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`
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)
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).
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
* change "pmbootstrap kconfig_check" to "pmbootstrap kconfig check"
* change "pmbootstrap menuconfig" to "pmbootstrap kconfig edit [-x|-g]"
(with legacy alias, because the first syntax was referenced to a lot)
* enable X11 interfaces: -x: xconfig, -g: gconfig
* new function to copy the xauthority file:
pmb.chroot.other.copy_xauthority()
* remove menufconfig() function from the kernel template and all kernel
aports ([skip ci] because it would rebuild all kernels and run out of
time). Alpine has dropped this as well, and it wouldn't work with the
new code anyway.
* Usage: pmbootstrap install --split
* Make obvious that export is the next step when split images are created
* Fix note for missing rootfs image on export
* Change wording from "system image" to "rootfs image"
* The idea was to show the note only when the rootfs image was not
generated yet. But this was broken, because the path we checked for
was missing the chroot path prefix (which is added now).
* Also don't display the message, when the split image files exist
The `msm-fb-refresher` updates the screen for msm based devices. It is
not needed for all devices, so we had some extra code in the initramfs,
that would only add it when the `deviceinfo_msm_refresher` variable was
set. However, we are able now to add files to initramfs hooks, so this
hack can be removed and simplify everything.
Changes:
* Remove `deviceinfo_msm_refresher` from all deviceinfos
* Add sanity check for it
* Move all `deviceinfo` sanity checks to an extra function
* `postmarketos-mkinitfs`: remove code for msm refresher
* `msm-fb-refresher`: add initramfs hook
Changes:
* `helpers/envkernel.sh`:
* installs everything needed for kernel compilation in the native
chroot
* mounts the kernel source to `/mnt/linux` inside the chroot
* creates `/mnt/linux/.output` and chowns it to the `pmos` user, that
folder will be used for the kernel build output
* sets up aliases for `make`, `pmbootstrap`, `pmbroot`, `kernelroot`
* new action `pmbootstrap work_migrate`: does the interactive work
folder migration if necessary, otherwise it doesn't output anything
* when calling this first, we can safely use all other commands
non-interactively without showing the output
Benefits:
* Fast setup (especially for people who are new to kernel
compilation
* No need to figure out distribution specific package names
(cross compilers!)
* No need to do a test build just to verify that the right
packages are installed
* Less error prone
* The right dependencies are always installed
* `ARCH` and `CROSS_COMPILE` variables always get set automatically
and based on `deviceinfo_arch`
* If the build environment is broken for some reason, just zap and
start over
* Easy to reproduce problems
Notes:
* `make menuconfig` works as well
* Sourcing was tested with `zsh`, `bash` and `fish`, it should be easy to
extend for other shells
* As discussed in IRC/matrix, we're removing `linux-postmarketos-lts`
for now. The kernel isn't used right now, and we save lots of
maintenance effort with not updating it every week or so.
* new config option `"kernel"` with possible values:
`"downstream", "mainline", "stable"` (downstream is always
`linux-$devicename`)
* ask for the kernel during `pmbootstrap init` if the device package
has kernel subpackages and install it in `_install.py`
* postmarketos-mkinitfs: display note instead of exit with error when
the `deviceinfo_dtb` file is missing (because we expect it to be
missing for downstream kernels)
* device-sony-amami:
* add kernel subpackages for downstream, mainline
* set `deviceinfo_dtb`
* device-qemu-amd64: add kernel subpackages for stable, lts, mainline
* test cases and test data for new functions
* test case that checks all aports for right usage of the feature:
* don't mix specifying kernels in depends *and* subpackages
* 1 kernel in depends is maximum
* kernel subpackages must have a valid name
* Test if devices packages reference at least one kernel
* Remove `_build_device_depends_note()` which informs the user that
`--ignore-depends` can be used with device packages to avoid building
the kernel. The idea was to make the transition easier after a change
we did months ago, and now the kernel doesn't always get built before
building the device package so it's not relevant anymore.
* pmb/chroot/other.py:
* Add autoinstall=True to kernel_flavors_installed(). When the flag
is set, the function makes sure that at least one kernel for the
device is installed.
* Remove kernel_flavor_autodetect() function, wherever it was used,
it has been replaced with kernel_flavors_installed()[0].
* pmb.helpers.frontend.py: remove code to install at least one kernel,
kernel_flavors_installed() takes care of that now.
* Change `pmbootstrap flasher flash_system` command to
`pmbootstrap flasher flash_rootfs`
* The old command still works, but all references have been changed to
the new command
* Remove obsolete `pmbootstrap flasher export` (that was changed to
`pmbootstrap export` a few months ago)
* Update `README.md` and ZSH auto completion
* Change the description of the generated rootfs image (not talking
about a system image anymore, mention that it has subpartitions)
* Better description of `pmbootstrap flasher flash_rootfs --partition`
When the native arch (e.g. `x86_64`) `APKINDEX` files are outdated, and
`pmbootstrap` gets instructed to build a linux package for a foreign
arch, then the `APKINDEX` cache did not get used anymore for the
current session. This means that every lookup of a package in an
`APKINDEX` caused the whole `APKINDEX` file to get parsed again instead
of using the cached version. This slowed it down so much that it felt
like `pmbootstrap` was looping forever.
How this happens in detail:
* Whenever pmbootstrap parses an `APKINDEX`, it fills up the
`args.cache["apkindex"]` dict with the parsed information and the
last modified date of the file.
* `pmbootstrap` checks the last modified date of the `APKINDEX` files
and updates them if they are older than 4 hours.
* When the bug appeared, then the cache was already filled up, then an
update happened and then `pmbootstrap` tried to read from the cache.
* So when reading from the `APKINDEX`, the cache gets ignored because
the last modified date is different.
* Up to this commit, the cache does not get deleted and filled up
again!
How to test:
Try these commands once without this commit, and then with this commit
applied:
```
$ sudo touch -m -t 201801010000 \
~/.local/var/pmbootstrap/cache_apk_x86_64/APKINDEX.*
$ pmbootstrap -v build linux-postmarketos-mainline --arch=armhf
```
Without the patch, you can see in `pmbootstrap log` that it is
resolving the dependencies properly, but very slowly. With the patch
the resolving happens almost instantly.
When parsing the depends of entries in the APKINDEX file, we ignore
all operators (<, =, >). (This is enough for our use case, was we only
do the dependency resolving to check which packages need to be built
and `apk` does the dependency resolving again before installing
anything).
We did not ignore the ~ character for fuzzy version compares, this is
fixed with this commit.
* pmbootstrap newapkbuild: Properly parse arguments
The `pmbootstrap newapkbuild` action wraps Alpine's `newapkbuild`. We
used to directly pass all arguments to `newapkbuild` without verifying
in Python whether they make sense or not. However, as `newpakbuild`
doesn't do strict sanity checks on the arguments, it is easy to end up
with unexpected behavior when using the command for the first time.
For example, `newapkbuild` allows either specifying a PKGNAME or SRCURL
as last parameter, and also allows setting a PKGNAME with the `-n`
parameter. It only makes sense to use that option when passing a
SRCURL.
With this commit, we duplicate the optins that should be passed through
to `newapkbuild` and use argparse to fully sanitize the options and
display a help page (`pmbootstrap newapkbuild -h`) that is consistent
with the other help pages.
Details:
* The `-f` (force) flag does not get passed through anymore. Instead we
use it in Python to skip asking if an existing aport should be
overwritten (the aports are outside of the chroot, so `newapkbuild`
can't handle it in a way that makes sense for pmbootstrap).
* Output of `newapkbuild` gets redirected to the log file now, as we
don't need it to display a help page.
* Don't verify the pkgver while creating the new APKBUILD. When passing
a SRCURL, the pkgver gets extracted from the end of the URL and may
not have a valid format yet (but we want the APKBUILD anyway).
* Stored options passed through in `pmb/config/__init__.py` and use it
in both `pmb/parse/arguments.py` and `pmb/helpers/frontend.py`.
* Only allow `-n` with SRCURL
* The postmarketOS aports folder gets specified with `--folder` now.
That way the generated help page is much closer to the original one
from `newapkbuild`. The default is `main`.
* Made the package type flags (CMake, autotools, ...) exclusive so only
one of them can be specified
In order to get cross-compilers, we generate a few aports (e.g.
binutils-armhf, gcc-armhf) automatically from Alpine's aports.
pmbootstrap was already able to perform a git checkout of Alpine's
aports repository. But it needed to be manually updated. Otherwise
the `pmbootstrap aportgen` command could actually downgrade the aport
instead of updating it to the current version.
After thinking about adding a dedicated pmbootstrap command for
updating git repositories, I thought it would be better to not open
that can of worms (pmbootstrap as general git wrapper? no thanks).
The solution implemented here compares the upstream aport version of
the git checkout of a certain package (e.g. gcc for gcc-armhf) with the
version in Alpine's binary package APKINDEX. When the aport version is
lower than the binary package version, it shows the user how to update
the git repository with just one command:
pmbootstrap chroot --add=git --user -- \
git -C /mnt/pmbootstrap-git/aports_upstream pull
Changes:
* `pmb.aportgen.core.get_upstream_aport()`: new function, that returns
the absolute path to the upstream aport on disk, after checking the
version of the aport against the binary package.
* Use that new function in pmb.aportgen.gcc and pmb.aportgen.binutils
* New function `pmb.helpers.repo.alpine_apkindex_path()`: updates the
APKINDEX if necessary and returns the absolute path to the APKINDEX.
This code was basically present already, but not as function, so now
we have a bit less overhead there.
* `pmbootstrap chroot`: new `--user` argument
* `pmb.parse.apkbuild`: make pkgname check optional, as it fails with
the official gcc APKBUILD before we modify it (the current APKBUILD
parser is not meant to be perfect, as this would require a full shell
parsing implementation).
* Extended `test_aportgen.py` and enabled it by default in
`testcases_fast.sh`. Previously it was disabled due to traffic
concerns (cloning the aports repo, but then again we do a full KDE
plasma mobile installation in Travis now, so that shouldn't matter
too much).
* `testcases_fast.sh`: With "test_aport_in_sync_with_git" removed
from the disabled-by-default list (left over from timestamp based
rebuilds), there were no more test cases disabled by default. I've
changed it, so now the qemu_running_processes test case is disabled,
and added an `--all` parameter to the script to disable no test
cases. Travis runs with the `--all` parameter while it's useful to
do a quick local test without `--all` in roughly 2 minutes instead of
10.
* `aports/cross/binutils-*`: Fix `_mirror` variable to point to current
default Alpine mirror (so the aportgen testcase runs through).
Since PR #1247 we are using the virtual package `.pmbootstrap` to mark
packages as not explcitly installed. In case `pmbootstrap` doesn't
finish the installation because of a bug, the `.pmbootstrap` virtual
package may not get uninstalled.
As virtual packages don't have the "timestamp" attribute set in the
package database (which indicates when the package was built), the
APKINDEX parser fails to parse them.
Changes:
* pmb.parse.apkindex.parse_next_block(): don't require the "timestamp"
attribute to be set (but the arch attribute instead, which is always
present)
* pmb.parse.apkindex.parse(): when a block does not have a `timestamp`
attribute, skip it, because it must be a virtual package.
* add test cases for both functions with a package database that
contains a virtual package.
`-m` is for deleting local compiled packages, for which there is no
aport with the same version. Prior to this change, this only worked
for packages where no aport exists, or for packages that are newer
than the aports.
That is, because we used the usual APKINDEX parsing logic, which
ignores old packages in the APKINDEX and only returns the one with the
highest version (that makes sense during dependency resolution).
Changes:
* New `pmb.parse.apkindex.parse_blocks()` function that returns a raw
list of blocks, instead of the dict with removed duplicates with
lower version you get from the usual `.parse()` function.
* Renamed each of the zap flags and their descriptions to make clear
what they are doing now.
```
short long (old) long (new)
-p --packages --pkgs-local
-m --mismatch-bins --pkgs-local-mismatch
-o, --old-bins --pkgs-online-mismatch
```
Use case: `mkbootimg` provides the `unpackbootimg` package. When
running `pmb.chroot.apk.install(args,"unpackbootimg")`, it was not
able to properly build the package.
Reproducing the error:
```
sudo rm ~/.local/var/pmbootstrap/packages/x86_64/mkbootimg*
pmbootstrap index
pmbootstrap --mirror-pmOS="" chroot --add=unpackbootimg
```
Or alternatively (simpler but less illustrative):
```
pmbootstrap build unpackbootimg --force
```
### Only download APKINDEX for relevant architectures
We're downloading the APKINDEX files for all architectures supported by
postmarketOS currently (x86, x86_64, armhf, aarch64). Most of the time,
we only need it for the native and device arch, so this PR reduces the
downloaded files to what is really necessary.
### Intuitive pmbootstrap update logic
* pmb.helpers.repo.update():
* Default is updating all arches where the APKBUILD files exist
* Add existing_only parameter
* Return True when files have been downloaded
* Properly print which arches will be updated
* Print update reason only in verbose log
* Add and improve comments
* pmb.parse.arguments(), update action:
* Add --non-existing parameter
* Default for --arch is None (instead of arch.native)
* pmb.helpers.frontend.update():
* Inform about --non-existing if no APKBUILDs have been updated
Here are the changes necessary in pmbootstrap to make proprietary
software installed onto the device (firmware and userspace drivers)
optional (#756). To full close the issue, we need to apply this concept
to all device packages we already have in a follow-up PR.
Changes:
* New config file options nonfree_firmware and nonfree_userland, which
we ask for during "pmbootstrap init" if there are non-free components
for the selected device.
* We find that out by checking the APKBUILD's subpakages: The non-free
packages are called $pkgname-nonfree-firmware and
$pkgname-nonfree-userland.
* During "pmbootstrap init" we also show the pkgdesc of these
subpackages. Parsing that is implemented in
pmb.parse._apkbuild.subpkgdesc(). It was not implemented as part of
the regular APKBUILD parsing, as this would need a change in the
output format, and it is a lot *less* code if done like in this
commit.
* pmb/parse/apkbuild.py was renamed to _apkbuild.py, and
pmb/install/install.py to _install.py: needed to call the function in
the usual way (e.g. pmb.parse.apkbuild()) but still being able to
test the individual functions from these files in the test suite.
We did the same thing for pmb/build/_package.py already.
* Install: New function get_nonfree_packages() returns the non-free
packages that will be installed, based on the user's choice in
"pmbootstrap init" and on the subpackages the device has.
* Added test cases and test data (APKBUILDs) for all new code,
refactored test/test_questions.py to have multiple functions for
testing the various questions / question types from
"pmbootstrap init" instead of having it all in one big function.
This allows to use another aport folder for testing the new
non-free related questions in init.
The message appears 20 or so times in a row when setting up a new
chroot. Right now we log it into the debug log (which is what
"pmbootstrap log" and "pmbootstrap --details-to-stdout" show). This
is annoying for normal pmbootstrap usage, and really clutters up the
Travis output.
This commit changes the log level to verbose, so it only shows up
when "-v" is passed to pmbootstrap.
* The APKINDEX parser used to return a dictionary with one package for
a given package name. This works for the installed packages database,
because there can only be one provider for a package. But when
parsing packages from binary repositories, we need to support
multiple providers for one package. It is now possible to get a
dictionary with either multiple providers, or just a single provider
for each package.
* Dependency parsing logic has been adjusted, to support multiple
providers. For multiple providers, the one with the same package
name as the package we are looking up is prefered. If there is none
(eg. "so:libEGL.so.1" is provided by "mesa-egl"), it prefers packages
that will be installed anyway, and after that packages that are
already installed. When all else fails, it just picks the first one
and prints a note in the "pmbootstrap log".
* Added testcases for all functions in pmb.parse.apkindex and
pmb.parse.depends
* pmbootstrap chroot has a new "--add" parameter to specify packages
that pmbootstrap should build if neccessary, and install in the
chroot. This can be used to quickly test the depencency resolution
of pmbootstrap without doing a full "pmbootstrap install".
Fixes#1122.
* New "pmbootstrap build --src=/local/source/path hello-world" syntax
* The local source path gets mounted inside the chroot
* From there, a copy of the source code gets created with rsync (so
we can write into the source folder if necessary, for better
compatibility with all kinds of APKBUILDs)
* After the aport gets copied into the chroot before building (as
usually), we extend the APKBUILD with overrides to make it use
mountpoint's source instead of downloading the package's source
from the web as usually
* The package built with the local source gets _pYYYYMMDDHHMMSS
appended to the pkgver
* linux-postmarketos-mainline: use $builddir, fix patch checksum
The 'necessary_kconfig_options' dictionary in pmb/config/__init__.py
now has the different architectures (space separated) as the keys and
the dictionary, which matches kernel config options and their
expected value, as its value.
For that purpose, the 'check' function in pmb/parse/kconfig.py was
modified, so that it takes the architecture from the kconfig filename
and uses it to find the needed kernel config options.
Closes#1218.
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.
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.
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 ...)
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.
This PR makes the workflow faster and pmbootstrap will
produce less traffic. Details:
* Check if it's possible to create and read from a device
node directly when initializing a chroot (closes#472)
* Copy the Qemu binary into the forign-arch chroots
before initializing them, so the post-install script
directly work during the chroot setup and we don't need
to call apk fix afterwards
* Use pmb.helpers.repo.update(), which only updates the
APKINDEX files if they are older than 4 hours, instead
of using apk's repo update function which always
downloads the APKINDEX files
* Chroot initialization
* Getting the initial APKINDEX to download apk-tools-static
* Updating the APKINDEX at the start of pmbootstrap install
* Fixed a bug in from_chroot_suffix: the buildroot_x86_64 has
architecture x86_64, not x86.
Small improvements:
* Allow to specify multiple packages to `pmbootstrap parse_apkbuild`
* Specifying no package will parse all packages (like kconfig_check)
(also `parse_apkbuild`)
* JSON output is sorted of `parse_apkbuild`
* Make pkgver check optional, so we can disable it in the device wizard test case
* Parse_apk* -> apk*_parse
* Don't let the user mess with globs (disallow '*' in pkgname)
Oliver Smith
Grant Miller
ryang
steamport
Marcin Mielniczuk
Daniele Debernardi
Luca Weiss
Alexander
Lucas Ramage
Pavel Machek