This was a fallout from the /usr/share relocation, only found and
reported now by z3ntu. If the device package happened to be installed
later, then the link would point to the wrong location.
Fixes 1d748a93dd
Allows things like podman/docker to work out of the box.
I looked at /etc/init.d/cgroups, and having this service enabled isn't a
problem if the kernel doesn't have cgroup support... in that situation,
the service script just simply exits (success).
[ci:skip-build] already built successfully in CI
Manual config changes:
* Enable SYSCON_REBOOT_MODE
* Enable LEDS_QCOM_LPG (not yet used but initial driver support present)
[ci:skip-build]: Already built successfully in CI.
For some reason kernel does not print anything to ttySAC3 even with
console=/dev/ttySAC3. If earlycon is specified it prints the early
stuff though.
Add ttySAC3 in both places anyways since that should be the correct
uart.
The riscv64 builder in Alpine is currently stuck, let's disable the
package for riscv64 temporarily so bpo isn't stuck on it for master. I
doubt anybody is using it on riscv64 currently.
[ci:skip-vercheck]
[ci:skip-build]: already built successfully in CI
EFI booting with systemd-boot requires building the kernel with ZBOOT -
the EFI decompressor stub. However, this breaks the old Android booting.
Detect when ZBOOT is enabled and install both vmlinuz.efi as well as the
old gzipped kernel image.
To help make this upgrade more seamless, continue to support building
the kernel with ZBOOT disabled and just print a warning.
Disable module stripping - we compress them anyway and a recent kernel
change seems to have caused some bug here.
This is a packaging change that doesn't require rebuilding the kernel.
[ci:skip-vercheck]
Co-developed-by: Jakob Hauser <jahau@rocketmail.com>
Co-developed-by: Pablo Correa Gomez <ablocorrea@hotmail.com>
[ci:skip-build]: Already built successfully in CI.
Something seems to have gone wrong in our rebuild and sync of pulseaudio
with alpine upstream. And user feedback points that building armv7 devices
with LTO is actually broken in our builders. In alpine, they enabled lto
in pulseaudio very long ago (f46174365b219fe431890df3df7e7dc488c9451e 08-09-22)
and nobody as complained so far. Since it's not really a phone-specific package,
one would hope that more people upstream would have noticed if it were broken
there. Since nobody has, we assume this is a downstream problem, and just
unconditionally disable lto in our build. We hope this will be just a temporary
workaround, until the patches that prompted the fork are no longer needed, and
we can simply use upstream pulseaudio, that shouldn't be broken.
Fixes#2368
[ci:skip-build] already built successfully in CI
As described in [0] let's switch away from RNDIS and use NCM instead.
Since we cannot force all kernels to switch at the same time, let's keep
a fallback to RNDIS in the setup function.
We can also remove usb_f_rndis from modprobe as the module gets loaded
automatically when needed, but instead we need to load libcomposite
manually so that $configfs/usb_gadget gets created.
[0] https://gitlab.com/postmarketOS/pmaports/-/issues/1797
[ci:skip-build] already built successfully in CI
With recent gcc versions we get errors like:
/linux/arch/arm/mm/proc-v7.S: Assembler messages:
/linux/arch/arm/mm/proc-v7.S:425: Error: junk at end of line, first unrecognized character is `#'
Seen on at least two samsung kernels based on 3.4 and 3.10. Fix issue
by backporting (part of) commit found in mainline.
