openwrt/staging/blocktrron.git
4 years agomediatek: mt7623: remove left-over fragments of mt7623a-rfb-emmc
Adrian Schmutzler [Thu, 30 Jan 2020 17:55:21 +0000 (18:55 +0100)]
mediatek: mt7623: remove left-over fragments of mt7623a-rfb-emmc

The image creation for the mt7623a-rfb-emmc has been removed during
a patch refresh without specific comment. The corresponding base-files
entries and DTS patches for 4.14 are still there.

Since mt7623 is pretty dead and nobody has missed this device, let's
just remove the rest.

Fixes: 050da2107a7e ("mediatek: backport upstream mediatek patches")
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Acked-by: John Crispin <john@phrozen.org>
4 years agoramips: use tpt DTS trigger for TP-Link TL-MR3020 v3 and TL-WA801ND v5
Jan Alexander [Wed, 29 Jan 2020 14:03:18 +0000 (15:03 +0100)]
ramips: use tpt DTS trigger for TP-Link TL-MR3020 v3 and TL-WA801ND v5

This converts the TP-Link TL-MR3020v3 board to use the WLAN throughput
LED trigger in order to react to all VAPs.

It also moves the WLAN trigger config of the TP-Link TL-WA801NDv5 to the
DTS and merges the now identical LAN LED configs.

Verified these changes on a TL-MR3020v3.

Signed-off-by: Jan Alexander <jan@nalx.net>
[changed commit title and extended commit message]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agoRevert "ramips: fix NETGEAR R6260 EEPROM offset"
David Bauer [Thu, 30 Jan 2020 11:37:28 +0000 (12:37 +0100)]
Revert "ramips: fix NETGEAR R6260 EEPROM offset"

This reverts commit 4716c843d62a5a4c37e7886f54bcf446aa26e786.

Netgear seems to use different partition layouts on the R6260, which
would require us to dynamically detect the position of (at least) the
factory partition.

Revert this fix to avoid breaking existing installations until a better
solution has been worked out.

Signed-off-by: David Bauer <mail@david-bauer.net>
4 years agokernel: add support for XTX xt25f128 SPI-NOR flash chips
Daniel Golle [Thu, 30 Jan 2020 09:31:26 +0000 (11:31 +0200)]
kernel: add support for XTX xt25f128 SPI-NOR flash chips

This fixes support for Teltonika RUT9xx which in recent versions of
the device uses xt25f128b flash.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
4 years agoramips: remove duplicate DEVICE_PACKAGES for TP-Link Archer C20i
Adrian Schmutzler [Wed, 29 Jan 2020 19:27:25 +0000 (20:27 +0100)]
ramips: remove duplicate DEVICE_PACKAGES for TP-Link Archer C20i

DEVICE_PACKAGES is specified twice for the same device. Remove the
first (=older) assignment.

Fixes: 40692f0fb55c ("ramips: mt7620: select only the matching mt76 driver")
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agoath79: add support for GL.iNet 6408/6416 (GL.iNet V1)
Adrian Schmutzler [Sun, 8 Dec 2019 19:18:58 +0000 (20:18 +0100)]
ath79: add support for GL.iNet 6408/6416 (GL.iNet V1)

This ports the GL.iNet 6408/6416 from ar71xx.

The GL-Connect GL.iNet v1 routers are basically a TP-Link TL-WR710N with
more DRAM/Flash and console/GPIO header in the same small form-factor.

Specifications:
 - SoC: Atheros AR9331
 - CPU: 400 MHz
 - Flash: 8/16 MiB
 - RAM: 64 MiB
 - WiFi: 2.4 GHz b/g/n (SoC)
 - Ethernet: 2x 100M ports (LAN/WAN)
 - USB: 1x 2.0

The difference between 6408 and 6416 is just the flash size. It looks like
only the 16 MiB version has been advertised, while the 6408 is a modified
version. There are also 1-port versions sold by third parties.

Installation:
Install the sysupgrade image via stock firmware GUI or upload it via uboot
(web-based). The device will be available at 192.168.1.1.

Attention: In ar71xx, the same board name is used for both flash versions.
So, please make sure you flash the correct ath79 image when upgrading.

This has been device-tested on a GL.iNet 6416.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agoopkg: update to latest Git HEAD
Jo-Philipp Wich [Sat, 25 Jan 2020 13:27:10 +0000 (14:27 +0100)]
opkg: update to latest Git HEAD

80d161e opkg: Fix -Wformat-overflow warning
c09fe20 libopkg: fix skipping of leading whitespace when parsing checksums

Fixes: CVE-2020-7982
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
4 years agoramips: fix NETGEAR R6260 EEPROM offset
David Bauer [Tue, 28 Jan 2020 20:19:16 +0000 (21:19 +0100)]
ramips: fix NETGEAR R6260 EEPROM offset

The EEPROM offset for the NETGEAR R6260 is incorrect, thus no valid
calibration data is used.

Fix this only for the NETGEAR R6260, as it's currently unknown whether
or not other boards are affected.

Signed-off-by: David Bauer <mail@david-bauer.net>
4 years agoramips: add support for GL.iNet microuter-N300
David Bauer [Sun, 5 Jan 2020 23:31:24 +0000 (00:31 +0100)]
ramips: add support for GL.iNet microuter-N300

The GL.iNet microuter-N300 (internally referred as MT300N-v4) is a
pocket-size travel router. It is essentially identical to the VIXMINI
(internally referred as MT300N-v3) but with double the RAM and
SPI-flash.

Additionally, set the label-mac for both the VIXMINI as well as the
microuter-N300.

Hardware
--------
SoC:   MediaTek MT7628NN
RAM:   128M DDR2
FLASH: 16M
LED:   Power - WLAN
BTN:   Reset
UART:  115200 8N1
       TX and RX are labled on the board as pads next to the SoC

Installation via web-interface
------------------------------
1. Visit the web-interface at 192.168.8.1
   Note: The ethernet port is by default WAN. So you need to connect to
   the router via WiFi

2. Navigate to the Update tab on the left side.

3. Select "Local Update"

4. Upload the OpenWrt sysupgrade image.
   Note: Make sure you select not to preserve the configuration.

Installation via U-Boot
-----------------------
1. Hold down the reset button while powering on the device.
   Wait for the LED to flash 5 times.

2. Assign yourself a static IPv4 in 192.168.1.0/24

3. Upload the OpenWrt sysupgrade image at 192.168.1.1.

Signed-off-by: David Bauer <mail@david-bauer.net>
4 years agohostapd: add back support for passing CSA events from sta/mesh to AP interfaces
Felix Fietkau [Tue, 28 Jan 2020 15:27:53 +0000 (16:27 +0100)]
hostapd: add back support for passing CSA events from sta/mesh to AP interfaces

Fixes handling CSA when using AP+STA or AP+Mesh
This change was accidentally dropped in commit 167028b75
("hostapd: Update to version 2.9 (2019-08-08)")

Signed-off-by: Felix Fietkau <nbd@nbd.name>
4 years agomac80211: backport airtime queue limits support
Felix Fietkau [Sat, 18 Jan 2020 17:44:00 +0000 (18:44 +0100)]
mac80211: backport airtime queue limits support

Signed-off-by: Felix Fietkau <nbd@nbd.name>
4 years agomac80211: backport fix for an no-ack tx status issue
Felix Fietkau [Sat, 18 Jan 2020 17:41:08 +0000 (18:41 +0100)]
mac80211: backport fix for an no-ack tx status issue

Signed-off-by: Felix Fietkau <nbd@nbd.name>
4 years agowireguard: bump to 0.0.20200128
Jason A. Donenfeld [Tue, 28 Jan 2020 15:55:16 +0000 (16:55 +0100)]
wireguard: bump to 0.0.20200128

This fixes a few small oversights for the 5.5 compat layer.

Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
4 years agohostapd: unconditionally enable ap/mesh for wpa-cli
Felix Fietkau [Tue, 28 Jan 2020 13:12:08 +0000 (14:12 +0100)]
hostapd: unconditionally enable ap/mesh for wpa-cli

Without this change, wpa-cli features depend on which wpad build variant was
used to build the wpa-cli package

Signed-off-by: Felix Fietkau <nbd@nbd.name>
4 years agoocteon: apply vendor_model scheme to device definition/image name
Adrian Schmutzler [Sat, 18 Jan 2020 01:18:49 +0000 (02:18 +0100)]
octeon: apply vendor_model scheme to device definition/image name

This updates the device definition name for octeon target to provide
more useful names for the images and be consistent with the increasing
number of targets following that scheme.

Since the target is not using device tree yet, this does not touch
board_name and thus sets BOARD_NAME in image Makefile to ensure
sysupgrade is still working.

While at it, move Build block before Device blocks and remove trailing
whitespace for CMDLINE.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agomediatek: use consistent naming scheme for device nodes
Adrian Schmutzler [Tue, 14 Jan 2020 14:38:51 +0000 (15:38 +0100)]
mediatek: use consistent naming scheme for device nodes

This harmonizes the device node names (and thus the image names, too)
between subtargets of the mediatek target. So far, each subtarget
has somewhat used its own naming scheme. Now, we use the vendor_device
syntax there, too.

Since DTS names have different patterns and the target only contains
a few devices, this does not replace DEVICE_DTS by a calculated
default value (like for other targets).

SUPPORTED_DEVICES is adjusted based on the node rename where necessary,
though it looks like for several older devices it was not set up
correctly so far.

While at it, this also changes the DTS name for u7623-02-emmc-512m
to all-lower-case.

Cc: John Crispin <john@phrozen.org>
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agokernel: use older kernel for explicitly setting dependencies
Adrian Schmutzler [Fri, 24 Jan 2020 22:40:22 +0000 (23:40 +0100)]
kernel: use older kernel for explicitly setting dependencies

It is generally more desirable to use older kernel versions for
dependencies, as this will require less changes when newer kernels
are added (they will by default select the newer packages).

Since we currently only have two kernels (4.14 and 4.19) in master,
this patch applies this logic by converting all LINUX_4_19 symbols
to their inverted LINUX_4_14 equivalents.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agohostapd: remove some bashisms
Sven Roederer [Wed, 25 Dec 2019 17:46:14 +0000 (18:46 +0100)]
hostapd: remove some bashisms

"[[" is a bash extension for test. As the ash-implementation is not
fully compatible we drop its usage.

Signed-off-by: Sven Roederer <devel-sven@geroedel.de>
[remove shebang, slightly facelift commit title/message]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years ago6in4/6in4.sh: remove some bashism (usage of [[)
Sven Roederer [Wed, 25 Dec 2019 17:43:45 +0000 (18:43 +0100)]
6in4/6in4.sh: remove some bashism (usage of [[)

"[[" is a bash extension for test. As the ash-implementation is not
fully compatible we drop its usage.

Signed-off-by: Sven Roederer <devel-sven@geroedel.de>
4 years agonetifd/config.sh: remove some bashism (usage of [[)
Sven Roederer [Wed, 25 Dec 2019 17:42:13 +0000 (18:42 +0100)]
netifd/config.sh: remove some bashism (usage of [[)

"[[" is a bash extension for test. As the ash-implementation is not
fully compatible we drop its usage.

Signed-off-by: Sven Roederer <devel-sven@geroedel.de>
4 years agomvebu: remove bashism
Sven Roederer [Wed, 25 Dec 2019 17:39:00 +0000 (18:39 +0100)]
mvebu: remove bashism

"[[" is a bash extension for test. As the ash-implementation is not
fully compatible we drop its usage.

Signed-off-by: Sven Roederer <devel-sven@geroedel.de>
[split patch, remove shebang, adjust commit title/message]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agobase-files: remove some bashisms
Sven Roederer [Wed, 25 Dec 2019 17:39:00 +0000 (18:39 +0100)]
base-files: remove some bashisms

"[[" is a bash extension for test. As the ash-implementation is
not fully compatible we drop its usage.
Also change to "=" for simple test, which is sufficient. (see d6ac8ca76c04ed)

Signed-off-by: Sven Roederer <devel-sven@geroedel.de>
[split patch, removed shebang]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agobuild: Add KBUILD_HOSTLDLIBS
Hauke Mehrtens [Fri, 17 Jan 2020 17:47:33 +0000 (18:47 +0100)]
build: Add KBUILD_HOSTLDLIBS

In Linux kernel commit 8377bd2b9ee1 ("kbuild: Rename HOST_LOADLIBES to
KBUILD_HOSTLDLIBS") HOST_LOADLIBES was renamed to KBUILD_HOSTLDLIBS.
This patch adapts the OpenWrt kernel build to this new variable. Without
this change the kernel host tools would not link against the libraries
found in the staging directory.

Signed-off-by: Hauke Mehrtens <hauke.mehrtens@intel.com>
4 years agokernel: sfp: add Nokia SFP fix from net-next
Scott Roberts [Fri, 24 Jan 2020 04:45:53 +0000 (21:45 -0700)]
kernel: sfp: add Nokia SFP fix from net-next

Add Nokia GPON ONT SFP fix for tx_fault in net-next.

Signed-off-by: Scott Roberts <ttocsr@gmail.com>
4 years agolantiq: ltq-ptm: vr9: fix skb handling in ptm_hard_start_xmit()
Martin Schiller [Tue, 21 Jan 2020 09:42:33 +0000 (10:42 +0100)]
lantiq: ltq-ptm: vr9: fix skb handling in ptm_hard_start_xmit()

Call skb_orphan(skb) to call the owner's destructor function and make
the skb unowned.

This is necessary to prevent sk_wmem_alloc of a socket from overflowing,
which leads to ENOBUFS errors on application level.

Signed-off-by: Martin Schiller <ms@dev.tdt.de>
4 years agombedtls: update to 2.16.4
Magnus Kroken [Sat, 25 Jan 2020 17:33:41 +0000 (18:33 +0100)]
mbedtls: update to 2.16.4

Fixes side channel vulnerabilities in mbed TLS' implementation of ECDSA.

Release announcement:
https://tls.mbed.org/tech-updates/releases/mbedtls-2.16.4-and-2.7.13-released

Security advisory:
https://tls.mbed.org/tech-updates/security-advisories/mbedtls-security-advisory-2019-12

Fixes:
 * CVE-2019-18222: Side channel attack on ECDSA

Signed-off-by: Magnus Kroken <mkroken@gmail.com>
4 years agoipq40xx: fix misplaced cells-sizes in WPJ419's dts
Christian Lamparter [Sun, 26 Jan 2020 00:12:50 +0000 (01:12 +0100)]
ipq40xx: fix misplaced cells-sizes in WPJ419's dts

This patch fixes the occurences of the following warning
message from the dtc:

Warning (reg_format): /soc/spi@78b5000/flash0@0/partitions/partition@0:reg:
property has invalid length (8 bytes) (#address-cells == 2, #size-cells == 1)

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
4 years agoipq-wifi: drop deprecated .bin support
Christian Lamparter [Sat, 18 Jan 2020 23:00:52 +0000 (00:00 +0100)]
ipq-wifi: drop deprecated .bin support

This patch converts the Qxwlan E2600AC image away from
the deprecated .bin file and to the new .qca4019 method.

As a result, we no longer need to carry around the
legacy support for handling .bin files.

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
4 years agoipq40xx: add support for EnGenius EMD1
Yen-Ting-Shen [Fri, 1 Nov 2019 06:00:40 +0000 (14:00 +0800)]
ipq40xx: add support for EnGenius EMD1

SOC:     IPQ4018 / QCA Dakota
CPU:     Quad-Core ARMv7 Processor rev 5 (v7l) Cortex-A7
DRAM:    256 MiB
NOR:     32 MiB
ETH:     Qualcomm Atheros QCA8072 (1 port)
WLAN1:   Qualcomm Atheros QCA4018 2.4GHz 802.11bgn 2:2x2
WLAN2:   Qualcomm Atheros QCA4018 5GHz 802.11a/n/ac 2:2x2
INPUT:   RESET Button
LEDS:    White, Blue, Red, Orange

Flash instruction:

From EnGenius firmware to OpenWrt firmware:

In Firmware Upgrade page, upgrade your openwrt-ipq40xx-generic-engenius_emd1-squashfs-factory.bin directly.

From OpenWrt firmware to EnGenius firmware:

1. Setup a TFTP server on your computer and configure static IP to 192.168.99.8
   Put the EnGenius firmware in the TFTP server directory on your computer.
2. Power up EMD1. Press 4 and then press any key to enter u-boot.
3. Download EnGenius firmware
   (IPQ40xx) # tftpboot 0x84000000 openwrt-ipq40xx-emd1-nor-fw-s.img
4. Flash the firmware
   (IPQ40xx) # imgaddr=0x84000000 && source 0x84000000:script
5. Reboot
   (IPQ40xx) # reset

Signed-off-by: Yen-Ting-Shen <frank.shen@senao.com>
[removed BOARD_NAME]
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
4 years agoipq40xx: Add support for D-Link DAP-2610
Fredrik Olofsson [Tue, 10 Sep 2019 09:25:53 +0000 (11:25 +0200)]
ipq40xx: Add support for D-Link DAP-2610

Specifications
==============
- SOC: IPQ4018
- RAM: DDR3 256MB
- Flash: SPI NOR 16MB
- WiFi:
    - 2.4GHz: IPQ4018, 2x2, front end SKY85303-11
    - 5GHz: IPQ4018, 2x2, front end SKY85717-21
- Ethernet: 1x 10/100/1000Mbps, POE 802.3af
- PHY: QCA8072
- UART: GND, blocked, 3.3V, RX, TX / 115200 8N1
- LED: 1x red / green
- Button: 1x reset / factory default
- U-Boot bootloader with tftp and "emergency web server" accessible
  using serial port.

Installation
============
Flash factory image from D-Link web UI. Constraints in the D-Link web UI
makes the factory image unnecessarily large. Flash again using
sysupgrade from inside OpenWrt to reclaim some flash space.

Return to stock D-Link firmware
===============================
Partition layout is preserved, and it is possible to return to the stock
firmware simply by downloading it from D-Link and writing it to the
firmware partition.

    # mtd -r write dap2610-firmware.bin firmware

Quirks
======
To be flashable from the D-Link http server, the firmware must be larger
then 6MB, and the size in the firmware header must match the actual file
size. Also, the boot loader verifies the checksum of the firmware before
each boot, thus the jffs2 must be after the checksum covered part. This
is solved in the factory image by having the rootfs at the very end of
the image (without pad-rootfs).

The sysupgrade image which does not have to be flashable from the D-Link
web UI may be smaller, and the checksum in the firmware header only
covers the kernel part of the image.

Signed-off-by: Fredrik Olofsson <fredrik.olofsson@anyfinetworks.com>
[added WRGG Variables to DEVICE_VARS, squashed spi pinconf/mux,
added emd1's gmac0 config,fix dtc warnings]
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
4 years agotools/gmp: update to 6.2.0
Hannu Nyman [Sat, 25 Jan 2020 09:24:38 +0000 (11:24 +0200)]
tools/gmp: update to 6.2.0

Update GNU gmp to version 6.2.0

Release notes:  https://gmplib.org/gmp6.2.html

Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
4 years agolantiq: tidy up image/Makefile
Adrian Schmutzler [Fri, 24 Jan 2020 17:51:36 +0000 (18:51 +0100)]
lantiq: tidy up image/Makefile

This harmonizes indent for Build blocks and removes multiple empty
lines.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agolantiq: fix setting SOC to DEFAULT_SOC
Adrian Schmutzler [Fri, 24 Jan 2020 17:48:47 +0000 (18:48 +0100)]
lantiq: fix setting SOC to DEFAULT_SOC

This adds the missing assignment of DEFAULT_SOC to the SOC variable
by default.

Fixes: 09ee51c614d9 ("lantiq: define SOC only once for uniform targets")
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agolantiq: define SOC only once for uniform targets
Adrian Schmutzler [Fri, 24 Jan 2020 17:38:53 +0000 (18:38 +0100)]
lantiq: define SOC only once for uniform targets

In lantiq there are several subtarget where all devices have the
same value set to the SOC variable for each device individually.

This patch introduces a non-device-dependent variable DEFAULT_SOC,
which is used if no specific SOC is set for a device, and thus reduces
the number of redundant definitions drastically.

This is applied to all subtargets except xway, as only the latter has
two different SOCs.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agolantiq: move DTS_DIR variable out of Device definition
Adrian Schmutzler [Fri, 24 Jan 2020 17:35:25 +0000 (18:35 +0100)]
lantiq: move DTS_DIR variable out of Device definition

The DTS_DIR variable is not a device variable, thus it should not
be set inside Device/Default but globally.

Fixes: c640370939d7 ("lantiq: use soc_vendor_device scheme on DTS file")
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agolantiq: fix model name for BT Home Hub 3 Type A
Adrian Schmutzler [Fri, 24 Jan 2020 17:32:11 +0000 (18:32 +0100)]
lantiq: fix model name for BT Home Hub 3 Type A

The number 3 was accidentally removed from the name during split
of DEVICE_TITLE.

Fixes: fd666870582f ("lantiq: split up DEVICE_TITLE")
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agoramips: define SOC only once for uniform targets
Adrian Schmutzler [Tue, 21 Jan 2020 20:54:50 +0000 (21:54 +0100)]
ramips: define SOC only once for uniform targets

In ramips, all devices in mt7621, mt76x8 and rt288x subtarget have
the same value set to the SOC variable for each device individually.

This patch introduces a non-device-dependent variable DEFAULT_SOC,
which is used if no specific SOC is set for a device, and thus reduces
the number of redundant definitions drastically.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agokernel: act_ctinfo: really fix backport this time
Kevin Darbyshire-Bryant [Fri, 24 Jan 2020 14:18:26 +0000 (14:18 +0000)]
kernel: act_ctinfo: really fix backport this time

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
4 years agokernel: act_ctinfo: fixup build error on 4.14
Kevin Darbyshire-Bryant [Fri, 24 Jan 2020 13:37:58 +0000 (13:37 +0000)]
kernel: act_ctinfo: fixup build error on 4.14

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
4 years agokernel: bump 4.19 to 4.19.98
Koen Vandeputte [Fri, 24 Jan 2020 10:20:03 +0000 (11:20 +0100)]
kernel: bump 4.19 to 4.19.98

Refreshed all patches.

Compile-tested on: cns3xxx
Runtime-tested on: cns3xxx

Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
4 years agokernel: bump 4.14 to 4.14.167
Koen Vandeputte [Fri, 24 Jan 2020 10:13:35 +0000 (11:13 +0100)]
kernel: bump 4.14 to 4.14.167

Refreshed all patches.

Compile-tested on: cns3xxx
Runtime-tested on: cns3xxx

Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
4 years agokernel: act_ctinfo: backport memory leak fix
Kevin Darbyshire-Bryant [Fri, 24 Jan 2020 11:21:47 +0000 (11:21 +0000)]
kernel: act_ctinfo: backport memory leak fix

[ Upstream commit 09d4f10a5e78d76a53e3e584f1e6a701b6d24108 ]

Implement a cleanup method to properly free ci->params

BUG: memory leak
unreferenced object 0xffff88811746e2c0 (size 64):
 comm "syz-executor617", pid 7106, jiffies 4294943055 (age 14.250s)
 hex dump (first 32 bytes):
   00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
   c0 34 60 84 ff ff ff ff 00 00 00 00 00 00 00 00  .4`.............
 backtrace:
   [<0000000015aa236f>] kmemleak_alloc_recursive include/linux/kmemleak.h:43 [inline]
   [<0000000015aa236f>] slab_post_alloc_hook mm/slab.h:586 [inline]
   [<0000000015aa236f>] slab_alloc mm/slab.c:3320 [inline]
   [<0000000015aa236f>] kmem_cache_alloc_trace+0x145/0x2c0 mm/slab.c:3549
   [<000000002c946bd1>] kmalloc include/linux/slab.h:556 [inline]
   [<000000002c946bd1>] kzalloc include/linux/slab.h:670 [inline]
   [<000000002c946bd1>] tcf_ctinfo_init+0x21a/0x530 net/sched/act_ctinfo.c:236
   [<0000000086952cca>] tcf_action_init_1+0x400/0x5b0 net/sched/act_api.c:944
   [<000000005ab29bf8>] tcf_action_init+0x135/0x1c0 net/sched/act_api.c:1000
   [<00000000392f56f9>] tcf_action_add+0x9a/0x200 net/sched/act_api.c:1410
   [<0000000088f3c5dd>] tc_ctl_action+0x14d/0x1bb net/sched/act_api.c:1465
   [<000000006b39d986>] rtnetlink_rcv_msg+0x178/0x4b0 net/core/rtnetlink.c:5424
   [<00000000fd6ecace>] netlink_rcv_skb+0x61/0x170 net/netlink/af_netlink.c:2477
   [<0000000047493d02>] rtnetlink_rcv+0x1d/0x30 net/core/rtnetlink.c:5442
   [<00000000bdcf8286>] netlink_unicast_kernel net/netlink/af_netlink.c:1302 [inline]
   [<00000000bdcf8286>] netlink_unicast+0x223/0x310 net/netlink/af_netlink.c:1328
   [<00000000fc5b92d9>] netlink_sendmsg+0x2c0/0x570 net/netlink/af_netlink.c:1917
   [<00000000da84d076>] sock_sendmsg_nosec net/socket.c:639 [inline]
   [<00000000da84d076>] sock_sendmsg+0x54/0x70 net/socket.c:659
   [<0000000042fb2eee>] ____sys_sendmsg+0x2d0/0x300 net/socket.c:2330
   [<000000008f23f67e>] ___sys_sendmsg+0x8a/0xd0 net/socket.c:2384
   [<00000000d838e4f6>] __sys_sendmsg+0x80/0xf0 net/socket.c:2417
   [<00000000289a9cb1>] __do_sys_sendmsg net/socket.c:2426 [inline]
   [<00000000289a9cb1>] __se_sys_sendmsg net/socket.c:2424 [inline]
   [<00000000289a9cb1>] __x64_sys_sendmsg+0x23/0x30 net/socket.c:2424

Fixes: 24ec483cec98 ("net: sched: Introduce act_ctinfo action")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Cc: Kevin 'ldir' Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
Cc: Cong Wang <xiyou.wangcong@gmail.com>
Cc: Toke Høiland-Jørgensen <toke@redhat.com>
Acked-by: Kevin 'ldir' Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
4 years agowireguard-tools: bump to 1.0.20200121
Jason A. Donenfeld [Tue, 21 Jan 2020 15:42:58 +0000 (16:42 +0100)]
wireguard-tools: bump to 1.0.20200121

* Makefile: remove pwd from compile output
* Makefile: add standard 'all' target
* Makefile: evaluate git version lazily

Quality of life improvements for packagers.

* ipc: simplify inflatable buffer and add fuzzer
* fuzz: add generic command argument fuzzer
* fuzz: add set and setconf fuzzers

More fuzzers and a slicker string list implementation. These fuzzers now find
themselves configuring wireguard interfaces from scratch after several million
mutations, which is fun to watch.

* netlink: make sure to clear return value when trying again

Prior, if a dump was interrupted by a concurrent set operation, we'd try
again, but forget to reset an error flag, so we'd keep trying again forever.
Now we do the right thing and succeed when we succeed.

* Makefile: sort inputs to linker so that build is reproducible

Earlier versions of make(1) passed GLOB_NOSORT to glob(3), resulting in the
linker receiving its inputs in a filesystem-dependent order. This screwed up
reproducible builds.

Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
4 years agowireguard: bump to 0.0.20200121
Jason A. Donenfeld [Tue, 21 Jan 2020 15:42:57 +0000 (16:42 +0100)]
wireguard: bump to 0.0.20200121

* Makefile: strip prefixed v from version.h

This fixes a mistake in dmesg output and when parsing the sysfs entry in the
filesystem.

* device: skb_list_walk_safe moved upstream

This is a 5.6 change, which we won't support here, but it does make the code
cleaner, so we make this change to keep things in sync.

* curve25519: x86_64: replace with formally verified implementation

This comes from INRIA's HACL*/Vale. It implements the same algorithm and
implementation strategy as the code it replaces, only this code has been
formally verified, sans the base point multiplication, which uses code
similar to prior, only it uses the formally verified field arithmetic
alongside reproducable ladder generation steps. This doesn't have a
pure-bmi2 version, which means haswell no longer benefits, but the
increased (doubled) code complexity is not worth it for a single
generation of chips that's already old.

Performance-wise, this is around 1% slower on older microarchitectures,
and slightly faster on newer microarchitectures, mainly 10nm ones or
backports of 10nm to 14nm. This implementation is "everest" below:

Xeon E5-2680 v4 (Broadwell)

armfazh: 133340 cycles per call
everest: 133436 cycles per call

Xeon Gold 5120 (Sky Lake Server)

armfazh: 112636 cycles per call
everest: 113906 cycles per call

Core i5-6300U (Sky Lake Client)

armfazh: 116810 cycles per call
everest: 117916 cycles per call

Core i7-7600U (Kaby Lake)

armfazh: 119523 cycles per call
everest: 119040 cycles per call

Core i7-8750H (Coffee Lake)

armfazh: 113914 cycles per call
everest: 113650 cycles per call

Core i9-9880H (Coffee Lake Refresh)

armfazh: 112616 cycles per call
everest: 114082 cycles per call

Core i3-8121U (Cannon Lake)

armfazh: 113202 cycles per call
everest: 111382 cycles per call

Core i7-8265U (Whiskey Lake)

armfazh: 127307 cycles per call
everest: 127697 cycles per call

Core i7-8550U (Kaby Lake Refresh)

armfazh: 127522 cycles per call
everest: 127083 cycles per call

Xeon Platinum 8275CL (Cascade Lake)

armfazh: 114380 cycles per call
everest: 114656 cycles per call

Achieving these kind of results with formally verified code is quite
remarkable, especialy considering that performance is favorable for
newer chips.

Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
4 years agoath10k-firmware: fix mirror hash
DENG Qingfang [Mon, 20 Jan 2020 17:06:22 +0000 (01:06 +0800)]
ath10k-firmware: fix mirror hash

Fix PKG_MIRROR_HASH hash mismatch.

Fixes: 641a93f0f226 ("ath10k-firmware: update wave 1 firmware to 10.2.4-1.0-00047")
Signed-off-by: DENG Qingfang <dengqf6@mail2.sysu.edu.cn>
[added missing commit description]
Signed-off-by: Petr Štetiar <ynezz@true.cz>
4 years agoprocd: update to version 2020-01-24
Petr Štetiar [Fri, 24 Jan 2020 05:41:56 +0000 (06:41 +0100)]
procd: update to version 2020-01-24

00aafc4f439e procd: show process's exit code
856b5f8be046 state: fix reboot causing shutdown inside LXC container
b44417c20c7f instance: provide error feedback if ujail binary is missing

Signed-off-by: Petr Štetiar <ynezz@true.cz>
4 years agokernel: fix dst reference leak in flow offload
Felix Fietkau [Thu, 23 Jan 2020 16:40:06 +0000 (17:40 +0100)]
kernel: fix dst reference leak in flow offload

Fixes a significant amount of leaked memory with lots of connections

Signed-off-by: Felix Fietkau <nbd@nbd.name>
4 years agoath79: fix SUPPORTED_DEVICES not matching ar71xx board names
Adrian Schmutzler [Thu, 23 Jan 2020 14:42:48 +0000 (15:42 +0100)]
ath79: fix SUPPORTED_DEVICES not matching ar71xx board names

Based on a script for comparison, this fixes (hopefully) all errors
in SUPPORTED_DEVICES for ar71xx->ath79 upgrade.

Devices where old string is removed as the device does not exist
in ar71xx:
- dlink_dir-859-a1
- tplink_archer-a7-v5
- tplink_cpe510-v3

Devices where string is changed because it did not match the board
name in ar71xx:
- tplink_tl-mr3220-v1
- tplink_tl-mr3420-v1
- tplink_tl-wr2543-v1
- tplink_tl-wr741nd-v4
- tplink_tl-wr841-v7
- ubnt_unifiac-mesh
- ubnt_unifiac-mesh-pro
- ubnt_unifiac-pro

For this device, the correct string could not be found, but we could
not determine the correct one. Thus, the string is removed for now:
- tplink_tl-wr740n-v4

The script for checking this is quite simple (note that newer
entries, i.e. ath79->ath79 upgrade, are displayed as missing):

  newpath=target/linux/ath79/image/
  oldpath=target/linux/ar71xx/base-files/lib/ar71xx.sh

  for s in $(grep -roh "SUPPORTED_DEVICES.*" $newpath | sed 's/SUPPORTED_DEVICES *.= *//'); do
    found="Missing"
    grep -q -r "\"$s\"" $oldpath && found="Found"
    echo "$s: $found."
  done

The errors might be filtered by appending 'grep "Missing"' to the script.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agoath79: move lzma-loader to the end of available RAM
Adrian Schmutzler [Wed, 22 Jan 2020 17:06:07 +0000 (18:06 +0100)]
ath79: move lzma-loader to the end of available RAM

In certain cases, the uncompressed initramfs image will overwrite
the lzma-loader, which is currently only 10 MB away from kernel image
start. To prevent this, change LZMA_TEXT_START to 24 MB, so loader
and compressed image have 8 MB at the end of RAM and uncompressed
image has 24 MB available.

This is only enabled for ath79 at the moment, as there we can be sure
that all devices have 32+ MB RAM and TARGET_INITRAMFS_COMPRESSION_LZMA
is not enabled there.

Despite, since lzma-loader is currently build specifically for ath79
anyway, there is no need to re-specify LOADADDR and LZMA_TEXT_START
in image/Makefile, so the values are set directly in
image/lzma-loader/Makefile and the overwrite in image/Makefile is
removed.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Tested-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
4 years agoath79: add support for MikroTik RouterBOARD wAP G-5HacT2HnD
Roger Pueyo Centelles [Thu, 28 Nov 2019 11:53:01 +0000 (12:53 +0100)]
ath79: add support for MikroTik RouterBOARD wAP G-5HacT2HnD

This commit adds support for the MikroTik RouterBOARD wAP G-5HacT2HnD
(wAP AC), a small weatherproof dual band, dual-radio 802.11ac
wireless AP with integrated omnidirectional anntennae and one
10/100/1000 Mbps Ethernet port.

See https://mikrotik.com/product/RBwAPG-5HacT2HnD for more info.

Specifications:
 - SoC: Qualcomm Atheros QCA9556
 - RAM: 64 MB
 - Storage: 16 MB NOR
 - Wireless:
   · Atheros AR9550 (SoC) 802.11b/g/n 2x2:2, 2 dBi antennae
   · Qualcomm QCA9880 802.11a/n/ac 3x3:3, 2 dBi antennae
 - Ethernet: Atheros AG71xx (SoC, AR8033), 1x 1000/100/10 port,
   passive PoE in

Working:
 - Board/system detection
 - Sysupgrade
 - Serial console
 - Ethernet
 - 2.4 GHz radio
 - 5 GHz radio and LED
 - Reset button

Not working/Unsupported:
 - 2.4 GHz LED
 - AP/CAP LED
 - ZT2046Q SPI temperature and voltage sensor

This adds the basic features for supporting MikroTik devices:
 - a common recipe for mikrotik images in common-mikrotik.mk
 - support for minor (MikroTik NOR) split firmware (only for
   generic subtarget so far)

Acknowledgments: Robert Marko <robimarko@gmail.com>
                 Andrew Cameron <apcameron@softhome.net>
                 Koen Vandeputte <koen.vandeputte@ncentric.com>
                 Chuanhong Guo <gch981213@gmail.com>

Signed-off-by: Roger Pueyo Centelles <roger.pueyo@guifi.net>
Co-developed-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Tested-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
4 years agoath79: allow to override AR8033 SGMII aneg status
David Bauer [Thu, 11 Apr 2019 15:59:44 +0000 (17:59 +0200)]
ath79: allow to override AR8033 SGMII aneg status

In order to make the QCA955x SGMII workaround work, the unsuccessful
SGMII autonegotiation on the AR8033 should not block the PHY
state-machine.

Otherwise, the ag71xx driver never becomes aware of the copper-side
link-establishment and the workaround is never executed.

Signed-off-by: David Bauer <mail@david-bauer.net>
[remove one trailing whitespace per file]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agoath79: add QCA955x SGMII link loss workaround
David Bauer [Thu, 11 Apr 2019 15:59:43 +0000 (17:59 +0200)]
ath79: add QCA955x SGMII link loss workaround

This commit adds a workaround for the loss of the SGMII link observed on
the QCA955x generation of SoCs. The workaround originates part from the
U-Boot source code, part from the implementation from AVM found in the
GPL tarball for the AVM FRITZ!WLAN Repeater 450E.

The bug results in a stuck SGMII link between the PHY device and the SoC
side. This has only been observed with the Atheros AR8033 PHY and most
likely all devices using such combination are affected.

It is worked around by reading a hidden SGMII status register and
issuing a SGMII PHY reset until the link becomes useable again.

Signed-off-by: David Bauer <mail@david-bauer.net>
4 years agorbextract: support devices directly showing ERD magic
Roger Pueyo Centelles [Thu, 28 Nov 2019 12:29:12 +0000 (13:29 +0100)]
rbextract: support devices directly showing ERD magic

Older ath79-based MikroTik devices have the ERD calibration data
compressed and stored different to newer IPQ40xx ones. This commit
adds support for these former ones.

Signed-off-by: Roger Pueyo Centelles <roger.pueyo@guifi.net>
Acked-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
4 years agorbextract: add package
Roger Pueyo Centelles [Thu, 28 Nov 2019 12:23:34 +0000 (13:23 +0100)]
rbextract: add package

This utility extracts the radio calibration data, as well as other
board-related information (model, serial number, etc.), from MikroTik
Routerboard devices' flash.

Signed-off-by: Roger Pueyo Centelles <roger.pueyo@guifi.net>
Acked-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
4 years agorbcfg: make package available for ath79
Roger Pueyo Centelles [Thu, 28 Nov 2019 12:11:59 +0000 (13:11 +0100)]
rbcfg: make package available for ath79

Signed-off-by: Roger Pueyo Centelles <roger.pueyo@guifi.net>
4 years agohostapd: fix faulty WMM IE parameters with ETSI regulatory domains
Felix Fietkau [Thu, 23 Jan 2020 13:51:58 +0000 (14:51 +0100)]
hostapd: fix faulty WMM IE parameters with ETSI regulatory domains

hostapd sets minimum values for CWmin/CWmax/AIFS and maximum for TXOP.
The code for applying those values had a few bugs leading to bogus values,
which caused significant latency and packet loss.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
4 years agokirkwood: exploit BOARD_NAME to set DEVICE_DTS and SUPPORTED_DEVICES
Adrian Schmutzler [Sat, 18 Jan 2020 01:21:15 +0000 (02:21 +0100)]
kirkwood: exploit BOARD_NAME to set DEVICE_DTS and SUPPORTED_DEVICES

Due to the history of the target, all devices added before a certain
point have the same device string in BOARD_NAME, DEVICE_DTS and added
to SUPPORTED_DEVICES.

Thus, we can set this one automatically for all devices where
BOARD_NAME is specified, removing the explicit DEVICE_DTS and
SUPPORTED_DEVICES addition there.

For new devices, nothing has changed, and just DEVICE_DTS has to
be set manually.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agosunxi: tidy up and sort alphabetically in image Makefiles
Adrian Schmutzler [Thu, 23 Jan 2020 13:21:19 +0000 (14:21 +0100)]
sunxi: tidy up and sort alphabetically in image Makefiles

This tidies up the image Makefiles for the sunxi target by:
- Move the if-condition for the subtarget to the parent Makefile
- Remove lots of unnecessary empty lines
- Sort device definitions alphabetically
- Harmonize line wrapping for DEVICE_PACKAGES

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agouboot-envtools: ath79: add support for glinet,gl-ar150
Kimmo Vuorinen [Wed, 22 Jan 2020 21:36:26 +0000 (23:36 +0200)]
uboot-envtools: ath79: add support for glinet,gl-ar150

Add ubootenv uci config for GL.inet GL-AR150

Signed-off-by: Kimmo Vuorinen <kimmo.vuorinen@gmail.com>
[commit title/message facelift]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agouboot-envtools: ar71xx: add support for gl-ar150/-domino/-mifi
Kimmo Vuorinen [Wed, 22 Jan 2020 21:34:58 +0000 (23:34 +0200)]
uboot-envtools: ar71xx: add support for gl-ar150/-domino/-mifi

Add ubootenv uci config for gl-ar150, gl-domino and gl-mifi

Signed-off-by: Kimmo Vuorinen <kimmo.vuorinen@gmail.com>
[commit message/title facelift]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agoar71xx: change u-boot-env to read-write for gl-ar150/-domino/mifi
Kimmo Vuorinen [Wed, 22 Jan 2020 21:38:35 +0000 (23:38 +0200)]
ar71xx: change u-boot-env to read-write for gl-ar150/-domino/mifi

Change u-boot-env partitions to be mounted as read-write for gl-ar150,
gl-domino and gl-mifi so uboot-envtools support is possible.

Signed-off-by: Kimmo Vuorinen <kimmo.vuorinen@gmail.com>
[commit title/message facelift]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agoath79: add support for TP-Link TL-WA850RE v2
Adrian Schmutzler [Thu, 16 Jan 2020 13:41:44 +0000 (14:41 +0100)]
ath79: add support for TP-Link TL-WA850RE v2

TP-Link TL-WA850RE v2 is a wall-plug N300 Wi-Fi range extender,
based on Qualcomm/Atheros QCA9533 v2.

Short specification:

- 550/391/195 MHz (CPU/DDR/AHB)
- 1x 10/100 Mbps Ethernet
- 32 MB of RAM (DDR1)
- 4 MB of FLASH
- 2T2R 2.4 GHz
- 2x internal antennas (embedded on PCB)
- 9x LED (all can be turned off with GPIO15), 2x button
- UART (J3) header on PCB

Flash instruction: use "factory" image directly in vendor GUI.

Warning: this device does not include any kind of recovery mechanism
in the bootloader and disassembling process is not trivial.

You can access vendor firmware over serial line using:
- login: root
- password: sohoadmin

Stock firmware uses label MAC address for WiFi and same with local
bit set for ethernet. Since this is difficult to reproduce with
the toolset of OpenWrt, we just keep both ethernet and WiFi to
the same address here.

This is the first tiny device with tplink-safeloader in ath79.
Firmware partition is only 3648k and thus even smaller than for
the tplink-4m(lzma) devices.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agoar71xx: use dynamic partitioning for TP-Link TL-WA850RE v2
Adrian Schmutzler [Thu, 16 Jan 2020 14:37:22 +0000 (15:37 +0100)]
ar71xx: use dynamic partitioning for TP-Link TL-WA850RE v2

This moves the TP-Link TL-WA850RE v2 to dynamic partitioning and
will allow to use this for ath79 as well.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agogeneric at803x: remove unneeded patches
David Bauer [Tue, 21 Jan 2020 20:11:20 +0000 (21:11 +0100)]
generic at803x: remove unneeded patches

- Remove the "RGMII TX delay fixup" hack and the associated
   DT-property. It was never used in a DT-based platform and
   solved a problem which can be mitigated by using correct
   delays on the MAC side.

 - Remove the patch to enable platform-data support for the
   at803x driver. It was only used by ar71xx which does not
   (and never will) support kernel 4.19 or later.

 - Remove the SmartEEE DT-configuration patch. As explained
   previously, this patch never disabled the Atheros SmartEEE
   implementation, but rather "standard" EEE. This can be done
   on device-tree compatible platforms by adding the
   "eee-broken-1000t" or "eee-broken-100tx" properties to the PHY
   node. As all usages of the old properties are migrated, this
   patch can be removed.

Signed-off-by: David Bauer <mail@david-bauer.net>
Acked-by: Christian Lamparter <chunkeey@gmail.com>
4 years agoath79: use upstream properties to disable EEE
David Bauer [Tue, 21 Jan 2020 20:11:19 +0000 (21:11 +0100)]
ath79: use upstream properties to disable EEE

The hack-patch which introduced the "at803x-disable-smarteee" for
disabling SmartEEE did in fact not disable SmartEEE but rather disabled
802.3az "standard" EEE.

This can be done by using the upstream properties "eee-broken-100tx" and
"eee-broken-1000t". EEE is then disabled by the PHY subsystem.

Tested on devolo WiFi pro 1200e.

Signed-off-by: David Bauer <mail@david-bauer.net>
4 years agoath79: add support for TP-Link TL-WA850RE v1
Adrian Schmutzler [Mon, 9 Dec 2019 13:51:07 +0000 (14:51 +0100)]
ath79: add support for TP-Link TL-WA850RE v1

This ports support for the TL-WA850RE v1 range extender from ar71xx
to ath79.

Specifications:
  Board: AP123 / AR9341 rev. 3
  Flash/RAM: 4/32 MiB
  CPU: 535 MHz
  WiFi: 2.4 GHz b/g/n
  Ethernet: 1 port (100M)

Flashing instructions:
Upload the factory image via the vendor firmware upgrade option.

Recovery:
Note that this device does not provide TFTP via ethernet like many
other TP-Link devices do. You will have to open the case if you
require recovery beyond failsafe.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agomac80211: add support for wds_bridge hostapd feature
Daniel Golle [Wed, 22 Jan 2020 12:14:14 +0000 (14:14 +0200)]
mac80211: add support for wds_bridge hostapd feature

hostapd allows putting WDS (4addr mode) clients into a separate bridge
other than the bridge regular (3addr mode) clients end up in. This is
useful for example giving WDS clients access to several VLANs
(trunking) while regular clients will end up inside a specific VLAN.

Add 'wds_bridge' config parameter for wifi-iface which contains the
name of the bridge. hostapd-mini already supports this feature, so all
needed is to add the UCI wrapping in mac80211.sh.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
4 years agokernel: sfp: re-attempt probing for phy
Jonas Gorski [Wed, 4 Dec 2019 22:11:34 +0000 (23:11 +0100)]
kernel: sfp: re-attempt probing for phy

Add patches retrying to probe the PHY to restore support for PHYs taking
longer to initialize without breaking modules without PHYs.

Patches taken from http://git.armlinux.org.uk/cgit/linux-arm.git/log/?h=phy

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
4 years agokernel: sfp: add two fixes submitted to upstream
Jonas Gorski [Wed, 4 Dec 2019 22:09:41 +0000 (23:09 +0100)]
kernel: sfp: add two fixes submitted to upstream

Add two small fixes for SFP that were submitted upstream.

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
4 years agomvebu: add uDPU update patches
Russell King [Wed, 27 Nov 2019 11:45:35 +0000 (11:45 +0000)]
mvebu: add uDPU update patches

Update the uDPU kernel support, fixing a number of issues:
* make ethernet work again by adding comphy definitions
* slow the I2C bus to give it more chance of actually working
* the SFP cages are designed to support up to 3W modules, which
  would be prevented from initialising without this patch.

Signed-off-by: Russell King <linux@armlinux.org.uk>
[jonas.gorski: rename to mvebu, refresh patches]
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
4 years agokernel: add SFP support for Methode DM7052 NBASE-T module
Russell King [Wed, 27 Nov 2019 11:45:30 +0000 (11:45 +0000)]
kernel: add SFP support for Methode DM7052 NBASE-T module

Add support for Methode DM7052 NBASE-T module to OpenWRT. These
patches are taken from my "phy" branch, and will be sent for the
next kernel merge window.

Signed-off-by: Russell King <linux@armlinux.org.uk>
[jonas.gorski: move patches to pending, refresh patches]
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
4 years agokernel: add backported phy/phylink/sfp patches
Russell King [Wed, 27 Nov 2019 11:45:25 +0000 (11:45 +0000)]
kernel: add backported phy/phylink/sfp patches

Backport the phy/phylink/sfp patches currently queued in netdev or in
mainline necessary to support GPON popular modules, specifically to
support Huawei and Nokia GPON modules.

Signed-off-by: Russell King <linux@armlinux.org.uk>
[jonas.gorski: include kernel version in file names, refresh patches]
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
4 years agokernel: move phylink patches from mvebu to generic
Russell King [Wed, 27 Nov 2019 11:45:20 +0000 (11:45 +0000)]
kernel: move phylink patches from mvebu to generic

Move two phylink patches from mvebu to generic, so that everyone can
benefit from them.

Signed-off-by: Russell King <linux@armlinux.org.uk>
[jonas.gorski: add kernel version to file names]
Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>
4 years agokernel: remove obsolete phylink/SFP patches
Russell King [Wed, 27 Nov 2019 11:45:15 +0000 (11:45 +0000)]
kernel: remove obsolete phylink/SFP patches

Remove the old phylink/SFP patches from the OpenWRT build; these will
be updated with a new set in subsequent.

450-reprobe_sfp_phy is also removed for several reasons:
1) it is not in mainline.
2) it breaks copper modules that do not have a PHY.
3) it makes backporting the current patch set harder.

Discussion is ongoing with the patch author for a mainline Linux kernel
patch for this.

Signed-off-by: Russell King <linux@armlinux.org.uk>
4 years agocurl: update to version 7.68.0 (security fix)
Jan Pavlinec [Fri, 10 Jan 2020 15:42:33 +0000 (16:42 +0100)]
curl: update to version 7.68.0 (security fix)

Fixes
CVE-2019-15601

Signed-off-by: Jan Pavlinec <jan.pavlinec@nic.cz>
4 years agotools/bison: update to 3.5
Hannu Nyman [Sat, 18 Jan 2020 18:56:33 +0000 (20:56 +0200)]
tools/bison: update to 3.5

Update bison to 3.5

Release notes:
https://lists.gnu.org/archive/html/info-gnu/2019-12/msg00002.html

Note for future: release notes mention that YYPRINT macro is
declared deprecated, but apparently still works for now. I found
one possible use of that in scripts/config/zconf.tab.c_shipped
That might be modernized at some point, but as the file is synced
with the one from upstream Linux, it might get fixed there.

Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
4 years agoath79: remove usused TPLINK_BOARD_NAME variable for DEVICE_VARS
Adrian Schmutzler [Tue, 21 Jan 2020 17:00:31 +0000 (18:00 +0100)]
ath79: remove usused TPLINK_BOARD_NAME variable for DEVICE_VARS

TPLINK_BOARD_NAME has been renamed to TPLINK_BOARD_ID a long time
ago (7d6c63d87542: "build: rename TPLINK_BOARD_NAME to
TPLINK_BOARD_ID" for ar71xx), and before introducing ath79 target
at all.

TPLINK_BOARD_NAME seems to have been introduced into ath79 target
only by mistake. It has never been used. Remove it.

Fixes: 53c474abbdfe ("ath79: add new OF only target for QCA MIPS silicon")
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agoRevert "kernel: mtd: Make subpartitions inherit parent's access mode"
Petr Štetiar [Tue, 21 Jan 2020 16:05:19 +0000 (17:05 +0100)]
Revert "kernel: mtd: Make subpartitions inherit parent's access mode"

This reverts commit de80424f706682e8bba27c60bcd2a9c1b4a5e875 which needs
more work and testing as it broke at least jffs2 overlays at least on
ath79 platform, marking them as read-only, thus unusable:

 jffs2_build_filesystem(): erasing all blocks after the end marker...
 jffs2: Erase at 0x009e0000 failed immediately: -EROFS. Is the sector locked?

Ref: http://lists.infradead.org/pipermail/openwrt-devel/2020-January/021344.html
Reported-by: Steve Brown <sbrown@ewol.com>
Signed-off-by: Petr Štetiar <ynezz@true.cz>
4 years agofstools: update to version 2020-01-21
Petr Štetiar [Tue, 21 Jan 2020 16:14:57 +0000 (17:14 +0100)]
fstools: update to version 2020-01-21

deb745f82b93 Revert "fstools: Add support to read-only MTD partitions (eg. recovery images)"

Signed-off-by: Petr Štetiar <ynezz@true.cz>
4 years agourngd: update to version 2020-01-21
Petr Štetiar [Tue, 21 Jan 2020 16:03:21 +0000 (17:03 +0100)]
urngd: update to version 2020-01-21

c7f7b6b65b82 Tag version 1.0.2
236b7a0aef21 Fix blocked entropy generation

Signed-off-by: Petr Štetiar <ynezz@true.cz>
4 years agozynq: derive DEVICE_DTS from device definition name
Adrian Schmutzler [Sat, 18 Jan 2020 01:21:14 +0000 (02:21 +0100)]
zynq: derive DEVICE_DTS from device definition name

In zynq target, the DEVICE_DTS variable is always set consistent
with the model part of the device definition name.

This patch replaces the redundant definitions for the individual
devices with a common recipe.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agomvebu: split base-files across subtargets
Adrian Schmutzler [Sat, 18 Jan 2020 01:21:13 +0000 (02:21 +0100)]
mvebu: split base-files across subtargets

For the mvebu target in particular, there is a lot of files in
base-files that are only relevant for one subtarget. Improve
overview and reduce size per subtarget by moving/splitting
base-files depending on the subtarget they belong to.

While at it, consolidate 01_leds by using the model part of
the board name as variable.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Acked-by: Tomasz Maciej Nowak <tomek_n@o2.pl>
4 years agomvebu: use SOC to derive DEVICE_DTS
Adrian Schmutzler [Sat, 18 Jan 2020 01:21:12 +0000 (02:21 +0100)]
mvebu: use SOC to derive DEVICE_DTS

This introduces the SOC variable to mvebu target to derive some of
the DEVICE_DTS variables based on the SOC prefix and the device
definition name.

Since DTS names and compatible are inconsistent also in the kernel
for this target, the scheme cannot be applied to all devices, though.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Acked-by: Tomasz Maciej Nowak <tomek_n@o2.pl>
4 years agomvebu: move subtarget image Makefile switch to parent Makefile
Adrian Schmutzler [Sat, 18 Jan 2020 01:21:11 +0000 (02:21 +0100)]
mvebu: move subtarget image Makefile switch to parent Makefile

This moves the if conditions for choosing which image Makefiles
are used to the parent image/Makefile. It seems more convenient
to have "codeflow" in the parent while the subtarget-specific
files only contain the definitions.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Acked-by: Tomasz Maciej Nowak <tomek_n@o2.pl>
4 years agoath79: add support for TP-Link TL-MR6400
Enrico Mioso [Fri, 17 Jan 2020 12:15:52 +0000 (13:15 +0100)]
ath79: add support for TP-Link TL-MR6400

This device is an LTE router supported in ar71xx so far.
As per original commit, hardware specifications (v1.0 EU):
- SoC: QCA9531
- Flash: Winbond W25Q64FV (8MiB)
- RAM: EtronTech EM6AB160TSE-5G (64MiB)
- Wireless: SoC platform only (2.4GHz b/g/n, 2x internal antenna)
- Ethernet: 2NIC (3x100M + 1x100M)
- WWAN: TP-LINK LTE MODULE (2x external detachable antenna)
- Power: DC 12V 1A

Flashing instructions:
You can flash via tftp recovery (serve factory image as /mr6400_tp_recovery.bin
on 192.168.0.66/24, connect to any ethernet port and power on device while
holding the reset button). Flashing via OEM web interface does not work.

Known issues:
- LTE module does not always come up during boot (showing USB enumeration errors). Similar behavior has been reported at least from one user for ar71xx, too. Turning USB off and on again will serve as a workaround.
- eth0 (LAN) always shows carrier as 1 even if no cable is plugged in (this works "correctly" on ar71xx)

Signed-off-by: Enrico Mioso <mrkiko.rs@gmail.com>
[several adjustments]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Tested-by: Filip Moc <lede@moc6.cz>
4 years agoar71xx: ubnt-rocket-m-ti: fix RSSI LED definitions
Lech Perczak [Sun, 17 Nov 2019 19:57:10 +0000 (20:57 +0100)]
ar71xx: ubnt-rocket-m-ti: fix RSSI LED definitions

When mapping for RSSI LEDs was defined for interface wlan0 on
Ubiquiti Rocket M Titanium, it missed connection to actual interface.
Therefore create the mapping to interface, so RSSI LEDs work without
additional configuration, after starting rssileds service.

While at that, split RSSI into ~equal intervals for 6 LEDs,
and remove coefficients needed for PWM LEDs, as this board does not
support PWM LEDs.

Finally, for complete support, enable 'rssileds' package in per-device
rootfs, so the indicator works out of box.

Signed-off-by: Lech Perczak <lech.perczak@gmail.com>
4 years agoprocd: update to latest git HEAD
Daniel Golle [Tue, 21 Jan 2020 10:52:12 +0000 (12:52 +0200)]
procd: update to latest git HEAD

 58c12f7 jail: add basic support for network namespaces
 ba69639 jail: create resolv.conf symlink for netns jails
 81b88b1 jail: more strict mount options for /tmp/resolv.conf.d/

Add new 'netns' flag for procd_add_jail to make ujail setup a new
network namespace for the jailed service.
See previous netifd commit for example configuration for netns jailed
service.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
4 years agonetifd: add basic support for jail network namespaces
Daniel Golle [Tue, 21 Jan 2020 08:18:10 +0000 (10:18 +0200)]
netifd: add basic support for jail network namespaces

Prepare netifd for handling procd service jails having their own
network namespace.
Intefaces having the jail attribute will only be brought up inside the
jail's network namespace by procd calling the newly introduced ubus
method 'netns_updown'.
Currently proto 'static' is supported and configuration changes are
not yet being handled (ie. you'll have to restart the jailed service
for changes to take effect).

Example /etc/config/network snippet:
config device 'veth0'
    option type 'veth'
    option name 'vhost0'
    option peer_name 'virt0'

config interface 'virt'
    option type 'bridge'
    list ifname 'vhost0'
    option proto 'static'
    option ipaddr '10.0.0.1'
    option netmask '255.255.255.0'

config interface 'virt0'
    option ifname 'virt0'
    option proto 'static'
    option ipaddr '10.0.0.2'
    option netmask '255.255.255.0'
    option gateway '10.0.0.1'
    option dns '10.0.0.1'
    option jail 'transmission'

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
4 years agoath79: fix SUPPORTED_DEVICES for WNDR4300 and WNDR3700v4
Michal Cieslakiewicz [Mon, 20 Jan 2020 20:34:02 +0000 (21:34 +0100)]
ath79: fix SUPPORTED_DEVICES for WNDR4300 and WNDR3700v4

Kernel partition increase to 4 MiBs for Netgear WNDR3700v4 and WNDR4300
routers breaks sysupgrade image compatibility with ar71xx builds.
Therefore, SUPPORTED_DEVICES variable has to be removed for both devices
from target makefile.

Reported-by: DENG Qingfang <dengqf6@mail2.sysu.edu.cn>
Signed-off-by: Michal Cieslakiewicz <michal.cieslakiewicz@wp.pl>
4 years agolibubox: update to version 2020-01-20
Petr Štetiar [Mon, 20 Jan 2020 15:22:07 +0000 (16:22 +0100)]
libubox: update to version 2020-01-20

 43a103ff17ee blobmsg: blobmsg_parse and blobmsg_parse_array oob read fixes
 5c0faaf4f5e2 tests: prefer dynamically allocated buffers
 1ffa41535369 blobmsg_json: prefer snprintf usage
 132ecb563da7 blobmsg: blobmsg_vprintf: prefer vsnprintf
 a2aab30fc918 jshn: prefer snprintf usage
 b0886a37f39a cmake: add a possibility to set library version
 a36ee96618a9 blobmsg: blobmsg_add_json_element() 64-bit values
 f0da3a4283b7 blobmsg_json: fix int16 serialization
 20a070f08139 tests: blobmsg/json: add more test cases
 379cd33d1992 tests: include json script shunit2 based testing

Acked-by: Jo-Philipp Wich <jo@mein.io>
Signed-off-by: Petr Štetiar <ynezz@true.cz>
4 years agokernel: mtd: Make subpartitions inherit parent's access mode
Bruno Pena [Sat, 9 Nov 2019 14:23:58 +0000 (15:23 +0100)]
kernel: mtd: Make subpartitions inherit parent's access mode

Currently it's not possible to effectively mark a "firmware" partition
as read-only. The sub-partitions "kernel", "rootfs" and "rootfs_data"
are always created as read-write (ignoring the parent access mode).

This patch enforces the access mode of sub-partitions to match the
parent partition, which is useful for recovery images that are meant
to be fully read-only to avoid accidental damage from end-user.

An example of such implementation (read-only firmware image) is the
recovery image used on the Zsun-SD100 [1].

Please note the related patch for fstools [2] to enable this read-only
concept.

[1] https://github.com/brunompena/zsun-resources
[2] http://lists.infradead.org/pipermail/openwrt-devel/2020-January/021043.html

Signed-off-by: Bruno Pena <brunompena@gmail.com>
[removed already obsolete 4.9 kernel patch]
Signed-off-by: Petr Štetiar <ynezz@true.cz>
4 years agofstools: update to version 2020-01-18
Petr Štetiar [Sat, 18 Jan 2020 13:49:11 +0000 (14:49 +0100)]
fstools: update to version 2020-01-18

 f5c7c1813f52 fstools: Add support to read-only MTD partitions (eg. recovery images)
 189b41b6b487 libblkid-tiny: fix f2fs labels by increasing label buffer

Signed-off-by: Petr Štetiar <ynezz@true.cz>
4 years agokernel: bump 4.19 to 4.19.97
Koen Vandeputte [Mon, 20 Jan 2020 11:25:02 +0000 (12:25 +0100)]
kernel: bump 4.19 to 4.19.97

Refreshed all patches.

Compile-tested on: cns3xxx
Runtime-tested on: cns3xxx

Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
4 years agokernel: bump 4.14 to 4.14.166
Koen Vandeputte [Mon, 20 Jan 2020 11:23:38 +0000 (12:23 +0100)]
kernel: bump 4.14 to 4.14.166

Refreshed all patches.

Compile-tested on: cns3xxx
Runtime-tested on: cns3xxx

Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
4 years agokernel: bump 4.19 to 4.19.96
Koen Vandeputte [Mon, 20 Jan 2020 10:04:32 +0000 (11:04 +0100)]
kernel: bump 4.19 to 4.19.96

Refreshed all patches.

Compile-tested on: cns3xxx
Runtime-tested on: cns3xxx

Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
4 years agokernel: bump 4.14 to 4.14.165
Koen Vandeputte [Mon, 20 Jan 2020 10:00:59 +0000 (11:00 +0100)]
kernel: bump 4.14 to 4.14.165

Refreshed all patches.

Compile-tested on: cns3xxx
Runtime-tested on: cns3xxx

Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
4 years agokernel: bump 4.19 to 4.19.95
Koen Vandeputte [Tue, 14 Jan 2020 15:34:24 +0000 (16:34 +0100)]
kernel: bump 4.19 to 4.19.95

Refreshed all patches.

Compile-tested on: cns3xxx
Runtime-tested on: cns3xxx

Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
4 years agokernel: bump 4.14 to 4.14.164
Koen Vandeputte [Tue, 14 Jan 2020 13:39:40 +0000 (14:39 +0100)]
kernel: bump 4.14 to 4.14.164

Refreshed all patches.

Compile-tested on: cns3xxx
Runtime-tested on: cns3xxx

Signed-off-by: Koen Vandeputte <koen.vandeputte@ncentric.com>
4 years agobrcm47xx: fix switch port order for Netgear WN2500RP V1
Walter Sonius [Fri, 27 Dec 2019 11:41:35 +0000 (12:41 +0100)]
brcm47xx: fix switch port order for Netgear WN2500RP V1

The Netgear WN2500RP V1 switch0 already works for LAN
however the port order for the LAN ports is inverted. Correct
physical port order watched from the back of the device is:
4 / 3 / 2 / 1
WAN port is absent on this device and therefore removed
from switch config.

Signed-off-by: Walter Sonius <walterav1984@gmail.com>
[move block to maintain alphabetic sorting]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
4 years agoramips: apply LED_POLARITY rt3050-esw on MT7628AN/MT7688
Maximilian Pachl [Wed, 13 Nov 2019 11:57:28 +0000 (12:57 +0100)]
ramips: apply LED_POLARITY rt3050-esw on MT7628AN/MT7688

The device tree property "mediatek,led_polarity" is ignored for
MT7628AN and MT7688. According to the datasheet both SoCs have
the matching register. Therefore the property should be applied
on these two devices as well.

Signed-off-by: Maximilian Pachl <m@ximilian.info>
Reviewed-by: Sungbo Eo <mans0n@gorani.run>
Tested-by: Sungbo Eo <mans0n@gorani.run>