Move apache to PCRE2 now that PCRE is flagged EOL and won't receive any
security update.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
(cherry picked from commit d14fe0c51c)
Bump apache to release 2.4.57 and refresh patch automatically.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
(cherry picked from commit 86f9af41c1)
Pcre (1) is unmaintained and reached its end of life in 2021.
The base system provides pcre2 exclusively since May.
Signed-off-by: Leon M. Busch-George <leon@georgemail.eu>
(cherry picked from commit 379946951c)
Replace my own patch with the upstream solution, which they issued
in response to my bug report.
(Two patches as they overlooked something on the first try.
Reference to https://savannah.gnu.org/bugs/index.php?63431 )
The nettle lib evaluation is now conditional to not having "--disable-ntlm".
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
(cherry picked from commit fd7da3333e)
Update wget to 1.21.3
* Remove patch 100-fix-hsts-time.patch as upstream has issued
its own version on the fixes
* Add a hack (and fixup autoreconf) to fix an upstream bug that
forces the nettle library into nossl even if NTLM is disabled.
Upstream bug filed: https://savannah.gnu.org/bugs/?63431
* Remove old maintainer who has not been active
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
(cherry picked from commit a694130993)
`time_t` on musl 1.2 is 64bit, while `long` is 32 bit. we will always get zero time with the original source on mips big endian.
Signed-off-by: Huangbin Zhan <zhanhb88@gmail.com>
(cherry picked from commit 0907651268)
It seems like the libpcre dependency was added by mistake.
While checking in the source code of czmq (Makefile.am, CMakeLists.txt),
I see there are several dependencies, but there isn't PCRE.
Fixes: 936a48a ("czmq: add new package")
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
(cherry picked from commit e3ab95185c)
include/libdrm/drm.h and others headers are needed to build libva.
Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
(cherry picked from commit cb8c0ba6c0)
libreswan makefile detects macos (darwin) and changes build logic
but OpenWrt is always Linux so it is required to specify linux as
target platfrom
This patch specifies Linux as a target platfrom
Signed-off-by: Sergey V. Lobanov <sergey@lobanov.in>
(cherry picked from commit 270e8eabc3)
zerotier as default has executable stack.
[ 11.343143] process '/usr/bin/zerotier-one' started with executable stack
executable stacks are not recommend, possibly provide a threat and there
seems to be no advantage of executable stack with zerotier-one - so let's
build it without instead.
Stack is executable on x86_64, but not on all archs, such as ramips.
Signed-off-by: Oskari Rauta <oskari.rauta@gmail.com>
(cherry picked from commit 56f30520f2)
configure script detects Darwin and uses flags incompatible with
Linux target build.
This patch uses fakeuname tool if host OS is MacOS to avoid Darwin
detection on target build.
Signed-off-by: Sergey V. Lobanov <sergey@lobanov.in>
(cherry picked from commit 6e9c701050)
This patch adds KERNEL=Linux to MAKE_FLAGS to avod Darwin detection.
If Makefile detects Darwin, it removes -lbsd from build flags, but
this flag is required due to target is always Linux, not bsd-like.
Signed-off-by: Sergey V. Lobanov <sergey@lobanov.in>
(cherry picked from commit 53a51fb3d1)
configure script detects Darwin and uses flags incompatible with
Linux target build.
This patch uses fakeuname tool if host OS is MacOS to avoid Darwin
detection on target build.
Signed-off-by: Sergey V. Lobanov <sergey@lobanov.in>
(cherry picked from commit 0a9a52ee77)
This package provides fakeuname host tool to avoid build platform
detection for invalid configure/build scripts that are not
friendly to cross-platform build.
This fake uname tool returns Linux as OS name and target Linux
version as OS version.
Signed-off-by: Sergey V. Lobanov <sergey@lobanov.in>
(cherry picked from commit 4b6fc857be)
Fixes CVEs:
- CVE-2023-2828: The overmem cleaning process has been improved, to
prevent the cache from significantly exceeding the configured
max-cache-size limit.
- CVE-2023-2911: A query that prioritizes stale data over lookup
triggers a fetch to refresh the stale data in cache. If the fetch is
aborted for exceeding the recursion quota, it was possible for named
to enter an infinite callback loop and crash due to stack overflow.
The complete list of changes is available in the upstream release
notes at
https://ftp.isc.org/isc/bind9/cur/9.18/doc/arm/html/notes.html#notes-for-bind-9-18-16
Signed-off-by: Noah Meyerhans <frodo@morgul.net>
(cherry picked from commit 9ac79ad469)
Imported patches included in debian and other package.
* 200-Fix-NULL-pointer-crashes-from-175.patch
CVE-2021-3502
A flaw was found in avahi 0.8-5. A reachable assertion is present in avahi_s_host_name_resolver_start function allowing a local attacker to crash the avahi service by requesting hostname resolutions through the avahi socket or dbus methods for invalid hostnames. The highest threat from this vulnerability is to the service availability.
* 201-Avoid-infinite-loop-in-avahi-daemon-by-handling-HUP-event.patch
CVE-2021-3468
A flaw was found in avahi in versions 0.6 up to 0.8. The event used to signal the termination of the client connection on the avahi Unix socket is not correctly handled in the client_work function, allowing a local attacker to trigger an infinite loop. The highest threat from this vulnerability is to the availability of the avahi service, which becomes unresponsive after this flaw is triggered.
* 202-avahi_dns_packet_consume_uint32-fix-potential-undefined-b.patch
avahi_dns_packet_consume_uint32 left shifts uint8_t values by 8, 16 and 24 bits to combine them into a 32-bit value. This produces an undefined behavior warning with gcc -fsanitize when fed input values of 128 or 255 however in testing no actual unexpected behavior occurs in practice and the 32-bit uint32_t is always correctly produced as the final value is immediately stored into a uint32_t and the compiler appears to handle this "correctly".
Cast the intermediate values to uint32_t to prevent this warning and ensure the intended result is explicit.
* 203-Do-not-disable-timeout-cleanup-on-watch-cleanup.patch
This was causing timeouts to never be removed from the linked list that tracks them, resulting in both memory and CPU usage to grow larger over time.
* 204-Emit-error-if-requested-service-is-not-found.patch
It currently just crashes instead of replying with error. Check return
value and emit error instead of passing NULL pointer to reply.
* 205-conf-file-line-lengths.patch
Allow avahi-daemon.conf file to have lines longer than 256 characters (new limit 1024).
Signed-off-by: Hirokazu MORIKAWA <morikw2@gmail.com>
(cherry picked from commit 779af4d40c)
Function start_service() is called whenever service may need reloading.
If SMB server is not running it could be simply because it has been
stopped. Reloading service in such case is not an error so:
1. Don't log error as it isn't one
2. Don't exit with error code as it was confusing procd
This change fixes scenario like:
/etc/init.d/ksmbd stop
/etc/init.d/wsdd2 reload
(previously above wasn't stopping wsdd2)
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
(cherry picked from commit 6020ca52bf)
The default runtime directory used by LXC is /run which doesn't exist
in OpenWrt. It causes errors like:
Failed to create lock for foo
lxc-create: foo: tools/lxc_create.c: main: 260 Failed to create lxc container
There has been workaround for that in the lxc-auto.init but it requires
installing "lxc-auto" package. Replacing that "ln -s" workaround with
Makefile specifying --with-runtime-path allows using pure "lxc" in
OpenWrt (without the "lxc-auto").
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
(cherry picked from commit 90fef036fe)
