openwrt/staging/blogic.git
7 months agomediatek: Moved recovery image creation to include/image-commands.mk
Roland Reinl [Sat, 30 Mar 2024 16:59:14 +0000 (17:59 +0100)]
mediatek: Moved recovery image creation to include/image-commands.mk

The recovery image is reqired for D-Link M30 as well. So I moved it to include/image-commands.mk to be able to use it for MT7622 and filogic devices.

Signed-off-by: Roland Reinl <reinlroland+github@gmail.com>
7 months agofilogic: Add LED driver for GCA230718
Roland Reinl [Tue, 26 Dec 2023 07:31:15 +0000 (08:31 +0100)]
filogic: Add LED driver for GCA230718

Add basic support for the LED driver for GCA230718.

 - I didn't find any documentation or driver for it, so the information below is purely based on my investigations
 - If there is already I driver for it, please tell me. Maybe I didn't search enough
 - I implemented a kernel module (leds-gca230718) to access the LEDs via DTS
 - The LED controller supports PWM for brightness control and ramp control for smooth blinking. This is not implemented in the driver
 - The LED controller supports toggling (on -> off -> on -> off) where the brightness of the LEDs can be set individually for each on cycle
 - Until now, only simple active/inactive control is implemented (like when the LEDs would have been connected via GPIO)
 - Controlling the LEDs requires three sequences sent to the chip. Each sequence consists of
   - A reset command (0x81 0xE4) written to register 0x00
   - A control command (for example 0x0C 0x02 0x01 0x00 0x00 0x00 0xFF 0x01 0x00 0x00 0x00 0xFF 0x87 written to register 0x03)
 - The reset command is always the same
 - In the control command
   - byte 0 is always the same
   - byte 1 (0x02 in the example above) must be changed in every sequence: 0x02 -> 0x01 -> 0x03)
   - byte 2 is set to 0x01 which disables toggling. 0x02 would be LED toggling without ramp control, 0x03 would be toggling with ramp control
   - byte 3 to 6 define the brightness values for the LEDs (R,G,B,W) for the first on cycle when toggling
   - byte 7 defines the toggling frequency (if toggling enabled)
   - byte 8 to 11 define the brightness values for the LEDs (R,G,B,W) for the second on cycle when toggling
   - byte 12 is constant 0x87

Signed-off-by: Roland Reinl <reinlroland+github@gmail.com>
7 months agokernel: bump 5.15 to 5.15.153
Hauke Mehrtens [Fri, 29 Mar 2024 14:45:50 +0000 (15:45 +0100)]
kernel: bump 5.15 to 5.15.153

Removed because they are upstream:
generic/backport-5.15/704-15-v5.19-net-mtk_eth_soc-move-MAC_MCR-setting-to-mac_finish.patch
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=linux-5.15.y&id=c5c0760adc260d55265c086b9efb350ea6dda38b

generic/pending-5.15/735-net-mediatek-mtk_eth_soc-release-MAC_MCR_FORCE_LINK-.patch
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=linux-5.15.y&id=448cc8b5f743985f6d1d98aa4efb386fef4c3bf2

generic/pending-5.15/736-net-ethernet-mtk_eth_soc-fix-PPE-hanging-issue.patch
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=linux-5.15.y&id=9fcadd125044007351905d40c405fadc2d3bb6d6

Add new configuration symbols for tegra target.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 months agokernel: bump 6.6 to 6.6.23
John Audia [Wed, 27 Mar 2024 08:11:32 +0000 (04:11 -0400)]
kernel: bump 6.6 to 6.6.23

Changelog: https://cdn.kernel.org/pub/linux/kernel/v6.x/ChangeLog-6.6.23

Removed upstreamed:
  pending-6.6/735-net-mediatek-mtk_eth_soc-release-MAC_MCR_FORCE_LINK-.patch[1]
  pending-6.6/736-net-ethernet-mtk_eth_soc-fix-PPE-hanging-issue.patch[2]
  mediatek/patches-6.6/232-clk-mediatek-mt7981-topckgen-flag-SGM_REG_SEL-as-cri.patch[3]

Manually rebased:
  mediatek/patches-6.6/100-dts-update-mt7622-rfb1.patch

Added:
  generic/backports-6.6/981-mtd-spinand-Add-support-for-5-byte-IDs.patch[4]

All other patches automatically rebased.

1. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.6.23&id=1f32abb474c1c9bdb21d9eda74c325a0b3a162e5
2. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.6.23&id=943c14ece95eb1cf98d477462aebcbfdfd714633
3. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.6.23&id=6ff01b314149d1cf59caebc29384f0beed21cba4
4. See comments in https://github.com/openwrt/openwrt/pull/14992 regarding broken flogic/xiaomi_redmi-router-ax6000-ubootmod

Build system: x86/64
Build-tested: x86/64/AMD Cezanne, flogic/xiaomi_redmi-router-ax6000-ubootmod, flogic/glinet_gl-mt6000
Run-tested: x86/64/AMD Cezannei, flogic/xiaomi_redmi-router-ax6000-ubootmod, flogic/glinet_gl-mt6000

Signed-off-by: John Audia <therealgraysky@proton.me>
7 months agoath79: add support for Huawei AP5030DN
Marco von Rosenberg [Sun, 31 Mar 2024 15:07:39 +0000 (17:07 +0200)]
ath79: add support for Huawei AP5030DN

Huawei AP5030DN is a dual-band, dual-radio 802.11ac Wave 1 3x3 MIMO
enterprise access point with two Gigabit Ethernet ports and PoE
support.

Hardware highlights:
- CPU: QCA9550 SoC at 720MHz
- RAM: 256MB DDR2
- Flash: 32MB SPI-NOR
- Wi-Fi 2.4GHz: QCA9550-internal radio
- Wi-Fi 5GHz: QCA9880 PCIe WLAN SoC
- Ethernet 1: 10/100/1000 Mbps Ethernet through Broadcom B50612E PHY
- Ethernet 2: 10/100/1000 Mbps Ethernet through Marvell 88E1510 PHY
- PoE: input through Ethernet 1 port
- Standalone 12V/2A power input
- Serial console externally available through RJ45 port
- External watchdog: SGM706 (1.6s timeout)

Serial console:
  9600n8 (9600 baud, no stop bits, no parity, 8 data bits)

MAC addresses:
  Each device has 32 consecutive MAC addresses allocated by
  the vendor, which don't overlap between devices.
  This was confirmed with multiple devices with consecutive
  serial numbers.
  The MAC address range starts with the address on the label.
  To be able to distinguish between the interfaces,
  the following MAC address scheme is used:
    - eth0 = label MAC
    - eth1 = label MAC + 1
    - radio0 (Wi-Fi 5GHz) = label MAC + 2
    - radio1 (Wi-Fi 2.4GHz) = label MAC + 3

Installation:
0. Connect some sort of RJ45-to-USB adapter to "Console" port of the AP

1. Power up the AP

2. At prompt "Press f or F  to stop Auto-Boot in 3 seconds",
   do what they say.
   Log in with default admin password "admin@huawei.com".

3. Boot the OpenWrt initramfs from TFTP using the hidden script
   "run ramboot". Replace IP address as needed:

   > setenv serverip 192.168.1.10
   > setenv ipaddr 192.168.1.1
   > setenv rambootfile
     openwrt-ath79-generic-huawei_ap5030dn-initramfs-kernel.bin
   > saveenv
   > run ramboot

4. Optional but recommended as the factory firmware cannot
   be downloaded publicly:
   Back up contents of "firmware" partition using the web interface or ssh:

   $ ssh root@192.168.1.1 cat /dev/mtd11 > huawei_ap5030dn_fw_backup.bin

5. Run sysupgrade using sysupgrade image. OpenWrt
   shall boot from flash afterwards.

Return to factory firmware (using firmware upgrade package downloaded from
non-public Huawei website):
1. Start a TFTP server in the directory where
   the firmware upgrade package is located

2. Boot to u-boot as described above

3. Install firmware upgrade package and format the config partitions:

   > update system FatAP5X30XN_SOMEVERSION.bin
   > format_fs

Return to factory firmware (from previously created backup):
1. Copy over the firmware partition backup to /tmp,
   for example using scp

2. Use sysupgrade with force to restore the backup:
   sysupgrade -F huawei_ap5030dn_fw_backup.bin

3. Boot AP to U-Boot as described above

Quirks and known issues
-----------------------

- On initial power-up, the Huawei-modified bootloader suspends both
ethernet PHYs (it sets the "Power Down" bit in the MII control
register). Unfortunately, at the time of the initial port, the kernel
driver for the B50612E/BCM54612E PHY behind eth0 doesn't have a resume
callback defined which would clear this bit. This makes the PHY unusable
since it remains suspended forever. This is why the backported kernel
patches in this commit are required which add this callback and for
completeness also a suspend callback.

- The stock firmware has a semi dual boot concept where the primary
kernel uses a squashfs as root partition and the secondary kernel uses
an initramfs. This dual boot concept is circumvented on purpose to gain
more flash space and since the stock firmware's flash layout isn't
compatible with mtdsplit.

- The external watchdog's timeout of 1.6s is very hard to satisfy
during bootup. This is why the GPIO15 pin connected to the watchdog input
is configured directly in the LZMA loader to output the CPU_CLK/4 signal
which keeps the watchdog happy until the wdt-gpio kernel driver takes
over. Because it would also take too long to read the whole kernel image
from flash, the uImage header only includes the loader which then reads
the kernel image from flash after GPIO15 is configured.

Signed-off-by: Marco von Rosenberg <marcovr@selfnet.de>
[fixed 6.6 backport patch naming]
Signed-off-by: David Bauer <mail@david-bauer.net>
7 months agokernel: fix build issue on macOS
Felix Fietkau [Sat, 30 Mar 2024 20:58:31 +0000 (21:58 +0100)]
kernel: fix build issue on macOS

On x86, the build failed while trying to compile tools/lib/string.c because
of a clash with the system provided implementation for strlcpy
Add ifdefs to prevent the conflict.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 months agotools: b43-tools: fix compilation with GCC14
Robert Marko [Fri, 29 Mar 2024 17:57:03 +0000 (18:57 +0100)]
tools: b43-tools: fix compilation with GCC14

GCC14 no longer treats integer types and pointer types as equivalent in
assignments (including implied assignments of function arguments and return
values), and instead fails the compilation with a type error.

So, as a workaround lets disable the newly introduced error
-Werror=int-conversion and just make it print a warning to enable compiling
with GCC14 as Fedora 40 now defaults to it.

Signed-off-by: Robert Marko <robimarko@gmail.com>
7 months agouboot-envtools: ath79: remove D-Link DIR-8x9 and DAP-1720 env config
Shiji Yang [Tue, 19 Mar 2024 13:25:52 +0000 (21:25 +0800)]
uboot-envtools: ath79: remove D-Link DIR-8x9 and DAP-1720 env config

The uboot-envtools can automatically parse the dts 'u-boot,env'
compatible string. So the env config file is now useless.

Signed-off-by: Shiji Yang <yangshiji66@qq.com>
7 months agoath79: move D-Link DAP-1720 A1 to tiny sub-target
Shiji Yang [Tue, 19 Mar 2024 13:25:52 +0000 (21:25 +0800)]
ath79: move D-Link DAP-1720 A1 to tiny sub-target

This device only has 64 MiB RAM and ath10k wireless driver will
consume a lot of memory. Let's move it to the tiny sub-target to
get extra 7 MiB of free space. In this way, we can extend their
lifetime to receive support for the next OpenWrt LTS version. This
patch also trims the duplicate "recovery.bin" image as it's the
same as the "factory.bin".

Signed-off-by: Shiji Yang <yangshiji66@qq.com>
7 months agoath79: move D-Link DIR-859 and DIR-869 series to tiny sub-target
Shiji Yang [Tue, 19 Mar 2024 13:25:52 +0000 (21:25 +0800)]
ath79: move D-Link DIR-859 and DIR-869 series to tiny sub-target

These devices only have 64 MiB RAM and ath10k wireless driver will
consume a lot of memory. Let's move them to the tiny sub-target to
get extra 7 MiB of free space. In this way, we can extend their
lifetime to receive support for the next OpenWrt LTS version. This
patch also trims the USB package for the non-existent USB port.

Signed-off-by: Shiji Yang <yangshiji66@qq.com>
7 months agoath79: tiny: enable NVMEM u-boot-env driver
Shiji Yang [Tue, 19 Mar 2024 13:25:52 +0000 (21:25 +0800)]
ath79: tiny: enable NVMEM u-boot-env driver

The upcoming D-Link devices to the tiny sub-target require it to
parse the u-env MAC address. The kernel size will increase by
about 1 KiB.

Signed-off-by: Shiji Yang <yangshiji66@qq.com>
7 months agoath79: move seama image recipe to the common Makefile
Shiji Yang [Tue, 19 Mar 2024 13:25:52 +0000 (21:25 +0800)]
ath79: move seama image recipe to the common Makefile

Move seama image recipe to the common Makefile in order for some
tiny sub-target D-Link devices can share it.

Signed-off-by: Shiji Yang <yangshiji66@qq.com>
7 months agoath79: disable building small Flash EnGenius devices
Shiji Yang [Tue, 19 Mar 2024 13:25:52 +0000 (21:25 +0800)]
ath79: disable building small Flash EnGenius devices

The buildbot has not successfully built them for a long time due
to small Flash size.

Signed-off-by: Shiji Yang <yangshiji66@qq.com>
7 months agodtc: fix compilation with both libraries
Rosen Penev [Thu, 22 Feb 2024 06:22:02 +0000 (22:22 -0800)]
dtc: fix compilation with both libraries

Upstream backports.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
7 months agotools/meson: static host and both libraries
Rosen Penev [Wed, 21 Feb 2024 21:16:35 +0000 (13:16 -0800)]
tools/meson: static host and both libraries

Host packages typically are statically linked to avoid rpath issues and
to avoid libraries not being found as a result. With target packages,
both libraries make the most sense as InstallDev typically installs
both, giving packages flexibility. Default this behavior.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
7 months agotools/meson: update to 1.3.2
Rosen Penev [Wed, 21 Feb 2024 21:13:34 +0000 (13:13 -0800)]
tools/meson: update to 1.3.2

Add explicit setup as implicit is deprecated.

Rename pkgconfig as no hyphen is deprecated.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
7 months agomediatek: filogic: add Netcore N60 support
Chukun Pan [Fri, 3 Nov 2023 15:02:35 +0000 (23:02 +0800)]
mediatek: filogic: add Netcore N60 support

Hardware specification:
  SoC: MediaTek MT7986A 4x A53
  Flash: ESMT F50L1G41LB 128MB
  RAM: W632GU6NB DDR3 256MB
  Ethernet: 1x 2.5G + 4x 1G
  WiFi1: MT7975N 2.4GHz 4T4R
  WiFi2: MT7975PN 5GHz 4T4R
  Button: Reset, WPS
  Power: DC 12V 2A

Flash instructions:
  1. Connect to the router using ssh or telnet,
     username: useradmin, password is the web
     login password of the router.
  2. Use scp to upload bl31-uboot.fip and flash:
     "mtd write xxx-preloader.bin spi0.0"
     "mtd write xxx-bl31-uboot.fip FIP"
     "mtd erase ubi"
  3. Connect to the router via the Lan port,
     set a static ip of your PC.
     (ip 192.168.1.254, gateway 192.168.1.1)
  4. Download initramfs image, reboot router,
     waiting for tftp recovery to complete.
  5. After openwrt boots up, perform sysupgrade.

Note:
  1. Back up all mtd partitions before flashing.

Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>
7 months agouboot-mediatek: add Netcore N60 support
Chukun Pan [Thu, 2 Nov 2023 15:16:28 +0000 (23:16 +0800)]
uboot-mediatek: add Netcore N60 support

The vendor uboot requires special fit verification.
So add a custom uboot build for this device.

Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>
7 months agoramips: mt7620a_tplink_archer-c5-v4: use LED_FUNCTION_WLAN_2GHZ and LED_FUNCTION_WLAN...
Mieczyslaw Nalewaj [Mon, 18 Mar 2024 10:07:35 +0000 (11:07 +0100)]
ramips: mt7620a_tplink_archer-c5-v4: use LED_FUNCTION_WLAN_2GHZ and LED_FUNCTION_WLAN_5GHZ

Use LED_FUNCTION_WLAN_2GHZ and LED_FUNCTION_WLAN_5GHZ instead function-enumerator

Signed-off-by: Mieczyslaw Nalewaj <namiltd@yahoo.com>
7 months agoimagebuilder: add check if target is sourced from feed
Florian Eckert [Wed, 3 May 2023 11:29:42 +0000 (13:29 +0200)]
imagebuilder: add check if target is sourced from feed

The image generation would fail, if the target is included from a feed.
To fix this, check if targets is found in the feed directory.

Signed-off-by: Florian Eckert <fe@dev.tdt.de>
Tested-by: Thomas Richard <thomas.richard@bootlin.com>
7 months agoimagebuilder: check if BOARD is located in the feeds sub directory
Florian Eckert [Thu, 25 Jan 2024 09:25:06 +0000 (10:25 +0100)]
imagebuilder: check if BOARD is located in the feeds sub directory

Fixes the regression so that targets that were installed via a feed can
also be build again with the Image Builder.

Fixes: 84ec8c4 ("imagebuilder: copy from buildroot only target/linux")
Signed-off-by: Florian Eckert <fe@dev.tdt.de>
Tested-by: Thomas Richard <thomas.richard@bootlin.com>
7 months agorockchip: remove redundant 'console' parameter from boot script
Tianling Shen [Mon, 25 Mar 2024 18:25:42 +0000 (02:25 +0800)]
rockchip: remove redundant 'console' parameter from boot script

ttyS2 is the default console used for all rockchip boards.
The redundant 'console=tty1' parameter now breaks the console due to
recent procd update.

Signed-off-by: Tianling Shen <cnsztl@immortalwrt.org>
7 months agorockchip: remove 'swiotlb' parameter from boot script
Tianling Shen [Mon, 25 Mar 2024 16:40:37 +0000 (00:40 +0800)]
rockchip: remove 'swiotlb' parameter from boot script

We have hardware IOMMU support and this is totally unnecessary.
The given value is also unreasonable, it's too small and causes
kernel panic in some cases:

[ 5706.856473] sdhci-dwcmshc fe310000.mmc: swiotlb buffer is full (sz: 28672 bytes), total 512 (slots), used 498 (slots)
[ 5706.864451] sdhci-dwcmshc fe310000.mmc: swiotlb buffer is full (sz: 65536 bytes), total 512 (slots), used 464 (slots)

This parameter seems to be added by mistake, so remove it.

Fixes: e35c7ab51fd1 ("rockchip: merge bootscript")
Signed-off-by: Tianling Shen <cnsztl@immortalwrt.org>
7 months agoqualcommax: default to kernel 6.6
Robert Marko [Fri, 29 Mar 2024 10:38:49 +0000 (11:38 +0100)]
qualcommax: default to kernel 6.6

6.6 has been in testing on qualcommax for a while so it should be in a
good shape, but lets default to it to get a wider audience.

Signed-off-by: Robert Marko <robimarko@gmail.com>
7 months agoRevert "tools/xz: update to 5.6.1" (CVE-2024-3094)
Petr Štetiar [Fri, 29 Mar 2024 16:59:01 +0000 (16:59 +0000)]
Revert "tools/xz: update to 5.6.1" (CVE-2024-3094)

This reverts commit 714c91d1a63f29650abaa9cf69ffa47cf2c70297 as probably
the upstream xz repository and the xz tarballs have been backdoored.

References: https://www.openwall.com/lists/oss-security/2024/03/29/4.
Signed-off-by: Petr Štetiar <ynezz@true.cz>
7 months agoconfig: Enable ext4 journaling by default.
Jordan Woyak [Tue, 26 Mar 2024 01:56:06 +0000 (20:56 -0500)]
config: Enable ext4 journaling by default.

Not having a journal by default is a major "gotcha".

Because openwrt does not fsck on boot, a power loss without journaling
can result in a dirty filesystem that openwrt will mount as read-only
which requires intervention to restore the router to working order.

Signed-off-by: Jordan Woyak <jordan.woyak@gmail.com>
7 months agokernel: kmod-phy-smsc: add dependency on crc16
Tomasz Maciej Nowak [Fri, 29 Mar 2024 13:58:35 +0000 (14:58 +0100)]
kernel: kmod-phy-smsc: add dependency on crc16

Introduced WoL feature needs CRC16 support.

Signed-off-by: Tomasz Maciej Nowak <tmn505@gmail.com>
7 months agoperf: fix build on linux 6.6
Felix Fietkau [Sun, 24 Mar 2024 14:43:02 +0000 (15:43 +0100)]
perf: fix build on linux 6.6

- use Makefile.perf to prevent overriding MAKEFLAGS
- fix path to PKG_CONFIG
- link libstdc++ statically (only used for demangling)

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 months agotools/cmake: disable xcode generator
Felix Fietkau [Sun, 24 Mar 2024 09:59:22 +0000 (10:59 +0100)]
tools/cmake: disable xcode generator

Fixes build on latest macOS SDK

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 months agotools/e2fsprogs: disable building fuse2fs to avoid depending on libfuse
Felix Fietkau [Sun, 24 Mar 2024 09:40:31 +0000 (10:40 +0100)]
tools/e2fsprogs: disable building fuse2fs to avoid depending on libfuse

Signed-off-by: Felix Fietkau <nbd@nbd.name>
7 months agokernel: bump 5.15 to 5.15.152
Hauke Mehrtens [Tue, 26 Mar 2024 00:18:15 +0000 (01:18 +0100)]
kernel: bump 5.15 to 5.15.152

Removed because it is upstream:
generic/backport-5.15/081-v5.17-regmap-allow-to-define-reg_update_bits-for-no-bus.patch
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=linux-5.15.y&id=fbddd48f1456db32b675fad95a902de38345902a

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 months agokernel: bump 5.15 to 5.15.151
Hauke Mehrtens [Mon, 25 Mar 2024 22:54:57 +0000 (23:54 +0100)]
kernel: bump 5.15 to 5.15.151

No manual changes needed.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
7 months agolinux-firmware: add missing LICENSE_FILES info
Florian Eckert [Wed, 13 Mar 2024 12:22:20 +0000 (13:22 +0100)]
linux-firmware: add missing LICENSE_FILES info

Where it is clear which lincense the firmware package has, the missing
information are added.

Signed-off-by: Florian Eckert <fe@dev.tdt.de>
7 months agolinux-firmware: add LICENSE_FILES and LICENSE file handling
Florian Eckert [Wed, 13 Mar 2024 12:21:34 +0000 (13:21 +0100)]
linux-firmware: add LICENSE_FILES and LICENSE file handling

The firmware blobs have all different licenses from the different
manufacturers of the binary blobs. This information is contained in the
upstream 'linux-firmware' repositroy.

This commit extends the package handling so that this information can be
added as an additional argument during packages generation.

Signed-off-by: Florian Eckert <fe@dev.tdt.de>
7 months agoarchs38: remove 5.15 support
Robert Marko [Thu, 28 Mar 2024 12:25:09 +0000 (13:25 +0100)]
archs38: remove 5.15 support

Since 6.1 is now default, drop 5.15 support.

Signed-off-by: Robert Marko <robimarko@gmail.com>
7 months agoarchs38: default to kernel 6.1
Robert Marko [Thu, 28 Mar 2024 12:24:36 +0000 (13:24 +0100)]
archs38: default to kernel 6.1

Default to kernel 6.1 on archs38.

Signed-off-by: Robert Marko <robimarko@gmail.com>
7 months agogeneric: 6.6: add kernel missing symbol
Mieczyslaw Nalewaj [Thu, 28 Mar 2024 15:11:31 +0000 (16:11 +0100)]
generic: 6.6: add kernel missing symbol

Add missing symbol CONFIG_SSIF_IPMI_BMC causing bcm53xx compilation errors.

Signed-off-by: Mieczyslaw Nalewaj <namiltd@yahoo.com>
7 months agotools/cmake: update to 3.29.0
Nick Hainke [Fri, 29 Mar 2024 06:18:07 +0000 (07:18 +0100)]
tools/cmake: update to 3.29.0

Changelog:
- https://www.kitware.com/cmake-3-29-0-available-for-download/

Refresh patches:
- 120-curl-fix-libressl-linking.patch
- 130-bootstrap_parallel_make_flag.patch
- 150-zstd-libarchive.patch

Signed-off-by: Nick Hainke <vincent@systemli.org>
7 months agotools/xz: update to 5.6.1
Nick Hainke [Thu, 28 Mar 2024 20:36:29 +0000 (21:36 +0100)]
tools/xz: update to 5.6.1

Change mirror to github.

Signed-off-by: Nick Hainke <vincent@systemli.org>
7 months agobcm47xx: refresh kernel 6.1 config
Rafał Miłecki [Thu, 28 Mar 2024 23:10:42 +0000 (00:10 +0100)]
bcm47xx: refresh kernel 6.1 config

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
7 months agobcm47xx: add testing support for kernel 6.1
Rafał Miłecki [Thu, 28 Mar 2024 12:57:07 +0000 (13:57 +0100)]
bcm47xx: add testing support for kernel 6.1

Tested on Luxul XWR-1750 (BCM4706). Working: switch (LAN, WAN), LEDs,
sysupgrade.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 months agobcm4908: refresh kernel 6.6 config
Rafał Miłecki [Thu, 28 Mar 2024 12:28:12 +0000 (13:28 +0100)]
bcm4908: refresh kernel 6.6 config

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 months agobcm4008: prepare to work on kernel 6.6
Rafał Miłecki [Thu, 28 Mar 2024 11:25:54 +0000 (12:25 +0100)]
bcm4008: prepare to work on kernel 6.6

Don't add KERNEL_TESTING_PATCHVER yet as there are some issues with it.

On TP-Link Archer C2300 serial console seems to stop working after
preinit:
> Press the [f] key and hit [enter] to enter failsafe mode
> Press the [1], [2], [3] or [4] key and hit [enter] to select the de�

On Netgear R8000P XHCI causes external abort:
[    2.139586] Internal error: synchronous external abort: 0000000096000210 [#1] SMP
[    2.147212] Modules linked in: xhci_plat_hcd(+) xhci_hcd ohci_platform ohci_hcd fsl_mph_dr_of ehci_platform ehci_fsl ehci_hcd gpio_button_hotplug(O) usbcore nls_base usb_common crc32c_generic
[    2.164774] CPU: 0 PID: 358 Comm: kmodloader Tainted: G           O       6.6.22 #0
[    2.172658] Hardware name: Netgear R8000P (DT)
[    2.177229] pstate: 60400005 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[    2.184395] pc : xhci_gen_setup+0x80/0x34c [xhci_hcd]
[    2.189591] lr : xhci_gen_setup+0x74/0x34c [xhci_hcd]
[    2.194788] sp : ffffffc0815d37b0
[    2.198194] x29: ffffffc0815d37b0 x28: ffffff8002000000 x27: ffffff8001172d88
[    2.205540] x26: ffffff8002000000 x25: 0000000000000000 x24: ffffffc078b603c0
[    2.212888] x23: ffffffc078b2a008 x22: ffffff8001172c10 x21: ffffff8002000000
[    2.220235] x20: ffffff8002000000 x19: ffffff8002000250 x18: 0000000000000000
[    2.227582] x17: 626d756e20737562 x16: 2064656e67697373 x15: ffffffffffffffff
[    2.234929] x14: ffffff80019e9915 x13: ffffff80019e9913 x12: 00000000ffffffea
[    2.242276] x11: 00000000ffffefff x10: 0000000000000062 x9 : 00000000ffffffd0
[    2.246760] bcm63138_nand ff801800.nand-controller: timeout waiting for command 0x4
[    2.249623] x8 : 0000000000000073 x7 : ffffffc0815d37c0 x6 : 0000000000000075
[    2.257513] bcm63138_nand ff801800.nand-controller: intfc status c00000e0
[    2.264855] x5 : 0000000000000081 x4 : 0000000000000000 x3 : ffffff8001b61800
[    2.279193] x2 : ffffffc080b5d000 x1 : ffffff80020003a8 x0 : ffffff8002000398
[    2.286540] Call trace:
[    2.289048]  xhci_gen_setup+0x80/0x34c [xhci_hcd]

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 months agobcm4908: switch to the kernel 6.1
Rafał Miłecki [Thu, 28 Mar 2024 11:01:44 +0000 (12:01 +0100)]
bcm4908: switch to the kernel 6.1

Tested on Netgear R8000P and TP-Link Archer C2300.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 months agobcm4908: use DT to mark NAND controller WP pin as not connected
Rafał Miłecki [Thu, 28 Mar 2024 11:00:04 +0000 (12:00 +0100)]
bcm4908: use DT to mark NAND controller WP pin as not connected

Use cleaner and more upstream-fit solution.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 months agoairoha: remove 5.15 support
Robert Marko [Tue, 26 Mar 2024 11:57:30 +0000 (12:57 +0100)]
airoha: remove 5.15 support

Now that 6.1 is default, remove 5.15 support.

Signed-off-by: Robert Marko <robimarko@gmail.com>
8 months agoairoha: default to kernel 6.1
Robert Marko [Tue, 26 Mar 2024 11:56:34 +0000 (12:56 +0100)]
airoha: default to kernel 6.1

6.1 has been present as testing for a while now, so default to it.

Signed-off-by: Robert Marko <robimarko@gmail.com>
8 months agokernel: backport brcmnand support for "brcm,wp-not-connected"
Rafał Miłecki [Thu, 28 Mar 2024 07:41:51 +0000 (08:41 +0100)]
kernel: backport brcmnand support for "brcm,wp-not-connected"

It's required to support NAND controllers with WP pin on boards that
don't have it connected to NAND chip.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 months agobcm53xx: refresh kernel 6.6 config
Rafał Miłecki [Wed, 27 Mar 2024 23:11:37 +0000 (00:11 +0100)]
bcm53xx: refresh kernel 6.6 config

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 months agobcm53xx: add testing support for kernel 6.6
Rafał Miłecki [Wed, 27 Mar 2024 22:59:46 +0000 (23:59 +0100)]
bcm53xx: add testing support for kernel 6.6

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 months agobcm53xx: switch to the kernel 6.1
Rafał Miłecki [Wed, 27 Mar 2024 10:58:38 +0000 (11:58 +0100)]
bcm53xx: switch to the kernel 6.1

Successfully verified on Luxul XWR-3150 (BCM47094). Tested LAN, WAN,
Wi-Fi, USB, sysupgrade.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 months agobcm53xx: refresh kernel 6.1 config
Rafał Miłecki [Wed, 27 Mar 2024 10:58:18 +0000 (11:58 +0100)]
bcm53xx: refresh kernel 6.1 config

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 months agokernel: fix NVMEM looking for OTP NVMEM cells in wrong DT node
Rafał Miłecki [Wed, 27 Mar 2024 23:03:34 +0000 (00:03 +0100)]
kernel: fix NVMEM looking for OTP NVMEM cells in wrong DT node

It was breaking NAND controllers drivers.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
8 months agosunxi: backport h616 thermal sensor support
Chukun Pan [Sat, 9 Mar 2024 15:16:25 +0000 (23:16 +0800)]
sunxi: backport h616 thermal sensor support

Backport H616 thermal sensor support from linux-next.
Tested on the Orange Pi Zero 3 (H618 SoC).

Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>
8 months agosunxi: add support for Orange Pi Zero 3
Chukun Pan [Sat, 18 Nov 2023 15:10:25 +0000 (23:10 +0800)]
sunxi: add support for Orange Pi Zero 3

Key features:
  Allwinner H618 SoC (Quad core Cortex-A53)
  1/1.5/2/4 GiB LPDDR4 DRAM
  1 USB 2.0 type C port (Power + OTG)
  1 USB 2.0 host port
  1Gbps Ethernet port
  Micro-HDMI port
  MicroSD slot

Installation:
  Write the image to SD Card with dd.

Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>
8 months agogeneric: 6.1: backport AXP PMIC support
Chukun Pan [Fri, 8 Dec 2023 15:10:21 +0000 (23:10 +0800)]
generic: 6.1: backport AXP PMIC support

Backport AXP15060, AXP313a and AXP192 support.
The AXP15060 PMIC is used for starfive boards,
and the AXP313a PMIC is used for sunxi boards.
Remove conflicting patches from starfive target.

Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>
8 months agouboot-sunxi: bump to 2024.01
Chukun Pan [Fri, 5 Jan 2024 15:16:16 +0000 (23:16 +0800)]
uboot-sunxi: bump to 2024.01

This version supports LPDDR4 DRAM of H618 SoC.

Runtime-tested:
  Olimex Olinuxino Micro (A20)
  Orange Pi Zero 3 (H618)
  Pine64 SoPine (A64)

Tested-by: Zoltan HERPAI <wigyori@uid0.hu>
Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>
8 months agokernel: qca-ssdk: fix C45 MDIO support on kernel 6.6
Robert Marko [Tue, 26 Mar 2024 11:36:27 +0000 (12:36 +0100)]
kernel: qca-ssdk: fix C45 MDIO support on kernel 6.6

Kernel 6.3 has introduced separate C45 read/write operations, and thus
split them out of the C22 operations completely so the old way of marking
C45 reads and writes via the register value does not work anymore.

This is causing SSDK to fail and find C45 only PHY-s such as Aquantia ones:
[   22.187877] ssdk_phy_driver_init[371]:INFO:dev_id = 0, phy_adress = 8, phy_id = 0x0 phytype doesn't match
[   22.209924] ssdk_phy_driver_init[371]:INFO:dev_id = 0, phy_adress = 0, phy_id = 0x0 phytype doesn't match

This in turn causes USXGMII MAC autoneg bit to not get set and then UNIPHY
autoneg will time out, causing the 10G ports not to work:
[   37.292784] uniphy autoneg time out!

So, lets detect C45 reads and writes by the magic BIT(30) in the register
argument and if so call separate C45 mdiobus read/write functions.

Signed-off-by: Robert Marko <robimarko@gmail.com>
8 months agotools/expat: update to version 2.6.2
Josef Schlehofer [Tue, 26 Mar 2024 09:37:37 +0000 (10:37 +0100)]
tools/expat: update to version 2.6.2

Release notes:
https://github.com/libexpat/libexpat/blob/R_2_6_2/expat/Changes

Fixes:
CVE-2024-28757

Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
8 months agogeneric: phy-mediatek-xfi-tphy: fix SGMII issue
Daniel Golle [Mon, 25 Mar 2024 21:58:49 +0000 (21:58 +0000)]
generic: phy-mediatek-xfi-tphy: fix SGMII issue

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: https://git01.mediatek.com/plugins/gitiles/openwrt/feeds/mtk-openwrt-feeds/+/b72d6cba92bf9e29fb035c03052fa1e86664a25b
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
8 months agorealtek: add Zyxel GS1900-8 v2
Goetz Goerisch [Sun, 24 Mar 2024 17:31:06 +0000 (18:31 +0100)]
realtek: add Zyxel GS1900-8 v2

The Zyxel GS1900-8 v2 or Rev.B1 is a newer variant of the GS1900-8, but
otherwise similar to the other GS1900 switches.

Differences
------------
* Front Button labeled RESTORE
* NO Power Switch on rear
* Serial Header next to the barrel power connector
* Part Number ends 0102F

Signed-off-by: Goetz Goerisch <ggoerisch@gmail.com>
8 months agomediatek: filogic: add Unielec U7981-01 support
Allen Zhao [Sat, 9 Mar 2024 02:44:58 +0000 (10:44 +0800)]
mediatek: filogic: add Unielec U7981-01 support

Hardware specification:
  SoC: MediaTek MT7981B 2x A53
  Flash: 8GB eMMC or 128 MB SPI-NAND
  RAM: 256MB
  Ethernet: 5x 10/100/1000 Mbps
  Switch: MediaTek MT7531AE
  WiFi: MediaTek MT7976C
  Button: Reset
  USB: M.2(B-key) for 4G/5G Module
  Power: DC 12V 1A
  UART: 3.3v, 115200n8
  --------------------------
  |         Layout         |
  |   -----------------    |
  | 4 | VCC RX TX GND | <= |
  |   -----------------    |
  --------------------------

The U-boot menu will automatically appear at startup, and then select
the required options through UP/DOWN Key.

NAND Flash and eMMC Flash instructions:
1. Set your computers IP adress to 192.168.1.2.
2. Run a TFTP server providing the sysupgrade.bin image.
3. Power on the router, into the U-Boot menu.
4. Select "2. Upgrade firmware"
5. Update sysupgrade.bin file name, input server IP and input device
   IP (if they deviate from the defaults)
6. Wait for automatic startup after burning

Signed-off-by: Allen Zhao <allenzhao@unielecinc.com>
8 months agoprocd: update to Git HEAD (2024-03-25)
Robert Marko [Mon, 25 Mar 2024 10:50:06 +0000 (11:50 +0100)]
procd: update to Git HEAD (2024-03-25)

79f8cfa58ee7 ci: add github test workflow
428f40e7984f test commit fixing warnings
63058d1f81a5 ci: enable ujail builds
49ea930a862c utils: add key-value offset support to get_cmdline_val()
ca8c30208d5e inittab: fallback when multiple "console=" is detected

Required for the recent Elecom multiple console commits.

Signed-off-by: Robert Marko <robimarko@gmail.com>
8 months agohostapd: fix Argument list too long build error
Robert Marko [Sun, 24 Mar 2024 19:54:26 +0000 (20:54 +0100)]
hostapd: fix Argument list too long build error

Currently, both CI and local builds of wpa-supplicant will fail with:
/bin/sh: Argument list too long

Its happening as the argument list for mkdir in build.rules is too large
and over the MAX_ARG_STRLEN limit.

It seems that recent introduction of APK compatible version schema has
increased the argument size and thus pushed it over the limit uncovering
the issue.

Fixes: e8725a932e16 ("treewide: use APK compatible version schema")
Signed-off-by: Robert Marko <robimarko@gmail.com>
8 months agotreewide: update PKG_MIRROR_HASH after APK version schema
Paul Spooren [Sat, 23 Mar 2024 12:07:25 +0000 (13:07 +0100)]
treewide: update PKG_MIRROR_HASH after APK version schema

With the change in version schema the downloaded files changed, too,
mostly the hash is now prefixed with a tilde `~` instead of a dash `-`.

Since each downloaded archive contains folder with the same name as the
archive, the checksum changed.

Signed-off-by: Paul Spooren <mail@aparcar.org>
8 months agompc85xx: add 6.6 testing kernel
Pawel Dembicki [Wed, 13 Mar 2024 11:35:04 +0000 (12:35 +0100)]
mpc85xx: add 6.6 testing kernel

This commit adds 6.6 kernel as testing verion.

Run tested:
p2020/p2020-rdb
p1010/TL-WDR4900-V1
p1010/BR200-WP

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
8 months agompc85xx: kernel: 6.6: refresh config
Pawel Dembicki [Wed, 13 Mar 2024 10:07:13 +0000 (11:07 +0100)]
mpc85xx: kernel: 6.6: refresh config

config-6.6 refreshed by 'make kernel_oldconfig'
p2020 config-default refreshed manually

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
8 months agompc85xx: kernel: 6.6: refresh patches
Pawel Dembicki [Wed, 13 Mar 2024 10:01:01 +0000 (11:01 +0100)]
mpc85xx: kernel: 6.6: refresh patches

Manualy refreshed:
101-powerpc-85xx-hiveap-330-support.patch
105-powerpc-85xx-panda-support.patch

Rest done by 'make target/linux/refresh'

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
8 months agokernel/mpc85xx: Restore kernel files for v6.1
Pawel Dembicki [Wed, 13 Mar 2024 09:38:57 +0000 (10:38 +0100)]
kernel/mpc85xx: Restore kernel files for v6.1

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.

See: https://lists.openwrt.org/pipermail/openwrt-devel/2023-October/041673.html
for the original discussion.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
8 months agokernel/mpc85xx: Create kernel files for v6.6 (from v6.1)
Pawel Dembicki [Wed, 13 Mar 2024 09:38:57 +0000 (10:38 +0100)]
kernel/mpc85xx: Create kernel files for v6.6 (from v6.1)

This is an automatically generated commit.

During a `git bisect` session, `git bisect --skip` is recommended.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
8 months agompc85xx: refresh kernel config
Pawel Dembicki [Wed, 13 Mar 2024 09:38:08 +0000 (10:38 +0100)]
mpc85xx: refresh kernel config

Done by 'make kernel_oldconfig'.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
8 months agokernel: generic: add missing symbol
Pawel Dembicki [Wed, 13 Mar 2024 09:35:47 +0000 (10:35 +0100)]
kernel: generic: add missing symbol

Found durring 'make kernel_oldconfig'.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
8 months agompc85xx: remove 5.15 support
Pawel Dembicki [Wed, 13 Mar 2024 07:50:06 +0000 (08:50 +0100)]
mpc85xx: remove 5.15 support

Kernel is switched to 6.1. Lets remove 5.15 support.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
8 months agompc85xx: switch to 6.1 kernel
Pawel Dembicki [Wed, 13 Mar 2024 07:48:39 +0000 (08:48 +0100)]
mpc85xx: switch to 6.1 kernel

We need to prepare for 6.6. Switch to 6.1 for wider testing.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
8 months agompc85xx: copy missing patch from 5.15 to 6.1
Pawel Dembicki [Wed, 13 Mar 2024 12:13:51 +0000 (13:13 +0100)]
mpc85xx: copy missing patch from 5.15 to 6.1

patch 150-arch-powerpc-simpleboot-prevent-overwrite-of-CPU1-sp.patch
was missed durring copy from 5.15 to 6.1.

This patch restore it.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
8 months agoimx: kernel: add imx8mp PCI support
Tim Harvey [Mon, 22 Jan 2024 23:30:18 +0000 (15:30 -0800)]
imx: kernel: add imx8mp PCI support

IMX8MP PCI support requires a few patches backported from Linux 6.2.

Signed-off-by: Tim Harvey <tharvey@gateworks.com>
8 months agoimx: add Gateworks Venice support
Tim Harvey [Fri, 12 Jan 2024 19:34:28 +0000 (11:34 -0800)]
imx: add Gateworks Venice support

Add support for Gateworks Venice imx8m family of boards:
 - required kernel modules for on-board devices
 - image generation
 - initial network config
 - sysupgrade support

The resulting compressed disk image
(bin/targets/imx/cortexa53/openwrt-imx-cortexa53-gateworks_venice-squashfs-img.gz)
can be installed on a Gateworks venice board via U-Boot:

u-boot=> tftpboot $loadaddr openwrt-imx-cortexa53-gateworks_venice-squashfs-img.gz && \
         gzwrite mmc $dev $loadaddr $filesize

WARNING: this will overwrite any boot firmware on the eMMC user hardware
partition which if being used will brick your board requiring JTAG to
re-program boot firmware and recover

The compressed disk image contains the partition table and filesystems only
and that it is expected that boot firmware is installed properly on the
eMMC boot0 hardware partition. The easiest way to ensure this is to
use the Gateworks JTAG adapter/process to install the latest boot firmware
as follows from a Linux host:
  wget http://dev.gateworks.com/jtag/jtag_usbv4
  chmod +x jtag_usbv4
  wget http://dev.gateworks.com/venice/images/firmware-venice-imx8mm.bin
  sudo ./jtag_usbv4 -p firmware-venice-imx8mm.bin

Signed-off-by: Tim Harvey <tharvey@gateworks.com>
8 months agoimx: add imx8m support
Tim Harvey [Mon, 8 Jan 2024 22:04:15 +0000 (14:04 -0800)]
imx: add imx8m support

Add imx8m support:
 - add a cortexa53 subtarget to imx
 - move ARCH and KERNELNAME to subtargets
 - account for kernel modules that are not used for cortexa53

No device-specific targets or firmware images are created yet but all
imx8m* dtbs will be built.

enabling CONFIG_TARGET_ROOTFS_INITRAMFS results in
openwrt-imx-cortexa53-imx8m-initramfs-kernel.bin which has been
successfully booted on an imx8mm-evk using the following:

u-boot=> tftpboot $fdt_addr_r image-imx8mm-evk.dtb && \
tftpboot $kernel_addr_r openwrt-imx-cortexa53-imx8m-initramfs-kernel.bin && \
booti $kernel_addr_r - $fdt_addr_r

Signed-off-by: Tim Harvey <tharvey@gateworks.com>
8 months agolinux-firmware: package Intel AX201 firmware
Petr Štetiar [Mon, 4 Mar 2024 07:16:19 +0000 (07:16 +0000)]
linux-firmware: package Intel AX201 firmware

Alexander reported following:

 iwlwifi 0000:00:14.3: Detected crf-id 0x3617, cnv-id 0x20000302 wfpm id 0x80000000
 iwlwifi 0000:00:14.3: PCI dev a0f0/0074, rev=0x351, rfid=0x10a100
 iwlwifi 0000:00:14.3: Direct firmware load for iwlwifi-QuZ-a0-hr-b0-77.ucode failed with error -2

It seems, that as of the current date, the highest firmware API version
supported by Linux 6.8-rc7 is still 77.

Closes: #14771
Signed-off-by: Petr Štetiar <ynezz@true.cz>
8 months agoath79: register ttyATH1 as OpenWrt console for ELECOM WAB-I1750-PS
INAGAKI Hiroshi [Mon, 11 Mar 2024 14:30:26 +0000 (23:30 +0900)]
ath79: register ttyATH1 as OpenWrt console for ELECOM WAB-I1750-PS

Add a hotplug script and add ttyATH1 on ELECOM WAB-I1750-PS to
/etc/inittab while booting for using that console as an OpenWrt console.

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
8 months agoath79: register all ttys as Linux console for ELECOM WAB-I1750-PS
INAGAKI Hiroshi [Tue, 12 Mar 2024 09:16:12 +0000 (18:16 +0900)]
ath79: register all ttys as Linux console for ELECOM WAB-I1750-PS

Register ttyS0 and ttyATH1 as Linux console on ELECOM WAB-I1750-PS.
ttyS0 provides "SERVICE" port and internal pin header for debugging and
recoverying by maker, ttyATH1 provides "SERIAL" port for configuration
by users.

Signed-off-by: INAGAKI Hiroshi <musashino.open@gmail.com>
8 months agokirkwood: resize kernel partition for kirkwood devices
Pawel Dembicki [Thu, 14 Mar 2024 12:57:25 +0000 (13:57 +0100)]
kirkwood: resize kernel partition for kirkwood devices

The 6.1 kernel has caused another increase in kernel size, and now
it's more than 3MB:

WARNING: Image file iom_ix4-200d-uImage is too big: 3170394 > 3145728
WARNING: Image file iom_ix2-200-uImage is too big: 3171494 > 3145728
WARNING: Image file linksys_e4200-v2-uImage is too big: 3171879 > 3145728
WARNING: Image file linksys_ea4500-uImage is too big: 3171871 > 3145728
WARNING: Image file linksys_ea3500-uImage is too big: 3171651 > 3145728

This causes problems for 5 devices:
  - Iomega StorCenter ix2/ix4
  - Linksys EA3500/EA4200/EA4500

They have enough resources for proper operation with 6.1, but all of
them had a 3MB kernel size limit. Let's keep them alive and
resize kernel partitions to 4MB.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
8 months agokirkwood: remove 5.15 kernel support
Pawel Dembicki [Tue, 5 Mar 2024 21:34:58 +0000 (22:34 +0100)]
kirkwood: remove 5.15 kernel support

Kernel was switched to 6.1. Lets drop 5.15 support.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
8 months agokirkwood: switch kernel version to 6.1
Pawel Dembicki [Tue, 5 Mar 2024 21:26:56 +0000 (22:26 +0100)]
kirkwood: switch kernel version to 6.1

6.1 testing version was introduced some time ago. Kernel size issues
are resolved now. Time to bump kernel version.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
8 months agokirkwood: Add missing package dependency
Pawel Dembicki [Wed, 6 Mar 2024 09:50:08 +0000 (10:50 +0100)]
kirkwood: Add missing package dependency

When build images with the 6.1 kernel for all devices and all kmods,
there is one dependancy error:

  pkg_hash_check_unresolved: cannot find dependency kmod-thermal
  for kmod-hwmon-gpiofan

This commit fix that issue.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
8 months agokirkwood: 6.1: backport Marvell 88E6171 init fix patch
Pawel Dembicki [Tue, 5 Mar 2024 20:52:52 +0000 (21:52 +0100)]
kirkwood: 6.1: backport Marvell 88E6171 init fix patch

After commit de5c9bf40c45 ("net: phylink: require supported_interfaces to
be filled") Marvell 88e6171 switches fail to be probed:

[    1.263852] mv88e6085 f1072004.mdio-bus-mii:10: switch 0x1710 detected: Marvell 88E6171, revision 2
[    2.177654] mv88e6085 f1072004.mdio-bus-mii:10: OF node /ocp@f1000000/mdio-bus@72004/switch@10/ports/port@5 of CPU port 5 lacks the required "phy-mode" property
[    2.194230] mv88e6085 f1072004.mdio-bus-mii:10: phylink: error: empty supported_interfaces
[    2.202554] error creating PHYLINK: -22
[    2.214109] mv88e6085: probe of f1072004.mdio-bus-mii:10 failed with error -22

Upstream handled it. Lets backport patch.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
8 months agokirkwood: Add missing phy-mode and fixed links
Pawel Dembicki [Tue, 5 Mar 2024 14:23:31 +0000 (15:23 +0100)]
kirkwood: Add missing phy-mode and fixed links

Copied from original Andrew's Lunn commit message:

    The DSA framework has got more picky about always having a phy-mode
    for the CPU port. The Kirkwood Ethernet is an RGMII port. Set the
    switch to impose the RGMII delays.

    Additionally, the cpu label has never actually been used in the
    binding, so remove it.

This commit backport change from upstream and fix downstream EA3500 dts.

Signed-off-by: Pawel Dembicki <paweldembicki@gmail.com>
8 months agotoolchain: gcc: switch default to 13
Nick Hainke [Fri, 12 May 2023 16:11:10 +0000 (18:11 +0200)]
toolchain: gcc: switch default to 13

Use GCC 13 instead of GCC 12 by default.

All target kernels are building with GCC 13.
Most packages from the feed are building fine.

The root file systems is getting a little bit smaller for MIPS 32 BE
and aarch64.

With GCC 12 I got these sizes for lantiq/xrx200:
7,005,867 openwrt-lantiq-xrx200-tplink_tdw8970-initramfs-kernel.bin

With GCC 13 I got these sizes for lantiq/xrx200:
6,989,754 openwrt-lantiq-xrx200-tplink_tdw8970-initramfs-kernel.bin

With GCC 12 I got these sizes for armsr/armv8:
13,083,836 openwrt-armsr-armv8-generic-ext4-combined.img.gz
 4,900,240 openwrt-armsr-armv8-generic-ext4-rootfs.img.gz
20,142,592 openwrt-armsr-armv8-generic-kernel.bin

With GCC 13 I got these sizes for armsr/armv8:
13,068,966 openwrt-armsr-armv8-generic-ext4-combined.img.gz
 4,893,078 openwrt-armsr-armv8-generic-ext4-rootfs.img.gz
20,142,592 openwrt-armsr-armv8-generic-kernel.bin

Signed-off-by: Nick Hainke <vincent@systemli.org>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
8 months agoramips: add support for TP-Link RE205 v3
Steffen Loley [Sat, 10 Feb 2024 15:50:50 +0000 (16:50 +0100)]
ramips: add support for TP-Link RE205 v3

TP-Link RE205 v3 is a wireless range extender with Ethernet and 2.4G and 5G
WiFi with external antennas.
It's based on MediaTek MT7628AN+MT7610EN like the RE200 v3/v4 but with
external antennas.

Specifications
--------------

- MediaTek MT7628AN (580 Mhz)
- 64 MB of RAM
- 8 MB of FLASH
- 2T2R 2.4 GHz and 1T1R 5 GHz
- 1x 10/100 Mbps Ethernet
- 5x LED (GPIO-controlled), 2x button
- UART connection holes on PCB (57600 8n1)

There are 2.4G and 5G LEDs in blue which are controlled separately.

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

Installation is identical to RE200 v3 devices as described at
https://openwrt.org/toh/tp-link/re200#installation

Web Interface
-------------

It is possible to upgrade to OpenWrt via the web interface. Simply flash
the -factory.bin from OEM. In contrast to a stock firmware, this will not
overwrite U-Boot.

Recovery
--------

U-Boot seems to be locked on newer versions, if not it can be accessed over
the UART as described in the link above.

Signed-off-by: Steffen Loley <slo-src@web.de>
8 months agouml: restore 6.1 config + patches
Christian Lamparter [Sat, 23 Mar 2024 19:35:28 +0000 (20:35 +0100)]
uml: restore 6.1 config + patches

keep the main 6.1 files around until they too will be deleted.

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
8 months agouml: add 6.6 testing kernel config + patches
Christian Lamparter [Sat, 23 Mar 2024 19:34:40 +0000 (20:34 +0100)]
uml: add 6.6 testing kernel config + patches

apart from refreshes to both config and patches, nothing has changed.

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
8 months agouml: remove 5.15 config + patches
Christian Lamparter [Sat, 23 Mar 2024 19:35:59 +0000 (20:35 +0100)]
uml: remove 5.15 config + patches

no longer in use as we migrate to 6.6 shortly.

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
8 months agouml: promote to 6.1 kernel
Christian Lamparter [Sat, 23 Mar 2024 19:32:20 +0000 (20:32 +0100)]
uml: promote to 6.1 kernel

there have been no reports of any problems with 6.1.
let's make it the default kernel.

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
8 months agoapm821xx: WNDR4700: fix fan-control
Christian Lamparter [Sat, 23 Mar 2024 18:50:22 +0000 (19:50 +0100)]
apm821xx: WNDR4700: fix fan-control

This service was unfunctional due to not having its executable bit set.
Furthermore, sysupgrade complains about the file being present in images.

Also, the tc654 driver doesn't provide pwm1_enable sysfs file, instead its
now called pwm1_mode.

Please note that the fan was always spinning. It should now turn off, when
there's enough the thermal headroom.

Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
8 months agotreewide: disable spectre mitigation on unaffected Arm64 targets
Qingfang Deng [Mon, 26 Feb 2024 08:38:53 +0000 (16:38 +0800)]
treewide: disable spectre mitigation on unaffected Arm64 targets

Cortex-A53 is not vulnerable to any Spectre variants.

Ref: https://developer.arm.com/Arm%20Security%20Center/Speculative%20Processor%20Vulnerability
Signed-off-by: Qingfang Deng <dqfext@gmail.com>
8 months agotreewide: disable spectre mitigation on unaffected Arm32 targets
Qingfang Deng [Mon, 26 Feb 2024 08:38:52 +0000 (16:38 +0800)]
treewide: disable spectre mitigation on unaffected Arm32 targets

Cortex-A5 and Cortex-A7 are not vulnerable to any Spectre variants.

Ref: https://developer.arm.com/Arm%20Security%20Center/Speculative%20Processor%20Vulnerability
Signed-off-by: Qingfang Deng <dqfext@gmail.com>
8 months agolibiwinfo: update to Git HEAD (2024-03-23)
Hauke Mehrtens [Sat, 23 Mar 2024 18:34:22 +0000 (19:34 +0100)]
libiwinfo: update to Git HEAD (2024-03-23)

3aa2b6b devices: add device id for MediaTek MT7601U
79a9615 devices: add device id for Realtek RTL8188CU and RTL8188FTV

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
8 months agorealtek: support common GPIOs on D-Link DGS-1210-16
Richard Kunze [Wed, 20 Mar 2024 18:26:27 +0000 (19:26 +0100)]
realtek: support common GPIOs on D-Link DGS-1210-16

D-Link DGS-1210-16 hangs when rebooting and has no support for the reset
button.

Fix both by enabling the same GPIOs for reboot and the reset button as
already used for D-Link DGS-1210-20 and D-Link DGS-1210-28.

Signed-off-by: Richard Kunze <kunze@tivano.de>
8 months agokernel: fix iptables offload build without nftables
Qingfang Deng [Tue, 12 Mar 2024 02:05:13 +0000 (10:05 +0800)]
kernel: fix iptables offload build without nftables

When rebasing patches on top of 6.1, a change that removed NF_TABLES
from NF_FLOW_TABLE's dependency was dropped accidentally, and iptables
flow offload module can only be built when nftables is selected.
Restore the change to fix it.

Fixes: 19a246bb6503 ("generic: 6.1: manually refresh hack patches")
Signed-off-by: Qingfang Deng <dqfext@gmail.com>
8 months agokernel: support RISC-V PMU
Qingfang Deng [Tue, 12 Mar 2024 02:08:30 +0000 (10:08 +0800)]
kernel: support RISC-V PMU

Add RISC-V PMU config symbols and enable them for riscv64.

Signed-off-by: Qingfang Deng <dqfext@gmail.com>