Libxml2 2.11.0 released

Platform
,
1

Download

https://download.gnome.org/sources/libxml2/2.11/libxml2-2.11.0.tar.xz (2.50M)
sha256sum: b1a2fca3f7722d6c35a4f24974a1b5800e4d84ec5cb756bc61ef2b8c75a19e36

Major changes

Protection against entity expansion attacks, also known as “billion laughs”
has been greatly improved. Malicious files should be detected reliably now
and false positives should be reduced. It is possible though that large
documents which make heavy use of entities are rejected now.

This release finally fixes symbol visibility on UNIX systems. Internal
symbols will now be hidden. While these symbols were never declared in public
headers, it was still possible to declare them manually. Now this won’t work.

All symbol information has been removed from the ELF version script to fix
link errors with --no-undefined-version. The version nodes are kept so it
should still be possible to run binaries linked against older versions.

About 90 memory errors in code paths handling malloc failures have been fixed.
While these issues shouldn’t impact security, this improves robustness under
memory pressure.

The XInclude engine has been reworked to properly support nested includes.

Several cases of quadratic behavior in the XML push parser have been fixed.

Refactoring has begun on some buffering and encoding code with the goal of
simplifying this part of the code base and improving error reporting.

Other highlights:

  • Consolidated private header files.
  • Major rework of the autoconf build.
  • Deprecated several outdated and internal functions.

Special thanks to Google’s Open Source Security Subsidies program for
sponsoring much of the work on this release!

Ongoing work on libxml2 relies on funding. For a list of important open
issues see https://gitlab.gnome.org/GNOME/libxml2/-/issues/507

Security

  • Fix use-after-free in xmlParseContentInternal() (David Kilzer)
  • xmllint: Fix use-after-free with --maxmem
  • parser: Fix OOB read when formatting error message
  • entities: Rework entity amplification checks

Regressions

  • parser: Fix regression in xmlParserNodeInfo accounting

Bug fixes

  • Fix memory errors in code handling malloc failures
  • encoding: Fix error code in asciiToUTF8
  • xpath: number(‘-’) should return NaN
  • xmlParseStartTag2() contains typo when checking for default definitions for
    an attribute in a namespace (David Kilzer)
  • uri: Fix handling of port numbers
  • error: Make sure that error messages are valid UTF-8
  • xinclude: Fix nested includes

Improvements

  • xmllint: Validate --maxmem integer option
  • xmlValidatePopElement() can return invalid value (-1) (David Kilzer)
  • parser: Rework EBCDIC code page detection
  • parser: Limit name length in xmlParseEncName
  • parser: Rework shrinking of input buffers
  • html: Rely on CUR_CHAR to grow the input buffer
  • parser: Rely on CUR_CHAR/NEXT to grow the input buffer
  • valid: Make xmlValidateElement non-recursive
  • html: Fix quadratic behavior in htmlParseTryOrFinish
  • xmllint: Fix memory leak with --pattern --stream
  • parser: Stop calling xmlParserInputShrink
  • html: Impose some length limits
  • valid: Allow xmlFreeValidCtxt(NULL)
  • parser: Stop calling xmlParserInputGrow
  • xinclude: Fix quadratic behavior in xmlXIncludeLoadTxt
  • xinclude: Abort immediately if max depth was exceeded
  • xpath: Only report the first error
  • error: Don’t move past current position
  • error: Limit number of parser errors
  • parser: Lower entity nesting limit with XML_PARSE_HUGE
  • parser: Don’t increase depth twice when parsing internal entities
  • parser: Improve detection of entity loops
  • parser: Only report a single entity error
  • libxml.h: Remove dubious definition of LIBXML_STATIC
  • html: Improve parsing of nested lists
  • memory: Don’t use locks in xmlMemUsed
  • encoding: Remove unused variable xmlDefaultCharEncodingHandler
  • Rework initialization code
  • Add .editorconfig
  • parser: Merge misc, prolog and epilog cases in push parser
  • parser: Fix ‘consumed’ accounting when switching encodings
  • html: Fix check for end of comment in push parser
  • parser: Fix push parser with 1-3 byte initial chunk
  • parser: Rewrite push parser boundary checks
  • reader: Switch to xmlParserInputBufferCreateMem
  • html: Don’t escape ASCII chars in href attributes
  • io: Don’t shrink memory input buffers
  • parser: Don’t call xmlSHRINK from push parser
  • parser: Ignore cdata argument in xmlParseCharData
  • parser: Rework push parser parser progress checks
  • io: Fix a few integer overflows in I/O statistics
  • io: Rework xmlParserInputBufferGrow with encodings
  • io: Remove xmlInputReadCallbackNop
  • io: Check for memory buffer early in xmlParserInputGrow
  • parser: Fix error message in xmlParseCommentComplex
  • Bypass proxy in nanoHTTP for hosts in “no_proxy” (Markus Jörg)
  • schemas: Fix infinite loop in xmlSchemaCheckElemSubstGroup
  • threads: Remove check for pthread_equal
  • xinclude: Rework XInclude cache
  • xinclude: Remove inefficient refcounting scheme
  • xmllint: Improve handling of empty XPath node sets
  • parser: Fix potential memory leak in xmlParseAttValueInternal
  • error: Don’t use initGenericErrorDefaultFunc
  • xpath: Lower XPath recursion limit on Windows
  • Stop including sys/types.h
  • Don’t define WIN32 macro
  • Make xmlNewSAXParserCtx take a const sax handler
  • Consolidate private header files
  • Remove internal macros from parserInternals.h
  • Move some HTML functions to correct header file
  • xmllint: Stop calling xmlSAXDefaultVersion
  • Introduce xmlNewSAXParserCtxt and htmlNewSAXParserCtxt
  • Don’t mess with parser options in htmlParseDocument
  • Remove useless call to htmlDefaultSAXHandlerInit
  • Remove htmlDefaultSAXHandler from non-SAX1 build
  • Don’t initialize SAX handler in htmlReadMemory
  • Fix htmlReadMemory mixing up XML and HTML functions
  • Don’t use default SAX handler to report unrelated errors
  • Create stream with buffer in xmlNewStringInputStream
  • xmlcatalog: Fix memory leaks

Code quality

  • xzlib: Fix implicit sign change in xz_open
  • parser: Simplify calculation of available buffer space
  • parser: Use size_t when subtracting input buffer pointers
  • parser: Check for integer overflow when updating checkIndex
  • xpath: Fix harmless integer overflow in xmlXPathTranslateFunction
  • schematron: Use logical and
  • relaxng: Remove useless if statement
  • schemas: Remove useless if statement
  • pattern: Merge identical branches
  • regexp: Add sanity check in xmlRegCalloc2
  • regexp: Simplify xmlRegAtomPush
  • encoding: Cast toupper argument to unsigned char
  • uri: Add explicit cast in xmlSaveUri
  • buf: Fix return value of xmlBufGetInputBase
  • parser: Fix integer overflow of input ID
  • parser: Remove useless ent->etype test in xmlParseReference
  • parser: Remove useless ent->children tests in xmlParseReference
  • xmlmemory.c: Remove xmlMemContentShow
  • libxml.h: Add comments and indentation
  • libxml.h: Don’t include stdio.h
  • xmlexports.h: Disable docs for internal macro XMLPUBLIC
  • parser: Simplify xmlParseConditionalSections
  • io: Rearrange code in xmlSwitchInputEncodingInt
  • warnings: Fix -Wstrict-prototypes warning
  • warnings: Remove set-but-unused variables
  • Fix compiler warnings in SAX2.c
  • Fix unused variable warning in python/types.c
  • Fix compiler warning in examples
  • Fix compiler warnings in fuzzing code
  • Remove unused code in nanohttp.c
  • Remove or annotate char casts
  • Don’t use sizeof(xmlChar) or sizeof(char)
  • Remove explicit integer casts

Deprecations

  • parser: Deprecate more internal functions
  • parser: Deprecate some parser input functions
  • parser: Deprecate xmlString*DecodeEntities
  • threads: Deprecate some internal functions
  • buf: Deprecate static/immutable buffers
  • Deprecate internal parser functions
  • Deprecate old HTML SAX API
  • Generate deprecation warnings for old SAX API
  • Mark more functions setting globals as deprecated
  • Mark more parser functions as deprecated
  • Mark most SAX1 functions as deprecated
  • Deprecate some global variables

Portability

  • autoconf: Warn about outdated C compilers
  • win32: Remove broken libxml2.def.src
  • Remove symbols from version script
  • catalog.c: Silence a cast warning on VS 2022 (Lukáš Tyrychtr)
  • libxml.h: Remove ancient LynxOS setup
  • Use python3 not python (Ross Burton)
  • xstc/fixup-tests.py: port to Python 3 (Ross Burton)
  • xstc/fixup-tests.py: unify whitespace (Ross Burton)
  • Remove hacky heuristic from b2dc5675 (Alex Richardson)
  • Avoid creating an out-of-bounds pointer by rewriting a check
    (Alex Richardson)
  • Hide internal functions
  • Correctly relocate internal pointers after realloc() (Alex Richardson)
  • Visual Studio builds: Allow silencing deprecation warnings (Chun-wei Fan)
  • Visual Studio: Define XML_DEPRECATED (Chun-wei Fan)
  • xmllint: Include <io.h> on Windows
  • warnings: Work around MSVC bug
  • sources: Silence C4013 warnings on Visual Studio (Chun-wei Fan)
  • python/setup.py.in: Improve Windows import patching (Chun-wei Fan)
  • python: Create .pyd on Windows
  • Fix Python build on Windows
  • Fix Windows compiler warnings in python/types.c
  • Fix libxml_PyFileGet
  • Remove BeOS support
  • Fix libxml_PyFileGet with stdout on macOS
  • Migrate from PyEval_ to PyObject_
  • Port build_glob.py to Python 3
  • Port genChRanges.py to Python 3
  • xmlexports.h: Remove LIBXML_FASTCALL optimization
  • Remove XMLCALL and XMLCDECL macros from public headers
  • Remove XMLDECL macro from .c files

Build systems

  • cmake: Link against dl and dld only when LIBXML2_WITH_MODULES is
    enabled (Alexander Kutelev)
  • autotools: Fix make distcheck
  • Remove RPM build, Makefile.tests, README.tests
  • libxml.m4: deprecate AM_PATH_XML2, wrap PKG_CHECK_MODULES instead
    (Ross Burton)
  • libxml.m4: fix -Wstrict-prototypes (Sam James)
  • cmake: Build static library with -DLIBXML_STATIC
  • autotools: Don’t use version script on Windows
  • autotools: Fix winsock detection
  • autotools: Only add network libraries if HTTP/FTP enabled
  • autotools: Disable parallel Python build
  • python: Don’t output missing generators during build
  • build: Remove check for broken ss_family
  • http: Simplify IPv6 checks
  • autotools: Fix network checks on Windows
  • Fix detection of GNU libiconv
  • cmake: Fix Python installation
  • cmake: Don’t check for Python 2
  • configure.ac: Also check for MSYS host
  • Improve network library detection
  • Detect ws2_32 with AC_SEARCH_LIBS
  • Rework network configure checks
  • Remove arg cast configure checks
  • Fix dlopen check
  • Remove HAVE_WIN32_THREADS configuration flag
  • Rework dlopen and pthread detection
  • Fix test in configure.ac
  • cmake: Enable GCC compiler warnings
  • Always link with -no-undefined
  • Use AM_CFLAGS and AM_LDFLAGS consistently
  • Remove -Wredundant-decls
  • Call AC_CHECK_* with multiple arguments
  • configure.ac: Remove checks for unused programs
  • Rework library detection in configure.ac
  • Rearrange configure.ac
  • Consolidate zlib and lzma detection
  • Remove “runtime debugging”
  • Consolidate simple API modules in configure.ac
  • Fix dependency resolution in configure.ac
  • Fix --with-valid --without-regexps build
  • Fix --with-schemas --without-xpath build
  • Don’t build unneeded .c source files
  • Move xmlIsXHTML to tree.c
  • Cleanup distribution settings in Makefile.am
  • Also clean *.pyc files for Python 2
  • Don’t distribute libxml2.spec

Tests

  • testchar: Add test for memory pull parser with encoding
  • fuzz: Also test init function of URI fuzzer
  • fuzz: Separate fuzzer for DTD validation
  • gitlab-ci: Enable all “integer” sanitizers
  • fuzz: Inject random malloc failures
  • fuzz: Support variable integer sizes in fuzz data
  • fuzz: Fix duplicate detection in fuzzEntityRecorder
  • fuzz: Set filename in xmlFuzzEntityLoader
  • fuzz: Allow xmlFuzzReadString(NULL)
  • fuzz: Fix Makefile dependencies
  • fuzz: Add test/recurse to seed corpus
  • fuzz: Add separate XInclude fuzzer
  • runsuite: Some errors are expected
  • testrecurse: Test entity expansion stats
  • testapi.c: Initialize catalog early
  • gentest.py: Fix memory leak in API tests
  • tests: Enable “runsuite” test
  • python/tests/reader2: use absolute paths everywhere (Ross Burton)
  • python/tests/reader2: always exit(1) if a test fails (Ross Burton)
  • testModule: exit if the module can’t be opened (Ross Burton)
  • CI: disable modules in gcc:static build (Ross Burton)
  • CI: fix CI on MinGW builds (Ross Burton)
  • python: Fix memory leak checks
  • tests: Check that xmlInitParser doesn’t allocate memory
  • tests: Fix use-after-free in Python tests
  • tests: Remove unneeded #includes
  • gitlab-ci: Make Test-Msvc exit if ctest fails
  • gitlab-ci: Treat compiler warnings as errors on MSVC
  • test: Add test for push parser boundaries
  • gitlab-ci: Upgrade image to Ubuntu 22.10, reenable MSan
  • gitlab-ci: Reenable LeakSanitizer
  • gitlab-ci: Fix llvm-symbolizer
  • xinclude: Don’t create result doc for test with errors
  • xinclude: Also test error messages
  • gitlab-ci: Allow cast-align warnings from clang
  • gitlab-ci: Fix tar invocation
  • gitlab-ci: Move MSVC test to separate script
  • gitlab-ci: Fix SUFFIX, remove MINGW_PATH
  • gitlab-ci: Consolidate CMake test scripts
  • gitlab-ci: Only install MinGW autotools if needed
  • gitlab-ci: Only install cmake MinGW package if needed
  • gitlab-ci: Install 7-Zip using the .msi
  • Use $MSYSTEM and ‘bash -lc’ in MinGW CI
  • Add CI job for MinGW/Autotools
  • Consolidate CI scripts
  • Allow empty MINGW_PACKAGE_PREFIX
  • Move Dockerfile to .gitlab-ci directory
  • testapi: Disable on Windows for now
  • Disable fuzzer tests if glob.h wasn’t found
  • Move automata test to runtest.c
  • Fix testapi when building --without-sax1

Documentation

  • doc: Remove ancient files
  • Remove ancient TODOs
  • html: Fix htmlInitAutoClose documentation
  • doc: Mention new location of XML catalog as breaking change
  • doc: Mention potentially breaking changes in NEWS
  • doc: Remove xmlDllMain from documentation and version script
  • doc: Mention ${sysconfdir} in man pages
  • doc: Document xmlcatalog --convert
  • doc: Document xmllint --nodict and --pedantic
  • doc: Fix indentation in source XML files
  • xmllint: Document --quiet option
  • Improve cross-references in API docs
  • Improve documentation of globals
  • Fix documentation parser
  • Support comments for global variables in documentation
  • Fix update call in apibuild.py
  • Don’t index anything in DOC_DISABLE sections
  • Fix warnings from apibuild.py
  • Start with documentation for maintainers
1 Like
2

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.