generic: 6.1, 6.6: mt7530: import pending PHY muxing detection patch

net: dsa: mt7530: detect PHY muxing when PHY is defined on switch MDIO bus

Signed-off-by: Arınç ÜNAL <arinc.unal@arinc9.com>

ramips: mt7621-dts: remove incorrect ethphy4 node for WAVLINK WL-WN573HX1

The ethernet-phy@4 node doesn't exist for WAVLINK WL-WN573HX1. Remove it
and the duplicate gmac0 node.

Signed-off-by: Arınç ÜNAL <arinc.unal@arinc9.com>

ramips: mt7621-dts: do not modify ethernet node for MeiG SLT866

Currently, the pinctrl-0 property on the ethernet node is modified to
exclude the rgmii1 and rgmii2 pin groups to be claimed with rgmii1 and
rgmii2 functions, respectively. Remove the modification of this property as
we need these pin groups to be claimed with the said functions for this
device.

Signed-off-by: Arınç ÜNAL <arinc.unal@arinc9.com>

zynq: 5.15: remove files

Remove files related to the old 5.15 kernel version as they are no
longer required.

Signed-off-by: Nick Hainke <vincent@systemli.org>

zynq: set default kernel version to 6.1

Set default kernel version to 6.1, facilitating simpler testing across
a broader user base.

Signed-off-by: Nick Hainke <vincent@systemli.org>

kernel: bump 6.1 to 6.1.89

Changelogs:
https://cdn.kernel.org/pub/linux/kernel/v6.x/ChangeLog-6.1.67
https://cdn.kernel.org/pub/linux/kernel/v6.x/ChangeLog-6.1.68
https://cdn.kernel.org/pub/linux/kernel/v6.x/ChangeLog-6.1.69

Upstreamed patches:
target/linux/generic/backport-6.1/740-v6.9-01-netfilter-flowtable-validate-pppoe-header.patch [1]
target/linux/generic/backport-6.1/740-v6.9-02-netfilter-flowtable-incorrect-pppoe-tuple.patch [2]
target/linux/generic/backport-6.1/790-48-STABLE-net-dsa-mt7530-trap-link-local-frames-regardless-of-.patch [3]
target/linux/generic/backport-6.1/790-50-v6.10-net-dsa-mt7530-fix-mirroring-frames-received-on-loca.patch [4]
target/linux/generic/backport-6.1/790-16-v6.4-net-dsa-mt7530-set-all-CPU-ports-in-MT7531_CPU_PMAP.patch [5]
target/linux/generic/backport-6.1/790-46-v6.9-net-dsa-mt7530-fix-improper-frames-on-all-25MHz-and-.patch [6]
target/linux/generic/backport-6.1/790-47-v6.10-net-dsa-mt7530-fix-enabling-EEE-on-MT7531-switch-on-.patch [7]
target/linux/mediatek/patches-6.1/220-v6.3-clk-mediatek-clk-gate-Propagate-struct-device-with-m.patch [8]
target/linux/mediatek/patches-6.1/222-v6.3-clk-mediatek-clk-mtk-Propagate-struct-device-for-com.patch [9]
target/linux/mediatek/patches-6.1/223-v6.3-clk-mediatek-clk-mux-Propagate-struct-device-for-mtk.patch [10]
target/linux/mediatek/patches-6.1/226-v6.3-clk-mediatek-clk-mtk-Extend-mtk_clk_simple_probe.patch [11]

Symbol changes:
MITIGATION_SPECTRE_BHI (new) [12]
SPECTRE_BHI_{ON,OFF} (deprecated) [12]

References:
[1] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.1.89&id=8bf7c76a2a207ca2b4cfda0a279192adf27678d7
[2] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.1.89&id=f1c3c61701a0b12f4906152c1626a5de580ea3d2
[3] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.1.89&id=19643bf8c9b5bb5eea5163bf2f6a3eee6fb5b99b
[4] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.1.89&id=e86c9db58eba290e858e2bb80efcde9e3973a5ef
[5] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.1.89&id=013c787d231188a6408e2991150d3c9bf9a2aa0b
[6] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.1.89&id=41a004ffba9b1fd8a5a7128ebd0dfa3ed39c3316
[7] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.1.89&id=7d51db455ca03e5270cc585a75a674abd063fa6c
[8] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.1.89&id=082b831488a41257b7ac7ffa1d80a0b60d98394d
[9] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.1.89&id=6f5f72a684a2823f21efbfd20c7e4b528c44a781
[10] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.1.89&id=a4fe8813a7868ba5867e42e60de7a2b8baac30ff
[11] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.1.89&id=c1d87d56af063c87961511ee25f6b07a5676d27d
[12] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.1.89&id=d844df110084ef8bd950a52194865f3f63b561ca
Signed-off-by: Shiji Yang <yangshiji66@qq.com>

generic: MIPS64: fix detect_memory_region() compilation error

1. Enable this feature only for 32-bit CPUs as MIPS64 can not
   access the full range unmapped uncached memory.

2. Backport this fix to the 6.1 old LTS kernel.

Signed-off-by: Shiji Yang <yangshiji66@qq.com>

qualcommax: ipq807x: uboot-envtools: yuncore ax880

There was no config in the uboot-envtools package, so there is no
generated /etc/fw_env.config for the fw_printenv and fw_setenv utils.
Since uboot-envtools 2024.01, there is a way to make these utils work
without /etc/fw_env.config if the DT has an env partition with the prop.:

compatible = "u-boot,env";

So, this commit adds the prop. above to the appsblenv:0 partition
in the yuncore ax880 DTS file.

Signed-off-by: Isaev Ruslan <legale.legale@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15305
Signed-off-by: Robert Marko <robimarko@gmail.com>

tools/elfutils: remove dirname

In the 0.191 update dirname was used instead of <libgen.h> to fix the
poisoned basename error:
/usr/include/libgen.h:35:9: error: attempt to use poisoned "basename"
       35 | #define basename        __xpg_basename

However, doing this has lead to libelf.a pulling in xmalloc, xstrdup and
friends and statically linking them thus leading to a symbol name conflict
with FRR host build and anything else that links against libelf and uses
xmalloc and friends.

Well, it turns out that upstream has added a helper[1] for basename so it
can compile with musl 1.2.5 which dropped the basename declaration, but it
also means that we must NOT include <libgen.h> and that poisoned error is
intentional and added to prevent duplicate basename definitions.
This also means that for macOS we dont need to do any additional header
inclusions as the new helper takes care of basename.

So, to fix the symbol conflict we can simply drop the <dirname.h> inclusion
and build from elfutils.

Tested on Fedora 40 as well as macOS 14.4.1.

[1] https://sourceware.org/git/?p=elfutils.git;a=commit;h=a2194f6b305bf0d0b9dd49dccd0a5c21994c8eea

Fixes: #24030
Fixes: b6f025b42429 ("tools/elfutils: update to 1.91")
Link: https://github.com/openwrt/openwrt/pull/15337
Signed-off-by: Robert Marko <robimarko@gmail.com>

ipq40xx: fix I2C pin config on Aruba AP-303H

Turn on SoC pull-ups on I2C pins, since there are no discrete pull-up
resistors on the bus.

Increase clock to 400 kHz. Both chips on the bus support 400 kHz. I
tested the ISL28022 at 4,000 reads/sec and didn't see any garbled output
or bus hangs, even with SoC drive strength reduced to 2 for the test.

Signed-off-by: Ryan Salsbury <ryanrs@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15334
Signed-off-by: Robert Marko <robimarko@gmail.com>

sifiveu: drop 6.1 support

Now that 6.6 is the default, remove the 6.1 config and patches.

Signed-off-by: Zoltan HERPAI <wigyori@uid0.hu>

sifiveu: switch default to 6.6

Switch the default kernel to 6.6.

Signed-off-by: Zoltan HERPAI <wigyori@uid0.hu>

mediatek: switch to fitblk for JDCloud RE-CP-03

Use the new fitblk driver.

Tested-By: Yangyu Chen <cyy@cyyself.name>
Signed-off-by: Tianling Shen <cnsztl@immortalwrt.org>

procd: make mDNS TXT record parsing more solid

mDNS broadcast can't accept empty TXT record and would fail
registration.

Current procd_add_mdns_service checks only if the first passed arg is
empty but don't make any verification on the other args permittins
insertion of empty values in TXT record.

Example:

procd_add_mdns "blah" \
"tcp" "50" \
"1" \
"" \
"3"

Produce:

{ "blah_50": { "service": "_blah._tcp.local", "port": 50, "txt": [ "1", "", "3" ] } }

The middle empty TXT record should never be included as it's empty.

This can happen with scripts that make fragile parsing and include
variables even if they are empty.

Prevent this and make the TXT record more solid by checking every
provided TXT record and include only the non-empty ones.

The fixed JSON is the following:

{ "blah_50": { "service": "_blah._tcp.local", "port": 50, "txt": [ "1", "3" ] } }

Fixes: b0d9dcf84dd0 ("procd: update to latest git HEAD")
Reported-by: Paul Donald <newtwen@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15331
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>

toolchain: glibc: update to glibc 2.38

This updates glibc to version 2.38.

Add --enable-crypt since the crypt function got disabled in 2.38, but we
still need it.
Also add the newly introduced --enable-fortify-source flag and hook it up
to the build system.

Switch to .zst archive while at it.

Signed-off-by: Kazuki H <kazukih0205@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/14259
[ switch to .zst and fix mirror HASH ]
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>

qualcommax: ipq60xx: mute warning on bootcount enable compile

ipq60xx bootcount script include /lib/functions that produce warning
when the script is enabled on image compilation. This script is already
included by /etc/rc.common hence it's not needed.

While at it also fix the format of the switch case.

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

arm-trusted-firmware-mvebu: refresh cryptopp hash

Well, it seems that cryptopp hash was never refreshed since calling
make package/boot/arm-trusted-firmware-mvebu/check FIXUP=1 V=s does not
actually refresh the download calls hashes so refresh it manually.

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

arm-trusted-firmware-mvebu: use SOURCE_VERSION instead of VERSION

Since ("download: don't overwrite VERSION variable") trying to download the
required sources for mvebu ATF will fail with:
Makefile:247: *** Download/mox-boot-builder is missing the SOURCE_VERSION field..  Stop.

This also broke the buildbot mvebu/cortex-a53 builds.

So, fix it by switching to SOURCE_VERSION instead.

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

mediatek: fix remaining invalid MT7531 switch reg IDs

See commit 8b66f1a06d05afbc1450978f6e20516c8308da25:
"mediatek: correct address of MT753x switch IC"

Fixes: ac8bfe316b81 ("generic: 6.1, 6.6: mt7530: import accepted patches")
Signed-off-by: Felix Fietkau <nbd@nbd.name>

apk: add package

APK (Alpine Package Keeper) is the package manager of Alpine Linux and
has multiple advantages over OPKG. While Alpine uses APK version 2, this
commit adds version 3 with a heavily optimised database structure and
additional feature making it suitable for OpenWrt.

This commit will be followed by many more to add APK build capabilities
to the OpenWrt build system, firstly enabling side by side builds of APK
and OPKG packages, later replacing OPKG entirely.

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

generic: 6.6: add kernel missing symbol CONFIG_SPI_BCMBCA_HSSPI

Add missing symbol CONFIG_SPI_BCMBCA_HSSPI causing bcm4908 compilation errors.

Signed-off-by: Mieczyslaw Nalewaj <namiltd@yahoo.com>
Link: https://github.com/openwrt/openwrt/pull/15320
Signed-off-by: Robert Marko <robimarko@gmail.com>

download: don't overwrite VERSION variable

In package-defaults.mk the VERSION variable contains the final package
version, which is a combination of `PKG_VERSION`, `PKG_RELEASE` and if
used, parameters of `PKG_SOURCE_VERSION`. While this works fine for
building, within the package Makefile the content VERSION varies:

When building a package from a release tarball, the content of VERSION
contains the actual package version, however when building from source
(i.e. Git), the VERSION is overwritten by `PKG_SOURCE_VERSION`.

To fix the overwrite, this commit switches from `VERSION` in download.mk
to `SOURCE_VERSION` which isn't used anywhere else yet.

As a result, Makefiles may pass the package version to be included
inside the binary.

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

mpc85xx: convert WS-AP3710i to simpleImage wrapper

Convert the Enterasys WS-AP3710i access point to use the simpleImage
wrapper.

This is necessary, as the bootlaoder does not align the DTB correctly
(and does not support altering the FDT loadaddress). Booting images with
kernels 5.15 and later can break depending on the alignment on the DTB
within the FIT image.

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

mbedtls: Remove Kconfig options removed from mbedtls 3.6.0

These options are not available in mbedtls 3.6.0 and selecting them
causes an error.

MBEDTLS_CERTS_C was removed in:
https://github.com/Mbed-TLS/mbedtls/commit/1aec64642cd0b9490c96fdd1fa9816741d93fce4

MBEDTLS_XTEA_C was removed in:
https://github.com/Mbed-TLS/mbedtls/commit/10e8cf5fef7e062927bfbec69ad85fbf491b4299

MBEDTLS_SSL_TRUNCATED_HMAC was removed in:
https://github.com/Mbed-TLS/mbedtls/commit/4a7010d1aaae937dfa328ef253d70d04b3b260ff

Fixes: 0e0664264386 ("mbedtls: Update to version 3.6.0")
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

libaudit: update to 3.1.4, join with daemon and utils, rename

Changes:
- new URL for sources (old address is dead)
- daemon and utils from packages feed are merged in here
  - only build once
  - no need to update at the same time in both places
- update to v3.1.4
  - removed unneeded patches
  - added audisp-syslog
  - removed audispd (no longer exists)
- rename and move to package/utils/audit
  - update new path in one dependent package

Signed-off-by: Marius Dinu <m95d+git@psihoexpert.ro>

toolchain: binutils: switch to version 2.42 by default

Change the default binutils version to 2.42.

Tested-by: Jonathan Lancett <j.lancett@ntlworld.com>
Tested-by: Andre Heider <a.heider@gmail.com>
Signed-off-by: Nick Hainke <vincent@systemli.org>

malta: refresh kernel 6.6 configuration

This refreshes the configuration on top of kernel 6.6 and activates it
as test kernel configuration.

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>

kernel/malta: 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.

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

Signed-off-by: Tony Ambardar <itugrok@yahoo.com>

kernel/malta: Create kernel files for v6.6 (from v6.1)

This is an automatically generated commit.

When doing `git bisect`, consider `git bisect --skip`.

Signed-off-by: Tony Ambardar <itugrok@yahoo.com>

scripts/kernel_bump: Fix commit text formatting

Text of a commit message body should wrap at 75 characters. Manual commits
are expected to do so, but automated commits *must* do so to avoid adding
repeated ugly commits.

Signed-off-by: Tony Ambardar <itugrok@yahoo.com>

uqmi: set dangling-pointer as error

With "ebfe8b4 CMakeLists: set no-dangling-pointer" the compilation
option is set in uqmi, and can therefore be removed from no-error.

Signed-off-by: Jean Thomas <jean.thomas@wifirst.fr>

uqmi: update to git HEAD

e7207be uqmi: print radio interfaces in serving system command
6ef41d6 uqmi: create function to print radio interface string
e25d042 uqmi: Add basic 5G NR support
3e782be uqmi: sync data from libqmi project
368d46c uqmi: support C reserved keywords in upstream JSON files
02e42c0 reorganize source code in common and uqmi specific parts
4591f0a .gitignore build/ directories
2b57ee1 uqmi: commands-uim: fix uninitialized use of card_application_state
7c77e77 data/code-gen: add support for indications
ddbf864 qmi-struct.h: add missing includes
5320c1d move qmi_get_error_str to into utils.c
1503bc7 dev.c: add missing import strings.h
bae945f commands-nas: add missing includes
9ffd0e2 commands: make `struct blob_buf status` public
a4fbdcc commands-nas: fix gcc warning
8ff632a dev.c: add comment to qmi_request_wait()
a043a74 CMakeLists: refactor SOURCES variable to allow later adding uqmid
ebfe8b4 CMakeLists: set no-dangling-pointer
c47125d CMakeLists: improve generated files
0f64b69 CMakeLists: update cmake minimum version to 3.5

As the built uqmi binary is now moved to a dedicated directory,
update the Makefile accordingly.

Signed-off-by: Jean Thomas <jean.thomas@wifirst.fr>

libunwind: update to 1.8.1

Rebased patches:
- 003-fix-missing-ef_reg-defs-with-musl.patch
- 004-ppc-musl.patch

Signed-off-by: Weijie Gao <hackpascal@gmail.com>

grub2: update to 2.12

compile tested: x86_64,i386,loongarch64

Signed-off-by: Weijie Gao <hackpascal@gmail.com>
{Refresh patches}
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>

mediatek: bpi-r4: store random MAC addresses for the BPi-R4

Despite coming with multiple I2C EEPROMs supposedly dedicated for that
purpose, the BPi-R4 does not seem to have factory assigned MAC addresses.
Hence, just like for all other BPi boards, store a randomly generated
MAC address on first boot and derive WAN and Wi-Fi MAC addresses from
that as well. Not perfect, but better than random on every boot.

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

mediatek: bpi-r3-mini: persist WiFi and WAN MAC addresses

Derive WAN and WiFi MAC addresses from the (random) MAC address stored
in the U-Boot environment.

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

kernel: bump 6.6 to 6.6.29

Changelog: https://cdn.kernel.org/pub/linux/kernel/v6.x/ChangeLog-6.6.29

Removed upstreamed:
generic/backport-6.6/740-v6.9-01-netfilter-flowtable-validate-pppoe-header.patch[1]
generic/backport-6.6/740-v6.9-02-netfilter-flowtable-incorrect-pppoe-tuple.patch[2]
generic/backport-6.6/790-29-v6.9-net-dsa-mt7530-fix-improper-frames-on-all-25MHz-and-.patch[3]
generic/backport-6.6/790-31-v6.10-net-dsa-mt7530-fix-enabling-EEE-on-MT7531-switch-on-.patch[4]
generic/backport-6.6/790-34-v6.10-net-dsa-mt7530-fix-mirroring-frames-received-on-loca.patch[5]
generic/backport-6.6/790-35-v6.10-net-dsa-mt7530-fix-port-mirroring-for-MT7988-SoC-swi.patch[6]
mediatek/patches-6.6/963-net-ethernet-mtk_eth_soc-fix-WED-wifi-reset.patch[7]

Manually rebased:
generic/backport-6.6/790-23-v6.9-net-dsa-mt7530-get-rid-of-priv-info-cpu_port_config.patch

All other patches automatically rebased.

1. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.6.29&id=a2471d271042ea18e8a6babc132a8716bb2f08b9
2. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.6.29&id=4ed82dd368ad883dc4284292937b882f044e625d
3. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.6.29&id=21b9d89d93422221cdda1b82fd075fa3c94a11d9
4. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.6.29&id=bd41ee1efd478852a0882ce5f136bc2b5e83eff2
5. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.6.29&id=d1be3960539249a8690ed09a29d0e3bf34189dd2
6. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.6.29&id=f8de1b6208bf71bd3102548d33dd8475573ad2ea
7. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.6.29&id=6855f724f19620c3ddff57c349e0abba797475b1

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

Signed-off-by: John Audia <therealgraysky@proton.me>

mbedtls: Update to version 3.6.0

This adds support for mbedtls 3.6.0.
The 3.6 version is the next LTS version of mbedtls.
This version supports TLS 1.3.

This switches to download using git. The codeload tar file misses some
git submodules.

Add some extra options added in mbedtls 3.6.0.

The size of the compressed ipkg increases:
230933 bin/packages/mips_24kc/base/libmbedtls13_2.28.7-r2_mips_24kc.ipk
300154 bin/packages/mips_24kc/base/libmbedtls14_3.6.0-r1_mips_24kc.ipk

The removed patch was integrated upstream.

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

hostapd: Fix compile against mbedtsl 3.6

Fix compile of the mbedtls extension for hostapd.

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

uencrypt: Fix mbedtls 3.6 compatibility

This makes it compile with mbedtls 3.6.0.

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

uencrypt: Fix compile warnings

keylen and ivlen are of type long and not size_t.

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

at91: 5.15: remove old config and patch files

6.1 is now default so remove old config and patch files.

Signed-off-by: Nick Hainke <vincent@systemli.org>

at91: set 6.1 as default kernel version

Adjust the kernel version to default 6.1 in order to facilitate testing
for a broader audience.

Signed-off-by: Nick Hainke <vincent@systemli.org>

zynq: 6.1: set testing kernel

Allow selecting 6.1 as testing kernel on zynq.

Tested-by: Luis Araneda <luaraneda@gmail.com> # Zybo Z7-20
Signed-off-by: Nick Hainke <vincent@systemli.org>

zynq: 6.1: refresh kernel config

Refresh the kernel config by runnnig make kernel_menuconfig.

Signed-off-by: Nick Hainke <vincent@systemli.org>

kernel/zynq: Restore kernel files for v5.15

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: Nick Hainke <vincent@systemli.org>

kernel/zynq: Create kernel files for v6.1 (from v5.15)

This is an automatically generated commit.

When doing `git bisect`, consider `git bisect --skip`.

Signed-off-by: Nick Hainke <vincent@systemli.org>

uboot-mvebu: backport patch for Turris Omnia to enable LTO

Fixes issue while building package uboot-mvebu on OpenWrt 23.05:

u-boot-with-spl.kwb exceeds file size limit:
  limit:  0xf0000 bytes
  actual: 0xf0100 bytes
  excess: 0x100 bytes
make[3]: *** [Makefile:1466: u-boot-with-spl.kwb] Error 1
make[3]: *** Deleting file 'u-boot-with-spl.kwb'
make[3]: Leaving directory '/workspaces/openwrt/build_dir/target-arm_cortex-a9+vfpv3-d16_musl_eabi/u-boot-omnia/u-boot-2024.04'
make[2]: *** [Makefile:83: /workspaces/openwrt/build_dir/target-arm_cortex-a9+vfpv3-d16_musl_eabi/u-boot-omnia/u-boot-2024.04/.built] Error 2

Signed-off-by: Josef Schlehofer <pepe.schlehofer@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15307
Signed-off-by: Robert Marko <robimarko@gmail.com>

ipq40xx: refresh patches again

CI tells us that we need to, so lets refresh them.

Fixes: 4265fe9348c0 ("ipq40xx: refresh patches")
Link: https://github.com/openwrt/openwrt/pull/15310
Signed-off-by: Robert Marko <robimarko@gmail.com>

package/network/utils/iptables: fix PKG_CPE_ID

cpe:/a:netfilter:iptables is the correct CPE ID for iptables:
https://nvd.nist.gov/products/cpe/search/results?keyword=cpe:2.3:a:netfilter:iptables

Fixes: c61a2395140d92cdd37d3d6ee43a765427e8e318 (add PKG_CPE_ID ids to package and tools)
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15297
Signed-off-by: Robert Marko <robimarko@gmail.com>

package/libs/libjson-c: fix PKG_CPE_ID

cpe:/a:json-c:json-c is the correct CPE ID for libjson-c:
https://nvd.nist.gov/products/cpe/search/results?keyword=cpe:2.3:a:json-c:json-c

Fixes: c61a2395140d92cdd37d3d6ee43a765427e8e318 (add PKG_CPE_ID ids to package and tools)
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15292
Signed-off-by: Robert Marko <robimarko@gmail.com>

package/network/services/dropbear: fix PKG_CPE_ID

cpe:/a:dropbear_ssh_project:dropbear_ssh is the correct CPE ID for dropbear:
https://nvd.nist.gov/products/cpe/search/results?keyword=cpe:2.3:a:dropbear_ssh_project:dropbear_ssh

Fixes: c61a2395140d92cdd37d3d6ee43a765427e8e318 (add PKG_CPE_ID ids to package and tools)
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15290
Signed-off-by: Robert Marko <robimarko@gmail.com>

tools/expat: fix PKG_CPE_ID

cpe:/a:libexpat_project:libexpat is the correct CPE ID for expat:
https://nvd.nist.gov/products/cpe/search/results?keyword=cpe:2.3:a:libexpat:libexpat

Fixes: c61a2395140d92cdd37d3d6ee43a765427e8e318 (add PKG_CPE_ID ids to package and tools)
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15291
Signed-off-by: Robert Marko <robimarko@gmail.com>

ipq-wifi: correct PKG_MIRROR_HASH

It seems that somehow a wrong hash has been used for ipq-wifi, so refresh
it.

Fixes: f10d55df9e0a ("ipq-wifi: update to Git HEAD (2024-04-26)")
Signed-off-by: Robert Marko <robimarko@gmail.com>

ipq40xx: refresh patches

CI tells us that we need to, so lets refresh them.

Signed-off-by: Robert Marko <robimarko@gmail.com>

netifd: packet_steering: fix shell error on unset steering_flows option

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

netifd: fix bogus warnings in packet_steering init script

Avoids warnings if options are unset

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

kernel: update fraglist GRO patch to the latest upstream submission

Cosmetic fixes + IPv6 NAT support

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

kernel: fix crash with multicast-to-unicast and fraglist GRO

Avoid calls to skb_copy, which creates invalid packets that cause a crash
on segmentation

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

generic: v6.1, v6.6: add patch to fix PHY-muxing on MT7530

Move accepted patches to backport folder, re-add previously removed patch
which caused havoc on MT7621 and add the (still pending) fix.

Fixes: d40691a5fb ("generic: 6.1, 6.6: mt7530: import pending patches")
Signed-off-by: Daniel Golle <daniel@makrotopia.org>

kernel: generic backport 6.1: rename mistyped .patch file

Rename mistyped file .patc to patch

Fixes: ac8bfe316b ("generic: 6.1, 6.6: mt7530: import accepted patches")
Signed-off-by: Antonio Flores <antflores627@gmail.com>

ubox: update to Git HEAD (2024-04-26)

85f1053019ca kmodloader: fix insmod path logic

Signed-off-by: Robert Marko <robimarko@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15288
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>

bcm4908: set kernel 6.6 as a testing version

Add kernel 6.6 for testing.

Signed-off-by: Mieczyslaw Nalewaj <namiltd@yahoo.com>

bcm4908: remove 5.15 support

Now that 6.1 is default, remove 5.15 support.

Signed-off-by: Mieczyslaw Nalewaj <namiltd@yahoo.com>

lldpd: fix reload bug: advertisements shall default to on

Because these capability advertisements default to on in lldpd, they
became absent at reload, and not restart, due to how the reload logic
works ( keep daemon running, send unconfigured and then the new config
via socket ), and it was not evident unless you happened to be looking
for it (e.g. via pcap or tcpdump). It was also not evident from the
manpage ( have now sent patches upstream ).

At reload time, the unconfigure logic disabled them unless they were
explicitly enabled (compare with other settings where 'unconfigure' just
resets them). Now they default to on/enabled at init time, and are
explicitly 'unconfigure'd at startup if the user disables them via:

lldp_mgmt_addr_advertisements=0
lldp_capability_advertisements=0

In other words: explicit is necessary to disable the advertisements.

The same applies to 'configure system capabilities enabled'. Technically
'unconfigure'd is the default but now it is explicit at reload.

Tested on: 23.05.3

Signed-off-by: Paul Donald <newtwen+github@gmail.com>

toolchain/nasm: fix PKG_CPE_ID

cpe:/a:nasm:netwide_assembler is the correct CPE ID for nasm:
https://nvd.nist.gov/products/cpe/search/results?keyword=cpe:2.3:a:nasm:netwide_assembler

Fixes: bcf02c5d3123a99c717ca33f1d7c6250acf0f33f (toolchain: assign PKG_CPE_ID)
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>

tools/flex: fix PKG_CPE_ID

cpe:/a:westes:flex is the correct CPE ID for flex:
https://nvd.nist.gov/products/cpe/search/results?keyword=cpe:2.3:a:westes:flex

Fixes: c61a2395140d92cdd37d3d6ee43a765427e8e318 (add PKG_CPE_ID ids to package and tools)
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>

tools/zlib: fix PKG_CPE_ID

cpe:/a:zlib:zlib is the correct CPE ID for zlib:
https://nvd.nist.gov/products/cpe/search/results?keyword=cpe:2.3:a:zlib:zlib

Fixes: c61a2395140d92cdd37d3d6ee43a765427e8e318 (add PKG_CPE_ID ids to package and tools)
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>

package/libs/pcre2: fix PKG_CPE_ID

cpe:/a:pcre:pcre2 is the correct CPE ID for pcre2:
https://nvd.nist.gov/products/cpe/search/results?keyword=cpe:2.3:a:pcre:pcre2

Fixes: c39b0646f3f2d96d40f601209859175af8537b6d (pcre2: import pcre2 from packages feed)
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>

qualcommax: IPQ807x: ZyXEL NBG7815: Fix random Wifi MAC
For this particualar device we get random MAC's for Wifi on each (re-)boot.
This is because art partition/pre caldata do not contain valid MAC addresses.

As we have now a new/better approach with ath11k_patch_mac we can use it for
this device too.

I'm using this approach for like two weeks and its working flawlessly.

Signed-off-by: Ulrich Stark <pwned-pixel@posteo.de>
qualcommax: IPQ807x: ZyXEL NBG7815: Fix random Wifi MAC
Changing order to 3/phy0/5G-1, 2/phy1/2G, 4/phy2/5G-2.

Signed-off-by: Ulrich Stark <pwned-pixel@posteo.de>

ustp: update to Git HEAD (2023-05-29)

a85a5bc83bde netif_utils: correctly close fd on read error

Signed-off-by: Paul Donald <newtwen+github@gmail.com>

ipq40xx: fix USB on Aruba AP-303H

Enable USB 3.0 controller, disable USB 2.0 controller.

The USB 2.0 port on the AP-303H is actually connected to the USB 3.0
controller's HS phy. Enable the HS phy only, since the SS lanes are not
brought out to the connector.

Signed-off-by: Ryan Salsbury <ryanrs@gmail.com>

ipq40xx: use nvmem ethernet MACs on Aruba AP-303H

Use NVMEM to assign "factory sticker" MAC address to WAN ethernet
interface. Set LAN address to sticker + 1.

Signed-off-by: Ryan Salsbury <ryanrs@gmail.com>

generic: 6.1, 6.6: remove patch which breaks WAN on MT7621

Importing pending patch "net: dsa: mt7530: move MT753X_MTRAP operations
for MT7530" broke WAN connectivity on most MT7621 which use PHY-muxing
to hook up either port 0 or port 4 to GMAC1.

Remove it for now until the author submits a fixed version.

Fixes: https://github.com/openwrt/openwrt/issues/15279
Fixes: https://github.com/openwrt/openwrt/issues/15273
Fixes: d40691a5fb ("generic: 6.1, 6.6: mt7530: import pending patches")
Signed-off-by: Daniel Golle <daniel@makrotopia.org>

kernel: move mv88e6xxx fix to generic backports

The backports introduced in commit d40756563c46 ("kernel: backport
phylink changes from mainline Linux") broke the mv88e6xxx DSA driver.
A backport to fix this was added to the kirkwood target, but as it is
used in multiple targets, and there's a kmod package for it, the fix
should be in generic backports.

This fixes the switch on the WatchGuard Firebox M300 when running the
6.1 testing kernel.

There is no need to backport the fix for the 6.6 kernel, as it was
included in 6.6.5.

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>

perf: fix build on PowerPC

Building perf's intel-pt-decoder fails on both PPC32 and PPC64:

/home/stijn/Development/OpenWrt/openwrt/staging_dir/toolchain-powerpc64_e5500_gcc-13.2.0_musl/lib/gcc/powerpc64-openwrt-linux-musl/13.2.0/../../../../powerpc64-openwrt-linux-musl/bin/ld.bfd:
/home/stijn/Development/OpenWrt/openwrt/build_dir/target-powerpc64_e5500_musl/linux-qoriq_generic/linux-6.1.86/tools/
perf-target-powerpc64_e5500_musl/perf-in.o: in function `insn_set_byte':
/home/stijn/Development/OpenWrt/openwrt/build_dir/target-powerpc64_e5500_musl/linux-qoriq_generic/linux-6.1.86/tools/perf/util/intel-pt-decoder/../../../arch/x86/include/asm/insn.h:64:
undefined reference to `__le32_to_cpu'

Add NO_AUXTRACE=1 to MAKE_FLAGS for LINUX_KARCH powerpc, which disables
build of intel-pt-decoder on both PPC32 and PPC64.

Signed-off-by: Stijn Tintel <stijn@linux-ipv6.be>

ipq-wifi: update to Git HEAD (2024-04-26)

fab9e29f6b92 ipq6018: update regdb in TPLink EAP610-Outdoor BDF
6d02b65fadf3 ipq8074: update RegDB in new submitted BDF
644ba9ea2e66 ipq6018: update RegDB in new submitted BDF

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

ipq40xx: set DEVICE_DTS_DIR to /qcom by default

Set DEVICE_DTS_DIR to /qcom by default instead of limiting it to
TESTING_KERNEL since we moved 6.6 to default version.

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

ipq806x: set DEVICE_DTS_DIR to /qcom by default

Set DEVICE_DTS_DIR to /qcom by default instead of limiting it to
TESTING_KERNEL since we moved 6.6 to default version.

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

bcm53xx: update 600-net-disable-GRO-by-default.patch after recent GRO change

Fixes build error

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

ipq40xx: drop 6.1 support

Drop support for ipq40xx for kernel 6.1.

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

ipq40xx: switch default to 6.6

Switch default kernel version for ipq40xx to 6.6.

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

ipq806x: drop 6.1 support

Drop support for ipq806x for kernel 6.1.

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

ipq806x: switch default to 6.6

Switch default kernel version for ipq806x to 6.6.

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

ath79: update 900-unaligned_access_hacks.patch after recent GRO change

Fixes build error

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

sunxi: update the 6.6 DTS_DIR hack

Update and simplify the 6.1 vs. 6.6 DTS_DIR hack until 6.1 support is gone.

Signed-off-by: Zoltan HERPAI <wigyori@uid0.hu>

netifd: add flow steering mode to the packet steering script

This allows directing processing of locally received packets to the CPUs
of the tasks receiving them

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

netifd: add a packet steering mode matching the old script

This spreads packet processing across all cores

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

kernel: improve GRO performance

For packets not belonging to a local socket, use fraglist GRO instead of
regular GRO. This make segmenting packets very cheap and avoids the need for
selectively disabling GRO

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

kernel: backport flow offload pppoe fix

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

tools/elfutils: refresh portability patch for macOS

Quilt refresh combined two sets of changes to the same file.

The switch from using libgen.h to dirname.h because of function poisoning
from gnulib's import of basename() was added as a new patch hunk instead
of an edit to the original one.

The original patch hunk was to fix build errors on an earlier version of
elfutils before the "dirname" module was being imported to fix further
build errors with the 0.191 version.

Tested-by: Georgi Valkov <gvalkov@gmail.com> # MacOS
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>

tools/elfutils: remove no-op copy of tdestroy()

A false tdestroy() function was added in order to make elfutils build on
macOS again. A previous commit added declarations for a real version of
tdestroy() into gnulib, which is already imported, as well as the
preprocessor flags and the triggers for the Makefile.am conditional in
order to include the source to be built.

Tested-by: Georgi Valkov <gvalkov@gmail.com> # MacOS
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>

tools/elfutils: use locally declared static reallocarray()

On macOS, stdlib.h in the standard include paths does not provide
reallocarray() while both elfutils and gnulib do, however they are
declared differently, leading to an error:

  ./system.h:101:1: error: static declaration of 'reallocarray' follows non-static declaration
  reallocarray (void *ptr, size_t nmemb, size_t size)

A normal "configure && make" build cycle results in both declarations
being enabled as a result of both elfutils and gnulib having completely
separate configure checks where gnulib uses an internal placeholder symbol
HAVE_REALLOCARRAY, and elfutils uses a standard autoconf macro
HAVE_DECL_REALLOCARRAY.

Fix this by excluding the import of the reallocarray module which causes
gnulib checks in the configure stage to not even consider whether to
declare reallocarray later on, so the decision is only between the
standard include stdlib.h and the elfutils header.

Tested-by: Georgi Valkov <gvalkov@gmail.com> # MacOS
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>

tools/elfutils: organize gnulib import build stage

Organize the Makefile lines involved in gnulib importing and its
workarounds. It improves readability and keeps git history organized.

Tested-by: Georgi Valkov <gvalkov@gmail.com> # MacOS
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>

tools/elfutils: override SUBDIRS variable of Makefile

Instead of editing the SUBDIRS variable with a patch, it can be overriden
at the end of the command line when invoking Make.

This tool has a series of recursive Makefiles in each subdirectory,
therefore SUBDIRS is set to a pattern of Make functions so that the result
is variable depending on the current subdirectory that Make is being
invoked in.

It's not necessary to have gnulib-cache.m4 in EXTRA_DIST since we don't
need to re-import after packaging this in the SDK, so get rid of the
entire patch hunk for ./Makefile.am

Tested-by: Georgi Valkov <gvalkov@gmail.com> # MacOS
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>

tools/elfutils: update to 1.91

Release Notes:
https://sourceware.org/pipermail/elfutils-devel/2024q1/006876.html

Manually refresh:
- 100-portability.patch

Change:
- replace libgen.h with gnulib "dirname" module for compilation errors:
    In file included from ./../libdw/libdwP.h:38,
                     from eblobjnote.c:42:
    /usr/include/libgen.h:35:9: error: attempt to use poisoned "basename"
       35 | #define basename        __xpg_basename
          |         ^

Tested-by: Georgi Valkov <gvalkov@gmail.com> # MacOS
Co-Developed-by: Nick Hainke <vincent@systemli.org>
Signed-off-by: Nick Hainke <vincent@systemli.org>
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>

tools/coreutils: update to 9.5

Update to latest stable release.

The following commits in gnulib caused a conflict in locally bootstrapped
coreutils with stable gnulib:

  8f4b4e52c991de2233b471f8e35a068866b31f01
  2749234203959df8d72cd8638d4e00a9fff450db

A module (strftime) was marked deprecated and replaced by another module
(nstrftime) in the version of gnulib that coreutils was released with
compared to the stable branch that we use for importing. Conflicts from
the previous version of coreutils are now gone, so other imported headers
are now good.

Refresh patch:
 - 000-bootstrap.patch

Remove upstreamed patch:
 - 001-bootstrap-sync.patch

Link: https://lists.gnu.org/archive/html/coreutils/2024-03/msg00132.html
Tested-by: Georgi Valkov <gvalkov@gmail.com> # MacOS
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>

tools/coreutils: update to 9.4

Update to latest stable release.

Add configure option to disable support for the Year 2038 problem.
(for now, as some versions of GCC do not yet support it)

Syncing bootstrap script fails, backport an upstream patch which can be
removed at next coreutils update.

Several headers from the stable gnulib branch cause build failure because
the changes in the imported versions are incompatible with the Makefile
that gets generated for coreutils. This version of coreutils was released
after being bootstrapped and autoreconf'ed with a significantly different
version of gnulib compared to our local gnulib, so skip importing them
(and restore the backup).

While at it, organize restoring the originally shipped version of files
into a Make foreach function.

Refresh patch:
- 000-bootstrap.patch

New patch:
- 001-bootstrap-sync.patch

Link: https://lists.gnu.org/archive/html/coreutils/2023-08/msg00099.html
Tested-by: Georgi Valkov <gvalkov@gmail.com> # MacOS
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>

tools/bison: hardcode path to m4 with STAGING_DIR_HOST

Force bison to ignore the M4 environment variable and hardcode it to the
locally built m4 during build operations using the relocatable path
variable STAGING_DIR_HOST.

This allows bison to continue to function while we are forcefully avoiding
autoreconf and other autoconf and automake-like operations by giving a
fake path to m4 with the M4 environment variable.

The specific path can still be overridden independently from the
environment within the line of invocation that runs bison by setting
STAGING_DIR_HOST within the command, so document this in the help printout.

Tested-by: Georgi Valkov <gvalkov@gmail.com> # MacOS
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>

gettext-full: update to 0.22.5

Release Announcement:
https://savannah.gnu.org/news/?group_id=425

Refresh:
- 200-libunistring-missing-link.patch

Tested-by: Georgi Valkov <gvalkov@gmail.com> # MacOS
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>