openwrt/staging/blocktrron.git
3 years agoRevert "swconfig: fix Broadcom b53 support"
Rafał Miłecki [Mon, 4 Oct 2021 05:12:36 +0000 (07:12 +0200)]
Revert "swconfig: fix Broadcom b53 support"

This reverts commit 8f9cd1af0f9c325a902dbd0e79e12015372e6bb0.

That commit was meant to add a single EXPORT_SYMBOL_GPL() but it
actually also added few .of_match_table-s. One commit should handle one
thing and should not introduce unrelated changes.

Regarding actual changes:
1. EXPORT_SYMBOL_GPL is not required as we don't build swconfig drivers
   as modules.
2. PHY drivers must not have .of_match_table. That is allowed for MDIO
   drivers. This could work for some time (although is didn't for me on
   bcm53xx) but does not with kernel 5.10. It causes a soft lockup and
   upstream developers confirmed it's an unsupported design.

Link: https://lore.kernel.org/netdev/2b1dc053-8c9a-e3e4-b450-eecdfca3fe16@gmail.com/t/#mf80e472f35ee23f7a75cbf5b1e101a17ab3a64a3
Cc: Tobias Schramm <tobleminer@gmail.com>
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
3 years agokernel: backport bgmac upstream commits from 5.15 / for 5.16
Rafał Miłecki [Thu, 7 Oct 2021 09:29:52 +0000 (11:29 +0200)]
kernel: backport bgmac upstream commits from 5.15 / for 5.16

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
3 years agompc85xx: switch to Kernel 5.10
Adrian Schmutzler [Sat, 2 Oct 2021 16:26:01 +0000 (18:26 +0200)]
mpc85xx: switch to Kernel 5.10

This has testing support for 7 months. Time to switch.

TL-WDR4900 is disabled due to kernel size limitation.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agokernel: 5.10: add missing symbols
Pawel Dembicki [Mon, 4 Oct 2021 20:30:58 +0000 (22:30 +0200)]
kernel: 5.10: add missing symbols

Three missing symbols were found during mpc85xx/p2020 compilation.

While at it, CONFIG_FSL_ENETC_MDIO is moved to generic config
for consistency.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
[move CONFIG_FSL_ENETC_MDIO, remove redundant definitions, adjust
commit message/title]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agobcm27xx: remove obsolete kernel 5.4
Adrian Schmutzler [Tue, 5 Oct 2021 18:51:18 +0000 (20:51 +0200)]
bcm27xx: remove obsolete kernel 5.4

With the upgrade to kernel 5.10 per default the old version is no
longer required to be in tree.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agofirmware-utils: bump to git HEAD
Kevin Darbyshire-Bryant [Tue, 5 Oct 2021 20:22:36 +0000 (21:22 +0100)]
firmware-utils: bump to git HEAD

f9ad6b3 Add more missing includes for byte swap operations

Basically stop it exploding on MacOS

Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
3 years agokernel: drop kmod-gpio-dev
Stijn Tintel [Fri, 1 Oct 2021 02:02:24 +0000 (05:02 +0300)]
kernel: drop kmod-gpio-dev

The GPIO_DEVICE symbol belonged to a custom driver that was removed from
OpenWrt in 2012. The symbol never existed in the upstream kernel.

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
3 years agobcm27xx: add RPI_AXIPERF symbol
Stijn Tintel [Thu, 30 Sep 2021 17:26:18 +0000 (20:26 +0300)]
bcm27xx: add RPI_AXIPERF symbol

When KERNEL_PERF_EVENTS is enabled in OpenWrt, or PERF_EVENTS in the
kernel config, the RPI_AXIPERF is exposed. Add it to the subtarget
kernel configs to avoid build failures.

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
3 years agobcm27xx: add SND_SOC_AD193X_{I2C,SPI} symbols
Stijn Tintel [Thu, 30 Sep 2021 15:36:23 +0000 (18:36 +0300)]
bcm27xx: add SND_SOC_AD193X_{I2C,SPI} symbols

When building an image for the bcm27xx target, some combinations of
config options will fail to build due the SND_SOC_AD193X_I2C and
SND_SOC_AD193X_SPI kernel config symbols being missing.

The problem only occurs on bcm27xx as the target contains a patch that
modifies the Kconfig file containing the symbols; in the vanilla kernel,
there is no string after the tristate keyword so the symbol is not
exposed.

The _I2C symbol depends on I2C, which is enabled in the kernel configs
of all bcm27xx subtargets.

The _SPI symbol depends on SPI_MASTER, which is exposed by either
kmod-mmc-spi, kmod-spi-bitbang, kmod-spi-dev, kmod-spi-bcm2835 or
kmod-spi-bcm2835-aux.

Both symbols are defined in the sound/soc/codecs directory, which is
only included when SND_SOC is enabled, so this problem doesn't occur
when kmod-sound-soc-core is not enabled. As the
kmod-sound-soc-bcm2835-i2s package disables the SND_SOC_AD193X_SPI
symbol, it also doesn't occur when kmod-sound-soc-bcm2835-i2s is
enabled.

As there are several possible config combinations that do exhibit this
problem, it is best to solve it by adding the missing symbols to the
subtarget kernel configs. By doing this we can remove them from the
kmod-sound-soc-bcm2835-i2s package.

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
3 years agotools/cmake: update to 3.21.3
Rosen Penev [Tue, 5 Oct 2021 01:30:33 +0000 (18:30 -0700)]
tools/cmake: update to 3.21.3

Signed-off-by: Rosen Penev <rosenp@gmail.com>
3 years agotools/bison: update to 3.8.2
Rosen Penev [Tue, 5 Oct 2021 01:30:12 +0000 (18:30 -0700)]
tools/bison: update to 3.8.2

Signed-off-by: Rosen Penev <rosenp@gmail.com>
3 years agobusybox: update to 1.34.1
Hannu Nyman [Sun, 3 Oct 2021 18:23:04 +0000 (21:23 +0300)]
busybox: update to 1.34.1

Update busybox to version 1.34.1, which is a minor
maintenance release. It contains just the two post-1.34.0
upstream patches that we earlier backported plus a few fixes
to awk.

* Remove the two backported upstream patches that are
  now unnecessary.

Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
3 years agofirmware-utils: update to version 2021-10-05
Petr Štetiar [Mon, 5 Jul 2021 09:54:26 +0000 (11:54 +0200)]
firmware-utils: update to version 2021-10-05

Includes following changes:

 db65821f006c cmake: fix missing install target
 3a0cfc856991 Add initial GitLab CI support
 8f47adea6f87 Add missing includes for byte swap operations
 fbafae9f8037 Convert to CMake based project

Additionaly moves source code into separate Git project repository and
converts the package build to utilize CMake.

Signed-off-by: Petr Štetiar <ynezz@true.cz>
[rmilecki: rebase, update to the latest repo git & rm -r src]
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
3 years agoramips: fix USW-Flex reversed switch-port order
David Bauer [Mon, 4 Oct 2021 15:53:10 +0000 (17:53 +0200)]
ramips: fix USW-Flex reversed switch-port order

Switch port order was reversed due to reading the internal labling
(which mismatches the one on the case).

Signed-off-by: David Bauer <mail@david-bauer.net>
3 years agouboot-mediatek: update to 2021.10
Daniel Golle [Thu, 19 Aug 2021 02:57:00 +0000 (03:57 +0100)]
uboot-mediatek: update to 2021.10

U-Boot 2021.10 has been released.
Rebase mediatek patches on top of new release and remove some patches
which have been merged upstream.

Tested on Bananapi BPi-R2 (mt7623), Bananapi BPi-R64 (mt7622) and
Linksys E8450 (mt7622).

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
3 years agodnsmasq: add explicit "set:" for client-matching options
Paul Fertser [Sat, 25 Sep 2021 11:00:23 +0000 (14:00 +0300)]
dnsmasq: add explicit "set:" for client-matching options

Bring the usage in line with the dnsmasq man page and the other options
where set: is mandatory.

No functional change.

Signed-off-by: Paul Fertser <fercerpav@gmail.com>
3 years agobuild: fix `which` detection on Fedora & MacOS
Paul Spooren [Sat, 2 Oct 2021 01:56:27 +0000 (15:56 -1000)]
build: fix `which` detection on Fedora & MacOS

Fix Fedora 34/35 issue where 'which' detection of 'which' wasn't working
because Fedora use alias and proc

Fixup of fca5ad55d2 prereq-build: fix `which` detection on Fedora

Reported-by: Jani Partanen <rtfm@iki.fi>
Suggest-by: Etienne Champetier <champetier.etienne@gmail.com>
Tested-by: Georgi Valkov <gvalkov@abv.bg>
Signed-off-by: Paul Spooren <mail@aparcar.org>
3 years agokernel: Deactivate some ARM64 errata workarounds
Hauke Mehrtens [Sat, 2 Oct 2021 22:42:49 +0000 (00:42 +0200)]
kernel: Deactivate some ARM64 errata workarounds

This deactivates the following workarounds for erratas in ARM64 CPUS:
CONFIG_ARM64_ERRATUM_1165522: Cortex-A76 cores (r0p0, r1p0, r2p0)
CONFIG_ARM64_ERRATUM_1286807: Cortex-A76 cores (r0p0 to r3p0)
CONFIG_ARM64_ERRATUM_1418040: Cortex-A76/Neoverse-N1 cores (r0p0 to r3p1)
CONFIG_CAVIUM_TX2_ERRATUM_219: Cavium ThunderX2
CONFIG_FUJITSU_ERRATUM_010001: Fujitsu-A64FX

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
3 years agosunxi: Refresh kernel 5.10 configuration
Hauke Mehrtens [Sat, 2 Oct 2021 22:12:05 +0000 (00:12 +0200)]
sunxi: Refresh kernel 5.10 configuration

The cortex53 subtarget was missing multiple config options and the other
targets just defined some options which are not needed.

Fixes: 83672f506d95 ("sunxi: add testing Linux 5.10")
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
3 years agosunxi: Activate CONFIG_SATA_PMP
Hauke Mehrtens [Sat, 2 Oct 2021 20:03:04 +0000 (22:03 +0200)]
sunxi: Activate CONFIG_SATA_PMP

The kernel 5.4 configuration activated SATA Port Multiplier support, add
it to kernel 5.10 too.

Fixes: 83672f506d95 ("sunxi: add testing Linux 5.10")
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
3 years agokirkwood: refresh kernel config
Sungbo Eo [Sun, 20 Jun 2021 13:59:06 +0000 (22:59 +0900)]
kirkwood: refresh kernel config

Refresh config with make kernel_oldconfig.

Tested-by: Pawel Dembicki <paweldembicki@gmail.com>
Signed-off-by: Sungbo Eo <mans0n@gorani.run>
3 years agokirkwood: solidify SATA_PMP config
Sungbo Eo [Sun, 20 Jun 2021 13:51:50 +0000 (22:51 +0900)]
kirkwood: solidify SATA_PMP config

SATA_PMP option is no longer exposed when no SATA host driver is enabled
since upstream linux commit bd322af15ce9 ("ata: make SATA_PMP option
selectable only if any SATA host driver is enabled").

Commit 1bb3f593ee73 ("kirkwood: update config for kernel 5.10") manually
added CONFIG_SATA_PMP=y to config file, but the config will disappear for
every kernel_oldconfig refresh.

To prevent this, a small hack is added, which selects SATA_HOST
automatically when SATA_PMP is enabled. This patch can be dropped if
SATA_MV is ever re-added into kernel config file.

Tested-by: Pawel Dembicki <paweldembicki@gmail.com>
Signed-off-by: Sungbo Eo <mans0n@gorani.run>
[Move patch to generic/hack-5.10]
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
3 years agoath79: add recipe for common setup with loader-okli-compile
Adrian Schmutzler [Sun, 26 Sep 2021 14:07:44 +0000 (16:07 +0200)]
ath79: add recipe for common setup with loader-okli-compile

These instructions are repeated for a few devices now, let's move
them to shared definition so we do not repeat ourselves too often.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agotreewide: use AUTORELEASE on all uboot-* packages
Adrian Schmutzler [Sat, 2 Oct 2021 17:03:32 +0000 (19:03 +0200)]
treewide: use AUTORELEASE on all uboot-* packages

Nobody ever updates PKG_RELEASE when changing devices or setup in
the various uboot-* packages. Use $(AUTORELEASE) so we still have
proper versioning there.

Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agorealtek: ensure output drivers are enabled in RTL8231
Paul Fertser [Fri, 1 Oct 2021 09:37:11 +0000 (12:37 +0300)]
realtek: ensure output drivers are enabled in RTL8231

The bootloader can leave the GPIO expander in a state which doesn't have
output drivers enabled so GPIOs will properly work for input but output
operations will have no effect.

To avoid disrupting the boot in case the bootloader left direction and
data registers in an inconsistent state (e.g. pulling SoC's reset to 0)
reconfigure everything as input.

Reviewed-by: Sander Vanheule <sander@svanheule.net>
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
3 years agobcm4908img: detect Linksys images
Rafał Miłecki [Fri, 10 Sep 2021 14:34:48 +0000 (16:34 +0200)]
bcm4908img: detect Linksys images

Linksys uses an extra 0x100 bytes long tail for BCM4908 images.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
3 years agobcm4908img: store offset of tail data
Rafał Miłecki [Fri, 10 Sep 2021 14:34:47 +0000 (16:34 +0200)]
bcm4908img: store offset of tail data

This simplifies some operations as it doesn't have to be caculated over
and over. It will also allow adding support for more vendor formats.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
3 years agomvebu: remove obsolete Kernel 5.4
Paul Spooren [Fri, 1 Oct 2021 21:28:36 +0000 (11:28 -1000)]
mvebu: remove obsolete Kernel 5.4

With the upgrade to Kernel 5.10 per default the old version is no longer
required to be in tree.

Signed-off-by: Paul Spooren <mail@aparcar.org>
3 years agoipq806x: remove obsolete Kernel 5.4
Paul Spooren [Fri, 1 Oct 2021 21:27:12 +0000 (11:27 -1000)]
ipq806x: remove obsolete Kernel 5.4

With the upgrade to Kernel 5.10 per default the old version is no longer
required to be in tree.

Signed-off-by: Paul Spooren <mail@aparcar.org>
3 years agogemini: remove obsolete Kernel 5.4
Paul Spooren [Fri, 1 Oct 2021 21:25:57 +0000 (11:25 -1000)]
gemini: remove obsolete Kernel 5.4

With the upgrade to Kernel 5.10 per default the old version is no longer
required to be in tree.

Signed-off-by: Paul Spooren <mail@aparcar.org>
3 years agomalta: remove obsolete Kernel 5.4
Paul Spooren [Fri, 1 Oct 2021 21:17:01 +0000 (11:17 -1000)]
malta: remove obsolete Kernel 5.4

With the upgrade to Kernel 5.10 per default the old version is no longer
required to be in tree.

Signed-off-by: Paul Spooren <mail@aparcar.org>
3 years agoarmvirt: remove obsolete Kernel 5.4
Paul Spooren [Fri, 1 Oct 2021 21:15:41 +0000 (11:15 -1000)]
armvirt: remove obsolete Kernel 5.4

With the upgrade to Kernel 5.10 per default the old version is no longer
required to be in tree.

Signed-off-by: Paul Spooren <mail@aparcar.org>
3 years agox86: remove obsolete Kernel 5.4
Paul Spooren [Fri, 1 Oct 2021 21:13:07 +0000 (11:13 -1000)]
x86: remove obsolete Kernel 5.4

With the upgrade to Kernel 5.10 per default the old version is no longer
required to be in tree.

Signed-off-by: Paul Spooren <mail@aparcar.org>
3 years agokernel: bump 5.4 to 5.4.150
Hauke Mehrtens [Sat, 2 Oct 2021 13:28:55 +0000 (15:28 +0200)]
kernel: bump 5.4 to 5.4.150

Manually rebased:
  generic/pending-5.4/800-bcma-get-SoC-device-struct-copy-its-DMA-params-to-th.patch
  mvebu/patches-5.4/021-arm64-dts-marvell-armada-37xx-Move-PCIe-comphy-handl.patch

Removed upstreamed:
  layerscape/patches-5.4/819-uart-0004-MLK-18137-fsl_lpuart-Fix-loopback-mode.patch

All others updated automatically.

Compile-tested on: lantiq/xrx200, armvirt/64
Runtime-tested on: lantiq/xrx200, armvirt/64

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
3 years agomvebu: add Globalscale MOCHAbin
Robert Marko [Mon, 27 Sep 2021 21:14:51 +0000 (23:14 +0200)]
mvebu: add Globalscale MOCHAbin

Globalscale MOCHAbin is a Armada 7040 based development board.

Specifications:
* Armada 7040 Quad core ARMv8 Cortex A-72 @ 1.4GHz
* 2 / 4 / 8 GB of DDR4 DRAM
* 16 GB eMMC
* 4MB SPI-NOR (Bootloader)
* 1x M.2-2280 B-key socket (for SSD expansion, SATA3 only)
* 1x M.2-2250 B-key socket (for modems, USB2.0 and I2C only)
* 1x Mini-PCIe 3.0 (x1, USB2.0 and I2C)
* 1x SATA 7+15 socket (SATA3)
* 1x 16-pin (2×8) MikroBus Connector
* 1x SIM card slot (Connected to the mini-PCIe and both M.2 slots)
* 2x USB3.0 Type-A ports via SMSC USB5434B hub
* Cortex 2x5 JTAG
* microUSB port for UART (PL2303GL/PL2303SA onboard)
* 1x 10G SFP+
* 1x 1G SFP (Connected to 88E1512 PHY)
* 1x 1G RJ45 with PoE PD (Connected to 88E1512 PHY)
* 4x 1G RJ45 ports via Topaz 88E6141 switch
* RTC with battery holder (SoC provided, requires CR2032 battery)
* 1x 12V DC IN
* 1x Power switch
* 1x 12V fan header (3-pin, power only)
* 1x mini-PCIe LED header (2x0.1" pins)
* 1x M.2-2280 LED header (2x0.1" pins)
* 6x Bootstrap jumpers
* 1x Power LED (Green)
* 3x Tri-color RGB LEDs (Controllable)
* 1x Microchip ATECC608B secure element

Note that 1G SFP and 1G WAN cannot be used at the same time as they are in
parallel connected to the same PHY.

Installation:

Copy dtb from build_dir to bin/ and run tftpserver there:
$ cp ./build_dir/target-aarch64_cortex-a72_musl/linux-mvebu_cortexa72/image-armada-7040-mochabin.dtb bin/targets/mvebu/cortexa72/
$ in.tftpd -L -s bin/targets/mvebu/cortexa72/

Connect to the device UART via microUSB port and power on the device.

Power on the device and hit any key to stop the autoboot.

Set serverip (host IP) and ipaddr (any free IP address on the same subnet), e.g:
$ setenv serverip 192.168.1.10 # Host
$ setenv ipaddr 192.168.1.15 # Device

Set the ethernet device (Example for the 1G WAN):
$ setenv ethact mvpp2-2

Ping server to confirm network is working:
$ ping $serverip
Using mvpp2-2 device
host 192.168.1.15 is alive

Tftpboot the firmware:
$ tftpboot $kernel_addr_r openwrt-mvebu-cortexa72-globalscale_mochabin-initramfs-kernel.bin
$ tftpboot $fdt_addr_r image-armada-7040-mochabin.dtb

Boot the image:
$ booti $kernel_addr_r - $fdt_addr_r

Once the initramfs is booted, transfer openwrt-mvebu-cortexa72-globalscale_mochabin-squashfs-sdcard.img.gz
to /tmp dir on the device.

Gunzip and dd the image:
$ gunzip /tmp/openwrt-mvebu-cortexa72-globalscale_mochabin-squashfs-sdcard.img.gz
$ dd if=/tmp/openwrt-mvebu-cortexa72-globalscale_mochabin-squashfs-sdcard.img of=/dev/mmcblk0 && sync

Reboot the device.

Hit any key to stop the autoboot.

Reset U-boot env and set the bootcmd:
$ env default -a
$ setenv bootcmd 'load mmc 0 ${loadaddr} boot.scr && source ${loadaddr}'

Optionally I would advise to edit the console env variable to remove earlycon as that
causes the kernel to never use the driver for the serial console.
Earlycon should be used only for debugging before the kernel can configure the console
and will otherwise cause various issues with the console.

$ setenv console 'console=ttyS0,115200'

Save and reset
$ saveenv
$ reset

OpenWrt should boot from eMMC now.

Signed-off-by: Robert Marko <robert.marko@sartura.hr>
3 years agolantiq: kernel: xway-nand: Move ECC engine setting to new DT binding
Daniel Kestrel [Sat, 7 Aug 2021 07:28:24 +0000 (09:28 +0200)]
lantiq: kernel: xway-nand: Move ECC engine setting to new DT binding

When trying to add support for another device with Micron NAND chips,
it was discovered that the default setting in the kernel source does
not work with Micron Chips, since the device trees setting is
overwritten and hard coded by the kernel xway_nand driver. This was
the original reason for this PR.
A kernel patch sets the default ECC mode to soft without overwriting
the device tree settings and the device tree for devices using it
are updated with new parameters because the old ones are deprecated
by torvalds/linux@533af69.
A patch for kernel 5.4 is provided to support the new settings
because kernel 5.4 does not support it.

Signed-off-by: Daniel Kestrel <kestrel1974@t-online.de>
3 years agolantiq: kernel: xway-nand: Fix setting on-die ECC engines in dts
Hauke Mehrtens [Sat, 2 Oct 2021 12:03:17 +0000 (14:03 +0200)]
lantiq: kernel: xway-nand: Fix setting on-die ECC engines in dts

This backports a fix proposed for upstream kernel to fix overwriting
the NAND ECC engine in device tree.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
3 years agouboot-lantiq: fix sha1.h header clash when system libmd installed
Alan Swanson [Fri, 1 Oct 2021 15:46:32 +0000 (16:46 +0100)]
uboot-lantiq: fix sha1.h header clash when system libmd installed

Backport of u-boot commit "includes: move openssl headers to include/u-boot"
https://github.com/u-boot/u-boot/commit/2b9912e6a7df7b1f60beb7942bd0e6fa5f9d0167

Fixes: FS#3955
Signed-off-by: Alan Swanson <reiver@improbability.net>
3 years agomediatek: mt7623: bpi-r2: add LED aliases
Daniel Golle [Fri, 1 Oct 2021 22:25:48 +0000 (23:25 +0100)]
mediatek: mt7623: bpi-r2: add LED aliases

Add aliases for LEDs in device tree of the BPi-R2.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
3 years agomediatek: remove files and patches for Linux 5.4
Daniel Golle [Fri, 1 Oct 2021 22:24:06 +0000 (23:24 +0100)]
mediatek: remove files and patches for Linux 5.4

All subtargets only provide files and patches for Linux 5.10 by now
so there is little use for the old Linux 5.4 stuff. Remove it.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
3 years agoarmvirt: switch to Kernel 5.10
Paul Spooren [Thu, 30 Sep 2021 05:15:05 +0000 (19:15 -1000)]
armvirt: switch to Kernel 5.10

Armvirt is a development and testing platform and should therefore use
the latest OpenWrt Kernel by default.

Tested via qemu.

Acked-by: Rosen Penev <rosenp@gmail.com>
Acked-by: Yousong Zhou <yszhou4tech@gmail.com>
Signed-off-by: Paul Spooren <mail@aparcar.org>
3 years agomalta: switch to Kernel 5.10
Paul Spooren [Thu, 30 Sep 2021 05:19:51 +0000 (19:19 -1000)]
malta: switch to Kernel 5.10

Malta is a development and testing platform and should therefore use the
latest OpenWrt Kernel by default.

Acked-by: Rosen Penev <rosenp@gmail.com>
Acked-by: Yousong Zhou <yszhou4tech@gmail.com>
Signed-off-by: Paul Spooren <mail@aparcar.org>
3 years agoprereq-build: revert "fix `which` detection on Fedora"
Paul Spooren [Fri, 1 Oct 2021 04:26:14 +0000 (18:26 -1000)]
prereq-build: revert "fix `which` detection on Fedora"

This reverts commit fca5ad55d292f8e481406123d0e95b3c85d39733.

Causes problems on MacOS host systems.

Signed-off-by: Paul Spooren <mail@aparcar.org>
3 years agoprereq-build: fix `which` detection on Fedora
Jani Partanen [Thu, 30 Sep 2021 20:51:37 +0000 (23:51 +0300)]
prereq-build: fix `which` detection on Fedora

Fix Fedora 34/35 issue where 'which' detection of 'which' wasn't working
because Fedora use alias and proc

Signed-off-by: Jani Partanen <rtfm@iki.fi>
[fix commit subject and message]
Signed-off-by: Paul Spooren <mail@aparcar.org>
3 years agomediatek: add FEATURES and order them alphabetically
Daniel Golle [Wed, 15 Sep 2021 18:10:46 +0000 (19:10 +0100)]
mediatek: add FEATURES and order them alphabetically

Add common features 'gpio', 'nand', 'pci', 'pcie', 'squashfs' and 'usb'
for all mediatek targets, add 'display' and 'usbgadget' for MT7623.
Sort features alphabetically while at it.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
3 years agouboot-mediatek: fix and make use of LEDs on BPi-R2
Daniel Golle [Sat, 18 Sep 2021 00:16:48 +0000 (01:16 +0100)]
uboot-mediatek: fix and make use of LEDs on BPi-R2

Fix BPi-R2 GPIO LEDs to indicate boot into production or recovery
firmware in DTS and define them in default environment.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
3 years agomediatek: mt7622: provide power on mPCIe in SATA mode
Daniel Golle [Thu, 16 Sep 2021 20:04:14 +0000 (21:04 +0100)]
mediatek: mt7622: provide power on mPCIe in SATA mode

User reported that R64 doesn't provide power to the mPCIe slot in case
the PCIe port is disabled as it is when selecting the SATA
configuration. As users may still want to use USB-connected mPCIe
modules in CN8 slot.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
3 years agokernel: bump 5.10 to 5.10.70
Rui Salvaterra [Thu, 30 Sep 2021 13:21:41 +0000 (14:21 +0100)]
kernel: bump 5.10 to 5.10.70

Deleted (upstreamed):
bcm53xx/patches-5.10/181-Revert-USB-bcma-Add-a-check-for-devm_gpiod_get.patch

Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
3 years agokernel: bump 5.10 to 5.10.69
Rui Salvaterra [Mon, 27 Sep 2021 08:20:47 +0000 (09:20 +0100)]
kernel: bump 5.10 to 5.10.69

Patches automatically refreshed.

Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
3 years agokernel: bump 5.10 to 5.10.68
Rui Salvaterra [Wed, 22 Sep 2021 20:31:09 +0000 (21:31 +0100)]
kernel: bump 5.10 to 5.10.68

Deleted (upstreamed):
generic/backport-5.10/610-v5.13-02-netfilter-Fix-fall-through-warnings-for-Clang.patch
generic/backport-5.10/792-v5.15-0001-net-dsa-b53-Fix-calculating-number-of-switch-ports.patch
generic/backport-5.10/792-v5.15-0002-net-dsa-b53-Set-correct-number-of-ports-in-the-DSA-s.patch
generic/backport-5.10/792-v5.15-0003-net-dsa-b53-Fix-IMP-port-setup-on-BCM5301x.patch
generic/backport-5.10/840-0001-PCI-of-Don-t-fail-devm_pci_alloc_host_bridge-on-miss.patch
generic/backport-5.10/840-0002-PCI-iproc-Fix-BCMA-probe-resource-handling.patch
generic/pending-5.10/498-mtd-mtdconcat-select-readwrite-function.patch

Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
3 years agokernel: bump 5.10 to 5.10.67
Rui Salvaterra [Sat, 18 Sep 2021 20:55:58 +0000 (21:55 +0100)]
kernel: bump 5.10 to 5.10.67

Deleted (upstreamed):
bcm27xx/patches-5.10/950-0312-vc4_hdmi-Set-HD_CTL_WHOLSMP-and-HD_CTL_CHALIGN_SET.patch

Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
3 years agokernel: bump 5.10 to 5.10.66
Rui Salvaterra [Thu, 16 Sep 2021 12:55:00 +0000 (13:55 +0100)]
kernel: bump 5.10 to 5.10.66

No patches refreshed.

Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
3 years agokernel: bump 5.10 to 5.10.65
Rui Salvaterra [Wed, 15 Sep 2021 09:31:37 +0000 (10:31 +0100)]
kernel: bump 5.10 to 5.10.65

Manually refreshed:
generic/pending-5.10/800-bcma-get-SoC-device-struct-copy-its-DMA-params-to-th.patch
bcm27xx/patches-5.10/950-0053-firmware-bcm2835-Support-ARCH_BCM270x.patch
bcm27xx/patches-5.10/950-0087-firmware-raspberrypi-Add-backward-compatible-get_thr.patch

Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
3 years agonetifd: update to the latest version
Felix Fietkau [Thu, 30 Sep 2021 09:33:11 +0000 (11:33 +0200)]
netifd: update to the latest version

186f6eaeba70 wireless: display log messages for setup/teardown/retry
fac471c4934a wireless: process and close script file descriptor when rerunning setup
62e2bb56f48e main: poll process log stream even if processes are killed
0e311d3f2d1a wireless: reset number of retries on config change
e467e0ff44c0 wireless: reset retry counter when setup succeeds
448ffc154fe7 wireless: fix index for stations

Signed-off-by: Felix Fietkau <nbd@nbd.name>
3 years agomac80211: allow retry of wifi setup if an iw interface add command fails
Felix Fietkau [Thu, 30 Sep 2021 09:31:19 +0000 (11:31 +0200)]
mac80211: allow retry of wifi setup if an iw interface add command fails

In some cases, spurious failures might be cleared by teardown and retry

Signed-off-by: Felix Fietkau <nbd@nbd.name>
3 years agocryptodev-linux: update to 1.12
Rosen Penev [Thu, 23 Sep 2021 19:57:52 +0000 (12:57 -0700)]
cryptodev-linux: update to 1.12

Remove upstream backport.

Use AUTORELEASE for simplicity.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
3 years agokernel: add missing symbol to kmod-kvm
Stijn Tintel [Tue, 28 Sep 2021 22:28:06 +0000 (01:28 +0300)]
kernel: add missing symbol to kmod-kvm

Even though TRACEPOINTS is not enabled in my kernel config, my build
fails due to KVM_MMU_AUDIT being missing. Add this symbol to kmod-kvm to
fix this.

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
3 years agokernel: order kmod-kvm symbols alphabetically
Stijn Tintel [Tue, 28 Sep 2021 22:26:10 +0000 (01:26 +0300)]
kernel: order kmod-kvm symbols alphabetically

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>
3 years agoprereq-build: recognize Python 3.10
Marcin Juszkiewicz [Tue, 28 Sep 2021 13:03:24 +0000 (15:03 +0200)]
prereq-build: recognize Python 3.10

Fedora 35 contains Python 3.10 as default version. Make use of it.

Signed-off-by: Marcin Juszkiewicz <marcin@juszkiewicz.com.pl>
[fix commit subject]
Signed-off-by: Paul Spooren <mail@aparcar.org>
3 years agonetifd: update to the latest version
Felix Fietkau [Tue, 28 Sep 2021 12:27:20 +0000 (14:27 +0200)]
netifd: update to the latest version

4d0c2ad3fd26 wireless: fix applying wireless devices attributes on hotplug events

Signed-off-by: Felix Fietkau <nbd@nbd.name>
3 years agonetifd: update to the latest version
Felix Fietkau [Mon, 27 Sep 2021 16:59:53 +0000 (18:59 +0200)]
netifd: update to the latest version

5a4ac30c7a15 netifd: rework/fix device free handling

Signed-off-by: Felix Fietkau <nbd@nbd.name>
3 years agoramips: add support for Netgear R7450
Dale Hui [Sun, 26 Sep 2021 04:08:06 +0000 (21:08 -0700)]
ramips: add support for Netgear R7450

Netgear R7450 is a clone of Netgear R6700v2

Specifications
==============
SoC: MediaTek MT7621AT
RAM: 256M DDR3
FLASH: 128M NAND
WiFi: MediaTek MT7615N an+ac
MediaTek MT7615N bgn
ETH: MediaTek MT7621AT
BTN: 1x Connect (WPS), 1x WLAN, 1x Reset
LED: Power (white/amber), WAN(white/amber), 2.4G(white), 5G(white),
USB(white) , GuestWifi(white) 4x LAN(white/amber), Wifi Button(white),
WPS Button(white)

Flash Instructions
==================
Login to netgear webinterface and flash factory.img

Signed-off-by: Dale Hui <strokes-races0b@icloud.com>
[fix model/compatible in DTS]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agoramips: add support for Netgear R6900v2
Dale Hui [Sun, 26 Sep 2021 04:07:58 +0000 (21:07 -0700)]
ramips: add support for Netgear R6900v2

Netgear R6900v2 is a clone of Netgear R6700v2

Specifications
==============
SoC: MediaTek MT7621AT
RAM: 256M DDR3
FLASH: 128M NAND
WiFi: MediaTek MT7615N an+ac
MediaTek MT7615N bgn
ETH: MediaTek MT7621AT
BTN: 1x Connect (WPS), 1x WLAN, 1x Reset
LED: Power (white/amber), WAN(white/amber), 2.4G(white), 5G(white),
USB(white) , GuestWifi(white) 4x LAN(white/amber), Wifi Button(white),
WPS Button(white)

Flash Instructions
==================
Login to netgear webinterface and flash factory.img

Signed-off-by: Dale Hui <strokes-races0b@icloud.com>
3 years agoramips: add Nighthawk AC2100 as name for Netgear R6700v2
Dale Hui [Sat, 21 Aug 2021 07:10:22 +0000 (00:10 -0700)]
ramips: add Nighthawk AC2100 as name for Netgear R6700v2

Netgear Nighthawk AC2100 is another name of the Netgear R6700v2.

Signed-off-by: Dale Hui <strokes-races0b@icloud.com>
[adjust commit message/title]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agoramips: make Netgear R7200 a separate device from R6700v2
Dale Hui [Sat, 21 Aug 2021 05:36:50 +0000 (22:36 -0700)]
ramips: make Netgear R7200 a separate device from R6700v2

With the various variants of Netgear R**** devices, make it more
obvious which image should be used for the R7200.

Signed-off-by: Dale Hui <strokes-races0b@icloud.com>
[provide proper commit message]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agorealtek: add Kernel 5.10 as testing version
INAGAKI Hiroshi [Wed, 5 May 2021 00:31:53 +0000 (09:31 +0900)]
realtek: add Kernel 5.10 as testing version

This patch adds "KERNEL_TESTING_PATCHVER:=5.10" to the Makefile in
realtek target to allow using Kernel 5.10 for testing.

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: enable uart1 on the devices with PoE support in 5.10
INAGAKI Hiroshi [Mon, 23 Aug 2021 09:07:05 +0000 (18:07 +0900)]
realtek: enable uart1 on the devices with PoE support in 5.10

On the devices with PoE support, the secondary UART (uart1) on the SoC
is used to communicate between the SoC and controller.

Enable the secondary UART on the following devices:

- D-Link DGS-1210-10P
- Netgear GS110TPP v1
- Netgear GS310TP v1
- ZyXEL GS1900-8HP v1/v2
- ZyXEL GS1900-10HP
- ZyXEL GS1900-24HP v2

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: use physical addresses in soc dtsi in 5.10
INAGAKI Hiroshi [Wed, 11 Aug 2021 10:06:09 +0000 (19:06 +0900)]
realtek: use physical addresses in soc dtsi in 5.10

Use physical addresses instead of virtual address in dts files.

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: use gpio-keys instead of "-polled" if SoC GPIO is used in 5.10
INAGAKI Hiroshi [Sat, 26 Jun 2021 03:29:24 +0000 (12:29 +0900)]
realtek: use gpio-keys instead of "-polled" if SoC GPIO is used in 5.10

The new backported GPIO driver supports interrupt, so use gpio-keys
instead of gpio-keys-polled for keys connected to the internal GPIO
controller.

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: update GPIO bindings in the dts files in dts-5.10
INAGAKI Hiroshi [Sat, 8 May 2021 08:51:29 +0000 (17:51 +0900)]
realtek: update GPIO bindings in the dts files in dts-5.10

this patch includes the following changes:

- adjust mapping for the new driver
  - GPIO 24 -> GPIO 0
  - GPIO 47 -> GPIO 0 (+ disabling system LED)

- disable pins in the invalid range
  (out of the range 0-31 of the new driver)
  - are these pins on the external RTL8231 (&gpio1)?
    - GPIO 67 (-> GPIO 3 on &gpio1?)
    - GPIO 94 (-> GPIO 30 on &gpio1?)

- drop "indirect-access-bus-id" property from gpio0 node in device dts
  files

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: fix kernel panic in DSA driver for 5.10
INAGAKI Hiroshi [Sat, 8 May 2021 08:28:58 +0000 (17:28 +0900)]
realtek: fix kernel panic in DSA driver for 5.10

dsa_to_port function in 5.10 returns dsa_port from the port list in
dsa_switch_tree, but the tree is built when the switch is registered
by dsa_register_switch and it's null in rtl83xx_mdio_probe.

So, we need to use dsa_to_port after the registration of the switch.

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: add pinmux node of LED_GLB_CTRL to rtl838x.dtsi in dts-5.10
INAGAKI Hiroshi [Fri, 7 May 2021 03:10:10 +0000 (12:10 +0900)]
realtek: add pinmux node of LED_GLB_CTRL to rtl838x.dtsi in dts-5.10

This patch adds a pinctrl-single pinmux node to allow disabling system
LED and enabling GPIO 0 (old driver: GPIO 24).

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: fix compile errors in dsa driver for 5.10
INAGAKI Hiroshi [Thu, 6 May 2021 15:38:22 +0000 (00:38 +0900)]
realtek: fix compile errors in dsa driver for 5.10

this patch fixes the following errors when compiling:

- dsa_switch_alloc is removed[1]

- a parameter "enum dsa_tag_protocol mprot" is added to dsa_tag_protocol
  in dsa_switch_ops (include/net/dsa.h)

- several paramters are added to "phylink_mac_link_up" in dsa_switch_ops
  (include/net/dsa.h)

  added:
    - int speed
    - int duplex
    - bool tx_pause
    - bool rx_pause

- a parameter "struct switchdev_trans *trans" is added to
  port_vlan_filtering in dsa_switch_ops (include/net/dsa.h)

[1]: https://lore.kernel.org/lkml/20191020031941.3805884-17-vivien.didelot@gmail.com/

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: fix compile errors in rtl838x_eth.c for 5.10
INAGAKI Hiroshi [Thu, 6 May 2021 14:22:19 +0000 (23:22 +0900)]
realtek: fix compile errors in rtl838x_eth.c for 5.10

this patch fixes the following errors when compiling:

- "unsigned int txqueue" is added as an additional parameter of
  ndo_tx_timeout in net_device_ops (include/linux/netdevice.h)

- "mac_link_state" in phylink_mac_ops (include/linux/phylink.h)
  is renamed to "mac_pcs_get_state" and changed the return value
  to void from int

- several parameters are added to "mac_link_up" in phylink_mac_ops
  (include/linux/phylink.h) and the order of the parameters is
  changed

  added:
    - int speed
    - int duplex
    - bool tx_pause
    - bool rx_pause

- a parameter "phy_interface_t *interface" is added to of_get_phy_mode
  (drivers/of/of_net.c) and returns the state instead of phy mode

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: add "soc" node to soc dtsi in dts-5.10
INAGAKI Hiroshi [Thu, 6 May 2021 11:46:42 +0000 (20:46 +0900)]
realtek: add "soc" node to soc dtsi in dts-5.10

Add a "soc" node as a simple-bus to rtl838x.dtsi and rtl930x.dtsi.

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: update soc dtsi in 5.10 for backported drivers
INAGAKI Hiroshi [Thu, 6 May 2021 10:40:04 +0000 (19:40 +0900)]
realtek: update soc dtsi in 5.10 for backported drivers

this patch updates SoC dtsi (rtl838x.dtsi, rtl930x.dtsi) for the
following backported drivers:

- gpio-realtek-otto (5.13)
- spi-realtek-rtl (5.12)
- irq-realtek-rtl (5.12)

And, disable SoC GPIO node (gpio0) in rtl930x.dtsi in dts-5.10.
Currently, the upstreamed driver doesn't support the GPIO controller on
RTL930x SoC.

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: cleanup and update soc dtsi in 5.10
INAGAKI Hiroshi [Wed, 11 Aug 2021 11:27:38 +0000 (20:27 +0900)]
realtek: cleanup and update soc dtsi in 5.10

the following changes are included in this patch:

- node is enabled by default, drop 'status = "okay"'
- adjust order of "compatible" lines and "reg" lines
- add a new blank line before fixed-link node in rtl830x.dtsi

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: copy dts directory for Kernel 5.10
INAGAKI Hiroshi [Wed, 11 Aug 2021 11:23:06 +0000 (20:23 +0900)]
realtek: copy dts directory for Kernel 5.10

This patch adds "dts-5.10" directory to use backported drivers.
There are several specification changes in the new drivers, so there
are some compatibility issues in using dts/dtsi files for 5.4.

The old DTS files are moved to "dts-5.4", so their corresponding
kernel version is obvious as well.

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
[change "dts" to "dts-5.4", adjust Makefile]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agorealtek: refresh and update config-5.10
INAGAKI Hiroshi [Wed, 5 May 2021 13:47:19 +0000 (22:47 +0900)]
realtek: refresh and update config-5.10

This patch adjusts config-5.10 by running kernel_menuconfig.

Note:

- disable psb6970 phy driver (unused in realtek target?)

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: refresh patches for Kernel 5.10
INAGAKI Hiroshi [Wed, 5 May 2021 05:02:49 +0000 (14:02 +0900)]
realtek: refresh patches for Kernel 5.10

This patch refresh all patches in patches-5.10/ to adjust for Kernel
5.10.

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: remove unnecessary line from rtl838x/Platform in 5.10
INAGAKI Hiroshi [Thu, 6 May 2021 09:17:54 +0000 (18:17 +0900)]
realtek: remove unnecessary line from rtl838x/Platform in 5.10

The following line is already defined in arch/mips/Kbuild.platforms by
300-mips-add-rtl838x-platform.patch.

platform-$(CONFIG_RTL838X) += rtl838x/

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: drop fixup_bigphys_addr from ioremap.h in 5.10
INAGAKI Hiroshi [Wed, 5 May 2021 06:20:03 +0000 (15:20 +0900)]
realtek: drop fixup_bigphys_addr from ioremap.h in 5.10

A macro with the same name is provided in asm/pgtable.h in Kernel 5.10,
use it and drop from ioremap.h.

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: fix "help" line in Kconfig in files/patches for 5.10
INAGAKI Hiroshi [Wed, 5 May 2021 05:18:43 +0000 (14:18 +0900)]
realtek: fix "help" line in Kconfig in files/patches for 5.10

In Kernel 5.10, "help" must be used instead of "---help---".

this patch fixes the following errors:

drivers/net/dsa/rtl83xx/Kconfig:7: syntax errorgit
drivers/net/dsa/rtl83xx/Kconfig:6: unknown statement "---help---"
drivers/net/dsa/rtl83xx/Kconfig:7:warning: ignoring unsupported character '.'
drivers/net/dsa/rtl83xx/Kconfig:7: unknown statement "This"
drivers/net/ethernet/Kconfig:170: syntax error
drivers/net/ethernet/Kconfig:169: unknown statement "---help---"
drivers/net/ethernet/Kconfig:170:warning: ignoring unsupported character '.'
drivers/net/ethernet/Kconfig:170: unknown statement "Say"
drivers/net/phy/Kconfig:331: syntax error
drivers/net/phy/Kconfig:330: unknown statement "---help---"
drivers/net/phy/Kconfig:331: unknown statement "Supports"

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: backport irq-realtek-rtl driver from 5.12 to 5.10
INAGAKI Hiroshi [Fri, 21 May 2021 14:16:37 +0000 (23:16 +0900)]
realtek: backport irq-realtek-rtl driver from 5.12 to 5.10

This patch backports "irq-realtek-rtl" driver to Kernel 5.10 from 5.12.
"MACH_REALTEK_RTL" is used as a platform name in upstream, but "RTL838X"
is used in OpenWrt, so update the dependency by the additional patch.

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: drop platform irq driver from 5.10
INAGAKI Hiroshi [Fri, 21 May 2021 14:20:43 +0000 (23:20 +0900)]
realtek: drop platform irq driver from 5.10

To use backported irq driver, drop old irq driver from realtek target
and call irqchip_init() in setup.c.

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: backport gpio-realtek-otto driver from 5.13 to 5.10
INAGAKI Hiroshi [Wed, 5 May 2021 13:05:39 +0000 (22:05 +0900)]
realtek: backport gpio-realtek-otto driver from 5.13 to 5.10

This patch backports "gpio-realtek-otto" driver to Kernel 5.10.
"MACH_REALTEK_RTL" is used as a platform name in upstream, but "RTL838X"
is used in OpenWrt, so update the dependency by the additional patch.

Note: GPIO mapping is changed in the upstreamed driver.

  old - new

   24 -  0
   25 -  1
   26 -  2
   27 -  3
   28 -  4
   29 -  5
   30 -  6
   31 -  7
   16 -  8
   17 -  9
   18 - 10
   19 - 11
   20 - 12
   21 - 13
   22 - 14
   23 - 15
    8 - 16
    9 - 17
   10 - 18
   11 - 19
   12 - 20
   13 - 21
   14 - 22
   15 - 23

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: drop rtl838x gpio driver from 5.10
INAGAKI Hiroshi [Thu, 24 Jun 2021 11:32:31 +0000 (20:32 +0900)]
realtek: drop rtl838x gpio driver from 5.10

To backport the upstreamed driver (gpio-realtek-otto) from 5.13, drop the
old driver from realtek target.

And, modify 301-gpio-add-rtl838x-driver.patch to remove rtl838x GPIO
support and rename it only for rtl8231 GPIO support.

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: backport spi-realtek-rtl driver from 5.12 to 5.10
INAGAKI Hiroshi [Thu, 6 May 2021 10:30:58 +0000 (19:30 +0900)]
realtek: backport spi-realtek-rtl driver from 5.12 to 5.10

This patch backports "spi-realtek-rtl" driver to Kernel 5.10 from 5.12.
"MACH_REALTEK_RTL" is used as a platform name in upstream, but "RTL838X"
is used in OpenWrt, so update the dependency by the additional patch.

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: drop rtl838x spi-nor driver from 5.10
INAGAKI Hiroshi [Thu, 24 Jun 2021 11:22:51 +0000 (20:22 +0900)]
realtek: drop rtl838x spi-nor driver from 5.10

To backport the upstreamed driver (spi-realtek-rtl) from 5.12, drop the
old driver from realtek target.

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
3 years agorealtek: copy config/files/patches to 5.10
INAGAKI Hiroshi [Wed, 5 May 2021 00:32:27 +0000 (09:32 +0900)]
realtek: copy config/files/patches to 5.10

this patch copies the following files from 5.4 to 5.10:

- config-5.4   -> config-5.10
- files-5.4/   -> files-5.10/
- patches-5.4/ -> patches-5.10/

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
[rebase on change in files-5.4]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agoramips: enable I2C_CHARDEV
David Bauer [Fri, 20 Aug 2021 13:52:51 +0000 (15:52 +0200)]
ramips: enable I2C_CHARDEV

Expose I2C busses with a chardev device. This is required to control the
PSE controller on the Ubiquiti UniFi Flex Switch.

Signed-off-by: David Bauer <mail@david-bauer.net>
3 years agoramips: add support for Ubiquiti USW-Flex
David Bauer [Fri, 20 Aug 2021 11:23:42 +0000 (13:23 +0200)]
ramips: add support for Ubiquiti USW-Flex

Hardware
--------
MediaTek MT7621AT
16M SPI-NOR Macronix MX25L12835FMI
Microchip PD69104B1 4-Channel PoE-PSE controller
TI TPS2373 PoE-PD controller

PoE-Controller
--------------

By default, the PoE outputs do not work with OpenWrt. To make them output
power, install the "poemgr" package from the packages feed.
This package can control the PD69104B1 PSE controller.

Installation
------------

1. Connect to the booted device at 192.168.1.20 using username/password
   "ubnt" via SSH.

2. Add the uboot-envtools configuration file /etc/fw_env.config with the
   following content

   $ echo "/dev/mtd1 0x0 0x1000 0x10000 1" > /etc/fw_env.config

3. Update the bootloader environment.

   $ fw_setenv boot_openwrt "fdt addr \$(fdtcontroladdr);
     fdt rm /signature; bootubnt"
   $ fw_setenv bootcmd "run boot_openwrt"

4. Transfer the OpenWrt sysupgrade image to the device using SCP.

5. Check the mtd partition number for bs / kernel0 / kernel1

   $ cat /proc/mtd

6. Set the bootselect flag to boot from kernel0

   $ dd if=/dev/zero bs=1 count=1 of=/dev/mtdblock4

7. Write the OpenWrt sysupgrade image to both kernel0 as well as kernel1

   $ dd if=openwrt.bin of=/dev/mtdblock6
   $ dd if=openwrt.bin of=/dev/mtdblock7

8. Reboot the device. It should boot into OpenWrt.

Restore to UniFi
----------------

To restore the vendor firmware, follow the Ubiquiti UniFi TFTP
recovery guide for access points. The process is the same for
the Flex switch.

Signed-off-by: David Bauer <mail@david-bauer.net>
3 years agoath79: add support for TP-Link CPE710-v1
Andrew Cameron [Mon, 12 Apr 2021 18:24:33 +0000 (13:24 -0500)]
ath79: add support for TP-Link CPE710-v1

TP-Link CPE710-v1 is an outdoor wireless CPE for 5 GHz with
one Ethernet port based on the AP152 reference board

Specifications:
- SoC: QCA9563-AL3A MIPS 74kc @ 775MHz, AHB @ 258MHz
- RAM: 128MiB DDR2 @ 650MHz
- Flash: 16MiB SPI NOR Based on the GD25Q128
- Wi-Fi 5Ghz: ath10k chip (802.11ac for up to 867Mbps on 5GHz wireless
  data rate) Based on the QCA9896
- Ethernet: one 1GbE port
- 23dBi high-gain directional 2×2 MIMO antenna and a dedicated metal
  reflector
- Power, LAN, WLAN5G Blue LEDs
- 3x Blue LEDs

Flashing instructions:
Flash factory image through stock firmware WEB UI or through TFTP
To get to TFTP recovery just hold reset button while powering on for
around 30-40 seconds and release.
Rename factory image to recovery.bin
Stock TFTP server IP:192.168.0.100
Stock device TFTP address:192.168.0.254

Signed-off-by: Andrew Cameron <apcameron@softhome.net>
[convert to nvmem, fix MAC assignment in 11-ath10k-caldata]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agokernel: 5.10: backport fix for lp55xx LED driver
Robert Marko [Fri, 17 Sep 2021 10:27:44 +0000 (12:27 +0200)]
kernel: 5.10: backport fix for lp55xx LED driver

This backports the upstream commit:
leds: lp55xx: Initialize enable GPIO direction to output

Without it under kernel 5.10 on Asus MAP-AC2200
the LED driver will fail probing:
[    1.947521] lp5523x: probe of 0-0032 failed with error -22

After the backported fix:
[    1.873236] lp5523x 0-0032: lp5523 Programmable led chip found

Signed-off-by: Robert Marko <robert.marko@sartura.hr>
Tested-by: Szabolcs Hubai <szab.hu@gmail.com> [ipq4029/gl-b1300]
Tested-by: Nick Hainke <vincent@systemli.org> [ipq4019/fritzbox-7530
                                               ipq4019/fritzbox-4040
                                               ipq4019/sxtsq-5ac]
Tested-by: Stefan Lippers-Hollmann <s.l-h@gmx.de> [ipq4019/map-ac2200]
3 years agoipq40xx: add testing support for kernel 5.10
Robert Marko [Sun, 12 Sep 2021 20:41:43 +0000 (22:41 +0200)]
ipq40xx: add testing support for kernel 5.10

Add kernel 5.10 as the testing kernel to ipq40xx to
get wider testing.

The following devices failed to build with buildbot settings and all
feeds installed (apparently due to kernel size):

 * cell-c rtl30vw
 * compex wpj428
 * devolo magic 2 next
 * engenius emr3500
 * glinet gl-b1300
 * glinet gl-s1300
 * qcom ap-dk01.1-c1
 * qcom ap-dk04.1-c1

Signed-off-by: Robert Marko <robert.marko@sartura.hr>
Tested-by: Nick Hainke <vincent@systemli.org> [ipq4019/fritzbox-7530
                                               ipq4019/fritzbox-4040
                                               ipq4019/sxtsq-5ac]
Tested-by: Szabolcs Hubai <szab.hu@gmail.com> [ipq4029/gl-b1300]
Tested-by: Stefan Lippers-Hollmann <s.l-h@gmx.de> [ipq4019/map-ac2200]
[add tested-by and note about failed devices]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agoipq40xx: 5.10: move AR40xx to MDIO drivers
Robert Marko [Sun, 12 Sep 2021 21:50:20 +0000 (23:50 +0200)]
ipq40xx: 5.10: move AR40xx to MDIO drivers

MDIO drivers were moved into their own sub directory of networking drivers.
This has caused the AR40xx driver to probe before MDIO drivers and that wont
work as it depends on the MDIO bus to be up so it can be fetched.

Lets solve it by moving the AR40xx into MDIO folder so they get probed like
before.

Signed-off-by: Robert Marko <robert.marko@sartura.hr>
3 years agoipq40xx: net: ethernet: edma: reject unsupported coalescing params
Robert Marko [Sun, 12 Sep 2021 21:26:59 +0000 (23:26 +0200)]
ipq40xx: net: ethernet: edma: reject unsupported coalescing params

Set ethtool_ops->supported_coalesce_params to let
the core reject unsupported coalescing parameters.

This driver did not previously reject unsupported parameters.

This is a required ethtool op since kernel 5.7.

Signed-off-by: Robert Marko <robert.marko@sartura.hr>
3 years agoipq40xx: net: ethernet: edma: update of_get_phy_mode() for 5.10
Robert Marko [Sun, 12 Sep 2021 21:00:51 +0000 (23:00 +0200)]
ipq40xx: net: ethernet: edma: update of_get_phy_mode() for 5.10

In kernel v5.5 of_get_phy_mode had its API changed, so its now returning 0
or errors instead of phymode.

Phymode is now returning by passing a pointer to phy_interface_t where it
will be stored.

Signed-off-by: Robert Marko <robert.marko@sartura.hr>
3 years agoipq40xx: 5.10: add kernel config
Robert Marko [Sun, 12 Sep 2021 20:40:39 +0000 (22:40 +0200)]
ipq40xx: 5.10: add kernel config

Copy config from 5.4 and run "make kernel_oldconfig".

Select default ("N") for all new symbols.

Signed-off-by: Robert Marko <robert.marko@sartura.hr>
[make commit message more explicit]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
3 years agoipq40xx: 5.10: refresh patches to apply
Robert Marko [Sun, 12 Sep 2021 20:36:12 +0000 (22:36 +0200)]
ipq40xx: 5.10: refresh patches to apply

Refresh the kernel patches on top of 5.10 so they apply.
Manually fixup the 705-net-add-qualcomm-ar40xx-phy.patch
to apply.

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