This is an automatically generated commit which aids following Kernel patch
history, as git will see the move and copy as a rename thus defeating the
purpose.
For the original discussion see:
https://lists.openwrt.org/pipermail/openwrt-devel/2023-October/041673.html
The Realtek mdio bus operations are based on the port number. As
of now the driver exposes a single bus with a 1:1 mapping between
phy and port. This will change in the future when the driver
exposes multiple busses. This will need a mapping - e.g.
- bus 0, phy 23 -> read/write port 23
- bus 1, phy 0 -> read/write port 24
Provide a rtmdio_phy_to_port() indirection function that will
lookup the port number based on a bus and a phy. For now it
will only run some basic sanity checks and return the given
input address (because of 1:1 mapping).
When the mdio driver gets enhanced for multiple busses it must
snchronize reads/writes to the single controller. Add a lock
to the control structure and guard the critical operations.
The ethernet driver currently uses the legacy sw() macros. Add a
regmap to the driver that is derived from the parent mfd (like in
the mdio drivers). As a first consumer make use of that in the IRQ
functions.
Shiji Yang [Tue, 31 Mar 2026 14:33:44 +0000 (22:33 +0800)]
kernel: modules: fix kmod-iavf dependencies
Add missing dependency kmod-ptp to fix the build error. We also
add symbol CONFIG_PTP_1588_CLOCK_OPTIONAL explicitly for kmod-ptp
because it is selected by CONFIG_PTP_1588_CLOCK. Fix:
Package kmod-iavf is missing dependencies for the following libraries:
ptp.ko
LAN Ports:
- lan1 is the normal white LAN port
- lan2 is labelled "LAN/Backhaul", which can either be LAN or WAN.
- Since this device is intended to be part of a mesh, this will usually
be a LAN port, hence why it's assigned to lan2 here
Notes:
- I've chosen "Halo Base Pro" as the device model, despite the website URL
showing WL-WN535M3, as the label only states "Halo Base Pro".
- Only the 5GHz WiFi address is labelled on the device
Shiji Yang [Thu, 19 Feb 2026 12:06:08 +0000 (20:06 +0800)]
kernel: modules: add new driver package phy-mtk-tphy
MediaTek T-PHY controller supports multiple usb2.0, usb3.0 ports,
PCIe and SATA. ARM mediatek and ramips/mt7621 targets require this
package to make the XHCI work properly.
Shiji Yang [Mon, 19 May 2025 11:03:43 +0000 (19:03 +0800)]
lantiq: dts: rename spi pins node name to spi-pins
The dtc will treat "spi" node as an SPI controller. This resulted
in lots of false warning reports. Workaround it by renaming the
pinctrl node name to "spi-pins". This patch fixes the following
dtc warnings:
vr9.dtsi:326.18-336.6: Warning (spi_bus_bridge): /fpi@10000000/pinmux@e100b10/spi: incorrect #address-cells for SPI bus
vr9.dtsi:326.18-336.6: Warning (spi_bus_bridge): /fpi@10000000/pinmux@e100b10/spi: incorrect #size-cells for SPI bus
ar9.dtsi:286.18-296.6: Warning (spi_bus_bridge): /fpi@10000000/pinmux@e100b10/spi: incorrect #address-cells for SPI bus
ar9.dtsi:286.18-296.6: Warning (spi_bus_bridge): /fpi@10000000/pinmux@e100b10/spi: incorrect #size-cells for SPI bus
amazonse.dtsi:166.18-176.6: Warning (spi_bus_bridge): /fpi@10000000/pinmux@e100b10/spi: incorrect #address-cells for SPI bus
amazonse.dtsi:166.18-176.6: Warning (spi_bus_bridge): /fpi@10000000/pinmux@e100b10/spi: incorrect #size-cells for SPI bus
seuros [Tue, 10 Mar 2026 23:32:04 +0000 (23:32 +0000)]
tools/e2fsprogs: fix build with system libblkid probe API
When building on a host with e2fsprogs >= 1.47.0 installed, configure
detects the system libblkid probe API (blkid_probe_get_partitions etc.)
and sets ac_cv_lib_blkid_blkid_get_cache=yes. This causes plausible.c
to be compiled with HAVE_BLKID_PROBE_GET_PARTITIONS defined, but the
host build links against the internal blkid stub, resulting in:
plausible.c: error: unknown type name 'blkid_probe'
Override the configure cache variables to prevent detection of the
system libblkid probe API, ensuring the internal stub is used
consistently.
The RTL8231 LED driver is using the .brightness_set op, which according
to the documentation should not sleep. Since the RTL8231 is controlled
via an MDIO bus, this is not the case.
When using a non-offloaded LED driver such as netdev, this may currently
trigger kernel warnings:
WARNING: CPU: 0 PID: 14 at kernel/time/timer.c:1754 call_timer_fn.isra.0+0x84/0x8c
timer: led_timer_function+0x0/0x1d8 preempt leak: 00000100 -> 00000000
Switch to the .brightness_set_blocking op to indicate the operation
might sleep.
Fixes: 6ef6014887c3 ("realtek: Add pinctrl support for RTL8231") Signed-off-by: Sander Vanheule <sander@svanheule.net>
These patches stop secondary CPUs before restart and wait for them
to go offline, fixing the mt7621 reboot deadlock without needing to
disable CONFIG_LIST_HARDENED.
rteth_930x_decode_tag() and rteth_931x_decode_tag() are basically
the same. The only difference is the port bitmask. Luckily on the
RTL930x devices the unused fifth bit is always zero and the mask
can be aligned with RTL931x. Take over the RTL931x reason debugging
for both targets.
Jonas Jelonek [Wed, 4 Feb 2026 22:02:00 +0000 (22:02 +0000)]
realtek: pcs: rtl931x: set SerDes to off early
Set the SerDes mode to OFF early in the setup process to have a clear
starting point. This was part of the media handling before as a leftover
from the SDK code import. In the SDK, this function didn't only care
about applying some media settings but was also some kind of mode
setting for fiber modes.
Jonas Jelonek [Sat, 28 Mar 2026 17:24:15 +0000 (18:24 +0100)]
realtek: pcs: condense fiber media types
Right now we operate with distinct media types for fiber using different
speeds. This is more of a leftover from the SDK then it really makes
sense design-wise. The set_media function from the SDK did a lot more
than just setting some media-specific parameters. As part of
deconstructing this, also reduce the fiber types to a single media type
and handle the speed-agnostic parts based on the hw_mode for now.
This also drops the check for 100M speed around a block of writes. This
check has it's origin in SDK code where a switch statement just didn't
handle this case. However, the rest of the SDK doesn't handle this case
either. While the explicit 100M support isn't verified yet, there's no
need to keep that check.
Jonas Jelonek [Fri, 27 Mar 2026 22:32:34 +0000 (23:32 +0100)]
realtek: pcs: collapse DAC media types
So far we had separate media types for different DAC cable lengths,
equal to how the SDK defines them. However, this seems overengineered.
The types '_50CM' and '_100CM' are always treated equally, same for
'_300CM' and '_500CM'. Not only in the RTL931x code, but also the
RTL930x code usually just makes a distinction between short and long.
Thus, make that a bit cleaner by reducing the DAC type set to '_SHORT'
and '_LONG' with DAC cables < 3m being considered short and those with
>= 3m being considered as long.
Jonas Jelonek [Fri, 27 Mar 2026 12:01:06 +0000 (13:01 +0100)]
realtek: pcs: rename port_media to sds_media
The term 'port media' was chosen due to how the code taken from the SDK
was organized. However, the SDK does a pretty good job at intermixing a
lot of things which cause confusion in the end. This also applies to the
media settings. To be correct, we do not really have port awareness in
the PCS driver (and probably shouldn't have) so we do not really deal
with a port's media. Instead, we set settings depending on what is
attached to the SerDes itself.
To fix this confusion, rename the enum, its fields and all usage from
'port_media' to 'sds_media'.
The NanoPi R5S LTS variant has a reset button wired to GPIO4_A0,
but the current OpenWrt rockchip 6.12 patch set lacks the
corresponding gpio-keys definition in rk3568-nanopi-r5s.dts.
As a result, pressing the button generates no reset event.
rockchip already includes kmod-gpio-button-hotplug by default,
so backporting the upstream DTS change restores the expected
behavior.
Rename the patch to
011-v6.17-arm64-dts-rockchip-Add-reset-button-to-NanoPi-R5S.patch
to match the upstream source.
The non-LTS NanoPi R5S leaves this pin unconnected, matching the
upstream rationale for using the shared rk3568-nanopi-r5s.dts.
The dsa probing currently disable multiple spanning tree for RTL93xx
and keeps the bootloader setup for RTL83xx untouched. Additionally it
does this by checking for the family id. Refactor this to a common
"disable msti" setup.
To make it clear: From now on RTL83xx will also run in MSTI disabled
state. Tested on Linksys LGS310C.
xingchi [Tue, 13 Jan 2026 08:01:20 +0000 (16:01 +0800)]
ramips: add support for Z-ROUTER ZR-2662
This commit adds support for Z-ROUTER ZR-2662 (also known as Routerich
AX1800 V2) wireless WiFi 6 router.
Hardware specification
---------------------
- SoC : MediaTek MT7621AT, MIPS, 880 MHz
- RAM : 256 MiB
- Flash : NAND 128 MiB (AMD/Spansion S34ML01G2)
- WLAN :
- 2.4 GHz : MediaTek MT7905D/MT7975 (14c3:7916), b/g/n/ax, MIMO 2x2
- 5 GHz : MediaTek MT7915E (14c3:7915), a/n/ac/ax, MIMO 2x2
- Ethernet : 10/100/1000 Mbps x4 (1x WAN, 3x LAN)
- USB : 1x 2.0
- UART : 3.3V, 115200n8, pins are silkscreened on the pcb
- Buttons : 1x Reset
- LEDs : 1x WiFi 2.4 GHz (green)
1x WiFi 5 GHz (green)
1x LAN (green)
1x WAN (green)
1x WAN no-internet (red)
- Power : 12 VDC, 1 A
Installation
------------
1. Run tftp server on your PC (IP: 192.168.2.2) and put OpenWrt initramfs
image (initramfs.bin) to the tftp root dir
2. Open the following link in the browser to enable telnet:
http://192.168.2.1/cgi-bin/telnet_ssh
3. Connect to the router (default IP: 192.168.2.1) using telnet shell
(credentials - user:admin)
4. Run the following commands in the telnet shell (this will install
OpenWrt initramfs image on nand flash):
cd /tmp
tftp -g -r initramfs.bin 192.168.2.2
mtd write initramfs.bin firmware
mtd erase firmware_backup
reboot
5. Copy OpenWrt sysupgrade image (sysupgrade.bin) to the /tmp dir of the
router
6. Connect to the router (IP: 192.168.1.1) using ssh shell and run
sysupgrade command:
sysupgrade -n /tmp/sysupgrade.bin
Return to stock
---------------
1. Copy stock firmware (stock.bin) to the /tmp dir of the router using scp
2. Run following command in the router shell:
cd /tmp
mtd write stock.bin firmware
reboot
Recovery
--------
Connect uart (pins are silkscreened on the pcb), interrupt boot process by
pressing any key, use u-boot menu to flash stock firmware image or OpenWrt
initramfs image.
MAC addresses
-------------
+---------+-------------------+-----------+
| | MAC | Algorithm |
+---------+-------------------+-----------+
| LAN | 24:0f:5e:xx:xx:4c | label |
| WAN | 24:0f:5e:xx:xx:4d | label+1 |
| WLAN 2g | 24:0f:5e:xx:xx:4e | label+2 |
| WLAN 5g | 24:0f:5e:xx:xx:4f | label+3 |
+---------+-------------------+-----------+
The WLAN 2.4 MAC was found in 'factory', 0x4
The LAN MAC was found in 'factory', 0xfff4
The WAN MAC was found in 'factory', 0xfffa
Note: This device is similar to Z-ROUTER ZR-2660, but with minor hardware
revisions. The firmware and configuration are compatible.
MAC addresses:
+---------+-------------------+-----------+
| | MAC | Algorithm |
+---------+-------------------+-----------+
| WAN | 80:af:ca:xx:xx:x1 | label+1 |
| LAN | 80:af:ca:xx:xx:x0 | label |
| WLAN 2g | 80:af:ca:xx:xx:x2 | label+2 |
+---------+-------------------+-----------+
Migration to OpenWrt:
- Download the RSA signed intermediate firmware from the Cudy website:
`openwrt-ramips-mt76x8-cudy_lt300-v3-squashfs-flash.bin`
- Connect computer to LAN and flash the intermediate firmware via OEM web interface
- OpenWrt is now accessible via 192.168.1.1
Revert back to OEM firmware:
- Set up a TFTP server on IP 192.168.1.88 and connect to the WAN port (upper port)
- Provide the Cudy firmware as recovery.bin in the TFTP server
- Press the reset button while powering on the device
- Recovery process is started now
- When recovery process is done, OEM firmware is accessible via 192.168.10.1 again
General information:
- No possibility to load a initramfs image via U-Boot because there is no option to interrupt U-Boot
Clemens Hopfer [Fri, 30 May 2025 12:03:40 +0000 (14:03 +0200)]
base-files: MAJOR/MINOR not sequential, use DISKSEQ instead
Export the unique, monotonic DISKSEQ sequence drive number instead of its
major/minor numbers to identify the boot disk and directly match the partition
in export_partdevice with PARTN.
The MINOR blockdevice numbers are not guaranteed sequential across disks, it
can happen that disks enumerate before their partitions are probed, resulting
in interleaved MINOR numbers breaking the partition offset calculation:
major minor #blocks name
259 0 250059096 nvme0n1
259 2 8192 nvme0n1p1
259 3 491520 nvme0n1p2
259 4 239 nvme0n1p128
259 1 250059096 nvme1n1
259 5 250057728 nvme1n1p1
Replace the deprecated EXTRA_CFLAGS with ccflags-y so that
the -DCONFIG_MT76_LEDS define is applied correctly by the kernel
build system. EXTRA_CFLAGS is no longer honored by recent
kbuilds[1]; ccflags-y is the supported variable and works
on kernels 6.12 and 6.18.
Note: OpenWrt combines the stock dual firmware partitions
for more overlay capacity, however the OpenWrt image cannot
exceed 13504k
Installation:
-------------
> When connected to CloudTrax, the local management login will be disabled to prevent settings conflicts.
Ensure the switch does not have a working internet connection or the local
web management interface is disabled.
Go to the web management page of the switch (may require factory reset).
By default the switch will use DHCP to obtain an IP address.
The default login user is `admin` with password `0p3nm3$h!`
On the left menu, click "Management" and then "Dual Image" and ensure that
"Partition 0" is selected as the active partition. If it is not, select
"Partition 0" and click "Apply" to save changes.
Click on "Upgrade" in the top right of the web interface. Select the
Active boot partition to update. Select the OpenWrt file ending
in `-initramfs-kernel.bin` as the update file to upload.
Upload the file and follow the prompts to upgrade the firmware.
Reboot the switch from the web UI after the firmware update is completed.
Wait for OpenWrt to finish booting (~2 minutes)
Use SSH or the Luci UI (if available) to perform the sysupgrade.
Copy the sysupgrade file ending in `-squashfs-sysupgrade.bin` to the switch:
```
scp -O openwrt-realtek-rtl838x-datto_l8-squashfs-sysupgrade.bin root@192.168.1.1:/tmp/
```
SSH to the switch and run `sysupgrade`:
```
ssh root@192.168.1.1
$ sysupgrade -n /tmp/openwrt-realtek-rtl838x-datto_l8-squashfs-sysupgrade.bin
```
OpenWrt will be installed. Note that first boot after installing requires ~3
minutes for the JFFS2 overlay to be formatted. When the Power LED stops blinking
in the first boot after `sysupgrade`, JFFS2 formatting is completed.
----
Revert back to stock firmware:
You will need a tftp server and the original Datto firmware.
Download the firmware for the S8-L/L8 from Datto:
https://networkinghelp.datto.com/help/Content/kb/Networking/Switches/KB360023113291.html
Rename `s8-l_fw_01.03.24_180823-1639.bix` to `vmlinux.bix`,
put `vmlinux.bix` in the root directory of your tftp server.
Connect a serial console to the UART header and power on the switch.
Interrupt U-Boot by typing `pac` when you see
`Enter correct key to stop autoboot:`
Run the following commands:
```
setenv serverip <tftp_server_ip>
setenv ipaddr <ip_on_same_subnet>
setenv netmask 255.255.255.0
run rtkon
run update_linux
run update_linux2
reset
```
The switch will boot the Datto firmware.
Signed-off-by: Hal Martin <hal.martin@gmail.com> Tested-By: Raylynn Knight <rayknight@me.com> Signed-off-by: Sander Vanheule <sander@svanheule.net>
John Audia [Fri, 29 Aug 2025 09:43:26 +0000 (05:43 -0400)]
x86/64: 6.12: 6.18: enable performance related options
Enable Turbo Boost Max 3.0 (INTEL_TURBO_MAX_3) to allow the kernel
scheduler to utilize preferred cores on modern Intel CPUs. This has
no effect on AMD systems.
Enable core scheduling (SCHED_CORE) to provide safer SMT task
placement and better isolation. While it can reduce peak throughput
slightly, it improves scheduling correctness on heterogeneous cores
and works across both Intel and AMD.
Enable the accelerated GHASH implementation using the CLMUL instruction
(CRYPTO_GHASH_CLMUL_NI_INTEL). This provides substantial performance
improvements for AES-GCM workloads (e.g. TLS, IPsec, OpenVPN) on CPUs
with the pclmulqdq instruction. Supported on both modern Intel and AMD
processors; falls back gracefully on older CPUs.
Run-tested twice, once with the 6.12 series and again with the 6.18.
This is an automatically generated commit which aids following Kernel patch
history, as git will see the move and copy as a rename thus defeating the
purpose.
For the original discussion see:
https://lists.openwrt.org/pipermail/openwrt-devel/2023-October/041673.html
wireguard-tools: use kconfig dependency for ip instead of runtime
The conditional runtime dependency on the ip package was originally
intended to only pull in the ip package when busybox ip is not
configured. However, in APK-based builds the BUSYBOX_CONFIG_*
variables may not be resolvable at package metadata generation
time, causing the ip dependency to be unconditionally baked into
the .apk package. This forces users to install ip-tiny or ip-full
even though busybox already provides the ip command.
Revert to the previous +@BUSYBOX_CONFIG_IP / +@BUSYBOX_CONFIG_FEATURE_IP_LINK
kconfig-level dependencies. These ensure busybox ip support is
enabled at config time without creating a runtime package dependency.
Both options default to y, and wireguard.uc only uses basic
ip link commands that busybox fully supports.
Like the PHY_C22() macro before add a helper that allows to define
a C45 based phy. It works basically the same with two parameters
PHY_C45(port_number, bus_address) where
- port_number is the absolute overall unique phy number
- bus_address is the location of the phy on the bus
As a first consumer adapt the Xikestor SKS8300-8T devicetree.
Shiji Yang [Tue, 31 Mar 2026 00:10:30 +0000 (08:10 +0800)]
kernel: tune 24kc instead of 34kc for mips32r2
GCC generates the same code for 24kc and 34kc. Since we have
converted all 34kc targets to the 24kc, it's better to switch
kernel -mtune to 24kc to avoid confusing developers.
airoha: backport minor fixup for Ethernet driver on Offload Scenario
Backport minor fixup merged upstream for Ethernet driver on Offload
Scenario. This is to continue the effort of keeping the Airoha Ethernet
driver synced with the upstream version.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Backport 2 patches from [1] that cleanup the i2c-rtl9300 upstream driver.
They have been long accepted and are already in mainline kernel as of 7.0-rc1.
The "100-rtl9300-i2c-add-more-speeds.patch" required refreshing after adding
these backport patches.
realtek: dsa: move fib_entries to config structure
The fib_entries attribute is a device specific constant.
Therefore move it into the configuration structure. Add
a comment why someone used 16K fib_entries for RTL931x
instead of the possible 32K.
Robert Marko [Tue, 31 Mar 2026 16:54:47 +0000 (18:54 +0200)]
generic: 6.18: hack: io_uring: zcrx: Use IS_REACHABLE() instead of IS_ENABLED()
We patch DMA_BUF to make it tristate, so once ALL_KMODS is selected it will
be built as a module even if previously disabled in the config.
So, since IO_URING_ZCRX does not depend on DMA_BUF linking will fail with:
aarch64-openwrt-linux-musl-ld: Unexpected GOT/PLT entries detected!
aarch64-openwrt-linux-musl-ld: Unexpected run-time procedure linkages detected!
aarch64-openwrt-linux-musl-ld: io_uring/zcrx.o: in function `io_release_dmabuf':
io_uring/zcrx.c:94:(.text+0x20): undefined reference to `dma_buf_unmap_attachment_unlocked'
aarch64-openwrt-linux-musl-ld: io_uring/zcrx.c:97:(.text+0x30): undefined reference to `dma_buf_detach'
aarch64-openwrt-linux-musl-ld: io_uring/zcrx.c:99:(.text+0x3c): undefined reference to `dma_buf_put'
aarch64-openwrt-linux-musl-ld: io_uring/zcrx.o: in function `io_import_dmabuf':
io_uring/zcrx.c:125:(.text+0x1b20): undefined reference to `dma_buf_get'
aarch64-openwrt-linux-musl-ld: io_uring/zcrx.c:132:(.text+0x1b34): undefined reference to `dma_buf_attach'
aarch64-openwrt-linux-musl-ld: io_uring/zcrx.c:139:(.text+0x1b48): undefined reference to `dma_buf_map_attachment_unlocked'
make[6]: *** [scripts/Makefile.vmlinux:72: vmlinux.unstripped] Error 1
So, lets use IS_REACHABLE() to check for CONFIG_DMA_SHARED_BUFFER instead
to avoid adding a dependency to CONFIG_DMA_SHARED_BUFFER.
Signed-off-by: Robert Marko <robert.marko@sartura.hr>
Use the compiled asus_rt-ax52-pro-initramfs.trx file from the this repo.
Connect the PC via LAN to one of the yellow router ports and wait until your PC to get a DHCP lease.
Browse to http://192.168.50.1 or http://www.asusrouter.com/
If your router is brand new, finish the setup process and log into the Web-UI.
Navigate to Administration → Firmware Upgrade or use this link http://www.asusrouter.com/Advanced_FirmwareUpgrade_Content.asp.
Upload the .trx file to router
Wait for it to reboot
trx image is initramfs version. You must upgrade to squashfs version.
Browse to http://192.168.1.1/cgi-bin/luci/admin/system/flash
Upload asus_rt-ax52-pro-squashfs-sysupgrade.bin and use sysupgrade -n
Wait for it to reboot
SSH to 192.168.1.1 and set a root password, or browse to http://192.168.1.1
-------Revert to stock asus firmware ---------:
1: Download the rt-ax52 firmware from ASUS official website. Save the firmware to tftp server directory and rename to RT-AX52.trx
2: Connect the PC with TFTP server to the RT-AX52. Set a static ip on the ethernet interface of your PC. (ip address: 192.168.1.70, subnet mask:255.255.255.0)
3: Conect to the serial console, power on again, interrupt the autoboot process by pressing '4' when prompted. $ ubi remove linux
$ ubi remove jffs2
$ ubi remove rootfs
$ ubi remove rootfs_data
$ ubi create linux 0x45fe000
$ reset
then the dut will reboot,interrupt the autoboot process by pressing '2' when prompted. 2: Load System code then write to Flash via TFTP.
Warning!! Erase Linux in Flash then burn new one. Are you sure?(Y/N) $: enter y
you will see the follow, type enter directly:
Input device IP (192.168.1.1) ==:
Input server IP (192.168.1.70) ==:
Input Linux Kernel filename (RT-AX52.trx) ==:
The Realtek target currently uses two phy macros to simplify the
device dts.
- EXTERNAL_PHY() to denote a phy attached to the SoC
- INTERNAL_PHY() to denote an internal PHY (inside the SoC)
There is no benefit doing this. The topology around a port/phy is
well defined by the port macros. They link port, phy, pcs and even
leds. The only consumer of the attribute "phy-is-integrated" is
inside the dsa driver and that is being refactored.
As a first step define a new more meaningful PHY_C22() macro that
describes a c22 capable phy. This does not need to care about the
external/internal relation. To make it even more useful for the
RTL93xx targets with multiple mdio busses give it two parameters
PHY_C22(port_number, bus_address) where
- port_number is the absolute overall unique phy number
- bus_address is the location of the phy on the bus
For RTL83xx these two parameters will usually be the same. Instead
of three steps (inventing the macro, converting the consumers and
removeing the old macor) do a one-step conversion for the existing
EXTERNAL_PHY() macro.
The dsa driver currently has different attributes to denote what
hardware is around a port:
- phy_is_integrated: true if phy is not driven by a serdes
- phy: the type of the attached phy (e.g. 0=NONE, 2=RTL8218B, ....)
- pcs: link to a serdes pcs instance
This is somehow redundant and especially the phy type should be only
part of the phy driver and is not needed by the dsa driver at all.
Remove the redundancy by simply keeping a boolean attribute "phy" that
flags a phy driven port and can be used similar to the pcs (pointer)
attribute. With that the driver can check phy/pcs as follows:
- if (ports[i].pcs) -> port has a dedicated serdes
- if (ports[i].phy) -> port has a dedicated phy
That implemented, the "phy-is-integrated" attribute of a phy can be
removed from the dts. This will be a separate commit. As a side effect
the following (annoying) boot message for kernel 6.18 gets fixed.
OF: /switchcore@1b000000/mdio-controller/mdio-bus@0/ethernet-phy@24:
Read of boolean property 'sfp' with a value.
Jonas Jelonek [Mon, 30 Mar 2026 09:46:29 +0000 (09:46 +0000)]
realtek: force in-band autoneg on XGS1250-12 A1 10G PHYs
In kernel 6.18, upstream added a change to the Aquantia PHY driver which
reports autoneg and inband capabilities as the PHY supports it, and
configures it accordingly in the PHY [1]. Due to how phylink works, it
then decides to turn off in-band signalling and prefer outband signalling
via MDIO.
We do not fully support running a USXGMII link with disabled
autonegotiation which leads to a non-working link between RTL93xx switch
and Aquantia PHYs running on USXGMII. To workaround this issue until
this support is added (if it is properly supported by the hardware),
force the Aquantia PHYs on affected devices to use inband signalling
instead of outband signalling. To achieve this, one can add
Jonas Jelonek [Mon, 30 Mar 2026 09:42:08 +0000 (09:42 +0000)]
realtek: dts: use SWITCH_PORT_LED for XGS1250-12
Switch the common DTS for Zyxel XGS1250-12 variants to the recently
added SWITCH_PORT_LED macro to reduce boilerplate and make the DTS
cleaner. As a side effect, this also assign labels to the port nodes so
they can be referenced by the variant-specific device tree sources.
Daniel Golle [Tue, 31 Mar 2026 02:13:41 +0000 (03:13 +0100)]
kernel: update MxL862xx DSA driver
Introduce fix which prevents the kernel from crashing in case the mxl862xx
driver fails to probe due to outdated firmware running on the switch.
Cancel all pending work and prevent rescheduling of counter polling in case
the driver errors out during probe.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
projects / thirdparty / openwrt.git / log
