalandpartialcompilationunitentries}).
2646 \section{String Offsets Table}
2647 \label{chap:stringoffsetstable}
2648 Each set of entries in the string offsets table contained in the
2649 \dotdebugstroffsets{} or \dotdebugstroffsetsdwo{}
2650 section begins with a header containing:
2651 \begin{enumerate}[1. ]
2652 \item \texttt{unit\_length} (\livelink{datarep:initiallengthvalues}{initial length}) \\
2653 A 4-byte or 12-byte length containing the length of
2654 the set of entries for this compilation unit, not
2655 including the length field itself. In the 32-bit
2656 DWARF format, this is a 4-byte unsigned integer
2657 (which must be less than \xfffffffzero); in the 64-bit
2658 DWARF format, this consists of the 4-byte value
2659 \wffffffff followed by an 8-byte unsigned integer
2660 that gives the actual length (see
2661 Section \refersec{datarep:32bitand64bitdwarfformats}).
2663 \item \texttt{version} (\addtoindex{uhalf}) \\
2664 A 2-byte version identifier containing the value
2665 \versiondotdebugstroffsets{}
2666 (see Appendix \refersec{app:dwarfsectionversionnumbersinformative}).
2667 \item \texttt{padding} (\addtoindex{uhalf}) \\
2670 This header is followed by a series of string table offsets
2671 that have the same representation as \DWFORMstrp.
2672 For the 32-bit DWARF format, each offset is 4 bytes long; for
2673 the 64-bit DWARF format, each offset is 8 bytes long.
2675 The \DWATstroffsetsbase{} attribute points to the first
2676 entry following the header. The entries are indexed
2677 sequentially from this base entry, starting from 0.
2679 \section{Address Table}
2680 \label{chap:addresstable}
2681 Each set of entries in the address table contained in the
2682 \dotdebugaddr{} section begins with a header containing:
2683 \begin{enumerate}[1. ]
2684 \item \texttt{unit\_length} (\livelink{datarep:initiallengthvalues}{initial length}) \\
2685 A 4-byte or 12-byte length containing the length of
2686 the set of entries for this compilation unit, not
2687 including the length field itself. In the 32-bit
2688 DWARF format, this is a 4-byte unsigned integer
2689 (which must be less than \xfffffffzero); in the 64-bit
2690 DWARF format, this consists of the 4-byte value
2691 \wffffffff followed by an 8-byte unsigned integer
2692 that gives the actual length (see
2693 Section \refersec{datarep:32bitand64bitdwarfformats}).
2696 \item \texttt{version} (\addtoindex{uhalf}) \\
2697 A 2-byte version identifier containing the value
2698 \versiondotdebugaddr{}
2699 (see Appendix \refersec{app:dwarfsectionversionnumbersinformative}).
2702 \item \texttt{address\_size} (\addtoindex{ubyte}) \\
2703 A 1-byte unsigned integer containing the size in
2704 bytes of an address (or the offset portion of an
2705 address for segmented addressing) on the target
2709 \item \texttt{segment\_size} (\addtoindex{ubyte}) \\
2710 A 1-byte unsigned integer containing the size in
2711 bytes of a segment selector on the target system.
2714 This header is followed by a series of segment/address pairs.
2715 The segment size is given by the \texttt{segment\_size} field of the
2716 header, and the address size is given by the \texttt{address\_size}
2717 field of the header. If the \texttt{segment\_size} field in the header
2718 is zero, the entries consist only of an addresses.
2720 The \DWATaddrbase{} attribute points to the first entry
2721 following the header. The entries are indexed sequentially
2722 from this base entry, starting from 0.
2724 \section{Range List Table}
2725 \label{app:rangelisttable}
2726 Each set of entries in the range list table contained in the
2727 \dotdebugranges{} section begins with a header containing:
2728 \begin{enumerate}[1. ]
2729 \item \texttt{unit\_length} (\livelink{datarep:initiallengthvalues}{initial length}) \\
2730 A 4-byte or 12-byte length containing the length of
2731 the set of entries for this compilation unit, not
2732 including the length field itself. In the 32-bit
2733 DWARF format, this is a 4-byte unsigned integer
2734 (which must be less than \xfffffffzero); in the 64-bit
2735 DWARF format, this consists of the 4-byte value
2736 \wffffffff followed by an 8-byte unsigned integer
2737 that gives the actual length (see
2738 Section \refersec{datarep:32bitand64bitdwarfformats}).
2741 \item \texttt{version} (\addtoindex{uhalf}) \\
2742 A 2-byte version identifier containing the value
2743 \versiondotdebugranges{}
2744 (see Appendix \refersec{app:dwarfsectionversionnumbersinformative}).
2747 \item \texttt{address\_size} (\addtoindex{ubyte}) \\
2748 A 1-byte unsigned integer containing the size in
2749 bytes of an address (or the offset portion of an
2750 address for segmented addressing) on the target
2754 \item \texttt{segment\_size} (\addtoindex{ubyte}) \\
2755 A 1-byte unsigned integer containing the size in
2756 bytes of a segment selector on the target system.
2759 This header is followed by a series of range list entries as
2760 described in Section \refersec{chap:locationlists}.
2761 The segment size is given by the
2762 \texttt{segment\_size} field of the header, and the address size is
2763 given by the \texttt{address\_size} field of the header. If the
2764 \texttt{segment\_size} field in the header is zero, the segment
2765 selector is omitted from the range list entries.
2767 The \DWATrangesbase{} attribute points to the first entry
2768 following the header. The entries are referenced by a byte
2769 offset relative to this base address.
2772 \section{Location List Table}
2773 \label{datarep:locationlisttable}
2774 Each set of entries in the location list table contained in the
2775 \dotdebugloc{} or \dotdebuglocdwo{} sections begins with a header containing:
2776 \begin{enumerate}[1. ]
2777 \item \texttt{unit\_length} (\livelink{datarep:initiallengthvalues}{initial length}) \\
2778 A 4-byte or 12-byte length containing the length of
2779 the set of entries for this compilation unit, not
2780 including the length field itself. In the 32-bit
2781 DWARF format, this is a 4-byte unsigned integer
2782 (which must be less than \xfffffffzero); in the 64-bit
2783 DWARF format, this consists of the 4-byte value
2784 \wffffffff followed by an 8-byte unsigned integer
2785 that gives the actual length (see
2786 Section \refersec{datarep:32bitand64bitdwarfformats}).
2789 \item \texttt{version} (\addtoindex{uhalf}) \\
2790 A 2-byte version identifier containing the value
2791 \versiondotdebugloc{}
2792 (see Appendix \refersec{app:dwarfsectionversionnumbersinformative}).
2795 \item \texttt{address\_size} (\addtoindex{ubyte}) \\
2796 A 1-byte unsigned integer containing the size in
2797 bytes of an address (or the offset portion of an
2798 address for segmented addressing) on the target
2802 \item \texttt{segment\_size} (\addtoindex{ubyte}) \\
2803 A 1-byte unsigned integer containing the size in
2804 bytes of a segment selector on the target system.
2807 This header is followed by a series of location list entries as
2808 described in Section \refersec{chap:locationlists}.
2809 The segment size is given by the
2810 \texttt{segment\_size} field of the header, and the address size is
2811 given by the \texttt{address\_size} field of the header. If the
2812 \texttt{segment\_size} field in the header is zero, the segment
2813 selector is omitted from the range list entries.
2815 The entries are referenced by a byte offset relative to the first
2816 location list following this header.
2819 \section{Dependencies and Constraints}
2820 \label{datarep:dependenciesandconstraints}
2822 The debugging information in this format is intended to
2824 \addtoindexx{DWARF section names!list of}
2834 \dotdebugpubnames{},
2835 \dotdebugpubtypes{},
2839 \dotdebugstroffsets{}
2840 sections of an object file, or equivalent
2841 separate file or database. The information is not
2842 word\dash aligned. Consequently:
2845 \item For the \thirtytwobitdwarfformat{} and a target architecture with
2846 32\dash bit addresses, an assembler or compiler must provide a way
2847 to produce 2\dash byte and 4\dash byte quantities without alignment
2848 restrictions, and the linker must be able to relocate a
2849 4\dash byte address or
2850 \addtoindexx{section offset!alignment of}
2851 section offset that occurs at an arbitrary
2854 \item For the \thirtytwobitdwarfformat{} and a target architecture with
2855 64\dash bit addresses, an assembler or compiler must provide a
2856 way to produce 2\dash byte, 4\dash byte and 8\dash byte quantities without
2857 alignment restrictions, and the linker must be able to relocate
2858 an 8\dash byte address or 4\dash byte
2859 \addtoindexx{section offset!alignment of}
2860 section offset that occurs at an
2861 arbitrary alignment.
2863 \item For the \sixtyfourbitdwarfformat{} and a target architecture with
2864 32\dash bit addresses, an assembler or compiler must provide a
2865 way to produce 2\dash byte, 4\dash byte and 8\dash byte quantities without
2866 alignment restrictions, and the linker must be able to relocate
2867 a 4\dash byte address or 8\dash byte
2868 \addtoindexx{section offset!alignment of}
2869 section offset that occurs at an
2870 arbitrary alignment.
2872 \textit{It is expected that this will be required only for very large
2873 32\dash bit programs or by those architectures which support
2874 a mix of 32\dash bit and 64\dash bit code and data within the same
2877 \item For the \sixtyfourbitdwarfformat{} and a target architecture with
2878 64\dash bit addresses, an assembler or compiler must provide a
2879 way to produce 2\dash byte, 4\dash byte and 8\dash byte quantities without
2880 alignment restrictions, and the linker must be able to
2881 relocate an 8\dash byte address or
2882 \addtoindexx{section offset!alignment of}
2883 section offset that occurs at
2884 an arbitrary alignment.
2887 \section{Integer Representation Names}
2888 \label{datarep:integerrepresentationnames}
2890 The sizes of the integers used in the lookup by name, lookup
2891 by address, line number and call frame information sections
2893 Table \ref{tab:integerrepresentationnames}.
2897 \setlength{\extrarowheight}{0.1cm}
2898 \begin{longtable}{c|l}
2899 \caption{Integer representation names} \label{tab:integerrepresentationnames}\\
2900 \hline \bfseries Representation name&\bfseries Representation \\ \hline
2902 \bfseries Representation name&\bfseries Representation\\ \hline
2904 \hline \emph{Continued on next page}
2909 \addtoindex{sbyte}& signed, 1\dash byte integer \\
2910 \addtoindex{ubyte}&unsigned, 1\dash byte integer \\
2911 \addtoindex{uhalf}&unsigned, 2\dash byte integer \\
2912 \addtoindex{uword}&unsigned, 4\dash byte integer \\
2918 \section{Type Signature Computation}
2919 \label{datarep:typesignaturecomputation}
2921 A type signature is computed only by the DWARF producer;
2922 \addtoindexx{type signature computation}
2923 it is used by a DWARF consumer to resolve type references to
2924 the type definitions that are contained in
2925 \addtoindexx{type unit}
2928 The type signature for a type T0 is formed from the
2929 \MDfive{}\footnote{\livetarg{def:MDfive}{MD5} Message Digest Algorithm,
2930 R.L. Rivest, RFC 1321, April 1992}
2931 hash of a flattened description of the type. The flattened
2932 description of the type is a byte sequence derived from the
2933 DWARF encoding of the type as follows:
2934 \begin{enumerate}[1. ]
2936 \item Start with an empty sequence S and a list V of visited
2937 types, where V is initialized to a list containing the type
2938 T0 as its single element. Elements in V are indexed from 1,
2941 \item If the debugging information entry represents a type that
2942 is nested inside another type or a namespace, append to S
2943 the type\textquoteright s context as follows: For each surrounding type
2944 or namespace, beginning with the outermost such construct,
2945 append the letter 'C', the DWARF tag of the construct, and
2946 the name (taken from
2947 \addtoindexx{name attribute}
2948 the \DWATname{} attribute) of the type
2949 \addtoindexx{name attribute}
2950 or namespace (including its trailing null byte).
2952 \item Append to S the letter 'D', followed by the DWARF tag of
2953 the debugging information entry.
2955 \item For each of the attributes in
2956 Table \refersec{tab:attributesusedintypesignaturecomputation}
2958 the debugging information entry, in the order listed,
2959 append to S a marker letter (see below), the DWARF attribute
2960 code, and the attribute value.
2963 \caption{Attributes used in type signature computation}
2964 \label{tab:attributesusedintypesignaturecomputation}
2965 \simplerule[\textwidth]
2967 \autocols[0pt]{c}{2}{l}{
2982 \DWATcontainingtype,
2986 \DWATdatamemberlocation,
3007 \DWATrvaluereference,
3011 \DWATstringlengthbitsize,
3012 \DWATstringlengthbytesize,
3017 \DWATvariableparameter,
3020 \DWATvtableelemlocation
3023 \simplerule[\textwidth]
3026 Note that except for the initial
3027 \DWATname{} attribute,
3028 \addtoindexx{name attribute}
3029 attributes are appended in order according to the alphabetical
3030 spelling of their identifier.
3032 If an implementation defines any vendor-specific attributes,
3033 any such attributes that are essential to the definition of
3034 the type should also be included at the end of the above list,
3035 in their own alphabetical suborder.
3037 An attribute that refers to another type entry T is processed
3038 as follows: (a) If T is in the list V at some V[x], use the
3039 letter 'R' as the marker and use the unsigned LEB128 encoding
3040 of x as the attribute value; otherwise, (b) use the letter 'T'
3041 as the marker, process the type T recursively by performing
3042 Steps 2 through 7, and use the result as the attribute value.
3044 Other attribute values use the letter 'A' as the marker, and
3045 the value consists of the form code (encoded as an unsigned
3046 LEB128 value) followed by the encoding of the value according
3047 to the form code. To ensure reproducibility of the signature,
3048 the set of forms used in the signature computation is limited
3056 \item If the tag in Step 3 is one of \DWTAGpointertype,
3057 \DWTAGreferencetype,
3058 \DWTAGrvaluereferencetype,
3059 \DWTAGptrtomembertype,
3060 or \DWTAGfriend, and the referenced
3061 type (via the \DWATtype{} or
3062 \DWATfriend{} attribute) has a
3063 \DWATname{} attribute, append to S the letter 'N', the DWARF
3064 attribute code (\DWATtype{} or
3065 \DWATfriend), the context of
3066 the type (according to the method in Step 2), the letter 'E',
3067 and the name of the type. For \DWTAGfriend, if the referenced
3068 entry is a \DWTAGsubprogram, the context is omitted and the
3069 name to be used is the ABI-specific name of the subprogram
3070 (for example, the mangled linker name).
3073 \item If the tag in Step 3 is not one of \DWTAGpointertype,
3074 \DWTAGreferencetype,
3075 \DWTAGrvaluereferencetype,
3076 \DWTAGptrtomembertype, or
3077 \DWTAGfriend, but has
3078 a \DWATtype{} attribute, or if the referenced type (via
3080 \DWATfriend{} attribute) does not have a
3081 \DWATname{} attribute, the attribute is processed according to
3082 the method in Step 4 for an attribute that refers to another
3086 \item Visit each child C of the debugging information
3087 entry as follows: If C is a nested type entry or a member
3088 function entry, and has
3089 a \DWATname{} attribute, append to
3090 \addtoindexx{name attribute}
3091 S the letter 'S', the tag of C, and its name; otherwise,
3092 process C recursively by performing Steps 3 through 7,
3093 appending the result to S. Following the last child (or if
3094 there are no children), append a zero byte.
3099 For the purposes of this algorithm, if a debugging information
3101 \DWATspecification{}
3102 attribute that refers to
3103 another entry D (which has a
3106 then S inherits the attributes and children of D, and S is
3107 processed as if those attributes and children were present in
3108 the entry S. Exception: if a particular attribute is found in
3109 both S and D, the attribute in S is used and the corresponding
3110 one in D is ignored.
3112 DWARF tag and attribute codes are appended to the sequence
3113 as unsigned LEB128 values, using the values defined earlier
3116 \textit{A grammar describing this computation may be found in
3117 Appendix \refersec{app:typesignaturecomputationgrammar}.
3120 \textit{An attribute that refers to another type entry should
3121 be recursively processed or replaced with the name of the
3122 referent (in Step 4, 5 or 6). If neither treatment applies to
3123 an attribute that references another type entry, the entry
3124 that contains that attribute should not be considered for a
3125 separate \addtoindex{type unit}.}
3127 \textit{If a debugging information entry contains an attribute from
3128 the list above that would require an unsupported form, that
3129 entry should not be considered for a separate
3130 \addtoindex{type unit}.}
3132 \textit{A type should be considered for a separate
3133 \addtoindex{type unit} only
3134 if all of the type entries that it contains or refers to in
3135 Steps 6 and 7 can themselves each be considered for a separate
3136 \addtoindex{type unit}.}
3139 Where the DWARF producer may reasonably choose two or more
3140 different forms for a given attribute, it should choose
3141 the simplest possible form in computing the signature. (For
3142 example, a constant value should be preferred to a location
3143 expression when possible.)
3145 Once the string S has been formed from the DWARF encoding,
3146 an \MDfive{} hash is computed for the string and the
3147 least significant 64 bits are taken as the type signature.
3149 \textit{The string S is intended to be a flattened representation of
3150 the type that uniquely identifies that type (that is, a different
3151 type is highly unlikely to produce the same string).}
3153 \textit{A debugging information entry should not be placed in a
3154 separate \addtoindex{type unit}
3155 if any of the following apply:}
3159 \item \textit{The entry has an attribute whose value is a location
3160 expression, and the location expression contains a reference to
3161 another debugging information entry (for example, a \DWOPcallref{}
3162 operator), as it is unlikely that the entry will remain
3163 identical across compilation units.}
3165 \item \textit{The entry has an attribute whose value refers
3166 to a code location or a \addtoindex{location list}.}
3168 \item \textit{The entry has an attribute whose value refers
3169 to another debugging information entry that does not represent
3175 \textit{Certain attributes are not included in the type signature:}
3178 \item \textit{The \DWATdeclaration{} attribute is not included because it
3179 indicates that the debugging information entry represents an
3180 incomplete declaration, and incomplete declarations should
3182 \addtoindexx{type unit}
3183 separate type units.}
3185 \item \textit{The \DWATdescription{} attribute is not included because
3186 it does not provide any information unique to the defining
3187 declaration of the type.}
3189 \item \textit{The \DWATdeclfile,
3191 \DWATdeclcolumn{} attributes are not included because they
3192 may vary from one source file to the next, and would prevent
3193 two otherwise identical type declarations from producing the
3194 same \MDfive{} hash.}
3196 \item \textit{The \DWATobjectpointer{} attribute is not included
3197 because the information it provides is not necessary for the
3198 computation of a unique type signature.}
3202 \textit{Nested types and some types referred to by a debugging
3203 information entry are encoded by name rather than by recursively
3204 encoding the type to allow for cases where a complete definition
3205 of the type might not be available in all compilation units.}
3207 \textit{If a type definition contains the definition of a member function,
3208 it cannot be moved as is into a type unit, because the member function
3209 contains attributes that are unique to that compilation unit.
3210 Such a type definition can be moved to a type unit by rewriting the DIE tree,
3211 moving the member function declaration into a separate declaration tree,
3212 and replacing the function definition in the type with a non-defining
3213 declaration of the function (as if the function had been defined out of
3216 An example that illustrates the computation of an \MDfive{} hash may be found in
3217 Appendix \refersec{app:usingtypeunits}.