We have two methods of cross-compiling:
* native: everything runs with the host architecture, QEMU is not
involved. This is the fastest, but requires the build system to be
working with it. We use this for all linux-* packages currently.
* distcc: everything runs through QEMU emulating the target arch,
*except* for the compiler. This is the most compatible approach
working with all packages.
When compiling `linux-*` packages natively, kernel scripts needed
during the build process get generated. Some of these are C files that
get compiled as executables. In native mode, these get compied to the
native architecture, in distcc mode to the target architecture.
The problem is, that we need these scripts compiled for the target
architecture in the kernel's dev package in order to compile kernel
modules outside of the kernel's package (e.g. wireguard).
It is not possible to just rewrite this logic to generate target-arch
binaries when running in native mode, because these binaries require
musl-dev, linux-headers and some other packages to be installed for the
target architecture inside the native chroot.
We solve this by introducing a new `kernel-scripts` package. which
contains just the binary scripts. In case the dev package was
cross-compiled, it depends on `kernel-scripts` and symlinks these
binaries. The `kernel-scripts` package always gets compiled in distcc
mode since it does not have a `linux-` prefix.
Fixes#1230.
Fixes#1227.
This also updates the hash in linux-postmarketos-mainline, because the
hash changed upstream (they updated their git version server-side?).
This port includes:
* Weston including touch and WiFi (using e.g. firmware-samsung-i9305) works.
* Touchscreen calibration for osk-sdl
* Sound configuration
* Udev rules for touch and pen, pen works!
Closes#441. Adjust bootimg_analyze code:
* Install mkbootimg (which now provides unpackbootimg) instead of
unpackbootimg. In theory, pmbootstrap should recognize this
automatically, however right now it does not yet handle this case.
* The file names of the extracted files have changed.
* Automatically generate a calibration matrix for libinput
This takes the calibration matrix for wayland and divides the pixel
offsets by the device width/height.
* Bump pkgrels of devices using devicepkg-dev
This causes new packages to be generated, using the new devicepkg-dev
version.
Right now, they appear on screen when using --details-to-stdout. This
does not work well with Travis CI and screws up the log.
Disabling the progress bars in abuild works just like Alpine does it in
their Travis CI script: Exporting SUDO_APK as
"abuild-apk --no-progress" instead of "abuild-apk".
test_check_checksums.py: Run "pmbootstrap build_init" before building
any packages, so it is a bit less verbose (downloading the APKINDEX
files etc.). Later we run the build init code again (because we use
--strict while building the packages), but then the APKINDEX files
are already present. So overall the log is a bit shorter before the
building starts. (It is still logged to the logfile, which gets
printed on error anyway.)
The device is Xiaomi Redmi 1S, based on MSM8296 platform, with Qualcomm
Snapdragon 400. The kernel is from lineageOS (3.4.0). Some additional patches
are needed for successful compilation. The screen dimension is 720x1280.
The touchscreen device name is ft5x06_i2c. Use this name for udev
instead of hardcoded device path. Also add a workaround for red
screen weston by setting weston_pixman to 2
The screen doesn't refresh. Enabling msm-fb-refresher as suggested.
Add udev rule to indicate whether the device is in charging state.
Green led indicates that the device is charging, whereas red light
indicates that the device is discharging.
For prima wlan module to work, an initialization procedure is needed (
echo 1 > /dev/wcnss_wlan ), before loading the prima wlan module
(wlan.ko). Also build the prima wlan as module.
Xiaomi Redmi 1S is using pronto wlan module, which needs some firmware
to run. Add the required wifi firmware to enable the wifi interface.
Add an openrc service to load wlan module after wcnss_service is
initialized.
* Testsuite: Run UIs in Qemu and check running processes (and other changes)
* When `pmbootstrap qemu` gets killed, it now takes down the Qemu process with it
* `test/check_checksums.py` got a new optional `--build` parameter, which makes
it build all changed packages instead of just checking the checksums
* We run this before running the testsuite now, so all changed packages get
built before running tests (otherwise tests would hang without any output
while a changed package is building)
* New testcase, that zaps all chroots, installs a specific UI (xfce4 and
plasma-mobile currently, easy to extend), runs it via Qemu and checks the
running processes via SSH.
* Version checking testcase: rewritten to include Alpine's testsuite file in
our source tree, so we don't need to clone their git repo anymore. Now it
is enabled for Travis.
* All this gives us a nice 10% code coverage boost
* Increased the `hello-world` pkgrel to verify that the Travis job is working.
* Various fixes
* Build device-packages for the device arch and don't raise an
exception, but print a note if --ignore-depends is not specified
and therefore the kernel gets installed, too.
* Don't use --force when building in Travis (because abuild doesn't
check the checksums then. Bug report on the way.)
* Don't run the building process in the background, but wait for its
completion
* Exit with 1 when showing usage in check_checksums.py
It used to have an entry for /mnt/pmbootstrap-packages, which only
makes sense while working on the chroot with pmbootstrap. After the
installation on the device, there's no repo in that path.
Furthermore, empty lines were added to the recovery installer script
for readability (thanks @ata2001!)
Follow up to #1162.
* `pmb.build.buildinfo()`: Used to record the build environment. It is
flawed because it scans the repo APKINDEX files instead of using the
actually installed packages list. When it was implemented we were not
able to do the latter. After this is removed, `pmb.parse.depends` can
be simplified (it needs to be rewritten for #1122).
* `pmb.helpers.repo.diff()` and `pmb.helpers.repo.files()`: These were
used exclusively by `pmb.build.buildinfo()`, to learn about which
files have been changed in the local repository folder after a
package was built. The idea was, that we could find subpackages that
way. But this information is present in the installed package list as
well, which is a much cleaner approach.
Nowadays pmb.config.build_device_architectures holds the native
architecture as well, so we don't need to explicitly download the
native architecture APKINDEX.
* Fail if mkbootimg/uboot-tools are not installed, but creating a
boot.img file / u-boot legacy image was requested via deviceinfo
(fixes#312)
* Fail if /boot/dt.img is missing, but we have a qcdt device
* Fail if the dtb file specified in deviceinfo does not exist
* Fail if mkbootimg etc. exit with error code
* Don't try to add the ext4 module into the initramfs. We always
compile it into the kernel. Instead, kconfig_check makes sure it
is enabled now. (fixes#1037)
* Add a note that modprobe warnings can be ignored mostly
zap -m:
* APKINDEX parsing: parse the "origin" field as well, so we know
where a subpackage comes from
* pmbootstrap zap -m: properly delete all packages, that do not
have an aport or where the aport has another version. This also
works with subpackages now,
we use the origin field to resolve it.
* Only reindex when packages have been deleted in "zap -m"
zap in general:
* Show the amount of cleared up space after the deletion instead
of "Done"
* Print "Shutdown complete" to "pmbootstrap log" instead of stdout
(we need to call it twice during zap now to get the space
calculation right)
* Add `--dry` argument to `pmbootstrap zap` (this was very useful
for debugging) to list the packages/chroots that would get
deleted
* Roughly output the command that would get executed to delete
files, so it's obvious what's going on in --dry mode. (% rm ...)
Current status:
- flashing only manually (intel phones have different bootimg format)
- usb networking works
- xorg works using fbdev
- packages needed for wayland are not available for x86
- touch works in xorg and osk-sdl
If you want to build a package without changing the version number,
please use `--force` from now on. For example:
pmbootstrap build --force hello-world
Prior to this commit, changes were detected automatically (timestamp
based rebuilds). However, that feature does not work as expected with
the binary package repository we have now, and depending on how you use
git, it has never worked. Close#1167, close#1156, close#1023 and
close#985. This commit also mentions --force when a package is up to date,
but the user requested to build it.
Preparation for #1122.
* `pmb.parse.apkindex.parse()`, removed strict parameter: This used to raise
an exception when two entries in the apkindex provided the same package.
Turns out this is *not* invalid after all, two packages can provide the same
soname for example (e.g. libhybris, mesa-egl). In an APKINDEX, sonames are
listed as they were packages ("so:libjpeg.so.8" etc.).
* Remove `pmbootstrap challenge` leftover code from reproducible builds effort,
which was a dead end. This code uses the broken strict feature.
The message showed up, when you apk-static download could not be
verified. What the user needs to do instead is checking if openssl
is installed, and possibly delete the http cache ("zap -hc").
Oliver Smith
Pierre Rudloff
Michael Hamann
Lucas Ramage
alive4ever
Pavel Machek
Bart Ribbers
Daniele Debernardi
ashirviskas
Emil Lauridsen
Alex Roth
Peter Mann
Alfie Day
Ryan Murthick
NotKit