Copyright © 2004, 2005 Free Standards Group
Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.1; with no Invariant Sections, with no Front-Cover Texts, and with no Back-Cover Texts. A copy of the license is included in the section entitled "GNU Free Documentation License".
Portions of the text are copyrighted by the following parties:
The Regents of the University of California
Free Software Foundation
Ian F. Darwin
Paul Vixie
BSDI (now Wind River)
Andrew G Morgan
Jean-loup Gailly and Mark Adler
Massachusetts Institute of Technology
These excerpts are being used in accordance with their respective licenses.
Linux is a trademark of Linus Torvalds.
UNIX a registered trademark of the Open Group in the United States and other countries.
LSB is a trademark of the Free Standards Group in the USA and other countries.
AMD is a trademark of Advanced Micro Devices, Inc.
Intel and Itanium are registered trademarks and Intel386 is a trademarks of Intel Corporation.
PowerPC and PowerPC Architecture are trademarks of the IBM Corporation.
OpenGL is a registered trademark of Silicon Graphics, Inc.
This is version 3.1 of the Linux Standard Base Core Specification. This specification is part of a family of specifications under the general title "Linux Standard Base". Developers of applications or implementations interested in using the LSB trademark should see the Free Standards Group Certification Policy for details.
The LSB defines a binary interface for application programs that are compiled and packaged for LSB-conforming implementations on many different hardware architectures. Since a binary specification shall include information specific to the computer processor architecture for which it is intended, it is not possible for a single document to specify the interface for all possible LSB-conforming implementations. Therefore, the LSB is a family of specifications, rather than a single one.
This document should be used in conjunction with the documents it references. This document enumerates the system components it includes, but descriptions of those components may be included entirely or partly in this document, partly in other documents, or entirely in other reference documents. For example, the section that describes system service routines includes a list of the system routines supported in this interface, formal declarations of the data structures they use that are visible to applications, and a pointer to the underlying referenced specification for information about the syntax and semantics of each call. Only those routines not described in standards referenced by this document, or extensions to those standards, are described in the detail. Information referenced in this way is as much a part of this document as is the information explicitly included here.
The specification carries a version number of either the form x.y or x.y.z. This version number carries the following meaning:
The first number (x) is the major version number. All versions with the same major version number should share binary compatibility. Any addition or deletion of a new library results in a new version number. Interfaces marked as deprecated may be removed from the specification at a major version change.
The second number (y) is the minor version number. Individual interfaces may be added if all certified implementations already had that (previously undocumented) interface. Interfaces may be marked as deprecated at a minor version change. Other minor changes may be permitted at the discretion of the LSB workgroup.
The third number (z), if present, is the editorial level. Only editorial changes should be included in such versions.
Since this specification is a descriptive Application Binary Interface, and not a source level API specification, it is not possible to make a guarantee of 100% backward compatibility between major releases. However, it is the intent that those parts of the binary interface that are visible in the source level API will remain backward compatible from version to version, except where a feature marked as "Deprecated" in one release may be removed from a future release.
Implementors are strongly encouraged to make use of symbol versioning to permit simultaneous support of applications conforming to different releases of this specification.
The Linux Standard Base (LSB) defines a system interface for compiled applications and a minimal environment for support of installation scripts. Its purpose is to enable a uniform industry standard environment for high-volume applications conforming to the LSB.
These specifications are composed of two basic parts: A common specification ("LSB-generic" or "generic LSB") describing those parts of the interface that remain constant across all implementations of the LSB, and an architecture-specific supplement ("LSB-arch" or "archLSB") describing the parts of the interface that vary by processor architecture. Together, the LSB-generic and the architecture-specific supplement for a single hardware architecture provide a complete interface specification for compiled application programs on systems that share a common hardware architecture.
The LSB-generic document shall be used in conjunction with an architecture-specific supplement. Whenever a section of the LSB-generic specification shall be supplemented by architecture-specific information, the LSB-generic document includes a reference to the architecture supplement. Architecture supplements may also contain additional information that is not referenced in the LSB-generic document.
The LSB contains both a set of Application Program Interfaces (APIs) and Application Binary Interfaces (ABIs). APIs may appear in the source code of portable applications, while the compiled binary of that application may use the larger set of ABIs. A conforming implementation shall provide all of the ABIs listed here. The compilation system may replace (e.g. by macro definition) certain APIs with calls to one or more of the underlying binary interfaces, and may insert calls to binary interfaces as needed.
The LSB is primarily a binary interface definition. Not all of the source level APIs available to applications may be contained in this specification.
This is the Core module of the Linux Standards Base (LSB). This module provides the fundamental system interfaces, libraries, and runtime environment upon which all conforming applications and libraries depend.
Interfaces described in this module are mandatory except where explicitly listed otherwise. Core interfaces may be supplemented by other modules; all modules are built upon the core.
The following referenced documents are indispensable for the application of this document. For dated references, only the edition cited applies. For undated references, the latest edition of the referenced document (including any amendments) applies.
Note: Where copies of a document are available on the World Wide Web, a Uniform Resource Locator (URL) is given for informative purposes only. This may point to a more recent copy of the referenced specification, or may be out of date. Reference copies of specifications at the revision level indicated may be found at the Free Standards Group's Reference Specifications site.
Table 2-1. Normative References
| Name | Title | URL |
|---|---|---|
| Filesystem Hierarchy Standard | Filesystem Hierarchy Standard (FHS) 2.3 | http://www.pathname.com/fhs/ |
| IEC 60559/IEEE 754 Floating Point | IEC 60559:1989 Binary floating-point arithmetic for microprocessor systems | http://www.ieee.org/ |
| ISO C (1999) | ISO/IEC 9899: 1999, Programming Languages --C | |
| ISO POSIX (2003) | ISO/IEC 9945-1:2003 Information technology -- Portable Operating System Interface (POSIX) -- Part 1: Base Definitions ISO/IEC 9945-2:2003 Information technology -- Portable Operating System Interface (POSIX) -- Part 2: System Interfaces ISO/IEC 9945-3:2003 Information technology -- Portable Operating System Interface (POSIX) -- Part 3: Shell and Utilities ISO/IEC 9945-4:2003 Information technology -- Portable Operating System Interface (POSIX) -- Part 4: Rationale Including Technical Cor. 1: 2004 | http://www.unix.org/version3/ |
| Itanium C++ ABI | Itanium C++ ABI (Revision 1.83) | http://refspecs.freestandards.org/cxxabi-1.83.html |
| Large File Support | Large File Support | http://www.UNIX-systems.org/version2/whatsnew/lfs20mar.html |
| SUSv2 | CAE Specification, January 1997, System Interfaces and Headers (XSH),Issue 5 (ISBN: 1-85912-181-0, C606) | http://www.opengroup.org/publications/catalog/un.htm |
| SUSv2 Commands and Utilities | The Single UNIX® Specification(SUS) Version 2, Commands and Utilities (XCU), Issue 5 (ISBN: 1-85912-191-8, C604) | http://www.opengroup.org/publications/catalog/un.htm |
| SVID Issue 3 | American Telephone and Telegraph Company, System V Interface Definition, Issue 3 ; Morristown, NJ, UNIX Press, 1989.(ISBN 0201566524) | |
| SVID Issue 4 | System V Interface Definition,Fourth Edition | |
| System V ABI | System V Application Binary Interface, Edition 4.1 | http://www.caldera.com/developers/devspecs/gabi41.pdf |
| System V ABI Update | System V Application Binary Interface - DRAFT - 17 December 2003 | http://www.caldera.com/developers/gabi/2003-12-17/contents.html |
| X/Open Curses | CAE Specification, May 1996, X/Open Curses, Issue 4, Version 2 (ISBN: 1-85912-171-3, C610), plus Corrigendum U018 | http://www.opengroup.org/publications/catalog/un.htm |
In addition, the specifications listed below provide essential background information to implementors of this specification. These references are included for information only.
Table 2-2. Other References
| Name | Title | URL |
|---|---|---|
| DWARF Debugging Information Format, Revision 2.0.0 | DWARF Debugging Information Format, Revision 2.0.0 (July 27, 1993) | http://refspecs.freestandards.org/dwarf/dwarf-2.0.0.pdf |
| DWARF Debugging Information Format, Revision 3.0.0 (Draft) | DWARF Debugging Information Format, Revision 3.0.0 (Draft) | http://refspecs.freestandards.org/dwarf/ |
| ISO/IEC TR14652 | ISO/IEC Technical Report 14652:2002 Specification method for cultural conventions | |
| ITU-T V.42 | International Telecommunication Union Recommendation V.42 (2002): Error-correcting procedures for DCEs using asynchronous-to-synchronous conversionITUV | http://www.itu.int/rec/recommendation.asp?type=folders&lang=e&parent=T-REC-V.42 |
| Li18nux Globalization Specification | LI18NUX 2000 Globalization Specification, Version 1.0 with Amendment 4 | http://www.li18nux.org/docs/html/LI18NUX-2000-amd4.htm |
| Linux Allocated Device Registry | LINUX ALLOCATED DEVICES | http://www.lanana.org/docs/device-list/devices.txt |
| PAM | Open Software Foundation, Request For Comments: 86.0 , October 1995, V. Samar & R.Schemers (SunSoft) | http://www.opengroup.org/tech/rfc/mirror-rfc/rfc86.0.txt |
| RFC 1321: The MD5 Message-Digest Algorithm | IETF RFC 1321: The MD5 Message-Digest Algorithm | http://www.ietf.org/rfc/rfc1321.txt |
| RFC 1831/1832 RPC & XDR | IETF RFC 1831 & 1832 | http://www.ietf.org/ |
| RFC 1833: Binding Protocols for ONC RPC Version 2 | IETF RFC 1833: Binding Protocols for ONC RPC Version 2 | http://www.ietf.org/rfc/rfc1833.txt |
| RFC 1950: ZLIB Compressed Data Format Specication | IETF RFC 1950: ZLIB Compressed Data Format Specification | http://www.ietf.org/rfc/rfc1950.txt |
| RFC 1951: DEFLATE Compressed Data Format Specification | IETF RFC 1951: DEFLATE Compressed Data Format Specification version 1.3 | http://www.ietf.org/rfc/rfc1951.txt |
| RFC 1952: GZIP File Format Specification | IETF RFC 1952: GZIP file format specification version 4.3 | http://www.ietf.org/rfc/rfc1952.txt |
| RFC 2440: OpenPGP Message Format | IETF RFC 2440: OpenPGP Message Format | http://www.ietf.org/rfc/rfc2440.txt |
| RFC 2821:Simple Mail Transfer Protocol | IETF RFC 2821: Simple Mail Transfer Protocol | http://www.ietf.org/rfc/rfc2821.txt |
| RFC 2822:Internet Message Format | IETF RFC 2822: Internet Message Format | http://www.ietf.org/rfc/rfc2822.txt |
| RFC 791:Internet Protocol | IETF RFC 791: Internet Protocol Specification | http://www.ietf.org/rfc/rfc791.txt |
| RPM Package Format | RPM Package Format V3.0 | http://www.rpm.org/max-rpm/s1-rpm-file-format-rpm-file-format.html |
| zlib Manual | zlib 1.2 Manual | http://www.gzip.org/zlib/ |
The libraries listed in Table 3-1 shall be available on a Linux Standard Base system, with the specified runtime names. The libraries listed in Table 3-2 are architecture specific, but shall be available on all LSB conforming systems. This list may be supplemented or amended by the architecture specific supplement.
Table 3-1. Standard Library Names
| Library | Runtime Name |
|---|---|
| libdl | libdl.so.2 |
| libcrypt | libcrypt.so.1 |
| libz | libz.so.1 |
| libncurses | libncurses.so.5 |
| libutil | libutil.so.1 |
| libpthread | libpthread.so.0 |
| librt | librt.so.1 |
| libpam | libpam.so.0 |
| libgcc_s | libgcc_s.so.1 |
Table 3-2. Standard Library Names defined in the Architecture Specific Supplement
| Library | Runtime Name |
|---|---|
| libm | See archLSB |
| libc | See archLSB |
| proginterp | See archLSB |
These libraries will be in an implementation-defined directory which the dynamic linker shall search by default.
A conforming implementation is necessarily architecture specific, and must provide the interfaces specified by both the generic LSB Core specification and its relevant architecture specific supplement.
Rationale: An implementation must provide at least the interfaces specified in these specifications. It may also provide additional interfaces.
A conforming implementation shall satisfy the following requirements:
A processor architecture represents a family of related processors which may not have identical feature sets. The architecture specific supplement to this specification for a given target processor architecture describes a minimum acceptable processor. The implementation shall provide all features of this processor, whether in hardware or through emulation transparent to the application.
The implementation shall be capable of executing compiled applications having the format and using the system interfaces described in this document.
The implementation shall provide libraries containing the interfaces specified by this document, and shall provide a dynamic linking mechanism that allows these interfaces to be attached to applications at runtime. All the interfaces shall behave as specified in this document.
The map of virtual memory provided by the implementation shall conform to the requirements of this document.
The implementation's low-level behavior with respect to function call linkage, system traps, signals, and other such activities shall conform to the formats described in this document.
The implementation shall provide all of the mandatory interfaces in their entirety.
The implementation may provide one or more of the optional interfaces. Each optional interface that is provided shall be provided in its entirety. The product documentation shall state which optional interfaces are provided.
The implementation shall provide all files and utilities specified as part of this document in the format defined here and in other referenced documents. All commands and utilities shall behave as required by this document. The implementation shall also provide all mandatory components of an application's runtime environment that are included or referenced in this document.
The implementation, when provided with standard data formats and values at a named interface, shall provide the behavior defined for those values and data formats at that interface. However, a conforming implementation may consist of components which are separately packaged and/or sold. For example, a vendor of a conforming implementation might sell the hardware, operating system, and windowing system as separately packaged items.
The implementation may provide additional interfaces with different names. It may also provide additional behavior corresponding to data values outside the standard ranges, for standard named interfaces.
A conforming application is necessarily architecture specific, and must conform to both the generic LSB Core specification and its relevant architecture specific supplement.
A conforming application shall satisfy the following requirements:
Its executable files shall be either shell scripts or object files in the format defined for the Object File Format system interface.
Its object files shall participate in dynamic linking as defined in the Program Loading and Linking System interface.
It shall employ only the instructions, traps, and other low-level facilities defined in the Low-Level System interface as being for use by applications.
If it requires any optional interface defined in this document in order to be installed or to execute successfully, the requirement for that optional interface shall be stated in the application's documentation.
It shall not use any interface or data format that is not required to be provided by a conforming implementation, unless:
If such an interface or data format is supplied by another application through direct invocation of that application during execution, that application shall be in turn an LSB conforming application.
The use of that interface or data format, as well as its source, shall be identified in the documentation of the application.
It shall not use any values for a named interface that are reserved for vendor extensions.
For the purposes of this document, the following definitions, as specified in the ISO/IEC Directives, Part 2, 2001, 4th Edition, apply:
For the purposes of this document, the following terms apply:
The architectural part of the LSB Specification which describes the specific parts of the interface that are platform specific. The archLSB is complementary to the gLSB.
The total set of interfaces that are available to be used in the compiled binary code of a conforming application.
The common part of the LSB Specification that describes those parts of the interface that remain constant across all hardware implementations of the LSB.
Describes a value or behavior that is not defined by this document but is selected by an implementor. The value or behavior may vary among implementations that conform to this document. An application should not rely on the existence of the value or behavior. An application that relies on such a value or behavior cannot be assured to be portable across conforming implementations. The implementor shall document such a value or behavior so that it can be used correctly by an application.
A file that is read by an interpreter (e.g., awk). The first line of the shell script includes a reference to its interpreter binary.
The set of interfaces that are available to be used in the source code of a conforming application.
Describes the nature of a value or behavior not defined by this document which results from use of an invalid program construct or invalid data input. The value or behavior may vary among implementations that conform to this document. An application should not rely on the existence or validity of the value or behavior. An application that relies on any particular value or behavior cannot be assured to be portable across conforming implementations.
Describes the nature of a value or behavior not specified by this document which results from use of a valid program construct or valid data input. The value or behavior may vary among implementations that conform to this document. An application should not rely on the existence or validity of the value or behavior. An application that relies on any particular value or behavior cannot be assured to be portable across conforming implementations.
Other terms and definitions used in this document shall have the same meaning as defined in Chapter 3 of the Base Definitions volume of ISO POSIX (2003).
Throughout this document, the following typographic conventions are used:
| function() | the name of a function | |
| command | the name of a command or utility | |
CONSTANT | a constant value | |
| parameter | a parameter | |
variable | a variable |
Throughout this specification, several tables of interfaces are presented. Each entry in these tables has the following format:
| name | the name of the interface | |
| (symver) | An optional symbol version identifier, if required. | |
| [refno] | A reference number indexing the table of referenced specifications that follows this table. |
For example,
refers to the interface named forkpty() with symbol versionGLIBC_2.0 that is defined in the
SUSv3 reference.
Note: Symbol versions are defined in the architecture specific supplements only.
This specification includes many interfaces described in ISO POSIX (2003). Unless otherwise specified, such interfaces should behave exactly as described in that specification. Any conflict between the requirements described here and the ISO POSIX (2003) standard is unintentional, except as explicitly noted otherwise.
Note: In addition to the differences noted inline in this specification, PDTR 24715 has extracted the differences between this specification and ISO POSIX (2003) into a single place. It is the long term plan of the Free Standards Group to converge the LSB Core Specification with ISO/IEC 9945 POSIX.
The LSB Specification Authority is responsible for deciding the meaning of conformance to normative referenced standards in the LSB context. Problem Reports regarding underlying or referenced standards in any other context will be referred to the relevant maintenance body for that standard.
The LSB is the base for several other specification projects under the umbrella of the Free Standards Group (FSG). This specification is the foundation, and other specifications build on the interfaces defined here. However, beyond those specifications listed as Normative References, this specification has no dependencies on other FSG projects.
Executable and Linking Format (ELF) defines the object format for compiled applications. This specification supplements the information found in System V ABI Update and is intended to document additions made since the publication of that document.
LSB-conforming applications shall assume that stack, heap and other allocated memory regions will be non-executable. The application must take steps to make them executable if needed.
LSB-conforming applications shall use the data representation as defined in the Arcitecture specific ELF documents.
In addition to the fundamental types specified in the architecture specific supplement, a 1 byte data type is defined here.
LSB-conforming implementations shall support the object file Executable and Linking Format (ELF), which is defined by the following documents:
this specification
an architecture specific supplement to this specification
As described in System V ABI, an ELF object file contains a number of sections.
The section header table is an array of
Elf32_Shdr or
Elf64_Shdr structures as
described in System V ABI. The
sh_type member shall be either a value from
Table 11-1, drawn from the System V
ABI, or one of the additional values specified in Table 11-2.
A section header's sh_type member specifies the sections's semantics.
The following section types are defined in the System V ABI and the System V ABI Update.
Table 11-1. ELF Section Types
Various sections hold program and control information. Sections in the lists below are used by the system and have the indicated types and attributes.
The following sections are defined in the System V ABI and the System V ABI Update.
Table 11-3. ELF Special Sections
| Name | Type | Attributes |
|---|---|---|
| .bss | SHT_NOBITS | SHF_ALLOC+SHF_WRITE |
| .comment | SHT_PROGBITS | 0 |
| .data | SHT_PROGBITS | SHF_ALLOC+SHF_WRITE |
| .data1 | SHT_PROGBITS | SHF_ALLOC+SHF_WRITE |
| .debug | SHT_PROGBITS | 0 |
| .dynamic | SHT_DYNAMIC | SHF_ALLOC+SHF_WRITE |
| .dynstr | SHT_STRTAB | SHF_ALLOC |
| .dynsym | SHT_DYNSYM | SHF_ALLOC |
| .fini | SHT_PROGBITS | SHF_ALLOC+SHF_EXECINSTR |
| .fini_array | SHT_FINI_ARRAY | SHF_ALLOC+SHF_WRITE |
| .hash | SHT_HASH | SHF_ALLOC |
| .init | SHT_PROGBITS | SHF_ALLOC+SHF_EXECINSTR |
| .init_array | SHT_INIT_ARRAY | SHF_ALLOC+SHF_WRITE |
| .interp | SHT_PROGBITS | SHF_ALLOC |
| .line | SHT_PROGBITS | 0 |
| .note | SHT_NOTE | 0 |
| .preinit_array | SHT_PREINIT_ARRAY | SHF_ALLOC+SHF_WRITE |
| .rodata | SHT_PROGBITS | SHF_ALLOC |
| .rodata1 | SHT_PROGBITS | SHF_ALLOC |
| .shstrtab | SHT_STRTAB | 0 |
| .strtab | SHT_STRTAB | SHF_ALLOC |
| .symtab | SHT_SYMTAB | SHF_ALLOC |
| .tbss | SHT_NOBITS | SHF_ALLOC+SHF_WRITE+SHF_TLS |
| .tdata | SHT_PROGBITS | SHF_ALLOC+SHF_WRITE+SHF_TLS |
| .text | SHT_PROGBITS | SHF_ALLOC+SHF_EXECINSTR |
Object files in an LSB conforming application may also contain one or more of the additional special sections described below.
Table 11-4. Additional Special Sections
| Name | Type | Attributes |
|---|---|---|
| .ctors | SHT_PROGBITS | SHF_ALLOC+SHF_WRITE |
| .data.rel.ro | SHT_PROGBITS | SHF_ALLOC+SHF_WRITE |
| .dtors | SHT_PROGBITS | SHF_ALLOC+SHF_WRITE |
| .eh_frame | SHT_PROGBITS | SHF_ALLOC |
| .eh_frame_hdr | SHT_PROGBITS | SHF_ALLOC |
| .gcc_except_table | SHT_PROGBITS | SHF_ALLOC |
| .gnu.version | SHT_GNU_versym | SHF_ALLOC |
| .gnu.version_d | SHT_GNU_verdef | SHF_ALLOC |
| .gnu.version_r | SHT_GNU_verneed | SHF_ALLOC |
| .got.plt | SHT_PROGBITS | SHF_ALLOC+SHF_WRITE |
| .jcr | SHT_PROGBITS | SHF_ALLOC+SHF_WRITE |
| .note.ABI-tag | SHT_NOTE | SHF_ALLOC |
| .stab | SHT_PROGBITS | 0 |
| .stabstr | SHT_STRTAB | 0 |
| .ctors | This section contains a list of global constructor function pointers. | |
| .data.rel.ro | This section holds initialized data that contribute to the program's memory image. This section may be made read-only after relocations have been applied. | |
| .dtors | This section contains a list of global destructor function pointers. | |
| .eh_frame | This section contains information necessary for frame unwinding during exception handling. See Section 11.6.1. | |
| .eh_frame_hdr | This section contains a pointer to the .eh_frame section which is accessible to the runtime support code of a C++ application. This section may also contain a binary search table which may be used by the runtime support code to more efficiently access records in the .eh_frame section. See Section 11.6.2. | |
| .gcc_except_table | This section holds Language Specific Data. | |
| .gnu.version | This section contains the Symbol Version Table. See Section 11.7.2. | |
| .gnu.version_d | This section contains the Version Definitions. See Section 11.7.3. | |
| .gnu.version_r | This section contains the Version Requirements. See Section 11.7.4. | |
| .got.plt | This section holds the read-only portion of the GLobal Offset Table. This section may be made read-only after relocations have been applied. | |
| .jcr | This section contains information necessary for registering compiled Java classes. The contents are compiler-specific and used by compiler initialization functions. | |
| .note.ABI-tag | Specify ABI details. See Section 11.8. | |
| .stab | This section contains debugging information. The contents are not specified as part of the LSB. | |
| .stabstr | This section contains strings associated with the debugging infomation contained in the .stab section. |
Symbols in a source program are translated by the compilation system into symbols that exist in the object file.
The LSB does not specify debugging information, however, some additional sections contain information which is encoded using the the encoding as specified by DWARF Debugging Information Format, Revision 2.0.0 with extensions defined here.
Note: The extensions specified here also exist in DWARF Debugging Information Format, Revision 3.0.0 (Draft). It is expected that future versions of the LSB will reference the final version of that document, and that the definitions here will be taken from that document instead of being specified here.
The DWARF Exception Header Encoding is used to describe the type of data used in the .eh_frame and .eh_frame_hdr section. The upper 4 bits indicate how the value is to be applied. The lower 4 bits indicate the format of the data.
Table 11-5. DWARF Exception Header value format
| Name | Value | Meaning |
|---|---|---|
| DW_EH_PE_absptr | 0x00 | The Value is a literal pointer whose size is determined by the architecture. |
| DW_EH_PE_uleb128 | 0x01 | Unsigned value is encoded using the Little Endian Base 128 (LEB128) as defined by DWARF Debugging Information Format, Revision 2.0.0. |
| DW_EH_PE_udata2 | 0x02 | A 2 bytes unsigned value. |
| DW_EH_PE_udata4 | 0x03 | A 4 bytes unsigned value. |
| DW_EH_PE_udata8 | 0x04 | An 8 bytes unsigned value. |
| DW_EH_PE_sleb128 | 0x09 | Signed value is encoded using the Little Endian Base 128 (LEB128) as defined by DWARF Debugging Information Format, Revision 2.0.0. |
| DW_EH_PE_sdata2 | 0x0A | A 2 bytes signed value. |
| DW_EH_PE_sdata4 | 0x0B | A 4 bytes signed value. |
| DW_EH_PE_sdata8 | 0x0C | An 8 bytes signed value. |
Table 11-6. DWARF Exception Header application
| Name | Value | Meaning |
|---|---|---|
| DW_EH_PE_pcrel | 0x10 | Value is relative to the current program counter. |
| DW_EH_PE_textrel | 0x20 | Value is relative to the beginning of the .text section. |
| DW_EH_PE_datarel | 0x30 | Value is relative to the beginning of the .got or .eh_frame_hdr section. |
| DW_EH_PE_funcrel | 0x40 | Value is relative to the beginning of the function. |
| DW_EH_PE_aligned | 0x50 | Value is aligned to an address unit sized boundary. |
One special encoding, 0xff (DW_EH_PE_omit), shall be used to indicate that no value ispresent.
In addition to the Call Frame Instructions defined in section 6.4.2 of DWARF Debugging Information Format, Revision 2.0.0, the following additional Call Frame Instructions may also be used.
Table 11-7. Additional DWARF Call Frame Instructions
| Name | Value | Meaning |
|---|---|---|
| DW_CFA_expression | 0x10 | The DW_CFA_expression instruction takes two operands: an unsigned LEB128 value representing a register number, and a DW_FORM_block value representing a DWARF expression. The required action is to establish the DWARF expression as the means by which the address in which the given register contents are found may be computed. The value of the CFA is pushed on the DWARF evaluation stack prior to execution of the DWARF expression. The DW_OP_call2, DW_OP_call4, DW_OP_call_ref and DW_OP_push_object_address DWARF operators (see Section 2.4.1 of DWARF Debugging Information Format, Revision 2.0.0) cannot be used in such a DWARF expression. |
| DW_CFA_offset_extended_sf | 0x11 | The DW_CFA_offset_extended_sf instruction takes two operands: an unsigned LEB128 value representing a register number and a signed LEB128 factored offset. This instruction is identical to DW_CFA_offset_extended except that the second operand is signed. |
| DW_CFA_def_cfa_sf | 0x12 | The DW_CFA_def_cfa_sf instruction takes two operands: an unsigned LEB128 value representing a register number and a signed LEB128 factored offset. This instruction is identical to DW_CFA_def_cfa except that the second operand is signed and factored. |
| DW_CFA_def_cfa_offset_sf | 0x13 | The DW_CFA_def_cfa_offset_sf instruction takes a signed LEB128 operand representing a factored offset. This instruction is identical to DW_CFA_def_cfa_offset except that the operand is signed and factored. |
| DW_CFA_GNU_args_size | 0x2e | The DW_CFA_GNU_args_size instruction takes an unsigned LEB128 operand representing an argument size. This instruction specifies the total of the size of the arguments which have been pushed onto the stack. |
| DW_CFA_GNU_negative_offset_extended | 0x2f | The DW_CFA_def_cfa_sf instruction takes two operands: an unsigned LEB128 value representing a register number and an unsigned LEB128 which represents the magnitude of the offset. This instruction is identical to DW_CFA_offset_extended_sf except that the operand is subtracted to produce the offset. This instructions is obsoleted by DW_CFA_offset_extended_sf. |
When using languages that support exceptions, such as C++, additional information must be provided to the runtime environment that describes the call frames that must be unwound during the processing of an exception. This information is contained in the special sections .eh_frame and .eh_framehdr.
Note: The format of the .eh_frame section is similar in format and purpose to the .debug_frame section which is specified in DWARF Debugging Information Format, Revision 3.0.0 (Draft). Readers are advised that there are some subtle difference, and care should be taken when comparing the two sections.
The .eh_frame section shall contain 1 or more Call Frame Information (CFI) records. The number of records present shall be determined by size of the section as contained in the section header. Each CFI record contains a Common Information Entry (CIE) record followed by 1 or more Frame Description Entry (FDE) records. Both CIEs and FDEs shall be aligned to an addressing unit sized boundary.
Table 11-8. Call Frame Information Format
| Common Information Entry Record |
| Frame Description Entry Record(s) |
Table 11-9. Common Information Entry Format
| Length | Required |
| Extended Length | Optional |
| CIE ID | Required |
| Version | Required |
| Augmentation String | Required |
| Code Alignment Factor | Required |
| Data Alignment Factor | Required |
| Return Address Register | Required |
| Augmentation Data Length | Optional |
| Augmentation Data | Optional |
| Initial Instructions | Required |
| Padding |
LengthA 4 byte unsigned value indicating the length in bytes of the CIE structure,
not including the Length field itself. If
Length contains the value 0xffffffff, then the
length is contained in the Extended Length field.
If Length contains the value 0, then this CIE shall
be considered a terminator and processing shall end.
Extended LengthA 8 byte unsigned value indicating the length in bytes of the CIE structure,
not including the Length and
Extended Length fields.
CIE IDA 4 byte unsigned value that is used to distinguish CIE records from FDE records. This value shall always be 0, which indicates this record is a CIE.
VersionA 1 byte value that identifies the version number of the frame information structure. This value shall be 1.
Augmentation StringThis value is a NUL terminated string that identifies the augmentation to the CIE or to the FDEs associated with this CIE. A zero length string indicates that no augmentation data is present. The augmentation string is case sensitive and shall be interpreted as described below.
Code Alignment FactorAn unsigned LEB128 encoded value that is factored out of all advance location instructions that are associated with this CIE or its FDEs. This value shall be multiplied by the delta argument of an adavance location instruction to obtain the new location value.
Data Alignment FactorA signed LEB128 encoded value that is factored out of all offset instructions that are associated with this CIE or its FDEs. This value shall be multiplied by the register offset argument of an offset instruction to obtain the new offset value.
Augmentation LengthAn unsigned LEB128 encoded value indicating the length in bytes of the Augmentation Data. This field is only present if the Augmentation String contains the character 'z'.
Augmentation DataA block of data whose contents are defined by the contents of the Augmentation String as described below. This field is only present if the Augmentation String contains the character 'z'. The size of this data is given by the Augentation Length.
Initial InstructionsInitial set of Call Frame Instructions. The number of instructions is determined by the remaining space in the CIE record.
PaddingExtra bytes to align the CIE structure to an addressing unit size boundary.
The Agumentation String indicates the presence of some optional fields, and how those fields should be intepreted. This string is case sensitive. Each character in the augmentation string in the CIE can be interpreted as below:
Table 11-10. Frame Description Entry Format
| Length | Required |
| Extended Length | Optional |
| CIE Pointer | Required |
| PC Begin | Required |
| PC Range | Required |
| Augmentation Data Length | Optional |
| Augmentation Data | Optional |
| Call Frame Instructions | Required |
| Padding |
LengthA 4 byte unsigned value indicating the length in bytes of the CIE structure,
not including the Length field itself. If
Length contains the value 0xffffffff, then the
length is contained the Extended Length field.
If Length contains the value 0, then this CIE shall
be considered a terminator and processing shall end.
Extended LengthA 8 byte unsigned value indicating the length in bytes of the CIE structure,
not including the Length field itself.
CIE PointerA 4 byte unsigned value that when subtracted from the offset of the current FDE yields the offset of the start of the associated CIE. This value shall never be 0.
PC BeginAn encoded value that indicates the address of the initial location associated with this FDE. The encoding format is specified in the Augmentation Data.
PC RangeAn absolute value that indicates the number of bytes of instructions associated with this FDE.
Augmentation LengthAn unsigned LEB128 encoded value indicating the length in bytes of the Augmentation Data. This field is only present if the Augmentation String in the associated CIE contains the character 'z'.
Augmentation DataA block of data whose contents are defined by the contents of the Augmentation String in the associated CIE as described above. This field is only present if the Augmentation String in the associated CIE contains the character 'z'. The size of this data is given by the Augentation Length.
Call Frame InstructionsA set of Call Frame Instructions.
PaddingExtra bytes to align the FDE structure to an addressing unit size boundary.
The .eh_frame_hdr section contains additional information about the .eh_frame section. A pointer to the start of the .eh_frame data, and optionally, a binary search table of pointers to the .eh_frame records are found in this section.
Data in this section is encoded according to Section 11.5.1.
Table 11-11. .eh_frame_hdr Section Format
| Encoding | Field |
|---|---|
| unsigned byte | version |
| unsigned byte | eh_frame_ptr_enc |
| unsigned byte | fde_count_enc |
| unsigned byte | table_enc |
| encoded | eh_frame_ptr |
| encoded | fde_count |
| binary search table |
This chapter describes the Symbol Versioning mechanism. All ELF objects may provide or depend on versioned symbols. Symbol Versioning is implemented by 3 section types: SHT_GNU_versym, SHT_GNU_verdef, and SHT_GNU_verneed.
The prefix Elfxx in the following descriptions and code fragments stands for either "Elf32" or "Elf64", depending on the architecture.
Versions are described by strings. The structures that are used for symbol versions also contain a member that holds the ELF hashing values of the strings. This allows for more efficient processing.
The special section .gnu.version which has a section type of SHT_GNU_versym shall contain the Symbol Version Table. This section shall have the same number of entries as the Dynamic Symbol Table in the .dynsym section.
The .gnu.version section shall contain an array of elements of type Elfxx_Half. Each entry specifies the version defined for or required by the corresponding symbol in the Dynamic Symbol Table.
The values in the Symbol Version Table are specific to the object in which they
are located. These values are identifiers that are provided by the the
vna_other member of the
Elfxx_Vernaux structure or the
vd_ndx member of the
Elfxx_Verdef structure.
The values 0 and 1 are reserved.
| 0 | The symbol is local, not available outside the object. | |
| 1 | The symbol is defined in this object and is globally available. |
All other values are used to identify version strings located in one of the other Symbol Version sections. The value itself is not the version associated with the symbol. The string identified by the value defines the version of the symbol.
The special section .gnu.version_d which has a section type of SHT_GNU_verdef shall contain symbol version definitions. The number of entries in this section shall be contained in the DT_VERDEFNUM entry of the Dynamic Section .dynamic. The sh_link member of the section header (see figure 4-8 in the System V ABI) shall point to the section that contains the strings referenced by this section.
The section shall contain an array of Elfxx_Verdef structures, as described in Figure 11-1, optionally followed by an array of Elfxx_Verdaux structures, as defined in Figure 11-2.
typedef struct {
Elfxx_Half vd_version;
Elfxx_Half vd_flags;
Elfxx_Half vd_ndx;
Elfxx_Half vd_cnt;
Elfxx_Word vd_hash;
Elfxx_Word vd_aux;
Elfxx_Word vd_next;
} Elfxx_Verdef; |
Figure 11-1. Version Definition Entries
vd_version | Version revision. This field shall be set to 1. | |
vd_flags | Version information flag bitmask. | |
vd_ndx | Version index numeric value referencing the SHT_GNU_versym section. | |
vd_cnt | Number of associated verdaux array entries. | |
vd_hash | Version name hash value (ELF hash function). | |
vd_aux | Offset in bytes to a corresponding entry in an array of Elfxx_Verdaux structures as defined in Figure 11-2 | |
vd_next | Offset to the next verdef entry, in bytes. |
typedef struct {
Elfxx_Word vda_name;
Elfxx_Word vda_next;
} Elfxx_Verdaux; |
Figure 11-2. Version Definition Auxiliary Entries
vda_name | Offset to the version or dependency name string in the section header, in bytes. | |
vda_next | Offset to the next verdaux entry, in bytes. |
The special section .gnu.version_r which has a section type of
SHT_GNU_verneed
shall contain required symbol version definitions. The number of entries in
this section shall be contained in the DT_VERNEEDNUM entry of the Dynamic
Section .dynamic.
The sh_link member of the section header (see figure 4-8 in
System V ABI)
shall point to the section that contains the strings referenced by this section.
The section shall contain an array of Elfxx_Verneed structures, as described in Figure 11-3, optionally followed by an array of Elfxx_Vernaux structures, as defined in Figure 11-4.
typedef struct {
Elfxx_Half vn_version;
Elfxx_Half vn_cnt;
Elfxx_Word vn_file;
Elfxx_Word vn_aux;
Elfxx_Word vn_next;
} Elfxx_Verneed; |
Figure 11-3. Version Needed Entries
typedef struct {
Elfxx_Word vna_hash;
Elfxx_Half vna_flags;
Elfxx_Half vna_other;
Elfxx_Word vna_name;
Elfxx_Word vna_next;
} Elfxx_Vernaux; |
Figure 11-4. Version Needed Auxiliary Entries
When loading a sharable object the system shall analyze version definition data from the loaded object to assure that it meets the version requirements of the calling object. This step is referred to as definition testing. The dynamic loader shall retrieve the entries in the caller's Elfxx_Verneed array and attempt to find matching definition information in the loaded Elfxx_Verdef table.
Each object and dependency shall be tested in turn. If a symbol definition is missing and the vna_flags bit for VER_FLG_WEAK is not set, the loader shall return an error and exit. If the vna_flags bit for VER_FLG_WEAK is set in the Elfxx_Vernaux entry, and the loader shall issue a warning and continue operation.
When the versions referenced by undefined symbols in the loaded object are found, version availability is certified. The test completes without error and the object shall be made available.
When symbol versioning is used in an object, relocations extend definition testing beyond the simple match of symbol name strings: the version of the reference shall also equal the name of the definition.
The same index that is used in the symbol table can be referenced in the SHT_GNU_versym section, and the value of this index is then used to acquire name data. The corresponding requirement string is retrieved from the Elfxx_Verneed array, and likewise, the corresponding definition string from the Elfxx_Verdef table.
If the high order bit (bit number 15) of the version symbolis set, the object cannot be used and the static linker shall ignore the symbol's presence in the object.
When an object with a reference and an object with the definition are being linked, the following rules shall govern the result:
The object with the reference and the object with the definitions both use
versioning. All described matching is processed in this case. A fatal error
shall be triggered when no matching definition can be found in the object whose
name is the one referenced by the vn_name element in the
Elfxx_Verneed entry.
The object with the reference does not use versioning, while the object with the definitions does. In this instance, only the definitions with index numbers 1 and 2 will be used in the reference match, the same identified by the static linker as the base definition. In cases where the static linker was not used, such as in calls to dlopen(), a version that does not have the base definition index shall be acceptable if it is the only version for which the symbol is defined.
The object with the reference uses versioning, but the object with the definitions specifies none. A matching symbol shall be accepted in this case. A fatal error shall be triggered if a corruption in the required symbols list obscures an outdated object file and causes a match on the object filename in the Elfxx_Verneed entry.
Neither the object with the reference nor the object with the definitions use versioning. The behavior in this instance shall default to pre-existing symbol rules.
Every executable shall contain a section named .note.ABI-tag of type SHT_NOTE. This section is structured as a note section as documented in the ELF spec. The section shall contain at least the following entry. The name field (namesz/name) contains the string "GNU". The type field shall be 1. The descsz field shall be at least 16, and the first 16 bytes of the desc field shall be as follows.
The first 32-bit word of the desc field shall be 0 (this signifies a Linux executable). The second, third, and fourth 32-bit words of the desc field contain the earliest compatible kernel version.