With the nokia N900 the user can choose between different keyboard
layouts and during the install the setup-keymap installs the choosen
layout.
This only changes the keyboard layout of the console, while installing
a UI using xorg, the layout was hardcoded to "us".
With a simple check for the existence of an xorg config containing
XkbLayout, this will run sed to replace the default with the layout
chosen by the user.
Tested with a clean install of the nokia N900 with the xfce4 UI.
elogind requires cgroup support in the kernel.
It does not start without it, resulting into errors like:
elogind-daemon[1654]: Failed to mount tmpfs at /sys/fs/cgroup: No such file or directory
elogind-daemon[1654]: Failed to mount cgroup at /sys/fs/cgroup/elogind: No such file or directory
elogind-daemon[1654]: Failed to allocate manager object: No such file or directory
dbus-daemon[1437]: [system] Activating service name='org.freedesktop.login1' requested by ':1.1' (uid=0 pid=1642 comm="/usr/bin/lightdm ") (using servicehelper)
dbus-daemon[1437]: [system] Failed to activate service 'org.freedesktop.login1': timed out (service_start_timeout=25000ms)
Most of the supported desktop environments (e.g. Weston) do not
start without elogind, so we should require CONFIG_CGROUPS=y in
all kernels to avoid this problem.
Make /home/pmos/build/.git point to the .git dir from pmaports.git, with
a symlink so abuild does not fail.
abuild expects the current working directory to be a subdirectory of a
cloned git repository (e.g. main/openrc from aports.git). If git is
installed, it will try to get the last git commit from that repository,
and place it in the resulting apk (.PKGINFO) as well as use the date
from that commit as SOURCE_DATE_EPOCH (for reproducible builds).
With that symlink, we actually make it use the last git commit from
pmaports.git for SOURCE_DATE_EPOCH and have that in the resulting apk's
.PKGINFO.
Fixes: #1841
When 'pmbootstrap build' is started with '--no-depends', we make sure
that binary packages of all dependencies exist before proceeding with a
build. Make sure that we have downloaded the APKINDEX for the given arch
first, otherwise this may complain that a binary package is missing
although it does exist.
This happened when using the "native" cross compile method, e.g. when
building kernels for different architectures. It would complain that
there is no binary package for "bash" (which is in Alpine obviously):
https://builds.sr.ht/~postmarketos/job/103882
When running pmbootstrap init, first select device vendor, then device
codename. Also fixed tests for new behavior and added some new ones for
new scenarios.
postmarketOS/pmaports!700 adds a new "deviceinfo_bootimg_dtb_second"
option that places the DTB in the "second" area of the Android boot
image.
Attempt to detect this automatically by checking the extracted
second binary for the FDT magic (0xd00dfeed).
Packages ending in -dev: just assume that the originating aport has the
same pkgname, except for the -dev at the end. Otherwise we may end up
with the wrong package.
For example, if something depends on plasma-framework-dev, and
plasma-framework is in Alpine, but plasma is in pmaports, then the
regular guess_main() algorithm below would pick plasma instead of
plasma-framework.
Fixes: build.postmarketos.org#52
mkbootimg from osm0sis changed the filename of the dtb section in [1]
which breaks the detection of qcdt in pmbootstrap.
Fix this by adjusting the filename pmbootstrap checks to match the
unpackbootimg behavior.
[1] https://github.com/osm0sis/mkbootimg/commit/d8222a4d
Currently, building with envkernel.sh entirely ignores the
makedepends listed in the kernel APKBUILD. Common dependencies
needed by most kernels are hardcoded in envkernel.sh.
However, some kernels may need extra dependencies either during
the build process or when the kernel is packaged.
Installing the makedepends when the build environment is initialized
is difficult because "source helpers/envkernel.sh" is not aware
of the exact kernel package that is going to be built later.
However, we can easily modify the packaging step
(i.e. pmbootstrap build --envkernel linux-...)
to install the required makedepends.
This fixes building "linux-postmarketos-qcom-msm8916" using envkernel,
since it requires "dtbTool" and "installkernel" to be installed when
the APKBUILD package() function is executed.
The sparse image tools (simg2img, img2simg, append2simg) are now
part of Alpine's android-tools where they are continously updated.
(See https://github.com/nmeum/android-tools/pull/8)
Therefore, "libsparse" now conflicts with "android-tools", which
causes the fastboot flasher to fail if "libsparse" is already
installed.
Install "android-tools" instead of "libsparse" before generating
a sparse image to avoid this problem.
Make it more obvious that we are generating packages during
'pmbootstrap init', when using the new device wizard:
[22:32:32] *** pmaport generated: /home/user/.local/var/pmbootstrap/cache_git/pmaports/device/device-openphoenux-neo900
[22:32:32] *** pmaport generated: /home/user/.local/var/pmbootstrap/cache_git/pmaports/device/linux-openphoenux-neo900
This file is used by the package browser frontend build by the alpine
developers. It uses the contents of the DESCRIPTION file to validate the
cache in the database.
Remove the recursive check, as it caused an infinite loop. It took a
very long time to complete anyway, even when it worked. The reasoning
for having the recursive check in the first place was, that we would
have device packages with arch=noarch set in the APKBUILD, but which
could only be built for a certain architecture (the device
architecture). Nowadays using arch=noarch in device packages is
forbidden, and we enforce that rule [1]. So the recursive arch check
isn't necessary anymore.
[1] ac6c0a2997
This reverts commit 6fb5b28e2f.
The -i option was removed from fastboot, so we can't use it anymore
unless we fork the package. There was only one device using it,
amazon-thor. I will add a note to the wiki page.
Fixes#1830.
Linking big libraries may take more than five minutes, so change the
default timeout to fifteen minutes to save users from frustration.
This is the same value that I'm using to build binary packages for the
repository.
Pushing directly to master without making a merge request, because this
is a trivial change.
Make sure, that "args.fork_alpine" is always present. Otherwise,
pmbootstrap will fail if the aportgen code is called by anything but
"pmbootstrap aportgen". For example, when the user is adding a new
device during "pmbootstrap init".
Fixes: 54e51759ad ("aportgen: add feature to fork upstream packages")
The important part about this patch is the change for armhf, which
adjusts the hostspec to the one used by Alpine.
Fixes a part of postmarketOS/pmaports#363
This MR add the --fork-alpine argument to the pmboostrap aportgen
command, which downloads the APKBUILD and related files and copies them
into the pmaports/temp folder.
At the moment, attempting to install to SD card when
flash_sparse is set to "true" will always fail because
/home/pmos/rootfs does not exist in that case.
Sparse images are only useful to speed up Fastboot flashing.
Nothing will be able to read it from the SD card.
The problem can therefore be avoided by simply skipping the
generation of the sparse image when SD card installation is used.
Remove u-boot from the native_cross_compile list as we now have
crossdirect. Compiling is not terribly slow anymore, and when not using
the "native" method, it is possible to depend on firmware packages like
arm-trusted-firmware-sun50i.
This reverts commit d7d7ccb672.
This allows for example for me to call the kconfig check function on the
.config file in my Linux tree: $ pmbootstrap kconfig check --file
.config and it reports me which kconfig options I need to enable.
From https://docs.python.org/3/library/re.html
"Backslashes are not handled in any special way
in a string literal prefixed with 'r'."
Signed-off-by: Steffen Pankratz <kratz00@gmx.de>
Properly handle the following two cases in APKBUILDs:
* depends="$depends ..."
* depends="${depends} ..."
First I've attempted to refactor the parsing code to do this in a more
generic way. But I've realized that it would make more sense to retire
the python based APKBUILD parsing approach altogether and finally use a
shell script parser. Let's discuss this in #1801.
Add check for UEVENT_HELPER for kernels >= 4.0.0. Change LBDAF check to
be only required for kernels <5.2.0. Adjust the config format and
checking code to support such range specific checks.
Compare a version against a check string. This will be used in
"pmbootstrap kconfig check", to only require certain options if the
pkgver is in a specified range.
This fixes regression from commits 0431a519 and 4daf9916.
pmbootstrap kconfig_edit raised an error while trying to checksum a package.
Error was:
(028793) [16:13:22] ERROR: 'module' object is not callable
(028793) [16:13:22] See also: <https://postmarketos.org/troubleshooting>
(028793) [16:13:22] Traceback (most recent call last):
File "/home/lexx/dev/pmos/pmbootstrap/pmb/__init__.py", line 63, in main
getattr(frontend, args.action)(args)
File "/home/lexx/dev/pmos/pmbootstrap/pmb/helpers/frontend.py", line 279, in kconfig
pmb.build.menuconfig(args, args.package)
File "/home/lexx/dev/pmos/pmbootstrap/pmb/build/menuconfig.py", line 158, in menuconfig
pmb.build.checksum(args, pkgname)
TypeError: 'module' object is not callable
The function was renamed, call new function.
(Some?) Mediatek kernels don't have the kernel in the top-level
directory but in a subdirectory called 'kernel' next to a folder
'mediatek' containing most device-specific code.
Adjust the menuconfig code to be able to work with that.
Download all sources and verify their checksums.
This will be used in pmaports.git CI, if ci:skip-build is set in the
commit message (currently it just skips the build, and we don't test if
the source checksums are valid or not).
Fix configure crash when building mozjs60 by mounting tmpfs as /dev/shm
in chroots. This is an important fix, because of this dependency chain,
that is currently broken in armhf and armv7 (see pmaports#244):
postmarketos-base -> networkmanager -> polkit -> mozjs60
Explicitly depend on mpc1. Our generated gcc aports use the !tracedeps
option, so we need to explicitly set the libraries it depends on.
This has mostly not been an issue, as we are installing our gcc
packages together with Alpine's gcc package, which causes the libraries
to get installed anyway.
Related: pmaports#236
abuild depends on the gcc binary in order to define a default CBUILD value.
When using an alternative gcc version (e.g. envkernel.sh with gcc6), the
gcc binary is not installed by the envkernel.sh script.
This change sets the CBUILD env variable so that abuild doesn't need to
depend on the gcc package.
Do not crash during "pmbootstrap zap" and other actions when the user
needs to type in the sudo password before running "du" to calculate the
work folder size. We're trying to parse the "du" output as integer, but
of course the 'Sorry, try again' message from sudo is not a valid
integer.
$ pmbootstrap --details-to-stdout zap
(008707) [14:00:57] Shutdown complete
(008707) [14:00:57] Calculate work folder size
(008707) [14:00:57] % sudo du --summarize --block-size=1 /home/luca/.local/var/pmbootstrap
[sudo] password for luca:
[sudo] password for luca: Sorry, try again.
12966293504 /home/luca/.local/var/pmbootstrap
(008707) [14:01:03] ERROR: invalid literal for int() with base 10: 'Sorry, try again.\n12966293504'
It would be a good idea to separate the sudo output from the process
output in general, see #1677 for a related proposal.
When the pmaport arch does not match, but there's a matching binary
package, return the binary package instead of the pmaport.
This is needed to fix "pmbootstrap repo_missing" for x86_64
networkmanager-qt, which indirectly depends on mesa (and mesa needs to
be taken from the binary packages for x86_64, not from the pmaports,
see #1741).
Split the part that actually checks the arch against the arches list
into pmb.helpers.pmaports.check_arches(arches, arch), and call it
from pmb.helpers.package.check_arch(args, pkgname, arch, binary).
This will be used in a follow up commit, where we have already resolved
the package data and only need to check the architecture.
Packages can add pmb:strict to their options to enable the --strict
mode. This can be used if packages need to get build in a clean chroot
or their make dependencies need to get removed.
Depend on new pmaports version, where crossdirect uses the native ccache
binary instead of going through the foreign arch ccache first and then
going through crossdirect.
Old:
ccache (foreign) -> crossdirect (native) -> gcc (native)
New:
crossdirect (native) -> ccache (native) -> gcc (native)
Set the PATH to the crossdirect binaries, and don't set CCACHE_PATH or
CCACHE_COMPILERCHECK from pmbootstrap anymore. crossdirect sets the
CCACHE_PATH to /native/usr/bin now, along with all other required
environment variables. CCACHE_COMPILERCHECK isn't necessary anymore,
because ccache will call gcc directly and therefore be able to use the
file's timestamp and size directly. Also passing that would not work
with the current crossdirect package.
Launch native cross compilers inside foreign chroot. Enable by default,
but allow disabling with --no-crossdirect for now. This option and the
distcc-sshd related code will be removed in the future.
Aborts the build if any dependencies would have to be build first. This
is useful for build.postmarketos.org, because we want to build exactly
one package in one build job. If dependencies would need to be built, we
made a mistake earlier, and not aborting the build makes it harder to
find that orginal mistake.
When a pmaport can not be built for the desired architecture, fall back
to the binary package (from postmarketOS or Alpine) if it exists. This
allows us to provide an updated version of mesa for arm arches, but
using Alpine's mesa package on x86* arches.
Follow up to last commit, turns out we need to use it in both functions
or otherwise at least gitlab-ci can fail, as seen in this MR:
https://gitlab.com/postmarketOS/pmaports/merge_requests/270
After using 'busybox tar' instead of 'tar' in both places in the
APKBUILDs, it went through.
Use 'busybox tar' to avoid 'tar: Child returned status 141' on some
machines (builds.sr.ht, gitlab-ci).
See pmaports#26.
[skip ci]: already ran through successfully
This adds flasher support for uuu, a utility used by NXP devices for
flashing images.
The flasher expects a uuu command list script to be installed in the
device rootfs at /usr/share/uuu/flash_script.lst.
This allows building u-boot much faster, previously it used to build
whole u-boot with distcc+qemu cross-compile, now it builds u-boot on
native chroot using cross-compilers.
In pmb.helpers.repo.alpine_apkindex_path(), default to
arch = args.arch_native *before* calling pmb.helpers.repo.update().
Because otherwise update() defaults to all arches instead of
args.arch_native.
This caused the APKINDEX files for all arches to get downloaded, as
pmb.chroot.apk_static.init() calls alpine_apkindex_path() without
setting arch = args.arch_native explicitly.
Related to a proprietary "T-Mobile's WiFi-Calling" feature. It should
not affect normal calls, so it should be safe to disable it. Having it
enabled causes sudo, ip and other commands to hang.
It is dangerous to flash the boot partition on the Ouya. Doing so can
result in a bricked device. Allow deviceinfo to blacklist partitions from
ever being flashed.
Ignore APKBUILDs that have "!pmb:kconfigcheck" in their options by
default in "pmbootstrap kconfig check", but print a note that they have
been skipped. Check all kernels with "pmbootstrap kconfig check -f".
This is necessary, because the Librem 5 devboard kernel's config does
not have CONFIG_DM_CRYPT enabled in their config, and we check for
that. As the device is still under heavy development, we will make our
lives easier by just using the upstream kernel config without any
changes and ignoring it in our check by default.
"breeze-icons" depends on "qt5-qtbase-dev", but
"pmbootstrap repo_missing" should return "qt5-qtbase" instead.
This patch fixes it, as one can see with:
$ pmbootstrap repo_missing --built breeze-icons --overview
Provides a quick way to incrementally compile a kernel and push it to
device.
Example usage.
Compile the kernel:
$ cd /src/linux/
$ source /src/pmbootstrap/helpers/envkernel.sh
$ make tegra_postmarketos_defconfig
$ make -jX
Package kernel and flash to device:
$ pmbootstrap build --envkernel linux-samsung-p4wifi
$ pmbootstrap flasher flash_kernel
Modify kernel source then incremental compile, package, and flash:
$ make -jX
$ pmbootstrap build --envkernel linux-samsung-p4wifi
$ pmbootstrap flasher flash_kernel
Set the log level to DEBUG if args (and therefore logging) failed to
initialize. This causes the default logging handler to be used, and
everything that gets written to the log (error message and trace!) will
get printed to stdout instead of not being printed anywhere at all.
This can be tested by removing x86_64 from the mapping in
pmb.parse.arch.alpine_native() (assuming the host is x86_64). Without
this patch, no meaningful error appears. With the patch, the real error
message, and a stack trace gets printed.
Test with flake8: *.py
./test/check_checksums.py:13:13: E117 over-indented
./pmb/config/init.py:97:8: F632 use ==/!= to compare str, bytes, and int literals
./pmb/parse/arguments.py:229:13: E117 over-indented
Increase pakage version and minimum required apk version. The latter
should be done with every new release from now on, so we always have a
recent version there. I've updated the pmbootstrep release checklist
accordingly: https://wiki.postmarketos.org/wiki/Pmbootstrap_release
This drops the --apparent-size parameter when calculating the size
required for rootfs, which seemed to return a size that was too small
for some devices. This also includes specifying the number of inodes to
support when formatting rootfs.
Fixes#1717
Allow changing the kernel partition for fastboot and heimdall in
deviceinfo and on the fly while doing "pmbootstrap flasher
flash_kernel". Also allow changing the partition for
"... flash_rootfs" with fastboot (this was only possible with
heimdall so far).
Introduce two new deviceinfo variables:
* flash_fastboot_partition_kernel
* flash_fastboot_partition_system
This is useful for devices with dual partitioning that have boot_a
and boot_b.
This allows embedding multiple firmware binaries into SD images.
Firmware images for most devices that require this functionality consist
mainly of u-boot, but some devices (e.g. librem5) have multiple firmware
images that need to be embedded in the SD image created by pmbootstrap.
This functionality uses two new deviceinfo parameters:
- deviceinfo_sd_embed_firmware: a comma-separated list of
binary:offset (where binary is under /usr/share/firmware)
- deviceinfo_sd_embed_firmware_step_size: The number of bytes for
each increment of the offset specified in the
deviceinfo_sd_embed_firmware (typically 1024 or 2048)
deviceinfo_write_uboot_spl has been obsoleted by these new parameters.
Rename pmb.helpers.devices.list() to
pmb.helpers.devices.list_codenames(). Python already has a list()
function, so we name our function to make the calls to the codenames
listing function inside pmb/helpers/devices.py less confusing.
This fixes list_apkbuilds which is dead code currently but is useful
for external software using pmbootstrap. It also adds list_deviceinfos
which does the same thing but for the deviceinfo data.
The plan is to add an API later, where both functions can be used.
APKBUILD should be indented with tabs not spaces.
This also resolves a bug which the replace_functions feature was always
trimming off the beginning of a line.
For example:
replace_functions = {"build": "return 0"}
would be formatted as "rn 0".
Python < 3.6 randomized the order of keys in dictionaries, unless
OrderedDict was used. Use OrderedDict to store the version suffixes.
When the order was randomized, the valid version string 3.0.0_pre1 did
not always pass the validation check. The suffix "pre" should always be
detected as such, but with the random order, it was sometimes detected
as "p" suffix (see below). The following letters "re" are not a valid
suffix_no (the number expected to follow the suffix) and so it failed.
suffixes = {
"pre": ["pre", ...],
"post": ["p". ...]
}
Display the resolved deviceinfo file path in exceptions from the
deviceinfo parser. Instead of messages like these:
RuntimeError: Please add 'deviceinfo_codename="lg-mako"' to: /home/user/.local/var/pmbootstrap/cache_git/pmaports/.gitlab-ci/testcases/../../device/device-lg-mako/deviceinfo
We get the more readable version:
RuntimeError: Please add 'deviceinfo_codename="lg-mako"' to: /home/user/.local/var/pmbootstrap/cache_git/pmaports/device/device-lg-mako/deviceinfo
deviceinfo_codneame holds the device's code name, so we can easily look
it up in the finished postmarketOS installation by reading
/etc/deviceinfo.
Related: postmarketOS/pmaports#157
Properly replace ~ with $HOME in all paths. Fix tab completion when
passing a path starting with ~ to pmaports:
$ pmbootstrap --aports ~/src/pmaports/ build linux-<TAB>
Make tab completion work again for packages, if the default pmaports
dir is used. In the last "args" code refactor, the variable replacing
code was moved into its own function. We did not call it in the
packagecompleter() yet, so it could not replace the $WORK variable in
the default pmaports path.
Show all config keys that can be queried and set in the
'pmbootstrap config -h' output and make tab completion work for the
key names.
I've set "metavar" and placed the variables in the helpstring. That
way, argparse will not generate a huge "positional arguments" string
that blows up the layout of the help output:
[{ccache_size,device,extra_packages,hostname,jobs,kernel,keymap,...
Make it possible to point the pmaports dir to an existing repository
clone, while not having to give the --aports/-p parameter with every
command.
pmbootstrap --aports=/my/pmaports/path init
If the config file exists already, the dir can also be set with:
pmbootstrap config aports /my/pmaports/path
Adds an optional deviceinfo variable, `deviceinfo_rootfs_image_sector_size`,
which specifies the logical sector size of the device's storage.
Some devices made after 2016 with UFS storage uses 4096 byte sectors
instead of the normal 512 bytes. The partition table in our rootfs
must match, otherwise the root filesystem won't mount on the device.
This change passes the sector size to `losetup` when creating the image
if the deviceinfo specifies it, so the image will have the correct
sector size.
If the deviceinfo doesn't specify the new option, the behaviour
is the same as previous versions of pmbootstrap.
Note that the sector size option only works on Linux 4.14 and above,
so pmbootstrap should be run on a >4.14 computer when installing to
devices with non-standard sector size.
To find if a device needs this parameter, run `fdisk -l` on the device.
If the output shows
`Note: sector size is 4096 (not 512)`
then add `deviceinfo_rootfs_image_sector_size="4096"` to the deviceinfo.
This is needed by the Pixel 3 XL (google-crosshatch) port.
See https://gitlab.com/postmarketOS/pmbootstrap/issues/1696.
Do not fail when an APKINDEX can not be downloaded, print a WARNING
instead. This matches, what apk does. Add a new allow_404 parameter to
pmb.helpers.http.download(), and use it in pmb.helpers.repo.update()
for downloading APKINDEX files. Cache the APKINDEX URLs that gave a 404
for the session, so we do not attempt to download these again.
This is needed for the new binary repository: the initial build is done
without existing APKINDEX files, so we must not fail in that case.
Change the cache format from args.cache["offline_msg_shown"] to
args.cache["pmb.helpers.repo.update"]["offline_msg_shown"]. This is in
preparation for saving more data in the cache of
pmb.helpers.repo.update in the next commit.
Do not fail anymore when attempting to start a new binary repository
build without any existing binary packages:
pmbootstrap -mp="" repo_missing
Find subpackages defined with subpackage functions:
subpackages="dev:mydevfunc"
Find provides defined with specific versions:
provides="mkbootimg=1.0.0"
Weston and Plasma Mobile might choose to draw on this virtual frame
buffer, just like it happened for the samsung-jflte. Disabling the
USE_VFB option fixed this, so let's make sure we don't have that option
enabled for any kernel.
Multiple -mp arguments can be used to list multiple mirrors:
$ pmbootstrap -mp=first -mp=second chroot -- cat /etc/apk/repositories
This is needed for the new build infrastructure, so we can have a WIP
repository to which we push packages until all of them are up to date,
and then publish all of them at once. Software like KDE/Plasma Mobile,
which expect a lot of packages to be updated from one version to
another will not end up with a half-way through upgrade that way.
Obscure feature: it was possible to specify a local path as
--mirror-pmOS. It would then get mounted to /mnt/postmarketos-mirror
inside the chroot, and be specified as such in the generated
/etc/apk/repositories file.
I had used this once for some testing scripts, but I am sure nobody is
using this anymore. The same can be achieved with running a local http
server anyway:
<https://wiki.postmarketos.org/wiki/Installing_packages_on_a_running_phone>
Removing this makes it easier to support multiple postmarketOS mirrors
(next commit).
-m is the Alpine mirror, -mp is the postmarketOS mirror. Use "URL" as
metavar and add help text that explains how to disable the postmarketOS
mirror (so all pmaports get built locally).
In pmb.helpers.package.get(), we are differentiating between packages
that do not exist at all, and packages that do not exist for the
specified architecture. Make sure to actuallyy download the foreign arch
APKINDEX files, before trying to find the packages in there. Otherwise
the "could not find aport, and could not find this package in any
APKINDEX" package would appear.
We ran into this when testing on sr.ht, because pmbootstrap runs on a
fresh install every time, where no old APKINDEX files are present
(which would work around the bug).
This was meant to be part of the previous commit already, and was
tested, but it was not submitted correctly. So here it is as separate
commit straight to master. It is trivial anyway.
Add a new action that lists all aports, for which no binary packages
exist. Only list packages that can be built for the relevant arch
(specified with --arch). This works recursively: when a package can be
built for a certain arch, but one of its dependencies
(or their depends) can not be built for that arch, then don't list it.
This action will be used for the new sr.ht based build infrastructure,
to figure out which packages need to be built ahead of time (so we can
trigger each of them as single build job). Determining the order of the
packages to be built is not determined with pmbootstrap, the serverside
code of build.postmarketos.org takes care of that.
For testing purposes, a single package can also be specified and the
action will list if it can be built for that arch with its
dependencies, and what needs to be built exactly.
Add pmb/helpers/package.py to hold functions that work on both pmaports
and (binary package) repos - in contrary to the existing
pmb/helpers/pmaports.py (see previous commit) and pmb/helpers/repo.py,
which only work with one of those.
Refactoring:
* pmb/helpers/pmaports.py: add a get_list() function, which lists all
aports and use it instead of writing the same glob loop over and over
* add pmb.helpers.pmaports.get(), which finds an APKBUILD and parses it
in one step.
* rename pmb.build._package.check_arch to ...check_arch_abort to
distinguish it from the other check_arch function
Move find_aport() and find_aport_guess_main() from pmb/build/other.py
to the new file pmb/helpers/pmaports.py.
Finding aports is not only needed when building packages, hence it
makes sense to move it out of pmb.build. The pmb/helpers/pmaports.py
file will have more pmaports related functions in a follow up commit.
Target arch is the system the package will run on. Host arch is the
system the package is compiled on.
For example kernel packages can be compiled on host arch x86_64 and
intended to run on target arch armhf. A build is necessary check
against host arch will always return True. The correct way is to check if
the package needs to be built for target arch.
Do not try to build and install dependencies for the package's
architecture when compiling in the "native" mode. That mode is
described here in detail:
https://wiki.postmarketos.org/wiki/Build_internals#Cross-compile_types
This makes it possible to cross compile kernels again, which need to
be built with GCC6. We have switched to Alpine's GCC6 package, but it is
not available for armhf/aarch64 on edge yet, because Alpine's build bots
are stuck (right now armhf is not even listed):
http://build.alpinelinux.org/
Huge thanks to ryang2478/Decatf for posting this patch in pmaports#138!
Clearly state which version is being used, and also display the message
when using build --force.
- Old:
WARNING: Package 'ubuntu-app-launch' in your aports folder has version
0_git20180604-r0, but the binary package repositories already have version
0_p20181101174257-r0! See also: <https://postmarketos.org/warning-repo2>
- New:
WARNING: package hello-world: aport version 1-r4 is lower than 1-r5 from
the binary repository. 1-r5 will be used when installing hello-world.
See also: <https://postmarketos.org/warning-repo2>
Find the main package by assuming it is a prefix of the subpkgname. We
do that, because in some APKBUILDs the subpkgname="" variable gets
filled with a shell loop and the APKBUILD parser in pmbootstrap can't
parse this right. (Intentionally, we don't want to implement a full
shell parser.)
This covers most use cases and saves a lot of build time. Can be
changed on demand. Again, this simplifies package building as part of
the new build infrastructure effort.
Use the device's architecture instead of noarch. Because the device
packages should never be built for other architectures, even if all
depends can be built for other arches as well.
This simplifies package building as part of the new build
infrastructure effort.
All existing pmaports will be changed shortly, along with a test case
in pmaports.
Set HOME to /root for commands started with pmb.chroot.root() and to
/home/pmos for commands started with pmb.chroot.user().
POSIX requires this variable to be set, see:
<http://pubs.opengroup.org/onlinepubs/009695399/basedefs/xbd_chap08.html>
And this prevents a crash in "adb", which takes and alternative code
path if HOME is not set, that does not work with musl (fixes#1638).
Thanks to @ryang2678 for figuring this out!
* create symlinks to the GCC8 patches (introduced in pmaports!35)
* link to the new vendorkernel reference wiki page
* use SPDX license in the license= field
* add comment above the compiler section of the APKBUILD
* remove empty line at the end of the APKBUILD
When the timeout occurs it is important to ensure clean up of child
processes. Killing only the direct process created by a command can
leave child processes running.
For example a pmbootstrap.py install will run apk add. This run command
creates multiple processes as follows:
(cmd line arguments snipped for readability)
$ ps -e -o pid,ppid,pgid,cmd
PID PPID PGID CMD
31738 23247 31738 python3 ./pmbootstrap.py -t 15 install --no-fde
31746 31738 31738 sudo env -i /bin/sh -c ... ;apk --no-progress add
31747 31746 31738 /bin/sh -c ... ;apk --no-progress add
31748 31747 31738 apk --no-progress add
The root process of the run command is PID 31746. We want to kill
the child processes too. Otherwise only running kill -9 31746 will leave
the processes 31747 and 31748 running.
* aportgen: modify code to allow generating gcc6-armhf and other gcc6
cross compiler packages
* package: when 'gcc6' is in the depends of a package, and the cross
compiling mode is "native" (as we do it with kernels), install the
gcc6 cross compiler instead of the usual one (gcc8)
Related: pmaports#103
Overview:
In order to execute foreign arch binaries on the host system, we are
using the Linux kernel's binfmt_misc feature in combination with
static builds of QEMU. Before this patch, the statically compiled
QEMU binaries were taken from Debian (mostly because I did not realize
that Alpine ships them as well). Now we can use the ones from the aport.
Benefits:
This allows us to easily update and patch the QEMU executables, we
don't need to be in sync with Debian's versions anymore.
Alpine's package is more modular, so we can save some download,
install, zap time, as well as disk space: setting up an armhf chroot
with pmbootstrap took ~102 MB before, now it's ~18 MB.
Detailed changes:
* Remove `cross/qemu-user-static-repack` aport
* Add `data/qemu-user-binfmt.txt` with the binfmt_misc flags for ELF
binaries of various arches (extracted from Debian's packaging)
* When parsing that file, don't write verbose messages to
`pmbootstrap log` anymore, only to the verbose log (can be enabled
with `pmbootstrap -v`)
* Rename `pmb.parse.arch.alpine_to_debian()` to ...`alpine_to_qemu()`
* Rename `arch_debian` to `arch_qemu`
This commit adds a test case, which makes sure that the KDE framework
and plasma framework version are always the same.
Additional changes:
* APKBUILD parser parses the URL now (that's the best way I found to
categorize the KDE aports in frameworks and other)
* Changed single quotes to double quotes in KDE APKBUILDs, so the
parser doesn't include the single quotes in the parsed result
* Added the test case to the gitlab CI config
With this patch, "pmbootstrap flasher" will fail with "the following
arguments are required: action_flasher". Without it, it just prints
"Done" and quits.
Overview:
Since Alpine updated to distcc 3.3 last week, pmbootstrap wasn't able to use
distcc for cross compilation anymore. It always falled back to running the
compiler in QEMU (which works, but is a lot slower). The reason for that is,
that distcc requires all compilers that are being used in a whitelist now.
This partially fixes CVE-2004-2687 in distccd, which allowed trivial remote
code execution by any process connecting to the distccd server. We only run
distccd on localhost, but still this can be used for privilege escalation of
sandboxed processes running on the host system (not part of pmbootstrap
chroots).
Because the CVE is only partially fixed (see the comment in
`pmb/chroot/distccd.py` for details), we make sure that only the building
chroots can talk to the distcc server by running distcc over ssh.
Details:
* Completely refactored `pmb/chroot/distccd.py` to run distcc over ssh
* Store the running distcc server's arguments as JSON now, not as INI
* Make debugging distcc issues easy:
* Set DISTCC_BACKOFF_PERIOD=0, so the distcc client will not ignore the
server after errors happened (this masks the original error!)
* New pmbootstrap parameters:
* `--distcc-nofallback`: avoids falling back to compiling with QEMU and not
throwing an error
* `--ccache-disable`: avoid ccache (when the compiler output is cached,
distcc does not get used)
* `--verbose` prints verbose output of the distcc too
* New test case, that uses the new pmbootstrap parameters to force
compilation through distcc, and shows the output of distcc and distccd in
verbose mode on error (as well as the log of sshd)
Alpine's `abuild` will uninstall all dependencies by default, when a
package build fails.
Leaving this configuration unchanged leads to unexpected behavior with
pmbootstrap: when executing `pmbootstrap build --strict` and pressing
`^C` during the build, pmbootstrap will stop, but an `apk` process
will be started in the background to remove the dependency packages.
Running `pmbootstrap shutdown` at this time will not work, because the
`apk` process is still running.
With this commit, dependencies don't get cleaned up from the chroots.
Follow-up to !1373, where `pmbootstrap flasher flash_system` was
replaced with `pmbootstrap flasher flash_rootfs`. We still had used
terms like "system partition" in a lot of places.
This commit replaces it everywhere, so it's clear that we're talking
about the pmOS rootfs (which may or may not be installed to Android's
system partition).
The test suite needed a `pmbootstrap shutdown` after running through,
before it could successfully run again.
Explanation:
This was caused by `test/test_pkgrel_bump.py`, which creates a
temporary work folder with every subfolder ("chroot_native",
"cache_apk_x86_64", ...) linked to the original work folder except for
the "packages" folder. At the end of the test case,
`pmbootstrap shutdown` gets executed and is expected to umount
everything as usual. But it does not umount anything because of the
symlinks, so `work/chroot_native/mnt/pmbootstrap-packages` points to
the fake packages folder of that test case, even after it is finished.
As a result, any test case that tries to access the packages folder in
the native chroot, will fail until `pmbootstrap shutdown` gets called.
Detailed Changes:
* Umount all folders inside the work folder, even if these are symlinks
* Remove obsolete reference to "disable timestamp based rebuilds" in a
comment in `test/test_pkgrel_bump.py`
* Run `pmbootstrap work_migrate` and `pmbootstrap shutdown` at the
beginning of `test/testcases_fast.sh`, in case the pkgrel_bump test
case was aborted before it could properly shutdown and to make it
more robust in general (user may have changed the mountpoints, work
folder may need to be migrated)
Prior to this commit, it was possible to type in packages with a
trailing comma in `pmbootstrap init` when asked for extra packages.
This leads to problems during `pmbootstrap install`, so now we disallow
it. Fixes#1540.
It is unexpected for quite a lot of people, that the chroot folders are
still mounted when a pmbootstrap command has finished. With this commit,
it will let the user know explicitly:
> NOTE: chroot is still active (use 'pmbootstrap shutdown' as necessary)
Close#1524
We are analyzing the `boot.img` with `file` before we send it to
`unpackbootimg`. File does not recognize all kinds of `boot.img` files,
which `unpackbootimg` can extract, so we need a way to skip this check.
Details:
* Add `-f` parameter, continues extraction with a warning if the file
seems to be invalid
* Tell the user that `-f` can be used if the `boot.img` is invalid and
it's not specified
* Consistent spelling of `boot.img` instead of `bootimg` in messages
Fixes#1608
Oliver Smith
Alexey Min
Daniele Debernardi
lambdadroid
Luca Weiss
Martijn Braam
Steffen Pankratz
Oliver Smith
Idan Horo
Robert Yang
Nick Reitemeyer
Zhuowei Zhang
Clayton Craft
Bhushan Shah
Grant Miller
Ion Agorria
Craig Comstock
raingloom
NotKit
Marti Raudsepp