openwrt/openwrt.git
2 years agoat91: add support for sama5d2 icp board
Claudiu Beznea [Mon, 11 Oct 2021 08:47:23 +0000 (11:47 +0300)]
at91: add support for sama5d2 icp board

Add support for SAMA5D2 ICP board.

Hardware:
- SoC: SAMA5D27
- RAM: 512 MB DDR3L
- MMC: One stanard SD card interface
- USB: One USB host switch 4 ports with power switch,
       One USB device type Micro-AB
- CAN: 2 interfaces
- Ethernet: One Gigabit Ethernet PHY through HSIC,
    One ETH switchport,
    One EtherCAT interface
- WiFi/BT: Footprint for IEEE 802.11 b/g/n Wi-Fi plus
   Bluetooth module (Wi-Fi/BT), suitable for
   Microchip WILC3000-MR110CA or WILC3000-MR110UA
- Debug port: One J-Link-OB/J-Link-CDC, one JTAG interface
- Leds: one RGB LED
- Buttons: reset, wakeup, 2 user buttons
- Expansion: one PIOBU/PIO connector, 3 mikrobus sockets
- Power mangament: PMIC (MCP16502), one power consumption device
                   (PAC1934)

Not working in Linux:
- EtherCAT interface: there is no Linux support integrated
- PAC1934: driver available at [1] but not integrated in Linux

Flashing:
- follow the procedure at [2]

[1] https://ww1.microchip.com/downloads/en/DeviceDoc/pac193x_linux_driver.zip
[2] https://www.linux4sam.org/bin/view/Linux4SAM/Sama5d2IcpMainPage#Create_a_SD_card_with_the_demo

Signed-off-by: Claudiu Beznea <claudiu.beznea@microchip.com>
2 years agoat91: enable specific sam9x kernel config flags
Claudiu Beznea [Mon, 11 Oct 2021 08:47:22 +0000 (11:47 +0300)]
at91: enable specific sam9x kernel config flags

Enable specific sam9x kernel config flags.

Signed-off-by: Claudiu Beznea <claudiu.beznea@microchip.com>
2 years agoat91: kernel: bump to 5.10
Claudiu Beznea [Mon, 11 Oct 2021 08:47:21 +0000 (11:47 +0300)]
at91: kernel: bump to 5.10

Bump at91 targets to kernel v5.10. With this patches and files for
wb45n and wb50n were removed as they are now included in upstream
kernel. Along with:
- this the kernel config for sama5d2 and sam9x targets has been
  refreshed (with make kernel_menuconfig + save);
- CONFIG_ARCH_AT91 and specific sam9x SoCs (AT91RM9200, AT91SAM9,
  SAM9X60) has been enabled such that sam9x SoCs to be able to boot.

Signed-off-by: Claudiu Beznea <claudiu.beznea@microchip.com>
2 years agowolfssl: enable ECC Curve 25519 by default
Stan Grishin [Thu, 7 Oct 2021 20:15:25 +0000 (20:15 +0000)]
wolfssl: enable ECC Curve 25519 by default

* fixes https://github.com/openwrt/packages/issues/16652
 see https://github.com/openwrt/packages/issues/16674#issuecomment-934983898

Signed-off-by: Stan Grishin <stangri@melmac.net>
2 years agosunxi: 5.10: enable sun8i-thermal
Chukun Pan [Sat, 2 Oct 2021 15:05:02 +0000 (23:05 +0800)]
sunxi: 5.10: enable sun8i-thermal

Enable the sun8i-thermal driver to allow reading the
temperature of the SoC.

As suggested by mans0n, disable this driver in the
a8 subtarget because it does not support yet.

Tested on NanoPi R1S H5.

Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>
2 years agouclibc++: remove
Rosen Penev [Sat, 27 Feb 2021 04:17:46 +0000 (20:17 -0800)]
uclibc++: remove

No package here depends on it. Furthermore, uClibc++ is a fairly buggy
C++ library and seems to be relatively inactive upstream.

It also lacks proper support for modern C++11 features.

The main benefit of it is size: 66.6 KB vs 287.3 KB on mips24kc. Static
linking and LTO can help bring the size down of packages that need it.

Added warning message to uclibc++.mk

Signed-off-by: Rosen Penev <rosenp@gmail.com>
Acked-by: Hauke Mehrtens <hauke@hauke-m.de>
2 years agou-boot.mk: always link host libraries static
Andre Heider [Tue, 19 Oct 2021 09:08:46 +0000 (11:08 +0200)]
u-boot.mk: always link host libraries static

Host libraries are only build static, so let's pass --static to
pkg-config globally and remove the then unnecessary patches doing
exactly that individually.

Signed-off-by: Andre Heider <a.heider@gmail.com>
2 years agotools/pkgconf: add PKG_CONFIG_EXTRAARGS for additional arguments
Andre Heider [Tue, 19 Oct 2021 09:08:46 +0000 (11:08 +0200)]
tools/pkgconf: add PKG_CONFIG_EXTRAARGS for additional arguments

This can be used my the OpenWrt build system to pass additional
arguments when required.

Signed-off-by: Andre Heider <a.heider@gmail.com>
2 years agou-boot.mk: fix pkg-config usage
Andre Heider [Wed, 13 Oct 2021 10:51:22 +0000 (12:51 +0200)]
u-boot.mk: fix pkg-config usage

Using Host/Exports doesn't work as intended, explicitly add the
required vars so that u-boot finds the required libraries when building
its tools.

Signed-off-by: Andre Heider <a.heider@gmail.com>
2 years agokernel: 5.10: qca8k: backport qca8k_setup tidy-up
Matthew Hagan [Sat, 16 Oct 2021 19:54:25 +0000 (20:54 +0100)]
kernel: 5.10: qca8k: backport qca8k_setup tidy-up

Tidy qca8k_setup for loops relating to port handling. No functional
changes.

Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
Signed-off-by: Matthew Hagan <mnhagan88@gmail.com>
2 years agokernel: 5.10: backport qca8k cpu_port_index fix in parse_port_config
Matthew Hagan [Sat, 16 Oct 2021 19:22:45 +0000 (20:22 +0100)]
kernel: 5.10: backport qca8k cpu_port_index fix in parse_port_config

Add cpu_port_index fix to apply settings to correct CPU port.

Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
Signed-off-by: Matthew Hagan <mnhagan88@gmail.com>
2 years agokernel: 5.10: backport qca8k feature additions
Matthew Hagan [Thu, 9 Sep 2021 19:40:21 +0000 (19:40 +0000)]
kernel: 5.10: backport qca8k feature additions

Backport Ansuel Smith's various qca8k feature additions:
- mac-power-sel support
- SGMII PLL explicit enable
- tx/rx clock phase to falling edge
- power-on-sel and LED open drain mode
- cpu port 6
- qca8328 support
- sgmii internal delay
- move port config to dedicated struct
- convert to yaml schema

Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
Signed-off-by: Matthew Hagan <mnhagan88@gmail.com>
2 years agokernel: 5.10: backport QCA83x PHY resume fix, DAC amplitude preferred master, debug...
Matthew Hagan [Fri, 1 Oct 2021 20:06:27 +0000 (21:06 +0100)]
kernel: 5.10: backport QCA83x PHY resume fix, DAC amplitude preferred master, debug reg names

Backport workaround for QCA8327 PHY resume, which does not properly support
genphy_suspend/resume. Also add DAC amplitude fix for the QCA8327 PHY,
set port to preferred master and add proper names to debug regs.

Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
Signed-off-by: Matthew Hagan <mnhagan88@gmail.com>
2 years agokernel: 5.10: backport at803x QCA83xx phy support improvements
Matthew Hagan [Mon, 20 Sep 2021 12:37:53 +0000 (12:37 +0000)]
kernel: 5.10: backport at803x QCA83xx phy support improvements

This commit add accepted upstream patches that improve & tidy qca83xx support.
1 - Split qca8327 to A & B variants, identifiable by phy_id
2 - Add suspend/resume support to qca8xx phys
3 - Tidy spacing and phy naming.

Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
Signed-off-by: Matthew Hagan <mnhagan88@gmail.com>
2 years agokernel: 5.10: backport at803x internal QCA8327 PHY support
Matthew Hagan [Tue, 14 Sep 2021 08:09:40 +0000 (08:09 +0000)]
kernel: 5.10: backport at803x internal QCA8327 PHY support

Add support for qca8327 internal phy needed for correct init of the
switch port. It does use the same qca8337 function and reg just with a
different id.

Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
Signed-off-by: Matthew Hagan <mnhagan88@gmail.com>
2 years agokernel: 5.10: backport qca8k legacy mdio mapping panic fix
Matthew Hagan [Sat, 11 Sep 2021 16:41:03 +0000 (16:41 +0000)]
kernel: 5.10: backport qca8k legacy mdio mapping panic fix

Add backport of Ansuel Smith's "net: dsa: qca8k: fix kernel panic with
legacy mdio mapping" patch.

Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
Signed-off-by: Matthew Hagan <mnhagan88@gmail.com>
2 years agokernel: 5.10: backport additional qca8k fixes
Matthew Hagan [Fri, 10 Sep 2021 17:59:49 +0000 (17:59 +0000)]
kernel: 5.10: backport additional qca8k fixes

Backport fixes including:
net: dsa: qca8k: fix missing unlock on error in qca8k_vlan_(add|del)
net: dsa: qca8k: check return value of read functions correctly
net: dsa: qca8k: add missing check return value in qca8k_phylink_mac_config()
net: dsa: qca8k: fix an endian bug in qca8k_get_ethtool_stats()
net: dsa: qca8k: check the correct variable in qca8k_set_mac_eee()

Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
Signed-off-by: Matthew Hagan <mnhagan88@gmail.com>
2 years agokernel: 5.10: backport qca8k stability improvements
Matthew Hagan [Thu, 9 Sep 2021 17:12:30 +0000 (17:12 +0000)]
kernel: 5.10: backport qca8k stability improvements

This is a backport of Ansuel Smith's "Multiple improvement to qca8k stability"
series. The QCA8337 switch is available on multiple platforms including
ipq806x, ath79 and bcm53xx.

Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
Signed-off-by: Matthew Hagan <mnhagan88@gmail.com>
2 years agomac80211: Add devm_platform_get_and_ioremap_resource()
Hauke Mehrtens [Sun, 24 Oct 2021 13:03:20 +0000 (15:03 +0200)]
mac80211: Add devm_platform_get_and_ioremap_resource()

This function is missing in kernel 5.4, but it is sued by ath10k.
This fixes the build of ath10k on some targets.

Fixes: cfe0eb7485d8 ("mac80211: Update to version 5.14.13-1")
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
2 years agokernel: 5.10: silence bogus "Missing #address-cells in interrupt provider" warnings
Christian Lamparter [Sun, 24 Oct 2021 00:45:25 +0000 (02:45 +0200)]
kernel: 5.10: silence bogus "Missing #address-cells in interrupt provider" warnings

Rosen reported strange dtc warnings that had their origin in
an upstream patch to 5.8-rc1. Upon further digging this
revealed an ongoing thread [0] discussing the topic:

> [...]I don't think we need a bunch of warning fix patches to add
> these everywhere. Also, the need for #address-cells pretty much makes
> no sense on any modern system. It is a relic from days when the bus
> (address) topology and interrupt topology were related.

and later on:
> So really, we only need to be checking for #address-cells in nodes
> with interrupt-map.

This patch backports just the patch which removed the warning message
(this is from the upstream dtc project [1] - but not the kernel).
the patch does not add the checking of the #address-cells in nodes
with interrupt-map.

[0] <https://lore.kernel.org/linux-devicetree/91e3405245c89f134676449cf3822285798d2ed2.1612189652.git.michal.simek@xilinx.com/>
[1] <https://git.kernel.org/pub/scm/utils/dtc/dtc.git/commit/?id=d8d1a9a77863a8c7031ae82a1d461aa78eb72a7b>
Link: <https://github.com/openwrt/openwrt/pull/4685>
Reported-by: Rosen Penev <rosenp@gmail.com>
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
2 years agoimage: define DEVICE_DTS_DELIMITER as per-device variable
Christian Lamparter [Sat, 23 Oct 2021 19:01:24 +0000 (21:01 +0200)]
image: define DEVICE_DTS_DELIMITER as per-device variable

DEVICE_DTS_DELIMITER needs to be in the DEFAULT_DEVICE_VARS
list to work as expected. This was missing from the original
version and got overlooked.

Fixes: fd6790864739 ("scripts: mkits.sh: Allow legacy @ mode for dts creation")
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
2 years agomac80211: Update to version 5.15-rc6-1
Hauke Mehrtens [Mon, 18 Oct 2021 22:44:43 +0000 (00:44 +0200)]
mac80211: Update to version 5.15-rc6-1

The removed patches were applied upstream.
The Cisco Aironet 802.11b driver was removed from backports, remove
it also from OpenWrt.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
2 years agomac80211: Update to version 5.14.13-1
Hauke Mehrtens [Sun, 10 Oct 2021 19:24:14 +0000 (21:24 +0200)]
mac80211: Update to version 5.14.13-1

The removed patches were applied upstream.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
2 years agomac80211: Update to version 5.13.19-1
Hauke Mehrtens [Sat, 9 Oct 2021 17:25:36 +0000 (19:25 +0200)]
mac80211: Update to version 5.13.19-1

The removed patches were applied upstream.

of_get_mac_address() was backported in our OpenWrt kernel, remove the
change from backports.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
2 years agomac80211: Update to version 5.12.19-1
Hauke Mehrtens [Sat, 9 Oct 2021 14:25:30 +0000 (16:25 +0200)]
mac80211: Update to version 5.12.19-1

The removed patches were applied upstream.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
2 years agomac80211: Update to version 5.11.22-1
Hauke Mehrtens [Mon, 4 Oct 2021 21:48:20 +0000 (23:48 +0200)]
mac80211: Update to version 5.11.22-1

The removed patches were applied upstream.
This backports version 5.11.22 and later does not support kernel
versions < 4.4, this allows us to remove some patches too.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
2 years agoethtool: update to v5.14
Hans Dedecker [Sat, 23 Oct 2021 19:39:41 +0000 (21:39 +0200)]
ethtool: update to v5.14

Update to newly released version 5.14

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
2 years agoRevert "ethtool: update to v5.14"
Hans Dedecker [Sat, 23 Oct 2021 19:37:27 +0000 (21:37 +0200)]
Revert "ethtool: update to v5.14"

This reverts commit 7630001427fa266fa61da0b2533e2c1054eababe

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
2 years agoethtool: update to v5.14
Hans Dedecker [Sat, 23 Oct 2021 19:30:55 +0000 (21:30 +0200)]
ethtool: update to v5.14

Update to newly released version 5.14

Signed-off-by: Hans Dedecker <dedeckeh@gmail.com>
2 years agoumdns: add missing syscall to seccomp filter
Michael Peleshenko [Thu, 21 Oct 2021 22:14:05 +0000 (18:14 -0400)]
umdns: add missing syscall to seccomp filter

The 'clock_gettime64', syscall is missing.
Found with 'utrace /usr/sbin/umdns' on an R7800.

Signed-off-by: Michael Peleshenko <mpeleshenko@gmail.com>
2 years agowireless-regdb: update to version 2021.08.28
Christian Lamparter [Sat, 23 Oct 2021 16:08:51 +0000 (18:08 +0200)]
wireless-regdb: update to version 2021.08.28

e983a25 Update regulatory rules for Ecuador (EC)
a0bcb88 wireless-regdb: Update regulatory rules for Norway (NO) on 6 and 60 GHz
cdf854d wireless-regdb: Update regulatory rules for Germany (DE) on 6GHz
86cba52 wireless-regdb: reduce bandwidth for 5730-5850 and 5850-5895 MHz in US
6fa2384 wireless-regdb: remove PTMP-ONLY from 5850-5895 MHz for US
9839e1e wireless-regdb: recent FCC report and order allows 5850-5895 immediately
42dfaf4 wireless-regdb: update 5725-5850 MHz rule for GB

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
2 years agobase-files: chmod 1777 /var/lock
Deomid Ryabkov [Sat, 23 Oct 2021 13:22:39 +0000 (16:22 +0300)]
base-files: chmod 1777 /var/lock

Per FHS 3.0, /var/lock is the location for lock files [1].
However its current permissions (755) are too restrictive
for use by unprivileged processes.
Debian and Ubuntu set them to 1777, and now so do we.

[1] <https://refspecs.linuxfoundation.org/FHS_3.0/fhs-3.0.html#varlockLockFiles>

Signed-off-by: Deomid Ryabkov <rojer@rojer.me>
[fixed typo in commit message, had to remove "rojer" due to git hooks]
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
2 years agoath79: ag71xx: Disable napi related interrupts during probe
Sven Eckelmann [Sat, 23 Oct 2021 09:27:35 +0000 (11:27 +0200)]
ath79: ag71xx: Disable napi related interrupts during probe

ag71xx_probe is registering ag71xx_interrupt as handler for the gmac0/gmac1
interrupts. The handler is trying to use napi_schedule to handle the
processing of packets. But the netif_napi_add for this device is
called a lot later in ag71xx_probe.

It can therefore happen that a still running gmac0/gmac1 is triggering the
interrupt handler with a bit from AG71XX_INT_POLL set in
AG71XX_REG_INT_STATUS. The handler will then call napi_schedule and the
napi code will crash the system because the ag->napi is not yet
initialized:

  libphy: Fixed MDIO Bus: probed
  CPU 0 Unable to handle kernel paging request at virtual address 00000000, epc == 00000000, ra == 81373408
  Oops[#1]:
  CPU: 0 PID: 1 Comm: swapper Not tainted 5.4.152 #0
  $ 0   : 00000000 00000001 00000000 8280bf28
  $ 4   : 82a98cb0 00000000 81620000 00200140
  $ 8   : 00000000 00000000 74657272 7570743a
  $12   : 0000005b 8280bdb9 ffffffff ffffffff
  $16   : 00000001 82a98cb0 00000000 8280bf27
  $20   : 8280bf28 81620000 ffff8b00 8280bf30
  $24   : 00000000 8125af9c
  $28   : 82828000 8280bed8 81610000 81373408
  Hi    : 00005fff
  Lo    : 2e48f657
  epc   : 00000000 0x0
  ra    : 81373408 __napi_poll+0x3c/0x11c
  Status: 1100dc03 KERNEL EXL IE
  Cause : 00800008 (ExcCode 02)
  BadVA : 00000000
  PrId  : 00019750 (MIPS 74Kc)
  Modules linked in:
  Process swapper (pid: 1, threadinfo=(ptrval), task=(ptrval), tls=00000000)
  Stack : ffff8afb ffff8afa 81620000 00200140 00000000 82a98cb0 00000008 0000012c
          81625620 81373684 ffffffff ffffffff ffffffef 00000008 816153d8 81620000
          815b0d60 815bbd54 00000000 81753700 8280bf28 8280bf28 8280bf30 8280bf30
          81753748 00000008 00000003 00000004 0000000c 00000100 3fffffff 8175373c
          816059f0 814ddb48 00000001 8160ab30 81615488 810618bc 00000006 00000000
          ...
  Call Trace:

  [<81373684>] net_rx_action+0xfc/0x26c
  [<814ddb48>] __do_softirq+0x118/0x2ec
  [<810618bc>] handle_percpu_irq+0x50/0x80
  [<8125ab8c>] plat_irq_dispatch+0x94/0xc8
  [<81004e98>] handle_int+0x138/0x144

  Code: (Bad address in epc)

  ---[ end trace a60d797432b656b2 ]---

The gmcc0/gmac1 must be brought in a state in which it doesn't signal a
AG71XX_INT_POLL related status bits as interrupt before registering the
interrupt handler. ag71xx_hw_start will take care of re-initializing the
AG71XX_REG_INT_ENABLE.

Fixes: f529a3742043 ("surprise :p")
Signed-off-by: Sven Eckelmann <sven@narfation.org>
2 years agokernel: backport GPIO LED patch for MT7530
Rosen Penev [Thu, 23 Sep 2021 02:16:32 +0000 (19:16 -0700)]
kernel: backport GPIO LED patch for MT7530

This allows to specify and control switch LEDs on devices using mt7530
(typically mediatek and ramips targets).

Normally these LED GPIOs are 0, 3, 6, 9, and 12. wan/lan assignment is
per device. GPIO 9 is normally inverted. so GPIO_ACTIVE_HIGH instead of
GPIO_ACTIVE_LOW.

Tested on Linksys E7350.

Refreshed all patches.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
2 years agoipq40xx: MR33: Fix LP5562 LED driver probe
Tan Zien [Fri, 22 Oct 2021 20:15:08 +0000 (04:15 +0800)]
ipq40xx: MR33: Fix LP5562 LED driver probe

Add the reg and color property to each channel node. This update is
to accommodate the multicolor framework.

Refer to:
<https://lore.kernel.org/all/20200622185919.2131-9-dmurphy@ti.com>
<https://lore.kernel.org/all/20210818070209.1540451-1-michal.vokac@ysoft.com>

Signed-off-by: Tan Zien <nabsdh9@gmail.com>
[replaced links to lore, wrote something of a commit message]
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
2 years agomt76: update to the latest version
Felix Fietkau [Sat, 23 Oct 2021 11:31:08 +0000 (13:31 +0200)]
mt76: update to the latest version

f6bde7ba82ee mt76: connac: fix unresolved symbols when CONFIG_PM is unset

Signed-off-by: Felix Fietkau <nbd@nbd.name>
2 years agomt76: update to the latest version
Felix Fietkau [Sat, 23 Oct 2021 05:31:07 +0000 (07:31 +0200)]
mt76: update to the latest version

9d288d7c0c73 mt76: mt7615: apply cached RF data for DBDC
4d2e81d9ecab mt76: mt7915: remove mt7915_mcu_add_he()
2c8c3bcd766c mt76: mt7915: rework .set_bitrate_mask() to support more options
577d45f6e4a0 mt76: mt7915: rework debugfs fixed-rate knob
b68af355f707 mt76: mt7915: fix endiannes warning mt7915_mcu_beacon_check_caps
b52adf981561 mt76: mt7915: add WA firmware log support
ed3d8569c400 mt76: mt7915: add debugfs knobs for MCU utilization
0f49a22314a5 mt76: mt7921: disable 4addr capability
754d4a37ab71 mt76: mt7921: fix mt7921s Kconfig
f81f42fe875f mt76: Print error message when reading EEPROM from mtd failed
fd25a550965d mt76: Make use of the helper macro kthread_run()
1d01257662a6 wireless: mediatek: mt7921: fix Wformat build warning

Signed-off-by: Felix Fietkau <nbd@nbd.name>
2 years agouci: update to git HEAD
Hauke Mehrtens [Fri, 22 Oct 2021 21:51:05 +0000 (23:51 +0200)]
uci: update to git HEAD

 cmake: Allow override of install directories

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
2 years agoprocd: update to git HEAD
Hauke Mehrtens [Fri, 22 Oct 2021 21:48:24 +0000 (23:48 +0200)]
procd: update to git HEAD

 jail: Fix build with glibc

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
2 years agolayerscape: Fix build in dtb
Hauke Mehrtens [Fri, 22 Oct 2021 20:09:16 +0000 (22:09 +0200)]
layerscape: Fix build in dtb

This fixes a kernel build problem.
The removed parts of the patch are already applied upstream.

Fixes: 9ad3ef27b9ce ("kernel: bump 5.4 to 5.4.153")
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
2 years agoapm821xx: WNDAP6X0: add missed uci-default for compat
Christian Lamparter [Wed, 20 Oct 2021 19:04:43 +0000 (21:04 +0200)]
apm821xx: WNDAP6X0: add missed uci-default for compat

This should have been included in the previous patch that
resized the kernel partition to fit bigger kernels.

Fixes: 7a6a3494450c ("apm821xx: WNDAP620 + WNDAP660: reorganize partitions for 5.10")
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
2 years agoapm821xx: implement new LED label naming scheme
Christian Lamparter [Thu, 13 Aug 2020 16:26:19 +0000 (18:26 +0200)]
apm821xx: implement new LED label naming scheme

This patch updates all current APM82181 devices over to that
"new LED naming scheme". This includes many updates to the
device-tree:
- dropped the deprecated, but beloved "label" property.
- rename all DT leds node names to led-#.
- add function and color properties.
- utilized panic-indicator property.
- dropped led- aliases (see below).

migration scripts for all devices are included.

For more information. See:
<https://www.kernel.org/doc/html/latest/leds/leds-class.html>

For the future: It looks like the color+function properties
won over the dt-alias / label. This will need to be wired up
into openwrt eventually. For APM821xx the situation is that
all devices have a dedicated power and fault indicator.

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
2 years agoath79: lzma-loader: fix & re-enable per-board CONFIG_BOARD_DEV
Christian Lamparter [Sat, 31 Jul 2021 20:56:29 +0000 (22:56 +0200)]
ath79: lzma-loader: fix & re-enable per-board CONFIG_BOARD_DEV

Back in the AR71XX days, the lzma-loader code could be customized
based on the $BOARD variable. These would be passed as a
compile-time -DCONFIG_BOARD_$DEVICE_MODEL flag to the compiler.
Hence, the lzma-loader would be able to include device-specific
fixups.

Note: There's still a fixup for the TpLink TL-WR1043ND V1 found
in the lzma-loader's board.c code. But since the days of AR71XX
I couldn't find a forum post or bug reported. So, I left it
as is to not break anything by enabling it.

=> If you have a TL-WR1043ND V1 and you have problem with
the ethernet: let me know. Because otherwise, the fixup
might simply no longer needed with ath79 and it can be removed.

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
2 years agoapm821xx: MX60(W): re-enable + allow bigger future kernels
Christian Lamparter [Fri, 1 Oct 2021 19:53:26 +0000 (21:53 +0200)]
apm821xx: MX60(W): re-enable + allow bigger future kernels

The MX60's kernel is limited to 0x3EFC00 by the values in
mkmerakifw.c. Since the initramfs method of loading the
kernel seems to be working, this patch does away with the
use of the mkmerakifw tool for the MX60(W).

But this will go along with a change in u-boot as well.
So before you upgrade, please attach the serial cable and
perform:

| setenv owrt510_boot run meraki_ubi owrt_bootargs\; run owrt_load1 owrt_bootkernel\; run owrt_load2 owrt_bootkernel
| setenv bootcmd run owrt510_boot
| saveenv

Note: You won't be able to use older OpenWrt releases without
switching the bootcmd back to owrt_boot!

Note2: We are no longer compatible with older OpenWrt MX60 installs.
the legacy BOARD_NAME and SUPPORTED_DEVICES can be dropped. This is
because upgrades from older images are not possible without uboot env
changes anymore. Also the bogus BLOCKSIZE value
(which was set to 63k back then, in order to get the kernel properly
aligned after the fdt + meraki header) can be set to the NANDs real
value. The FDT size (which was needed for alignment) can now be
slimmed down as well.

Co-developed-by: Martin Kennedy <hurricos@gmail.com>
Signed-off-by: Martin Kennedy <hurricos@gmail.com>
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
2 years agoapm821xx: MBL: HDD regulators overhaul for MBL DUO
Christian Lamparter [Fri, 15 Oct 2021 19:30:58 +0000 (21:30 +0200)]
apm821xx: MBL: HDD regulators overhaul for MBL DUO

Takimata reported on the OpenWrt forum in thread [0], that his
MyBook Live Duo wasn't booting OpenWrt 21.02 after upgrading
from the previous OpenWrt 19.07.

The last logged entries on his console

|[    0.531599] sata1-regulator GPIO handle specifies active low - ignored
|[    0.538391] sata0-regulator GPIO handle specifies active low - ignored
|[    0.759791] ata2: SATA link down (SStatus 0 SControl 300)
|[    0.765251] ata1: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
|[    5.909555] ata1.00: qc timeout (cmd 0xec)
|[    5.913656] ata1.00: failed to IDENTIFY (I/O error, err_mask=0x4)
|[    6.231757] ata1: SATA link down (SStatus 0 SControl 300)

This extract clearly showed that the HDD on which OpenWrt is installed,
simply disappeared after the SATA power regulators had been initialized.

The reason why this worked with OpenWrt 19.07 was because the kernel
config symbol CONFIG_REGULATOR=y was not set in the target's config-4.14.

(This shows that the MBL Single does differ from the DUO in that
it does not have programmable power regulators for the HDDs.)

[0] <https://forum.openwrt.org/t/21-02-0-and-snapshot-fail-to-boot-on-my-book-live-duo/106585>

Reported-by: Takimata (forum)
Tested-by: Takimata (forum)
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
2 years agoath79: mikrotik: use 64 KiB SPI NOR erase sectors
Roger Pueyo Centelles [Tue, 31 Aug 2021 14:45:17 +0000 (16:45 +0200)]
ath79: mikrotik: use 64 KiB SPI NOR erase sectors

This patch removes CONFIG_MTD_SPI_NOR_USE_4K_SECTORS from the default
symbols for the ath79/mikrotik target.

MikroTik devices hold some of their user-configurable settings in the
soft_config partition, which is typically sized 4 KiB, of the SPI NOR
flash memory. Previously, in the ar71xx target, it was possible to use
64 KiB erase sectors but also smaller 4 KiB ones when needed. This is
no longer the case in ath79 with newer kernels so, to be able to write
to these 4 KiB small partitions without erasing 60 KiB around, the
CONFIG_MTD_SPI_NOR_USE_4K_SECTORS symbol was added to the defaults.
However, this ended up making sysupgrade images which were built with
64 KiB size blocks not to keep settings (e.g., the files under
/etc/config/) over the flashing process.

Using 4 KiB erase sector size on the sysupgrade images (by setting
BLOCKSIZE = 4k) allows keeping settings over a flashing process, but
renders the process terribly slow, possibly causing a user to
mistakenly force a manual device reboot while the process is still on-
going. Instead, ditching the 4 KiB erase sectors for the default
64 KiB erase size provides normal SPI write speed and sysupgrade times,
at the expense of not being able to modify the soft_config partition
(which is rarely a required thing).

An OpenWrt patch for MTD_SPI_NOR_USE_4K_SECTORS_LIMIT may once have
allowed to use different per-partition erase sector sizes. Due to
changes on recent kernels it now only works on a per-device basis.
Also, partial eraseblock write can be performed in ath79 with kernels
5.4 and lower, by copying the blocks from the 64 KiB, erasing the whole
sector and restoring those blocks not meant to be modified. A kernel
bump had that patch broken for a long time, but got fixed in bf2870c.

Note: the settings in the soft_config partition can be reset to their
defaults by holding the reset button for 5 seconds (and less than 10
seconds) at device boot.

Fixes: FS#3492 (sysupgrade […] loses settings...)
Fixes: a66eee63368e (ath79: add mikrotik subtarget)
Signed-off-by: Roger Pueyo Centelles <roger.pueyo@guifi.net>
2 years agobcm53xx: bridge all LAN ports on Linksys EA9500
Rafał Miłecki [Fri, 22 Oct 2021 11:46:22 +0000 (13:46 +0200)]
bcm53xx: bridge all LAN ports on Linksys EA9500

External switch ports need to be bridged too.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
2 years agotools/isl: update the download URL
Rui Salvaterra [Thu, 21 Oct 2021 19:09:49 +0000 (20:09 +0100)]
tools/isl: update the download URL

isl.gforge.inria.fr has been dead since early this month [1]. Switch to
libisl.sourceforge.io for the time being.

[1] https://groups.google.com/g/isl-development/c/JGaMo2VUu_8

Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
2 years agotoolchain: switch packaged toolchain to tar.xz
Paul Spooren [Thu, 21 Oct 2021 01:54:26 +0000 (15:54 -1000)]
toolchain: switch packaged toolchain to tar.xz

Currently the tar.bz2 while ImageBuilder and SDK switched to tar.xz.
Unify it for faster compression since it will make use of
multi-threading.

Signed-off-by: Paul Spooren <mail@aparcar.org>
2 years agobcm53xx: switch to the upstream DSA-based b53 driver
Rafał Miłecki [Mon, 18 Oct 2021 13:04:17 +0000 (15:04 +0200)]
bcm53xx: switch to the upstream DSA-based b53 driver

1. Drop swconfig
2. Simplify network setup
3. Verify network config
4. Disable Buffalo WZR-900DHP for now - it misses ports definition

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Reviewed-By: Christian Lamparter <chunkeey@gmail.com>
2 years agobcm47xx: add kernel 5.10 support
Rafał Miłecki [Thu, 21 Oct 2021 15:26:56 +0000 (17:26 +0200)]
bcm47xx: add kernel 5.10 support

It's for *development* only as it doesn't work with lzma-loader due to
bigger size.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
2 years agohostapd: ubus: fix uninitialized pointer
David Bauer [Thu, 21 Oct 2021 15:09:28 +0000 (17:09 +0200)]
hostapd: ubus: fix uninitialized pointer

This fixes passing a bogus non-null pointer to the ubus handler in case
the transition request is rejected.

Signed-off-by: David Bauer <mail@david-bauer.net>
2 years agoath79: use correct USB package for DIR-505
David Bauer [Thu, 21 Oct 2021 13:51:48 +0000 (15:51 +0200)]
ath79: use correct USB package for DIR-505

AR9331 requires kmod-usb2-chipidea to use the USB ports. Include the
correct package so they can be used with the base image.

Signed-off-by: David Bauer <mail@david-bauer.net>
2 years agohostapd: fix up patches after the last commit
Felix Fietkau [Thu, 21 Oct 2021 10:37:09 +0000 (12:37 +0200)]
hostapd: fix up patches after the last commit

Signed-off-by: Felix Fietkau <nbd@nbd.name>
2 years agohostapd: fix a race condition on adding AP mode wds sta interfaces
Felix Fietkau [Wed, 20 Oct 2021 19:13:10 +0000 (21:13 +0200)]
hostapd: fix a race condition on adding AP mode wds sta interfaces

Both hostapd and netifd attempt to add a VLAN device to a bridge.
Depending on which one wins the race, bridge vlan settings might be incomplete,
or hostapd might run into an error and refuse to service the client.
Fix this by preventing hostapd from adding interfaces to the bridge and
instead rely entirely on netifd handling this properly

Signed-off-by: Felix Fietkau <nbd@nbd.name>
2 years agonetifd: update to the latest version
Felix Fietkau [Wed, 20 Oct 2021 19:12:31 +0000 (21:12 +0200)]
netifd: update to the latest version

c61a1d432b34 wireless: fix creating AP mode WDS station interfaces
f78bdec2ed5f wireless: fix handling vif attributes on reload with mode change

Signed-off-by: Felix Fietkau <nbd@nbd.name>
2 years agokernel: 5.10: backport Marvell 88E1510/2 PHY SFP support
Robert Marko [Fri, 8 Oct 2021 16:49:46 +0000 (18:49 +0200)]
kernel: 5.10: backport Marvell 88E1510/2 PHY SFP support

Backport upstream SFP support for the Marvell 88E1510/2 PHY-s.

Globalscale MOCHAbin uses this PHY for the hybrid
WAN port that has 1G SFP and 1G RJ45 with PoE PD
connected to it.

This allows the SFP port to be used on it as well as
parsing the SFP module details with ethtool.

Signed-off-by: Robert Marko <robert.marko@sartura.hr>
2 years agokernel: 5.10: backport 100 BaseX SFP support
Robert Marko [Fri, 8 Oct 2021 16:49:45 +0000 (18:49 +0200)]
kernel: 5.10: backport 100 BaseX SFP support

Backport upstream support for 100Base-FX, 100Base-LX, 100Base-PX and
100Base-BX10 SFP modules.

This is a prerequisite for the Globalscale MOCHAbin hybrid 1G
SFP/Copper support backporting.

Signed-off-by: Robert Marko <robert.marko@sartura.hr>
2 years agoltq-vdsl-app: add error vector counters to the ubus metrics
Andre Heider [Sun, 3 Oct 2021 10:07:41 +0000 (12:07 +0200)]
ltq-vdsl-app: add error vector counters to the ubus metrics

These are useful stats to debug vector related line deteriorations,
see [0].

Example output:
    "erb": {
    "sent": 169925,
    "discarded": 0
    }

[0] https://forum.openwrt.org/t/vectoring-on-lantiq-vrx200-vr9-missing-callback-for-sending-error-samples/104046

Signed-off-by: Andre Heider <a.heider@gmail.com>
2 years agoltq-vdsl-app: prepare for multiple mei ioctls
Andre Heider [Sun, 3 Oct 2021 10:05:51 +0000 (12:05 +0200)]
ltq-vdsl-app: prepare for multiple mei ioctls

Refactor so that the outer function opens and closes the mei fd and
passes it around, just as with the main fd.

That also allows us to use the IOCTL macro in get_vector_status() and
clean up accordingly.

Switch to AUTORELEASE while at it.

Signed-off-by: Andre Heider <a.heider@gmail.com>
2 years agoath10k-ct: update to version from 2021-09-22
Andrew Robbins [Tue, 5 Oct 2021 15:00:16 +0000 (11:00 -0400)]
ath10k-ct: update to version from 2021-09-22

Add in a fix for 160Mhz dfs on 5.10 and higher.
Add support for 5.13 and 5.15 kernels.
Add of_get_mac_address support for 5.15 driver.

Signed-off-by: Andrew Robbins <andrew@robbinsa.me>
2 years agokernel: bump 5.10 to 5.10.75
Rui Salvaterra [Wed, 20 Oct 2021 12:07:59 +0000 (13:07 +0100)]
kernel: bump 5.10 to 5.10.75

Deleted (upstreamed):
bcm27xx/patches-5.10/950-0735-xhci-guard-accesses-to-ep_state-in-xhci_endpoint_res.patch [1]

[1] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=linux-5.10.y&id=dc3e0a20dbb9dbaa22f4a33dea34230f8c663c40

Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
2 years agokernel: bump 5.10 to 5.10.74
Rui Salvaterra [Sun, 17 Oct 2021 17:31:09 +0000 (18:31 +0100)]
kernel: bump 5.10 to 5.10.74

Patches automatically refreshed.

Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
2 years agokernel: bump 5.10 to 5.10.73
Rui Salvaterra [Wed, 13 Oct 2021 10:02:37 +0000 (11:02 +0100)]
kernel: bump 5.10 to 5.10.73

Patches automatically refreshed.

Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
2 years agokernel: bump 5.4 to 5.4.155
John Audia [Wed, 20 Oct 2021 12:32:55 +0000 (08:32 -0400)]
kernel: bump 5.4 to 5.4.155

All patches automatically rebased.

Signed-off-by: John Audia <graysky@archlinux.us>
2 years agokernel: bump 5.4 to 5.4.154
John Audia [Sun, 17 Oct 2021 13:14:42 +0000 (09:14 -0400)]
kernel: bump 5.4 to 5.4.154

All patches automatically rebased.

Signed-off-by: John Audia <graysky@archlinux.us>
2 years agokernel: bump 5.4 to 5.4.153
John Audia [Wed, 13 Oct 2021 19:14:17 +0000 (15:14 -0400)]
kernel: bump 5.4 to 5.4.153

Removed upstreamed:
  backport-5.4/070-v5.5-MIPS-BPF-Restore-MIPS32-cBPF-JIT.patch

All other patches automatically rebased.

Signed-off-by: John Audia <graysky@archlinux.us>
2 years agowolfssl: fix compile when enable-devcrypto is set
Ivan Pavlov [Fri, 15 Oct 2021 05:11:10 +0000 (08:11 +0300)]
wolfssl: fix compile when enable-devcrypto is set

fixing linking error when --enable-devcrypto=yes
fixes: 7d92bb050961 wolfssl: update to 4.8.1-stable

Signed-off-by: Ivan Pavlov <AuthorReflex@gmail.com>
2 years agoiw: sync nl80211 with kernel backports
Hauke Mehrtens [Sun, 10 Oct 2021 13:41:43 +0000 (15:41 +0200)]
iw: sync nl80211 with kernel backports

The nl80211 was out of sync with the version used in our backports. This
broke the configuration of the antenna gain.

Fixes: 2bfac61483db ("mac80211: backport support for BSS color changes")
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
2 years agorealtek: switch to kernel 5.10
Adrian Schmutzler [Tue, 19 Oct 2021 19:08:58 +0000 (21:08 +0200)]
realtek: switch to kernel 5.10

The usual testers did their tests. Now we need testers who use the
master builds.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
2 years agomac80211: add missing patch chunk for mac80211_hwsim
Felix Fietkau [Wed, 20 Oct 2021 12:04:24 +0000 (14:04 +0200)]
mac80211: add missing patch chunk for mac80211_hwsim

Fixes build error

Signed-off-by: Felix Fietkau <nbd@nbd.name>
2 years agomt76: update to the latest version
Felix Fietkau [Tue, 19 Oct 2021 10:43:03 +0000 (12:43 +0200)]
mt76: update to the latest version

ebf5b2336591 mt7615/mt7915: fix hwmon device name
8d12f9ed275d mt76: mt7615: mt7622: fix ibss and meshpoint
e7883cdc0b4e mt76: mt7915: improve code readability in mt7915_mcu_sta_bfer_ht
831d5967abb9 wireless: fix spelling of A-MSDU in HE capabilities
f09cb04be261 wireless: align some HE capabilities with the spec
0eeba8f2952d wireless: align HE capabilities A-MPDU Length Exponent Extension
655a6c65b8a7 mt76: mt7915: introduce mt7915_mcu_beacon_check_caps()
4440025d0ba9 mt76: mt7915: fix txbf starec TLV issues
87d2fb6fbff5 mt76: mt7915: improve starec readability of txbf
53c6a3cb7f6b mt76: mt7915: fix sta_rec_wtbl tag len
f517845e4f28 mt76: mt7915: rework starec TLV tags
1df017bc39a3 mt76: mt7915: fix muar_idx in mt7915_mcu_alloc_sta_req()
6724b0a9a748 mt76: mt7915: set VTA bit in tx descriptor
f1f505cbbb30 mt76: mt7915: set muru platform type
8c9d4b38d258 mt76: mt7915: remove dead code in mt7915_get_et_stats
d0ccc4297935 mt76: rely on phy pointer in mt76_register_debugfs_fops routine signature
0af0af82bb97 mt76: mt7915: introduce mt76 debugfs sub-dir for ext-phy
10e85d62f213 mt76: mt7915: improve code readability for xmit-queue handler
b6051f7713d2 mt76: sdio: export mt76s_alloc_rx_queue and mt76s_alloc_tx routines
9a97c38a309d mt76: mt7921: get rid of unused variable in mt7921_tx_complete_skb
c02847c05884 mt76: mt7921: get rid of unused variable in mt7921_mac_tx_free
60dd47a0a62e mt76: mt7915: remove dead code in debugfs code
ce74fc020d81 mt76: mt7921: add MU EDCA cmd support
c062f6920356 mt76: mt7921: refactor mac.c to be bus independent
bfa909c833e8 mt76: mt7921: refactor dma.c to be pcie specific
6556bddf26d2 mt76: mt7921: refactor mcu.c to be bus independent
1c8418207c86 mt76: mt7921: refactor init.c to be bus independent
6cf8248c1a44 mt76: mt7921: add MT7921_COMMON module
77600b0c10ac mt76: connac: move mcu reg access utility routines in mt76_connac_lib module
65362a00d07d mt76: mt7663s: rely on mcu reg access utility
956206bb55c3 mt76: mt7921: make all event parser reusable between mt7921s and mt7921e
f0dedcf6aaf0 mt76: mt7921: use physical addr to unify register access
5079d5b0b13c mt76: sdio: move common code in mt76_sdio module
26257594398b mt76: sdio: introduce parse_irq callback
e353424f1b07 mt76: sdio: extend sdio module to support CONNAC2
ddab3dd25f94 mt76: connac: extend mcu_get_nic_capability
b2d9a1748a41 mt76: mt7921: rely on mcu_get_nic_capability
e6ce5d9cbda0 mt76: mt7921: refactor mt7921_mcu_send_message
ce3706a65ccd mt76: mt7921: introduce mt7921s support
3143118baf53 mt76: mt7921s: add reset support
645eac64bece mt76: mt76x0: correct VHT MCS 8/9 tx power eeprom offset
d54796787cb7 mt76: move mt76_sta_stats in mt76.h
094e085abf5a mt76: move mt76_ethtool_worker_info in mt76 module
f80ab6dde63d mt76: mt7915: run mt7915_get_et_stats holding mt76 mutex
4a11cb67dc27 mt76: mt7915: move tx amsdu stats in mib_stats
486da6fa2512 mt76: do not reset MIB counters in get_stats callback
d8837b7c8dcd mt76: mt7921: add some more MIB counters
5ffe086fcd1b mt76: mt7921: introduce stats reporting through ethtool
69154ae23f6b mt76: mt7921: add sta stats accounting in mt7921_mac_add_txs_skb
4b65fbc4e203 mt76: mt7921: move tx amsdu stats in mib_stats
35b8025f466b mt76: mt7921: add per-vif counters in ethtool
cfbbd861eb12 mt76: mt7915: enable HE UL MU-MIMO
a0b94987df80 mt76: mt7915: rework mt7915_mcu_sta_muru_tlv()
5fbb686e9c0c mt76: mt7915: fix missing HE phy cap
b649678c18ca mt76: mt7915: change max rx len limit of hw modules

Signed-off-by: Felix Fietkau <nbd@nbd.name>
2 years agomac80211: backport a few trivial patches
Felix Fietkau [Tue, 19 Oct 2021 09:19:26 +0000 (11:19 +0200)]
mac80211: backport a few trivial patches

No functional changes, just some renames to make it easier to keep mt76 in
sync with upstream

Signed-off-by: Felix Fietkau <nbd@nbd.name>
2 years agotools/llvm-bpf: add llvm+clang build suitable for compiling code to eBPF
Felix Fietkau [Sun, 17 Oct 2021 15:50:53 +0000 (17:50 +0200)]
tools/llvm-bpf: add llvm+clang build suitable for compiling code to eBPF

Preparation for building packages that ship eBPF code

Signed-off-by: Felix Fietkau <nbd@nbd.name>
2 years agoinclude/cmake.mk: fix host builds with CMAKE_BINARY_SUBDIR
Felix Fietkau [Sun, 17 Oct 2021 15:01:23 +0000 (17:01 +0200)]
include/cmake.mk: fix host builds with CMAKE_BINARY_SUBDIR

Use it in the same way as for target builds

Signed-off-by: Felix Fietkau <nbd@nbd.name>
2 years agoREADME: mention video feed
Paul Spooren [Thu, 14 Oct 2021 18:33:29 +0000 (08:33 -1000)]
README: mention video feed

The video feed just got support for Wayland which allows OpenWrt devices
to run as a Kiosk (displaying browser content via a display).

The availability of these packages should be mentioned to users.

Signed-off-by: Paul Spooren <mail@aparcar.org>
2 years agoncurses: add tmux terminfo
Jitao Lu [Sun, 17 Oct 2021 06:16:57 +0000 (14:16 +0800)]
ncurses: add tmux terminfo

They're preferred terminal descriptions for tmux, with additional support to
some special characters and italic fonts. More info can be found at:
https://github.com/tmux/tmux/wiki/FAQ

Fixes: FS#3404
Signed-off-by: Jitao Lu <dianlujitao@gmail.com>
2 years agohostapd: fix goto loop for ubus assoc handler
David Bauer [Sun, 17 Oct 2021 21:29:57 +0000 (23:29 +0200)]
hostapd: fix goto loop for ubus assoc handler

When a ubus event handler denies a association with a non-zero return
value, the code jumps to preceeding code, creating an endless loop until
the event handler accepts the assc request.

Move the ubus handler further up the code to avoid creating such a loop.

Signed-off-by: David Bauer <mail@david-bauer.net>
2 years agowireguard-tools: add uci option to disable wireguard peers
Stepan Henek [Sun, 25 Apr 2021 21:32:29 +0000 (23:32 +0200)]
wireguard-tools: add uci option to disable wireguard peers

Right now when I want to temporarily disable wg peer I need to delete
the entire peer section. This is not such a good solution because I
loose the previous configuration of the peer.

This patch adds `disabled` option to peer config which causes that
the config section is ignored.

Signed-off-by: Stepan Henek <stepan.henek@nic.cz>
[use $(AUTORELEASE)]
Signed-off-by: Paul Spooren <mail@aparcar.org>
2 years agonftables: bump to 1.0.0
Stijn Tintel [Thu, 14 Oct 2021 19:08:44 +0000 (22:08 +0300)]
nftables: bump to 1.0.0

This introduces support for hardware flow offloading, which was added in
in nftables 0.9.9.

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
Acked-by: Jo-Philipp Wich <jo@mein.io>
2 years agotegra: switch to kernel 5.10
Adrian Schmutzler [Mon, 18 Oct 2021 19:27:33 +0000 (21:27 +0200)]
tegra: switch to kernel 5.10

This target has testing support for kernel 5.10 for four months now.
Time to switch the default.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
Acked-by: Tomasz Maciej Nowak <tmn505@gmail.com>
2 years agobcm53xx: enable Linksys EA6300 & EA9200 builds
Rafał Miłecki [Mon, 18 Oct 2021 13:59:43 +0000 (15:59 +0200)]
bcm53xx: enable Linksys EA6300 & EA9200 builds

Both should be supported since:
1. Adding NVMEM driver for NVRAM
2. Using NVRAM info for determining active firmware partition

Linksys EA9500 uses very similar design and works fine.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
2 years agoprocd: update to git HEAD
Daniel Golle [Sun, 17 Oct 2021 20:21:49 +0000 (21:21 +0100)]
procd: update to git HEAD

 9b1e035 jail: netifd: code cosmetics
 d2a2ecc jail: netifd: fix error handling issue reported by coverity
 e1d7cee jail: netifd: check target netns fd before using it
 59f7699 uxc: add missing 'break' statement

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
2 years agowolfssl: remove --enable-sha512 configure switch
Andre Heider [Sat, 2 Oct 2021 08:40:00 +0000 (10:40 +0200)]
wolfssl: remove --enable-sha512 configure switch

It's the default anyway and this just looks confusing, as if it wasn't.

Switch to AUTORELEASE while at it.

The binary size is unchanged.

Signed-off-by: Andre Heider <a.heider@gmail.com>
2 years agowolfssl: always build with --enable-reproducible-build
Andre Heider [Sat, 2 Oct 2021 08:50:49 +0000 (10:50 +0200)]
wolfssl: always build with --enable-reproducible-build

This gates out anything that might introduce semantically frivolous jitter,
maximizing chance of identical object files.

The binary size shrinks by 8kb:
1244352 staging_dir/target-mipsel_24kc_musl/usr/lib/libwolfssl.so.4.8.1.39c36f2f
1236160 staging_dir/target-mipsel_24kc_musl/usr/lib/libwolfssl.so.4.8.1.39c36f2f

Signed-off-by: Andre Heider <a.heider@gmail.com>
2 years agowolfssl: build with WOLFSSL_ALT_CERT_CHAINS
Andre Heider [Wed, 6 Oct 2021 08:54:48 +0000 (10:54 +0200)]
wolfssl: build with WOLFSSL_ALT_CERT_CHAINS

"Alternate certification chains, as oppossed to requiring full chain
validataion. Certificate validation behavior is relaxed, similar to
openssl and browsers. Only the peer certificate must validate to a trusted
certificate. Without this, all certificates sent by a peer must be
used in the trust chain or the connection will be rejected."

This fixes e.g. uclient-fetch and curl connecting to servers using a Let's
Encrypt certificate which are cross-signed by the now expired
DST Root CA X3, see [0].

This is the recommended solution from upstream [1].

The binary size increases by ~12.3kb:
1236160 staging_dir/target-mipsel_24kc_musl/usr/lib/libwolfssl.so.4.8.1.39c36f2f
1248704 staging_dir/target-mipsel_24kc_musl/usr/lib/libwolfssl.so.4.8.1.39c36f2f

[0] https://github.com/openwrt/packages/issues/16674
[1] https://github.com/wolfSSL/wolfssl/issues/4443#issuecomment-934926793

Signed-off-by: Andre Heider <a.heider@gmail.com>
[bump PKG_RELEASE]
Signed-off-by: David Bauer <mail@david-bauer.net>
2 years agoramips: remove kmod-mt7663-firmware-sta from device packages
Felix Fietkau [Sun, 17 Oct 2021 14:05:34 +0000 (16:05 +0200)]
ramips: remove kmod-mt7663-firmware-sta from device packages

This firmware should only be used for mobile devices (e.g. laptops), where
AP mode functionality is typically not used. This firmware supports a lot
of power saving offload functionality at the expense of AP mode support.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
2 years agoramips: fix dtc warnings for telco-electronics_x1
Rosen Penev [Fri, 15 Oct 2021 02:15:49 +0000 (19:15 -0700)]
ramips: fix dtc warnings for telco-electronics_x1

In all other dts files, the entire block is not edited like this.
They're edited separately.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
2 years agozynq: switch to kernel 5.10
Luis Araneda [Sun, 10 Oct 2021 02:55:06 +0000 (23:55 -0300)]
zynq: switch to kernel 5.10

Use kernel 5.10 by default

compile-tested: all devices from target (wth ALL_KMODS)
run-tested: Digilent Zybo Z7-20

Signed-off-by: Luis Araneda <luaraneda@gmail.com>
2 years agozynq: kernel: update config for 5.10
Luis Araneda [Sun, 10 Oct 2021 02:55:05 +0000 (23:55 -0300)]
zynq: kernel: update config for 5.10

Update config with make kernel_oldconfig

Signed-off-by: Luis Araneda <luaraneda@gmail.com>
2 years agozynq: kernel: copy config from 5.4 to 5.10
Luis Araneda [Sun, 10 Oct 2021 02:55:04 +0000 (23:55 -0300)]
zynq: kernel: copy config from 5.4 to 5.10

Signed-off-by: Luis Araneda <luaraneda@gmail.com>
2 years agozynq: kernel: remove wireless extensions symbols
Luis Araneda [Sun, 10 Oct 2021 02:55:03 +0000 (23:55 -0300)]
zynq: kernel: remove wireless extensions symbols

This fixes compilation of several wireless drivers that
require support for the old wireless extension to work.
One example is kmod-hermes.

The symbols are set to "y" on generic configuration.
But they were wrongly disabled on the target-specific
configuration.

Signed-off-by: Luis Araneda <luaraneda@gmail.com>
2 years agozynq: kernel: refresh config
Luis Araneda [Sun, 10 Oct 2021 02:55:02 +0000 (23:55 -0300)]
zynq: kernel: refresh config

using "make kernel_oldconfig"

Several configs are now part of generic

Signed-off-by: Luis Araneda <luaraneda@gmail.com>
2 years agobuildsystem: add CONFIG_SECCOMP
Florian Eckert [Fri, 15 Oct 2021 07:19:12 +0000 (09:19 +0200)]
buildsystem: add CONFIG_SECCOMP

Until now, this feature was switched on via the kernel configuration
option KERNEL_SECCOMP.

The follwing change a7f794cd2aa104fdbd4c6e38f9b76373bf9b96e1 now requires that
the package procd-seccomp must also enabled for buildinmg.

However, this is not the case we have no dependency and the imagebuilder
cannot build the image, because of the implicit package selection.

This change adds a new configuration option CONFIG_SECCOMP.
The new option  has the same behaviour as the configuration
option CONFIG_SELINUX.

If the CONFIG_SECCOMP is selected then the package procd-seccomp and
KERNEL_SECCOMP is enabled for this build.

Signed-off-by: Florian Eckert <fe@dev.tdt.de>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
2 years agomediatek: add EEPROM data for BPi-R64 2.4GHz wmac
Daniel Golle [Fri, 20 Aug 2021 17:31:48 +0000 (18:31 +0100)]
mediatek: add EEPROM data for BPi-R64 2.4GHz wmac

EEPROM data extracted from vendor image found at
http://forum.banana-pi.org/t/bpi-r64-mt7622-mac80211-wifi-driver/10246/77
http://forum.banana-pi.org/uploads/short-url/jworbyBYpvrw9VQ2sx92B9z6DWS.bin

MAC address in the EEPROM has been zero'd which results in random
address on boot.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
2 years agokernel: backport a rewrite of the mips eBPF JIT implementation
Felix Fietkau [Wed, 13 Oct 2021 15:49:31 +0000 (17:49 +0200)]
kernel: backport a rewrite of the mips eBPF JIT implementation

This adds support for eBPF JIT for 32 bit targets and significantly improves
correctness.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
2 years agohostapd: ubus: add BSS transtiton request method
David Bauer [Sat, 5 Dec 2020 19:17:35 +0000 (20:17 +0100)]
hostapd: ubus: add BSS transtiton request method

The existing wnm_disassoc_imminent ubus method only supports issuing a
bss transition request with the disassoc imminent flag set.
For use-cases, where the client is requested to roam to another BSS
without a pending disassoc, this existing method is not suitable.

Add a new bss_transition_request ubus method, which provides a more
universal way to dispatch a transition request. It takes the following
arguments:

Required:
addr: String - MAC-address of the STA to send the request to (colon-seperated)

Optional:
abridged - Bool - Indicates if the abridged flag is set
disassociation_imminent: Bool - Whether or not the disassoc_imminent
                         flag is set
disassociation_timer: I32 - number of TBTTs after which the client will
                      be disassociated
validity_period: I32 - number of TBTTs after which the beacon
                 candidate list (if included) will be invalid
neighbors: blob-array - Array of strings containing neighbor reports as
           hex-string

Signed-off-by: David Bauer <mail@david-bauer.net>
2 years agohostapd: ubus: add notification for BSS transition response
David Bauer [Wed, 21 Oct 2020 18:50:01 +0000 (20:50 +0200)]
hostapd: ubus: add notification for BSS transition response

To allow steering daemons to be aware of the STA-decided transition
target, publish WNM transition responses to ubus. This way, steerings
daemons can learn about STA-chosen targets and send a better selection
of transition candidates.

Signed-off-by: David Bauer <mail@david-bauer.net>
2 years agotools/fakeroot: explicitly pass CPP variable
Felix Fietkau [Wed, 13 Oct 2021 14:36:16 +0000 (16:36 +0200)]
tools/fakeroot: explicitly pass CPP variable

For some reason, the generated configure script fails to properly set up
the internal preprocessor command variable, causing the host OS check for
Darwin to fail after the last update.
Explicitly setting CPP fixes this issue

Signed-off-by: Felix Fietkau <nbd@nbd.name>
2 years agox86/64: enable MMIO_CMDLINE_DEVICES for firecracker support
Paul Spooren [Mon, 11 Oct 2021 20:33:26 +0000 (10:33 -1000)]
x86/64: enable MMIO_CMDLINE_DEVICES for firecracker support

This Kernel option allows to run OpenWrt witin a `firecracker` micro VM.

Firecracker is a KVM-based tool for superfast booting VMs on x86_64 and
aarch64. It makes rootfs available to the guest as a virtio-mmio device
and passes its address via the kernel cmdline. A kernel without
CONFIG_VIRTIO_MMIO_CMDLINE_DEVICES will not recognize the rootfs
virtio-mmio device.

Suggested-by: Packet Please <pktpls@systemli.org>
Signed-off-by: Paul Spooren <mail@aparcar.org>