Fix for:
missing depend(s): so:libicuuc.so.70
While at it, I've checked if this is the most recent version of this
mobile fork, and what the upstream stance on this is. Meanwhile there's
a feature branch in the official dino.git repository, so I've adjusted
the APKBUILD. The commit is still the same, and apparently upstream
won't merge it, but rather switch the whole application to libadwaita.
Related: https://github.com/dino/dino/issues/178
Related: https://github.com/dino/dino/issues/1182
[ci:skip-build] already built successfully in CI
Helps #1479
The Disable-fullscreen patch from Purism is in principle useful
in mobile, as exiting fullscreen is at least troublesome. However,
the patch has not been upstreamed and, although we could carry it
in alpine for the 42 cycle, the 43 cycle will come with GTK4 support
and the patch will not rebase easily. Therefore, instead of waiting
6 months to drop, lets just drop it now.
[ci:skip-build] already built successfully in CI
Updates the u-boot on the Librem 5 to a branch that is much closer to
mainline u-boot.
- adds support for booting from SD card to the u-boot boot script
- m4 is no longer necessary, at least not right now (may be re-added by
purism in the future for supporting some low power modes)
- adds a patch to enable u-boot console on uart 4 (on the m.2 connector)
for the debug board
- drops the patch to enable efidebug command, this isn't necessary
anymore with mainline u-boot
Co-authored-by: Bart Ribbers <bribbers@disroot.org>
AppStream data for postmarketOS has been removed until the new format
without a double "--" is deployed. Otherwise transition will be a pain
and we would have to take care of it downstream.
Relates #1384
This requires a kernel >=5.15, and uses a better focus implementation
that doesn't require "sudo i2ctransfer"
[ci:skip-build] already built successfully in CI
Depencency chain:
gnome-software -> gnome-softare-plugin-apk -> gnome-software-lib -> gnome-software
is broken.
Application should not depend on its own plugin,
instead plugin should probably be installed using
install_if="gnome-software apk-tools".
This is needed to unblock BPO:
master/x86_64/gnome-software-9999_git20211203-r0.apk (missing depends: gnome-software)
[ci:skip-build] already built successfully in CI
The plugin now is linked dynamically against gnome-software
and pmos fork does not touch the library, only the ui,
so it should be save to drop this fork
This enables an auto-update mechanism on post-upgrade for the u-boot
firmware on the librem 5 and librem 5 devkit. This mechanism is largely
copied from the implementation in the pinephone u-boot package.
efidebug is useful for booting the kernel via EFI mode in u-boot.
Fix current version of epiphany not being installable because of soname
bump in libportal:
/ # apk add epiphany
ERROR: unable to select packages:
so:libportal.so.0 (no such package):
required by: epiphany-9999_git20210916-r0[so:libportal.so.0]
Upgrade to 41.3 while at it, and drop the firefox sync dialog patch
since it doesn't apply cleanly anymore. Only the disable fullscreen
patch is really important, Pablo only included the firefox sync patch
because it applied cleanly.
This packages the megapixels fork that purism has supporting the librem
5.
The `provides="megapixels"` is there so that apk doesn't try to install
both megapixels and this package side-by-side. If the user explicitly
installs megapixels-purism, apk will remove megapixels first. The
original megapixels is still preferred by apk by default.
[ci:skip-build]: already built successfully in CI
alsa-* 1.2.6 breaks audio switching and microphones on several devices,
with some outstanding issues still open upstream.
This 'fork' is meant to be very short-lived, to restore audio on devices
while we figure out how to fix it on 1.2.6.
see: https://gitlab.com/postmarketOS/pmaports/-/issues/1334
This will be upgraded soon, however let's let this broken URL not block
the initial v21.12 bring up for now. Again, same checksum as the
previous tarball.
For edge we'll probably remove this package and instead Alpine's version
instead. [ci:skip-vercheck]
Upgrade u-boot to the latest in purism's librem5 branch, and use newer
DDR firmware (syncs version with what purism is using)
The patch that was dropped here was merged, so it's not necessary to
carry it anymore.
[ci:skip-build]: already built successfully in CI
Purism's g-c-c fork is stuck in version 3.38. This has
multiple problems, including carrying old unnecessary
patches and producing some incompatibility with newer
GNOME releases available in Alpine. In consequence,
we fetch the source from upstream and only apply those
patches that make sense for our usecase. When the Purism
patches apply without issues, then we fetch them directly
from their repo.
* The first patch has been droped from PureOS, which currently
ships stock debian 0.99.11. Therefore the fix must have been
applied somewhere else, same as it happened with the torch patches.
* The second patch is already merged upstream and present in latest
0.99.13 release in alpine.
With Phosh 0.13.0, Logind is used without patches for the torch
as UPower doesn't want to handle this. Therefore, we can drop
the torch patches to UPower.
[ci:skip-build] already built successfully in CI
I believe everything here has been merged in master, so let's build that
for now (since it includes at least 1 additional fix not in the patches
that were dropped here)
- switch sources to official repo https://source.denx.de/u-boot
- use tag v2021.07
- extract patches from pine64-org
- enable DMA transfers from eMMC and mSD (u-boot from Megi)
[ci:skip-build] already built successfully in CI
This branch includes a patch to resolve a deadlock when Phosh is
starting up. This patch was merged in aports edge.
The rest of the patches in this branch are minor fixes, so I don't
expect any regressions.
fixes#1142
[ci:skip-build] already built successfully in CI
Commits cfb5d41b and ead17254 updated the default battery levels for
warnings and actions in the pinephone and the pinetab. Since then,
the chances of the phone dying for lack of battery power have completely
dissapeared. However, the battery icon in phosh was never updated to show
a "caution" battery level. The value was hardcoded and ignored the setting
in the config file. This commit fixes that issue
* Prevent interactive warning to pop once already approved when
[over/down]clocking DRAM
* Make sure $freq is always set
Co-Authored-By: Oliver Smith <ollieparanoid@postmarketos.org>
This patch fixes:
fastboot: error: Couldn't parse partition size '0x'.
I don't get the bug myself, but it was reported that this works as
expected. Let's merge it to pmaports temporarily while it's getting
upstreamed.
Related: https://github.com/nmeum/android-tools/pull/27
Related: pmbootstrap issue 2007
[ci:skip-build]: already built successfully in CI
uboot-tools is provided by u-boot-tools but sometimes apk still gets
confused. This change prevents errors such as
pine64-pinephone:~$ sudo apk upgrade --verbose
WARNING: Failed to perform initial self-upgrade, continuing with full upgrade.
ERROR: unable to select packages:
uboot-tools (virtual):
provided by: u-boot-tools
required by: device-pine64-pinephone-0.25-r1[uboot-tools]
[ci:ignore-count]
[ci:skip-build]: already built in CI successfully
Latest patches from Purism, including some they backported from
upstream.
fixes#1058
The debian dir was dropped from the archives in this fork repo, but the
number of patches has been dramatically reduced.
A new mechanism (thanks @ollieparanoid) is added here to compare the
list of patches stored in pmaports when this package was upgraded to the
list of patches in purism's repo @ some commit this package uses, so
that any difference can be detected (pacakge will fail to build). If
there's a mismatch, the patches from this repo should be updated/added
to match what's in purism's repo @ the commit, and purism_patches.series
should be updated to match the debian/patches/series file in the purism
repo.
Fix build error:
CMake Error at /usr/lib/cmake/Qt5Svg/Qt5SvgConfig.cmake:111 (find_package):
Could not find a configuration file for package "Qt5Widgets" that is
compatible with requested version "5.15.3".
The following configuration files were considered but not accepted:
/usr/lib/cmake/Qt5Widgets/Qt5WidgetsConfig.cmake, version: 5.15.2
MR for Alpine:
https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/20307
[ci:skip-vercheck]: pkgrel>0 on purpose
still missing:
* audio
* lcd backlight control
* touchscreen driver
* front camera
* accelerometer mount fix (driver seem not to support mount-matrix)
* 3D acceleration
* HDMI support
(Also sync with Alpine while I'm at it...)
This is necessary for mobile data to work on MSM8916-based devices
after some changes in the network driver (bam-dmux) to operate in
Raw-IP mode instead of Ethernet mode (which is somewhat broken).
* Add u-boot build with a lot of patches that makes the display work in
u-boot for boot selection
* Upgrade the rockchip kernel to 5.11 mainline with config for the
rk3399 devices built-in
* Make the rockpro64 and pinebook pro use the newer kernel
[ci:skip-build]
This is 3.24.27 w/ Purism's mobile UI patches in their fork. According
to Purism, 3.24.27 resolves some stability issues (like a segfault when
scrolling)
[ci:skip-build]: already built successfully in CI
Changes from purism:
- 2 second boot delay removed
- This includes an updated DDR memory training firmware version.
- patch to work around wifi device not being available on boot
Package changes:
- u-boot installed in location consistent with other pmaports/u-boot
packages
- update-u-boot script added for upgrading u-boot on the device
- devkit u-boot split off into a sep. subpackage
Let pmbootstrap build the package in the native chroot. This is
important for the v21.03 branch, which needs to ship its own
cross-embedded toolchain, as the one from Alpine is only in testing and
therefore not available in the Alpine v3.13 branch that pmOS v21.03 is
based on. By using the native branch, it is enough to build the
cross-embedded toolchain for x86_64, and we can avoid qemu related
errors when attempting to build it for arm architectures.
As a nice benefit, this makes compilation of the package faster on edge,
too.
[ci:skip-build]: already built successfully in CI
The previous implementation of this caused upower to select a thing
called 'flash_strobe' in the L5, which doesn't operate like a
flashlight. This changes the detection logic to look at the suffix, on
the pinephone the device ends in ':flash', and on the L5 it ends in
':torch'.
fixed#967
This patch is in purism's fork and it removes messages like this when
starting upower on the librem5 by properly detecting/setting the type:
TI:15:11:12 did not recognise USB path /sys/devices/platform/soc@0/30800000.bus/30a20000.i2c/i2c-0/0-003f/power_supply/tps6598x-source-psy-0-003f, please report
...
TI:15:11:12 did not recognise USB path /sys/devices/platform/soc@0/30800000.bus/30a50000.i2c/i2c-3/3-006a/power_supply/bq25890-charger, please report
Handle USSD QMI indication messages.
Add support for UCS2 USS Data coding scheme.
Check for User Action TLV type.
Downstream the accepted ofono patch as temporary solution.
This has upstreamed fixes for call audio on the L5, but upstream hasn't
made a release yet. In the meantime, this is needed for calls to work on
5.11.0 and later kernels
[ci:skip-build]: won't finish in time
The TLVs are documented in GobiAPI. I pass 0xff for the call ID, as the
stock RIL appears to always do. I would guess it means "current foreground
call."
The call ID is returned in TLV 0x10, but I didn't implement parsing of
that.
Co-authored-by: Joey Hewitt <joey@joeyhewitt.com>
Acked-by: Alexey Andreyev <aa13q@ya.ru>
GNOME settings daemon notifies the user that the device will
suspend soon through a notification.
However, this triggers the notification LED which stays on during
suspend, causing unnecessary power consumption.
Plasma 5.20.90 was being built while a libgps upgrade happened. It seems
that even though the package had already been built for every arch, it
had not for x86 yet and thus our plasma-workspace is still depending on
the older package and fails to install because it isn't available
anymore
Just bump the pkgrel so it now gets correctly build against libgps.so.27
Seems there are more places where ecm_find_qmlmodule is used
To fix the build from failing on armv7 we should disable them all
[ci:skip-build]: don't try to build all depends, takes too long
[ci:skip-vercheck]: unblocking armv7 build, no need to bump pkgrel
The build is still failing, probably because ecm_find_qmlmodule is not
only present in the main CMakeLists.txt, but also in others.
$ git grep ecm_find_qmlmodule
CMakeLists.txt:ecm_find_qmlmodule(QtQuick 2.3)
CMakeLists.txt:ecm_find_qmlmodule(QtQuick.Controls 1.2)
CMakeLists.txt:ecm_find_qmlmodule(QtQuick.Layouts 1.3)
CMakeLists.txt:ecm_find_qmlmodule(QtQuick.Window 2.1)
CMakeLists.txt:ecm_find_qmlmodule(QtMultimedia 5.0)
CMakeLists.txt:ecm_find_qmlmodule(org.kde.kquickcontrolsaddons 2.0)
CMakeLists.txt:ecm_find_qmlmodule(org.kde.plasma.core 2.0)
CMakeLists.txt:ecm_find_qmlmodule(org.kde.plasma.components 2.0)
kcmkwin/kwindesktop/CMakeLists.txt:ecm_find_qmlmodule(org.kde.plasma.core 2.0)
kcmkwin/kwineffects/CMakeLists.txt:ecm_find_qmlmodule(org.kde.plasma.core 2.0)
[ci:skip-build]: don't try to build all depends, takes too long
[ci:skip-vercheck]: unblocking arm build, no need to bump pkgrel
Seems we missed these when forking Plasma and now the armv7 builders are
failing on it
[ci:skip-build] Won't succeed anyway as the packages haven't been
uploaded to the repos yet, so pmbootstrap will try to build the entirety
of Plasma and fail because it takes too long
[ci:skip-vercheck] No need to bump pkgrel
2.0.15 is a development 'version' (it'll never be released), and will be
replaced by 2.0.16 when that is released. The reason for forking this
is because there are some issues that are resolved here that prevent
SDL2 from working on the Librem 5. The fixes cannot be easily
backported to 2.0.14, hence the fork and upgrade.
fixes#950
[ci:skip-build] Never succeeds in time
[ci:skip-vercheck] We need our Mauikit to be a rel newer than in Alpine
repos, but the CI doesn't like it
This includes a big rewrite in kwin which should increase the
performance a whole lot, and some awesome other stuff
Fix about dialogs not being adaptive and all other changes by Purism,
which are the reason why we use the fork in the first place. The patches
didn't get applied in the previous version we had packaged in
postmarketOS, because the patches are in debian/patches now.
Remove check-version.py, it's in the source tree now.
bpo is currently failing to build images with the on-device installer,
because calamares needs to be rebuilt against libboost_python38.so. Fork
it until that's resolved. [ci:skip-vercheck]
[ci:skip-build]: already built successfully in CI
Related: https://builds.sr.ht/~postmarketos/job/398948#task-img_installer-307
This syncs u-boot with the upstream branch, and includes the following
new features:
- support for board rev in u-boot
- fix uart4
[ci:skip-build]: already built successfully in CI
This packages the libhandy work in dino's feature/handy branch, which
makes the UI quite usable on mobile displays. A couple of windows don't
work well yet (e.g., the setting window), but chatting/omemo/file
transfer all work pretty well.
Currently phosh is broken in postmarketOS edge. Fork wlroots from Alpine
and revert the new consistency check that results in the breakage. I've
submitted the same to Alpine, but let's get it in pmOS now so it is
fixed ASAP.
Alpine MR: https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/14522
[ci:skip-vercheck]: new package added with -r1 on purpose
plasma-nano (albeit saying it is 5.20.1) is pointing to the master branch
and kept as-is (there is no updates in master and I don't see a reason to
downgrade it to the actual 5.20.2 tag which is a few small commits behind
the master).
Remove patches:
'0001-gtk-meson.build-add-new-hdy-files.patch' and
'10-Revert-gdkseatdefault-Grab-touch-events-where-applic.patch'
appear in the code downloaded from Purism at precisely the
location of the patches.
[ci:skip-build]: already built successfully in CI
For some reason, the ModemManager build tends to freeze when built
with QEMU user emulation for arm*. Changing the build to use a single
thread only (-j1) avoids that, although the build is slower of course.
Also limit building to "armhf armv7 aarch64" since the forks are not
needed on any other architectures, to reduce build times a bit. The
other architectures can just use the upstream packages from Alpine.
This allows to connect the modem to the Internet with oFono.
I have verified that this does not break anything if the "rmnet0"
network interface is missing. Plus, all mainline devices currently
covered by the package should also be able to use the new "BAM DMUX"
network driver that is used as network interface to the modem.
(Note: This works differently on newer SoCs, but they also need
something different in oFono...)
