uboot-mediatek: update to v2026.01

Remove upstreamed patches:
  100-08-cmd-mtd-add-markbad-subcommand-for-NMBM-testing.patch [1]
  101-01-mtd-spinand-add-support-for-FORESEE-F35SQA002G.patch [2]
  101-02-mtd-spinand-add-support-for-FORESEE-F35SQA001G.patch [3]
  110-mtd-spi-nand-add-support-for-FudanMicro-FM25S01A.patch [4]

Some SPI-NAND driver macro definitions and function parameters have
been changed in the latest release[3]. Hence we also had to rework
the related local patches to follow the upstream changes.

Tested on MT7981 SPI-NOR/EMMC.

[1] https://github.com/u-boot/u-boot/commit/21c1098cf443ea225090520cbe3ac7fc9e75cd33
[2] https://github.com/u-boot/u-boot/commit/2a0f8e7da0e5db93df8d48c5aab05037e071ca1e
[3] https://github.com/u-boot/u-boot/commit/2cbdd3e4491b6fc278c54d2a490939ecccc76127
[4] https://github.com/u-boot/u-boot/commit/8b984b5a397bc5fe829ca95e64f990c1e287beaa

Signed-off-by: Shiji Yang <yangshiji66@outlook.com>
[daniel@makrotopia.org: tested MT7622 SNAND and SPI-NOR]
Signed-off-by: Daniel Golle <daniel@makrotopia.org>

gemini: Fix up the rootfs device for SQ201

The rootfs partition is /dev/mtdblock3, fix it up.

Link: https://github.com/openwrt/openwrt/pull/21662
Signed-off-by: Linus Walleij <linusw@kernel.org>

gemini: Add DTS fix from upstream kernel

This contains small but important DTS fixes for the
RedBoot partitions.

Link: https://github.com/openwrt/openwrt/pull/21662
Signed-off-by: Linus Walleij <linusw@kernel.org>

imagebuilder: skip repository file when standalone

Standalone image builder doesn't have a repositories file as all
packages are included, which causes:

ERROR: failed to read repositories: PATH_TO_BUILDER/repositories: No such file or directory

The images are still built, so this is more of an informational error.

Pass related argument to apk only when CONFIG_IB_STANDALONE is not set.

Fixes: a8d17c21 ("imagebuilder: actually support IB from buildbot")
Signed-off-by: George Sapkin <george@sapk.in>
Link: https://github.com/openwrt/openwrt/pull/21658
Signed-off-by: Robert Marko <robimarko@gmail.com>

uboot-microchipsw: add FIP update flashing shortcut

Add env variable to make it easy to flash an FIP update.

Signed-off-by: Robert Marko <robert.marko@sartura.hr>

microchipsw: update DTS with the latest pending upstream

Update the DTS to the latest pending upstream version, it looks like this
should get merged soon upstream.

Signed-off-by: Robert Marko <robert.marko@sartura.hr>

d1: Enable LED driver for D1 builds

This enables the kernel configuration option
to build the SUN50I_A100 driver, which is
the one driver used by the D1 for the RGB
LED controller.

Signed-off-by: Gabriel Roper <lockheedmaniac@gmail.com>

d1: Add device tree configuration for D1 RGB LED Controller

This adds the device tree patches written by Samuel Holland
to the kernel. The driver was merged into mainline Linux in 6.8,
but the device tree patches were never merged into mainline.

Signed-off-by: Gabriel Roper <lockheedmaniac@gmail.com>

d1: refresh kernel config

Run 'make kernel_oldconfig' to get an up-to-date config.

Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>

d1: enable overlay support

Add the loop module and mkfs tool to use f2fs/ext4
based overlays. This fixes the following bug:

mount_root: unable to create loop device
mount_root: no usable overlay filesystem found, using tmpfs overlay

Fixes: 99545b4bb1faf7c3 ("d1: add new target")
Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>

d1: enable usbgadget feature

The USB of most d1 devices works in otg or peripheral mode.
Enable this feature to use usbgadget and fix musb errors:

musb-sunxi 4100000.usb: Invalid or missing 'dr_mode' property
musb-sunxi 4100000.usb: probe with driver musb-sunxi failed with error -22

Fixes: 99545b4bb1faf7c3 ("d1: add new target")
Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>

uboot-d1: update to 2024.01

Update to version 2024.01 to remove upstream patches.
The remaining patches come from this repository:
https://github.com/smaeul/u-boot/tree/d1-wip

[for Dongshan Nezha STU, Nezha D1]
Tested-by: Zoltan HERPAI <wigyori@uid0.hu>
Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>

kexec-tools: bump to 2.0.32

Change log is available at: https://github.com/horms/kexec-tools/compare/v2.0.28...v2.0.32

Manually refreshed patch:
- 020-i386-improve-basename-compatibility.patch

Remove upstreamed patch:
- 010-Fix-building-on-x86_64-with-binutils-2.41.patch

Signed-off-by: Roc Lai <laipeng668@qq.com>
Link: https://github.com/openwrt/openwrt/pull/21623
Signed-off-by: Robert Marko <robimarko@gmail.com>

kexec-tools: add aarch64 architecture support

To update kexec and kdump dependencies for supporting the aarch64 architecture.

Signed-off-by: Roc Lai <laipeng668@qq.com>
Link: https://github.com/openwrt/openwrt/pull/21623
Signed-off-by: Robert Marko <robimarko@gmail.com>

ath79: rb91x_nand: fix some issues in probe

Add devm for ioremap to proper cleanup.

Test gpiod_direction_input for failure.

Remove rb91x_nand_release and fix wrong usage.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/21165
Signed-off-by: Robert Marko <robimarko@gmail.com>

realtek: phy: add RTL8218D initialization

The RTL8218D currently relies on proper U-Boot configuration. In
case that is not possible, provide a basic setup sequence that can
bring the PHY "alive". The SDK provides multiple configuration
sequences for two operation modes (XSGMII or QSGMII) and the different
SoC families. Due to limited testing resources only provide a setup
for RTL93xx devices and both modes at the moment.

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21551
Signed-off-by: Robert Marko <robimarko@gmail.com>

dnsmasq: bump release to 2.92

bump dnsmasq to latest 2.92

updated 200-ubus_dns.patch
no changes to 100-remove-old-runtime-kernel-support.patch
all remaining patches not required

Changelog for version 2.92 https://thekelleys.org.uk/dnsmasq/CHANGELOG

Signed-off-by: gongzi miao <miaogongzi0227@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/21598
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

wifi-scripts: iwinfo.uc: cache survey data in iface object

Extract survey fetching into get_survey() and store results in iface.survey,
allowing access to full survey info (not just noise) for later use.

Signed-off-by: Felix Fietkau <nbd@nbd.name>

wifi-scripts: refactor iwinfo.uc to support dynamic data updates

Moved interface discovery and data population into an exported update()
function that can be called on-demand to refresh wireless interface
information. This allows using iwinfo.uc as a library inside daemons.

Signed-off-by: John Crispin <john@phrozen.org>

udebug: update to Git HEAD (2026-01-16)

c5c493ed5787 udebug-cli: fix reference to invalid variable in error message
875e1a7af6ca udebug-cli: add logdump command

Signed-off-by: Felix Fietkau <nbd@nbd.name>

build: fix stale package install stamps for unselected variants

When a package variant changes from =y to =m, its .install stamp file
was not being cleaned up, causing both variants to appear in the
package install list.

Fix by making compile depend on prepare-package-install, ensuring
the stamp file is always cleared. Only =y packages recreate it.

Signed-off-by: Felix Fietkau <nbd@nbd.name>

build: auto-generate CONFLICTS for package variants sharing PROVIDES

Packages using PROVIDES to declare virtual package alternatives (like
tc-tiny, tc-bpf, tc-full all providing 'tc') could be simultaneously
selected as =y, causing installation conflicts. The PROVIDES mechanism
only handles dependency resolution, not mutual exclusion.

Add add_implicit_provides_conflicts() to automatically generate CONFLICTS
from default variants to non-default variants sharing the same PROVIDES.
This ensures only one variant can be built-in (=y) at a time.

Skip generating implicit conflicts when the non-default already has
explicit CONFLICTS with the default, to avoid Kconfig dependency cycles
with the select-based dependency resolution.

Signed-off-by: Felix Fietkau <nbd@nbd.name>

mediatek: dts: mt7981: fix spi clocks

Fix spi1 sel-clk source and add missing assigned-clocks.

Signed-off-by: Shiji Yang <yangshiji66@outlook.com>
Link: https://github.com/openwrt/openwrt/pull/21108
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

mediatek: dts: mt7981: remove internal phy LED pinctrl

This default pinctrl doesn't work at all. We must add pinctrl phandle
to the ethernet-phy node or mac node. This patch removed the internal
phy LED pinctrl group definitions from the SOC dtsi because it is
currently only used by one device. Some incorrect LED pinctrls also
have been removed because these devices do not have hardware phy LED.

Signed-off-by: Shiji Yang <yangshiji66@outlook.com>
Link: https://github.com/openwrt/openwrt/pull/21108
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

mediatek: dts: mt7981: add mediatek,wed-pcie syscon phandle

The PCIe NIC WED offload features depend on it. Also add the missing
"syscon" compatible for the wed_pcie node as it's accessed by
syscon_regmap_lookup_by_phandle().

Signed-off-by: Shiji Yang <yangshiji66@outlook.com>
Link: https://github.com/openwrt/openwrt/pull/21108
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

mediatek: dts: mt7981: load xhci phy efuse by default

This should improve the stability of the USB 3.0 port[1].

[1] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=linux-6.12.y&id=6f2b033cb883f64ad084a75f13634242c7e179a6
Signed-off-by: Shiji Yang <yangshiji66@outlook.com>
Link: https://github.com/openwrt/openwrt/pull/21108
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

mediatek: dts: mt7981: remove useless nodes

These stuffs were copied from the vendor SDK. There are currently
no drivers compatible with them. The pending upstream patches did
not include them either.

Signed-off-by: Shiji Yang <yangshiji66@outlook.com>
Link: https://github.com/openwrt/openwrt/pull/21108
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

econet: add clock/reset support

This is in preparation for the next update to EcoNet Ethernet driver
which will be using real resets rather than bit-bashing. Patches are
backported up to the current state of clk-next because I intend on
upstreaming these patches soon.

Signed-off-by: Caleb James DeLisle <cjd@cjdns.fr>
Link: https://github.com/openwrt/openwrt/pull/21545
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

rockchip: fix set default serial_{addr,port} in boot script

On some SoCs like RK3399, U-Boot no longer export the serial
address in $stdout variable, it has been changed to something like
`stdout=serial,vidconsole` since video supported was added. So
detect $soc variable instead, this also makes the script more clear.

Addtional note: rk3566 is reported as rk3568, rk3582 is reported as
rk3588 in $soc.

Signed-off-by: Tianling Shen <cnsztl@immortalwrt.org>
Link: https://github.com/openwrt/openwrt/pull/21610
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

ucode: update to Git 85922056ef7 (2026-01-16)

45bf891e1d8d rtnl: add module documentation
8bbf01215ce3 nl80211: add module documentation
40a6aeb698e1 nl80211: add new attribute "mlo_links" for per link information
376e8733609e ubus: complete pending requests when disconnecting locally
32449bfb69d6 io: add ptsname, tcgetattr, tcsetattr, grantpt, unlockpt
76bf32679d86 digest: fix jsdoc to reveal sha512 functions
58b4597fa2ae build: drop remnant
6ef0b3ada3fd build: respect bin/lib paths
28132276a426 build: adjust flags and definitions
cb1b1c1a097b zlib: make chunk size configurable
6f80655c88c3 debian: refresh packaging
9fc4889c6e8e debian: adjust LTO build
a503a49f5cef debian: build "digest" and "zlib" modules
ffc48a2a4db9 socket: fix off-by-one in uv_to_sockaddr()

Fixes: https://github.com/jow-/ucode/issues/366
dropped patch 130 - integrated at source.
refreshed patches

https://github.com/openwrt/openwrt/pull/21585
Signed-off-by: Paul Donald <newtwen+github@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/21585
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

realtek: mdio: derive phy smi-bus from parent node

The dts for RTL93xx devices has duplicate data about the
smi bus of a phy node. The parent node declares the number
of the bus and the realtek,smi-address attribute does the
same.

Remove the bus part from the realtek,smi-address attribute
and lookup the bus from the parent node. While we are here
remove all realtek,smi-address attributes where phy id
matches the bus address. The driver will use that as a
fallback.

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21438
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

realtek: dts: convert devices to mdio bus 1-3

Switch the mdio bus topology for devices that have their phys
attached to bus 1-3. This does not have any impact because
the mdio driver was completly redisgned

With this commit the bus id is stored twice. Once in the (new)
bus and in the (old) realtek,smi-address property. E.g.

&mdio_bus1 {
  reg = <1>; <<< bus id
  phy24: ethernet-phy@24 {
    reg = <26>;
    compatible = "ethernet-phy-ieee802.3-c22";
    realtek,smi-address = <1 2>; <<< bus & address id
  };
};

This redundancy will be removed later.

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21438
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

realtek: dts: add mdio bus 1-3 to RTL93xx

RTL93xx devices have 4 smi busses (0-3). Add them to the dts.

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21438
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

realtek: mdio/dsa: refactor mdio bus initialization

The mdio driver currently determines the smi bus and address from the
realtek,smi-address attribute of the phy. To better reflect the
topology and align with upstream, the phys should be relocated below
their associated bus. As an interim solution the following dts notation
is in focus.

mdio_ctrl: mdio-controller {
  mdio_bus0: mdio-bus@0 {
    reg = <0>;
    phy0: ethernet-phy@0 {
      reg = <0>;
      compatible = "ethernet-phy-ieee802.3-c45";
      realtek,smi-address = <8>;
    };

  &mdio_bus1 {
    reg = <1>;
    phy16: ethernet-phy@16 {
      reg = <16>;
      compatible = "ethernet-phy-ieee802.3-c45";
      realtek,smi-address = <2>;
  };
}

With this

- the phy reg property still denotes the port number
- the bus number can be derived from the parent bus node.
- the bus address is taken from realtek,smi-address

Refactor bus initialization so it can handle phy nodes below
multiple bus nodes.

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21438
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

realtek: mdio: rename iterator to "addr"

During initialization the mdio driver registers phys with the
iterator "pn". To make clear that it is a phy address rename it
to "addr".

While we are here improve the upper bound check of the maximum
possible address. This is the family specific cpu port and not
the generic upper bound constant for all devices.

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21438
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

realtek: mdio: register mdio bus at controller node

In the future the mdio controller will have multiple busses
defined in the dts below the controller node. Nevertheless
it will still hand out only one single bus to the kernel.
Attach the (exported) bus to the controller node instead of
the single (dts) bus subnode.

With this change the mdio lookup in the dsa driver must
be changed to point to the mdio controller node too.

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21438
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

realtek: dsa: Remove family check from port_get_stp_state()

The device specific stp_get() functions can return the state
of a given port individually. No need to disassemble the
device specific state table. Additionally change function
prefix.

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21527
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

realtek: dsa: enrich stp_get() return value

The family specific stp_get() functions have a new return value
that gives the stp status of a given port. Instead of a constant
zero value provide the real data.

While we are here align the RTL930x implementation with the
other ones.

- Drop the debug output
- Make use of priv->r->tbl_access_data_0()

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21527
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

realtek: dsa: convert stp_get signature

The device specific stp_get() function returns a device specific
state to the generic caller. So the caller must use a if/else
statement depending on the family to look into the retrieved data.

Change the signature of the function so it can return a device
independent state of a given port in the future. Additionally
use the new function prefix.

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21527
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

netfilter: remove nf-conntrack6

nf_conntrack_ipv6 is no longer a separate kernel module, so remove it.

Signed-off-by: Qingfang Deng <dqfext@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/21612
Signed-off-by: Robert Marko <robimarko@gmail.com>

wifi-scripts: fix WiFi 6E discovery for 6GHz 320MHz operation

WiFi 6E (802.11ax) clients cannot discover 6GHz APs operating at
320MHz because the HE Operation element contains uninitialized
center frequency values.

For EHT320 mode, the code sets eht_oper_centr_freq_seg0_idx but not
the corresponding HE values. Later, the HE values are copied from
VHT values, but VHT is not used on 6GHz, leaving he_oper_chwidth
and he_oper_centr_freq_seg0_idx at 0. This causes WiFi 6E clients
to see incorrect channel width information, making the AP invisible
to them during scanning.

Fix this by:
1. Setting he_oper_chwidth to 3 (160MHz) for EHT320 mode
2. Computing he_oper_centr_freq_seg0_idx based on the 160MHz segment
   that contains the primary channel
3. Preserving these pre-set values instead of overwriting them with
   uninitialized VHT values

WiFi 7 clients continue to see 320MHz operation via the EHT Operation
element, while WiFi 6E clients can now discover and connect at 160MHz.

Signed-off-by: Ryan Chen <rchen14b@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/21588
Signed-off-by: Robert Marko <robimarko@gmail.com>

microchipsw: lan969x: use network driver as a module

Due to PTP conflicts with CONFIG_ALL_KMODS which causes the end images
to end up with no switchdev driver included, lets switch to using the
switchdev and PHY drivers as modules instead.

Fixes: 88a404a2d177 ("microchipsw: add support for Microchip LAN969x switches")
Signed-off-by: Robert Marko <robert.marko@sartura.hr>

kernel: modules: package Microchip SparX-5 switchdev driver as a module

Package the Microchip SparX-5 switchdev driver as a module.

Signed-off-by: Robert Marko <robert.marko@sartura.hr>

strace: use bundled kernel headers

Build against the bundled kernel headers instead of using the current
kernel ones. This ensures strace is using the kernel headers it is
written against, and not a random one that may contain breaking uapi
changes (which happen from time to time).

Fixes build against 6.18 final and recent LTS/stable kernels that got
minor uapi breakages (rename of a 6.18 introduced #define and a struct
field).

Signed-off-by: Jonas Gorski <jonas.gorski@gmail.com>

Revert "kernel: pcs-mtk-lynxi: backport fix for in-band status capabilities"

This reverts commit 7390e63374a9e0bd62c2448f362479de269b841a.

This broke SFP handling on some Mediatek SoCs.
Some SFPs are configured for inband/2500base-x link mode and were
rejected with this change.

Fixes: https://github.com/openwrt/openwrt/issues/21436
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

realtek: make all link speeds work for RTL8221B in Zyxel XGS1210-12 rev B1

SGMII only works correctly on this device if inband auto-negotiation is
enabled. Configure the PHY for SGMII and in-band mode in the device tree
to make this happen.

For 2.5G link speeds the PHY will still switch to 2500Base-X without AN.

The same configuration also works on RTL8226, so it is fine to apply
this change to the A1 revision of XGS1010-12/XGS1210-12 as well.

Signed-off-by: Jan Hoffmann <jan@3e8.eu>
Link: https://github.com/openwrt/openwrt/pull/21605
Signed-off-by: Robert Marko <robimarko@gmail.com>

ramips: mt76x8: add support for Cudy LT400E v1

This commit adds support for the Cudy LT400E v1,
an indoor 4G/LTE WiFi router.
It has 4x10/100 Mbps Ethernet ports, a 2.4 GHz radio,
and a Quectel 4G/LTE EC200A modem.

Hardware:

SoC: MediaTek MT7628NN
Flash: 8 MiB SPI
RAM: 64 MiB
Ethernet: 4x10/100 Mbps
WiFi: 2.4 GHz (embedded MT7628, 802.11n)
WWAN: Quectel EC200A-EL 4G/LTE Cat 4 modem
SIM: nano-SIM slot
Buttons: 1 Reset button, 1 WPS button
LEDs: 10x (blue)
Power: 12 V DC barrel jack

Installation:
The stock firmware is a customized variant of OpenWrt, which implements
a signature check that only allows flashing official firmware. Cudy
offers, on their website [1][6], intermediate OpenWrt firmware images
which do not implement the signature check. After flashing the
intermediate image, the upstream (i.e., official) OpenWrt image can be
installed.  See the device's entry in OpenWrt's Wiki for more details [2].

Cellular network:
The Quectel EC200A modem is internally connected to the SoC via USB, and
operates in ECM mode, supported by the CDC Ether driver [3]. When
inserting a SIM card in the slot, if no PIN is needed, the network
interface "usb0" is brought up and the cellular connection works out of
the box. To interact with the modem directly (unlock the PIN, change APN
and PDP settings, etc.), AT commands can be sent with picocom to serial
port /dev/ttyUSB2 [4].

Recovery:
The stock firmware can be recovered via TFTP, as the bootloader (based
on U-Boot) implements a recovery client. Set up a TFTP server on your
computer with IP 192.168.88.1/24 serving the stock firmware from Cudy's
website renamed to "recovery.bin". Press the Reset button (keep it
pressed), power on the device, wait for the TFTP server to send the
recovery.bin file, then release the Reset button. The router will take a
couple of minutes to reboot and set up the stock firmware. See the
device's entry in OpenWrt's Wiki for more details [2].

Discussion:
Use the dedicated thread in OpenWrt's forum to discuss any aspects [5].

Acknowledgments:
The code to support this device is largely based on the sources kindly
provided by Cudy in compliance with the GPL license. Thanks for
supporting the OpenWrt community.

Links:
[1] https://www.cudy.com/blogs/faq/openwrt-software-download
[2] https://openwrt.org/toh/cudy/lt400e_v1
[3] https://openwrt.org/docs/guide-user/network/wan/wwan/ethernetoverusb_cdc
[4] https://openwrt.org/docs/guide-user/network/wan/wwan/at_commands
[5] https://forum.openwrt.org/t/242466
[6] https://drive.google.com/drive/folders/1BKVarlwlNxf7uJUtRhuMGUqeCa5KpMnj

Co-authored-by: rogerpueyo <roger.pueyo@guifi.net>
Signed-off-by: Kamil Jońca <kjonca@onet.pl>
Link: https://github.com/openwrt/openwrt/pull/21246
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

kernel: bump 6.12 to 6.12.66

Changelog: https://cdn.kernel.org/pub/linux/kernel/v6.x/ChangeLog-6.12.66

All patches automatically rebased

Build system: x86/64
Build-tested: flogic/glinet_gl-mt6000
Run-tested: flogic/glinet_gl-mt6000

Signed-off-by: John Audia <therealgraysky@proton.me>
Link: https://github.com/openwrt/openwrt/pull/21591
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

mediatek: filogic: add support for Tenda BE12 Pro

This patch adds support for Tenda BE12 Pro.

The stock bootloader requires a specific 16-byte header (Magic: "God1")
to correctly verify and boot the Linux kernel from flash. A new image
command `tenda-mkdualimageheader` is added to generate this header and
append the checksum/size information required for the device to boot
successfully.

Hardware Specification:
- SoC: MediaTek MT7987A (Quad-core ARM Cortex-A53 2.0 GHz)
- RAM: 512 MiB DDR4
- Flash: 128 MiB SPI-NAND
- WiFi: MediaTek MT7992E (WiFi 7)
  - 2.4GHz: b/g/n/ax/be (4x4 MIMO)
  - 5GHz: a/n/ac/ax/be (4x4 MIMO)
- Ethernet:
  - 3x 10/100/1000 Mbps LAN (Airoha AN8855AE Switch)
  - 1x 10/100/1000/2500 Mbps LAN (SoC internal PHY)
  - 1x 10/100/1000/2500 Mbps WAN (Airoha EN8811H PHY)
- Buttons: Reset, WPS
- LEDs: 2x (Blue: System/Run, Red: Alarm)
- UART: 115200 8n1 (VCC, RX, TX, GND)

Flash Layout:
- 0x000000 - 0x300000 : Bootloader
- 0x300000 - 0x380000 : u-boot-env
- 0x380000 - 0x780000 : Factory
- 0x780000 - 0xd80000 : kernel
- 0xd80000 - 0x6780000 : ubi (rootfs)
- 0x6780000 - 0x6b80000 : CFG
- 0x6b80000 - 0x6f80000 : MISC2

MAC Addresses (NVMEM):
- Base MAC located at Factory partition offset 0x4
- gmac2 (WAN) : Base - 2 (Label MAC)
- gmac0 (LAN) : Base - 1
- gmac1 (LAN) : Base - 3
- 2.4 GHz     : Base
- 5 GHz       : Base + 1

Installation:
1. Connect via UART and interrupt U-Boot by pressing 'Ctrl+C' during boot.
2. Set up a TFTP server on a computer (IP 192.168.1.2) containing
   'openwrt-initramfs.bin'.
3. Load the initramfs image:
   tftpboot openwrt-initramfs.bin
4. Boot the image:
   bootm
5. Once booted into OpenWrt initramfs, perform a sysupgrade with the
   sysupgrade image.

Signed-off-by: Chen Minqiang <ptpt52@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/21461
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

ucode: enable ucode-mod-zlib

The module exists in ucode and has been present for a while
but has not been enabled. It provides the ucode zlib module
for handling gzip and zlib compression in ucode scripts.

The package is ~ 5Kb. Installed ~18Kb.

Signed-off-by: Paul Donald <newtwen+github@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/21417
Signed-off-by: Robert Marko <robimarko@gmail.com>

kernel: net: phy: realtek: backport upstream commits

Backport a bunch of upstream commits simplifying the RealTek PHY driver
and re-unifying the C22 and C45 driver instances.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>

rockchip: add HINLINK H28K support

Hardware:
- RK3528 SoC
- 1x USB 2.0
- 8/32GB eMMC
- 1/2/4GB LPDDR4
- 1x 1GbE RTL8111H
- 1x 1GbE RTL8211F
- MicroSD card slot

Installation:
  Use dd or balenaEtcher to flash the firmware.

Note:
  The RK3528 SoC does not support MSI-X, so it can't
  adjust the interrupts of r8169 to other CPU cores.

Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>

uboot-rockchip: add HINLINK H28K support

Add pending uboot support for the HINLINK H28K.

Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>

rockchip: correct MAC address for Radxa E20C

Use the unique MAC address in EEPROM for eth0.
The MAC address of RTL8111H (eth1) is correct.

Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>

rockchip: configure WAN port LED of Radxa E20C

The WAN(eth1) port LED on the Radxa E20C is completely off.
Since the r8169 driver supports configuring LEDs, using it.

Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>

generic: restore network port name for r8169 LED

Currently, the LED names generated by the r8169 driver are the
network port names renamed by udev, such as enp1s0. Add a hack
patch to restore the interface name to match OpenWrt.

Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>

x86: cs5535audio: add support for subtarget legacy

Build kernel module for cs5535 internal audio for target x86 "legacy" also
(so far only for target x86 "geode")

Signed-off-by: Pierre-Emmanuel Frot <piemfrot@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/21596
Signed-off-by: Robert Marko <robimarko@gmail.com>

mediatek: fix gmac definition for cudy m3000

The gmac definition has an offset of 1 compared to the label MAC.
Fixes: a55ab9e1343e85021253e9c55f67adf33ad68f09 ("mediatek: filogic: prevent faulty mac address assignment")
Signed-off-by: Maximilian Baumgartner <aufhaxer@googlemail.com>
Link: https://github.com/openwrt/openwrt/pull/21560
Signed-off-by: Robert Marko <robimarko@gmail.com>

generic: filter out more rust related symbols

More rust symbols that are detected using scripts or based on the rust
version have been added. Filter them out.

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
Link: https://github.com/openwrt/openwrt/pull/21554
Signed-off-by: Robert Marko <robimarko@gmail.com>

realtek: pcs: rtl930x: rework SerDes mode setting

Rework the SerDes mode setting to have a better logical structure,
similar to how RTL931x currently does it. Mode setting is grouped into a
MAC setup and IP core setup. Most modes just need to trigger the MAC
setup and then they work, otherwise some complex sequence is needed.

To achieve that, rename several functions and group their content
differently. While at it, rename some constants to use the RTPCS_ prefix.

Signed-off-by: Jonas Jelonek <jelonek.jonas@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/21565
Signed-off-by: Robert Marko <robimarko@gmail.com>

realtek: pcs: rtl930x: rename internal mode helpers

RTL930x has some helpers which take care of writing the actual mode
value to the corresponding register. They exist for the MAC setup SerDes
mode, the USXGMII submode and the SerDes IP core mode.

To reflect that, adjust the naming of these helpers accordingly. Most
importantly, prefix the helper names with '__'. This is a common
convention to denote helpers/functions which are internal, not intended
for general use and skip certain checks. Those conditions apply to these
helpers. Though they are still used quite often here, this might change
with further cleanups.

Signed-off-by: Jonas Jelonek <jelonek.jonas@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/21565
Signed-off-by: Robert Marko <robimarko@gmail.com>

realtek: pcs: rtl930x: use SerDes hw mode in calibration

Completely replace usage of PHY_INTERFACE_MODE_* in the vast amount of
calibration functions with the SerDes hardware mode.

Signed-off-by: Jonas Jelonek <jelonek.jonas@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/21565
Signed-off-by: Robert Marko <robimarko@gmail.com>

realtek: pcs: rtl930x: use SerDes hw mode for mode setting

Set and configure the SerDes mode using the SerDes hardware mode types
instead of PHY_INTERFACE_MODE_*.

Signed-off-by: Jonas Jelonek <jelonek.jonas@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/21565
Signed-off-by: Robert Marko <robimarko@gmail.com>

realtek: pcs: rtl930x: use SerDes hw mode in TX config

Use SerDes hardware mode in RTL930x SerDes TX config instead of
PHY_INTERFACE_MODE_* modes.

Signed-off-by: Jonas Jelonek <jelonek.jonas@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/21565
Signed-off-by: Robert Marko <robimarko@gmail.com>

realtek: pcs: rtl930x: use SerDes hw mode in patching

Switch to using the SerDes hardware mode in RTL930x SerDes patching aka
rtpcs_930x_sds_patch.

Signed-off-by: Jonas Jelonek <jelonek.jonas@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/21565
Signed-off-by: Robert Marko <robimarko@gmail.com>

realtek: pcs: rtl930x: use SerDes hw mode in setup entry

Start using the SerDes hardware in the RTL930x SerDes setup entrypoint.
Use the generic mapper (which will be moved out later to pcs_config) to
determine the mode. In the next steps, switch to solely using that
step-by-step.

Signed-off-by: Jonas Jelonek <jelonek.jonas@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/21565
Signed-off-by: Robert Marko <robimarko@gmail.com>

realtek: eth: adapt counter() and irq() functions

Convert the update_counter() and net_irq() functions to the
new prefix. For better readability rename "cntr" to "counter".

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21569
Signed-off-by: Robert Marko <robimarko@gmail.com>

realtek: eth: use device specifc hardware reset

The hardware reset is a combined function for all device families.
So it uses several family specific branches to decide what is
really needed. Refactor this into device specific variants to make
it easier to understand.

The cpu port and nic reset sequence is the same for all families.
Provide this with a dedicated function.

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21569
Signed-off-by: Robert Marko <robimarko@gmail.com>

realtek: image: switch cameo devices from gzip to lzma

The cameo based RTL83xx devices have a simple U-Boot that only
supports gzip compression. Actual models are mainly D-Link
DGS-1210 and the Apresia Light GS120GT-SS. Reduce the image
size by ~1MB by switching over to lzma compression and rt-loader.

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21577
Signed-off-by: Robert Marko <robimarko@gmail.com>

qualcommax: ipq60xx: add JDCloud RE-CS-07 support

Specifications:
  SoC:     Qualcomm IPQ6010 1.8GHz
  RAM:     K4A8G165WC-BCTD x2 2GiB
  ETH:     QCA8075 (3x LAN, 1x WAN)
  Flash:   ICMAX 8GB eMMC
  Power:   DC 12V 2A
  USB:     1x 3.0
  Button:  Reset

Flash instructions:
  1. Download the initramfs image, rename it to
     initramfs.itb, host it with the tftp server.
  2. Enter "jdmt018R" to interrupt U-Boot
     when the following statement appears:
     "disabled console and autoboot in 2 seconds"
  3. Run these commands in U-Boot console:
     tftpboot initramfs.itb
     bootm
  4. After openwrt boots up, use scp or luci web
     to upload sysupgrade.bin to upgrade.

Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>
Link: https://github.com/openwrt/openwrt/pull/19840
Signed-off-by: Robert Marko <robimarko@gmail.com>

qualcommax: ipq60xx: add JDCloud RE-CS-02 support

Specifications:
  SoC:     Qualcomm IPQ6010 1.8GHz
  RAM:     K4A4G165WF-BCTD x2 1GiB
  Flash:   FORESEE 128GB/256GB eMMC
  ETH:     4x 1Gbps + 1x 2.5Gbps
  WLAN1:   QCN5022 2.4GHz AX 2x2
  WLAN2:   QCN5052 5.8GHz AX 2x2
  WLAN2:   QCN9024 5.2GHz AX 4x4
  Button:  Reset, Screen, Wps
  Power:   DC 12V 3A
  USB:     1x 3.0

Flash instructions:
  1. Download the initramfs image, rename it to
     initramfs.itb, host it with the tftp server.
  2. If the stock firmware version is 4.3.0.r4211
     or higher, enter "jdmt018R" to interrupt
     U-Boot when the following statement appears:
     "disabled console and autoboot in 2 seconds"
  3. Run these commands in U-Boot console:
     tftpboot initramfs.itb
     bootm
  4. After openwrt boots up, use scp or luci web
     to upload sysupgrade.bin to upgrade.

Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>
Link: https://github.com/openwrt/openwrt/pull/19840
Signed-off-by: Robert Marko <robimarko@gmail.com>

qualcommax: ipq60xx: add JDCloud RE-SS-01 support

Specifications:
  SoC:     Qualcomm IPQ6000 1.2GHz
  RAM:     NT52CB256MB16DP 512MiB
  Flash:   FORESEE 64GB/128GB eMMC
  ETH:     QCA8075 (3x LAN, 1x WAN)
  WLAN1:   QCN5022 2.4GHz AX 2x2
  WLAN2:   QCN5052 5GHz AX 2x2
  Power:   DC 12V 2A
  Button:  Reset, Wps
  USB:     1x 3.0

Flash instructions:
  1. Download the initramfs image, rename it to
     initramfs.itb, host it with the tftp server.
  2. If the stock firmware version is 4.3.0.r4211
     or higher, enter "jdmt018R" to interrupt
     U-Boot when the following statement appears:
     "disabled console and autoboot in 2 seconds"
  3. Run these commands in U-Boot console:
     tftpboot initramfs.itb
     bootm
  4. After openwrt boots up, use scp or luci web
     to upload sysupgrade.bin to upgrade.

Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>
Link: https://github.com/openwrt/openwrt/pull/19780
Signed-off-by: Robert Marko <robimarko@gmail.com>

qualcommax: ipq50xx: match QDSK ports unit address with port id

Match the unit address of both QSDK switch ports with the node's port_id
property. Purely cleanup, no functional change.

Signed-off-by: George Moussalem <george.moussalem@outlook.com>
Link: https://github.com/openwrt/openwrt/pull/21456
Signed-off-by: Robert Marko <robimarko@gmail.com>

realtek: imagebuilder: add guard around loader generation

The imagebuilder cannot compile source, so we must guard against
generation of artifact targets that require this.  Without this
guard we see an error when building the zyxel,gs1920-24hp-v1 profile.

  Create standalone rt-loader, loading uimage from address 0xb40c0000
  mips-openwrt-linux-musl-gcc -fpic -msoft-float -Iinclude -c -o ... src/startup.S
  make[4]: mips-openwrt-linux-musl-gcc: No such file or directory

Signed-off-by: Eric Fahlgren <ericfahlgren@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/21556
Signed-off-by: Robert Marko <robimarko@gmail.com>

ci: rework and move build-on-comment action

Run this only when a PR is created, so either author or anyone who can
edit the PR message (all maintainers) can trigger such build.

Instead of maintaining it in the main repository, move it to the
external repository to bloat this repository less with further CI
commits.

Signed-off-by: Paul Spooren <mail@aparcar.org>

ramips: fix gmac definition for cudy ap1300 outdoor

The gmac definition has an offset of 1 at the moment. This leads to an
off by one error in downstream projects that rely on the package label mac.

Signed-off-by: Steffen Förster <nemesis@chemnitz.freifunk.net>
Link: https://github.com/openwrt/openwrt/pull/21543
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

realtek: eth: convert prefix of net_device_ops

Rename some of the net_device_ops to the new driver prefix.

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21476
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

realtek: eth: adapt prefix of tx functions

Rename transmit related functions with the new prefix.

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21476
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

realtek: eth: adapt prefix for mac functions

Convert the mac functions to the new prefix.

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21476
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

realtek: eth: convert tag decoder/encoder to new prefix

Rename the header tag encoding/decoding functions according
to the new prefix naming convention.

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21476
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

realtek: eth: rename priv to ctrl (as in other drivers)

All Realtek drivers use ctrl (control) als the central driver
structure. Align that for the ethernet driver. Additionally
use the rteth prefix for the device independent structure
definition. Its current prefix "rtl838x_" is only confusing.

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21476
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

wifi-scripts: fix wifi failure on specifying value of DAE client

The code to be replaced is a glorious no-op. A default value for
config.radius_das_client does not need to be assigned. This parameter
already has non-empty value: see the enclosing 'if' block.

As a result, the value of config.radius_das_client never gets modified
to contain both dae_client and dae_secret. This breaks hostapd.add_iface()
that expects config.radius_das_client to contain both dae_client and
dae_secret separated by a whitespace.

Fixes: #21519
Signed-off-by: Val Kulkov <val.kulkov@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/21522
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

uboot-rockchip: bump to 2026.01

Drop upstreamed patches, refresh others.

Tested on Radxa ROCK 5B+.

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>

arm-trusted-firmware-rockchip: bump to 2.14.0

Tested on Radxa ROCK 5B+.

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>

realtek: rt-loader: fix chip revision printout

There is currently a mismatch in the detection of the chip type.
rt-loader and the kernel give different revisions. E.g.

rt-loader: Running on RTL9313 (chip id 6567A) with 256MB
kernel: Realtek RTL9313 rev B (6567) SoC with 256 MB

Realtek internal version numbering is

- for RTL838x: 1=A, 2=B, ...
- for others: 0=A, 1=B, ...

rt-loader does not differentiate that. Adapt the calculation to
give a consistent picture.

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21498
Signed-off-by: Robert Marko <robimarko@gmail.com>

realtek: phy: simplify RTL8214FC configuration

Initialization of the RTL8214FC was coded like "put the cart
before the horse". Configuration was called from probing
and only when called for the last port.

Testing showed that there is no need to overcomplicate things.
Reorganize the setup as follows:

- Let probe() & config() do what they are meant for
- Split the config part between a package and a port sequence

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21508
Signed-off-by: Robert Marko <robimarko@gmail.com>

realtek: phy: drop unused firmware setup for RTL8214FC

The RTL8214FC is initialized with a minimum driver sequence.
The firmware usage has been made obsolete. Drop leftovers.
These include:

- old code
- firmware file
- kernel config (to include the firmware file into the build)

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21508
Signed-off-by: Robert Marko <robimarko@gmail.com>

realtek: phy: Use single initialization for RTL8214FC

There are currently two code paths to initialize the RTL8214FC. One is
for RTL838x, the other for RTL839x. With the recently added minimum
setup sequence the phy will come up and pass traffic on any device.
Use this als the reference implementation and drop the usage of the
magic firmware based setup.

Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21508
Signed-off-by: Robert Marko <robimarko@gmail.com>

realtek: dts: fix Zyxel GS1920 port section

For the GS1920 the build system throws errors like

../dts/rtl8392_zyxel_gs1920-24hp-v1.dts:256.19-29:
Warning (reg_format): /switch@1b000000/ports/port@0:reg:
property has invalid length (4 bytes)
(#address-cells == 2, #size-cells == 1)

The dts misses the address and size properties for the
ports section. Fix that.

Fixes: 2a55846 ("realtek: add support for ZyXEL GS1920-24HPv1")
Signed-off-by: Markus Stockhausen <markus.stockhausen@gmx.de>
Link: https://github.com/openwrt/openwrt/pull/21534
Signed-off-by: Robert Marko <robimarko@gmail.com>

kernel: bump 6.12 to 6.12.65

Changelog: https://cdn.kernel.org/pub/linux/kernel/v6.x/ChangeLog-6.12.65

Manually rebased:
  generic/hack-6.12/904-debloat_dma_buf.patch
  econet/patches-6.12/010-v6.16-MAINTAINERS-Add-entry-for-newly-added-EcoNet-platfor.patch

Build system: x86/64
Build-tested: flogic/glinet_gl-mt6000
Run-tested: flogic/glinet_gl-mt6000

Signed-off-by: John Audia <therealgraysky@proton.me>
Link: https://github.com/openwrt/openwrt/pull/21431
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

kernel: bump 6.12 to 6.12.64

Changelog: https://cdn.kernel.org/pub/linux/kernel/v6.x/ChangeLog-6.12.64

Manually rebased:
  bcm27xx/patches-6.12/950-0019-BCM2708-Add-core-Device-Tree-support.patch
  bcm27xx/patches-6.12/950-0446-i2c-designware-Add-support-for-bus-clear-feature.patch
  bcm27xx/patches-6.12/950-0875-PCI-brcmstb-Add-bcm2712-support.patch
  ipq40xx/patches-6.12/401-mmc-sdhci-msm-comment-unused-sdhci_msm_set_clock.patch

Removed upstreamed:
  generic-backport/403-v6.19-mtd-mtdpart-ignore-error-ENOENT-from-parsers-on-subp.patch[1]
  bcm27xx/patches-6.12/950-0873-PCI-brcmstb-Reuse-config-structure.patch[2]
  bcm27xx/patches-6.12/950-0888-PCI-brcmstb-don-t-use-ASPM-state-defines-for-registe.patch[3]
  bcm27xx/patches-6.12/950-0951-media-i2c-imx219-Restore-the-1920x1080-to-using-a-1-.patch[4]

All other patches automatically rebased.

1. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.12.64&id=846ceb1a94460f7331e879c02ff2dd505435b51b
2. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.12.64&id=8d185636a6299ff9d2e9eec3a4a25026c13d2351
3. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.12.64&id=38aa6ca6285ff76a7570e5b9acd1151f5cea783a
4. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.12.64&id=620f9d7bcf771b532bce67ffecf2d97759a4747f

Build system: x86/64
Build-tested: flogic/glinet_gl-mt6000, ramips/tplink_archer-a6-v3
Run-tested: flogic/glinet_gl-mt6000, ramips/tplink_archer-a6-v3

Signed-off-by: John Audia <therealgraysky@proton.me>
Co-Authored-by: Mieczyslaw Nalewaj <namiltd@yahoo.com>
Link: https://github.com/openwrt/openwrt/pull/21431
[Fixed: 950-0019-BCM2708-Add-core-Device-Tree-support.patch]
[remove 950-0888-PCI-brcmstb-don-t-use-ASPM-state-defines-for-registe.patch]
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

bcm27xx: refresh kernel patches

The patch 950-0019-BCM2708-Add-core-Device-Tree-support.patch creates a
symbolic link. It looks like quilt does not support this and never
refreshed it.

I removed the symbolic link, refreshed the patch and then added it
back again.

Link: https://github.com/openwrt/openwrt/pull/21431
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

mac80211: fix build for 6.12.64 bump

Disable the backport shim on 6.12.64 and newer where the function
is provided by the kernel.

References:
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/include/linux/hrtimer.h?h=v6.12.64&id=12493e7e888d266453d1c3898dc160178749afb3
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/include/linux/hrtimer.h?h=v6.12.64&id=d579cc549298a02ba90b7a6d20a2ba8e160f3f31

Co-Authored-by: Pascal Ernster <git@hardfalcon.net>
Signed-off-by: John Audia <therealgraysky@proton.me>
Link: https://github.com/openwrt/openwrt/pull/21431
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

netifd: use stable IAID for DHCPv4

Commit 9151c7015ed2 introduced support for the global DHCP DUID to
generate a RFC4361-style client identifier.
However, the IAID introduced in those changes is based on ifindex, which
is subject to changes and causes issues on environments requiring a stable
IAID.

This commit switches the IAID to a stable one based on MD5.

Fixes: 9151c7015ed2 ("netifd: use the global DHCP DUID for DHCPv4")
Link: https://github.com/openwrt/openwrt/pull/21489
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

base-files: introduce a function to generate IAID

Add new function "network_generate_iface_iaid()" to generate a stable IAID
from an interface name.

Link: https://github.com/openwrt/openwrt/pull/21489
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

airoha: show link rate and duplex

Implement the .get_link_ksettings to get the rate, duplex, and
auto-negotiation status.

Signed-off-by: Aleksander Jan Bajkowski <olek2@wp.pl>
Link: https://github.com/openwrt/openwrt/pull/21530
Signed-off-by: Robert Marko <robimarko@gmail.com>

rockchip: add `"compat_version": "1.1"` for Radxa E52C

This is something I missed in https://github.com/openwrt/openwrt/pull/20608

Fixes: 1f1db75432 ("rockchip: make NIC name predictable for Radxa E52C/ROCK 5 ITX/ROCK 5T")
Signed-off-by: FUKAUMI Naoki <naoki@radxa.com>
Link: https://github.com/openwrt/openwrt/pull/21533
Signed-off-by: Robert Marko <robimarko@gmail.com>

kernel: add kmod-hwmon-adt7470

This driver supports the ADT7470 thermal monitoring chip, which is used
in the ECS4100-12PH switch.

Fixes: fa9f92595197 ("realtek/rtl839x: Edgecore ECS4100-12PH support")
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
Link: https://github.com/openwrt/openwrt/pull/21505
Signed-off-by: Robert Marko <robimarko@gmail.com>

libiwinfo: update to Git HEAD (2026-01-14)

f5dd57a84cc3 nl80211: fix 5Ghz frequency calculations

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

mpc85xx: fix failsafe iface selection for mpc85xx boards

Some mpc85xx boards still boot with failsafe configured on a non-LAN
interface. Align the preinit interface with the first DSA port or the
interface that LAN is connected to.

This makes failsafe reachable on devices where the default selection does
not map to the primary LAN port.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/21516
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>