mdadm: backport musl patch

Needed to fix compilation with GCC14.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16912
Signed-off-by: Robert Marko <robimarko@gmail.com>

build: remove broken dependency of metadata on toplevel .config variables

Instead of relying on .config symbols for metadata, alter the DEFAULT
variable of affected packages. Fixes enabling opkg vs apk among others.

Signed-off-by: Felix Fietkau <nbd@nbd.name>

spidev_test: fix version for APK

Refactor version of spidev_test for APK.

Signed-off-by: Mieczyslaw Nalewaj <namiltd@yahoo.com>
Link: https://github.com/openwrt/openwrt/pull/16917
Signed-off-by: Robert Marko <robimarko@gmail.com>

config: Use APK as package manager by default

This commit comes after a long period of hard work, starting back in
early 2021 as a proof of concept.

Thanks to the Alpine Linux project for creating such a nice package
manager. Thanks to everyone involved; this is going to be great!

Signed-off-by: Paul Spooren <mail@aparcar.org>

nftables: update to 1.1.1

Release Announcement:
https://lwn.net/Articles/982283/
https://lwn.net/Articles/992760/

Link: https://github.com/openwrt/openwrt/pull/16769
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

libnftnl: update to 1.2.8

Release Announcement:
https://www.spinics.net/lists/netfilter-devel/msg87287.html
https://www.spinics.net/lists/netfilter-devel/msg88520.html

Link: https://github.com/openwrt/openwrt/pull/16769
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

lantiq: use devm for mutex_init

It's common to avoid calling mutex_destroy when done. It's not correct strictly speaking.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16832
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

lantiq: use more devm for i2c

i2c_add_numbered_adapter is the wrong function to use here. It requires
setting nr to some value, otherwise it behaves the same as
i2c_add_adapter. nr is not set.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16825
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

gdb: Fix description for apk, remove backtick

Remove backtick from gdb description text, as that seems to
be recognized as a shell action by compilation with apk,
causing error.

Example from test buildbot:
 rstrip.sh: /builder/shared-workdir/build/sdk/build_dir/target-aarch64_generic_musl/gdb-15.2/ipkg-aarch64_generic/gdb/usr/bin/gdb: executable
 bash: -c: line 1: unexpected EOF while looking for matching ``'
 bash: -c: line 2: syntax error: unexpected end of file
 make[3]: *** [Makefile:123: /builder/shared-workdir/build/sdk/bin/packages/aarch64_generic/base/gdb-15.2-r1.apk] Error 2

Local compilation:
 rstrip.sh: /OpenWrt/aarch64/build_dir/target-aarch64_cortex-a53_musl/gdb-15.2/ipkg-aarch64_cortex-a53/gdb/usr/bin/gdb: executable
 bash: -c: line 1: unexpected EOF while looking for matching ``'
 make[2]: *** [Makefile:123: /OpenWrt/aarch64/bin/packages/aarch64_cortex-a53/base/gdb-15.2-r1.apk] Error 2

Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
Link: https://github.com/openwrt/openwrt/pull/16908
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

wolfssl: Adjust version for apk

Adjust wolfssl version for apk by removing the "-stable"
from the OpenWrt version, although it is still needed for
upstream download archive name.

Define PKG_BUILD_DIR accordingly.

Utilize new short version to simplify ABI_VERSION calculation.

Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
Link: https://github.com/openwrt/openwrt/pull/16906
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>

base-files: Remove pkg_check when using apk

/sbin/pkg_check uses opkg and is not even packaged when using the
default opkg configuration. remove it when using apk too.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

kernel: bump 6.6 to 6.6.60

Changelog: https://cdn.kernel.org/pub/linux/kernel/v6.x/ChangeLog-6.6.60

Removed upstreamed:
generic/backport-6.6/409-mtd-spi-nor-winbond-fix-w25q128-regression.patch

All other patches automatically rebased.

1. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.6.60&id=27a58a19bd20a7afe369da2ce6d4ebea70768acd

Build system: x86/64
Build-tested: x86/64/AMD Cezanne, flogic/glinet_gl-mt6000, flogic/xiaomi_redmi-router-ax6000-ubootmod, ramips/tplink_archer-a6-v3
Run-tested: x86/64/AMD Cezanne, flogic/glinet_gl-mt6000, flogic/xiaomi_redmi-router-ax6000-ubootmod, ramips/tplink_archer-a6-v3

Signed-off-by: John Audia <therealgraysky@proton.me>
Link: https://github.com/openwrt/openwrt/pull/16892
Signed-off-by: Nick Hainke <vincent@systemli.org>

imagebuilder: fix OPKG package inclusion

In 451e2ce006 the code would only move packages separated by dashes,
however OPKG uses underscores to split between package name and version.
Result was the lack of such packages in the packed ImageBuilder.

Use dashes or underscores depending on the select package manger.

Fixes: 451e2ce006 "imagebuilder: fix package inclusion and APK index"
Signed-off-by: Paul Spooren <mail@aparcar.org>

treewide: remove THIS_MODULE assignment

Matches upstream coccinelle check: api/platform_no_drv_owner.cocci.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16846
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>

hostapd: split long lines

These two were getting rather long.

Signed-off-by: Leon M. Busch-George <leon@georgemail.eu>
Link: https://github.com/openwrt/openwrt/pull/16849
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>

menuconfig: add help description for Stack Smashing Protection options

Descriptions taken from:
https://wiki.osdev.org/Stack_Smashing_Protector

Signed-off-by: Marius Dinu <m95d+git@psihoexpert.ro>
Link: https://github.com/openwrt/openwrt/pull/16897
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>

apk: Implement apk list --full patch

Implement apk list --full patch to mimik opkg package info.

Link: https://github.com/openwrt/openwrt/pull/16759
Link: https://github.com/openwrt/openwrt/pull/16759
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>

apk: Bump to version 2024-11-08

Bump APK to version 2024-11-08, multiple fixup and pending arch patch
for ARM Big Endian merged upstream.

Link: https://github.com/openwrt/openwrt/pull/16759
Link: https://github.com/openwrt/openwrt/pull/16759
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>

ramips: mt7621: add missing regulator-boot-on

What seems to be happening is that the kernel requests an ACTIVE_LOW
gpio initially and sets it to high later based on gpios in dts.

This seems to break some devices where the bootloader sets it to high.

Fixes: e612900ae0 ("ramips: mt7621: convert usb power to regulators")
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16877
Signed-off-by: Robert Marko <robimarko@gmail.com>

imagebuilder: init APK dirs every time

The `--initdb` command creates basic folders required by APK,
previoiusly it would only run a single time when package_index is
actually called. Since the function isn't called if nothing changes,
`--initdb` doesn't initialize the rootfs again.

This commit moves it to package_reload, which runs every time.

Signed-off-by: Paul Spooren <mail@aparcar.org>

imagebuilder: cleanup package inclusion

Previously three different ways to include packages in an ImageBuilder
existed:

* buildbot: include libc, kernel (and base-files) in $(IB_LDIR)
* not buildbot, standalone: include all packages in ./packages/
* not buildbot, not standalone: include libc, kernel (and base-files) in
  ./packages/

First of, the separation between *buildbot* and *not buildbot, not
standalone* is not required, we can just always copy packages to
./packages instead of ever using the special place $(IB_LDIR).
Doing so drops the need to handle the extra case and also allows to
clean up the OPKG package installation, which no longer requries the
`firstword` logic, things are now always at ./packages.

Signed-off-by: Paul Spooren <mail@aparcar.org>

imagebuilder: fix package inclusion and APK index

We need to exclude packages from the APK index which must not be
upgraded. To do so, the packages `libc`, `kernel` and `base-files` are
excluded to APK never suggestes them as upgradable.

The previous logic would however match packages like `libcomerr0`,
causing build failures. Make the copying and excluding logic more
precise by adding a single dash.

Signed-off-by: Paul Spooren <mail@aparcar.org>

imagebuilder: fix APK for packages dir and cache

This commit solves multiple issues. First of just install the three
special packages base-files, libc and kernel directly from the index. In
upstream indexes, those will never appear to prevent accidental upgrades
may breaking the system.

Next, enable caching for the ImageBuilder, which speeds up consecutive
builds from ~33 seconds to ~5 seconds. Using cache however makes APK
create the folder `/var/cache/apk/` which conflicts with the base-files
installation, which ships a symlink from `/var` to `/tmp`, so specify
`--no-cache` for the rootfs initialization.

Lastly, drop the use of `apk update` since APK automatically does that.

Signed-off-by: Paul Spooren <mail@aparcar.org>

Revert "build: set PATH for post-install scripts"

This reverts commit 21b5ac862eed7ac373361c914cf6098cab04f503.
The approach is flawed and cannot work in the ImageBuilder.
Using /etc/uci-defaults/ which is run on the target is the only
possible way.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>

ath79: Add support for Sophos AP15C

The Sophos AP15C uses the same hardware as the AP15, but has a reset button.

Based on:
commit 6f1efb289837 ("ath79: add support for Sophos AP100/AP55 family")
author    Andrew Powers-Holmes <andrew@omnom.net>
          Fri, 3 Sep 2021 15:53:57 +0200 (23:53 +1000)
committer Hauke Mehrtens <hauke@hauke-m.de>
          Sat, 16 Apr 2022 16:59:29 +0200 (16:59 +0200)

Unique to AP15C:
 - Reset button
 - External RJ45 serial console port

Flashing instructions:

This firmware can be flashed either via a compatible Sophos SG or XG
firewall appliance, which does not require disassembling the device, or via
the U-Boot console available on the internal UART header.

To flash via XG appliance:
 - Register on Sophos' website for a no-cost Home Use XG firewall license
 - Download and install the XG software on a compatible PC or virtual
   machine, complete initial appliance setup, and enable SSH console access
 - Connect the target AP device to the XG appliance's LAN interface
 - Approve the AP from the XG Web UI and wait until it shows as Active
   (this can take 3-5 minutes)
 - Connect to the XG appliance over SSH and access the Advanced Console
   (Menu option 5, then menu option 3)
 - Run `sudo awetool` and select the menu option to connect to an AP via
   SSH. When prompted to enable SSH on the target AP, select Yes.
 - Wait 2-3 minutes, then select the AP from the awetool menu again. This
   will connect you to a root shell on the target AP.
 - Copy the firmware to /tmp/openwrt.bin on the target AP via SCP/TFTP/etc
 - Run `mtd -r write /tmp/openwrt.bin astaro_image`
 - When complete, the access point will reboot to OpenWRT.

To flash via U-Boot serial console:
 - Configure a TFTP server on your PC, and set IP address 192.168.99.8 with
   netmask 255.255.255.0
 - Copy the firmware .bin to the TFTP server and rename to 'uImage_AP15C'
 - Open the target AP's enclosure and locate the 4-pin 3.3V UART header [4]
 - Connect the AP ethernet to your PC's ethernet port
 - Connect a terminal to the UART at 115200 8/N/1 as usual
 - Power on the AP and press a key to cancel autoboot when prompted
 - Run the following commands at the U-Boot console:
    - `tftpboot`
    - `cp.b $fileaddr 0x9f070000 $filesize`
    - `boot`
 - The access point will boot to OpenWRT.

Signed-off-by: David Lutz <kpanic@hirnduenger.de>

vrx518_tc: allow larger MTU to support RFC4638

vrx518_tc currently sets the interface maximum MTU to the ethernet default
of 1500 bytes by default via ether_setup() called from ptm_setup().

To support 1508 byte baby jumbo frames (RFC4638) for PPPoE connections
over VDSL links as already supported by the VR9 ltq_ptm driver ([1], [2])
set the interface maximum MTU to MAX_MTU.

MAX_MTU is defined in dcdp/inc/tc_common.h to 2002 bytes and this value is
used in ptm_change_mtu() and elsewhere as the maximum MTU, however this is
short circuited by checks against the interface maximum MTU.

[1]: https://forum.openwrt.org/t/fritzbox-7530-and-rfc4638-baby-jumbo-frames/181327
[2]: https://git.openwrt.org/?p=openwrt/openwrt.git;a=commit;h=8a2a20e71e2909f84dab47e51dfda9e292a6c1ae

Signed-off-by: Andrew MacIntyre <andymac@pcug.org.au>
Link: https://github.com/openwrt/openwrt/pull/16856
Signed-off-by: Robert Marko <robimarko@gmail.com>

rampis: mt7621: Edgerouter-X adjust IMAGE_SIZE to reflect new kernel size

IMAGE_SIZE was previously set to kernel1 + ubi size = 256768k, now
kernel1 is 6MB adjust this value to add 3072k to total image size.

Signed-off-by: Tim Lunn <tim@feathertop.org>
Link: https://github.com/openwrt/openwrt/pull/15194
Signed-off-by: Petr Štetiar <ynezz@true.cz>

ramips: mt7621: Edgerouter-X drop factory image

Edgerouter-X factory images have not built automatically since 19.x due
to images being over 3MB. While it was possible to build custom images
with very stripped down config, this is no longer possible with the size
increases of linux 6.1 and 6.6.

Drop code for generation of factory images, if some dev later wishes to
try custom images they can revert this commit.

Signed-off-by: Tim Lunn <tim@feathertop.org>
Link: https://github.com/openwrt/openwrt/pull/15194
Signed-off-by: Petr Štetiar <ynezz@true.cz>

rampis: mt7621: Re-enable Edgerouter-X image builds

With the new layout providing 6MB for kernel there will be no issues
with kernel size affecting build of images.

Re-enable image builds for Edgerouter-X and X-SFP.

Signed-off-by: Tim Lunn <tim@feathertop.org>
Tested-by: Mauri Sandberg <maukka@ext.kapsi.fi>
Link: https://github.com/openwrt/openwrt/pull/15194
Signed-off-by: Petr Štetiar <ynezz@true.cz>

rampis: mt7621: Edgerouter-X update compat version to 2.0

Use compat version to indicate that the new layout for larger kernels
is in place. This handled by the patch to ubnt.sh to always select
the kernel1 slot for flashing and as active kernel slot.

Signed-off-by: Tim Lunn <tim@feathertop.org>
Tested-by: Mauri Sandberg <maukka@ext.kapsi.fi>
Link: https://github.com/openwrt/openwrt/pull/15194
Signed-off-by: Petr Štetiar <ynezz@true.cz>

rampis: mt7621: Edgerouter-X set compat version message

Refer user to the wiki page for instructions on how to migrate to the
new kernel layout.

Signed-off-by: Tim Lunn <tim@feathertop.org>
Tested-by: Mauri Sandberg <maukka@ext.kapsi.fi>
Link: https://github.com/openwrt/openwrt/pull/15194
Signed-off-by: Petr Štetiar <ynezz@true.cz>

rampis: mt7621: resize Edgerouter-X kernel size to 6MB

With the updated partition layout set in dts, set the KERNEL_SIZE
parameter to 6MB allowing builds of Linux 6.1 and later to fit.

Signed-off-by: Tim Lunn <tim@feathertop.org>
Tested-by: Mauri Sandberg <maukka@ext.kapsi.fi>
Link: https://github.com/openwrt/openwrt/pull/15194
Signed-off-by: Petr Štetiar <ynezz@true.cz>

rampis: mt7621: Edgerouter-X disable toggling of kernel slots

Uboot selects which kernel slot to boot based on a flag in the factory
mtd partition. Patch ubnt.sh to ensure always flash to kernel1 slot and
update flag if required.

Signed-off-by: Tim Lunn <tim@feathertop.org>
Tested-by: Mauri Sandberg <maukka@ext.kapsi.fi>
Link: https://github.com/openwrt/openwrt/pull/15194
Signed-off-by: Petr Štetiar <ynezz@true.cz>

rampis: mt7621: Convert Edgerouter-X dts to single kernel config

The OEM layout for Edgerouter X provides for two 3MB kernel slots.
As of linux 6.1 the kernel images no longer fit and as such
Edgerouter X builds have been disabled in Main.

Revise the layout to make kernel1 slot 6MB and drop kernel2 slot.

This patch applies the required changes to the dts file.

Signed-off-by: Tim Lunn <tim@feathertop.org>
Tested-by: Mauri Sandberg <maukka@ext.kapsi.fi>
Link: https://github.com/openwrt/openwrt/pull/15194
Signed-off-by: Petr Štetiar <ynezz@true.cz>

netifd: update to Git HEAD (2024-11-08)

e93f6c3fc729 main: fix format string related warnings for log/debug printf functions
6ab44a2413f9 fix signed/unsigned comparison warnings
4fe997b61d7d system-dummy: add system_if_apply_settings_after_up
992d33cb42a6 ubus: add notifications on wireless device state changes
34eb11eb6f5c device/interface: add "tags" attribute from config to status dump

Signed-off-by: Felix Fietkau <nbd@nbd.name>

wifi-scripts: add free-form "tags" array option for wifi-iface

This allows annotating wifi interfaces in the config in a way that can be
queried through wifi status. One example use case is to mark wifi interfaces
for use with specific services without having to explicitly reference the
(often unnamed) sections from elsewhere.

Signed-off-by: Felix Fietkau <nbd@nbd.name>

kernel: modules: package driver for generic MIPI DBI TFT panels

Package kmod-drm-panel-mipi-dbi as well as modules it depends on in
order to support a wide range of MIPI DBI complaint SPI-connected
TFT panels.

See https://github.com/notro/panel-mipi-dbi/ for more information on
how to use specific panels.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>

ath79: linksys_ea4500-v3: take over the space for user data

The last "syscfg" partition of the OEM firmware turns out to be a
UBIFS used to store user data, just as the "rootfs_data" of OpenWrt,
so it should be reasonable to absorb it into the "ubi" partition.

Factory installations via either OEM firmware or tftp, or by forcibly
flashing factory image to mtd5 (firmware) partition with mtd tool are
confirmed working, but the UBI remaining inside "syscfg" partition
could break upgrade. Fortunately, installing kmod-mtd-rw and erasing
"syscfg" partition before upgrade is confirmed working, in which case,
"ubi" will automatically expand to the blank space once occupied by
the former mtd8 (syscfg), with the total block number increased, but
the UBIFS for rootfs_data will not automatically claim the newly
available space (since it is created when mtd8 still exists, and
sysupgrade does not set "autoresize" flag to rootfs_data). These space
will be claimed during the next upgrade, when rootfs_data is removed
and created again.

Fixes: 50f727b7737d1 ("ath79: add support for Linksys EA4500 v3")
Signed-off-by: Edward Chow <equu@openmail.cc>
Link: https://github.com/openwrt/openwrt/pull/14791
Signed-off-by: Robert Marko <robimarko@gmail.com>

kernel: r8125: update to v9.014.01

Changelog: https://github.com/openwrt/rtl8125/compare/9.013.02...9.014.01

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

procd: update to Git HEAD (2024-11-06)

109fa41b2321 system: fix description value from os-release

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>

fritz-tools: fix mixed indentation

Signed-off-by: David Bauer <mail@david-bauer.net>

kernel: r8168: refresh patch

Patches weren't refreshed when r8168 was updated to v8.054.00 in a85e18b53f0f.

Fixes: a85e18b53f0f ("kernel: r8168: update to v8.054.00")
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

build: set PATH for post-install scripts

post-install scripts may need to call executables installed to
STAGING_DIR_HOSTPKG which is not part of the PATH set to TARGET_PATH in
rules.mk.
Set PATH for post-install scripts to TARGET_PATH_PKG.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Link: https://github.com/openwrt/openwrt/pull/16865
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>

mediatek: set 7 MACs per wifi band in board.json on OpenWrt One

Mass production units will get 16 assigned MAC addresses. This allows each phy
to spawn up to 7 VAPs which will each have unique MAC without needing the
private bit.

Signed-off-by: John Crispin <john@phrozen.org>

mediatek: set mac increment to 7 for the 5G radio on OpenWrt One

Mass production units will get 16 assigned MAC addresses. This allows each phy
to spawn up to 7 VAPs which will each have unique MAC without needing the
private bit.

Signed-off-by: John Crispin <john@phrozen.org>

packages: drop DEVICE specific dependency for omnia

Both packages `ombnia-mcu-firmware` and `omnia-mcutool` would depend on
a specific device. The buildbots however build all devices and therefore
the package isn't build at all, due to unmet dependencies.

While this didn't cause issues with OPKG, APK fails actively due to the
missing packages. Drop the specific dependency, however wants to install
unrelated firmware on any device can do that anyway.

Signed-off-by: Paul Spooren <mail@aparcar.org>

kernel: r8126: update to v10.014.01

Changelog: https://github.com/openwrt/rtl8126/compare/10.013.00...10.014.01

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

mac80211: compatibility with kernel 6.6.59

Fix genlmsg_multicast_allns() build error on kernel 6.6.59.
Based on kernel patch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit?h=v6.6.59&id=e0f83d268974dab0361d11904dfc9acec53f96a6

Signed-off-by: Mieczyslaw Nalewaj <namiltd@yahoo.com>
Link: https://github.com/openwrt/openwrt/pull/16835
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

kernel: bump 6.6 to 6.6.59

Changelog: https://cdn.kernel.org/pub/linux/kernel/v6.x/ChangeLog-6.6.59

Removed upstreamed:
 - generic/backport-6.6/777-netfilter-xtables-fix-typo-causing-some-targets-to-not-load-on-IPv6.patch[1]
 - generic/backport-6.6/780-24-v6.12-r8169-avoid-unsolicited-interrupts.patch[2]

All other patches automatically rebased.

1. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.6.59&id=433742ba96baf30c21e654ce3e698ad87100593b
2. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.6.59&id=7d6d46b429804b1a182106e27e2f8c0e84689e1a

Signed-off-by: Mieczyslaw Nalewaj <namiltd@yahoo.com>
Link: https://github.com/openwrt/openwrt/pull/16835
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

nu801: Mark as nonshared to build in step 1

Mark the package as nonshared to build it in the target specific build
step 1 of the build bots instead of the architecture generic build step 2.

In the build step 2 it may be left out if we build it using a different
target.

Fixes: #16857
Link: https://github.com/openwrt/openwrt/pull/16859
Signed-off-by: Robert Marko <robimarko@gmail.com>

ramips: Fix EX400 imagebuilder

A factory image for DNA EX400 depends on an initramfs image and they
were explicitly removed from the imagebuilder recently. Now the factory
image creation fails miserably and it also affects custom image creation
with the firmware selector.

Add the initramfs kernel to the staging so that it's shipped with the
imagebuilder. Also remove a image build target added solely for DNA EX400.

Tested by creating a factory and syspupgrade images locally with
the imagebuilder and verified their functionality.

Related work
  c85348d9abf4 ("imagebuilder: remove initramfs image files")

Fixes: fea2264d9fdd ("ramips: mt7621: Add DNA Valokuitu Plus EX400")
Signed-off-by: Mauri Sandberg <maukka@ext.kapsi.fi>
---
v4: use append-image-stage, remove Build/kernel-initramfs-bin
v3: adjust commit subject
v2: remove fix for inconsistent line ending elsewhere in the file
Link: https://github.com/openwrt/openwrt/pull/16659
Signed-off-by: Robert Marko <robimarko@gmail.com>

uboot-mediatek: fix boot media for mt7981 RFB NOR profile

Signed-off-by: Enrico Mioso <mrkiko.rs@gmail.com>
CC: Daniel Golle <daniel@makrotopia.org>

gpio-button-hotplug: use flexible array member

zero length arrays are deprecated.

Fixes coccinelle warning:

WARNING use flexible-array member instead

Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16848
Signed-off-by: Robert Marko <robimarko@gmail.com>

openwrt-keyring: fix missing 24.10 usign key by installing it

@KA2107 reported that opkg is not able to verify the artifact signatures
produced by buildbot using the usign 24.10 release keys. So lets fix it
by actually adding the 24.10 usign key with d310c6f2833e97f7 fingerprint
into the openwrt-keyring package.

Fixes: #16850
Reported-by: @KA2107
Fixes: a535cfc09e1c ("openwrt-keyring: add OpenWrt 24.10 release build usign key")
Signed-off-by: Petr Štetiar <ynezz@true.cz>

mtd: Add backported Winbond fix

Upstream stable is slow at picking this up and several systems
are regressing. Add the patch locally in OpenWrt.

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

mediatek: YunCore AX835: fix voltage regulator

Specifying GPIO_ACTIVE_HIGH on the GPIO for the voltage regulator doesn't
suffice. The regulator itself requires enable-active-high to be set.

Fixes: #16292
Signed-off-by: Leon M. Busch-George <leon@georgemail.eu>
Link: https://github.com/openwrt/openwrt/pull/16839
Signed-off-by: Robert Marko <robimarko@gmail.com>

ath79: gpio-rb91x-key: use gpiochip_get_data

No need for a custom function that does the same thing.

Oversight from devm conversion.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16748
Signed-off-by: Robert Marko <robimarko@gmail.com>

ath79: gpio-latch-miktorik: use gpiochip_get_data

Oversight from devm conversion. No need for a custom static inline
function.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16747
Signed-off-by: Robert Marko <robimarko@gmail.com>

ramips: mt7621: convert usb power to regulators

These things are regulators. Should silence dmesg messages about using
dummy regulators.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16804
Signed-off-by: Robert Marko <robimarko@gmail.com>

kernel: can: package driver for Geschwister Schneider interface

Package gs_usb CAN driver, which is for Geschwister Schneider and
bytewerk.org candleLight compatible USB/CAN interfaces.

Signed-off-by: Chuanhong Guo <gch981213@gmail.com>

mediatek: refresh patches

The mediatek target requires refreshing after recent additions.

Fixes: cfe8e6e75f ("mediatek: add support for Realtek RTL8261n 10G PHYs")
Fixes: ddfae94a14 ("mediatek: add support for swapping the polarity on usxgmii interfaces")
Signed-off-by: Daniel Golle <daniel@makrotopia.org>

mediatek: filogic: BPi-R4: wire up pinctrl for pwm-fan

Assign pwm function of PWM0 pin to the pwm-fan.
This is mostly just cosmetics as it basically reflects the default
setting of that pin.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>

mediatek: pinctrl-mt7988: add missing PWM pingroups

Add additionals possible pinctrl group for pwm2~7 on pins
pin  4 (GPIO_A)        pwm7
pin 58 (JTAG_JTDI)     pwm2
pin 59 (JTAG_JTDO)     pwm3
pin 60 (JTAG_JTMS)     pwm4
pin 61 (JTAG_JTCLK)    pwm5
pin 62 (JTAG_JTRST_N)  pwm6

They can be useful e.g. on the BPi-R4 as in that way pwm2~6 can be exposed
on the 26-pin header (pwm6 always, pwm2~5 instead of the full UART).

Signed-off-by: Daniel Golle <daniel@makrotopia.org>

uboot-mediatek: import patches improving menu navigation

Using the arrow keys to navigate the U-Boot menu often leads to being
dropped into the U-Boot shell unexpectedly.
This can be prevented in most cases by improving the logic to detect the
arrow key ESC sequence and only reprinting the menu if actually needed.
Also enable CONFIG_SERIAL_RX_BUFFER for all boards as it helps preventing
the remaining cases.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>

ipq40xx: use devm for mutex_init

Avoids having to call mutex_destroy in remove.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16755
Signed-off-by: Robert Marko <robimarko@gmail.com>

bcm27xx-gpu-fw: fix package

Install definition was accidentally removed in 7afdbc0955a5 and IB fails.

Fixes: 7afdbc0955a5 ("bcm27xx-gpu-fw: update to latest version")
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

kernel: KERNEL_NET_L3_MASTER_DEV default to y if !SMALL_FLASH

Follow-up for 45d541bb409355f090b971d96ebebd8610ef84a7

This change allows features such as kmod-vrf

```
KERNEL_NET_L3_MASTER_DEV=n

x86_64 generic
bzImage 6,927,360 bytes
aarch64 coretex-a53
kernel  4,268,836 bytes

KERNEL_NET_L3_MASTER_DEV=y

x86_64 generic
bzImage 6,931,456 bytes
aarch64 coretex-a53
kernel  4,273,042 bytes

Delta:
x86_64 generic
+4096 bytes
aarch64 coretex-a53
+4206 bytes

x86_64 generic vrf.ko - 258,792 bytes
aarch64 coretex-a53 vrf.ko - 263,632 bytes
```

See:
https://forum.openwrt.org/t/vrf-support-testing-out-evpn-at-home/181108
https://forum.openwrt.org/t/please-enable-net-l3-master-dev-in-kernel-build-by-default/201825

Signed-off-by: Paul Donald <newtwen+github@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16819
Signed-off-by: Robert Marko <robimarko@gmail.com>

mediatek: filogic: add Arcadyan Mozart platform

* Mediatek MT7988A (4x Cortex-A73, up to 1.8 GHz clock speed)
* 8 GiB eMMC
* 4 GiB DDR4 RAM
* 2x 10000M + 1x 1000M  ports
* MT7996 Tri-band (2.4G, 5G, 6G) 4T4R 802.11be Wi-Fi
* 2 buttons (Reset, WPS)
* 2x LED for each GMAC
* USB-C PD power input

Signed-off-by: John Crispin <john@phrozen.org>

mediatek: add support for swapping the polarity on usxgmii interfaces

This patch comes from the MTK SDK.

Signed-off-by: John Crispin <john@phrozen.org>

mediatek: add support for Realtek RTL8261n 10G PHYs

There is no upstream driver yet. Merge the RTL SDK driver for now.

Signed-off-by: John Crispin <john@phrozen.org>

uboot-mediatek: add support for arcadyan,mozart

Signed-off-by: John Crispin <john@phrozen.org>

openwrt-keyring: add OpenWrt 24.10 release build usign key

 fbae29d730f8 usign: add OpenWrt 24.10 release build public key

Signed-off-by: Petr Štetiar <ynezz@true.cz>

mediatek: mt7988: fix peripheral SPI busses

The clocks for SPI busses were named wrongly which resulted in the
spi-mt65xx driver not requesting them. This has apparently been
worked around by marking the clocks required for SPI0 which is used
for SPI-NOR and SPI-NAND flash chips as critical.
Fix the device tree for all 3 generic SPI host controllers and no
longer mark clocks as critical.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>

ath79: Really fix 5GHz on QCA9886 variant of ZTE MF286

Fix in commit 25eead21c5ab ("ath79: fix 5GHz on QCA9886 variant of ZTE MF286")
was incomplete. A user of such variant popped up, and in the boot log
after installation, we discovered that QCA9886 expects different
pre-calibration data size, than the older QCA9880 variant:

ath10k_pci 0000:00:00.0: qca9888 hw2.0 target 0x01000000 chip_id 0x00000000 sub 0000:0000
ath10k_pci 0000:00:00.0: kconfig debug 0 debugfs 1 tracing 0 dfs 1 testmode 0
ath10k_pci 0000:00:00.0: firmware ver 10.4b-ct-9888-fW-13-5ae337bb1 api 5 features mfp,peer-flow-ctrl,txstatus-noack,wmi-10.x-CT,ratemask-CT,regdump-CT,txrate-CT,flush-all-CT,pingpong-CT,ch-regs-CT,nop-CT,set-special-CT,tx-rc-CT,cust-stats-CT,txrate2-CT,beacon-cb-CT,wmi-block-ack-CT,wmi-bcn-rc-CT crc32 59e741e7
ath10k_pci 0000:00:00.0: invalid calibration data length in nvmem-cell 'pre-calibration': 2116 != 12064
ath10k_pci 0000:00:00.0: Loading BDF type 0
ath10k_pci 0000:00:00.0: failed to fetch board data for bus=pci,vendor=168c,device=0056,subsystem-vendor=0000,subsystem-device=0000 from ath10k/QCA9888/hw2.0/board-2.bin
ath10k_pci 0000:00:00.0: failed to fetch board-2.bin or board.bin from ath10k/QCA9888/hw2.0
ath10k_pci 0000:00:00.0: failed to fetch board file: -12
ath10k_pci 0000:00:00.0: could not probe fw (-12)

Explicitly define a pre-calibration nvmem-cell for this variant, and use
it instead of the calibration one, which is shorter.

Fixes: 25eead21c5ab ("ath79: fix 5GHz on QCA9886 variant of ZTE MF286")
Signed-off-by: Lech Perczak <lech.perczak@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16809
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

bcm27xx: update to latest RPi patches

The patches were generated from the RPi repo with the following command:
git format-patch v6.6.58..rpi-6.6.y

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

bcm27xx-gpu-fw: update to latest version

Use release tar instead of downloading separate files.

Full changelog: https://github.com/raspberrypi/firmware/compare/1.20240424...1.20241008

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

ramips: m4r v4: fix typo

It should be debounce-interval, as with the others.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16802
Signed-off-by: Robert Marko <robimarko@gmail.com>

CI: labeler: add bcm27xx-utils

bcm27xx-utils can only be built for bcm27xx target.

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

bcm27xx-utils: update to latest version

Fixes build with GCC 14.

Full changelog: https://github.com/raspberrypi/utils/compare/451b9881b72cb994c102724b5a7d9b93f97dc315...6a2a6becebbc38fde34a94386457ac8210f9119b

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

ipq806x: Revert "ipq806x: swap lan leds for Meraki MR52"

This reverts commit ec8f647d168fa8f3b1eedd9b5fe665f793f3a659, as with the
current kernel version, the change actually causes the same bug it once
may have fixed -- that is, the leds are now again reversed.

I suspect this was due to a switch to a newer kernel version between when
the patch was submitted and now reversing the order of the interfaces, so
that eth0 / the LAN interface is also the interface used for PoE, and eth1
/ the WAN interface is the non-PoE interface.

Signed-off-by: Rafal Boni <rafal.boni@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16779
Signed-off-by: Robert Marko <robimarko@gmail.com>

ramips: ignore errors from trx utility

Ignore errors from trx utility to handle case with image too big.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>

image: ignore errors from more commands

Ignore errors in more image commands to handle case where the image is
too big and check-image validation fails.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>

imagebuilder: fix copying of missing kernel/libc

Those packages were not copied due to OPKG using an underscore while APK
uses dashes. Remove that char to copy kernel/libc for either APK/OPKG.

Signed-off-by: Paul Spooren <mail@aparcar.org>

image: ignore errors from mktplinkfw command

Rework tplink-v2-header and tplink-v2-image Build define to ignore error
if mktplinkfw2 errors out.

This is to handle situation when the image is too big and can't be
generated or prev check-image calls deleted the source file as it's too
big.

This aligns to the pattern used by tplink-v1-image.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>

build: correctly ignore Initramfs image on error

Correctly ignore Initramfs image copy on error. This follows the pattern with
sysupgrade image where an image might fail as it's too big or the
generation command fails and there is nothing to copy to the bin
directory.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>

mxs: drop removal of firewall4 package

Drop removal of firewall4 package for I2SE Duckbill device.

With OPKG the firewall4 package was installed anyway as it's a
dependency of luci-app-firewall and was silently installed again later
in such condition. Drop it to fix support for APK.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>

mvebu: cortexa9: drop removal of firewall4 package

Drop removal of firewall4 package for Synology DS213j device.

With OPKG the firewall4 package was installed anyway as it's a
dependency of luci-app-firewall and was silently installed again later
in such condition. Drop it to fix support for APK.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>

build: Honor DEFAULT_VARIANT for APK packages

Previously APK would complain as it wasn't sure which package to
install by default when multiple packages would provide the same name.

Now, give the package a higher provider priority to make APK
automatically select the "default" package.

Signed-off-by: Paul Spooren <mail@aparcar.org>

build: don't include kernel/libc in package index

The same that is done in `ipkg-make-index.sh` should happen with APK.
If the pseudo packages, only added to add dependency constraints, are
added to the index, APK happily "upgrades" them and installs updated
kmods, too. However, the Kernel itself is never installed via a regular
package.

Fixes: https://github.com/openwrt/openwrt/issues/16808
Signed-off-by: Paul Spooren <mail@aparcar.org>

kernel: mtdsplit_minor: accept bootimage filename

RouterBOOT v7 on NOR devices no longer accepts the YAFFS kernel ELF
method of booting. It will accept an NPK image named bootimage.
Adjust mtdsplit_minor to accept this second possible boot file name.
Use the conservative value of 127 for YAFFS max name length (used when
YAFFS compiled with unicode support) vs 255.

Signed-off-by: John Thomson <git@johnthomson.fastmail.com.au>
Acked-by: Thibaut VARENE <hacks@slashdirt.org>
Link: https://github.com/openwrt/openwrt/pull/16780
Signed-off-by: Robert Marko <robimarko@gmail.com>

kernel: mtdsplit_minor: return 0 if not fatal

Introduced with Linux 6.7, in commit:
5c2f7727d437 ("mtd: mtdpart: check for subpartitions parsing result"),
when a parser returns an error, this will be passed up, and
consequently, all parent mtd partitions get torn down.

Adjust the MiNOR mtdsplit driver to only return an error if there is a
critical problem in reading from the mtd device or allocating memory.
Otherwise return 0 to indicate that no partitions were found.
Also add logging to indicate what went wrong.

This mtdsplit parser makes a very limited check of the first YAFFS
header. For example, this will not match expectations when initially booting
an initramfs image with OEM on MTD.

Signed-off-by: John Thomson <git@johnthomson.fastmail.com.au>
Acked-by: Thibaut VARENE <hacks@slashdirt.org>
Link: https://github.com/openwrt/openwrt/pull/16780
Signed-off-by: Robert Marko <robimarko@gmail.com>

ramips: linkit: replace driver with gpio-hog

It does the same thing minus a few dmesg prints.

Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16788
Signed-off-by: Robert Marko <robimarko@gmail.com>

ramips: Fix WiFi on Sercomm NA502 and NA502s

The WiFi chips are actually on PCIe1 and PCIe2, PCIe0 is empty. Fix the
assignment so that WiFi works properly again.

Signed-off-by: Andreas Böhler <dev@aboehler.at>
Link: https://github.com/openwrt/openwrt/pull/16807
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

bcm27xx-utils: fix PKG_MIRROR_HASH

The mirror hash has changed after 8009342.

Fixes: 8009342f43c5 ("bcm27xx-utils: fix version for APK")
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

generic: backport upstream r8169 irq patch

This commit backports an additional patch from linux v6.12 release.

Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>

imagebuilder: always add local packages.adb

It doesn't matter if `IB_STANDALONE` is enabled or not, local packages
must always be considered.

Signed-off-by: Paul Spooren <mail@aparcar.org>

package: fix key creation for SDK

Prior to this commit keys would only be generated if `make` is called
alone, but not for something like `make package/busybox/compile`.

The exact reasons are in the depth of make magic, so this is sheer luck!

Signed-off-by: Paul Spooren <mail@aparcar.org>

include/package-pack: drop version from provides

Adding a version to the provides causes it to conflict with other
packages that provides the same package, further details are available
here: https://gitlab.alpinelinux.org/alpine/apk-tools/-/blob/master/doc/apk-package.5.scd#L199

This was intitally done, if I remember correctly, to support depending
on the specific kernel modules. Due to patches to APK, versions
containing hashes work now, too, so this is no longer required.

Only add the version to packages that define an ABI version since other
packages depend against the package name plus ABI version.

While at it, format the now rather complex call.

Fixes: https://github.com/openwrt/openwrt/issues/16795
Signed-off-by: Paul Spooren <mail@aparcar.org>

armsr: don't limit kmods to armsr armv8 subtarget

Drop limitation on depending on only armv8 for armsr target as those
module should support both 32 and 64 bits systems.

Only thunderx-net actually require 64 bit systems.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>

dsl-vrx200-firmware-xdsl: fix PKG_VERSION for APK

The version was a mix of strings, hex numbers and semantic numbers.

Switch the PKG_VERSION to something digestible by APK and introduce
PKG_SOURCE_VERSION to handle the actual filename.

While at it, drop the redundant PKG_B_NAME which was the same as
PKG_NAME anyway.

Signed-off-by: Paul Spooren <mail@aparcar.org>