* Bump buildroot to update package/pigz
* Enable parallel gzip for faster Docker pulls
Docker checks if unpigz is available, and if so uses it to unpack
container layers with multiple CPU cores. This should make Docker pulls
faster, especially on lower end hardware.
Add timeout to expect call when waiting for the OS reboot after
switching slots. While it never fails for me locally, it regularly
breaks tests in GHA.
Since update to Systemd v256.x the Range header requires the num_entries part
and fails if it's not provided, which we worked around by [1]. With this patch
that was already accepted upstream, the workaround shouldn't be necessary
anymore.
[1] https://github.com/home-assistant/supervisor/pull/5827
Add test that OS update works - use the whole stack using CLI to update to the
latest stable version (unless executed manually on the latest stable release,
this version should never be the same as the currently tested one).
With this test in place, we can also test command for switching the slots, so
add an extra test for that too.
Fixes#4103
Add driver for Marvell PHYs, such as 88E1543(4L) on an ASRock C3758D4I-4L
board. Adding it to x86 config only, as it seems it's not widely used anywhere
else.
Fixes#4025
Bump Hailo stuff to the latest version. While this is a breaking change for
add-ons depending on the driver, the most commonly used one (i.e. Frigate)
didn't bump to v4.20.1 on their stable channel either, so it shouldn't have
significant impact. We agreed with @blakeblackshear that once HAOS bumps the
Hailo driver in HAOS 16, Frigate will follow.
Backport /boots endpoint for Systemd so we can use it in Supervisor to get the
actual list of boots. Should be available upstream since Systemd v258, for v256
minor tweaks were needed.
When creating OVA image, the CPU is slacking at the end of the build because it
is creating three ZIP archives, each one on a single CPU only. As we're
creating only single-entry archives, we can use pigz to use all cores.
The actual speedup on my machine (16C/32T) reflects the number of cores - it
takes around 2 seconds instead of 1 minute.
Since
127c420335
change in package/systemd, this option is patched by systemd build because
userspace FW loading has never been supported with Systemd. This should have no
runtime effect, just clear the warning about disabled option.
Fix build job to write config option for channel switching from #4043 to the
actual config. As it was written to .config in the top-level build directory,
it was never correctly applied.
* package/vcgencmd: add tool for RPi VideoCore commands
This tool is used by rpi-eeprom-update and is fairly lightweight binary without
dependencies. Use it as-is from raspberry/utils repo.
* package/rpi-eeprom: change package to install EEPROM userspace scripts
* configs: enable rpi-eeprom for rpi4, rpi4-64, rpi5-64 and yellow
On Pi5 and Yellow also enable flashrom so the firmware can be installed
directly without recovery being involved. On Yellow/CM4 this can't be done
without config.txt changes though (SPI and pinmuxing needs to be enabled) but
the image is shared there and users may eventually use the tools if they want,
so install BCM2711 on Yellow too. The "officially recommended" method is
rpiboot though, which is also documented in Yellow docs.
Because we use custom compatible strings in Yellow DTS's, the firmware loader
first attempts to load a firmware with this compatible in its name. Because it
doesn't exists, it shows error like this one before falling back to a more
generic one:
brcmfmac mmc1:0001:1: Direct firmware load for brcm/brcmfmac43455-sdio.raspberrypi,5-compute-module-ha-yellow.bin failed with error -2
While these errors are mostly harmless, add symlinks with our compatible in the
name to suppress them. Instead of patching upstream
package/brcmfmac_sdio-firmware-rpi which installs the firmware files, add them
to yellow overlay to make maintenance easier.
Latest RPi firmware package contains module options that supposedly improve
stability, with details described in [1].
Since the feature_disable mask also disables the dump_obss feature, this change
would also mitigate `brcmf_set_channel: set chanspec ... fail` messages still
seen in some environments even after #3719.
Fixes#3367
[1] 2788cb549a
Update to latest binary of 43455 firmware and add missing symlinks which
suppress warnings/file not found errors when loading the firmware on CM5.
* buildroot 50fcf58bfa...62bf5c5af5 (2):
> package/brcmfmac_sdio-firmware-rpi: add CM5/Pi 500 symlinks to 43455 FW
> package/brcmfmac_sdio-firmware-rpi: bump version to 4eec7f2
* Fix U-Boot config to access all RAM on 16 GB CM5
U-Boot defconfig used for Yellow checks only 4 DRAM banks, however, CM5 with 16
GB has the memory spread across 8 banks. Add a patch (submitted upstream) to
the defconfig to get access to the whole RAM.
Fixes#3989
* Add Upstream tag with link to uboot patchwork
Add input allowing to override the channel that's used for hassio image
downloads and in runtime for Supervisor updates, building on the option added
in #3618.
The new default is dev for dev builds, for GH releases keep using the stable
channel both for releases and pre-releases (so we could catch any stable issues
before beta is moved to stable).
To keep it DRY and idiomatic, create a new in-repo GH action for running
commands in the build container.
* Make usage of top-level make easier, drop 'all' target
Make it easier when using top-level make - proxy all possible commands to
Buildroot make and only wrap build for individual target builds. This way it's
still possible to run e.g. 'make ova' which would read the defconfig and run
the build, while we can also use the top-level make in the same way as it's in
vanilla Buildroot.
Target 'all' was dropped in favor of Buildroot 'make' without any arguments -
as it's fairly pointless to run all builds sequentially. With the current 19
targets it would take about a day even on a decent hardware and the build
artifacts would be lost in the process.
* Show warning only if BR2_DEFCONFIG changes
* Wait for 10s or input if defconfig differs
* Fall back to buildroot make in top-level make
To make running Buildroot commands easier, define .DEFAULT rule and fall back
to targets from Buildroot with necessary variables set. This makes
"savedefconfig" redundant as it's been simply passed to BR.
* Also implicitly fall back to 'clean' target
* Fix typo
* Update RPi kernel to 6.12.20
Update to latest stable RPi kernel and remove unnecessary 6.6.y kernel config
fragments.
* Refresh RPi and Yellow patches
Rebase all patches on 6.12.20, remove patches that are already present
upstream.
* Update Yellow device trees for 6.12.20
Upstream changes broke our downstream device trees. While the CM4 fix was
trivial, there were more changes in the CM5 device tree due to adaptation to
upstream code. To simplify future maintenance, DTS was refactored to reuse CM5
DTS include and override only what's necessary.
* Bump buildroot to update to matching package/rpi-firmware
* buildroot ead21eb6d2...cd82256125 (1):
> package/rpi-firmware: bump version to f49a396 (1.20250326)
