5 years agokernel: Add support spi-nor, Eon EN25QH32
YuheiOKAWA [Wed, 16 May 2018 10:54:26 +0000 (19:54 +0900)]
kernel: Add support spi-nor, Eon EN25QH32

Support Add spi-nor chip EN25QH32. JEDEC is 1c7016.

Can't boot issue, new revision hardware is EN25QH32 flash chip.
example, MZK-DP150N (ramips).

Signed-off-by: YuheiOKAWA <>
5 years agokernel: Add configfs support for USB HID gadget
Harvey Phillips [Sat, 19 May 2018 19:34:05 +0000 (20:34 +0100)]
kernel: Add configfs support for USB HID gadget

Added a rule to to build usb_f_hid.ko for configfs support
Tested on a Raspberry Pi Zero W

Signed-off-by: Harvey Phillips <>
5 years agoath10k-firmware: Fix mirror hash sum
Hauke Mehrtens [Sat, 19 May 2018 13:20:46 +0000 (15:20 +0200)]
ath10k-firmware: Fix mirror hash sum

This now matches what was generated locally on my PC and the file on the
mirror server.

Fixes: 349fe46103359 ("ath10k-firmware: Update QCA988X firmware to the latest version")
Signed-off-by: Hauke Mehrtens <>
5 years agoopenwrt-keyring: bundle latest usign certificates
Jo-Philipp Wich [Sun, 20 May 2018 17:37:58 +0000 (19:37 +0200)]
openwrt-keyring: bundle latest usign certificates

Includes the public usign certificates used by the 18.06.* release builds.

Signed-off-by: Jo-Philipp Wich <>
5 years agobase-files: depend on openwrt-keyring
Jo-Philipp Wich [Sun, 20 May 2018 17:34:36 +0000 (19:34 +0200)]
base-files: depend on openwrt-keyring

Signed-off-by: Jo-Philipp Wich <>
5 years agoopenwrt-keyring: rename from lede-keyring
Jo-Philipp Wich [Sun, 20 May 2018 17:32:29 +0000 (19:32 +0200)]
openwrt-keyring: rename from lede-keyring

Also let the new openwrt-keyring package provide lede-keyring for backwards

Signed-off-by: Jo-Philipp Wich <>
5 years agowireguard: bump to 20180519
Jason A. Donenfeld [Sat, 19 May 2018 03:55:58 +0000 (05:55 +0200)]
wireguard: bump to 20180519

* chacha20poly1305: add mips32 implementation

"The OpenWRT Commit" - this significantly speeds up performance on cheap
plastic MIPS routers, and presumably the remaining MIPS32r2 super computers
out there.

* timers: reinitialize state on init
* timers: round up instead of down in slack_time
* timers: remove slack_time
* timers: clear send_keepalive timer on sending handshake response
* timers: no need to clear keepalive in persistent keepalive

Andrew He and I have helped simplify the timers and remove some old warts,
making the whole system a bit easier to analyze.

* tools: fix errno propagation and messages

Error messages are now more coherent.

* device: remove allowedips before individual peers

This avoids an O(n^2) traversal in favor of an O(n) one. Before systems with
many peers would grind when deleting the interface.

Signed-off-by: Jason A. Donenfeld <>
5 years agomt76: update to the latest version
Felix Fietkau [Fri, 18 May 2018 16:13:55 +0000 (18:13 +0200)]
mt76: update to the latest version

73edb22 mt76: discard early received packets if not running yet
0b8d1dd mt76: fix beacon timer drift

Signed-off-by: Felix Fietkau <>
5 years agokernel: whitespace fixes
John Crispin [Fri, 18 May 2018 09:09:06 +0000 (11:09 +0200)]
kernel: whitespace fixes

Signed-off-by: John Crispin <>
5 years agox86: disable CONFIG_EFI_VARS
Alif M. Ahmad [Tue, 27 Feb 2018 02:01:04 +0000 (02:01 +0000)]
x86: disable CONFIG_EFI_VARS

Disable CONFIG_EFI_VARS, since it suffers from sysfs limitation (no
support for variable longer than 1024 bytes).

kmod-fs-efivarfs is the replacement of this, which enables mounting
efivarfs file system and doesn't suffer from 1024 bytes limitation.

Signed-off-by: Alif M. Ahmad <>
5 years agohostapd: fix IEEE 802.11r (fast roaming) defaults
Gospod Nassa [Fri, 18 May 2018 08:39:33 +0000 (09:39 +0100)]
hostapd: fix IEEE 802.11r (fast roaming) defaults

Use ft_psk_generate_local=1 by default, as it makes everything else fairly
trivial. All of the r0kh/r1kh and key management stuff goes away and hostapd
fairly much does it all for us.

We do need to provide nas_identifier, which can be derived from the BSSID,
and we need to generate a mobility_domain, for which we default to the first
four chars of the md5sum of the SSID.

The complex manual setup should also still work, but the defaults also
now work easily out of the box. Verified by manually running hostapd
(with the autogenerated config) and watching the debug output:

wlan2: STA ac:37:43:a0:a6:ae WPA: FT authentication already completed - do not start 4-way handshake

 This was previous submitted to LEDE in

[dwmw2: Rewrote commit message]
Signed-off-by: Gospod Nassa <>
Signed-off-by: David Woodhouse <>
5 years agoiwinfo: bump to latest git HEAD
John Crispin [Fri, 18 May 2018 07:37:53 +0000 (09:37 +0200)]
iwinfo: bump to latest git HEAD

e59f925 hardware: add device ids for QCA9984, 88W8887 and 88W8964 radios
2a82f87 nl80211: back out early when receiving FAIL-BUSY reply
77c32f0 nl80211: fix code calculating average signal and rate

Signed-off-by: John Crispin <>
5 years agowireguard: no longer need portability patch
Kevin Darbyshire-Bryant [Thu, 17 May 2018 22:08:21 +0000 (23:08 +0100)]
wireguard: no longer need portability patch

Drop package/network/services/wireguard/patches/100-portability.patch

Instead pass 'PLATFORM=linux' to make since we are always building FOR

Signed-off-by: Kevin Darbyshire-Bryant <>
5 years agoramips: add support for YUKAI Engineering Inc. BOCCO
YuheiOKAWA [Thu, 22 Feb 2018 16:58:23 +0000 (01:58 +0900)]
ramips: add support for YUKAI Engineering Inc. BOCCO

BOCCO is a communication robot provided by YUKAI Engineering Inc.

SoC: MT7620A
MEM: 256MB
Flash: 8MB
NAND: 512MB (non support)
Include Sound DAC and AMP.
No Wired Ethernet.

Signed-off-by: YuheiOKAWA <>
5 years agoar71xx: WNDR3700: enable RFKILL function for Wifi button
Michal Cieslakiewicz [Thu, 22 Feb 2018 22:11:08 +0000 (23:11 +0100)]
ar71xx: WNDR3700: enable RFKILL function for Wifi button

Wifi button for WNDR3700 dual band routers has been renamed to "rfkill"
and its emitted keycode changed to KEY_RFKILL. This allows OpenWrt/LEDE
to support it 'out of the box' without additional tweaking.

Until this patch, button had been sending BTN_2 keycode which was
ignored by default (no action script present). To get expected behaviour
of switching radios on and off, user had to manually rename/link
'rfkill' script to name 'BTN_2' in /etc/rc.button directory.

This patch follows similar changes for other Netgear ar71xx routers,
for example WNR2000v3. It applies cleanly to both latest trunk and 17.01
branch. Tested on WNDR3700v1.

Signed-off-by: Michal Cieslakiewicz <>
5 years agobrcm2708: add squashfs rootfs image
Christian Lamparter [Tue, 27 Mar 2018 17:42:18 +0000 (19:42 +0200)]
brcm2708: add squashfs rootfs image

This patch adds a image with squashfs as the root filesystem.
A rootfs_data partition will be generated on the first boot
and placed inside the rootfs partition (just after the squashfs

 - it is possible to migrate from an existing -ext4
   installation and back via sysupgrade.
 - existing partition layout will not be lost.
 - slightly smaller image size.
 - support for attendedsysupgrade

 - needs f2fs + tools as well. This is because fs-tools decides on the
   blocksize of the sdcard. So either f2fs or ext4 can get choosen as
   the rootfs_data filesystem (depends on the size of the root partition).
 - rootfs_data is placed into the rootfs partition. This makes
   it difficult for tools that expect a /dev/mmc0pX device.
   It also makes it difficult for data recovery tools since they
   might not expect to find a embedded partition or will be

For people with existing build configurations: make sure to include mkf2fs
and f2fsck package into the image... Otherwise the new -squashfs image will
boot of a ram-overlay and won't keep the configurations after a reboot.

Cc: Álvaro Fernández Rojas <>
Cc: Paul Spooren <>
Cc: Daniel Golle <>
Signed-off-by: Christian Lamparter <>
Acked-by: Daniel Golle <>
5 years agox86: Add APU3 reference to x86 board.d
Kristian Evensen [Thu, 15 Mar 2018 12:30:27 +0000 (13:30 +0100)]
x86: Add APU3 reference to x86 board.d

There is a new APU-model available, APU3. The device is configured in
the same way as the APU1 and APU2, so the same LED/network setup can be

I considered changing the case to pc-engines-apu*, but I chose to follow
the existing pattern and add the full board name.

Signed-off-by: Kristian Evensen <>
5 years agoapm821xx: backport and reassign crypto4xx patches
Christian Lamparter [Sat, 28 Apr 2018 20:08:09 +0000 (22:08 +0200)]
apm821xx: backport and reassign crypto4xx patches

This patch backports several patches that went upstream into
Herbert Xu's cryptodev-2.6 tree:

crypto: Use zeroing memory allocator instead of allocator/memset
crypto: crypto4xx - performance optimizations
crypto: crypto4xx - convert to skcipher
crypto: crypto4xx - avoid VLA use
crypto: crypto4xx - add aes-ctr support
crypto: crypto4xx - properly set IV after de- and encrypt
crypto: crypto4xx - extend aead fallback checks
crypto: crypto4xx - put temporary dst sg into request ctx

The older, outstanding patches from 120-wxyz series have been
upstreamed as well and therefore they have been reassigned to
fit into the series.

Signed-off-by: Christian Lamparter <>
5 years agoapm821xx: don't select swconfig for the MR24
Christian Lamparter [Sat, 28 Apr 2018 20:08:08 +0000 (22:08 +0200)]
apm821xx: don't select swconfig for the MR24

The swconfig package is part of the DEFAULT_PACKAGES list in
the apm821xx's nand subtarget. It's enabled by default because
the MX60(W) and WNDR4700 need it for their initramfs. However
the package is not necessary for the MR24's sysupgrade image, as
the MR24 does not have a switch.

Signed-off-by: Christian Lamparter <>
5 years agoipq40xx: fix ethernet on ap-dk01.1
Roman Yeryomin [Fri, 4 May 2018 14:43:46 +0000 (17:43 +0300)]
ipq40xx: fix ethernet on ap-dk01.1

Not sure how it worked before but ethernet is disabled
by default in ipq4019 dtsi.

Signed-off-by: Roman Yeryomin <>
5 years agokernel: bump 4.14 to 4.14.41
Koen Vandeputte [Thu, 17 May 2018 16:41:26 +0000 (18:41 +0200)]
kernel: bump 4.14 to 4.14.41

Refreshed all patches

Dropped upstreamed patches:

updated patches:

Added new ARM64 symbol: CONFIG_ARM64_ERRATUM_1024718

Compile-tested on: cns3xxx, imx6, mvebu (arm64), x86_64
Runtime-tested on: cns3xxx, imx6, x86_64

Signed-off-by: Koen Vandeputte <>
5 years agokernel: bump 4.9 to 4.9.100
Koen Vandeputte [Thu, 17 May 2018 16:41:25 +0000 (18:41 +0200)]
kernel: bump 4.9 to 4.9.100

Refreshed all patches

Added new ARM64 symbol: ARM64_ERRATUM_1024718

Compile-tested on: ar71xx
Runtime-tested on: ar71xx

Signed-off-by: Koen Vandeputte <>
5 years agoath3k: Add firmware support for QCA Rome
Robert Marko [Mon, 2 Apr 2018 22:44:43 +0000 (00:44 +0200)]
ath3k: Add firmware support for QCA Rome

Add needed firmware for newer QCA Rome Bluetooth family.
This enables use of bluetooth with ath3k driver on QCA9377/9378 devices.

Signed-off-by: Robert Marko <>
5 years agobrcm63xx: Tune the network configuration for several routers
Daniel Gonzalez Cabanelas [Wed, 21 Feb 2018 17:24:58 +0000 (18:24 +0100)]
brcm63xx: Tune the network configuration for several routers

- DGND3700v1: fix port numbers, they're in reverse order
- FaST2704n: there is no WAN port identified as such, only 4 LAN
- AD1018: configure the "FIBRE" port as WAN

Signed-off-by: Daniel Gonzalez Cabanelas <>
5 years agoramips: mmc: Sync with staging driver
Rosen Penev [Wed, 16 May 2018 23:07:51 +0000 (16:07 -0700)]
ramips: mmc: Sync with staging driver

Mostly whitespace cleanups. Some unneeded code was removed.

MMC init was also moved to the probe function as in

The cleanup commits are over 100, making it hard to do them individually.

Tested on GnuBee PC1 with an SD card being used as swap.

Signed-off-by: Rosen Penev <>
v2: Fixed compile issue with mt7620

5 years agoath10k-firmware: Update QCA988X firmware to the latest version
Timo Sigurdsson [Wed, 16 May 2018 22:33:56 +0000 (00:33 +0200)]
ath10k-firmware: Update QCA988X firmware to the latest version

This patch updates the QCA988X firmware to the latest revision
found in the ath10k-firmware and linux-firmware repositories.

Tested on TP-Link Archer C7 v2 (ar71xx).

Signed-off-by: Timo Sigurdsson <>
5 years agoath79: remove unused copy-file build step
Matthias Schiffer [Wed, 9 May 2018 22:04:16 +0000 (00:04 +0200)]
ath79: remove unused copy-file build step

Signed-off-by: Matthias Schiffer <>
5 years agoath79: fix TP-Link TL-WR1043ND v1 image build
Matthias Schiffer [Wed, 9 May 2018 20:34:47 +0000 (22:34 +0200)]
ath79: fix TP-Link TL-WR1043ND v1 image build

Both initramfs and flashed images are built and boot. No Ethernet, no WLAN,
probably further issues, so the image is not added to TARGET_DEVICES for

Signed-off-by: Matthias Schiffer <>
5 years agowireguard: bump to 20180514
Kevin Darbyshire-Bryant [Wed, 16 May 2018 15:14:15 +0000 (16:14 +0100)]
wireguard: bump to 20180514

52be69b version: bump snapshot
4884b45 ncat-client-server: add wg-quick variant
a333551 wg-quick: add darwin implementation
f5bf84d compat: backport for OpenSUSE 15
fe1ae1b wg-quick: add wg symlink
ecc1c5f wg-quick: add android implementation
3e6bb79 tools: reorganize for multiplatform wg-quick
b289d12 allowedips: Fix graphviz output after endianness patch

Refresh cross compile compatibility patch

Signed-off-by: Kevin Darbyshire-Bryant <>
5 years agonetfilter: fix hardware offload regression (FS#1551)
Felix Fietkau [Thu, 17 May 2018 05:53:57 +0000 (07:53 +0200)]
netfilter: fix hardware offload regression (FS#1551)

The relevant code needs to be enabled even if flow table support is only
compiled as module

Signed-off-by: Felix Fietkau <>
5 years agoath79: add AVM FRITZ!WLAN Repeater 300E
Mathias Kresin [Fri, 2 Mar 2018 18:28:50 +0000 (19:28 +0100)]
ath79: add AVM FRITZ!WLAN Repeater 300E

No known issues, everything works fine.

Signed-off-by: Mathias Kresin <>
5 years agoath79: ar724x: fix pll settings
Mathias Kresin [Sun, 13 May 2018 11:02:30 +0000 (13:02 +0200)]
ath79: ar724x: fix pll settings

Add the syscon compatible, otherwise used functions like
syscon_regmap_lookup_by_phandle() will return an error and setting the
ethernet pll data wont work at all.

Fix the pll register width. Writing to registers out of the range via
syscon isn't possible and returns an error. On ar7242 the last pll
register - Current Audio Modulation Logic Output - is at 0x1805003c.

Signed-off-by: Mathias Kresin <>
5 years agoath79: add and enforce image metadata
Mathias Kresin [Mon, 7 May 2018 20:41:04 +0000 (22:41 +0200)]
ath79: add and enforce image metadata

Add metadata to images and make the metdata mandatory.

Signed-off-by: Mathias Kresin <>
5 years agoath79: cleanup image build code
Mathias Kresin [Mon, 7 May 2018 19:09:44 +0000 (21:09 +0200)]
ath79: cleanup image build code

Remove recipes for not yet migrated boards. Drop variables which aren't
required due to the switch to devicetree like boardname.

Drop the legacy image build code. Boards added to this target shouldn't
use the legacy image build code anyway.

Signed-off-by: Mathias Kresin <>
5 years agoath79: sysupgrade: drop unused variables
Mathias Kresin [Mon, 7 May 2018 18:58:20 +0000 (20:58 +0200)]
ath79: sysupgrade: drop unused variables

None of the variables are used at the moment.

Signed-off-by: Mathias Kresin <>
5 years agoath79: sysupgrade: drop unused platform checks
Mathias Kresin [Mon, 7 May 2018 18:55:20 +0000 (20:55 +0200)]
ath79: sysupgrade: drop unused platform checks

None of these checks are used at the moment and most of them aren't
required if the image metadata validation is used.

Signed-off-by: Mathias Kresin <>
5 years agoath79: sysupgrade: remove obsolete sysupgrade watchdog kill
Mathias Kresin [Mon, 7 May 2018 18:51:13 +0000 (20:51 +0200)]
ath79: sysupgrade: remove obsolete sysupgrade watchdog kill

The watchdog kill command was meant for busybox watchdog. Busybox watchdog
was replaced by the procd watchdog mid 2013 with commit df7ce9301a25
("busybox: disable the watchdog utility by default"), which makes the kill
command obsolete since quite some time.

Signed-off-by: Mathias Kresin <>
5 years agoath79: fix dts files
Mathias Kresin [Tue, 8 May 2018 05:22:38 +0000 (07:22 +0200)]
ath79: fix dts files

Add the SoC compatible to the individual dts files. Rename the dts files
to match the common pattern.

Remove dts files wich aren't used and no image in ar71xx exists.

Signed-off-by: Mathias Kresin <>
5 years agoath79: enable wireless
Mathias Kresin [Tue, 8 May 2018 04:40:54 +0000 (06:40 +0200)]
ath79: enable wireless

PCI attached wireless works fine. So enable the wireless config symbols.

Signed-off-by: Mathias Kresin <>
5 years agoath79: cleanup kernel config
Mathias Kresin [Tue, 15 May 2018 15:34:18 +0000 (17:34 +0200)]
ath79: cleanup kernel config

Remove outdated or not required kernel config settings.

Signed-off-by: Mathias Kresin <>
5 years agoath79: drop platform data patches
Mathias Kresin [Fri, 6 Apr 2018 19:27:30 +0000 (21:27 +0200)]
ath79: drop platform data patches

Signed-off-by: Mathias Kresin <>
5 years agoocteontx: make board.d files executable
Mathias Kresin [Mon, 7 May 2018 18:20:19 +0000 (20:20 +0200)]
octeontx: make board.d files executable

Add the executable permission to the files to ensure they run on

Signed-off-by: Mathias Kresin <>
5 years agoath79: make board.d files executable
Mathias Kresin [Fri, 6 Apr 2018 04:53:01 +0000 (06:53 +0200)]
ath79: make board.d files executable

Add the executable permission to the files to ensure they run on

Signed-off-by: Mathias Kresin <>
5 years agolantiq: kernel 4.14: use vbus-supply devicetree property
Mathias Kresin [Sun, 22 Apr 2018 15:03:31 +0000 (17:03 +0200)]
lantiq: kernel 4.14: use vbus-supply devicetree property

It isn't a phy supply, as the phy is still up if the GPIO is low. It
rather is the supply for the vbus. A correct setting/definition will be
relevant as soon as USB peripheral mode is supported.

Signed-off-by: Mathias Kresin <>
5 years agokernel: dwc2: add support for host mode external vbus supply
Mathias Kresin [Sun, 22 Apr 2018 14:55:02 +0000 (16:55 +0200)]
kernel: dwc2: add support for host mode external vbus supply

Backport patch adding support for the vbus-supply devicetree property,
which allows to specific GPIOs (via fixed regulators) to enable
vbus/usb power.

Signed-off-by: Mathias Kresin <>
5 years agolantiq: complete AVM FRITZ!Box 3370 support
Mathias Kresin [Fri, 9 Jun 2017 17:28:36 +0000 (19:28 +0200)]
lantiq: complete AVM FRITZ!Box 3370 support

Rename the image and use a compatible string which indicates that only
hardware revision 2 and higher is supported.

It allows to use the wireless LED, as HWRev 1 uses GPIO#39 for the
wireless LED and starting with HWRev 2 GPIO#35 is used for the wireless

The HWREV can be checked by connecting to the fritzbox right after power
on via ftp:

  ftp> quote GETENV HWSubRevision

Within the same HW revision 5 of the Fritz!Box 3370 different NAND flash
chips are used. Usually it isn't a big deal but depending on the used
NAND flash chip, the ECC calculation is done different (and incompatible
of course).

Boards with a Micron MT29F1G08ABADA NAND flash chip are using the NAND
chip to calculate the ECC (on-die). Boards with a Hynix HY27UF081G2M NAND
flash chip are doing the ECC calculation in software.

Supporting both with a single DTS isn't possible. It might be possible
to add a patch selecting the ECC mode dynamicaly based on the found NAND
flash chip. But such a patch has no chance to get accepted upstream and
most likely need to be touched with every kernel update.

Instead two images are created. One for Micron NAND flash chip and one
for Hynix NAND flash chip. So far no pattern is known to identify the
used flash chip without opening the box.

Add the power off GPIO. At least EVA version 2186 sets/keeps the GPIO as
input, which will cause a reboot 30sec after power on. For boards with
EVA version 2186 the installation is tricky as it has to be finished
within the 30sec time frame.

The EVA version can be checked by connecting to the fritzbox right after
power on via ftp:

  ftp> quote GETENV urlader-version

The ath9k eeprom/caldata is at a different and offset and stored in
reverse order (from the last byte to the beginning) on the flash.
Reverse the bits to bring the data into the format expected by the
ath9k driver.

Since the ath9k eeprom is stored in reverse order on flash, we can not
use the mac address from the on flash eeprom. Get the MAC address from
the tffs instead.

Within the same HW revision 5 of the Fritz!Box 3370 both version of the
vr9 SoC are used. During preparation of kernel 4.14 support, all
devicetree source files were changed to load the vr9 v1.1 and vr9 v1.2
gphy firmware, which fixed the embedded phys for boards using the
version 1.2 of the vr9 SoC.

While at it, add a trigger to make use of the LAN LED. Setup the
build-in switch and add a hint for LuCI two show the ports in order
matching the labels on the case

Add support for the second USB port and provide the volatage GPIOs. Use
GPIO#21 as PCIe reset pin. The lan led is connected to GPIO#38.

Name the rootfs partition ubi and remove the mtd/rootfs related kernel
bootargs to use the OpenWrt autoprobing based on the partition name.

Enable sysupgrade support to allow an upgrade from a running system.
Since sysupgrade wasn't supported till now, drop image build code which
was added to allow a sysupgrade from earlier OpenWrt versions.

Build images that allow an (initial) installation via EVA bootloader.

To install OpenWrt via Eva bootloader, within the first seconds after
power on a ftp connection need to be established to the FRITZ!Box at and the the following ftp commands need to be run:

  ftp> quote USER adam2
  ftp> quote PASS adam2
  ftp> binary
  ftp> debug
  ftp> passive
  ftp> quote SETENV linux_fs_start 0
  ftp> quote MEDIA FLSH
  ftp> put /path/to/openwrt-lantiq-xrx200-FRITZ3370-eva-kernel.bin mtd1
  ftp> put /path/to/openwrt-lantiq-xrx200-FRITZ3370-eva-filesystem.bin mtd0

Signed-off-by: Andrea Merello <>
Signed-off-by: Mathias Kresin <>
5 years agolantiq: switch to kernel 4.14
Mathias Kresin [Wed, 21 Mar 2018 07:39:41 +0000 (08:39 +0100)]
lantiq: switch to kernel 4.14

Use kernel 4.14 by default.

Signed-off-by: Mathias Kresin <>
5 years agoltq-xdsl-app: start after led script
Mathias Kresin [Wed, 7 Mar 2018 20:38:03 +0000 (21:38 +0100)]
ltq-xdsl-app: start after led script

During handshake we are highjack and reset a LED to the configured trigger
afterwards. ltq-xdsl-app need to start after the LED init script, to
ensure that the LED init script doesn't re-highjack the LED we are
currently using for handshake indication.

Drop the comment about the atm dependency. The dependency was fixed quite
some time ago by using hotplug scripts for br2684ctl.

Signed-off-by: Mathias Kresin <>
5 years agocurl: bump to 7.60.0
Hans Dedecker [Wed, 16 May 2018 12:09:48 +0000 (14:09 +0200)]
curl: bump to 7.60.0

Refresh patches; remove 320-mbedtls_dont_use_deprecated_sha256_function
patch as upstream fixed

For changes in version 2.60 see

Signed-off-by: Hans Dedecker <>
5 years agokmod-sched-cake: bump to latest 20180515
Kevin Darbyshire-Bryant [Tue, 15 May 2018 21:00:40 +0000 (22:00 +0100)]
kmod-sched-cake: bump to latest 20180515

Following changes as part of the kernel
upstreaming attempts.  And fix a slight fsck up
when calculating overheads for GSO packets.

Signed-off-by: Kevin Darbyshire-Bryant <>
5 years agoebtables: update to latest git 2018-05-15
Hans Dedecker [Tue, 15 May 2018 12:00:37 +0000 (14:00 +0200)]
ebtables: update to latest git 2018-05-15

66a9701 ebtables: Fix build errors and warnings
9fff3d5 include: Fix musl libc compatibility
b1cdae8 extensions: Add string filter to ebtables

Signed-off-by: Hans Dedecker <>
5 years agomt76: update to the latest version
Felix Fietkau [Tue, 15 May 2018 11:18:22 +0000 (13:18 +0200)]
mt76: update to the latest version

b2ecc52 mt76x2: fix avg_rssi estimation
fd58b28 mt76x2: add a polling delay in mt76x2_mac_stop routine
a78673d mt76: fix sending encrypted broadcast packets for secondary interfaces
e87f925 mt76x2: apply coverage class on slot time too

Signed-off-by: Felix Fietkau <>
5 years agoRevert "ramips: mmc: Sync with staging drivers"
John Crispin [Tue, 15 May 2018 10:42:40 +0000 (12:42 +0200)]
Revert "ramips: mmc: Sync with staging drivers"

This reverts commit 2d401925b9bdf34de800fff57f593a3665de2a6a.

Signed-off-by: John Crispin <>
5 years agostaging: mt7621-eth: fix return value check in mt7621_gsw_probe()
Wei Yongjun [Mon, 14 May 2018 18:14:25 +0000 (11:14 -0700)]
staging: mt7621-eth: fix return value check in mt7621_gsw_probe()

In case of error, the function devm_ioremap_resource() returns ERR_PTR()
and never returns NULL. The NULL test in the return value check should
be replaced with IS_ERR().

Fixes: f079b6406348 ("staging: mt7621-eth: add gigabit switch driver (GSW)")
Signed-off-by: Wei Yongjun <>
Signed-off-by: Greg Kroah-Hartman <>
5 years agostaging: mt7621-eth: fix return value check in mtk_connect_phy_node()
Wei Yongjun [Mon, 14 May 2018 18:14:24 +0000 (11:14 -0700)]
staging: mt7621-eth: fix return value check in mtk_connect_phy_node()

In case of error, the function of_phy_connect() returns NULL pointer not
ERR_PTR(). The IS_ERR() test in the return value check should be
replaced with NULL test.

Fixes: e3cbf478f846 ("staging: mt7621-eth: add the drivers core files")
Signed-off-by: Wei Yongjun <>
Signed-off-by: Greg Kroah-Hartman <>
5 years agostaging: mt7621-eth: fix return value check in mtk_probe()
Wei Yongjun [Mon, 14 May 2018 18:14:23 +0000 (11:14 -0700)]
staging: mt7621-eth: fix return value check in mtk_probe()

In case of error, the function devm_ioremap_resource() returns ERR_PTR()
and never returns NULL. The NULL test in the return value check should
be replaced with IS_ERR().

Fixes: e3cbf478f846 ("staging: mt7621-eth: add the drivers core files")
Signed-off-by: Wei Yongjun <>
Signed-off-by: Greg Kroah-Hartman <>
5 years agoramips: pci: sync with staging driver
Rosen Penev [Mon, 14 May 2018 18:14:22 +0000 (11:14 -0700)]
ramips: pci: sync with staging driver

This is an amagalmation of two upstream commits dealing with whitespace
and dead code removal. I'm synching instead of having two separate commits
as they go out of order compared to previous commits here.

Tested on GnuBee PC1.

Signed-off-by: Rosen Penev <>
5 years agoramips: mmc: Sync with staging drivers
Rosen Penev [Mon, 14 May 2018 18:14:21 +0000 (11:14 -0700)]
ramips: mmc: Sync with staging drivers

Mostly whitespace cleanups. Some unneeded code was removed.

MMC init was also moved to the probe function as in

The cleanup commits are over 100, making it hard to do them individually.

Tested on GnuBee PC1 with an SD card being used as swap.

Signed-off-by: Rosen Penev <>
5 years agowireguard: Add support for ip6prefix config option
Toke Høiland-Jørgensen [Sun, 13 May 2018 18:15:50 +0000 (20:15 +0200)]
wireguard: Add support for ip6prefix config option

This makes it easier to distribute prefixes over a wireguard tunnel
interface, by simply setting the ip6prefix option in uci (just like with
other protocols).

Obviously, routing etc needs to be setup properly for things to work; this
just adds the config option so the prefix can be assigned to other

Signed-off-by: Toke Høiland-Jørgensen <>
5 years agolinux-firmware: Add firmware for usb-serial-ti-usb
David Woodhouse [Thu, 16 Feb 2017 00:23:27 +0000 (00:23 +0000)]
linux-firmware: Add firmware for usb-serial-ti-usb

Signed-off-by: David Woodhouse <>
5 years agoramips: Add i2c support for mt7620n
Matthias Badaire [Tue, 8 May 2018 13:42:05 +0000 (15:42 +0200)]
ramips: Add i2c support for mt7620n

I2c is supported on mt7620n like on mt7620a when checking the datasheet

Signed-off-by: Matthias Badaire <>
5 years agokirkwood: support for buttons in Audi and Viper
Alberto Bursi [Wed, 9 May 2018 20:38:20 +0000 (22:38 +0200)]
kirkwood: support for buttons in Audi and Viper

Both these devices have a wps and a reset button on
GPIO pins, which need kmod-gpio-button-hotplug
package to work.

Add this package to their default package config.

Troubleshooted and tested on a Viper.

Signed-off-by: Alberto Bursi <>
5 years agoath79: fix gmac compatible in ar9330.dtsi and ag71xx_setup_gmac_933x
Johann Neuhauser [Sat, 12 May 2018 20:25:35 +0000 (22:25 +0200)]
ath79: fix gmac compatible in ar9330.dtsi and ag71xx_setup_gmac_933x

1. compatible property in node gmac was wrong

2. ag71xx_setup_gmac_933x should use np of gmac-config and
   not the pointer to gmac. gmac is only used for the reg address.

Signed-off-by: Johann Neuhauser <>
5 years agomac80211: refactor non-{sae,dfs} mesh initialization
Daniel Golle [Mon, 14 May 2018 21:55:24 +0000 (23:55 +0200)]
mac80211: refactor non-{sae,dfs} mesh initialization

Refactor mesh initialization into a separate function, do some cleaning
on the way to make the code more readable.
 * Move iw mesh setup to new mac80211_setup_mesh()
 * fallback on 'ssid' parameter in case 'mesh_id' isn't set
 * move setting of freq variable to shared code as it is needed for
   both, the wpa_supplicant and the iw based setup.

Signed-off-by: Daniel Golle <>
5 years agohostapd: fix VHT80 for encrypted mesh channel settings
Sven Eckelmann [Mon, 14 May 2018 12:45:25 +0000 (14:45 +0200)]
hostapd: fix VHT80 for encrypted mesh channel settings

The max_oper_chwidth settings was parsed incorrectly for big endian system.
This prevented the system to switch to VHT80 (or VHT160). Instead they were
mapped to:

* HT20:   20MHz
* VHT20:  20MHz
* HT40:   40MHz
* VHT40:  40MHz
* VHT80:  40MHz
* VHT160: 40MHz

This happened because each max_oper_chwidth setting in the config file was
parsed as "0" instead of the actual value.

Fixes: a4322eba2b12 ("hostapd: fix encrypted mesh channel settings")
Signed-off-by: Sven Eckelmann <>
5 years agomac80211: Re-enable encrypted 11s meshpoint
Sven Eckelmann [Mon, 14 May 2018 13:11:45 +0000 (15:11 +0200)]
mac80211: Re-enable encrypted 11s meshpoint

The commit 574e4377fad5 ("mac80211: properly setup mesh interface") uses
the variable $wpa to decide whether encrypted meshpoint is requested by the
user or not. But the variable $wpa will only be set correctly after the
function wireless_vif_parse_encryption is called.

Fixes: 574e4377fad5 ("mac80211: properly setup mesh interface")
Signed-off-by: Sven Eckelmann <>
5 years agomvebu: add support for WRT32X (venom)
Michael Gray [Sun, 13 May 2018 13:29:30 +0000 (23:29 +1000)]
mvebu: add support for WRT32X (venom)

Linksys WRT32X (Venom) is identical in hardware to the WRT3200ACM
with a different flash layout and boots zImage rather than uImage.

- Marvell Armada 385 88F6820 (2x 1.8GHz)
- 256MB of Flash
- 512MB of RAM
- 2.4GHz (bgn) and 5GHz (an+ac wave 2)
- 4x 1Gbps LAN + 1x 1Gbps WAN
- 1x USB 3.0 and 1x USB 2.0/eSATA (combo port)

Flash instruction:
Apply factory image via web-gui.

Signed-off-by: Michael Gray <>
5 years agofirewall3: update to latest git HEAD
John Crispin [Mon, 14 May 2018 14:22:57 +0000 (16:22 +0200)]
firewall3: update to latest git HEAD

b45e162 helpers: fix the set_helper in the rule structure
f742ba7 helpers.conf: support also tcp in the CT sip helper
08b2c61 helpers: make the proto field as a list rather than one option

Signed-off-by: John Crispin <>
5 years agogemini: Augment DIR-685 device tree for RTL8366RB
Linus Walleij [Fri, 11 May 2018 17:16:48 +0000 (19:16 +0200)]
gemini: Augment DIR-685 device tree for RTL8366RB

This makes the ethernet and RTL8366RB come up on the
D-Link DIR-685.

Signed-off-by: Linus Walleij <>
5 years agogemini: create zImage for DIR-685
Linus Walleij [Fri, 11 May 2018 17:10:16 +0000 (19:10 +0200)]
gemini: create zImage for DIR-685

Signed-off-by: Linus Walleij <>
5 years agoramips: remove unnecessary resource details.
NeilBrown [Sat, 12 May 2018 19:57:48 +0000 (12:57 -0700)]
ramips: remove unnecessary resource details.

These resources are extracted from devicetree, so they aren't
needed here.

Signed-off-by: NeilBrown <>
Signed-off-by: Greg Kroah-Hartman <>
5 years agoramips: remove conditional compilation.
NeilBrown [Sat, 12 May 2018 19:57:47 +0000 (12:57 -0700)]
ramips: remove conditional compilation.

Code currently defines:

and then compiles code only if they are defined.
We might want to disable some of these via devicetree one
day, but for now just remove the #defines and the
conditions - all the code for different ports is
easy to identify.

Signed-off-by: NeilBrown <>
Signed-off-by: Greg Kroah-Hartman <>
5 years agoramips: improve interrupt mapping
NeilBrown [Sat, 12 May 2018 19:57:46 +0000 (12:57 -0700)]
ramips: improve interrupt mapping

As the Interrupts for the PCI adapters are listed in
devicetree we shouldn't need to have them explicit in the code.

The simplest way to do this is to use of_irq_parse_and_map_pci()
and specify an interrupt-map which identifies the different
PCI hosts by bus/slot numbers.
This has the advantage that the hwirq number are mapped to virq
numbers for us, so the ugly hack can go.

Signed-off-by: NeilBrown <>
Signed-off-by: Greg Kroah-Hartman <>
5 years agoramips: Remove redundant owner assignment
Christian Lütke-Stetzkamp [Sat, 12 May 2018 19:57:45 +0000 (12:57 -0700)]
ramips: Remove redundant owner assignment

Remove the owner assignment form the platform driver as
platform_driver_register() already initializes the owner.
Found using coccinelle.

Signed-off-by: Christian Lütke-Stetzkamp <>
Signed-off-by: Greg Kroah-Hartman <>
5 years agoramips: Move PCI driver to files directory
Rosen Penev [Sat, 12 May 2018 19:57:44 +0000 (12:57 -0700)]
ramips: Move PCI driver to files directory

This will make it easier to port upstream code.

Signed-off-by: Rosen Penev <>
5 years agomac80211: properly setup mesh interface
Daniel Golle [Sun, 13 May 2018 03:20:39 +0000 (05:20 +0200)]
mac80211: properly setup mesh interface

Setup wpa_supplicant for encrypted mesh or when using DFS channels and
adjust interface setup to pass fixed frequency for mesh mode.

Signed-off-by: Daniel Golle <>
5 years agohostapd: fix mesh+AP
Daniel Golle [Sat, 12 May 2018 19:08:18 +0000 (21:08 +0200)]
hostapd: fix mesh+AP

Fix encrypted (or DFS) AP+MESH interface combination in a way similar
to how it's done for AP+STA and fix netifd shell script.
Refresh patches while at it.

Signed-off-by: Daniel Golle <>
5 years agorpcd: update to lastest HEAD
Jo-Philipp Wich [Mon, 14 May 2018 05:16:13 +0000 (07:16 +0200)]
rpcd: update to lastest HEAD

8206219 uci: fix memory leak in rpc_uci_replace_savedir()
10f7878 exec: close stdout and stderr streams on child signal
92d0d75 uci: use correct sort index when reordering sections

Signed-off-by: Jo-Philipp Wich <>
5 years agowireguard: bump to 0.0.20180513
Kevin Darbyshire-Bryant [Sun, 13 May 2018 19:05:15 +0000 (20:05 +0100)]
wireguard: bump to 0.0.20180513

6b4a340 version: bump snapshot
faa2103 compat: don't clear header bits on RHEL
4014532 compat: handle RHEL 7.5's recent backports
66589bc queueing: preserve pfmemalloc header bit
37f114a chacha20poly1305: make gcc 8.1 happy
926caae socket: use skb_put_data
724d979 wg-quick: preliminary support for go implementation
c454c26 allowedips: simplify arithmetic
71d44be allowedips: produce better assembly with unsigned arithmetic
5e3532e allowedips: use native endian on lookup
856f105 allowedips: add selftest for allowedips_walk_by_peer
41df6d2 embeddable-wg-library: zero attribute padding
9a1bea6 keygen-html: add zip file example
f182b1a qemu: retry on 404 in wget for race

Signed-off-by: Kevin Darbyshire-Bryant <>
5 years agoat91: sama5d4: fix bad mkubifs options
Jo-Philipp Wich [Sun, 13 May 2018 16:33:47 +0000 (18:33 +0200)]
at91: sama5d4: fix bad mkubifs options

Drop the "-x lzo" compression type override from MKUBIFS_OPTS since we
do not support it in mkfs.ubifs. Attempting to use LZO compression results
in "Error: bad compressor name".

The build system also appends an explicit "--compr=zlib" to the mkfs command
line, turning the "-x" flag into a no-op anyway.

Signed-off-by: Jo-Philipp Wich <>
5 years agokernel: let kmod-rtl8366-smi conditionally depend on kmod-of-mdio
Jo-Philipp Wich [Sun, 13 May 2018 16:21:59 +0000 (18:21 +0200)]
kernel: let kmod-rtl8366-smi conditionally depend on kmod-of-mdio

We cannot depend on CONFIG_OF in the module definition context as this symbol
is not defined for OpenWrt menuconfig. Depend on the targets that appear to
need the kmod-of-mdio module instead.

The target dependency list may not be complete, it is based on the build
failures encountered by the build bots.

Fixes: dc629d9cf5 ("kernel: fix kmod-switch-rtl8366-smi dependency")
Signed-off-by: Jo-Philipp Wich <>
5 years agokernel: use accepted version of bcm47xxpart fix commit
Rafał Miłecki [Sat, 12 May 2018 21:34:43 +0000 (23:34 +0200)]
kernel: use accepted version of bcm47xxpart fix commit

Signed-off-by: Rafał Miłecki <>
5 years agokernel: backport mtd patch for minor partitioning cleanup
Rafał Miłecki [Sat, 12 May 2018 21:12:32 +0000 (23:12 +0200)]
kernel: backport mtd patch for minor partitioning cleanup

Signed-off-by: Rafał Miłecki <>
5 years agokernel: fix build error in flow offload code with CONFIG_NETFILTER=n
Felix Fietkau [Tue, 8 May 2018 21:17:36 +0000 (23:17 +0200)]
kernel: fix build error in flow offload code with CONFIG_NETFILTER=n

Signed-off-by: Felix Fietkau <>
5 years agodnsmasq: bump to 2.80test2
Kevin Darbyshire-Bryant [Sat, 12 May 2018 14:37:33 +0000 (15:37 +0100)]
dnsmasq: bump to 2.80test2

Refresh patches and backport upstream to current HEAD:

1f1873a Log warning on very large cachesize config, instead of truncating it.
0a496f0 Do unsolicited RAs for interfaces which appear after dnsmasq startup.
e27825b Fix logging in previous.
1f60a18 Retry SERVFAIL DNSSEC queries to a different server, if possible.
a0088e8 Handle query retry on REFUSED or SERVFAIL for DNSSEC-generated queries.
34e26e1 Retry query to other servers on receipt of SERVFAIL rcode.
6b17335 Add packet-dump debugging facility.
07ed585 Add logging for DNS error returns from upstream and local configuration.
0669ee7 Fix DHCP broken-ness when --no-ping AND --dhcp-sequential-ip are set.
f84e674 Be persistent with broken-upstream-DNSSEC warnings.

Compile & run tested: ar71xx Archer C7 v2

Signed-off-by: Kevin Darbyshire-Bryant <>
5 years agobcm53xx: backport the first bunch of 4.18 BCM5301X patches
Rafał Miłecki [Sat, 12 May 2018 13:38:53 +0000 (15:38 +0200)]
bcm53xx: backport the first bunch of 4.18 BCM5301X patches

Signed-off-by: Rafał Miłecki <>
5 years agoat91: cleanup kernel configurations
Jo-Philipp Wich [Fri, 11 May 2018 16:24:00 +0000 (18:24 +0200)]
at91: cleanup kernel configurations

The subtarget configurations contained a lot of reduntant symbols, arbitrary
overrides and symbols set to =m.

Remove most of these redundant settings to shrink the subtarget configurations
to a manageable size.

Signed-off-by: Jo-Philipp Wich <>
5 years agoat91: sama5d2: remove CONFIG_XZ_DEC override
Jo-Philipp Wich [Fri, 11 May 2018 14:40:34 +0000 (16:40 +0200)]
at91: sama5d2: remove CONFIG_XZ_DEC override

Fixes the following error spotted by the buildbot:

    Package kmod-fs-squashfs is missing dependencies for the following libraries:

Signed-off-by: Jo-Philipp Wich <>
5 years agoat91: add missing config symbols
Jo-Philipp Wich [Fri, 11 May 2018 14:01:26 +0000 (16:01 +0200)]
at91: add missing config symbols

Signed-off-by: Jo-Philipp Wich <>
5 years agokernel: rtl8366_smi: guard of_mdiobus_register() call
Jo-Philipp Wich [Fri, 11 May 2018 13:07:56 +0000 (15:07 +0200)]
kernel: rtl8366_smi: guard of_mdiobus_register() call

Make the call to of_mdiobus_register() dependant on CONFIG_OF to avoid
pulling in an unwanted dependency on of_mdio.ko.

Fixes: 3dc523f232 kernel: add OF support to rtl8355 driver
Signed-off-by: Jo-Philipp Wich <>
5 years agodnsmasq: add specific interface procd triggers
hux [Mon, 7 May 2018 22:37:48 +0000 (00:37 +0200)]
dnsmasq: add specific interface procd triggers

Right now interface.update events are sent out by netifd upon interface state,
route, address (lifetime), prefix lifetime changes.
Dnsmasq is only interested in interface state changes and currently adds an
interface trigger for all the "interface.*" events.
In combination with commit 23bba9cb330, which triggers a SIGHUP signal to dnsmasq,
IPv6 address/prefix lifetime changes on the wan will trigger dnsmasq reloads which
can become frequent in case of shorter lifetimes.

To avoid frequent dnsmasq reload, this patch adds specific interface triggers.
During dnsmasq init it loops dhcp uci section; if the value of the ignore option
is set to 0, then the corresponding interface trigger is not installed.
Otherwise, if the ignore option value is 1, then procd_add_interface_trigger is
called which adds the interface trigger.

Signed-off-by: hux <>
5 years agokernel: fix kmod-switch-rtl8366-smi dependency
John Crispin [Thu, 10 May 2018 08:22:19 +0000 (10:22 +0200)]
kernel: fix kmod-switch-rtl8366-smi dependency

Signed-off-by: John Crispin <>
5 years agoar71xx: fix and improve ALFA Network Tube2H support
Piotr Dymacz [Wed, 9 May 2018 18:04:01 +0000 (20:04 +0200)]
ar71xx: fix and improve ALFA Network Tube2H support

Fix ART offset (make it universal for 8/16 MB versions of the board) and
while at it, include also GPIO setup for h/w watchdog (EM6324QYSP5B).

Fixes: FS#1532
Signed-off-by: Piotr Dymacz <>
5 years agokernel: fix chipidea2 dependencies
John Crispin [Wed, 9 May 2018 17:55:10 +0000 (19:55 +0200)]
kernel: fix chipidea2 dependencies

Signed-off-by: John Crispin <>
5 years agonghttp2: bump to 1.32.0
Hans Dedecker [Wed, 9 May 2018 09:24:50 +0000 (11:24 +0200)]
nghttp2: bump to 1.32.0

572735e4 Update manual pages
e8d693c3 Bump up version number to 1.32.0, LT revision to 30:2:16
f44dfcd9 Update AUTHORS
1f1b0d93 Update manual pages
ce8c749b Merge pull request #1173 from nghttp2/asio-client-sni
3e4f257b asio: Support client side SNI
86fab997 Upgrade neverbleed to the latest master
c3ecd445 Merge pull request #1171 from nghttp2/h2load-rate-and-duration
c65ca20a h2load: -r and --duration are mutually exclusive
a5c408c5 Ignore all input after calling session_terminate_session
06379b28 Fix treatment of padding
e04de48e Merge pull request #1162 from nghttp2/libressl
00964642 Use LIBRESSL_IN_USE instead of defined(LIBRESSL_VERSION_NUMBER)
8d0b4544 libressl 2.7 has X509_VERIFY_PARAM_*
d8a34131 libressl 2.7 has SSL_CTX_get0_certificate
5db17d0a Compile with libressl 2.7.2
3febaef1 Bump up LT revision to 30:1:16 due to v1.31.1 release
b1bd6035 Fix frame handling
b48bcb21 examples: Use C style comment in .c files
6f3ce2c7 examples: Remove unused lambda capture
2f9121cf Merge branch 'Sp1l-Sp1l/allow-no-npn'
e65e7711 Add comment on #endif
636ef51b Fix compile error with -Wunused-function
400934e5 [PATCH] Allow building without NPN
4c3a3acf Merge pull request #1146 from vszakats/cmakestaticlib
9aa6002c Merge pull request #1144 from hellojaewon/master
f342260b cmake: add ENABLE_STATIC_LIB option to build static lib
a6dd4970 Fix typo
842509da Don't allow 101 HTTP status code because HTTP/2 removes HTTP Upgrade
4add618a Bump up version number to 1.32.0-DEV

Signed-off-by: Hans Dedecker <>
5 years agoath79: Ubiquiti Airmax M: add relocate-kernel to invalidate cache
Matthias Schiffer [Thu, 26 Apr 2018 18:04:48 +0000 (20:04 +0200)]
ath79: Ubiquiti Airmax M: add relocate-kernel to invalidate cache

Apply ar71xx fix for broken Ubiquiti bootloader to ath79 as well.

Signed-off-by: Matthias Schiffer <>
5 years agoramips: move mtk-mmc init to probe function to avoid breaking NAND flash
Felix Fietkau [Tue, 8 May 2018 15:45:00 +0000 (17:45 +0200)]
ramips: move mtk-mmc init to probe function to avoid breaking NAND flash

The driver messes with the pin control settings - MMC and NAND are
mutually exclusive

Signed-off-by: Felix Fietkau <>
5 years agoramips: clean up the mt7621 nand flash driver hacks, add support for the xiaomi mir3g...
Felix Fietkau [Tue, 8 May 2018 15:44:23 +0000 (17:44 +0200)]
ramips: clean up the mt7621 nand flash driver hacks, add support for the xiaomi mir3g flash chip

Signed-off-by: Felix Fietkau <>
5 years agokmod-sched-cake: bump to latest cake 2018-05-07
Kevin Darbyshire-Bryant [Tue, 8 May 2018 08:39:22 +0000 (08:39 +0000)]
kmod-sched-cake: bump to latest cake 2018-05-07

No functional change.  Code tidy ups.

735eaf2 Make sure we don't reallocate q->tins (we didn't anyway but his
really makes sure)
6c5ad6e Get rid of __GFP_NOWARN flag for memory allocation
2a37333 Don't need the wrapper for kvfree, and no need to check before calling it
2b1c631 Whitespace fix
7fe6e28 compat tidyup (for older kernel versions <4.4)
93b805c pedant tidy up superfluous semicolons on switch statements

Signed-off-by: Kevin Darbyshire-Bryant <>
5 years agobrcm47xx: switch to the kernel 4.14
Rafał Miłecki [Sat, 5 May 2018 22:06:14 +0000 (00:06 +0200)]
brcm47xx: switch to the kernel 4.14

The biggest (and the only?) disadventage of this is obviously an
increased image size.

For mips74k the size of vmlinux goes up from 4186684 B to 4701436 B.
Most devices use LZMA compressed kernel so probably more important is
vmlinux.lzma size which goes up from 1342945 B to the 1508498 B.

Still this isn't something that should stop target kernel bump. There
are various adventages of kernel 4.14. If kernel / image size is a
serious concern for anyone, it's perfectly possible to use previous
release which is pretty solid for the brcm47xx.

Signed-off-by: Rafał Miłecki <>
Cc: Paul Wassi <>
Cc: Hauke Mehrtens <>