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 (\DWTAGlouserMARK\DWTAGhiuserMARK{} DW\_TAG,
27 \DWATlouserMARK\DWAThiuserMARK{} DW\_AT,
28 \DWENDlouserMARK\DWENDhiuserMARK{} DW\_END,
29 \DWATElouserMARK\DWATEhiuserMARK{} DW\_ATE,
30 \DWOPlouserMARK\DWOPhiuserMARK{} DW\_OP,
31 \DWLANGlouserMARK\DWLANGhiuserMARK{}DW\_LANG,
32 \DWLNElouserMARK\DWLNEhiuserMARK{} DW\_LNE,
33 \DWCClouserMARK\DWCChiuserMARK{} DW\_CC or
34 \DWCFAlouserMARK\DWCFAhiuserMARK{} DW\_CFA
35 respectively) followed by
36 \_lo\_user or \_hi\_user.
37 \textit{For example, for DIE tags, the special
38 labels are \DWTAGlouserNAME{} and \DWTAGhiuserNAME.}
39 Values in the range between \textit{prefix}\_lo\_user
40 and \textit{prefix}\_hi\_user inclusive,
41 are reserved for vendor specific extensions. Vendors may
42 use values in this range without conflicting with current or
43 future system\dash defined values. All other values are reserved
44 for use by the system.
46 \textit{There may also be codes for vendor specific extensions
47 between the number of standard line number opcodes and
48 the first special line number opcode. However, since the
49 number of standard opcodes varies with the DWARF version,
50 the range for extensions is also version dependent. Thus,
51 \DWLNSlouserTARG{} and
52 \DWLNShiuserTARG{} symbols are not defined.
55 Vendor defined tags, attributes, base type encodings, location
56 atoms, language names, line number actions, calling conventions
57 and call frame instructions, conventionally use the form
58 \text{prefix\_vendor\_id\_name}, where
59 \textit{vendor\_id}\addtoindexx{vendor id} is some identifying
60 character sequence chosen so as to avoid conflicts with
63 To ensure that extensions added by one vendor may be safely
64 ignored by consumers that do not understand those extensions,
65 the following rules should be followed:
66 \begin{enumerate}[1. ]
68 \item New attributes should be added in such a way that a
69 debugger may recognize the format of a new attribute value
70 without knowing the content of that attribute value.
72 \item The semantics of any new attributes should not alter
73 the semantics of previously existing attributes.
75 \item The semantics of any new tags should not conflict with
76 the semantics of previously existing tags.
78 \item Do not add any new forms of attribute value.
83 \section{Reserved Values}
84 \label{datarep:reservedvalues}
85 \subsection{Error Values}
86 \label{datarep:errorvalues}
87 \addtoindexx{reserved values!error}
90 \addtoindexx{error value}
91 a convenience for consumers of DWARF information, the value
92 0 is reserved in the encodings for attribute names, attribute
93 forms, base type encodings, location operations, languages,
94 line number program opcodes, macro information entries and tag
95 names to represent an error condition or unknown value. DWARF
96 does not specify names for these reserved values, since they
97 do not represent valid encodings for the given type and should
98 not appear in DWARF debugging information.
101 \subsection{Initial Length Values}
102 \label{datarep:initiallengthvalues}
103 \addtoindexx{reserved values!initial length}
105 An \livetarg{datarep:initiallengthvalues}{initial length field} is one of the length fields that occur
107 of those DWARF sections that
113 \dotdebugpubnames{}, and
114 \dotdebugpubtypes{}) or the length field
115 that occurs at the beginning of the CIE and FDE structures
116 in the \dotdebugframe{} section.
118 In an \addtoindex{initial length field}, the values \wfffffffzero through
119 \wffffffff are reserved by DWARF to indicate some form of
120 extension relative to \addtoindex{DWARF Version 2}; such values must not
121 be interpreted as a length field. The use of one such value,
122 \xffffffff, is defined below
123 (see Section \refersec{datarep:32bitand64bitdwarfformats});
125 the other values is reserved for possible future extensions.
129 \section{Relocatable, Split, Executable and Shared Objects}
130 \label{datarep:executableobjectsandsharedobjects}
132 \subsection{Relocatable Objects}
134 \subsection{Split DWARF Objects}
135 \label{datarep:splitdwarfobjects}
136 A DWARF producer may partition the debugging
137 information such that the majority of the debugging
138 information can remain in individual object files without
139 being processed by the linker. The first partition contains
140 debugging information that must still be processed by the linker,
141 and includes the following:
144 The line number tables, range tables, frame tables, and
145 accelerated access tables, in the usual sections:
146 \dotdebugline, \dotdebugranges, \dotdebugframe,
147 \dotdebugpubnames, \dotdebugpubtypes{} and \dotdebugaranges,
150 An address table, in the \dotdebugaddr{} section. This table
151 contains all addresses and constants that require
152 link-time relocation, and items in the table can be
153 referenced indirectly from the debugging information via
154 the \DWFORMaddrx{} form, and by the \DWOPaddrx{} and
155 \DWOPconstx{} operators.
157 A skeleton compilation unit, as described in Section
158 \refersec{chap:skeletoncompilationunitentries},
159 in the \dotdebuginfo{} section.
161 An abbreviations table for the skeleton compilation unit,
162 in the \dotdebugabbrev{} section.
164 A string table, in the \dotdebugstr{} section. The string
165 table is necessary only if the skeleton compilation unit
166 uses either indirect string form, \DWFORMstrp{} or
169 A string offsets table, in the \dotdebugstroffsets{}
170 section. The string offsets table is necessary only if
171 the skeleton compilation unit uses the \DWFORMstrx{} form.
173 The attributes contained in the skeleton compilation
174 unit can be used by a DWARF consumer to find the object file
175 or DWARF object file that contains the second partition.
177 The second partition contains the debugging information that
178 does not need to be processed by the linker. These sections
179 may be left in the object files and ignored by the linker
180 (that is, not combined and copied to the executable object), or
181 they may be placed by the producer in a separate DWARF object
182 file. This partition includes the following:
185 The full compilation unit, in the \dotdebuginfodwo{} section.
186 Attributes in debugging information entries may refer to
187 machine addresses indirectly using the \DWFORMaddrx{} form,
188 and location expressions may do so using the \DWOPaddrx{} and
189 \DWOPconstx{} forms. Attributes may refer to range table
190 entries with an offset relative to a base offset in the
191 range table for the compilation unit.
193 \item Separate type units, in the \dotdebugtypesdwo{} section.
196 Abbreviations table(s) for the compilation unit and type
197 units, in the \dotdebugabbrevdwo{} section.
199 \item Location lists, in the \dotdebuglocdwo{} section.
202 A skeleton line table (for the type units), in the
203 \dotdebuglinedwo{} section (see
204 Section \refersec{chap:skeletoncompilationunitentries}).
206 \item Macro information, in the \dotdebugmacinfodwo{} section.
208 \item A string table, in the \dotdebugstrdwo{} section.
210 \item A string offsets table, in the \dotdebugstroffsetsdwo{}
214 Except where noted otherwise, all references in this document
215 to a debugging information section (for example, \dotdebuginfo),
216 applies also to the corresponding split DWARF section (for example,
219 \subsection{Executable Objects}
220 \label{chap:executableobjects}
221 The relocated addresses in the debugging information for an
222 executable object are virtual addresses.
224 \subsection{Shared Objects}
225 \label{datarep:sharedobjects}
227 addresses in the debugging information for a shared object
228 are offsets relative to the start of the lowest region of
229 memory loaded from that shared object.
231 \textit{This requirement makes the debugging information for
232 shared objects position independent. Virtual addresses in a
233 shared object may be calculated by adding the offset to the
234 base address at which the object was attached. This offset
235 is available in the run\dash time linker\textquoteright s data structures.}
239 \section{32-Bit and 64-Bit DWARF Formats}
240 \label{datarep:32bitand64bitdwarfformats}
241 \hypertarget{datarep:xxbitdwffmt}{}
242 \addtoindexx{32-bit DWARF format}
243 \addtoindexx{64-bit DWARF format}
244 There are two closely related file formats. In the 32\dash bit DWARF
245 format, all values that represent lengths of DWARF sections
246 and offsets relative to the beginning of DWARF sections are
247 represented using 32\dash bits. In the 64\dash bit DWARF format, all
248 values that represent lengths of DWARF sections and offsets
249 relative to the beginning of DWARF sections are represented
250 using 64\dash bits. A special convention applies to the initial
251 length field of certain DWARF sections, as well as the CIE and
252 FDE structures, so that the 32\dash bit and 64\dash bit DWARF formats
253 can coexist and be distinguished within a single linked object.
255 The differences between the 32\dash\ and 64\dash bit
257 detailed in the following:
258 \begin{enumerate}[1. ]
260 \item In the 32\dash bit DWARF format, an
261 \addtoindex{initial length field}
263 \addtoindexx{initial length field!encoding}
264 Section \refersec{datarep:initiallengthvalues})
265 is an unsigned 32\dash bit integer (which
266 must be less than \xfffffffzero); in the 64\dash bit DWARF format,
267 an \addtoindex{initial length field} is 96 bits in size,
270 \item The first 32\dash bits have the value \xffffffff.
272 \item The following 64\dash bits contain the actual length
273 represented as an unsigned 64\dash bit integer.
276 \textit{This representation allows a DWARF consumer to dynamically
277 detect that a DWARF section contribution is using the 64\dash bit
278 format and to adapt its processing accordingly.}
280 \item Section offset and section length
281 \hypertarget{datarep:sectionoffsetlength}{}
282 \addtoindexx{section length!use in headers}
284 \addtoindexx{section offset!use in headers}
285 in the headers of DWARF sections (other
286 \addtoindexx{initial length field}
288 \addtoindex{initial length}
289 fields) are listed following. In the 32\dash bit DWARF format these
290 are 32\dash bit unsigned integer values; in the 64\dash bit DWARF format,
292 \addtoindexx{section length!in .debug\_aranges header}
294 \addtoindexx{section length!in .debug\_pubnames header}
296 \addtoindexx{section length!in .debug\_pubtypes header}
297 unsigned integer values.
301 Section &Name & Role \\ \hline
302 \dotdebugaranges{} & \addtoindex{debug\_info\_offset} & offset in \dotdebuginfo{} \\
303 \dotdebugframe{}/CIE & \addtoindex{CIE\_id} & CIE distinguished value \\
304 \dotdebugframe{}/FDE & \addtoindex{CIE\_pointer} & offset in \dotdebugframe{} \\
305 \dotdebuginfo{} & \addtoindex{debug\_abbrev\_offset} & offset in \dotdebugabbrev{} \\
306 \dotdebugline{} & \addtoindex{header\_length} & length of header itself \\
307 \dotdebugpubnames{} & \addtoindex{debug\_info\_offset} & offset in \dotdebuginfo{} \\
308 & \addtoindex{debug\_info\_length} & length of \dotdebuginfo{} \\
310 \dotdebugpubtypes{} & \addtoindex{debug\_info\_offset} & offset in \dotdebuginfo{} \\
311 & \addtoindex{debug\_info\_length} & length of \dotdebuginfo{} \\
313 \dotdebugtypes{} & \addtoindex{debug\_abbrev\_offset} & offset in \dotdebugabbrev{} \\
314 & \addtoindex{type\_offset} & offset in \dotdebugtypes{} \\
319 The \texttt{CIE\_id} field in a CIE structure must be 64 bits because
320 it overlays the \texttt{CIE\_pointer} in a FDE structure; this implicit
321 union must be accessed to distinguish whether a CIE or FDE is
322 present, consequently, these two fields must exactly overlay
323 each other (both offset and size).
325 \item Within the body of the \dotdebuginfo{} or \dotdebugtypes{}
326 section, certain forms of attribute value depend on the choice
327 of DWARF format as follows. For the 32\dash bit DWARF format,
328 the value is a 32\dash bit unsigned integer; for the 64\dash bit DWARF
329 format, the value is a 64\dash bit unsigned integer.
332 Form & Role \\ \hline
333 \DWFORMrefaddr& offset in \dotdebuginfo{} \\
334 \DWFORMsecoffset& offset in a section other than \\
335 &\dotdebuginfo{} or \dotdebugstr{} \\
336 \DWFORMstrp&offset in \dotdebugstr{} \\
337 \DWOPcallref&offset in \dotdebuginfo{} \\
341 \item Within the body of the \dotdebugpubnames{} and
343 sections, the representation of the first field
344 of each tuple (which represents an offset in the
346 section) depends on the DWARF format as follows: in the
347 32\dash bit DWARF format, this field is a 32\dash bit unsigned integer;
348 in the 64\dash bit DWARF format, it is a 64\dash bit unsigned integer.
351 \item In the body of the \dotdebugstroffsets{} and \dotdebugstroffsetsdwo{}
352 sections, the size of entries in the body depend on the DWARF
353 format as follows: in the 32-bit DWARF format, entries are 32-bit
354 unsigned integer values; in the 64-bit DWARF format, they are
355 64-bit unsigned integers.
357 \item In the body of the \dotdebugaddr{}, \dotdebugloc{} and \dotdebugranges{}
358 sections, the contents of the address size fields depends on the
359 DWARF format as follows: in the 32-bit DWARF format, these fields
360 contain 4; in the 64-bit DWARF format these fields contain 8.
364 The 32\dash bit and 64\dash bit DWARF format conventions must \emph{not} be
365 intermixed within a single compilation unit.
367 \textit{Attribute values and section header fields that represent
368 addresses in the target program are not affected by these
371 A DWARF consumer that supports the 64\dash bit DWARF format must
372 support executables in which some compilation units use the
373 32\dash bit format and others use the 64\dash bit format provided that
374 the combination links correctly (that is, provided that there
375 are no link\dash time errors due to truncation or overflow). (An
376 implementation is not required to guarantee detection and
377 reporting of all such errors.)
379 \textit{It is expected that DWARF producing compilers will \emph{not} use
380 the 64\dash bit format \emph{by default}. In most cases, the division of
381 even very large applications into a number of executable and
382 shared objects will suffice to assure that the DWARF sections
383 within each individual linked object are less than 4 GBytes
384 in size. However, for those cases where needed, the 64\dash bit
385 format allows the unusual case to be handled as well. Even
386 in this case, it is expected that only application supplied
387 objects will need to be compiled using the 64\dash bit format;
388 separate 32\dash bit format versions of system supplied shared
389 executable libraries can still be used.}
393 \section{Format of Debugging Information}
394 \label{datarep:formatofdebugginginformation}
396 For each compilation unit compiled with a DWARF producer,
397 a contribution is made to the \dotdebuginfo{} section of
398 the object file. Each such contribution consists of a
399 compilation unit header
400 (see Section \refersec{datarep:compilationunitheader})
402 single \DWTAGcompileunit{} or
403 \DWTAGpartialunit{} debugging
404 information entry, together with its children.
406 For each type defined in a compilation unit, a contribution may
407 be made to the \dotdebugtypes{}
408 section of the object file. Each
409 such contribution consists of a
410 \addtoindex{type unit} header
411 (see Section \refersec{datarep:typeunitheader})
412 followed by a \DWTAGtypeunit{} entry, together with
415 Each debugging information entry begins with a code that
416 represents an entry in a separate
417 \addtoindex{abbreviations table}. This
418 code is followed directly by a series of attribute values.
420 The appropriate entry in the
421 \addtoindex{abbreviations table} guides the
422 interpretation of the information contained directly in the
424 \dotdebugtypes{} section.
427 Multiple debugging information entries may share the same
428 abbreviation table entry. Each compilation unit is associated
429 with a particular abbreviation table, but multiple compilation
430 units may share the same table.
431 \subsection{Unit Headers}
432 \label{datarep:unitheaders}
434 \subsubsection{Compilation Unit Header}
435 \label{datarep:compilationunitheader}
436 \begin{enumerate}[1. ]
438 \item \texttt{unit\_length} (\livelink{datarep:initiallengthvalues}{initial length}) \\
439 \addttindexx{unit\_length}
440 A 4\dash byte or 12\dash byte
441 \addtoindexx{initial length}
442 unsigned integer representing the length
443 of the \dotdebuginfo{}
444 contribution for that compilation unit,
445 not including the length field itself. In the \thirtytwobitdwarfformat,
446 this is a 4\dash byte unsigned integer (which must be less
447 than \xfffffffzero); in the \sixtyfourbitdwarfformat, this consists
448 of the 4\dash byte value \wffffffff followed by an 8\dash byte unsigned
449 integer that gives the actual length
450 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
453 \item \texttt{version} (\addtoindex{uhalf}) \\
454 A 2\dash byte unsigned integer representing the version of the
455 DWARF information for the compilation unit \addtoindexx{version number!compilation unit}
456 (see Appendix \refersec{app:dwarfsectionversionnumbersinformative}).
457 The value in this field is 4.
459 \item \addttindex{debug\_abbrev\_offset} (\livelink{datarep:sectionoffsetlength}{section offset}) \\
461 \addtoindexx{section offset!in .debug\_info header}
462 4\dash byte or 8\dash byte unsigned offset into the
464 section. This offset associates the compilation unit with a
465 particular set of debugging information entry abbreviations. In
466 the \thirtytwobitdwarfformat, this is a 4\dash byte unsigned length;
467 in the \sixtyfourbitdwarfformat, this is an 8\dash byte unsigned length
468 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
470 \item \texttt{address\_size} (\addtoindex{ubyte}) \\
471 A 1\dash byte unsigned integer representing the size in bytes of
472 \addttindexx{address\_size}
473 an address on the target architecture. If the system uses
474 \addtoindexx{address space!segmented}
475 segmented addressing, this value represents the size of the
476 offset portion of an address.
481 \subsubsection{Type Unit Header}
482 \label{datarep:typeunitheader}
484 The header for the series of debugging information entries
485 contributing to the description of a type that has been
486 placed in its own \addtoindex{type unit}, within the
487 \dotdebugtypes{} section,
488 consists of the following information:
489 \begin{enumerate}[1. ]
491 \item \texttt{unit\_length} (\livelink{datarep:initiallengthvalues}{initial length}) \\
492 \addttindexx{unit\_length}
493 A 4\dash byte or 12\dash byte unsigned integer
494 \addtoindexx{initial length}
495 representing the length
496 of the \dotdebugtypes{} contribution for that type unit,
497 not including the length field itself. In the \thirtytwobitdwarfformat,
498 this is a 4\dash byte unsigned integer (which must be
499 less than \xfffffffzero); in the \sixtyfourbitdwarfformat, this
500 consists of the 4\dash byte value \wffffffff followed by an
501 8\dash byte unsigned integer that gives the actual length
502 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
505 \item \texttt{version} (\addtoindex{uhalf}) \\
506 A 2\dash byte unsigned integer representing the version of the
507 DWARF information for the
508 type unit\addtoindexx{version number!type unit}
509 (see Appendix \refersec{app:dwarfsectionversionnumbersinformative}).
510 The value in this field is 4.
512 \item \addttindex{debug\_abbrev\_offset} (\livelink{datarep:sectionoffsetlength}{section offset}) \\
514 \addtoindexx{section offset!in .debug\_types header}
515 4\dash byte or 8\dash byte unsigned offset into the
517 section. This offset associates the type unit with a
518 particular set of debugging information entry abbreviations. In
519 the \thirtytwobitdwarfformat, this is a 4\dash byte unsigned length;
520 in the \sixtyfourbitdwarfformat, this is an 8\dash byte unsigned length
521 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
523 \item \texttt{address\_size} (ubyte) \\
524 A 1\dash byte unsigned integer representing the size
525 \addtoindexx{size of an address}
527 \addttindexx{address\_size}
528 an address on the target architecture. If the system uses
529 \addtoindexx{address space!segmented}
530 segmented addressing, this value represents the size of the
531 offset portion of an address.
533 \item \texttt{type\_signature} (8\dash byte unsigned integer) \\
534 \addtoindexx{type signature}
536 \addttindexx{type\_signature}
537 64\dash bit unique signature (see Section
538 \refersec{datarep:typesignaturecomputation})
539 of the type described in this type
542 \textit{An attribute that refers (using
543 \DWFORMrefsigeight{}) to
544 the primary type contained in this
545 \addtoindex{type unit} uses this value.}
547 \item \texttt{type\_offset} (\livelink{datarep:sectionoffsetlength}{section offset}) \\
548 \addttindexx{type\_offset}
549 A 4\dash byte or 8\dash byte unsigned offset
550 \addtoindexx{section offset!in .debug\_types header}
551 relative to the beginning
552 of the \addtoindex{type unit} header.
553 This offset refers to the debugging
554 information entry that describes the type. Because the type
555 may be nested inside a namespace or other structures, and may
556 contain references to other types that have not been placed in
557 separate type units, it is not necessarily either the first or
558 the only entry in the type unit. In the \thirtytwobitdwarfformat,
559 this is a 4\dash byte unsigned length; in the \sixtyfourbitdwarfformat,
560 this is an 8\dash byte unsigned length
561 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
565 \subsection{Debugging Information Entry}
566 \label{datarep:debugginginformationentry}
568 Each debugging information entry begins with an unsigned LEB128
569 number containing the abbreviation code for the entry. This
570 code represents an entry within the abbreviations table
571 associated with the compilation unit containing this entry. The
572 abbreviation code is followed by a series of attribute values.
574 On some architectures, there are alignment constraints on
575 section boundaries. To make it easier to pad debugging
576 information sections to satisfy such constraints, the
577 abbreviation code 0 is reserved. Debugging information entries
578 consisting of only the abbreviation code 0 are considered
581 \subsection{Abbreviations Tables}
582 \label{datarep:abbreviationstables}
584 The abbreviations tables for all compilation units
585 are contained in a separate object file section called
587 As mentioned before, multiple compilation
588 units may share the same abbreviations table.
590 The abbreviations table for a single compilation unit consists
591 of a series of abbreviation declarations. Each declaration
592 specifies the tag and attributes for a particular form of
593 debugging information entry. Each declaration begins with
594 an unsigned LEB128 number representing the abbreviation
595 code itself. It is this code that appears at the beginning
596 of a debugging information entry in the
599 section. As described above, the abbreviation
600 code 0 is reserved for null debugging information entries. The
601 abbreviation code is followed by another unsigned LEB128
602 number that encodes the entry\textquoteright s tag. The encodings for the
603 tag names are given in
604 Table \refersec{tab:tagencodings}.
607 \setlength{\extrarowheight}{0.1cm}
608 \begin{longtable}{l|l}
610 \caption{Tag encodings} \label{tab:tagencodings} \\
611 \hline \bfseries Tag name&\bfseries Value\\ \hline
613 \bfseries Tag name&\bfseries Value \\ \hline
615 \hline \emph{Continued on next page}
617 \hline \ddag\ \textit{New in DWARF Version 5}
619 \DWTAGarraytype{} &0x01 \\
620 \DWTAGclasstype&0x02 \\
621 \DWTAGentrypoint&0x03 \\
622 \DWTAGenumerationtype&0x04 \\
623 \DWTAGformalparameter&0x05 \\
624 \DWTAGimporteddeclaration&0x08 \\
626 \DWTAGlexicalblock&0x0b \\
628 \DWTAGpointertype&0x0f \\
629 \DWTAGreferencetype&0x10 \\
630 \DWTAGcompileunit&0x11 \\
631 \DWTAGstringtype&0x12 \\
632 \DWTAGstructuretype&0x13 \\
633 \DWTAGsubroutinetype&0x15 \\
634 \DWTAGtypedef&0x16 \\
635 \DWTAGuniontype&0x17 \\
636 \DWTAGunspecifiedparameters&0x18 \\
637 \DWTAGvariant&0x19 \\
638 \DWTAGcommonblock&0x1a \\
639 \DWTAGcommoninclusion&0x1b \\
640 \DWTAGinheritance&0x1c \\
641 \DWTAGinlinedsubroutine&0x1d \\
643 \DWTAGptrtomembertype&0x1f \\
644 \DWTAGsettype&0x20 \\
645 \DWTAGsubrangetype&0x21 \\
646 \DWTAGwithstmt&0x22 \\
647 \DWTAGaccessdeclaration&0x23 \\
648 \DWTAGbasetype&0x24 \\
649 \DWTAGcatchblock&0x25 \\
650 \DWTAGconsttype&0x26 \\
651 \DWTAGconstant&0x27 \\
652 \DWTAGenumerator&0x28 \\
653 \DWTAGfiletype&0x29 \\
655 \DWTAGnamelist&0x2b \\
656 \DWTAGnamelistitem&0x2c \\
657 \DWTAGpackedtype&0x2d \\
658 \DWTAGsubprogram&0x2e \\
659 \DWTAGtemplatetypeparameter&0x2f \\
660 \DWTAGtemplatevalueparameter&0x30 \\
661 \DWTAGthrowntype&0x31 \\
662 \DWTAGtryblock&0x32 \\
663 \DWTAGvariantpart&0x33 \\
664 \DWTAGvariable&0x34 \\
665 \DWTAGvolatiletype&0x35 \\
666 \DWTAGdwarfprocedure&0x36 \\
667 \DWTAGrestricttype&0x37 \\
668 \DWTAGinterfacetype&0x38 \\
669 \DWTAGnamespace&0x39 \\
670 \DWTAGimportedmodule&0x3a \\
671 \DWTAGunspecifiedtype&0x3b \\
672 \DWTAGpartialunit&0x3c \\
673 \DWTAGimportedunit&0x3d \\
674 \DWTAGcondition&\xiiif \\
675 \DWTAGsharedtype&0x40 \\
676 \DWTAGtypeunit{} &0x41 \\
677 \DWTAGrvaluereferencetype{} &0x42 \\
678 \DWTAGtemplatealias{} &0x43 \\
679 \DWTAGcoarraytype~\ddag &0x44 \\
680 \DWTAGgenericsubrange~\ddag &0x45 \\
681 \DWTAGdynamictype~\ddag & 0x46 \\
682 \DWTAGatomictype~\ddag & 0x47 \\
683 \DWTAGlouser&0x4080 \\
684 \DWTAGhiuser&\xffff \\
688 Following the tag encoding is a 1\dash byte value that determines
689 whether a debugging information entry using this abbreviation
690 has child entries or not. If the value is
692 the next physically succeeding entry of any debugging
693 information entry using this abbreviation is the first
694 child of that entry. If the 1\dash byte value following the
695 abbreviation\textquoteright s tag encoding is
696 \DWCHILDRENnoTARG, the next
697 physically succeeding entry of any debugging information entry
698 using this abbreviation is a sibling of that entry. (Either
699 the first child or sibling entries may be null entries). The
700 encodings for the child determination byte are given in
701 Table \refersec{tab:childdeterminationencodings}
703 Section \refersec{chap:relationshipofdebugginginformationentries},
704 each chain of sibling entries is terminated by a null entry.)
708 \setlength{\extrarowheight}{0.1cm}
709 \begin{longtable}{l|l}
710 \caption{Child determination encodings}
711 \label{tab:childdeterminationencodings}
712 \addtoindexx{Child determination encodings} \\
713 \hline \bfseries Children determination name&\bfseries Value \\ \hline
715 \bfseries Children determination name&\bfseries Value \\ \hline
717 \hline \emph{Continued on next page}
721 \DWCHILDRENno&0x00 \\
722 \DWCHILDRENyes&0x01 \\ \hline
727 Finally, the child encoding is followed by a series of
728 attribute specifications. Each attribute specification
729 consists of two parts. The first part is an unsigned LEB128
730 number representing the attribute\textquoteright s name. The second part
731 is an unsigned LEB128 number representing the attribute\textquoteright s
732 form. The series of attribute specifications ends with an
733 entry containing 0 for the name and 0 for the form.
736 \DWFORMindirectTARG{} is a special case. For
737 attributes with this form, the attribute value itself in the
740 section begins with an unsigned
741 LEB128 number that represents its form. This allows producers
742 to choose forms for particular attributes
743 \addtoindexx{abbreviations table!dynamic forms in}
745 without having to add a new entry to the abbreviations table.
747 The abbreviations for a given compilation unit end with an
748 entry consisting of a 0 byte for the abbreviation code.
751 Appendix \refersec{app:compilationunitsandabbreviationstableexample}
752 for a depiction of the organization of the
753 debugging information.}
756 \subsection{Attribute Encodings}
757 \label{datarep:attributeencodings}
759 The encodings for the attribute names are given in
760 Table \refersec{tab:attributeencodings}.
762 The attribute form governs how the value of the attribute is
763 encoded. There are nine classes of form, listed below. Each
764 class is a set of forms which have related representations
765 and which are given a common interpretation according to the
766 attribute in which the form is used.
768 Form \DWFORMsecoffsetTARG{}
770 \addtoindexx{rangelistptr class}
772 \addtoindexx{macptr class}
774 \addtoindexx{loclistptr class}
776 \addtoindexx{lineptr class}
777 namely \livelink{chap:classlineptr}{lineptr},
778 \livelink{chap:classloclistptr}{loclistptr},
779 \livelink{chap:classmacptr}{macptr} or
780 \livelink{chap:classrangelistptr}{rangelistptr}; the list
781 of classes allowed by the applicable attribute in
782 Table \refersec{tab:attributeencodings}
783 determines the class of the form.
785 \textit{In DWARF V3 the forms \DWFORMdatafour{} and
786 \DWFORMdataeight{} were
787 \addtoindexx{lineptr class}
789 \addtoindexx{rangelistptr class}
791 \addtoindexx{macptr class}
793 \addtoindexx{loclistptr class}
794 class constant \addtoindexx{constant class}
795 or one of the classes
796 \livelink{chap:classlineptr}{lineptr},
797 \livelink{chap:classloclistptr}{loclistptr},
798 \livelink{chap:classmacptr}{macptr} or
799 \livelink{chap:classrangelistptr}{rangelistptr}, depending on context. In
801 \DWFORMdatafour{} and
802 \DWFORMdataeight{} are members of class
803 constant in all cases.
805 \DWFORMsecoffset{} replaces
806 their usage for the other classes.}
809 Each possible form belongs to one or more of the following classes:
812 \item \livelinki{chap:classaddress}{address}{address class} \\
813 \livetarg{datarep:classaddress}{}
814 Represented as either:
816 \item An object of appropriate size to hold an
817 address on the target machine
819 The size is encoded in the compilation unit header
820 (see Section \refersec{datarep:compilationunitheader}).
821 This address is relocatable in a relocatable object file and
822 is relocated in an executable file or shared object.
824 \item An indirect index into a table of addresses (as
825 described in the previous bullet) in the
826 \dotdebugaddr{} section (\DWFORMaddrxTARG).
827 The representation of a \DWFORMaddrxNAME{} value is an unsigned
828 \addtoindex{LEB128} value, which is interpreted as a zero-based
829 index into an array of addresses in the \dotdebugaddr{} section.
830 The index is relative to the value of the \DWATaddrbase{} attribute
831 of the associated compilation unit.
835 \item \livelink{chap:classblock}{block} \\
836 \livetarg{datarep:classblock}{}
837 Blocks come in four forms:
839 \begin{myindentpara}{1cm}
840 A 1\dash byte length followed by 0 to 255 contiguous information
841 bytes (\DWFORMblockoneTARG).
844 \begin{myindentpara}{1cm}
845 A 2\dash byte length followed by 0 to 65,535 contiguous information
846 bytes (\DWFORMblocktwoTARG).
849 \begin{myindentpara}{1cm}
850 A 4\dash byte length followed by 0 to 4,294,967,295 contiguous
851 information bytes (\DWFORMblockfourTARG).
854 \begin{myindentpara}{1cm}
855 An unsigned LEB128 length followed by the number of bytes
856 specified by the length (\DWFORMblockTARG).
859 In all forms, the length is the number of information bytes
860 that follow. The information bytes may contain any mixture
861 of relocated (or relocatable) addresses, references to other
862 debugging information entries or data bytes.
864 \item \livelinki{chap:classconstant}{constant}{constant class} \\
865 \livetarg{datarep:classconstant}{}
866 There are six forms of constants. There are fixed length
867 constant data forms for one, two, four and eight byte values
872 and \DWFORMdataeightTARG).
873 There are also variable length constant
874 data forms encoded using LEB128 numbers (see below). Both
875 signed (\DWFORMsdataTARG) and unsigned
876 (\DWFORMudataTARG) variable
877 length constants are available
880 The data in \DWFORMdataone,
882 \DWFORMdatafour{} and
884 can be anything. Depending on context, it may
885 be a signed integer, an unsigned integer, a floating\dash point
886 constant, or anything else. A consumer must use context to
887 know how to interpret the bits, which if they are target
888 machine data (such as an integer or floating point constant)
889 will be in target machine byte\dash order.
891 \textit{If one of the \DWFORMdataTARG\textless n\textgreater
892 forms is used to represent a
893 signed or unsigned integer, it can be hard for a consumer
894 to discover the context necessary to determine which
895 interpretation is intended. Producers are therefore strongly
896 encouraged to use \DWFORMsdata{} or
897 \DWFORMudata{} for signed and
898 unsigned integers respectively, rather than
899 \DWFORMdata\textless n\textgreater.}
902 \item \livelinki{chap:classexprloc}{exprloc}{exprloc class} \\
903 \livetarg{datarep:classexprloc}{}
904 This is an unsigned LEB128 length followed by the
905 number of information bytes specified by the length
906 (\DWFORMexprlocTARG).
907 The information bytes contain a DWARF expression
908 (see Section \refersec{chap:dwarfexpressions})
909 or location description
910 (see Section \refersec{chap:locationdescriptions}).
912 \item \livelinki{chap:classflag}{flag}{flag class} \\
913 \livetarg{datarep:classflag}{}
914 A flag \addtoindexx{flag class}
915 is represented explicitly as a single byte of data
917 implicitly (\DWFORMflagpresentTARG).
919 first case, if the \nolink{flag} has value zero, it indicates the
920 absence of the attribute; if the \nolink{flag} has a non\dash zero value,
921 it indicates the presence of the attribute. In the second
922 case, the attribute is implicitly indicated as present, and
923 no value is encoded in the debugging information entry itself.
925 \item \livelinki{chap:classlineptr}{lineptr}{lineptr class} \\
926 \livetarg{datarep:classlineptr}{}
927 This is an offset into
928 \addtoindexx{section offset!in class lineptr value}
930 \dotdebugline{} or \dotdebuglinedwo{} section
932 It consists of an offset from the beginning of the
934 section to the first byte of
935 the data making up the line number list for the compilation
937 It is relocatable in a relocatable object file, and
938 relocated in an executable or shared object. In the
939 \thirtytwobitdwarfformat, this offset is a 4\dash byte unsigned value;
940 in the \sixtyfourbitdwarfformat, it is an 8\dash byte unsigned value
941 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
944 \item \livelinki{chap:classloclistptr}{loclistptr}{loclistptr class} \\
945 \livetarg{datarep:classloclistptr}{}
946 This is an offset into the
950 It consists of an offset from the
951 \addtoindexx{section offset!in class loclistptr value}
954 section to the first byte of
955 the data making up the
956 \addtoindex{location list} for the compilation unit.
957 It is relocatable in a relocatable object file, and
958 relocated in an executable or shared object. In the
959 \thirtytwobitdwarfformat, this offset is a 4\dash byte unsigned value;
960 in the \sixtyfourbitdwarfformat, it is an 8\dash byte unsigned value
961 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
964 \item \livelinki{chap:classmacptr}{macptr}{macptr class} \\
965 \livetarg{datarep:classmacptr}{}
967 \addtoindexx{section offset!in class macptr value}
969 \dotdebugmacinfo{} or \dotdebugmacinfo{} section
971 It consists of an offset from the
972 beginning of the \dotdebugmacinfo{}
973 section to the first byte of
974 the data making up the macro information list for the compilation
976 It is relocatable in a relocatable object file, and
977 relocated in an executable or shared object. In the
978 \thirtytwobitdwarfformat, this offset is a 4\dash byte unsigned value;
979 in the \sixtyfourbitdwarfformat, it is an 8\dash byte unsigned value
980 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
983 \item \livelinki{chap:classrangelistptr}{rangelistptr}{rangelistptr class} \\
984 \livetarg{datarep:classrangelistptr}{}
986 \addtoindexx{section offset!in class rangelistptr value}
987 offset into the \dotdebugranges{} section
990 offset from the beginning of the
991 \dotdebugranges{} section
992 to the beginning of the non\dash contiguous address ranges
993 information for the referencing entity.
995 a relocatable object file, and relocated in an executable or
996 shared object. In the \thirtytwobitdwarfformat, this offset
997 is a 4\dash byte unsigned value; in the 64\dash bit DWARF
998 format, it is an 8\dash byte unsigned value (see Section
999 \refersec{datarep:32bitand64bitdwarfformats}).
1002 \textit{Because classes \livelink{chap:classlineptr}{lineptr},
1003 \livelink{chap:classloclistptr}{loclistptr},
1004 \livelink{chap:classmacptr}{macptr} and
1005 \livelink{chap:classrangelistptr}{rangelistptr}
1006 share a common representation, it is not possible for an
1007 attribute to allow more than one of these classes}
1011 \item \livelinki{chap:classreference}{reference}{reference class} \\
1012 \livetarg{datarep:classreference}{}
1013 There are three types of reference.
1016 \addtoindexx{reference class}
1017 first type of reference can identify any debugging
1018 information entry within the containing unit.
1021 \addtoindexx{section offset!in class reference value}
1022 offset from the first byte of the compilation
1023 header for the compilation unit containing the reference. There
1024 are five forms for this type of reference. There are fixed
1025 length forms for one, two, four and eight byte offsets
1031 and \DWFORMrefeightTARG).
1032 There is also an unsigned variable
1033 length offset encoded form that uses unsigned LEB128 numbers
1034 (\DWFORMrefudataTARG).
1035 Because this type of reference is within
1036 the containing compilation unit no relocation of the value
1039 The second type of reference can identify any debugging
1040 information entry within a
1041 \dotdebuginfo{} section; in particular,
1042 it may refer to an entry in a different compilation unit
1043 from the unit containing the reference, and may refer to an
1044 entry in a different shared object. This type of reference
1045 (\DWFORMrefaddrTARG)
1046 is an offset from the beginning of the
1048 section of the target executable or shared object;
1049 it is relocatable in a relocatable object file and frequently
1050 relocated in an executable file or shared object. For
1051 references from one shared object or static executable file
1052 to another, the relocation and identification of the target
1053 object must be performed by the consumer. In the
1054 \thirtytwobitdwarfformat, this offset is a 4\dash byte unsigned value;
1055 in the \sixtyfourbitdwarfformat, it is an 8\dash byte
1057 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
1059 \textit{A debugging information entry that may be referenced by
1060 another compilation unit using
1061 \DWFORMrefaddr{} must have a
1062 global symbolic name.}
1064 \textit{For a reference from one executable or shared object to
1065 another, the reference is resolved by the debugger to identify
1066 the shared object or executable and the offset into that
1067 object\textquoteright s \dotdebuginfo{}
1068 section in the same fashion as the run
1069 time loader, either when the debug information is first read,
1070 or when the reference is used.}
1072 The third type of reference can identify any debugging
1073 information type entry that has been placed in its own
1074 \addtoindex{type unit}. This type of
1075 reference (\DWFORMrefsigeightTARG) is the
1076 \addtoindexx{type signature}
1077 64\dash bit type signature
1078 (see Section \refersec{datarep:typesignaturecomputation})
1082 \textit{The use of compilation unit relative references will reduce the
1083 number of link\dash time relocations and so speed up linking. The
1084 use of the second and third type of reference allows for the
1085 sharing of information, such as types, across compilation
1088 \textit{A reference to any kind of compilation unit identifies the
1089 debugging information entry for that unit, not the preceding
1092 \item \livelinki{chap:classstring}{string}{string class} \\
1093 \livetarg{datarep:classstring}{}
1094 A string is a sequence of contiguous non\dash null bytes followed by
1096 \addtoindexx{string class}
1097 A string may be represented:
1099 \item immediately in the debugging information entry itself
1100 (\DWFORMstringTARG),
1102 \addtoindexx{section offset!in class string value}
1103 offset into a string table contained in
1104 the \dotdebugstr{} section of the object file
1106 In the \thirtytwobitdwarfformat, the representation of a
1108 value is a 4\dash byte unsigned offset; in the \sixtyfourbitdwarfformat,
1109 it is an 8\dash byte unsigned offset
1110 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
1111 \item as an indirect offset into the string table using an
1112 index into a table of offsets contained in the
1113 \dotdebugstroffsets{} section of the object file (\DWFORMstrxTARG).
1114 The representation of a \DWFORMstrxNAME{} value is an unsigned
1115 \addtoindex{LEB128} value, which is interpreted as a zero-based
1116 index into an array of offsets in the \dotdebugstroffsets{} section.
1117 The offset entries in the \dotdebugstroffsets{} section have the
1118 same representation as \DWFORMstrp{} values.
1120 Any combination of these three forms may be used within a single compilation.
1122 If the \DWATuseUTFeight{}
1123 \addtoindexx{use UTF8 attribute}\addtoindexx{UTF-8} attribute is specified for the
1124 compilation, partial, skeleton or type unit entry, string values are encoded using the
1125 UTF\dash 8 (\addtoindex{Unicode} Transformation Format\dash 8) from the Universal
1126 Character Set standard (ISO/IEC 10646\dash 1:1993). Otherwise,
1127 the string representation is unspecified.
1129 \textit{The \addtoindex{Unicode} Standard Version 3 is fully compatible with
1130 ISO/IEC 10646\dash 1:1993. It contains all the same characters
1131 and encoding points as ISO/IEC 10646, as well as additional
1132 information about the characters and their use.}
1134 \textit{Earlier versions of DWARF did not specify the representation
1135 of strings; for compatibility, this version also does
1136 not. However, the UTF\dash 8 representation is strongly recommended.}
1140 In no case does an attribute use
1141 \addtoindexx{rangelistptr class}
1143 \addtoindexx{loclistptr class}
1145 \addtoindexx{lineptr class}
1147 \addtoindexx{macptr class}
1148 classes \livelink{chap:classlineptr}{lineptr},
1149 \livelink{chap:classloclistptr}{loclistptr}, \livelink{chap:classmacptr}{macptr} or
1150 \livelink{chap:classrangelistptr}{rangelistptr} to point into either the
1151 \dotdebuginfo{} or \dotdebugstr{} section.
1153 The form encodings are listed in
1154 Table \refersec{tab:attributeformencodings}.
1158 \setlength{\extrarowheight}{0.1cm}
1159 \begin{longtable}{l|l|l}
1160 \caption{Attribute encodings}
1161 \label{tab:attributeencodings}
1162 \addtoindexx{attribute encodings} \\
1163 \hline \bfseries Attribute name&\bfseries Value &\bfseries Classes \\ \hline
1165 \bfseries Attribute name&\bfseries Value &\bfseries Classes\\ \hline
1167 \hline \emph{Continued on next page}
1169 \hline \ddag\ \textit{New in DWARF Version 5}
1171 \DWATsibling&0x01&\livelink{chap:classreference}{reference}
1172 \addtoindexx{sibling attribute!encoding} \\
1173 \DWATlocation&0x02&\livelink{chap:classexprloc}{exprloc},
1174 \livelink{chap:classloclistptr}{loclistptr}
1175 \addtoindexx{location attribute!encoding} \\
1176 \DWATname&0x03&\livelink{chap:classstring}{string}
1177 \addtoindexx{name attribute!encoding} \\
1178 \DWATordering&0x09&\livelink{chap:classconstant}{constant}
1179 \addtoindexx{ordering attribute!encoding} \\
1180 \DWATbytesize&0x0b&\livelink{chap:classconstant}{constant},
1181 \livelink{chap:classexprloc}{exprloc},
1182 \livelink{chap:classreference}{reference}
1183 \addtoindexx{byte size attribute!encoding} \\
1184 \DWATbitoffset&0x0c&\livelink{chap:classconstant}{constant},
1185 \livelink{chap:classexprloc}{exprloc},
1186 \livelink{chap:classreference}{reference}
1187 \addtoindexx{bit offset attribute!encoding} \\
1188 \DWATbitsize&0x0d&\livelink{chap:classconstant}{constant},
1189 \livelink{chap:classexprloc}{exprloc},
1190 \livelink{chap:classreference}{reference}
1191 \addtoindexx{bit size attribute!encoding} \\
1192 \DWATstmtlist&0x10&\livelink{chap:classlineptr}{lineptr}
1193 \addtoindexx{statement list attribute!encoding} \\
1194 \DWATlowpc&0x11&\livelink{chap:classaddress}{address}
1195 \addtoindexx{low PC attribute!encoding} \\
1196 \DWAThighpc&0x12&\livelink{chap:classaddress}{address},
1197 \livelink{chap:classconstant}{constant}
1198 \addtoindexx{high PC attribute!encoding} \\
1199 \DWATlanguage&0x13&\livelink{chap:classconstant}{constant}
1200 \addtoindexx{language attribute!encoding} \\
1201 \DWATdiscr&0x15&\livelink{chap:classreference}{reference}
1202 \addtoindexx{discriminant attribute!encoding} \\
1203 \DWATdiscrvalue&0x16&\livelink{chap:classconstant}{constant}
1204 \addtoindexx{discriminant value attribute!encoding} \\
1205 \DWATvisibility&0x17&\livelink{chap:classconstant}{constant}
1206 \addtoindexx{visibility attribute!encoding} \\
1207 \DWATimport&0x18&\livelink{chap:classreference}{reference}
1208 \addtoindexx{import attribute!encoding} \\
1209 \DWATstringlength&0x19&\livelink{chap:classexprloc}{exprloc},
1210 \livelink{chap:classloclistptr}{loclistptr}
1211 \addtoindexx{string length attribute!encoding} \\
1212 \DWATcommonreference&0x1a&\livelink{chap:classreference}{reference}
1213 \addtoindexx{common reference attribute!encoding} \\
1214 \DWATcompdir&0x1b&\livelink{chap:classstring}{string}
1215 \addtoindexx{compilation directory attribute!encoding} \\
1216 \DWATconstvalue&0x1c&\livelink{chap:classblock}{block},
1217 \livelink{chap:classconstant}{constant},
1218 \livelink{chap:classstring}{string}
1219 \addtoindexx{constant value attribute!encoding} \\
1220 \DWATcontainingtype&0x1d&\livelink{chap:classreference}{reference}
1221 \addtoindexx{containing type attribute!encoding} \\
1222 \DWATdefaultvalue&0x1e&\livelink{chap:classconstant}{constant},
1223 \livelink{chap:classreference}{reference},
1224 \livelink{chap:classflag}{flag}
1225 \addtoindexx{default value attribute!encoding} \\
1226 \DWATinline&0x20&\livelink{chap:classconstant}{constant}
1227 \addtoindexx{inline attribute!encoding} \\
1228 \DWATisoptional&0x21&\livelink{chap:classflag}{flag}
1229 \addtoindexx{is optional attribute!encoding} \\
1230 \DWATlowerbound&0x22&\livelink{chap:classconstant}{constant},
1231 \livelink{chap:classexprloc}{exprloc},
1232 \livelink{chap:classreference}{reference}
1233 \addtoindexx{lower bound attribute!encoding} \\
1234 \DWATproducer&0x25&\livelink{chap:classstring}{string}
1235 \addtoindexx{producer attribute!encoding} \\
1236 \DWATprototyped&0x27&\livelink{chap:classflag}{flag}
1237 \addtoindexx{prototyped attribute!encoding} \\
1238 \DWATreturnaddr&0x2a&\livelink{chap:classexprloc}{exprloc},
1239 \livelink{chap:classloclistptr}{loclistptr}
1240 \addtoindexx{return address attribute!encoding} \\
1241 \DWATstartscope&0x2c&\livelink{chap:classconstant}{constant},
1242 \livelink{chap:classrangelistptr}{rangelistptr}
1243 \addtoindexx{start scope attribute!encoding} \\
1244 \DWATbitstride&0x2e&\livelink{chap:classconstant}{constant},
1245 \livelink{chap:classexprloc}{exprloc},
1246 \livelink{chap:classreference}{reference}
1247 \addtoindexx{bit stride attribute!encoding} \\
1248 \DWATupperbound&0x2f&\livelink{chap:classconstant}{constant},
1249 \livelink{chap:classexprloc}{exprloc},
1250 \livelink{chap:classreference}{reference}
1251 \addtoindexx{upper bound attribute!encoding} \\
1252 \DWATabstractorigin&0x31&\livelink{chap:classreference}{reference}
1253 \addtoindexx{abstract origin attribute!encoding} \\
1254 \DWATaccessibility&0x32&\livelink{chap:classconstant}{constant}
1255 \addtoindexx{accessibility attribute!encoding} \\
1256 \DWATaddressclass&0x33&\livelink{chap:classconstant}{constant}
1257 \addtoindexx{address class attribute!encoding} \\
1258 \DWATartificial&0x34&\livelink{chap:classflag}{flag}
1259 \addtoindexx{artificial attribute!encoding} \\
1260 \DWATbasetypes&0x35&\livelink{chap:classreference}{reference}
1261 \addtoindexx{base types attribute!encoding} \\
1262 \DWATcallingconvention&0x36&\livelink{chap:classconstant}{constant}
1263 \addtoindexx{calling convention attribute!encoding} \\
1264 \DWATcount&0x37&\livelink{chap:classconstant}{constant},
1265 \livelink{chap:classexprloc}{exprloc},
1266 \livelink{chap:classreference}{reference}
1267 \addtoindexx{count attribute!encoding} \\
1268 \DWATdatamemberlocation&0x38&\livelink{chap:classconstant}{constant},
1269 \livelink{chap:classexprloc}{exprloc},
1270 \livelink{chap:classloclistptr}{loclistptr}
1271 \addtoindexx{data member attribute!encoding} \\
1272 \DWATdeclcolumn&0x39&\livelink{chap:classconstant}{constant}
1273 \addtoindexx{declaration column attribute!encoding} \\
1274 \DWATdeclfile&0x3a&\livelink{chap:classconstant}{constant}
1275 \addtoindexx{declaration file attribute!encoding} \\
1276 \DWATdeclline&0x3b&\livelink{chap:classconstant}{constant}
1277 \addtoindexx{declaration line attribute!encoding} \\
1278 \DWATdeclaration&0x3c&\livelink{chap:classflag}{flag}
1279 \addtoindexx{declaration attribute!encoding} \\
1280 \DWATdiscrlist&0x3d&\livelink{chap:classblock}{block}
1281 \addtoindexx{discriminant list attribute!encoding} \\
1282 \DWATencoding&0x3e&\livelink{chap:classconstant}{constant}
1283 \addtoindexx{encoding attribute!encoding} \\
1284 \DWATexternal&\xiiif&\livelink{chap:classflag}{flag}
1285 \addtoindexx{external attribute!encoding} \\
1286 \DWATframebase&0x40&\livelink{chap:classexprloc}{exprloc},
1287 \livelink{chap:classloclistptr}{loclistptr}
1288 \addtoindexx{frame base attribute!encoding} \\
1289 \DWATfriend&0x41&\livelink{chap:classreference}{reference}
1290 \addtoindexx{friend attribute!encoding} \\
1291 \DWATidentifiercase&0x42&\livelink{chap:classconstant}{constant}
1292 \addtoindexx{identifier case attribute!encoding} \\
1293 \DWATmacroinfo&0x43&\livelink{chap:classmacptr}{macptr}
1294 \addtoindexx{macro information attribute!encoding} \\
1295 \DWATnamelistitem&0x44&\livelink{chap:classreference}{reference}
1296 \addtoindexx{name list item attribute!encoding} \\
1297 \DWATpriority&0x45&\livelink{chap:classreference}{reference}
1298 \addtoindexx{priority attribute!encoding} \\
1299 \DWATsegment&0x46&\livelink{chap:classexprloc}{exprloc},
1300 \livelink{chap:classloclistptr}{loclistptr}
1301 \addtoindexx{segment attribute!encoding} \\
1302 \DWATspecification&0x47&\livelink{chap:classreference}{reference}
1303 \addtoindexx{specification attribute!encoding} \\
1304 \DWATstaticlink&0x48&\livelink{chap:classexprloc}{exprloc},
1305 \livelink{chap:classloclistptr}{loclistptr}
1306 \addtoindexx{static link attribute!encoding} \\
1307 \DWATtype&0x49&\livelink{chap:classreference}{reference}
1308 \addtoindexx{type attribute!encoding} \\
1309 \DWATuselocation&0x4a&\livelink{chap:classexprloc}{exprloc},
1310 \livelink{chap:classloclistptr}{loclistptr}
1311 \addtoindexx{location list attribute!encoding} \\
1312 \DWATvariableparameter&0x4b&\livelink{chap:classflag}{flag}
1313 \addtoindexx{variable parameter attribute!encoding} \\
1314 \DWATvirtuality&0x4c&\livelink{chap:classconstant}{constant}
1315 \addtoindexx{virtuality attribute!encoding} \\
1316 \DWATvtableelemlocation&0x4d&\livelink{chap:classexprloc}{exprloc},
1317 \livelink{chap:classloclistptr}{loclistptr}
1318 \addtoindexx{vtable element location attribute!encoding} \\
1319 \DWATallocated&0x4e&\livelink{chap:classconstant}{constant},
1320 \livelink{chap:classexprloc}{exprloc},
1321 \livelink{chap:classreference}{reference}
1322 \addtoindexx{allocated attribute!encoding} \\
1323 \DWATassociated&0x4f&\livelink{chap:classconstant}{constant},
1324 \livelink{chap:classexprloc}{exprloc},
1325 \livelink{chap:classreference}{reference}
1326 \addtoindexx{associated attribute!encoding} \\
1327 \DWATdatalocation&0x50&\livelink{chap:classexprloc}{exprloc}
1328 \addtoindexx{data location attribute!encoding} \\
1329 \DWATbytestride&0x51&\livelink{chap:classconstant}{constant},
1330 \livelink{chap:classexprloc}{exprloc},
1331 \livelink{chap:classreference}{reference}
1332 \addtoindexx{byte stride attribute!encoding} \\
1333 \DWATentrypc&0x52&\livelink{chap:classaddress}{address},
1334 \livelink{chap:classconstant}{constant}
1335 \addtoindexx{entry pc attribute!encoding} \\
1336 \DWATuseUTFeight&0x53&\livelink{chap:classflag}{flag}
1337 \addtoindexx{use UTF8 attribute!encoding}\addtoindexx{UTF-8} \\
1338 \DWATextension&0x54&\livelink{chap:classreference}{reference}
1339 \addtoindexx{extension attribute!encoding} \\
1340 \DWATranges&0x55&\livelink{chap:classrangelistptr}{rangelistptr}
1341 \addtoindexx{ranges attribute!encoding} \\
1342 \DWATtrampoline&0x56&\livelink{chap:classaddress}{address},
1343 \livelink{chap:classflag}{flag},
1344 \livelink{chap:classreference}{reference},
1345 \livelink{chap:classstring}{string}
1346 \addtoindexx{trampoline attribute!encoding} \\
1347 \DWATcallcolumn&0x57&\livelink{chap:classconstant}{constant}
1348 \addtoindexx{call column attribute!encoding} \\
1349 \DWATcallfile&0x58&\livelink{chap:classconstant}{constant}
1350 \addtoindexx{call file attribute!encoding} \\
1351 \DWATcallline&0x59&\livelink{chap:classconstant}{constant}
1352 \addtoindexx{call line attribute!encoding} \\
1353 \DWATdescription&0x5a&\livelink{chap:classstring}{string}
1354 \addtoindexx{description attribute!encoding} \\
1355 \DWATbinaryscale&0x5b&\livelink{chap:classconstant}{constant}
1356 \addtoindexx{binary scale attribute!encoding} \\
1357 \DWATdecimalscale&0x5c&\livelink{chap:classconstant}{constant}
1358 \addtoindexx{decimal scale attribute!encoding} \\
1359 \DWATsmall{} &0x5d&\livelink{chap:classreference}{reference}
1360 \addtoindexx{small attribute!encoding} \\
1361 \DWATdecimalsign&0x5e&\livelink{chap:classconstant}{constant}
1362 \addtoindexx{decimal scale attribute!encoding} \\
1363 \DWATdigitcount&0x5f&\livelink{chap:classconstant}{constant}
1364 \addtoindexx{digit count attribute!encoding} \\
1365 \DWATpicturestring&0x60&\livelink{chap:classstring}{string}
1366 \addtoindexx{picture string attribute!encoding} \\
1367 \DWATmutable&0x61&\livelink{chap:classflag}{flag}
1368 \addtoindexx{mutable attribute!encoding} \\
1369 \DWATthreadsscaled&0x62&\livelink{chap:classflag}{flag}
1370 \addtoindexx{thread scaled attribute!encoding} \\
1371 \DWATexplicit&0x63&\livelink{chap:classflag}{flag}
1372 \addtoindexx{explicit attribute!encoding} \\
1373 \DWATobjectpointer&0x64&\livelink{chap:classreference}{reference}
1374 \addtoindexx{object pointer attribute!encoding} \\
1375 \DWATendianity&0x65&\livelink{chap:classconstant}{constant}
1376 \addtoindexx{endianity attribute!encoding} \\
1377 \DWATelemental&0x66&\livelink{chap:classflag}{flag}
1378 \addtoindexx{elemental attribute!encoding} \\
1379 \DWATpure&0x67&\livelink{chap:classflag}{flag}
1380 \addtoindexx{pure attribute!encoding} \\
1381 \DWATrecursive&0x68&\livelink{chap:classflag}{flag}
1382 \addtoindexx{recursive attribute!encoding} \\
1383 \DWATsignature{} &0x69&\livelink{chap:classreference}{reference}
1384 \addtoindexx{signature attribute!encoding} \\
1385 \DWATmainsubprogram{} &0x6a&\livelink{chap:classflag}{flag}
1386 \addtoindexx{main subprogram attribute!encoding} \\
1387 \DWATdatabitoffset{} &0x6b&\livelink{chap:classconstant}{constant}
1388 \addtoindexx{data bit offset attribute!encoding} \\
1389 \DWATconstexpr{} &0x6c&\livelink{chap:classflag}{flag}
1390 \addtoindexx{constant expression attribute!encoding} \\
1391 \DWATenumclass{} &0x6d&\livelink{chap:classflag}{flag}
1392 \addtoindexx{enumeration class attribute!encoding} \\
1393 \DWATlinkagename{} &0x6e&\livelink{chap:classstring}{string}
1394 \addtoindexx{linkage name attribute!encoding} \\
1395 \DWATstringlengthbitsize{} \ddag&0x6f&
1396 \livelink{chap:classconstant}{constant}
1397 \addtoindexx{string length attribute!size of length} \\
1398 \DWATstringlengthbytesize{} \ddag&0x70&
1399 \livelink{chap:classconstant}{constant}
1400 \addtoindexx{string length attribute!size of length} \\
1401 \DWATrank~\ddag&0x71&
1402 \livelink{chap:DWATrank}{constant, exprloc}
1403 \addtoindexx{rank attribute!encoding} \\
1404 \DWATstroffsetsbase~\ddag&0x72&
1405 \livelink{chap:classstring}{reference}
1406 \addtoindexx{string offsets base!encoding} \\
1407 \DWATaddrbase~\ddag &0x73&
1408 \livelink{chap:DWATaddrbase}{reference}
1409 \addtoindexx{address table base!encoding} \\
1410 \DWATrangesbase~\ddag&0x74&
1411 \livelink{chap:DWATrangesbase}{reference}
1412 \addtoindexx{ranges base!encoding} \\
1413 \DWATdwoid~\ddag &0x75&
1414 \livelink{chap:DWATdwoid}{constant}
1415 \addtoindexx{split DWARF object id!encoding} \\
1416 \DWATdwoname~\ddag &0x76&
1417 \livelink{chap:DWATdwname}{string}
1418 \addtoindexx{split DWARF object file name!encoding} \\
1419 \DWATreference~\ddag &0x77&
1420 \livelink{chap:DWATreference}{flag} \\
1421 \DWATrvaluereference~\ddag &0x78&
1422 \livelink{chap:DWATrvaluereference}{flag} \\
1423 \DWATlouser&0x2000 & --- \addtoindexx{low user attribute encoding} \\
1424 \DWAThiuser&\xiiifff& --- \addtoindexx{high user attribute encoding} \\
1431 \setlength{\extrarowheight}{0.1cm}
1432 \begin{longtable}{l|l|l}
1433 \caption{Attribute form encodings} \label{tab:attributeformencodings} \\
1434 \hline \bfseries Form name&\bfseries Value &\bfseries Classes \\ \hline
1436 \bfseries Form name&\bfseries Value &\bfseries Classes\\ \hline
1438 \hline \emph{Continued on next page}
1443 \DWFORMaddr &0x01&\livelink{chap:classaddress}{address} \\
1444 \textit{Reserved} &0x02& \\
1445 \DWFORMblocktwo &0x03&\livelink{chap:classblock}{block} \\
1446 \DWFORMblockfour &0x04&\livelink{chap:classblock}{block} \\
1447 \DWFORMdatatwo &0x05&\livelink{chap:classconstant}{constant} \\
1448 \DWFORMdatafour &0x06&\livelink{chap:classconstant}{constant} \\
1449 \DWFORMdataeight &0x07&\livelink{chap:classconstant}{constant} \\
1450 \DWFORMstring&0x08&\livelink{chap:classstring}{string} \\
1451 \DWFORMblock&0x09&\livelink{chap:classblock}{block} \\
1452 \DWFORMblockone &0x0a&\livelink{chap:classblock}{block} \\
1453 \DWFORMdataone &0x0b&\livelink{chap:classconstant}{constant} \\
1454 \DWFORMflag&0x0c&\livelink{chap:classflag}{flag} \\
1455 \DWFORMsdata&0x0d&\livelink{chap:classconstant}{constant} \\
1456 \DWFORMstrp&0x0e&\livelink{chap:classstring}{string} \\
1457 \DWFORMudata&0x0f&\livelink{chap:classconstant}{constant} \\
1458 \DWFORMrefaddr&0x10&\livelink{chap:classreference}{reference} \\
1459 \DWFORMrefone&0x11&\livelink{chap:classreference}{reference} \\
1460 \DWFORMreftwo&0x12&\livelink{chap:classreference}{reference} \\
1461 \DWFORMreffour&0x13&\livelink{chap:classreference}{reference} \\
1462 \DWFORMrefeight&0x14&\livelink{chap:classreference}{reference} \\
1463 \DWFORMrefudata&0x15&\livelink{chap:classreference}{reference} \\
1464 \DWFORMindirect&0x16&(see Section \refersec{datarep:abbreviationstables}) \\
1465 \DWFORMsecoffset{} &0x17&\livelink{chap:classlineptr}{lineptr}, \livelink{chap:classloclistptr}{loclistptr}, \livelink{chap:classmacptr}{macptr}, \livelink{chap:classrangelistptr}{rangelistptr} \\
1466 \DWFORMexprloc{} &0x18&\livelink{chap:classexprloc}{exprloc} \\
1467 \DWFORMflagpresent{} &0x19&\livelink{chap:classflag}{flag} \\
1468 \DWFORMstrx &0x1a&\livelink{chap:classstring}{string} \\
1469 \DWFORMaddrx &0x1b&\livelink{chap:classaddess}{address} \\
1470 \DWFORMrefsigeight &0x20&\livelink{chap:classreference}{reference} \\
1477 \section{Variable Length Data}
1478 \label{datarep:variablelengthdata}
1479 \addtoindexx{variable length data|see {LEB128}}
1481 \addtoindexx{Little Endian Base 128|see{LEB128}}
1482 encoded using \doublequote{Little Endian Base 128}
1483 \addtoindexx{little-endian encoding|see{endian attribute}}
1485 \addtoindexx{LEB128}
1486 LEB128 is a scheme for encoding integers
1487 densely that exploits the assumption that most integers are
1490 \textit{This encoding is equally suitable whether the target machine
1491 architecture represents data in big\dash\ endian or little\dash endian
1492 order. It is \doublequote{little\dash endian} only in the sense that it
1493 avoids using space to represent the \doublequote{big} end of an
1494 unsigned integer, when the big end is all zeroes or sign
1497 Unsigned LEB128 (ULEB128) numbers are encoded as follows:
1498 \addtoindexx{LEB128!unsigned, encoding as}
1499 start at the low order end of an unsigned integer and chop
1500 it into 7\dash bit chunks. Place each chunk into the low order 7
1501 bits of a byte. Typically, several of the high order bytes
1502 will be zero; discard them. Emit the remaining bytes in a
1503 stream, starting with the low order byte; set the high order
1504 bit on each byte except the last emitted byte. The high bit
1505 of zero on the last byte indicates to the decoder that it
1506 has encountered the last byte.
1508 The integer zero is a special case, consisting of a single
1511 Table \refersec{tab:examplesofunsignedleb128encodings}
1512 gives some examples of unsigned LEB128 numbers. The
1513 0x80 in each case is the high order bit of the byte, indicating
1514 that an additional byte follows.
1517 The encoding for signed, two\textquoteright s complement LEB128 (SLEB128)
1518 \addtoindexx{LEB128!signed, encoding as}
1519 numbers is similar, except that the criterion for discarding
1520 high order bytes is not whether they are zero, but whether
1521 they consist entirely of sign extension bits. Consider the
1522 32\dash bit integer -2. The three high level bytes of the number
1523 are sign extension, thus LEB128 would represent it as a single
1524 byte containing the low order 7 bits, with the high order
1525 bit cleared to indicate the end of the byte stream. Note
1526 that there is nothing within the LEB128 representation that
1527 indicates whether an encoded number is signed or unsigned. The
1528 decoder must know what type of number to expect.
1529 Table \refersec{tab:examplesofunsignedleb128encodings}
1530 gives some examples of unsigned LEB128 numbers and
1531 Table \refersec{tab:examplesofsignedleb128encodings}
1532 gives some examples of signed LEB128
1535 \textit{Appendix \refersec{app:variablelengthdataencodingdecodinginformative}
1536 \addtoindexx{LEB128!examples}
1537 gives algorithms for encoding and decoding these forms.}
1541 \setlength{\extrarowheight}{0.1cm}
1542 \begin{longtable}{l|l|l}
1543 \caption{Examples of unsigned LEB128 encodings}
1544 \label{tab:examplesofunsignedleb128encodings}
1545 \addtoindexx{LEB128 encoding!examples} \\
1546 \hline \bfseries Number&\bfseries First byte &\bfseries Second byte \\ \hline
1548 \bfseries Number&\bfseries First Byte &\bfseries Second byte\\ \hline
1550 \hline \emph{Continued on next page}
1556 128& 0 + 0x80 & 1 \\
1557 129& 1 + 0x80 & 1 \\
1558 130& 2 + 0x80 & 1 \\
1559 12857& 57 + 0x80 & 100 \\
1566 \setlength{\extrarowheight}{0.1cm}
1567 \begin{longtable}{l|l|l}
1568 \caption{Examples of signed LEB128 encodings} \label{tab:examplesofsignedleb128encodings} \\
1569 \hline \bfseries Number&\bfseries First byte &\bfseries Second byte \\ \hline
1571 \bfseries Number&\bfseries First Byte &\bfseries Second byte\\ \hline
1573 \hline \emph{Continued on next page}
1579 127& 127 + 0x80 & 0 \\
1580 -127& 1 + 0x80 & 0x7f \\
1581 128& 0 + 0x80 & 1 \\
1582 -128& 0 + 0x80 & 0x7f \\
1583 129& 1 + 0x80 & 1 \\
1584 -129& 0x7f + 0x80 & 0x7e \\
1591 \section{DWARF Expressions and Location Descriptions}
1592 \label{datarep:dwarfexpressionsandlocationdescriptions}
1593 \subsection{DWARF Expressions}
1594 \label{datarep:dwarfexpressions}
1597 \addtoindexx{DWARF Expression!operator encoding}
1598 DWARF expression is stored in a \nolink{block} of contiguous
1599 bytes. The bytes form a sequence of operations. Each operation
1600 is a 1\dash byte code that identifies that operation, followed by
1601 zero or more bytes of additional data. The encodings for the
1602 operations are described in
1603 Table \refersec{tab:dwarfoperationencodings}.
1606 \setlength{\extrarowheight}{0.1cm}
1607 \begin{longtable}{l|c|c|l}
1608 \caption{DWARF operation encodings} \label{tab:dwarfoperationencodings} \\
1609 \hline & &\bfseries No. of &\\
1610 \bfseries Operation&\bfseries Code &\bfseries Operands &\bfseries Notes\\ \hline
1612 & &\bfseries No. of &\\
1613 \bfseries Operation&\bfseries Code &\bfseries Operands &\bfseries Notes\\ \hline
1615 \hline \emph{Continued on next page}
1617 \hline \ddag\ \textit{New in DWARF Version 5}
1620 \DWOPaddr&0x03&1 & constant address \\
1621 & & &(size is target specific) \\
1623 \DWOPderef&0x06&0 & \\
1625 \DWOPconstoneu&0x08&1&1\dash byte constant \\
1626 \DWOPconstones&0x09&1&1\dash byte constant \\
1627 \DWOPconsttwou&0x0a&1&2\dash byte constant \\
1628 \DWOPconsttwos&0x0b&1&2\dash byte constant \\
1629 \DWOPconstfouru&0x0c&1&4\dash byte constant \\
1630 \DWOPconstfours&0x0d&1&4\dash byte constant \\
1631 \DWOPconsteightu&0x0e&1&8\dash byte constant \\
1632 \DWOPconsteights&0x0f&1&8\dash byte constant \\
1633 \DWOPconstu&0x10&1&ULEB128 constant \\
1634 \DWOPconsts&0x11&1&SLEB128 constant \\
1635 \DWOPdup&0x12&0 & \\
1636 \DWOPdrop&0x13&0 & \\
1637 \DWOPover&0x14&0 & \\
1638 \DWOPpick&0x15&1&1\dash byte stack index \\
1639 \DWOPswap&0x16&0 & \\
1640 \DWOProt&0x17&0 & \\
1641 \DWOPxderef&0x18&0 & \\
1642 \DWOPabs&0x19&0 & \\
1643 \DWOPand&0x1a&0 & \\
1644 \DWOPdiv&0x1b&0 & \\
1645 \DWOPminus&0x1c&0 & \\
1646 \DWOPmod&0x1d&0 & \\
1647 \DWOPmul&0x1e&0 & \\
1648 \DWOPneg&0x1f&0 & \\
1649 \DWOPnot&0x20&0 & \\
1651 \DWOPplus&0x22&0 & \\
1652 \DWOPplusuconst&0x23&1&ULEB128 addend \\
1653 \DWOPshl&0x24&0 & \\
1654 \DWOPshr&0x25&0 & \\
1655 \DWOPshra&0x26&0 & \\
1656 \DWOPxor&0x27&0 & \\
1658 \DWOPbra&0x28&1 & signed 2\dash byte constant \\
1665 \DWOPskip&0x2f&1&signed 2\dash byte constant \\ \hline
1667 \DWOPlitzero & 0x30 & 0 & \\
1668 \DWOPlitone & 0x31 & 0& literals 0 .. 31 = \\
1669 \ldots & & &\hspace{0.3cm}(\DWOPlitzero{} + literal) \\
1670 \DWOPlitthirtyone & 0x4f & 0 & \\ \hline
1672 \DWOPregzero & 0x50 & 0 & \\*
1673 \DWOPregone & 0x51 & 0® 0 .. 31 = \\*
1674 \ldots & & &\hspace{0.3cm}(\DWOPregzero{} + regnum) \\*
1675 \DWOPregthirtyone & 0x6f & 0 & \\ \hline
1677 \DWOPbregzero & 0x70 &1 & SLEB128 offset \\*
1678 \DWOPbregone & 0x71 & 1 &base register 0 .. 31 = \\*
1679 ... & & &\hspace{0.3cm}(\DWOPbregzero{} + regnum) \\*
1680 \DWOPbregthirtyone & 0x8f & 1 & \\ \hline
1682 \DWOPregx{} & 0x90 &1&ULEB128 register \\
1683 \DWOPfbreg{} & 0x91&1&SLEB128 offset \\
1684 \DWOPbregx{} & 0x92&2 &ULEB128 register, \\*
1685 & & &SLEB128 offset \\
1686 \DWOPpiece{} & 0x93 &1& ULEB128 size of piece \\
1687 \DWOPderefsize{} & 0x94 &1& 1-byte size of data retrieved \\
1688 \DWOPxderefsize{} & 0x95&1&1-byte size of data retrieved \\
1689 \DWOPnop{} & 0x96 &0& \\
1691 \DWOPpushobjectaddress&0x97&0 & \\
1692 \DWOPcalltwo&0x98&1& 2\dash byte offset of DIE \\
1693 \DWOPcallfour&0x99&1& 4\dash byte offset of DIE \\
1694 \DWOPcallref&0x9a&1& 4\dash\ or 8\dash byte offset of DIE \\
1695 \DWOPformtlsaddress&0x9b &0& \\
1696 \DWOPcallframecfa{} &0x9c &0& \\
1697 \DWOPbitpiece&0x9d &2&ULEB128 size, \\*
1699 \DWOPimplicitvalue{} &0x9e &2&ULEB128 size, \\*
1700 &&&\nolink{block} of that size\\
1701 \DWOPstackvalue{} &0x9f &0& \\
1702 \DWOPimplicitpointer{}~\ddag &0xa0& 2 &4- or 8-byte offset of DIE, \\*
1703 &&&SLEB128 constant offset \\
1704 \DWOPaddrx~\ddag&0xa1&1&ULEB128 indirect address \\
1705 \DWOPconstx~\ddag&0xa2&1&ULEB128 indirect constant \\
1706 \DWOPlouser{} &0xe0 && \\
1707 \DWOPhiuser{} &\xff && \\
1713 \subsection{Location Descriptions}
1714 \label{datarep:locationdescriptions}
1716 A location description is used to compute the
1717 location of a variable or other entity.
1719 \subsection{Location Lists}
1720 \label{datarep:locationlists}
1722 Each entry in a \addtoindex{location list} is either a location list entry,
1723 a base address selection entry, or an
1724 \addtoindexx{end of list entry!in location list}
1728 \subsubsection{Location List Entries in Non-Split Objects}
1729 A \addtoindex{location list} entry consists of two address offsets followed
1730 by a 2\dash byte length, followed by a block of contiguous bytes
1731 that contains a DWARF location description. The length
1732 specifies the number of bytes in that block. The two offsets
1733 are the same size as an address on the target machine.
1736 A base address selection entry and an
1737 \addtoindexx{end of list entry!in location list}
1738 end of list entry each
1739 consist of two (constant or relocated) address offsets. The two
1740 offsets are the same size as an address on the target machine.
1742 For a \addtoindex{location list} to be specified, the base address of
1743 \addtoindexx{base address selection entry!in location list}
1744 the corresponding compilation unit must be defined
1745 (see Section \refersec{chap:normalandpartialcompilationunitentries}).
1747 \subsubsection{Location List Entries in Split Objects}
1748 An alternate form for location list entries is used in split objects.
1749 Each entry begins with a one-byte code that indicates the kind of entry
1750 that follows. The encodings for these constants are given in
1751 Table \refersec{tab:locationlistentryencodingvalues}.
1754 \setlength{\extrarowheight}{0.1cm}
1755 \begin{longtable}{l|c}
1756 \caption{Location list entry encoding values} \label{tab:locationlistentryencodingvalues} \\
1757 \hline \bfseries Location list entry encoding name&\bfseries Value \\ \hline
1759 \bfseries Location list entry encoding name&\bfseries Value\\ \hline
1761 \hline \emph{Continued on next page}
1765 \DWLLEendoflistentry & 0x0 \\
1766 \DWLLEbaseaddressselectionentry & 0x01 \\
1767 \DWLLEstartendentry & 0x02 \\
1768 \DWLLEstartlengthentry & 0x03 \\
1769 \DWLLEoffsetpairentry & 0x04 \\
1773 \section{Base Type Attribute Encodings}
1774 \label{datarep:basetypeattributeencodings}
1776 The encodings of the
1777 \hypertarget{chap:DWATencodingencodingofbasetype}{}
1779 \addtoindexx{encoding attribute!encoding}
1782 attribute are given in
1783 Table \refersec{tab:basetypeencodingvalues}
1786 \setlength{\extrarowheight}{0.1cm}
1787 \begin{longtable}{l|c}
1788 \caption{Base type encoding values} \label{tab:basetypeencodingvalues} \\
1789 \hline \bfseries Base type encoding name&\bfseries Value \\ \hline
1791 \bfseries Base type encoding name&\bfseries Value\\ \hline
1793 \hline \emph{Continued on next page}
1797 \DWATEaddress&0x01 \\
1798 \DWATEboolean&0x02 \\
1799 \DWATEcomplexfloat&0x03 \\
1801 \DWATEsigned&0x05 \\
1802 \DWATEsignedchar&0x06 \\
1803 \DWATEunsigned&0x07 \\
1804 \DWATEunsignedchar&0x08 \\
1805 \DWATEimaginaryfloat&0x09 \\
1806 \DWATEpackeddecimal&0x0a \\
1807 \DWATEnumericstring&0x0b \\
1808 \DWATEedited&0x0c \\
1809 \DWATEsignedfixed&0x0d \\
1810 \DWATEunsignedfixed&0x0e \\
1811 \DWATEdecimalfloat{} & 0x0f \\
1812 \DWATEUTF{} & 0x10 \\
1813 \DWATElouser{} & 0x80 \\
1814 \DWATEhiuser{} & \xff \\
1819 The encodings of the constants used in the
1820 \DWATdecimalsign{} attribute
1822 Table \refersec{tab:decimalsignencodings}.
1825 \setlength{\extrarowheight}{0.1cm}
1826 \begin{longtable}{l|c}
1827 \caption{Decimal sign encodings} \label{tab:decimalsignencodings} \\
1828 \hline \bfseries Decimal sign code name&\bfseries Value \\ \hline
1830 \bfseries Decimal sign code name&\bfseries Value\\ \hline
1832 \hline \emph{Continued on next page}
1837 \DWDSunsigned{} & 0x01 \\
1838 \DWDSleadingoverpunch{} & 0x02 \\
1839 \DWDStrailingoverpunch{} & 0x03 \\
1840 \DWDSleadingseparate{} & 0x04 \\
1841 \DWDStrailingseparate{} & 0x05 \\
1847 The encodings of the constants used in the
1848 \DWATendianity{} attribute are given in
1849 Table \refersec{tab:endianityencodings}.
1852 \setlength{\extrarowheight}{0.1cm}
1853 \begin{longtable}{l|c}
1854 \caption{Endianity encodings} \label{tab:endianityencodings}\\
1855 \hline \bfseries Endian code name&\bfseries Value \\ \hline
1857 \bfseries Endian code name&\bfseries Value\\ \hline
1859 \hline \emph{Continued on next page}
1864 \DWENDdefault{} & 0x00 \\
1865 \DWENDbig{} & 0x01 \\
1866 \DWENDlittle{} & 0x02 \\
1867 \DWENDlouser{} & 0x40 \\
1868 \DWENDhiuser{} & \xff \\
1873 \section{Accessibility Codes}
1874 \label{datarep:accessibilitycodes}
1875 The encodings of the constants used in the
1876 \DWATaccessibility{}
1878 \addtoindexx{accessibility attribute!encoding}
1880 Table \refersec{tab:accessibilityencodings}.
1883 \setlength{\extrarowheight}{0.1cm}
1884 \begin{longtable}{l|c}
1885 \caption{Accessibility encodings} \label{tab:accessibilityencodings}\\
1886 \hline \bfseries Accessibility code name&\bfseries Value \\ \hline
1888 \bfseries Accessibility code name&\bfseries Value\\ \hline
1890 \hline \emph{Continued on next page}
1895 \DWACCESSpublic&0x01 \\
1896 \DWACCESSprotected&0x02 \\
1897 \DWACCESSprivate&0x03 \\
1903 \section{Visibility Codes}
1904 \label{datarep:visibilitycodes}
1905 The encodings of the constants used in the
1906 \DWATvisibility{} attribute are given in
1907 Table \refersec{tab:visibilityencodings}.
1910 \setlength{\extrarowheight}{0.1cm}
1911 \begin{longtable}{l|c}
1912 \caption{Visibility encodings} \label{tab:visibilityencodings}\\
1913 \hline \bfseries Visibility code name&\bfseries Value \\ \hline
1915 \bfseries Visibility code name&\bfseries Value\\ \hline
1917 \hline \emph{Continued on next page}
1923 \DWVISexported&0x02 \\
1924 \DWVISqualified&0x03 \\
1929 \section{Virtuality Codes}
1930 \label{datarep:vitualitycodes}
1932 The encodings of the constants used in the
1933 \DWATvirtuality{} attribute are given in
1934 Table \refersec{tab:virtualityencodings}.
1937 \setlength{\extrarowheight}{0.1cm}
1938 \begin{longtable}{l|c}
1939 \caption{Virtuality encodings} \label{tab:virtualityencodings}\\
1940 \hline \bfseries Virtuality code name&\bfseries Value \\ \hline
1942 \bfseries Virtuality code name&\bfseries Value\\ \hline
1944 \hline \emph{Continued on next page}
1949 \DWVIRTUALITYnone&0x00 \\
1950 \DWVIRTUALITYvirtual&0x01 \\
1951 \DWVIRTUALITYpurevirtual&0x02 \\
1959 \DWVIRTUALITYnone{} is equivalent to the absence of the
1963 \section{Source Languages}
1964 \label{datarep:sourcelanguages}
1966 The encodings of the constants used
1967 \addtoindexx{language attribute, encoding}
1969 \addtoindexx{language name encoding}
1972 attribute are given in
1973 Table \refersec{tab:languageencodings}.
1975 % If we don't force a following space it looks odd
1977 and their associated values are reserved, but the
1978 languages they represent are not well supported.
1979 Table \refersec{tab:languageencodings}
1981 \addtoindexx{lower bound attribute!default}
1982 default lower bound, if any, assumed for
1983 an omitted \DWATlowerbound{} attribute in the context of a
1984 \DWTAGsubrangetype{} debugging information entry for each
1988 \setlength{\extrarowheight}{0.1cm}
1989 \begin{longtable}{l|c|c}
1990 \caption{Language encodings} \label{tab:languageencodings}\\
1991 \hline \bfseries Language name&\bfseries Value &\bfseries Default Lower Bound \\ \hline
1993 \bfseries Language name&\bfseries Value &\bfseries Default Lower Bound\\ \hline
1995 \hline \emph{Continued on next page}
1998 \dag \ \textit{See text} \\ \ddag \ \textit{New in \addtoindex{DWARF Version 5}}
2001 \DWLANGCeightynine &0x0001 &0 \\
2002 \DWLANGC{} &0x0002 &0 \\
2003 \DWLANGAdaeightythree{} \dag &0x0003 &1 \addtoindexx{Ada} \\
2004 \DWLANGCplusplus{} &0x0004 &0 \\
2005 \DWLANGCobolseventyfour{} \dag &0x0005 &1 \\
2006 \DWLANGCoboleightyfive{} \dag &0x0006 &1 \\
2007 \DWLANGFortranseventyseven &0x0007 &1 \\
2008 \DWLANGFortranninety &0x0008 &1 \\
2009 \DWLANGPascaleightythree &0x0009 &1 \\
2010 \DWLANGModulatwo &0x000a &1 \\
2011 \DWLANGJava &0x000b &0 \\
2012 \DWLANGCninetynine &0x000c &0 \\
2013 \DWLANGAdaninetyfive{} \dag &0x000d &1 \addtoindexx{Ada} \\
2014 \DWLANGFortranninetyfive &0x000e &1 \\
2015 \DWLANGPLI{} \dag &0x000f &1 \\
2016 \DWLANGObjC{} &0x0010 &0 \\
2017 \DWLANGObjCplusplus{} &0x0011 &0 \\
2018 \DWLANGUPC{} &0x0012 &0 \\
2019 \DWLANGD{} &0x0013 &0 \\
2020 \DWLANGPython{} \dag &0x0014 &0 \\
2021 \DWLANGOpenCL{} \dag \ddag &0x0015 &0 \\
2022 \DWLANGGo{} \dag \ddag &0x0016 &0 \\
2023 \DWLANGModulathree{} \dag \ddag &0x0017 &1 \\
2024 \DWLANGHaskell{} \dag \ddag &0x0018 &0 \\
2025 \DWLANGCpluspluszerothree{} \ddag &0x0019 &0 \\
2026 \DWLANGCpluspluseleven{} \ddag &0x001a &0 \\
2027 \DWLANGOCaml{} \ddag &0x001b &0 \\
2028 \DWLANGlouser{} &0x8000 & \\
2029 \DWLANGhiuser{} &\xffff & \\
2034 \section{Address Class Encodings}
2035 \label{datarep:addressclassencodings}
2037 The value of the common
2038 \addtoindexi{address}{address class!attribute encoding}
2043 \section{Identifier Case}
2044 \label{datarep:identifiercase}
2046 The encodings of the constants used in the
2047 \DWATidentifiercase{} attribute are given in
2048 Table \refersec{tab:identifiercaseencodings}.
2051 \setlength{\extrarowheight}{0.1cm}
2052 \begin{longtable}{l|c}
2053 \caption{Identifier case encodings} \label{tab:identifiercaseencodings}\\
2054 \hline \bfseries Identifier case name&\bfseries Value \\ \hline
2056 \bfseries Identifier case name&\bfseries Value\\ \hline
2058 \hline \emph{Continued on next page}
2062 \DWIDcasesensitive&0x00 \\
2064 \DWIDdowncase&0x02 \\
2065 \DWIDcaseinsensitive&0x03 \\
2069 \section{Calling Convention Encodings}
2070 \label{datarep:callingconventionencodings}
2071 The encodings of the constants used in the
2072 \DWATcallingconvention{} attribute are given in
2073 Table \refersec{tab:callingconventionencodings}.
2076 \setlength{\extrarowheight}{0.1cm}
2077 \begin{longtable}{l|c}
2078 \caption{Calling convention encodings} \label{tab:callingconventionencodings}\\
2079 \hline \bfseries Calling convention name&\bfseries Value \\ \hline
2081 \bfseries Calling convention name&\bfseries Value\\ \hline
2083 \hline \emph{Continued on next page}
2089 \DWCCprogram&0x02 \\
2097 \section{Inline Codes}
2098 \label{datarep:inlinecodes}
2100 The encodings of the constants used in
2101 \addtoindexx{inline attribute!encoding}
2103 \DWATinline{} attribute are given in
2104 Table \refersec{tab:inlineencodings}.
2108 \setlength{\extrarowheight}{0.1cm}
2109 \begin{longtable}{l|c}
2110 \caption{Inline encodings} \label{tab:inlineencodings}\\
2111 \hline \bfseries Inline code name&\bfseries Value \\ \hline
2113 \bfseries Inline Code name&\bfseries Value\\ \hline
2115 \hline \emph{Continued on next page}
2120 \DWINLnotinlined&0x00 \\
2121 \DWINLinlined&0x01 \\
2122 \DWINLdeclarednotinlined&0x02 \\
2123 \DWINLdeclaredinlined&0x03 \\
2128 % this clearpage is ugly, but the following table came
2129 % out oddly without it.
2131 \section{Array Ordering}
2132 \label{datarep:arrayordering}
2134 The encodings of the constants used in the
2135 \DWATordering{} attribute are given in
2136 Table \refersec{tab:orderingencodings}.
2140 \setlength{\extrarowheight}{0.1cm}
2141 \begin{longtable}{l|c}
2142 \caption{Ordering encodings} \label{tab:orderingencodings}\\
2143 \hline \bfseries Ordering name&\bfseries Value \\ \hline
2145 \bfseries Ordering name&\bfseries Value\\ \hline
2147 \hline \emph{Continued on next page}
2152 \DWORDrowmajor&0x00 \\
2153 \DWORDcolmajor&0x01 \\
2159 \section{Discriminant Lists}
2160 \label{datarep:discriminantlists}
2162 The descriptors used in
2163 \addtoindexx{discriminant list attribute!encoding}
2165 \DWATdiscrlist{} attribute are
2166 encoded as 1\dash byte constants. The
2167 defined values are given in
2168 Table \refersec{tab:discriminantdescriptorencodings}.
2170 % Odd that the 'Name' field capitalized here, it is not caps elsewhere.
2172 \setlength{\extrarowheight}{0.1cm}
2173 \begin{longtable}{l|c}
2174 \caption{Discriminant descriptor encodings} \label{tab:discriminantdescriptorencodings}\\
2175 \hline \bfseries Descriptor name&\bfseries Value \\ \hline
2177 \bfseries Descriptor name&\bfseries Value\\ \hline
2179 \hline \emph{Continued on next page}
2191 \section{Name Lookup Tables}
2192 \label{datarep:namelookuptables}
2194 Each set of entries in the table of global names contained
2195 in the \dotdebugpubnames{} and
2196 \dotdebugpubtypes{} sections begins
2197 with a header consisting of:
2198 \begin{enumerate}[1. ]
2200 \item \texttt{unit\_length} (\livelink{datarep:initiallengthvalues}{initial length}) \\
2201 \addttindexx{unit\_length}
2202 A 4\dash byte or 12\dash byte unsigned integer
2203 \addtoindexx{initial length}
2204 representing the length
2205 of the \dotdebuginfo{}
2206 contribution for that compilation unit,
2207 not including the length field itself. In the
2208 \thirtytwobitdwarfformat, this is a 4\dash byte unsigned integer (which must be less
2209 than \xfffffffzero); in the \sixtyfourbitdwarfformat, this consists
2210 of the 4\dash byte value \wffffffff followed by an 8\dash byte unsigned
2211 integer that gives the actual length
2212 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
2214 \item version (\addtoindex{uhalf}) \\
2215 A 2\dash byte unsigned integer representing the version of the
2216 DWARF information for the name lookup table
2217 \addtoindexx{version number!name lookup table}
2218 (see Appendix \refersec{app:dwarfsectionversionnumbersinformative}).
2219 The value in this field is 2.
2222 \item \addtoindex{debug\_info\_offset} (section offset) \\
2224 \addtoindexx{section offset!in name lookup table set of entries}
2225 4\dash byte or 8\dash byte
2227 \dotdebuginfo{} or \dotdebuginfodwo{}
2228 section of the compilation unit header.
2229 In the \thirtytwobitdwarfformat, this is a 4\dash byte unsigned offset;
2230 in the \sixtyfourbitdwarfformat, this is an 8\dash byte unsigned offsets
2231 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
2233 \item \addtoindex{debug\_info\_length} (\livelink{datarep:sectionoffsetlength}{section length}) \\
2234 \addtoindexx{section length!in .debug\_pubnames header}
2236 \addtoindexx{section length!in .debug\_pubtypes header}
2237 4\dash byte or 8\dash byte length containing the size in bytes of the
2238 contents of the \dotdebuginfo{}
2239 section generated to represent
2240 this compilation unit. In the \thirtytwobitdwarfformat, this is
2241 a 4\dash byte unsigned length; in the \sixtyfourbitdwarfformat, this
2242 is an 8-byte unsigned length
2243 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
2248 This header is followed by a series of tuples. Each tuple
2249 consists of a 4\dash byte or 8\dash byte offset followed by a string
2250 of non\dash null bytes terminated by one null byte.
2252 DWARF format, this is a 4\dash byte offset; in the 64\dash bit DWARF
2253 format, it is an 8\dash byte offset.
2254 Each set is terminated by an
2255 offset containing the value 0.
2259 \section{Address Range Table}
2260 \label{datarep:addrssrangetable}
2262 Each set of entries in the table of address ranges contained
2263 in the \dotdebugaranges{}
2264 section begins with a header containing:
2265 \begin{enumerate}[1. ]
2266 % FIXME The unit length text is not fully consistent across
2269 \item \texttt{unit\_length} (\livelink{datarep:initiallengthvalues}{initial length}) \\
2270 \addttindexx{unit\_length}
2271 A 4-byte or 12-byte length containing the length of the
2272 \addtoindexx{initial length}
2273 set of entries for this compilation unit, not including the
2274 length field itself. In the \thirtytwobitdwarfformat, this is a
2275 4-byte unsigned integer (which must be less than \xfffffffzero);
2276 in the \sixtyfourbitdwarfformat, this consists of the 4-byte value
2277 \wffffffff followed by an 8-byte unsigned integer that gives
2279 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
2281 \item version (\addtoindex{uhalf}) \\
2282 A 2\dash byte version identifier representing the version of the
2283 DWARF information for the address range table
2284 (see Appendix \refersec{app:dwarfsectionversionnumbersinformative}).
2285 This value in this field \addtoindexx{version number!address range table} is 2.
2288 \item debug\_info\_offset (\livelink{datarep:sectionoffsetlength}{section offset}) \\
2290 \addtoindexx{section offset!in .debug\_aranges header}
2291 4\dash byte or 8\dash byte offset into the
2292 \dotdebuginfo{} section of
2293 the compilation unit header. In the \thirtytwobitdwarfformat,
2294 this is a 4\dash byte unsigned offset; in the \sixtyfourbitdwarfformat,
2295 this is an 8\dash byte unsigned offset
2296 (see Section \refersec{datarep:32bitand64bitdwarfformats}).
2298 \item address\_size (ubyte) \\
2299 A 1\dash byte unsigned integer containing the size in bytes of an
2300 \addtoindexx{address\_size}
2302 \addtoindexx{size of an address}
2303 (or the offset portion of an address for segmented
2304 \addtoindexx{address space!segmented}
2305 addressing) on the target system.
2307 \item segment\_size (ubyte) \\
2309 \addtoindexx{segment\_size}
2310 1\dash byte unsigned integer containing the size in bytes of a
2311 segment selector on the target system.
2315 This header is followed by a series of tuples. Each tuple
2316 consists of a segment, an address and a length.
2318 size is given by the \addtoindex{segment\_size} field of the header; the
2319 address and length size are each given by the address\_size
2320 field of the header.
2321 The first tuple following the header in
2322 each set begins at an offset that is a multiple of the size
2323 of a single tuple (that is, the size of a segment selector
2324 plus twice the \addtoindex{size of an address}).
2325 The header is padded, if
2326 necessary, to that boundary. Each set of tuples is terminated
2327 by a 0 for the segment, a 0 for the address and 0 for the
2328 length. If the \addtoindex{segment\_size} field in the header is zero,
2329 the segment selectors are omitted from all tuples, including
2330 the terminating tuple.
2333 \section{Line Number Information}
2334 \label{datarep:linenumberinformation}
2336 The \addtoindexi{version number}{version number!line number information}
2337 in the line number program header is \versiondotdebugline{}
2338 (see Appendix \refersec{app:dwarfsectionversionnumbersinformative}).
2340 The boolean values \doublequote{true} and \doublequote{false}
2341 used by the line number information program are encoded
2342 as a single byte containing the value 0
2343 for \doublequote{false,} and a non-zero value for \doublequote{true.}
2345 The encodings for the standard opcodes are given in
2346 \addtoindexx{line number opcodes!standard opcode encoding}
2347 Table \refersec{tab:linenumberstandardopcodeencodings}.
2349 % Odd that the 'Name' field capitalized here, it is not caps elsewhere.
2351 \setlength{\extrarowheight}{0.1cm}
2352 \begin{longtable}{l|c}
2353 \caption{Line number standard opcode encodings} \label{tab:linenumberstandardopcodeencodings}\\
2354 \hline \bfseries Opcode name&\bfseries Value \\ \hline
2356 \bfseries Opcode name&\bfseries Value\\ \hline
2358 \hline \emph{Continued on next page}
2364 \DWLNSadvancepc&0x02 \\
2365 \DWLNSadvanceline&0x03 \\
2366 \DWLNSsetfile&0x04 \\
2367 \DWLNSsetcolumn&0x05 \\
2368 \DWLNSnegatestmt&0x06 \\
2369 \DWLNSsetbasicblock&0x07 \\
2370 \DWLNSconstaddpc&0x08 \\
2371 \DWLNSfixedadvancepc&0x09 \\
2372 \DWLNSsetprologueend&0x0a \\*
2373 \DWLNSsetepiloguebegin&0x0b \\*
2374 \DWLNSsetisa&0x0c \\*
2381 The encodings for the extended opcodes are given in
2382 \addtoindexx{line number opcodes!extended opcode encoding}
2383 Table \refersec{tab:linenumberextendedopcodeencodings}.
2386 \setlength{\extrarowheight}{0.1cm}
2387 \begin{longtable}{l|c}
2388 \caption{Line number extended opcode encodings} \label{tab:linenumberextendedopcodeencodings}\\
2389 \hline \bfseries Opcode name&\bfseries Value \\ \hline
2391 \bfseries Opcode name&\bfseries Value\\ \hline
2393 \hline \emph{Continued on next page}
2395 \hline \ddag~\textit{New in DWARF Version 5}
2398 \DWLNEendsequence &0x01 \\
2399 \DWLNEsetaddress &0x02 \\
2400 \DWLNEdefinefile &0x03 \\
2401 \DWLNEsetdiscriminator &0x04 \\
2402 \DWLNEdefinefileMDfive~\ddag &0x05 \\
2403 \DWLNElouser &0x80 \\
2404 \DWLNEhiuser &\xff \\
2410 The encodings for the file entry format are given in
2411 \addtoindexx{line number opcodes!file entry format encoding}
2412 Table \refersec{tab:linenumberfileentryformatencodings}.
2415 \setlength{\extrarowheight}{0.1cm}
2416 \begin{longtable}{l|c}
2417 \caption{Line number file entry format \mbox{encodings}} \label{tab:linenumberfileentryformatencodings}\\
2418 \hline \bfseries File entry format name&\bfseries Value \\ \hline
2420 \bfseries File entry format name&\bfseries Value\\ \hline
2422 \hline \emph{Continued on next page}
2427 \DWLNFtimestampsize & 0x01 \\
2428 \DWLNFMDfive & 0x02 \\
2433 \section{Macro Information}
2434 \label{datarep:macroinformation}
2436 The source line numbers and source file indices encoded in the
2437 macro information section are represented as unsigned LEB128
2438 numbers as are the constants in a
2439 \DWMACINFOvendorext{} entry.
2441 The macinfo type is encoded as a single byte.
2443 \addtoindexx{macinfo types!encoding}
2445 Table \refersec{tab:macinfotypeencodings}.
2449 \setlength{\extrarowheight}{0.1cm}
2450 \begin{longtable}{l|c}
2451 \caption{Macinfo type encodings} \label{tab:macinfotypeencodings}\\
2452 \hline \bfseries Macinfo type name&\bfseries Value \\ \hline
2454 \bfseries Macinfo type name&\bfseries Value\\ \hline
2456 \hline \emph{Continued on next page}
2461 \DWMACINFOdefine&0x01 \\
2462 \DWMACINFOundef&0x02 \\
2463 \DWMACINFOstartfile&0x03 \\
2464 \DWMACINFOendfile&0x04 \\
2465 \DWMACINFOvendorext&\xff \\
2471 \section{Call Frame Information}
2472 \label{datarep:callframeinformation}
2474 In the \thirtytwobitdwarfformat, the value of the CIE id in the
2475 CIE header is \xffffffff; in the \sixtyfourbitdwarfformat, the
2476 value is \xffffffffffffffff.
2478 The value of the CIE \addtoindexi{version number}{version number!call frame information}
2479 is 4 (see Appendix \refersec{app:dwarfsectionversionnumbersinformative}).
2481 Call frame instructions are encoded in one or more bytes. The
2482 primary opcode is encoded in the high order two bits of
2483 the first byte (that is, opcode = byte $\gg$ 6). An operand
2484 or extended opcode may be encoded in the low order 6
2485 bits. Additional operands are encoded in subsequent bytes.
2486 The instructions and their encodings are presented in
2487 Table \refersec{tab:callframeinstructionencodings}.
2490 \setlength{\extrarowheight}{0.1cm}
2491 \begin{longtable}{l|c|c|l|l}
2492 \caption{Call frame instruction encodings} \label{tab:callframeinstructionencodings} \\
2493 \hline &\bfseries High 2 &\bfseries Low 6 & & \\
2494 \bfseries Instruction&\bfseries Bits &\bfseries Bits &\bfseries Operand 1 &\bfseries Operand 2\\ \hline
2496 & \bfseries High 2 &\bfseries Low 6 & &\\
2497 \bfseries Instruction&\bfseries Bits &\bfseries Bits &\bfseries Operand 1 &\bfseries Operand 2\\ \hline
2499 \hline \emph{Continued on next page}
2504 \DWCFAadvanceloc&0x1&delta & \\
2505 \DWCFAoffset&0x2®ister&ULEB128 offset \\
2506 \DWCFArestore&0x3®ister & & \\
2507 \DWCFAnop&0&0 & & \\
2508 \DWCFAsetloc&0&0x01&address & \\
2509 \DWCFAadvancelocone&0&0x02&1\dash byte delta & \\
2510 \DWCFAadvanceloctwo&0&0x03&2\dash byte delta & \\
2511 \DWCFAadvancelocfour&0&0x04&4\dash byte delta & \\
2512 \DWCFAoffsetextended&0&0x05&ULEB128 register&ULEB128 offset \\
2513 \DWCFArestoreextended&0&0x06&ULEB128 register & \\
2514 \DWCFAundefined&0&0x07&ULEB128 register & \\
2515 \DWCFAsamevalue&0&0x08 &ULEB128 register & \\
2516 \DWCFAregister&0&0x09&ULEB128 register &ULEB128 offset \\
2517 \DWCFArememberstate&0&0x0a & & \\
2518 \DWCFArestorestate&0&0x0b & & \\
2519 \DWCFAdefcfa&0&0x0c &ULEB128 register&ULEB128 offset \\
2520 \DWCFAdefcfaregister&0&0x0d&ULEB128 register & \\
2521 \DWCFAdefcfaoffset&0&0x0e &ULEB128 offset & \\
2522 \DWCFAdefcfaexpression&0&0x0f &BLOCK \\
2523 \DWCFAexpression&0&0x10&ULEB128 register & BLOCK \\
2525 \DWCFAoffsetextendedsf&0&0x11&ULEB128 register&SLEB128 offset \\
2526 \DWCFAdefcfasf&0&0x12&ULEB128 register&SLEB128 offset \\
2527 \DWCFAdefcfaoffsetsf&0&0x13&SLEB128 offset & \\
2528 \DWCFAvaloffset&0&0x14&ULEB128&ULEB128 \\
2529 \DWCFAvaloffsetsf&0&0x15&ULEB128&SLEB128 \\
2530 \DWCFAvalexpression&0&0x16&ULEB128&BLOCK \\
2531 \DWCFAlouser&0&0x1c & & \\
2532 \DWCFAhiuser&0&\xiiif & & \\
2536 \section{Non-contiguous Address Ranges}
2537 \label{datarep:noncontiguousaddressranges}
2539 Each entry in a \addtoindex{range list}
2540 (see Section \refersec{chap:noncontiguousaddressranges})
2542 \addtoindexx{base address selection entry!in range list}
2544 \addtoindexx{range list}
2545 a base address selection entry, or an end
2548 A \addtoindex{range list} entry consists of two relative addresses. The
2549 addresses are the same size as addresses on the target machine.
2551 A base address selection entry and an
2552 \addtoindexx{end of list entry!in range list}
2553 end of list entry each
2554 \addtoindexx{base address selection entry!in range list}
2555 consist of two (constant or relocated) addresses. The two
2556 addresses are the same size as addresses on the target machine.
2558 For a \addtoindex{range list} to be specified, the base address of the
2559 \addtoindexx{base address selection entry!in range list}
2560 corresponding compilation unit must be defined
2561 (see Section \refersec{chap:normalandpartialcompilationunitentries}).
2563 \section{String Offsets Table}
2564 \label{chap:stringoffsetstable}
2565 Each set of entries in the string offsets table contained in the
2566 \dotdebugstroffsets{} or \dotdebugstroffsetsdwo{}
2567 section begins with a header containing:
2568 \begin{enumerate}[1. ]
2569 \item \texttt{unit\_length} (\livelink{datarep:initiallengthvalues}{initial length}) \\
2570 A 4-byte or 12-byte length containing the length of
2571 the set of entries for this compilation unit, not
2572 including the length field itself. In the 32-bit
2573 DWARF format, this is a 4-byte unsigned integer
2574 (which must be less than \xfffffffzero); in the 64-bit
2575 DWARF format, this consists of the 4-byte value
2576 \wffffffff followed by an 8-byte unsigned integer
2577 that gives the actual length (see
2578 Section \refersec{datarep:32bitand64bitdwarfformats}).
2580 \item \texttt{version} (\addtoindex{uhalf}) \\
2581 A 2-byte version identifier containing the value
2582 \versiondotdebugstroffsets{}
2583 (see Appendix \refersec{app:dwarfsectionversionnumbersinformative}).
2584 \item \texttt{padding} (\addtoindex{uhalf}) \\
2587 This header is followed by a series of string table offsets
2588 that have the same representation as \DWFORMstrp.
2589 For the 32-bit DWARF format, each offset is 4 bytes long; for
2590 the 64-bit DWARF format, each offset is 8 bytes long.
2592 The \DWATstroffsetsbase{} attribute points to the first
2593 entry following the header. The entries are indexed
2594 sequentially from this base entry, starting from 0.
2596 \section{Address Table}
2597 \label{chap:addresstable}
2598 Each set of entries in the address table contained in the
2599 \dotdebugaddr{} section begins with a header containing:
2600 \begin{enumerate}[1. ]
2601 \item \texttt{unit\_length} (\livelink{datarep:initiallengthvalues}{initial length}) \\
2602 A 4-byte or 12-byte length containing the length of
2603 the set of entries for this compilation unit, not
2604 including the length field itself. In the 32-bit
2605 DWARF format, this is a 4-byte unsigned integer
2606 (which must be less than \xfffffffzero); in the 64-bit
2607 DWARF format, this consists of the 4-byte value
2608 \wffffffff followed by an 8-byte unsigned integer
2609 that gives the actual length (see
2610 Section \refersec{datarep:32bitand64bitdwarfformats}).
2613 \item \texttt{version} (\addtoindex{uhalf}) \\
2614 A 2-byte version identifier containing the value
2615 \versiondotdebugaddr{}
2616 (see Appendix \refersec{app:dwarfsectionversionnumbersinformative}).
2619 \item \texttt{address\_size} (\addtoindex{ubyte}) \\
2620 A 1-byte unsigned integer containing the size in
2621 bytes of an address (or the offset portion of an
2622 address for segmented addressing) on the target
2626 \item \texttt{segment\_size} (\addtoindex{ubyte}) \\
2627 A 1-byte unsigned integer containing the size in
2628 bytes of a segment selector on the target system.
2631 This header is followed by a series of segment/address pairs.
2632 The segment size is given by the \texttt{segment\_size} field of the
2633 header, and the address size is given by the \texttt{address\_size}
2634 field of the header. If the \texttt{segment\_size} field in the header
2635 is zero, the entries consist only of an addresses.
2637 The \DWATaddrbase{} attribute points to the first entry
2638 following the header. The entries are indexed sequentially
2639 from this base entry, starting from 0.
2641 \section{Range List Table}
2642 \label{app:rangelisttable}
2643 Each set of entries in the range list table contained in the
2644 \dotdebugranges{} section begins with a header containing:
2645 \begin{enumerate}[1. ]
2646 \item \texttt{unit\_length} (\livelink{datarep:initiallengthvalues}{initial length}) \\
2647 A 4-byte or 12-byte length containing the length of
2648 the set of entries for this compilation unit, not
2649 including the length field itself. In the 32-bit
2650 DWARF format, this is a 4-byte unsigned integer
2651 (which must be less than \xfffffffzero); in the 64-bit
2652 DWARF format, this consists of the 4-byte value
2653 \wffffffff followed by an 8-byte unsigned integer
2654 that gives the actual length (see
2655 Section \refersec{datarep:32bitand64bitdwarfformats}).
2658 \item \texttt{version} (\addtoindex{uhalf}) \\
2659 A 2-byte version identifier containing the value
2660 \versiondotdebugranges{}
2661 (see Appendix \refersec{app:dwarfsectionversionnumbersinformative}).
2664 \item \texttt{address\_size} (\addtoindex{ubyte}) \\
2665 A 1-byte unsigned integer containing the size in
2666 bytes of an address (or the offset portion of an
2667 address for segmented addressing) on the target
2671 \item \texttt{segment\_size} (\addtoindex{ubyte}) \\
2672 A 1-byte unsigned integer containing the size in
2673 bytes of a segment selector on the target system.
2676 This header is followed by a series of range list entries as
2677 described in Section \refersec{chap:locationlists}.
2678 The segment size is given by the
2679 \texttt{segment\_size} field of the header, and the address size is
2680 given by the \texttt{address\_size} field of the header. If the
2681 \texttt{segment\_size} field in the header is zero, the segment
2682 selector is omitted from the range list entries.
2684 The \DWATrangesbase{} attribute points to the first entry
2685 following the header. The entries are referenced by a byte
2686 offset relative to this base address.
2689 \section{Location List Table}
2690 \label{datarep:locationlisttable}
2691 Each set of entries in the location list table contained in the
2692 \dotdebugloc{} or \dotdebuglocdwo{} sections begins with a header containing:
2693 \begin{enumerate}[1. ]
2694 \item \texttt{unit\_length} (\livelink{datarep:initiallengthvalues}{initial length}) \\
2695 A 4-byte or 12-byte length containing the length of
2696 the set of entries for this compilation unit, not
2697 including the length field itself. In the 32-bit
2698 DWARF format, this is a 4-byte unsigned integer
2699 (which must be less than \xfffffffzero); in the 64-bit
2700 DWARF format, this consists of the 4-byte value
2701 \wffffffff followed by an 8-byte unsigned integer
2702 that gives the actual length (see
2703 Section \refersec{datarep:32bitand64bitdwarfformats}).
2706 \item \texttt{version} (\addtoindex{uhalf}) \\
2707 A 2-byte version identifier containing the value
2708 \versiondotdebugloc{}
2709 (see Appendix \refersec{app:dwarfsectionversionnumbersinformative}).
2712 \item \texttt{address\_size} (\addtoindex{ubyte}) \\
2713 A 1-byte unsigned integer containing the size in
2714 bytes of an address (or the offset portion of an
2715 address for segmented addressing) on the target
2719 \item \texttt{segment\_size} (\addtoindex{ubyte}) \\
2720 A 1-byte unsigned integer containing the size in
2721 bytes of a segment selector on the target system.
2724 This header is followed by a series of location list entries as
2725 described in Section \refersec{chap:locationlists}.
2726 The segment size is given by the
2727 \texttt{segment\_size} field of the header, and the address size is
2728 given by the \texttt{address\_size} field of the header. If the
2729 \texttt{segment\_size} field in the header is zero, the segment
2730 selector is omitted from the range list entries.
2732 The entries are referenced by a byte offset relative to the first
2733 location list following this header.
2736 \section{Dependencies and Constraints}
2737 \label{datarep:dependenciesandconstraints}
2739 The debugging information in this format is intended to
2741 \addtoindexx{DWARF section names!list of}
2751 \dotdebugpubnames{},
2752 \dotdebugpubtypes{},
2755 \dotdebugstroffsets{}
2758 sections of an object file, or equivalent
2759 separate file or database. The information is not
2760 word\dash aligned. Consequently:
2763 \item For the \thirtytwobitdwarfformat{} and a target architecture with
2764 32\dash bit addresses, an assembler or compiler must provide a way
2765 to produce 2\dash byte and 4\dash byte quantities without alignment
2766 restrictions, and the linker must be able to relocate a
2767 4\dash byte address or
2768 \addtoindexx{section offset!alignment of}
2769 section offset that occurs at an arbitrary
2772 \item For the \thirtytwobitdwarfformat{} and a target architecture with
2773 64\dash bit addresses, an assembler or compiler must provide a
2774 way to produce 2\dash byte, 4\dash byte and 8\dash byte quantities without
2775 alignment restrictions, and the linker must be able to relocate
2776 an 8\dash byte address or 4\dash byte
2777 \addtoindexx{section offset!alignment of}
2778 section offset that occurs at an
2779 arbitrary alignment.
2781 \item For the \sixtyfourbitdwarfformat{} and a target architecture with
2782 32\dash bit addresses, an assembler or compiler must provide a
2783 way to produce 2\dash byte, 4\dash byte and 8\dash byte quantities without
2784 alignment restrictions, and the linker must be able to relocate
2785 a 4\dash byte address or 8\dash byte
2786 \addtoindexx{section offset!alignment of}
2787 section offset that occurs at an
2788 arbitrary alignment.
2790 \textit{It is expected that this will be required only for very large
2791 32\dash bit programs or by those architectures which support
2792 a mix of 32\dash bit and 64\dash bit code and data within the same
2795 \item For the \sixtyfourbitdwarfformat{} and a target architecture with
2796 64\dash bit addresses, an assembler or compiler must provide a
2797 way to produce 2\dash byte, 4\dash byte and 8\dash byte quantities without
2798 alignment restrictions, and the linker must be able to
2799 relocate an 8\dash byte address or
2800 \addtoindexx{section offset!alignment of}
2801 section offset that occurs at
2802 an arbitrary alignment.
2805 \section{Integer Representation Names}
2806 \label{datarep:integerrepresentationnames}
2808 The sizes of the integers used in the lookup by name, lookup
2809 by address, line number and call frame information sections
2811 Table \ref{tab:integerrepresentationnames}.
2815 \setlength{\extrarowheight}{0.1cm}
2816 \begin{longtable}{c|l}
2817 \caption{Integer representation names} \label{tab:integerrepresentationnames}\\
2818 \hline \bfseries Representation name&\bfseries Representation \\ \hline
2820 \bfseries Representation name&\bfseries Representation\\ \hline
2822 \hline \emph{Continued on next page}
2827 \addtoindex{sbyte}& signed, 1\dash byte integer \\
2828 \addtoindex{ubyte}&unsigned, 1\dash byte integer \\
2829 \addtoindex{uhalf}&unsigned, 2\dash byte integer \\
2830 \addtoindex{uword}&unsigned, 4\dash byte integer \\
2836 \section{Type Signature Computation}
2837 \label{datarep:typesignaturecomputation}
2839 A type signature is computed only by the DWARF producer;
2840 \addtoindexx{type signature computation}
2841 it is used by a DWARF consumer to resolve type references to
2842 the type definitions that are contained in
2843 \addtoindexx{type unit}
2846 The type signature for a type T0 is formed from the
2847 \MDfive{}\footnote{\livetarg{def:MDfive}{MD5} Message Digest Algorithm,
2848 R.L. Rivest, RFC 1321, April 1992}
2849 hash of a flattened description of the type. The flattened
2850 description of the type is a byte sequence derived from the
2851 DWARF encoding of the type as follows:
2852 \begin{enumerate}[1. ]
2854 \item Start with an empty sequence S and a list V of visited
2855 types, where V is initialized to a list containing the type
2856 T0 as its single element. Elements in V are indexed from 1,
2859 \item If the debugging information entry represents a type that
2860 is nested inside another type or a namespace, append to S
2861 the type\textquoteright s context as follows: For each surrounding type
2862 or namespace, beginning with the outermost such construct,
2863 append the letter 'C', the DWARF tag of the construct, and
2864 the name (taken from
2865 \addtoindexx{name attribute}
2866 the \DWATname{} attribute) of the type
2867 \addtoindexx{name attribute}
2868 or namespace (including its trailing null byte).
2870 \item Append to S the letter 'D', followed by the DWARF tag of
2871 the debugging information entry.
2873 \item For each of the attributes in
2874 Table \refersec{tab:attributesusedintypesignaturecomputation}
2876 the debugging information entry, in the order listed,
2877 append to S a marker letter (see below), the DWARF attribute
2878 code, and the attribute value.
2881 \caption{Attributes used in type signature computation}
2882 \label{tab:attributesusedintypesignaturecomputation}
2883 \simplerule[\textwidth]
2885 \autocols[0pt]{c}{2}{l}{
2900 \DWATcontainingtype,
2904 \DWATdatamemberlocation,
2925 \DWATrvaluereference,
2929 \DWATstringlengthbitsize,
2930 \DWATstringlengthbytesize,
2935 \DWATvariableparameter,
2938 \DWATvtableelemlocation
2941 \simplerule[\textwidth]
2944 Note that except for the initial
2945 \DWATname{} attribute,
2946 \addtoindexx{name attribute}
2947 attributes are appended in order according to the alphabetical
2948 spelling of their identifier.
2950 If an implementation defines any vendor-specific attributes,
2951 any such attributes that are essential to the definition of
2952 the type should also be included at the end of the above list,
2953 in their own alphabetical suborder.
2955 An attribute that refers to another type entry T is processed
2956 as follows: (a) If T is in the list V at some V[x], use the
2957 letter 'R' as the marker and use the unsigned LEB128 encoding
2958 of x as the attribute value; otherwise, (b) use the letter 'T'
2959 as the marker, process the type T recursively by performing
2960 Steps 2 through 7, and use the result as the attribute value.
2962 Other attribute values use the letter 'A' as the marker, and
2963 the value consists of the form code (encoded as an unsigned
2964 LEB128 value) followed by the encoding of the value according
2965 to the form code. To ensure reproducibility of the signature,
2966 the set of forms used in the signature computation is limited
2967 to the following: \DWFORMsdata,
2972 \item If the tag in Step 3 is one of \DWTAGpointertype,
2973 \DWTAGreferencetype,
2974 \DWTAGrvaluereferencetype,
2975 \DWTAGptrtomembertype,
2976 or \DWTAGfriend, and the referenced
2977 type (via the \DWATtype{} or
2978 \DWATfriend{} attribute) has a
2979 \DWATname{} attribute, append to S the letter 'N', the DWARF
2980 attribute code (\DWATtype{} or
2981 \DWATfriend), the context of
2982 the type (according to the method in Step 2), the letter 'E',
2983 and the name of the type. For \DWTAGfriend, if the referenced
2984 entry is a \DWTAGsubprogram, the context is omitted and the
2985 name to be used is the ABI-specific name of the subprogram
2986 (for example, the mangled linker name).
2989 \item If the tag in Step 3 is not one of \DWTAGpointertype,
2990 \DWTAGreferencetype,
2991 \DWTAGrvaluereferencetype,
2992 \DWTAGptrtomembertype, or
2993 \DWTAGfriend, but has
2994 a \DWATtype{} attribute, or if the referenced type (via
2996 \DWATfriend{} attribute) does not have a
2997 \DWATname{} attribute, the attribute is processed according to
2998 the method in Step 4 for an attribute that refers to another
3002 \item Visit each child C of the debugging information
3003 entry as follows: If C is a nested type entry or a member
3004 function entry, and has
3005 a \DWATname{} attribute, append to
3006 \addtoindexx{name attribute}
3007 S the letter 'S', the tag of C, and its name; otherwise,
3008 process C recursively by performing Steps 3 through 7,
3009 appending the result to S. Following the last child (or if
3010 there are no children), append a zero byte.
3015 For the purposes of this algorithm, if a debugging information
3017 \DWATspecification{}
3018 attribute that refers to
3019 another entry D (which has a
3022 then S inherits the attributes and children of D, and S is
3023 processed as if those attributes and children were present in
3024 the entry S. Exception: if a particular attribute is found in
3025 both S and D, the attribute in S is used and the corresponding
3026 one in D is ignored.
3028 DWARF tag and attribute codes are appended to the sequence
3029 as unsigned LEB128 values, using the values defined earlier
3032 \textit{A grammar describing this computation may be found in
3033 Appendix \refersec{app:typesignaturecomputationgrammar}.
3036 \textit{An attribute that refers to another type entry should
3037 be recursively processed or replaced with the name of the
3038 referent (in Step 4, 5 or 6). If neither treatment applies to
3039 an attribute that references another type entry, the entry
3040 that contains that attribute should not be considered for a
3041 separate \addtoindex{type unit}.}
3043 \textit{If a debugging information entry contains an attribute from
3044 the list above that would require an unsupported form, that
3045 entry should not be considered for a separate
3046 \addtoindex{type unit}.}
3048 \textit{A type should be considered for a separate
3049 \addtoindex{type unit} only
3050 if all of the type entries that it contains or refers to in
3051 Steps 6 and 7 can themselves each be considered for a separate
3052 \addtoindex{type unit}.}
3055 Where the DWARF producer may reasonably choose two or more
3056 different forms for a given attribute, it should choose
3057 the simplest possible form in computing the signature. (For
3058 example, a constant value should be preferred to a location
3059 expression when possible.)
3061 Once the string S has been formed from the DWARF encoding,
3062 an \MDfive{} hash is computed for the string and the
3063 least significant 64 bits are taken as the type signature.
3065 \textit{The string S is intended to be a flattened representation of
3066 the type that uniquely identifies that type (that is, a different
3067 type is highly unlikely to produce the same string).}
3069 \textit{A debugging information entry should not be placed in a
3070 separate \addtoindex{type unit}
3071 if any of the following apply:}
3075 \item \textit{The entry has an attribute whose value is a location
3076 expression, and the location expression contains a reference to
3077 another debugging information entry (for example, a \DWOPcallref{}
3078 operator), as it is unlikely that the entry will remain
3079 identical across compilation units.}
3081 \item \textit{The entry has an attribute whose value refers
3082 to a code location or a \addtoindex{location list}.}
3084 \item \textit{The entry has an attribute whose value refers
3085 to another debugging information entry that does not represent
3091 \textit{Certain attributes are not included in the type signature:}
3094 \item \textit{The \DWATdeclaration{} attribute is not included because it
3095 indicates that the debugging information entry represents an
3096 incomplete declaration, and incomplete declarations should
3098 \addtoindexx{type unit}
3099 separate type units.}
3101 \item \textit{The \DWATdescription{} attribute is not included because
3102 it does not provide any information unique to the defining
3103 declaration of the type.}
3105 \item \textit{The \DWATdeclfile,
3107 \DWATdeclcolumn{} attributes are not included because they
3108 may vary from one source file to the next, and would prevent
3109 two otherwise identical type declarations from producing the
3110 same \MDfive{} hash.}
3112 \item \textit{The \DWATobjectpointer{} attribute is not included
3113 because the information it provides is not necessary for the
3114 computation of a unique type signature.}
3118 \textit{Nested types and some types referred to by a debugging
3119 information entry are encoded by name rather than by recursively
3120 encoding the type to allow for cases where a complete definition
3121 of the type might not be available in all compilation units.}
3123 \textit{If a type definition contains the definition of a member function,
3124 it cannot be moved as is into a type unit, because the member function
3125 contains attributes that are unique to that compilation unit.
3126 Such a type definition can be moved to a type unit by rewriting the DIE tree,
3127 moving the member function declaration into a separate declaration tree,
3128 and replacing the function definition in the type with a non-defining
3129 declaration of the function (as if the function had been defined out of
3132 An example that illustrates the computation of an \MDfive{} hash may be found in
3133 Appendix \refersec{app:usingtypeunits}.