Engine support is deprecated in OpenSSL 3.0 and for OpenSSL 3.0 the default
is to disable engine support as engine support is deprecated. For ath79 architecture
build with autodetection engine support fails, so explicitly set off for now.
Signed-off-by: Ivan Pavlov <AuthorReflex@gmail.com>
With heavy system logging which goes by default into `/var/log/messages`
log file which is usually placed in tmpfs/RAM one can trigger OOM killer
fairly easily, thus killing random processes and in some cases making
system unusable.
This is likely happening due to the fact, that Linux by default uses 1/2
of available RAM for tmpfs, which might be for example an issue on low
RAM devices with ath10k wireless.
So let's fix it by adding logrotate functionality which should limit the
size of `/var/log/messages` log file to 1M by default, but could be
tweaked by config knob if needed be.
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Maintainer: Tom Stöveken <tom@naaa.de>
Compile tested: SDK for OpenWrt 21.02.3
Run tested: xRX200 rev 1.2, AVM FRITZ!Boxwq 7360 V2, OpenWrt 21.02.3
Description:
Updated to version 0.14.0
Signed-off-by: Tom Stöveken <tom@naaa.de>
Update the package to a commit that fixes an issue with removing PCP
mappings from nftables.
This also allows us to fix the nftables miniupnpd implementation on
openwrt.
In this new implementation, a table is created at the start of miniupnpd
and it is dedicated to miniupnpd with a priority above the firewall4
table. This allows miniupnpd to go ahead of the drop rules of firewall4
and forward traffic as needed. There was the possibility of adding a
chain inside the firewall4 table, but this would raise an issue where
if firewall4 was reloaded the port forwardings would be lost and
miniupnpd could be out of sync. When miniupnpd is stopped the table is
deleted, taking the port forwardings with it.
Some of this commit is based of msylgj's work, mainly the logic of the
init/hotplug scripts and the makefile build parameters.
Signed-off-by: ZiMing Mo <msylgj@immortalwrt.org>
Signed-off-by: Tiago Gaspar <tiagogaspar8@gmail.com>
Always use pthread_mutexattr_settype() the
pthread_mutexattr_setkind_np() function is not available in the
glibc and musl version used by OpenWrt.
This fixes the following compile error:
arc-openwrt-linux-gnu/bin/ld: gnunet_fuse-mutex.o: in function `GNUNET_mutex_create':
mutex.c:(.text+0x14): undefined reference to `pthread_mutexattr_setkind_np'
arc-openwrt-linux-gnu/bin/ld: mutex.c:(.text+0x14): undefined reference to `pthread_mutexattr_setkind_np'
arc-openwrt-linux-gnu/bin/ld: mutex.c:(.text+0x76): undefined reference to `pthread_mutexattr_setkind_np'
arc-openwrt-linux-gnu/bin/ld: mutex.c:(.text+0x76): undefined reference to `pthread_mutexattr_setkind_np'
collect2: error: ld returned 1 exit status
This patch was taken from:
https://aur.archlinux.org/cgit/aur.git/plain/pthread_mutexattr_settype.patch?h=gnunet-fuse
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Currently, lua-eco will add dependencies to all SSL libraries that are
selected, even though it will only use one of them. That means that the
package downloaded from the regular repository will install OpenSSL,
wolfSSL and mbedTLS, even though it will only use OpenSSL.
Fix that by adding a built option so that the default can be changed at
build-time. To maintain the author's intention, a default symbol is
computed based on what libraries are being built into the image, or just
selected as a module. Originally, the order or preference was OpenSSL,
wolfSSL, then mbedTLS.
One change was made to the original order: if OpenSSL and wolfSSL are
both selected as module, and mbedTLS is not built into the image,
wolfSSL will be preferred over OpenSSL. This is being done to keep the
package consistent with OpenWRT's selection of wolfSSL as the default
SSL library. If they are both included in the image, then OpenSSL will
be preferred.
The order of preference is:
1. If at least one library is included in the image, use the first of
OpenSSL, wolfSSL, and mbedTLS that is included in the image.
2. If at least one library is selected, but none included in the image,
prefer wolfSSL, then OpenSSL, then mbedTLS.
Signed-off-by: Eneas U de Queiroz <cotequeiroz@gmail.com>
This adds conflicts between the variants,
because they provide the same files, and it should not be
possible to install them side by side. Otherwise, it might happen that
half files would be from one variant and the other half from the
other.
Also, adds provides as if you request to install ``vim`` and
``vim-full``, then the request could be satisfied even they collide,
because ``vim-full`` provides ``vim`` package.
Signed-off-by: Karel Kočí <cynerd@email.cz>
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
[add commit message]
The full variant should conflict with the default variant. This prevents that
libgd and libgd-full could be installed side by side, and also, the full
variant should provide the libgd. Otherwise, if you install libgd-full,
you can not install vnstat.
Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
This commit restores the patch for libboost_context for mips64 given
that the upstream update added in 1.80.0 does not fully fixes the issue.
Signed-off-by: Carlos Miguel Ferreira <carlosmf.pt@gmail.com>
Fixes: https://github.com/openwrt/packages/issues/19210
If libjpeg isn't selected by another package, all is fine.
But if it is selected, the stress-ng build will see it and try to build the
jpeg stressor. This would usually fail sometime and link-time.
In any case, it's better to just pick-up libjpeg as a dependency of
stress-ng. If people want to stress their system with this tool, they can
probably expect libjpeg as well.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
