kernel: backport upstream GSO improvements

Fixes some corner cases regarding segmenting packets that were assembled
by GRO.

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

ramips: mt76x8: add support for Yuncore M300

Hardware:
 - SoC: MediaTek MT7628DAN (MIPS 580MHz)
 - Flash: 8 MiB Spansion S25FL064K
 - RAM: 64 MiB (built-into SoC)
 - WLAN: 2.4 GHz (MT7628)
 - Ethernet: 1x 10/100 Mbps WAN, 1x 10/100 LAN (MT7628)
 - Buttons: 1 Reset button
 - LEDs: 1x Red, 1x Green
 - Serial console: unpopulated header, 57600 8n1 (RX only)
 - Power: 12 VDC, 1 A

There are unpopulated areas on the board for 5 GHz WiFi via PCIe as well
as (most likely) Quectel EG25-G 4G module. As both are not populated on
my board support for both is missing for now.

Installation:
The installation can be done via the recovery HTTP server which is built
into the bootloader. Hold down the reset button while connecting the
device to power and keep holding a bit more than 3 seconds. Connect to
http://192.168.188.253/ and upload sysupgrade.bin file.

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

kernel: Fix section mismatch in ubi

Fix ubiblock_create_from_param() ubiblock_create_auto_rootfs section mismatch.
Without this, the system upgrade will not work if the kernel was compiled with clang-18.

WARNING: modpost: vmlinux: section mismatch in reference: ubiblock_notify+0x190 (section: .text) -> ubiblock_create_auto_rootfs (section: .init.text)

```
[33342.080771] Call trace:
[33342.083205]  ubiblock_create_auto_rootfs+0x0/0xd0
[33342.087902]  blocking_notifier_call_chain+0xb0/0x1a0
[33342.092857]  ubi_volume_notify+0xcc/0xdc
[33342.096773]  ubi_create_volume+0x520/0x684
[33342.100861]  ubi_cdev_ioctl+0x2ac/0x834
[33342.104688]  __arm64_sys_ioctl+0x14f0/0x15f4
[33342.108947]  invoke_syscall+0x44/0xc8
[33342.112601]  do_el0_svc+0x78/0xa8
[33342.115907]  el0_svc+0x24/0x50
[33342.118951]  el0t_64_sync_handler+0x88/0xf0
[33342.123126]  el0t_64_sync+0x150/0x154
[33342.126784] Code: ???????? ???????? ???????? ???????? (????????)
```
Signed-off-by: Romanov Danila <pervokur@gmail.com>

config: kernel: Add support for configuring BTRFS to be built-in

Add the KERNEL_BTRFS_FS config option so that targets can select
whether BTRFS support must be built-in.

Select this option (alongside KERNEL_BTRFS_FS_POSIX_ACL) from the
layerscape/armv8_64b subtarget instead of enabling it in
target/linux/layerscape/armv8_64b/config-* files.

Move disabling of CONFIG_BTRFS_FS_CHECK_INTEGRITY into generic configs.

This makes it possible for OpenWRT to be built with built-in BTRFS
support on specific boards, instead of whole targets.

Signed-off-by: Marek Behún <kabel@kernel.org>
Link: https://github.com/openwrt/openwrt/pull/15990
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

mediatek: Add support for Linksys EA7500 v3

Specification:
- MT7629 CPU
- MT7531 switch
- MT7761N and MT7762N wifi
- 256 MB RAM
- 128 MB NAND flash with dual-boot partitions
- 2 buttons: WPS and reset
- 1 WAN port (1G)
- 4 LAN ports (1G)
- 1 USB port

Limitations (same as other MT7629/MT7761N/MT7762N devices):
- Wifi is not working
- Second core is not working (kernel error message "CPU1: failed to come online")

Disassembly:
- There are two screws under the front rubber feet and two under the label on the bottom (in the corners towards the back, you should be able to feel them).

Serial Interface:
- UART pin header is already soldered on the board. Pinning from front to back:
1 - VCC
2 - TX
3 - RX
4 - n/a
5 - GND

GPIO:
- 1 white LED, connected to GPIO 52
- 1 reset button, connected to GPIO 60
- 1 WPS button, connected to GPIO 58

MAC Adresses:
- The MAC address printed on the device label is used for LAN and WAN
- The MAC address is stored in the devinfo partition in ASCII format (hw_mac_addr=aa:bb:cc:dd:ee)
- 2.4 GHz wifi uses MAC of the device label + 1
- 5 GHz wifi uses MAC of the device label + 2

Flashing:
- OpenWrt is only runnig in the first partition of dual boot
- To ensure to be able to go back to the factory image, flash the last OEM firmware via OEM web interface. This will ensure that the OEM firmware is present on both partitions
- Because of dual boot partitions, flashing via OEM interface is not supported
- Start a TFTP server and provide the initramfs image. Default settings:
  - Router IP: 192.168.1.1
  - TFTP server IP: 192.168.1.100
  - TFTP file name: 7531.bin
- Open the device, connect UART and select " 1. System Load Linux to SDRAM via TFTP." during startup
- Adapt the settings to your environment, if required
- After initramfs is booted, flash the sysupgrade image

Return to OEM firmware:
- Run the following commands in OpenWrt to switch to the second partition
  fw_setenv boot_part 2
  fw_setenv bootimage 2
- Reboot the device. OEM firmware will start up again

Signed-off-by: Roland Reinl <reinlroland+github@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16067
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

rockchip: fix sysupgrade for Radxa ROCK Pi S

SUPPORTED_DEVICES is required for Radxa ROCK Pi S.

Signed-off-by: FUKAUMI Naoki <naoki@radxa.com>
Link: https://github.com/openwrt/openwrt/pull/16167
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

mac80211: add rtw89 driver

This commits adds the RTW89 driver from Realtek.
Supports the Realtek 8851BE/8852AE/8852BE/8852CE PCIe wireless chips.

Signed-off-by: Antonio Flores <antflores627@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16131
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

bmips: inteno-xg6846: Add DSA LED definitions

This adds the LED definitons for the XG6846 DSA port LEDs.
These are standard properties compatible with the existing
Marvell 88e6xxx DT bindings and fully standardized so this
is fine to add. They will be used by the in-flight Marvell
88e6xxx LEDs support patch.

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
[add empty lines between leds, remove default-state="off"]
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

bmips: inteno-xg6846: Add EEPROM node

For some troublesome devices it is necessary to obtain direct access
to the SFP module EEPROM so define it in the device tree.

Suggested-by: Paul Donald <newtwen@gmail.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

bmips: inteno-xg6846: Activate USB port

Some versions of the Inteno XG6846 has a USB port mounted.
For these machines the corresponing USB port nodes need to
be enabled.

Suggested-by: Henrik Ginstmark <henrik@ginstmark.se>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
[reorder DTS alphabetically]
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

uhttpd: Decrease the default validity time of certificate

The recommended maximum validity period is currently 397 days
and some browsers throw warning with longer periods.

Reference to
https://cabforum.org/working-groups/server/baseline-requirements/
 6.3.2 Certificate operational periods and key pair usage periods
 Subscriber Certificates issued on or after 1 September 2020
 SHOULD NOT have a Validity Period greater than 397 days and
 MUST NOT have a Validity Period greater than 398 days.

Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
Link: https://github.com/openwrt/openwrt/pull/15366
Signed-off-by: Robert Marko <robimarko@gmail.com>

uhttpd: Include new extensions in uhttpd self-signed certs

The introduction of MacOS Catalina includes new requirements for self-signed certificates.
See: https://support.apple.com/en-us/HT210176
These new requirements include the addition of two TLS server certificate extensions.
- extendedKeyUsage
- subjectAltName
The extendedKeyUsage must be set to serverAuth.
The subjectAltName must be set to the DNS name of the server.
In the absense of these new extensions, when the LUCI web interface is configured to use HTTPS and
self-signed certs, MacOS user running Google Chrome browsers will not be able to access the LUCI web enterface.
If you are generating self-signed certs which do not include that extension, Chrome will
report "NET::ERR_CERT_INVALID" instead of "NET::ERR_CERT_AUTHORITY_INVALID".  You can click through to
ignore the latter, but not the former.

This change updates the uhttpd init script to generate self-signed cert that meets the new requirements.
Signed-off-by: Pat Fruth <pat@patfruth.com>
Link: https://github.com/openwrt/openwrt/pull/15366
Signed-off-by: Robert Marko <robimarko@gmail.com>

px5g-mbedtls: add subjectAltName and extendedKeyUsage to SSL certs

To better acommodate with the current browsers' requirements, also
self-signed certificates should have subjectAltName and
extendedKeyUsage defined in the self-signed x509 SSL certificates.

The following case sensitive options are now possible:
-addext subjectAltName=DNS:...
-addext subjectAltName=EMAIL:...
-addext subjectAltName=IP:...
-addext subjectAltName=URI:...
-addext extendedKeyUsage=serverAuth OR -addext extendedKeyUsage=any

Initial draft by Paul Donald <newtwen@gmail.com>

Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
Link: https://github.com/openwrt/openwrt/pull/15366
Signed-off-by: Robert Marko <robimarko@gmail.com>

firmware: omnia-mcu-firmware: Bump to 4.1

Bump `omnia-mcu-firmware` to version 4.1.

This version fixes the following issue on boards with GD32 MCU:
* the user has old GD32 MCU bootloader and application (version 2.0)
* the user upgraded MCU application firmware to newer version (from
  2.99 to 4.0)
* the user wants to upgrade application again, but it is impossible,
  because when MCU application firmware jumps into the old MCU
  bootloader firmware (2.0), the old bootloader firmware gets stuck in
  exception
* the user has to restart the board and upgrade the bootloader firmware
  first, which is not ideal, since if bootloader firmware upgrade is
  interrupted, the board gets bricked

Therefore the `omnia-mcutool` utility version 0.3-rc3 will refuse to
upgrade MCU application firmware to versions 2.99 to 4.0 if the MCU
bootloader firmware is at version 2.0.

For users to be able to upgrade MCU application firmware on GD32
boards, they will need this new 4.1 version.

Users that already upgraded the MCU application firmware to a version
version between 2.99 and 4.0 (using a previous version of the
`omnia-mcutool` utility) have no other choice but to upgrade MCU
bootloader firmware as well.

Signed-off-by: Marek Behún <kabel@kernel.org>
Link: https://github.com/openwrt/openwrt/pull/16159
Signed-off-by: Robert Marko <robimarko@gmail.com>

tools/cmake: Update to 3.30.2

Update cmake to version 3.30.2
Release notes: https://cmake.org/cmake/help/v3.30/release/3.30.html

Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
Link: https://github.com/openwrt/openwrt/pull/16059
Signed-off-by: Robert Marko <robimarko@gmail.com>

mpc85xx: fix wdr4900 ethernet

997acc7f86ca985cba52f7ea8b72f0661a1e3c52 split this PHY driver up such
that external QCA switches now use CONFIG_QCA83XX_PHY. Fix it here so
that ethernet works again.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16154
Signed-off-by: Robert Marko <robimarko@gmail.com>

ramips: add support for Xiaomi R4AC v2 (intl)

The second edition of international version of Mi Router 4A 100M is
very similar to the non-international one, but has another wireless chip.

 Installation
--------------

1. Initialize build-in firmware (use webgui for 192.168.31.1)
  You should install root password

2. Run OpenWRTInvasion for the first time (probably it will fail)
  Version 0.0.10 is working as well as 0.0.1.

3. Run OpenWRTInvasion for the second time
  It will create an access to your router

4. Upload sysupgrade image to router (/tmp/fw.bin)
  pc# nc -l 8080 < …/ramips/mt76x8/…-100m-intl-v2-squashfs-sysupgrade.bin
  router# nc 192.168.31.175 8080 > /tmp/fw.bin

5. Flash new firmware
  router# run mtd -r write /tmp/fw.bin OS1

6. Check result
  Wait about 5-10 minutes after flash. Router should reboot itself and
  turn left led from orange to blue.

In case of failure one can use Xiaomi 4a 100m debrick tool
(it uploads special image via tftpd in recovery mode)
After that you can start again from step 1.

Another actions are very similar to original Mi Router 4A 100M

 Original mtd paritions:
-------------------------

```
Creating 9 MTD partitions on "raspi":
0x000000000000-0x000001000000 : "ALL"
0x000000000000-0x000000020000 : "Bootloader"
0x000000020000-0x000000030000 : "Config"
0x000000030000-0x000000040000 : "Factory"
0x000000040000-0x000000050000 : "crash"
0x000000050000-0x000000060000 : "cfg_bak"
0x000000060000-0x000000160000 : "overlay"
0x000000160000-0x000000dc0000 : "OS1"
0x000000dc0000-0x000001000000 : "disk"
with special sub-partition
0x0000002c0000-0x000000dc0000 : "rootfs"
```

We will use OS1+disk space:
```
0x000000160000-0x000001000000 : "firmware"
```

Co-authored-by: Nita Vesa <nita.vesa@elektrik.link>
Signed-off-by: Anton Stratonnikov <billic@yandex.ru>
Link: https://github.com/openwrt/openwrt/pull/14304
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

kernel: bump 6.1 to 6.1.104

https://cdn.kernel.org/pub/linux/kernel/v6.x/ChangeLog-6.1.104

All patches automatically rebased.

1 mm: restrict the pcp batch scale factor to avoid too long latency
a new kconfig option (PCP_BATCH_SCALE_MAX) is added to
set the max batch scale factor.Whose default value is 5,
and users can reduce it when necessary.

https://lore.kernel.org/all/20231016053002.756205-5-ying.huang@intel.com/T/#u

Build system: bcm4908 bcm53xx

Signed-off-by: Zxl hhyccc <zxlhhy@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16141
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

kernel: bump 6.6 to 6.6.45 and update config-6.6

Build on at leasst x86/64 failed without adding the new ksym:
CONFIG_PCP_BATCH_SCALE_MAX=5

According to www.kernelconfig.io[1], this option seems to apply
to all arches so I placed it in target/linux/generic/config-6.6

Changelog: https://cdn.kernel.org/pub/linux/kernel/v6.x/ChangeLog-6.6.45

Removed upstreamed:
generic/backport-6.6/719-v6.11-net-phy-realtek-add-support-for-RTL8366S-Gigabit-PHY.patch[2]

All other patches automatically rebased.

1. https://www.kernelconfig.io/config_pcp_batch_scale_max?q=&kernelversion=6.10.3&arch=x86
2. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.6.45&id=b45cbfa204b2a0985eb85dcb33d51714ee089bb9

Build system: x86/64
Build-tested: x86/64/AMD Cezanne, flogic/xiaomi_redmi-router-ax6000-ubootmod, ramips/tplink_archer-a6-v3
Run-tested: x86/64/AMD Cezanne, flogic/xiaomi_redmi-router-ax6000-ubootmod, ramips/tplink_archer-a6-v3

Signed-off-by: John Audia <therealgraysky@proton.me>
Link: https://github.com/openwrt/openwrt/pull/16144
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

toolchain: gcc: update GCC 14 to 14.2.0

All patches automatically rebased.

Signed-off-by: Thomas Weißschuh <thomas@t-8ch.de>
Link: https://github.com/openwrt/openwrt/pull/16047
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

hostapd: Add support for APuP

Add support for hostapd Access Point Micro Peering

Signed-off-by: Gioacchino Mazzurco <gio@polymathes.cc>
Link: https://gitlab.com/g10h4ck/hostap/-/commits/APuP
Link: https://github.com/openwrt/openwrt/pull/15442
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

lantiq: arv7525pw: use nvmem for eeprom

NVMEM is the upstream replacement for this. ralink,mtd-eeprom is
deprecated. The others need to stay as there's byte swapping going on.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16084
Signed-off-by: Robert Marko <robimarko@gmail.com>

uboot-envtools: Add support for Orange Pi R1 Plus & LTS

Add support this boards to envtools config
This commit integrates the latest changes from new U-Boot, which includes important updates to the DTSI files for the Orange Pi R1 Plus and Orange Pi R1 Plus LTS boards.

Signed-off-by: Vyacheslav Ivanov <islavaivanov76@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16090
Signed-off-by: Robert Marko <robimarko@gmail.com>

imx: drop 6.1 support

Drop config and files for Linux 6.1.

Signed-off-by: Mieczyslaw Nalewaj <namiltd@yahoo.com>
Link: https://github.com/openwrt/openwrt/pull/16107
Signed-off-by: Robert Marko <robimarko@gmail.com>

imx: use kernel 6.6 by default

Switch to Linux kernel version 6.6.

Signed-off-by: Mieczyslaw Nalewaj <namiltd@yahoo.com>
Link: https://github.com/openwrt/openwrt/pull/16107
Signed-off-by: Robert Marko <robimarko@gmail.com>

ath10k-ct: bump to version 6.9

Use ath10k-ct 6.9 to better match mac80211 backports 6.9.x

Drop patch 010 that is merged upstream.
Add patch 001 to fix version to 6.9 (overlooked by upstream).
Refresh patches.

Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
Link: https://github.com/openwrt/openwrt/pull/16036
Signed-off-by: Robert Marko <robimarko@gmail.com>

mediatek: mt7623: get rid of fitblk_get_bootdev

Also migrate mt7623 to new fitblk support scripts which simplify
sysupgrade when using uImage.FIT. This had been forgotten previously.

Fixes: 4448d6325f ("mediatek: make use of common uImage.FIT upgrade functions")
Signed-off-by: Daniel Golle <daniel@makrotopia.org>

base-files: get rid of forgotten traces of fitblk_get_bootdev

The function fitblk_get_bootdev doesn't exist any more, using it in
export_bootdevice anyway never made much sense and only worked for
classic block devices.
Just drop /dev/fit* handling there, it isn't needed anywhere.

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

ramips: mt7621_nand: use clk_get_optional_enabled

Simplifies the code by removing clk_disable_unprepare.

Also removed gotos and used dev_err_probe.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16133
Signed-off-by: Robert Marko <robimarko@gmail.com>

mediatek: fix a use-after-free kernel panic in wed code

Fix a use-after-free bug in mtk_wed_setup_tc_block_cb()
which leads to kernel panic when setup multiple ap
interfaces on one band of mt798x.

Signed-off-by: Zheng Zhang <everything411@qq.com>
Link: https://github.com/openwrt/openwrt/pull/16118
Signed-off-by: Robert Marko <robimarko@gmail.com>

mediatek: add script to trigger scrubbing of FIP-in-UBI

Read the 'fip' static volume in order to trigger scrubbing in case of
detecting flipped bits while reading.
We have to do this in Linux because we never read or touch the 'fip'
volume and the UBISPL implementation in ARM TrustedFirmware-A does NOT
handle scrubbing itself.

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

generic: import patch lowering bitflip_threshold on SPI-NAND

Reporting an unclean read from SPI-NAND only when the maximum number
of correctable bitflip errors has been hit seems a bit late.
UBI LEB scrubbing, which depends on the lower MTD device reporting
correctable bitflips, then only kicks in when it's almost too late.

Set bitflip_threshold to 75% of the ECC strength, which is also the
default for raw NAND.

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

arm-trusted-firmware-mediatek: fix NAND read failure on SNFI

A bug has plagued bl2 which caused failure to boot and bricked Linksys
E8450 and Belkin RT3200 devices in case of correctable bitflips being
detected during a read operation. A simple logic error resulted in read
to be considered errornous instead of just continueing in case of
correctable bitflips.

Address this by importing a patch fixing that logic error.

The issue, which has been dubbed as the "OpenWrt Kiss of Death", and is
now a thing of the past.

Users should preemptively update bl2 to prevent their devices being at
risk.

Link: https://github.com/mtk-openwrt/arm-trusted-firmware/pull/11
Signed-off-by: Daniel Golle <daniel@makrotopia.org>

generic: 6.6: add Winbond W25N01KV support

It seems that some Xiaomi AX3000T boards changed to using Winbond W25N01KV
SPI-NAND which is not supported in OpenWrt nor upstream kernel.

So, add a pending patch to support it as upstream supports rest of the KV
revision models.

Fixes: #16002
Link: https://github.com/openwrt/openwrt/pull/16088
Signed-off-by: Robert Marko <robimarko@gmail.com>

rockchip: add led and network config to nanopi r6s

We need to configure the led and network config for this board on
start as per the others

Signed-off-by: Ben Whitten <ben.whitten@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15607
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

rockchip: add NanoPi R6S support

Add patches for the nanopi r6s board, backporting from 6.9 where
basic support is landing.

Signed-off-by: Ben Whitten <ben.whitten@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15607
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

uboot-rockchip: add nanopi r6s rk3588s board support

Backporting support for the NanoPi R6S from upstream
uboot.

Signed-off-by: Ben Whitten <ben.whitten@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15607
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

uboot-rockchip: backport upstream dts sync

Upstream uboot have merged in kernel dts files, we need
the update for the rk3588 boards.

Signed-off-by: Ben Whitten <ben.whitten@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15607
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

rkbin: add rk3588 support

Adding support for the rk3588 platform

Signed-off-by: Ben Whitten <ben.whitten@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15607
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

kernel: r8126: add RSS variant

Instead of enabling RSS support, let's introduce a variant and let users
choose between both variants since it can cause network issues.

Signed-off-by: Milinda Brantini <C_A_T_T_E_R_Y@outlook.com>
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

kernel: r8125: add RSS variant

Instead of enabling RSS support, let's introduce a variant and let users
choose between both variants since it can cause network issues.

Signed-off-by: Milinda Brantini <C_A_T_T_E_R_Y@outlook.com>
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

package: add kmod-r8168 ethernet driver

r8168 is an out of tree driver provided by Realtek for RTL8168 devices.

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

package: add kmod-r8125 ethernet driver

r8125 is an out of tree driver provided by Realtek for RTL8125 devices.

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

package: add kmod-r8126 ethernet driver

r8126 is an out of tree driver provided by Realtek for RTL8126 devices.

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

uboot-envtools: add env settings for ubnt,unifi-6-lr-v3

Using the same configuration as my earlier Unifi 6-LRv2s:

```bash
$ cat /etc/fw_env.config
/dev/mtd3 0x0 0x1000 0x1000 1
$ fw_printenv
arch=arm
baudrate=115200
board=mt7622_evb
board_name=mt7622_evb
bootcmd=bootubnt
bootdelay=3
bootfile=uImage
cpu=armv7
device_model=U6-LR
ethact=mtk_eth
ethaddr=<redacted>
ipaddr=<redacted>
is_ble_stp=true
is_default=true
loadaddr=0x5007FF28
macaddr=<redacted>
serverip=<redacted>
soc=mt7622
stderr=serial
stdin=serial
stdout=serial
vendor=mediatek
```

Signed-off-by: Joel Low <joel@joelsplace.sg>
Link: https://github.com/openwrt/openwrt/pull/16127
Signed-off-by: Robert Marko <robimarko@gmail.com>

mediatek: EAP111: add label-mac-device

Add the label-mac-device alias so that label MAC is set and can later
be used in userspace.

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

sdk: fix APK key creation

The keys are created differently compared to the old OPKG keys. Instead
of being part of base-files/configure, they are created as a Makefile
requirement of `package/compile`, which is a cleaner solution.

This requirement would only be added to non SDK environments, however
APK always requires keys to be available. Add an `else` case for the SDK
and create keys.

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

tools: firmware-utils: update to Git HEAD (2024-08-09)

26c7f05 nec-usbatermfw: add tool for NEC "USB ATERM" format

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

qualcommax: ipq60xx: add support for Cambium Networks XE3-4

Cambium Networks XE3-4 is a tri-radio Wi-Fi 6/6E 4×4/2×2 AP.

Hardware:
    Model:    Cambium Networks XE3-4
    CPU:      IPQ6010/AP-CP01-C3, SoC Version: 1.0 @ 800 MHz
    Memory:   1 GiB
    Flash:    512 MiB Macronix MX30UF2G18AC + W25Q128FW
    Ethernet: 1x 1 GbE   (QCA8072)
              1x 2.5 GbE (QCA8081)
    Buttons:  1x Reset
    Serial:   TX, RX, GND
    Baudrate: 115200
    Radios:   Qualcomm Atheros IPQ6018 802.11ax - 2x2 - 2GHz
              Qualcomm Atheros IPQ6018 802.11ax - 2x2 - 5GHz
              Qualcomm Atheros QCN9074 802.11ax - 4x4 - 5GHz or 6GHz
              BLE 4.1
    Power:    32.0W 802.3bt5 PoE++
              25.5W 802.3at with USB, BT disabled
    Size:     215mm x 215mm
    Ports:    1x USB 2.0
    Antenna:  6 GHz: 6.29 dBi, Omni    30 dBm
              5 GHz: 6.12 dBi, Omni    31 dBm
              2.4 GHz: 4.85 dBi, Omni  29 dBm
    LEDs:     Multi-color status LEDs
    Mounting: Wall, ceiling or T-bar

Installation: Serial connection
1. Open the AP to get access to the board. Connect RX, TX and GND.
2. Power on the AP, and short the CS pin of the SPI flash with
   one of the APs GND pins.
3. Transfer the initramfs image with TFTP
   (Default server IP is 192.168.0.120)
   # tftpboot factory.ubi
4. Flash the rootfs partition
   # flash rootfs
5. Reboot the AP
   # reset

Signed-off-by: Kristian Skramstad <kristian+github@83.no>
Link: https://github.com/openwrt/openwrt/pull/15633
Signed-off-by: Robert Marko <robimarko@gmail.com>

rockchip: rock 3a: fix image check failed

Fixes the image check failed on system upgrade
  "Image check failed:
   upgrade: Device radxa,rock3a not supported by this image
   upgrade: Supported devices: radxa,rock-3a"

Signed-off-by: Tomas Lara <tl849670@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16064
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

mediatek: increase phy assert time for jdcloud re-cp-03

According to RTL8221B's datasheet, the PHY requires at least 10ms
for assert and 68ms (recommended) for de-assert. So increase the
assert/de-assert time to 15ms and 68ms respectively.

Fixes: c0c3234e1720 ("mediatek: add support for JDCloud RE-CP-03")
Signed-off-by: Tianling Shen <cnsztl@immortalwrt.org>
Link: https://github.com/openwrt/openwrt/pull/16106
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

kernel: rtl8367b: restore rlvid definitions

Restore RTL8367B_CHIP_VER_RLVID_SHIFT and RTL8367B_CHIP_VER_RLVID_MASK definitions
removed in commit c30e0eb2a33e.

Signed-off-by: Mieczyslaw Nalewaj <namiltd@yahoo.com>
Link: https://github.com/openwrt/openwrt/pull/16108
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

toolchain: mold: Update to 2.33.0

Update mold to 2.33.0
Link: https://github.com/rui314/mold/releases/tag/v2.33.0
Signed-off-by: Carsten Schuette <schuettecarsten@googlemail.com>
Link: https://github.com/openwrt/openwrt/pull/16048
Signed-off-by: Robert Marko <robimarko@gmail.com>

bmips: refactor Inteno XG6846 device tree

Refactor Inteno XG6846 device tree to be in line with other bmips devices.
Also expose USB LED automatically.

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

bcm27xx: align and enable hwmon/thermal symbols

- Enable CONFIG_HWMON and CONFIG_THERMAL_HWMON on all subtargets.
- Drop kmod-thermal from bcm2712.
- Add CONFIG_SENSORS_RASPBERRYPI_HWMON generic symbol.

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

libunistring: update to 1.2

Release notes:
  https://git.savannah.gnu.org/gitweb/?p=libunistring.git;a=blob_plain;f=NEWS

Signed-off-by: Aleksey Vasilenko <aleksey.vasilenko@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16065
Signed-off-by: Robert Marko <robimarko@gmail.com>

archs38: enable features rootfs-part

target/linux/archs38/image/Makefile calls gen_axs10x_sdcard_img.sh
with $(CONFIG_TARGET_ROOTFS_PARTSIZE).
Make sure a rootfs partition is built and usable.

Signed-off-by: Thomas Weißschuh <thomas@t-8ch.de>
Link: https://github.com/openwrt/openwrt/pull/16098
Signed-off-by: Robert Marko <robimarko@gmail.com>

kernel: mtdtests: update for 6.1 as well

6.1.103 and 6.6.44 introduced breakage complaining about missing mtd_test.ko
for some targets.

Signed-off-by: Zxl hhyccc <zxlhhy@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16093
Signed-off-by: Robert Marko <robimarko@gmail.com>

kernel: bump 6.1 to 6.1.103

https://cdn.kernel.org/pub/linux/kernel/v6.x/ChangeLog-6.1.103

Removed upstreamed:
generic/backport-6.1/412-v6.3-01-spidev-Add-Silicon-Labs-EM3581-device-compatible.patch

See: See: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.1.103&id=c6c419ed94788ca13334b4acd50167b7f6d00a06

All other patches automatically rebased.

Build system: bcm4908 bcm53xx

Signed-off-by: Zxl hhyccc <zxlhhy@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16093
Signed-off-by: Robert Marko <robimarko@gmail.com>

kernel: switch crc32 implementation back to default

Commit ec885796c05a switched the crc32 implementation from default to
byte-at-a-time algorithm, which runs slower but consumes less memory.
A decade has passed, and we have already abandoned targets that had
small memory, so switch it back to default for faster speed.

Signed-off-by: Qingfang Deng <qingfang.deng@siflower.com.cn>

mvebu: GL-MV1000: let u-boot-env be writable again

Allows easily changing boot media for GL-MV1000.

Signed-off-by: Enrico Mioso <mrkiko.rs@gmail.com>

mvebu: enable CONFIG_MTD_SPI_NOR_USE_VARIABLE_ERASE=y config option

Enable the CONFIG_MTD_SPI_NOR_USE_VARIABLE_ERASE kernel option to allow for
U-Boot environment writing. This might be hiding a problem somewhere else,
since the w25q128fw chip supports 32K erases, still this change makes it
much easier to switch the GL-MV1000 boot media without an UART cable
connection.

Thanks to @robimarko and @hacks for the precious hints and suggesting a
better approach.

Signed-off-by: Enrico Mioso <mrkiko.rs@gmail.com>

mvebu: GL-MV1000: add custom boot script

This allows booting from internal eMMC or SD card just changing the
U-Boot mmc_dev variable.
In particular, setting mmc_dev to 1 will result in booting from the SD card.
Setting the variable to 0 will result in internal eMMC boot (the default).
Should the variable be unset or an error condition occur while reading
from SD card, internal MMC booting will be tried.

Signed-off-by: Enrico Mioso <mrkiko.rs@gmail.com>

mpc85xx: remove 6.1 kernel support

mpc85xx was switched to 6.6. We can remove 6.1 support now.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16087
Signed-off-by: Robert Marko <robimarko@gmail.com>

mpc85xx: switch to kernel 6.6

It's time for wide tests.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16087
Signed-off-by: Robert Marko <robimarko@gmail.com>

bcm27xx: update to latest RPi patches

The patches were generated from the RPi repo with the following command:
git format-patch v6.6.44..rpi-6.6.y

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

apk: switch to index-trust branch

Initially APK would sign packages and package index and verify
signatures individually. With the latest change, all packages inside a
trusted index are automatically trusted.

This is important within the OpenWrt eco-system since signing the index
happens on another machine than the package creation.

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

base-files: improve Dell EMC Edge620 (x86) product support

This adds auto-configuration of network ports on Dell EMC Edge620 (x86) product.
It is similar in specs/features to some of the Sophos x86-based appliances, but:

1. Serial console terminal is built in and requires just the micro-USB cable
2. Comes with both MMC (16Gb) and SSD (256Gb) installed
3. Comes with 6 ethernet ports all 6 are functional when no SFP is used
4. Comes with two SFP cages and not one, like some of revision 3 Sophos products
5. Unlike Sophos devices, there are no non-wireless models of Edge 620,
   it comes with Qualcomm Atheros QCA9880 radio

These devices can be now found both second-hand and new at online marketplaces below
(sometimes well below) US $100, I believe they make great candidates for running OpenWrt.

The ethernet network ports on the case are marked GE1 thru to GE6 with the
following mapping once booted into OpenWrt:

```
GE1: eth2: pci0000:00/0000:00:0b.0/0000:02:00.2
GE2: eth3: pci0000:00/0000:00:0b.0/0000:02:00.3
GE3: eth0: pci0000:00/0000:00:0b.0/0000:02:00.0
GE4: eth1: pci0000:00/0000:00:0b.0/0000:02:00.1
GE5: eth7: pci0000:00/0000:00:17.0/0000:07:00.1
GE6: eth6: pci0000:00/0000:00:17.0/0000:07:00.0
```

Dell's instructions for [standard configuration](https://infohub.delltechnologies.com/en-us/l/dell-emc-edge-620-advanced-activation-guide/dell-emc-sd-wan-edge-620-standard-configuration/)
recommend using GE3, GE4, GE5, or GE6 for WAN, I've selected the GE6 as the sole
WAN port under OpenWrt with the rest of ethernet ports assigned to LAN.

Please merge before 24.xx is forked and if possible, cherry-pick for 23.05
if there's no ETA for 24.xx forking.

PS. @Hurricos I'm struggling with ixgbe mappings on Sophos devices which use
very similar hardware to Dell EMC, so even tho I know the sys paths for ethernet ports,
I'd prefer to do a separate commit to properly map ethernet ports to match the case markings
for this device at some point later.

Signed-off-by: Stan Grishin <stangri@melmac.ca>

generic: 6.6: add backported bmips CBR patches

Move v6.11 backported bmips CBR patches from bmips target to generic.

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

kernel: Add kmod-mfd-test for kernel 6.6 only

Compilation of mtd_test.ko should be added only for kernel 6.6 or above.

Fixes 26df88a ("kernel: Add kmod-mfd-test")

Signed-off-by: Mieczyslaw Nalewaj <namiltd@yahoo.com>
Link: https://github.com/openwrt/openwrt/pull/16085
Signed-off-by: Robert Marko <robimarko@gmail.com>

utils: omnia-mcutool: Bump to 0.3-rc3

Bump omnia-mcutool to 0.3-rc3:

* The `--upgrade` option will now work even if MCU is in bootloader (for
  example if previous upgrade was aborted).

* On boards with GD32 MCUs, `omnia-mcutool` will now refuse to upgrade
  application firmware to version lower than 4.1 if bootloader version
  is 2.0 (the original for first batch of boards with GD32 MCUs) since
  these versions of application and bootloader are not compatible.

  If user already upgraded to such a combination, an upgrade of
  bootloader firmware is required.

  The `--upgrade` option will inform about this and will automatically
  upgrade bootloader firmware if the `--force` option is given.

  (Note that version 4.1 of the MCU firmware was will be released soon,
   once it is properly tested.)

* Various other improvements.

Signed-off-by: Marek Behún <kabel@kernel.org>
Link: https://github.com/openwrt/openwrt/pull/16086
Signed-off-by: Robert Marko <robimarko@gmail.com>

kernel: bump 6.6 to 6.6.44

Changelog: https://cdn.kernel.org/pub/linux/kernel/v6.x/ChangeLog-6.6.44

Removed upstreamed:
rockchip/patches-6.6/044-v6.11-arm64-dts-rockchip-Add-sdmmc-related-properties-on-r.patch[1]
rockchip/patches-6.6/045-v6.11-arm64-dts-rockchip-Add-pinctrl-for-UART0-to-rk3308-r.patch[2]
rockchip/patches-6.6/046-v6.11-arm64-dts-rockchip-Add-mdio-and-ethernet-phy-nodes-t.patch[3]
rockchip/patches-6.6/048-v6.11-arm64-dts-rockchip-Update-WIFi-BT-related-nodes-on-r.patch[4]
rockchip/patches-6.6/310-PCI-dw-rockchip-Fix-initial-PERST-GPIO-value.patch[5]

Manually rebased:
bcm27xx/patches-6.6/950-0526-mfd-Add-rp1-driver.patch
ramips/patches-6.6/810-uvc-add-iPassion-iP2970-support.patch

All other patches automatically rebased.

1. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.6.44&id=96155dc8c99e76a0e58932ca5f88148a37af3617
2. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.6.44&id=30ee9e5ecc5fd6854a1690ab11e6e6b4851ac9ab
3. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.6.44&id=0c4c1b7daf23bccf0da5b65b3cd5b32bfa5a5af4
4. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.6.44&id=a86a5685675834732789df202e73153db59d8dcc
5. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.6.44&id=a30211c6704899de0616ac0f97601122d899c86d

Build system: x86/64
Build-tested: x86/64/AMD Cezanne
Run-tested: x86/64/AMD Cezanne

Signed-off-by: John Audia <therealgraysky@proton.me>
Link: https://github.com/openwrt/openwrt/pull/16061
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

kernel: Add kmod-mfd-test

6.6.44 introduced breakage complaining about missing mtd_test.ko
for some targets.

Signed-off-by: John Audia <therealgraysky@proton.me>
Link: https://github.com/openwrt/openwrt/pull/16061
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

kernel: rtl8367b: use realtek,extif property

Use realtek,extif property instead of realtek,extif0 to extif2
by extending it with the cpu_port parameter.
The extif number is automatically calculated based on cpu_port.

Signed-off-by: Mieczyslaw Nalewaj <namiltd@yahoo.com>
Link: https://github.com/openwrt/openwrt/pull/15749
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

kernel: rtl8367b: store chip in smi->rtl8367b_chip

Store the chip type in smi->rtl8367b_chip

Signed-off-by: Mieczyslaw Nalewaj <namiltd@yahoo.com>
Link: https://github.com/openwrt/openwrt/pull/15749
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

kernel: rtl8367b: remove rlvid analysis

Remove the rlvid analysis because for the rtl8367b family chips supported
by the driver (rtl8367rb and rtl8367r-vb), rlvid is always equal to 1.
So the code for rlvid equal to 0 is completely unnecessary.

Signed-off-by: Mieczyslaw Nalewaj <namiltd@yahoo.com>
Link: https://github.com/openwrt/openwrt/pull/15749
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

kernel: rtl8367: use realtek,extif property

Use realtek,extif property instead of realtek,extif0 and realtek,extif1
by extending it with the cpu_port parameter.
The extif number is automatically calculated based on cpu_port.

Signed-off-by: Mieczyslaw Nalewaj <namiltd@yahoo.com>
Link: https://github.com/openwrt/openwrt/pull/15749
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

kernel: add missing symbol

Enabling KERNEL_DEBUG_INFO_BTF and KERNEL_KPROBE_EVENTS on 6.6 exposes
CONFIG_PROBE_EVENTS_BTF_ARGS in the kernel config. Add a build option
for it to fix build failures with KERNEL_DEBUG_INFO_BTF and
KERNEL_KPROBE_EVENTS enabled on targets using the 6.6 kernel.

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

zynq: drop 6.1 support

Drop config for Linux 6.1.

Signed-off-by: Mieczyslaw Nalewaj <namiltd@yahoo.com>
Link: https://github.com/openwrt/openwrt/pull/16030
Signed-off-by: Robert Marko <robimarko@gmail.com>

zynq: use kernel 6.6 by default

Switch to Linux kernel version 6.6.

Signed-off-by: Mieczyslaw Nalewaj <namiltd@yahoo.com>
Link: https://github.com/openwrt/openwrt/pull/16030
Signed-off-by: Robert Marko <robimarko@gmail.com>

hostapd: fix SAE H2E security vulnerability

This patch backports fixes for a security vulnerability impacting the
hostapd implementation of SAE H2E.

As upgrading hostapd would require more testing, the second mitigation
step which involves backporting several patches was adopted as outlined
in the official advisory[1].

An explanation of the impact of the vulnerability is provided from the
advisory[1]:

This vulnerability allows the attacker to downgrade the negotiated group
to another enabled group if both the AP and STA have enabled SAE H2E and
multiple groups. It should be noted that the H2E option is not enabled
by default and the attack is not applicable to the default option, i.e.,
hunting-and-pecking, since it does not have any downgrade protection for
group negotiation. In addition, the default configuration for enabled
SAE groups in hostapd is to enable only a single group, so the
vulnerability is not applicable unless hostapd has been explicitly
configured to enable more groups for SAE.

[1]: https://w1.fi/security/2024-2/sae-h2h-and-incomplete-downgrade-protection-for-group-negotiation.txt

Signed-off-by: Rany Hany <rany_hany@riseup.net>
Link: https://github.com/openwrt/openwrt/pull/16042
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

utils: Add the omnia-mcutool utility

Add a new utility, omnia-mcutool, which main purpose is to upgrade the
firmware on the microcontroller on the Turris Omnia router. Depends on
omnia-mcu-firmware, and the upgrade process is pretty simple:

  omnia-mcutool --upgrade

Besides firmware upgrade, the utility can be used to show and configure
various firmware settings.

Signed-off-by: Marek Mojík <marek.mojik@nic.cz>
Signed-off-by: Marek Behún <kabel@kernel.org>
Link: https://github.com/openwrt/openwrt/pull/13799
Signed-off-by: Robert Marko <robimarko@gmail.com>

firmware: Add CZ.NIC Turris Omnia MCU firmware

Add a new package, omnia-mcu-firmware, containing firmware binaries for
the microcontroller on the Turris Omnia router.

Signed-off-by: Marek Mojík <marek.mojik@nic.cz>
Signed-off-by: Marek Behún <kabel@kernel.org>
Link: https://github.com/openwrt/openwrt/pull/13799
Signed-off-by: Robert Marko <robimarko@gmail.com>

mvebu: Add kmod-turris-omnia-mcu

Add support for the MCU driver on CZ.NIC's Turris Omnia. This adds
the ability to do a true board poweroff, and to configure various
features (for example the user may configure that after poweroff, the
router should automatically wake up at a specific time).

Signed-off-by: Marek Mojík <marek.mojik@nic.cz>
Signed-off-by: Marek Behún <kabel@kernel.org>
Link: https://github.com/openwrt/openwrt/pull/13799
Signed-off-by: Robert Marko <robimarko@gmail.com>

gpio-button-hotplug: add vendor button handling

Handle the KEY_VENDOR key in gpio-button-hotplug driver. This is used
by Turris Omnia.

Signed-off-by: Marek Behún <kabel@kernel.org>
Link: https://github.com/openwrt/openwrt/pull/13799
Signed-off-by: Robert Marko <robimarko@gmail.com>

mvebu: 6.6: Backport Turris Omnia MCU patches from 6.11

This backports patches
  dt-bindings: firmware: add cznic,turris-omnia-mcu binding
  platform: cznic: Add preliminary support for Turris Omnia MCU
  platform: cznic: turris-omnia-mcu: Add support for MCU connected GPIOs
  platform: cznic: turris-omnia-mcu: Add support for poweroff and wakeup
  platform: cznic: turris-omnia-mcu: Add support for MCU watchdog
  platform: cznic: turris-omnia-mcu: Add support for MCU provided TRNG
  ARM: dts: turris-omnia: Add MCU system-controller node
  ARM: dts: turris-omnia: Add GPIO key node for front button
  platform: cznic: turris-omnia-mcu: Depend on OF
  platform: cznic: turris-omnia-mcu: Depend on WATCHDOG
  platform: cznic: turris-omnia-mcu: fix Kconfig dependencies
that will be released in 6.11 into mvebu/patches-6.6.

Signed-off-by: Marek Behún <kabel@kernel.org>
Link: https://github.com/openwrt/openwrt/pull/13799
Signed-off-by: Robert Marko <robimarko@gmail.com>

ipq40xx: Drop nand features from Chromium image

This target doesn't actually use NAND.

Signed-off-by: Brian Norris <computersforpeace@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16001
Signed-off-by: Robert Marko <robimarko@gmail.com>

Revert "ath79: remove GPIO driver earlier registration hack"

This reverts commit f444dea428cdcafd78fb75004a942da24cabd48c.

It seems that some devices using GPIO WDT have really short WDT timeouts
and when using module_platform_driver registration it happens too late
and thus WDT will timeout and reset the board.

So, for now lets return the postcore_initcall hack for now.

Fixes: f444dea428cd ("ath79: remove GPIO driver earlier registration hack")
Signed-off-by: Joan Moreau <jom@grosjo.net>
Link: https://github.com/openwrt/openwrt/pull/16035
Signed-off-by: Robert Marko <robimarko@gmail.com>

generic: 6.6: backport upstream RealTek PHY patches

Replace downstream RealTek PHY patches with backported ones.

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

layerscape: remove kernel 6.1 support

Layerscape was switched to 6.6 kernel. Now is time to remove 6.1
support.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16037
Signed-off-by: Robert Marko <robimarko@gmail.com>

layerscape: switch to 6.6 kernel

Let's start wide tests 6.6 kernel.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16037
Signed-off-by: Robert Marko <robimarko@gmail.com>

mediatek: refresh patches

CI says patches need to be refreshed, so do so.

Fixes: 6bb334c5cf1c ("mediatek: fix u-boot env layout NVMEM definitions")
Link: https://github.com/openwrt/openwrt/pull/16051
Signed-off-by: Robert Marko <robimarko@gmail.com>

netifd: update to git HEAD

 68c8a4f system-linux: re-apply ethtool on phy attachment
 890929b wireless: add support for defining wifi interfaces via procd service data
 b57e40b wireless: use blobmsg_parse_attr
 7a6532f proto-shell: add proto property for skipping device config
 33ec3da CMake: bump the minimum required CMake version to 3.5

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

generic: 6.6: update block nvmem patchset

While discussions are still ongoing, update the block NVMEM provider
patchset to the level submitted upstream to allow testing and validation.

Link: https://patchwork.kernel.org/project/linux-block/list/?series=875202
Signed-off-by: Daniel Golle <daniel@makrotopia.org>

mediatek: fix u-boot env layout NVMEM definitions

s/u-boot,env-layout/u-boot,env/g

Signed-off-by: Enrico Mioso <mrkiko.rs@gmail.com>

kirkwood: remove 6.1 support

Kirkwood was switched to 6.6. We can remove 6.1 support now.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15998
Signed-off-by: Robert Marko <robimarko@gmail.com>

kirkwood: remove accidentally commited file

generic.mk was commited by mistake. It isn't used. Let's remove it.

Fixes: ccbdb212f36f ("kirkwood: Add missing package dependency")
Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15998
Signed-off-by: Robert Marko <robimarko@gmail.com>

kirkwood: switch to kernel 6.6

Let's switch to kernel 6.6 for wide tests.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15998
Signed-off-by: Robert Marko <robimarko@gmail.com>

rockchip: add RTL8723DU support for Radxa ROCK Pi E v3.0

Radxa ROCK Pi E v3.0 has a RTL8723DU Wi-Fi 4 on-board device. enable
it.

Signed-off-by: FUKAUMI Naoki <naoki@radxa.com>
Link: https://github.com/openwrt/openwrt/pull/15910
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

rockchip: add RTL8723DS support for Radxa ROCK Pi S

Radxa ROCK Pi S has a RTL8723DS Wi-Fi 4 on-board device. enable it.

Signed-off-by: FUKAUMI Naoki <naoki@radxa.com>
Link: https://github.com/openwrt/openwrt/pull/15910
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>