According to many bugreports [0][1][2] the default ath10k-ct kernel
module is unusable on devices with just 64 MiB RAM or with 128 MiB and
dual ath10k cards. The target boards boot but eventually oom-killer
starts to interfere with normal operation, so the current state is
effectively broken.
Since the two patches in question have a performance impact (and
possibly some other unexpected side-effects) a dedicated build variant
is added so that users of the low RAM devices can still benefit from all
the ath10k-ct advantages.
According to testing [3] results, the issue can be experienced even with
"a 256MB device with three radios". Measured performance impact of
implementing small buffers was lowering "the maximum 5 GHz throughput on
an IPQ40xx device without RPS/XPS optimizations from 494/432 Mbit/s for
TCP transfers (download/upload) to 438/343 Mbit/s"
The patches were apparently inspired by QSDK tweaks used by ODMs for the
affected devices.
[0] http://lists.infradead.org/pipermail/openwrt-devel/2019-December/020573.html
[1] https://github.com/openwrt/openwrt/pull/1077
[2] https://bugs.openwrt.org/index.php?do=details&task_id=2664
[3] https://github.com/freifunk-gluon/gluon/pull/1440#issue-195607701
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
[Remove double CONFIG_ATH10K-CT_LEDS entry]
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
(cherry picked from commit 1ac627024d)
This patch resolves recursive dependency warning on a feed package:
$ make defconfig
Collecting package info: done
tmp/.config-package.in:104721:error: recursive dependency detected!
For a resolution refer to Documentation/kbuild/kconfig-language.txt
subsection "Kconfig recursive dependency limitations"
tmp/.config-package.in:104721:symbol PACKAGE_nfs-kernel-server depends on NFS_KERNEL_SERVER_V4
For a resolution refer to Documentation/kbuild/kconfig-language.txt
subsection "Kconfig recursive dependency limitations"
feeds/packages/net/nfs-kernel-server/Config.in:4:symbol NFS_KERNEL_SERVER_V4 depends on PACKAGE_nfs-kernel-server
#
# configuration written to .config
#
19.07 branch uses kernel 4.14 only, so CONFIG_LINUX_4_9 symbol is not
needed anyway.
Ref: https://github.com/openwrt/packages/issues/10490
Signed-off-by: Sungbo Eo <mans0n@gorani.run>
Unloading and reloading the modules fails, as platform_device_put() does not
release resources fully.
root@OpenWrt:/# insmod i2c-gpio-custom bus0=0,18,0,5
[ 196.860620] Custom GPIO-based I2C driver version 0.1.1
[ 196.871162] ------------[ cut here ]------------
[ 196.880517] WARNING: CPU: 0 PID: 1365 at fs/sysfs/dir.c:31 0x80112158
[ 196.893431] sysfs: cannot create duplicate filename '/devices/platform/i2c-gpio.0'
...
[ 197.513200] kobject_add_internal failed for i2c-gpio.0 with -EEXIST, don't try to register things with the same name in the same directory.
This patch fixes it by replacing platform_device_put() to
platform_device_unregister().
Fixes: da77408537 ("i2c-gpio-custom: minor bugfix")
Fixes: 3bc81edc70 ("package: fix w1-gpio-custom package (closes#6770)")
Signed-off-by: Sungbo Eo <mans0n@gorani.run>
(cherry picked from commit a22b7a60d9)
This update doesn't include:
3b1e0a7bdfee brcmfmac: add support for SAE authentication offload
be898fed355e brcmfmac: send port authorized event for FT-802.1X
due to nl80211 dependencies.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
(cherry picked from commit c3aa33bf70)
This also fixes mac80211_prepare_vif iw set channel in monitor or
mesh mode.
Signed-off-by: Santiago Piccinini <spiccinini@altermundi.net>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
[daniel@makrotopia.org: fixed commit message]
(cherry picked from commit c7fb12beb1)
Commit b3d8b3a introduced a new test:
[ -n "$noscan" -a "$noscan" -gt 0 ] && hostapd_noscan=1
But if length of "$noscan" is zero (noscan is not set) this doesn't stop
the shell to evaluate the rest of the test.
root@hank2:~# [ -n "$noscan" -a "$noscan" -gt 0 ]
ash: out of range
root@hank2:~#
So when radios are brought up this shows in the log:
Sat Nov 23 10:51:38 2019 daemon.info procd: - init complete -
Sat Nov 23 10:52:24 2019 daemon.notice netifd: radio1 (1243): sh: out of range
Sat Nov 23 10:52:25 2019 user.notice firewall: Reloading firewall due to ifup of wan (eth0.2)
Sat Nov 23 10:52:25 2019 daemon.notice netifd: radio0 (1242): sh: out of range
Sat Nov 23 10:52:26 2019 authpriv.info dropbear[1536]: Not backgrounding
This commit sets noscan to 0 if unset and removes the gratuitous length
check, preventing the warning.
Signed-off-by: Sebastian Kemper <sebastian_ml@gmx.net>
(cherry picked from commit 28d84331f4)
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
This potentially fixes some issues seen on IBSS
when interfaces go out of range and then re-appear.
Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
The skb_get_hash_perturb() function now takes a siphash_key_t instead of
an u32. This was changed in commit 55667441c84f ("net/flow_dissector:
switch to siphash"). Use the correct type in the fq header file
depending on the kernel version.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Stefan Lippers-Hollmann <s.l-h@gmx.de>
(cherry picked from commit eaa047179a)
AutoLoad parameter must match the exact kernel module name. Fix it.
Fixes: 125f1ce9ad ("kernel: video: add DRM core and IMX DRM support for HDMI/LVDS")
Signed-off-by: Sungbo Eo <mans0n@gorani.run>
(cherry picked from commit 6990510aca)
Instead of depending on kmod-usb2 make it depend on the normal USB
dependencies. This should hopefully fix some problems seen in the build
bot builds for powerpc_8540.
In addition also activate DRIVER_11N_SUPPORT support.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
(cherry picked from commit 3ff3b044c0)
The QCA953x only supports 25 MHz refclk, however some OEMs set an
invalid bootstrap value for the REF_CLK option, which would break the
clock detection in ath9k.
Force the QCA953x refclk to 25MHz in ath9k, as this is (according to the
datasheet) the only valid frequency.
Signed-off-by: David Bauer <mail@david-bauer.net>
(cherry picked from commit 4c6fe32468)
Import patches from upstream to sync 19.07 with master:
9f3e3323e996 rt2x00: allow to specify watchdog interval
2034afe4db4a rt2800: add helpers for reading dma done index
759c5b599cf4 rt2800: initial watchdog implementation
09db3b000619 rt2800: add pre_reset_hw callback
710e6cc1595e rt2800: do not nullify initialization vector data
e403fa31ed71 rt2x00: add restart hw
0f47aeeada2a rt2800: do not enable watchdog by default
41a531ffa4c5 rt2x00usb: fix rx queue hang
3b902fa811cf rt2x00usb: remove unnecessary rx flag checks
1dc244064c47 rt2x00: no need to check return value of debugfs_create functions
706f0182b1ad rt2800usb: Add new rt2800usb device PLANEX GW-USMicroN
95844124385e rt2x00: clear IV's on start to fix AP mode regression
567a9b766b47 rt2x00: do not set IEEE80211_TX_STAT_AMPDU_NO_BACK on tx status
14d5e14c8a6c rt2x00: clear up IV's on key removal
13fa451568ab Revert "rt2800: enable TX_PIN_CFG_LNA_PE_ bits per band"
--pending-- rt2800: remove errornous duplicate condition
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
The kconfig symbol is an invisible one since its introduction. It is
not supposed to be enabled on its own.
Resolves FS#1821
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
(cherry picked from commit 4bf9bec361)
Improves rate control responsiveness and performance
Signed-off-by: Felix Fietkau <nbd@nbd.name>
[reworked to apply on 4.19.79 mac80211 + renumbered + refreshed]
Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
Upon writing to "remove" file, debugfs_remove_recursive() blocks while
holding rtnl_lock. This is because debugfs' file_ops callbacks are
executed in debugfs_use_file_*() context which prevents file removal.
Fix this by only flagging the device for removal and then do the cleanup
in file_ops.release callback which is executed out of that context.
Signed-off-by: Ali MJ Al-Nasrawy <alimjalnasrawy@gmail.com>
(cherry picked from commit c2635b871d)
Since v3.11, netdevice notification data are of type
"struct netdev_notifier_info". Handle it as such!
This should fix a critical bug in which devices are unable get released
because trelay does not release resources in response to UNREGISTER
event spamming the log with something like:
unregister_netdevice: waiting for eth0.1 to become free. Usage count = 1
Signed-off-by: Ali MJ Al-Nasrawy <alimjalnasrawy@gmail.com>
(cherry picked from commit 77cfc0739d)
The destination buffer size `d_len` is passed to `lzma_inflate` as a
pointer. Therefore, it needs to be dereferenced to compare its content.
Signed-off-by: Christian Franke <nobody@nowhere.ws>
(cherry picked from commit d544bc84a0)
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
5e8cd86f90da ath10k-ct: Backport ap-vlan code from 5.2 to 4.20 and 4.19 drivers.
0c518586bd7f ath10k-ct: Fix a few warning splats.
Adds AP VLAN.
Refreshed all patches.
Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
add module to support Emulex OneConnect
common in 10Gbit SFP+ cards by Dell/HP/IBM
supports OneConnect OCe10xxx OCe11xxx OCe14xxx,
LightPulse LPe12xxx
Signed-off-by: Alberto Bursi <alberto.bursi@outlook.it>
(cherry picked from commit 827f47749b)
This makes brcmfmac use the same wiphy after PCIe reset to help user
space handle corner cases (e.g. firmware crash).
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
(cherry picked from commit f39f4b2f6d)
Patch getting RAM info got upstreamed. A debugging fs entry for testing
reset feature was added.
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
(cherry picked from commit 681acdcc54)
This changes the default PKG_BUILD_DIR to take BUILD_VARIANT into
account (if set), so that packages do not need to manually override
PKG_BUILD_DIR just to handle variants.
This also updates most base packages with variants to use the updated
default PKG_BUILD_DIR.
Signed-off-by: Jeffery To <jeffery.to@gmail.com>
(cherry picked from commit e545fac8d9)
After commit e82a4d9cfb ("config: regenerate *_shipped sources") the mconf
parser became more strict as a side effect and started to spew a series of
warnings when evaluating our generated kconfig sources:
tmp/.config-package.in:705:warning: ignoring unsupported character '@'
The root cause of these warnings is a wrong use of the @SYMBOL dependency
syntax in various Makefile. Fix the corresponding Makefiles by turning
`@SYM||@SYM2` expressions into the proper `@(SYM||SYM2)` form.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
(cherry picked from commit 74739c4228)
By default, set BE tx queue TXOP limit to 2.0 in the hostapd config
Many vendor drivers are doing similar things to boost throughput.
On MT7612 under ideal conditions, it improves tx throughput from 470 Mbit/s
to about 570 Mbit/s.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
(cherry-picked from commit 8650201f10)
Close cooperation with Lorenzo Bianconi resulted
in these patches which fix all remaining seen issues
when using dynack.
Fix link losses when:
- Late Ack's are not seen or not present
- switching from too low static coverage class to dynack on a live link
These are fixed by setting the Ack Timeout/Slottime to
the max possible value for the currently used channel width when
a new station has been discovered.
When traffic flows, dynack is able to adjust to optimal values
within a few packets received (typically < 1 second)
These changes have been thoroughly tested on ~60 offshore devices
all interconnected using mesh over IBSS and dynack enabled on all.
Distances between devices varied from <100m up to ~35km
Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
(cherry picked from commit f6e8ba0238fe349b7529357793e2fb18635819ed)
the ath10k-ct package ships multiple versions of the ath10k-ct driver,
OpenWrt currently only uses the version 4.19, but we still ship some
patches for older versions. Remove all patches only touching older
versions and also remove the patch for older versions from patches which
do the same changes to multiple versions of ath10k-ct.
This removes some unneeded patches, the end binary should stay the same.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
(cherry picked from commit 10fe5ca362)
Changes:
ath10k: Improve PMF/MPF mgt frame check
And add a driver for 5.2 (beta, not even tested yet) kernel.
Refresh patches.
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
(cherry picked from commit 49b3dcb2ab)
This updates to backports-4.19.57-1 which contains the wireless
subsystem and driver from kernel 4.19.57.
The removed patches are applied upstream.
Tested-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
This doesn't include 9ff8614a3dbe ("brcmfmac: use separate Kconfig file
for brcmfmac") due to a few conflicts with backports changes.
An important change is:
[PATCH 2/7] brcmfmac: change the order of things in brcmf_detach()
which fixes a rmmod crash in the brcmf_txfinalize().
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
(cherry picked from commit db8e08a5a4)
This patch unifies the polled and interrupt-driven gpio_keys code
paths as well implements consistent handling of the debounce
interval set for the GPIO buttons and switches.
Hotplug events will only be fired if
1. The input changes its state and remains stable for the duration
of the debounce interval (default is 5 ms).
2. In the initial stable (no state-change for duration of the
debounce interval) state once the driver module gets loaded.
Switch type inputs will always report their stable state.
Unpressed buttons will not trigger an event for the initial
stable state. Whereas pressed buttons will trigger an event.
This is consistent with upstream's gpio-key driver that uses
the input subsystem (and dont use autorepeat).
Prior to this patch, this was handled inconsistently for interrupt-based
an polled gpio-keys. Hence this patch unifies the shared logic into the
gpio_keys_handle_button() function and modify both implementations to
handle the initial state properly.
The changes described in 2. ) . can have an impact on the
failsafe trigger. Up until now, the script checked for button
state changes. On the down side, this allowed to trigger the
failsafe by releasing a held button at the right time. On the
plus side, the button's polarity setting didn't matter.
Now, the failsafe will only engage when a button was pressed
at the right moment (same as before), but now it can
theoretically also trigger when the button was pressed the
whole time the kernel booted and well into the fast-blinking
preinit phase. However, the chances that this can happen are
really small. This is because the gpio-button module is usually
up and ready even before the preinit state is entered. So, the
initial pressed button event gets lost and most devices behave
as before.
Bisectors: If this patch causes a device to permanently go into
failsafe or experience weird behavior due to inputs, please
check the following:
- the GPIO polarity setting for the button
- the software-debounce value
Run-tested for 'gpio-keys' and 'gpio-keys-polled' on
- devolo WiFi pro 1200e
- devolo WiFi pro 1750c
- devolo WiFi pro 1750x
- Netgear WNDR4700
- Meraki MR24
- RT-AC58U
Signed-off-by: David Bauer <mail@david-bauer.net>
Signed-off-by: Christian Lamparter <chunkeey@gmail.com> [further
cleanups, simplification and unification]
(cherry picked from commit 27f3f493de)
While testing 4.19 build on malta/be64, I've encountered following
error:
gpio-button-hotplug/gpio-button-hotplug.c:529:18: error: implicit
declaration of function 'gpio_to_desc'
which is caused by the missing include fixed by this patch.
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
(cherry picked from commit dd6d82112a)
b43legacy needs ssb support and we do not compile the mips74 subtarget
of the brcm47xx target with SSB support. This causes a build failure in
the mac80211 package and only some of the kernel modules are being
created.
I am not aware of any device with a BRCM47xx mips74 CPU which uses a
b43legacy compatible device.
Fixes: FS#2334
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
(cherry picked from commit e05310b9b8)
Felix Fietkau
Paul Fertser
Sungbo Eo
Rafał Miłecki
Santiago Piccinini
Sebastian Kemper
Hauke Mehrtens
Koen Vandeputte
David Bauer
Daniel Golle
Yousong Zhou
Ali MJ Al-Nasrawy
Christian Franke
Alberto Bursi
Chuanhong Guo
Jeffery To
Jo-Philipp Wich
Kevin Darbyshire-Bryant
Petr Štetiar