Add building the 6.18 kernel for airoha as testing.
Signed-off-by: Kenneth Kasilag <kenneth@kasilag.me>
Link: https://github.com/openwrt/openwrt/pull/21019
Signed-off-by: Jonas Jelonek <jelonek.jonas@gmail.com>
Use the new kmod-phy-realtek driver for kernel 6.18 on
Gemtek W1700K.
The driver selection is gated using `KERNEL_PATCHVER` to
avoid build failures on 6.12.
The device tree selection is also gated by `KERNEL_PATCHVER`
to select the appropriate device tree per kernel version.
Once kernel 6.18 is tested stable for airoha these changes
can be simplified and the obsolete properties removed.
Signed-off-by: Kenneth Kasilag <kenneth@kasilag.me>
Link: https://github.com/openwrt/openwrt/pull/21019
Signed-off-by: Jonas Jelonek <jelonek.jonas@gmail.com>
Patch `600-12-clk-en7523-Add-reset-controller-support-for-EN7523-S.patch`
was accepted in 6.19-rc1 (1c0608d860db973ad09b5a9ccb19b76ae07622a3).
Properly tag the patch for future reference.
Signed-off-by: Kenneth Kasilag <kenneth@kasilag.me>
Link: https://github.com/openwrt/openwrt/pull/21019
Signed-off-by: Jonas Jelonek <jelonek.jonas@gmail.com>
Due to hardware bugs, the PCIE Gen3 IP in Airoha AN7581 requires a
special reset procedure:
> MAC reset asserted thru the SCU block
> PERST asserted thru the SCU for the desired PCIE lane
> PHY initialization runs, clocks enabled
> MAC reset deasserted
> EQ config (if needed) written
> PERST deasserted
The existing code currently toggles PERST for all three PCIE
ports every time mtk_pcie_en7581_power_up is called, resulting
in PCIE link down issues.
This issue was discovered during porting of the 6.18 kernel to
AN7581. It is not entirely clear how the hardware seemed to
function under kernel 6.12 with similar driver code, it is
presumed that differences in assert/deassert times for PERST in
6.18 changed which exposed the bug.
Signed-off-by: Kenneth Kasilag <kenneth@kasilag.me>
Link: https://github.com/openwrt/openwrt/pull/21019
Signed-off-by: Jonas Jelonek <jelonek.jonas@gmail.com>
Remove all patches that have already been upstreamed for kernel 6.18.
Signed-off-by: Kenneth Kasilag <kenneth@kasilag.me>
Link: https://github.com/openwrt/openwrt/pull/21019
Signed-off-by: Jonas Jelonek <jelonek.jonas@gmail.com>
Update the kernel configuration in for kernel 6.18.
Signed-off-by: Kenneth Kasilag <kenneth@kasilag.me>
Link: https://github.com/openwrt/openwrt/pull/21019
Signed-off-by: Jonas Jelonek <jelonek.jonas@gmail.com>
This is an automatically generated commit.
When doing `git bisect`, consider `git bisect --skip`.
Signed-off-by: Kenneth Kasilag <kenneth@kasilag.me>
Link: https://github.com/openwrt/openwrt/pull/21019
Signed-off-by: Jonas Jelonek <jelonek.jonas@gmail.com>
Set max RX size configuration (AIROHA_MAX_RX_SIZE) to 0x3f00.
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Link: https://github.com/openwrt/openwrt/pull/23585
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Disable unused crypto algorithms. If needed, install required packages.
Suggested-by: Qingfang Deng <dqfext@gmail.com>
Signed-off-by: Aleksander Jan Bajkowski <olek2@wp.pl>
Link: https://github.com/openwrt/openwrt/pull/23536
Signed-off-by: Jonas Jelonek <jelonek.jonas@gmail.com>
Apply minor fixup for PPE_MTU configuration and LRO queue configuration.
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
[ improve commit title/description ]
Link: https://github.com/openwrt/openwrt/pull/23566
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Add hardware TCP Large Receive Offload (LRO) support to the airoha_eth
driver, leveraging the EN7581/AN7583 SoC's 8 dedicated LRO hardware queues
mapped to RX queues 24–31. LRO hw offloading does not support
Scatter-Gather (SG) so it is required to increase the page_pool allocation
order to 2 for RX queues 24–31 (LRO queues).
Performance comparison between GRO and hw LRO has been carried out using
a 10Gbps NIC:
GRO: ~2.7 Gbps
LRO: ~8.1 Gbps
Tested-by: Madhur Agrawal <madhur.agrawal@airoha.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Link: https://github.com/openwrt/openwrt/pull/23530
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
EN7581 or AN7583 SoCs support connecting multiple external SerDes (e.g.
Ethernet or USB SerDes) to GDM3 or GDM4 ports via a hw arbiter that
manages the traffic in a TDM manner. As a result multiple net_devices can
connect to the same GDM{3,4} port and there is a theoretical "1:n"
relation between GDM ports and net_devices.
┌─────────────────────────────────┐
│ │ ┌──────┐
│ P1 GDM1 ├────►MT7530│
│ │ └──────┘
│ │ ETH0 (DSA conduit)
│ │
│ PSE/FE │
│ │
│ │
│ │ ┌─────┐
│ P0 CDM1 ├────►QDMA0│
│ P4 P9 GDM4 │ └─────┘
└──┬─────────────────────────┬────┘
│ │
┌──▼──┐ ┌────▼────┐
│ PPE │ │ ARB │
└─────┘ └─┬─────┬─┘
│ │
┌──▼──┐┌─▼───┐
│ ETH ││ USB │
└─────┘└─────┘
ETH1 ETH2
This series introduces support for multiple net_devices connected to the
same Frame Engine (FE) GDM port (GDM3 or GDM4) via an external hw
arbiter. Please note GDM1 or GDM2 does not support the connection with
the external arbiter.
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Link: https://github.com/openwrt/openwrt/pull/23481
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
PHY as21xxx used on Airoha EN7581 GDM4 requires in-band status control.
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Link: https://github.com/openwrt/openwrt/pull/23481
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Backport GDM2 loopback fixup for Ethernet driver. This should be the last
patch before introduction of Multi-Serdes support series.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
While implementing standalone PCS support for DSA, it was found that making
the MAC driver passing the available_pcs array is limiting and problematic
for memory handling and allocation. To better handle this, change the logic
and make phylink allocate the struct and make the MAC driver implement a
function in phylink_config .fill_available_pcs to fill the PCS array.
Update the Airoha and Mediatek driver to reflect this new implementation.
Link: https://github.com/openwrt/openwrt/pull/23413
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Introduce the capability to hw offload via netfilter flowtable APIs the
IP TOS info. Implement the sw offloading for DSCP field via the
netfilter flowtable APIs.
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Link: https://github.com/openwrt/openwrt/pull/23423
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Airoha reported some additional bug and fixes were pushed for the ethernet
driver. Backport the additional patch merged upstream and refresh all
affected patch.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
After the standalone Airoha PCS driver and pcs-handle binding, wan (gdm2)
and lan2 (gdm4) netdevs probe but do not pass traffic.
Similar to the Aeonsemi PHY it seems also the RTL PHY needs in-band to the
PCS for USXGMII to work correctly. This still needs to be better
investigated but in the meantime lets apply this workaround to restore
traffic.
Signed-off-by: Daniel Pawlik <pawlik.dan@gmail.com>
[ improve commit description ]
Link: https://github.com/openwrt/openwrt/pull/23383
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
It was discovered that the internal PHY interrupt are not always connected
and PHY link up/down is not correctly detected.
This is the case of the Nokia Valyrian board that suffer from this problem.
To handle this, drop the internal PHY interrupt property from DTSI and add
it only to the Gemtek W1700K DTS where it was reported to work correctly.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
The AN7531N SoC has currently problems communicating using `phy_mmd_...`
when irqbalance is active. But when there is a communication error in the
interrupt handler, the interrupt will be disabled. This can usually be seen
in the logs as:
irq 53: nobody cared (try booting with the "irqpoll" option)
CPU: 0 UID: 0 PID: 0 Comm: swapper/0 Tainted: G O 6.12.77 #0
Tainted: [O]=OOT_MODULE
Hardware name: Gemtek W1700K (OpenWrt U-Boot layout) (DT)
Call trace:
dump_backtrace.part.0+0xbc/0xcc
show_stack+0x14/0x20
dump_stack_lvl+0x58/0x74
dump_stack+0x14/0x1c
__report_bad_irq+0x48/0xf8
note_interrupt+0x2f4/0x340
handle_irq_event+0xac/0xe0
handle_simple_irq+0xa8/0xfc
handle_irq_desc+0x30/0x54
generic_handle_irq+0x1c/0x24
airoha_irq_handler+0x90/0xe0
__handle_irq_event_percpu+0x44/0x11c
handle_irq_event+0x40/0xe0
handle_fasteoi_irq+0xb4/0x240
handle_irq_desc+0x30/0x54
generic_handle_domain_irq+0x18/0x20
gic_handle_irq+0x60/0xec
do_interrupt_handler+0x4c/0x84
el1_interrupt+0x30/0x4c
el1h_64_irq_handler+0x14/0x1c
el1h_64_irq+0x68/0x6c
handle_softirqs+0x98/0x210
__do_softirq+0x10/0x18
____do_softirq+0xc/0x20
call_on_irq_stack+0x30/0x50
do_softirq_own_stack+0x18/0x20
irq_exit_rcu+0x80/0xb8
el1_interrupt+0x34/0x4c
el1h_64_irq_handler+0x14/0x1c
el1h_64_irq+0x68/0x6c
default_idle_call+0x24/0x34
do_idle+0x98/0xf0
cpu_startup_entry+0x30/0x38
kernel_init+0x0/0x130
console_on_rootfs+0x0/0x64
__primary_switched+0x80/0x88
handlers:
[<000000002b75be58>] irq_default_primary_handler threaded
[<000000006d87ada6>] phy_interrupt
Disabling IRQ #53
This is not a problem with the rtl8261n driver because it is not
registering an interrupt handler. But with the kernel realtek PHY driver, a
interrupt handler is registered which can trigger this problem on bootup.
To avoid is, disable the interrupt and use the PHY polling mode also wit
the upcoming realtek PHY driver support for RTL8261.
Co-authored-by: Aleksander Jan Bajkowski <olek2@wp.pl>
Co-authored-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Daniel Pawlik <pawlik.dan@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/23078
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Some W1700K and XR1701G boards with Realtek RTL8261N/RTL8261BE 10G PHYs
fail to bring up the USXGMII link on cold boot. The PHY enters a bad
state during initialization and the link stays down permanently until
power cycle.
Root cause: the GPIO reset assert/deassert timing (40ms/150ms) is too
short for the RTL8261N to complete its internal firmware load. The OEM
firmware uses 200ms/200ms.
Increase both PHY reset timings to 200ms/200ms to match OEM values.
Confirmed to fix intermittent boot failures on both W1700K (Gemtek)
and XR1701G boards.
Signed-off-by: Ryan Chen <rchen14b@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/22564
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
With new PCS implementation, the pcs property needs to be updated to the
new name of 'pcs-handle' to correctly work with PCS fwnode
consumer/provider.
Fixes: 4d5f579a81 ("airoha: update PCS node in DTSI for new PCS implementation")
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Update the PCS node in AN7581/AN7583 .dtsi for new PCS implementation.
The #pcs-cell is now needed for the produced/consumer implementation.
Also add entry for USB and PCIe PCS for AN7581 but keep them disabled
by default. USB and PCIe PCS for AN7583 node will be added later once PHY
code will stabilize.
Link: https://github.com/openwrt/openwrt/pull/23271
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Migrate Airoha PCS/Ethernet pending patch to PCS standalone implementation.
This new implementation drop the hack of reading and accessing the dev from
a different device and drop the legacy pcs_create/drop implementation in
favor of fwnode one with a provider/consumer approach.
This is also to sync with the proposed series posted upstream for revision.
The new PCS patch for AN7581 implement full support for USB and PCIe PCS.
Link: https://github.com/openwrt/openwrt/pull/23271
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Backport pending PCS standalone feature for kernel 6.12 and all the
required dependency patch.
All affected patch automatically refreshed.
Link: https://github.com/openwrt/openwrt/pull/23271
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
The vendor firmware checks for a bmt header in the last 528 erase blocks
of flash. The OpenWrt partition table did not respect that requirement,
and therefore the vendor and openwrt chainloader fight over those blocks
on every boot, potentially corrupting data stored in UBI blocks there.
This commit increases the size of the reserved_bmt partition to avoid
that fight.
Although the vendor bootloader only seems to touch the final 250 erase
blocks[1], the original vendor firmware system partition ended at
0x1be00000[2], so to be conservative, the consensus is to use that as
the end of mtd2 (ubi) partition and leave the last 528 blocks for mtd3
(reserved_bmt).
From https://openwrt.org/toh/gemtek/mxf-w1700k:
[1] OEM bootlog: [ 5.324337] bmt pool size: 250
[2] OEM bootlog: [ 5.478927] 0x000008600000-0x00001be00000 : "system"
Adds a compat_version to warn users to re-install to accomodate the
shrunken mtd2 ubi partition.
Fixed two nits suggested by Claude, zero padding a hex value and removed
an extra space.
Removed the wildcard setting of compat_version for other boards, as
suggested by Robert Marko, set compat_version only for the w1700k.
Reported-by: Loïc Yhuel <loic.yhuel@gmail.com>
Signed-off-by: Russell Senior <russell@personaltelco.net>
Link: https://github.com/openwrt/openwrt/pull/23061
Signed-off-by: Robert Marko <robimarko@gmail.com>
Disable IPSEC by default. If needed, install required packages.
Reduce compressed image size by 80 kB.
Signed-off-by: Aleksander Jan Bajkowski <olek2@wp.pl>
Link: https://github.com/openwrt/openwrt/pull/23269
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Airoha reported some bug in the TX/RX descriptor handling and PPE. Backport
the fix for such bug merged in net staging tree.
It's expected that these patch will be dropped in future minor kernel
version when submitted to stable staging tree.
All affected patch automatically refreshed.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
The audio should only be enabled when the sound
node is enabled. This fixes the following error:
an7581-audio 1fbe2200.afe: probe with driver an7581-audio failed with error -2
Fixes: 7b55651 ("airoha: enable I2S sound driver and add nodes for eMMC RFB board")
Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>
Link: https://github.com/openwrt/openwrt/pull/22660
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
The Kite EVB device wrongly select an invalid package for NPU firmware
where the correct one should be the airoha-en7581-npu-firmware one.
Fix the wrong package to restore compilation of the Airoha target.
Fixes: 0cf516751a ("airoha: an7581: generalize eMMC DTS and add Kite variant")
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Backport patch merged upstream that optimize the QDMA rx queue descriptor
setup by configuring the CPU IDX only when needed.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
It seems that upstream one of the patch had a compilation erro caused by
merging net and net-next and it was fixed silently in a merge commit.
Fix this error in the affected patch.
Fixes: 155c610962 ("airoha: backport additional patch for memory leak and multi-serdes rework")
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Backport upstream memory leak patch merged upstream and even more
preliminary patch for multi-serdes rewrk.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
The Airoha AN7581 EVB board switched to full UBI support. Update the
partition table to reflect the new implementation.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Generalize the eMMC DTS to common node and add a variant with the Kite WiFi
chip installed. Also rename the eMMC model name and add the Eagle name to
it.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Add specific DTSI for NPU firmware based on the supported WiFi chip and
update the supported device by including the relevant new DTSI.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
This symbol is selected by CONFIG_BPF, which was already enabled
on generic config-6.12.
Signed-off-by: Shiji Yang <yangshiji66@outlook.com>
Link: https://github.com/openwrt/openwrt/pull/22730
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
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>
Drop airoha,x2-mode related property from the DTSI and enable them only in
the relevant devices that use a PCIe0 line. x2-mode should be enabled only
for the WiFi card that would use it as it might cause problem for WiFi card
that expect 2 single PCIe x1 line instead of a single PCIe x2 line.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
