| Patch | Description | Author | Forwarded | Bugs | Origin | Last update |
|---|---|---|---|---|---|---|
| use-pkgconfig-for-libre2.patch | Use pkg-config to detect libre2 | Andrej Shadura <andrew.shadura@collabora.co.uk> | no | 2021-12-03 | ||
| 01_Makefile_am.patch | Change the locations built by install-data-hook to match debian install. =================================================================== |
Debian QA Group <packages@qa.debian.org> | no | 2021-12-03 | ||
| configure.diff | configure =================================================================== |
Debian QA Group <packages@qa.debian.org> | no | 2021-12-03 | ||
| 02_sphinx.conf.patch | Changed default locations to match debian locations. =================================================================== |
Debian QA Group <packages@qa.debian.org> | no | 2021-12-03 | ||
| add-multiarch-libdir.patch | add-multiarch-libdir | Debian QA Group <packages@qa.debian.org> | no | 2021-12-03 | ||
| fix-2540.patch | [PATCH] fixed #2504 daemon crash on snippet generation for empty documents with html_strip enabled | Stanislav Klinov <tomat@sphinxsearch.com> | no | 2016-09-06 | ||
| 06-CVE-2020-29050.patch | [PATCH] Fix random file reading by scattered snippets Issue was that load_files_scattered option just concatenated prefix with given name and then opened the file. So, if you provide something like '/etc/password' as file, or '../../../etc/passwd' even with non-zero prefix, it will unfortunately work. After this commit such behavior possible ONLY if user explicitly set `snippets_file_prefix` to empty string or '/'; another cases will not cause uncontroled reading. That fixes #866. [basilgello: Back-port to 2.2.11: - Refactor sphNormalizePath to use only available classes. - Do not use 'g_sSnippetsFilePrefix' in src/sphinxexcerpt.h, use 'tOptions.m_sFilePrefix' instead, as for 2.2.11, all two instances are assigned to 'g_sSnippetsFilePrefix'. - Adjust context. ] The following test passes, adapted from commit 66b5761ad258c60b1866a8e1333f86e74f48035 at https://github.com/manticoresoftware/manticoresearch : ==== /* cd test && g++ -o test test.cpp -I../src -I../config -I/usr/include/postgresql \ -I/usr/include/mariadb -DHAVE_CONFIG_H -L../src -lsphinx -pthread \ -lm -ldl -lstemmer -lmariadb -lpq -lexpat -lz */ void ASSERT_STREQ(const char *src, const char *dst) { auto normalized = sphNormalizePath(src); if (!strcmp(dst, normalized.scstr())) { auto _src = src; auto _dst = dst; if (src == nullptr) _src = "nullptr"; if (dst == nullptr) _dst = "nullptr"; std::cout << "OK : '" << _src << "' = '" << _dst << "'" << std::endl; } else { auto _src = src; auto _dst = dst; if (src == nullptr) _src = "nullptr"; if (dst == nullptr) _dst = "nullptr"; std::cout << "FAIL: '" << _src << "' != '" << _dst << "' ( = '" << normalized.scstr() << "')" << std::endl; } } int main() { ASSERT_STREQ ( "/", "/" ); ASSERT_STREQ ( "/..//bbb", "/bbb" ); ASSERT_STREQ ( "/quite/long/path/../../../etc/passwd", "/etc/passwd" ); ASSERT_STREQ ( "/aaa/bbb/ccc/ddd/../../../../../../../", "/" ); ASSERT_STREQ ( "", "" ); ASSERT_STREQ ( nullptr, "" ); ASSERT_STREQ ( "aaa/", "aaa" ); ASSERT_STREQ ( "aaa/.", "aaa" ); ASSERT_STREQ ( "aaa/././././////././", "aaa" ); ASSERT_STREQ ( "aaa/////", "aaa" ); ASSERT_STREQ ( "aaa/bbb/ccc", "aaa/bbb/ccc" ); ASSERT_STREQ ( "aaa/bbb/ccc/ddd/..", "aaa/bbb/ccc" ); ASSERT_STREQ ( "aaa/bbb/ccc/ddd/../../..", "aaa" ); ASSERT_STREQ ( "aaa/bbb/ccc/ddd/../../../xxx", "aaa/xxx" ); ASSERT_STREQ ( "aaa/bbb/ccc/ddd/../../../..", "" ); ASSERT_STREQ ( "aaa/bbb/ccc/ddd/../../../../", "" ); ASSERT_STREQ ( "aaa/bbb/ccc/ddd/../../../../../../../", "../../.." ); ASSERT_STREQ ( "..//bbb", "../bbb" ); return 0; } ==== Fixes CVE-2020-29050. |
klirichek <alexey@manticoresearch.com> | no | 2019-07-02 | ||
| support-mysql8-clients.patch | Support MySQL 8 clients The handshsake implementation of version 2.2.11 does not support MySQL 8 clients, not allowing them to connect to searchd using the MySQL protocol. It was fixed in version 3.3.1 according to the release notes: http://sphinxsearch.com/docs/sphinx3.html#version-3.1.1-17-oct-2018 . This patch was cherry-picked from manticoresearch project which is a sphinxsearch fork with the same handshake implementation. |
"Aleksey N. Vinogradov" <a.n.vinogradov@gmail.com> | no | other, https://github.com/manticoresoftware/manticoresearch/commit/ed98fdd7230c1d7298e1000fd10812338c6ba698 | 2019-12-06 | |
| config-default-to-localhost.patch | Fixes #1011 default listens bind to localhost in sample configs CVE-2019-14511 |
Adrian Nuta <adriannuta@gmail.com> | no | debian | 2019-08-21 | |
| install-paths.patch | Update the sample/minimal conf files to match installation paths | Andrej Shadura <andrew.shadura@collabora.co.uk> | no | 2021-12-03 | ||
| re2-drop-set_utf8.patch | use RE2::Options.set_encoding() instead of set_utf8() | Vitaly Lipatov <lav@altlinux.ru> | no | 2021-04-24 | ||
| cross.patch | Use pkg-config instead of mysql_config sphinxsearch fails to cross build from source, because it fails detecting mysql using mysql_config. During cross compilation, mysql_config does not work at all. For that reason, it should not be used at all. The attached patch looks up mysql using pkg-config, which presently is the standard tools for discovering compiler and linker flags. It makes sphinxsearch cross buildable. Please be aware: * In order to use PKG_CHECK_MODULES, it must not occur in a shell if/else branch. Such branches must be expressed using AS_IF to make the use of AC_REQUIRE work. * mariadb presently lacks a dependency on libssl-dev, but it emits -lssl as linker flag, see #917135. * In order to use the patch, you need to add pkg-config to Build-Depends. I tried writing the patch in a way that is upstreamable as it is useful to other distributions (e.g. yocto). |
Helmut Grohne <helmut@subdivi.de> | no | debian | 2018-12-23 | |
| fix-pthread-hurd.patch | Don’t call pthread_attr_setstacksize if PTHREAD_STACK_MIN is not defined Fixed build on Hurd See https://www.gnu.org/software/hurd/open_issues/libgomp_pthread_attr_setstacksize_pthread_stack_min.html <azeem> neal: libgomp (GNU's implementation of OpenMP) uses PTHREAD_STACK_MIN, which we do not define apparently <neal> azeem: We have fixed sized stacks. <neal> so the pthread_attr_setstacksize will fail once you define PTHREAD_STACK_MIN) |
Andrej Shadura <andrew.shadura@collabora.co.uk> | no | 2021-12-03 |