Backport driver from upcoming Linux 6.10 and put a pending fix on top
to make sure the netdev trigger offloading behaves as expected.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
1. Enable this feature only for 32-bit CPUs as MIPS64 can not
access the full range unmapped uncached memory.
2. Backport this fix to the 6.1 old LTS kernel.
Signed-off-by: Shiji Yang <yangshiji66@qq.com>
Importing pending patch "net: dsa: mt7530: move MT753X_MTRAP operations
for MT7530" broke WAN connectivity on most MT7621 which use PHY-muxing
to hook up either port 0 or port 4 to GMAC1.
Remove it for now until the author submits a fixed version.
Fixes: https://github.com/openwrt/openwrt/issues/15279
Fixes: https://github.com/openwrt/openwrt/issues/15273
Fixes: d40691a5fb ("generic: 6.1, 6.6: mt7530: import pending patches")
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
The backports introduced in commit d40756563c ("kernel: backport
phylink changes from mainline Linux") broke the mv88e6xxx DSA driver.
A backport to fix this was added to the kirkwood target, but as it is
used in multiple targets, and there's a kmod package for it, the fix
should be in generic backports.
This fixes the switch on the WatchGuard Firebox M300 when running the
6.1 testing kernel.
There is no need to backport the fix for the 6.6 kernel, as it was
included in 6.6.5.
Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
For packets not belonging to a local socket, use fraglist GRO instead of
regular GRO. This make segmenting packets very cheap and avoids the need for
selectively disabling GRO
Signed-off-by: Felix Fietkau <nbd@nbd.name>
net: dsa: mt7530: explain exposing MDIO bus of MT7531AE better
net: dsa: mt7530: do not pass port variable to mt7531_rgmii_setup()
net: dsa: mt7530: use priv->ds->num_ports instead of MT7530_NUM_PORTS
net: dsa: mt7530: get rid of mac_port_validate member of mt753x_info
net: dsa: mt7530: refactor MT7530_PMEEECR_P()
net: dsa: mt7530: get rid of function sanity check
net: dsa: mt7530: define MAC speed capabilities per switch model
net: dsa: mt7530: return mt7530_setup_mdio & mt7531_setup_common on error
net: dsa: mt7530: move MT753X_MTRAP operations for MT7530
net: dsa: mt7530: refactor MT7530_HWTRAP and MT7530_MHWTRAP
net: dsa: mt7530: refactor MT7530_MFC and MT7531_CFC, add MT7531_QRY_FFP
net: dsa: mt7530: rename mt753x_bpdu_port_fw enum to mt753x_to_cpu_fw
net: dsa: mt7530: rename p5_intf_sel and use only for MT7530 switch
net: dsa: mt7530: refactor MT7530_PMCR_P()
net: dsa: mt7530: disable EEE abilities on failure on MT7531 and MT7988
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Backport patches for support of generic spi-nor from SFDP data for
kernel 6.1.
Kernel 5.15 have major rework of the info flags and it's not trustable
to backport this amount of changes and expect correct function of it.
All affected patches automatically refreshed using make
target/linux/refresh.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Add pending patches to add LED_FUNCTION_MOBILE and LED_FUNCTION_SPEED_*
definitions for Fortinet FortiGate devices and IIJ SA-W2.
Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
First patch allows to inquire and modify Energy-Efficient-Ethernet
(EEE) settings via ethtool and thereby override the default setting of
a board done via bootstrap pins.
The second patch fixes a long-standing issue with STP (and similar
protocols) when using boards (or SoCs) governed by the mt7530 DSA
driver.
Both patches could also be (dirty-)applied to Linux 5.15, but I'd
rather just wait for that to happen via linux-stable to avoid the
mess.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Backport lots upstream changes, many of them fixes, for the mt7530 DSA
driver. Some of them may or may not find they way into Linux 6.1
stable, some certainly won't because they are fixes for backported
commits which aren't even present in Linux 6.1 upstream.
Apart from adding new patches, also remove mutated patch
723-net-mt7531-ensure-all-MACs-are-powered-down-before-r.patch
which should never have been added for Linux 6.1 -- it was applied
already upstream but coincidentally would fuzzy-apply in the wrong
place as well (for MT7530 instead of MT7531). While that didn't really
hurt anyone it is just unneeded.
The other deleted patch
795-mt7530-register-OF-node-for-internal-MDIO-bus.patch
has been replaced by an equivalent commit with a more complete patch
description by upstream maintainer Arınç Ünal.
The remaining differences compared to the upstream driver are:
* C22/C45 MDIO ops aren't split
Upstream did that, backporting it would require making changes to
*all* DSA drivers
* 'slave' -> 'user', 'master' -> 'conduit' language change in DSA
* support for selecting preferred CPU port on MT7531
Also this would require too many DSA framework changes potentially
affecting other devices. If we ever really use Linux 6.1 in a
release (I hope not) we can still reconsider to make the effort to
backport that.
In addition to some minor bug fixes and style improvements the switch
should now behave more conformant when it comes to link-local frames,
and we will again be able to cleanly pick patches from upstream.
MAINTAIERS NOTE:
Three patches are already part of Linux stable and should be removed with
the next minor kernel version bump:
789-STABLE-01-net-dsa-mt7530-prevent-possible-incorrect-XTAL-frequ.patch
789-STABLE-02-net-dsa-mt7530-fix-link-local-frames-that-ingress-vl.patch
789-STABLE-03-net-dsa-mt7530-fix-handling-of-all-link-local-frames.patch
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Make sure all patches can be applied to a git tree using 'git am'
by adding missing patch headers where needed.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Fix issue of transmitting abnormal data which leads to link problems
in 1G and 2.5G SerDes modes (SGMII, 1000Base-X, 2500Base-X) on the
MediaTek MT7988 SoC.
Link: b72d6cba92
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
A patch to resolve an issue was found in MediaTek's GPL-licensed SDK:
In the mtk_ppe_stop() function, the PPE scan mode is not disabled before
disabling the PPE. This can potentially lead to a hang during the process
of disabling the PPE.
Without this patch, the PPE may experience a hang during the reboot test.
Reference: b40da332df
Suggested-by: Bc-bocun Chen <bc-bocun.chen@mediatek.com>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Clearing bit MAC_MCR_FORCE_LINK which forces the link down too early
can result in MAC ending up in a broken/blocked state.
Fix this by handling this bit in the .mac_link_up and .mac_link_down
calls instead of in .mac_finish.
Suggested-by: Mason-cw Chang <Mason-cw.Chang@mediatek.com>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
ath25 has been on life support for the last couple of releases, eventually
leading to marking it as source-only in 2023.
It has been basically only touched to do a kernel bump so that we can make
the new OpenWrt release which was a challenge due to small RAM amount.
However, with the attempt of kernel 6.1 update it turns out that kernel
cannot even finish booting due to RAM constraints, so its time to let this
target go.
Signed-off-by: Robert Marko <robimarko@gmail.com>
Drop the flow-hash of the skb when forwarding to the L2TP netdev.
This avoids the L2TP qdisc from using the flow-hash from the outer
packet, which is identical for every flow within the tunnel.
This does not affect every platform but is specific for the ethernet
driver. It depends on the platform including L4 information in the
flow-hash.
Signed-off-by: David Bauer <mail@david-bauer.net>
When building the UBI NVMEM provider on 32-bit platforms a compiler
warning is triggered due to different sizeof(int).
Fix this by using integer types with well-defined size.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Kernel warns about comparision of different types without cast when
building the fitblk driver on 32-bit platforms.
Fix this by using `min_t(size_t, ...`.
Fixes: 8fc5457869 ("kernel: add pending fitblk uImage.FIT sub-image block driver")
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Use device_set_node to make sure OF node gets assigned on block
devices to be used as NVMEM providers. While block partitions were
already working fine as NVMEM providers, bare block devices such as
mmcblk0boot1 will not work without this change.
Fixes: fc153aa8d9 ("kernel: import pending patches adding support for NVMEM on UBI and MMC")
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Add PHY driver for Airoha EN8811H PHY and package it as kernel module.
The PHY needs to load firmware from rootfs, so there is no point in
having the driver built-into the kernel.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Replace previous patch adding paths and SerDes modes with patch series
pending upstream adding dedicated drivers for XFI T-PHY and USXGMII PCS,
extends LynxI PCS to be a standalone platform driver and as a consequence
makes much less changes to the actual Ethernet driver mtk_eth_soc.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Let's pick a bunch of useful phylink changes which allow us to keep
drivers in sync with mainline Linux.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Add 'fitblk' driver to replace the rejected/deprecated uImage.FIT
partition parser.
To use the new driver, add phandle /chosen/rootdisk and point it to
the MTD partition, UBI volume or block device holding the uImage.FIT.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Similar to supporting nvmem-layouts on MTD devices, also allow referencing
UBI and MMC devices in DT.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Backport two commits from Linux 6.3 wiring up device node parents of
ubi devices (pointing to their MTD parent) as well as ubiblock devices
(poiting to their parent UBI volume).
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Now that QCA807x interface mode check was upstreamed, use the upstreamed
version and mark it as such.
Signed-off-by: Robert Marko <robimarko@gmail.com>
Currently, we are checking whether the PHY package mode matches the
individual PHY interface modes at PHY package probe time, but at that time
we only know the PHY package mode and not the individual PHY interface
modes as of_get_phy_mode() that populates it will only get called once the
netdev to which PHY-s are attached to is being probed and thus this check
will always fail and return -EINVAL.
So, lets move this check to .config_init_once as at that point individual
PHY interface modes should be populated.
Fixes: 16364e4100 ("generic: 6.1: backport QCA807x PHY patches")
Signed-off-by: Robert Marko <robimarko@gmail.com>
Backport QCA807x PHY patches merged upstream that introduce the new
concept of PHY package.
Also add in generic config the new Kconfig CONFIG_QCA807X_PHY.
All affected patch automatically refreshed with make
target/linux/refresh.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
