Get rid of config.mirror_alpine and mirrors_postmarketos and make sure
they get migrated over for existing users.
mirrors_postmarketos being a list was always a bit off, but now we have
per-aports mirrors which make a lot more sense for what we're trying to
do with systemd.
Signed-off-by: Caleb Connolly <caleb@postmarketos.org>
We used to recursively descend into the dependencies of the packages we
mark for install, this was needed so that we build all the packages we
want to.
However, the new package builder now does this for us. So simplify and
speed this up a bit by not doing it in apk.install().
Additionally, soft-remove support for passing in packages to delete by
having the package name start with a "!". This was useful at some point,
but we now handle this much better in pmaports.
Signed-off-by: Caleb Connolly <caleb@postmarketos.org>
Generalise pmb.helpers.other.cache with a more python decorator.
The Cache decorator takes a list of function arguments to use as cache
keys, keyword args can be used to restrict caching so that it is skipped
entirely unless the attribute has a specific value.
For example, pmb.helpers.pmaports.get() has the decorator:
@Cache("pkgname", subpackages=True)
This means the return value will be cached only when subpackages is
True, otherwise it will always miss.
Signed-off-by: Caleb Connolly <caleb@postmarketos.org>
Add a new config section "mirrors", to replace the mirrors_alpine and
mirrors_postmarketos options. This will allow for more flexibility since
we can then handle the systemd staging repo (and others like plasma
nightly) with relative ease.
The loading/saving is fixed and now properly avoids writing out default
values, this way if the defaults are changed the user won't be stuck
with old values in their pmbootstrap.cfg.
Signed-off-by: Caleb Connolly <caleb@postmarketos.org>
Move pmb/parse/arch.py over to core and refactor it as an Arch type,
similar to how Chroot was done. Fix all the uses (that I can find) of
arch in the codebase that need adjusting.
The new Arch type is an Enum, making it clear what architectures can be
represented and making it much easier to reason about. Since we support
~5 (kinda) different representations of an Architecture (Alpine, Kernel,
target triple, platform, and QEMU), we now formalise that the Alpine
format is what we represent internally, with methods to convert to any
of the others as-needed.
Signed-off-by: Caleb Connolly <caleb@postmarketos.org>
Using Paths as cache keys is not really ideal, plus it's the absolute
path on the host, weird.
For now, add a cache_key() function to get just the part of the path we
care about and use it as the key. This probably ought to be revisisted
later.
Signed-off-by: Caleb Connolly <caleb@postmarketos.org>
Keeping the Config class in types seemed kinda weird and was just done
as a workaround to some cyclical imports. But now things are more in
shape let's move it to core.
Signed-off-by: Caleb Connolly <caleb@postmarketos.org>
Use a different binary repo depending on the source repository for the
package. This makes it possible to split out systemd packages into their
own repository.
Signed-off-by: Caleb Connolly <caleb@postmarketos.org>
Not sure if we should allow flash_method to be unset, but eh, simplify
usage by adding the none fallback here.
Signed-off-by: Caleb Connolly <caleb@postmarketos.org>
Introduce a Deviceinfo class and use it rather than the dictionary. This
gives us sweet sweet autocomplete, and lays the foundation for having a
proper deviceinfo validator in the future.
Additionally, continue refactoring out args...
Signed-off-by: Caleb Connolly <caleb@postmarketos.org>
Cease merging pmbootstrap.cfg into args, implement a Context type to let
us pull globals out of thin air (as an intermediate workaround) and rip
args out of a lot of the codebase.
This is just a first pass, after this we can split all the state that
leaked over into Context into types with narrower scopes (like a
BuildContext(), etc).
Signed-off-by: Caleb Connolly <caleb@postmarketos.org>
Defaulting to the native chroot isn't necessarily intuitive. Let's
require this be specified in full.
Signed-off-by: Caleb Connolly <caleb@postmarketos.org>
With the new chroot type, we can now write fancy paths in the pythonic
way. Convert most of the codebase over, as well as adding various other
type hints.
Signed-off-by: Caleb Connolly <caleb@postmarketos.org>
We use a custom verbose log level in pmbootstrap, unfortunately it isn't
possible to correctly type this due to some limitations in the logging
library [1], [2].
Given that our usecase is fairly simple, we can just wrap the module
with our own so we only have to tell mypy to ignore the error once
instead of at every callsite.
[1]: https://github.com/cryptax/droidlysis/issues/15
[2]: https://github.com/python/typing/discussions/980
Signed-off-by: Caleb Connolly <caleb@postmarketos.org>
The init_cache() function just assigned some default constants, simplify
this by just declaring it that way to begin with, as well as adding type
hints.
Signed-off-by: Caleb Connolly <caleb@postmarketos.org>
Introduce a new module: pmb.core to contain explicitly typed pmbootstrap
API. The first component being Suffix and SuffixType. This explicitly
defines what suffixes are possible, future changes should aim to further
constrain this API (e.g. by validating against available device
codenames or architectures for buildroot suffixes).
Additionally, migrate the entire codebase over to using pathlib.Path.
This is a relatively new part of the Python standard library that uses a
more object oriented model for path handling. It also uses strong type
hinting and has other features that make it much cleaner and easier to
work with than pure f-strings. The Chroot class overloads the "/"
operator the same way the Path object does, allowing one to write paths
relative to a given chroot as:
builddir = chroot / "home/pmos/build"
The Chroot class also has a string representation ("native", or
"rootfs_valve-jupiter"), and a .path property for directly accessing the
absolute path (as a Path object).
The general idea here is to encapsulate common patterns into type hinted
code, and gradually reduce the amount of assumptions made around the
codebase so that future changes are easier to implement.
As the chroot suffixes are now part of the Chroot class, we also
implement validation for them, this encodes the rules on suffix naming
and will cause a runtime exception if a suffix doesn't follow the rules.
This is needed for sphinx autoprogram since that expects an
argparse.ArgumentParser, and arguments() returns some argparse
"Namespace" obj. Useful for sphinx/autoprogram and maybe other things
later that want to get at pmb's full args.
A common pattern in APKBUILDs, is to introduce custom variables prefixed
with underscores that get then used in makedepends and other variables.
For example:
_wlrootsmakedepends="
eudev-dev
# ...
"
makedepends="
# ...
$_wlrootsmakedepends
"
Adjust the APKBUILD parser code, so it parses all top-level variables
and can use them further below when referenced inside other variables.
Before returning the parsed APKBUILD data, remove all variables that are
not in pmbootstrap's list of known APKBUILD parsing attributes (so the
result is the same).
I've compared "pmbootstrap apkbuild_parse" (which parses all APKBUILDs
in the currently checked out pmaports dir), before and after this
change, and the result is the same except for having more variables
successfully replaced.
- Performance Note-
This new implementation is actually faster than the previous one,
because we don't need to iterate through all known keys on each line of
the APKBUILDs. On my machine, average of 3 runs, parsing all APKBUILDs
of current pmaports master takes about half as long as with the previous
implementation.
$ time pmbootstrap -q apkbuild_parse >/dev/null
-> old code: 0.954
-> new code: 0.483
Make the handling of the custom NonBugError and BuildFailedError
exceptions more consistent with the handling of other exceptions, by
printing "ERROR: " infront of the actual error text. Then we don't need
to duplicate that where we raise the errors. pmbootstrap prints "ERROR"
in red.
Reimplement "pmbootstrap status" to be just a simple and useful status
overview. The previous version ran a bunch of checks every time, and
would fail on these even if pmaports was used for normal development:
* "non-official" branch checked out in pmaports
* pmaports.git is not clean
The information about aports.git was also considered not so useful upon
revisiting this command, since it is only used for "pmbootstrap
aportgen". Most users don't need this, and if the user runs this
command, it will tell if aports.git is outdated.
All of the above made the previous version unpleasant to use and I
suspect most people stopped using the command after trying it out a few
times and seeing the irrelevant but loud NOK complaints.
New version:
$ pmbootstrap status
Channel: edge (pmaports: master_staging_systemd)
Device: qemu-amd64 (x86_64, kernel: virt)
UI: console
systemd: no (default for selected UI)
Old version (without --details it only shows NOK checks):
$ pmbootstrap status --details
[00:55:20] *** CONFIG ***
[00:55:20] Device: qemu-amd64 (x86_64, "QEMU amd64")
[00:55:20] Kernel: virt
[00:55:20] User Interface: console
[00:55:20]
[00:55:20] *** GIT REPOS ***
[00:55:20] Path: /home/user/.local/var/pmbootstrap/cache_git
[00:55:20] - aports_upstream (master)
[00:55:20] - pmaports (master)
[00:55:20]
[00:55:20] *** CHECKS ***
[00:55:20] [OK ] Chroots zapped recently (or non-existing)
[00:55:20] [OK ] aports_upstream: on official channel branch
[00:55:20] [OK ] aports_upstream: workdir is clean
[00:55:20] [OK ] aports_upstream: tracking proper remote branch 'origin/master'
[00:55:20] [OK ] aports_upstream: up to date with remote branch
[00:55:20] [OK ] aports_upstream: remote information updated recently (via git fetch/pull)
[00:55:20] [OK ] pmaports: on official channel branch
[00:55:20] [OK ] pmaports: workdir is clean
[00:55:20] [OK ] pmaports: tracking proper remote branch 'origin/master'
[00:55:20] [OK ] pmaports: up to date with remote branch
[00:55:20] [OK ] pmaports: remote information updated recently (via git fetch/pull)
[00:55:20]
[00:55:20] NOTE: chroot is still active (use 'pmbootstrap shutdown' as necessary)
[00:55:20] DONE!
This works around an issue where some armhf apps are compiled with
instructions that are not compatible with aarch64, for example apk uses
MCR, and that does not exist on aarch64 causing apk to fail in a chroot
with SIGILL. It's impossible to build armhf images on aarch64 hosts, and
this also fixes a crash when running `pmbootstrap zap -a`, since pmb zap
will init an armhf chroot and blow up with apk generates a SIGILL.
For reference, here's how I arrived at the conclusion that apk (and gdb,
and probably others...) are executing invalid instructions on aarch64
when not using binfmt+qemu emulation:
$ uname -m
aarch64
$ file sbin/apk.static
sbin/apk.static: ELF 32-bit LSB executable, ARM, EABI5 version 1 (SYSV), statically linked, BuildID[sha1]=997a8ef97b17cb4951a6220b7807a66fed67bc10, stripped
$ gdb sbin/apk.static
(gdb) r
Starting program: /run/user/10000/foo/sbin/apk.static
Program received signal SIGILL, Illegal instruction.
0x0012843c in ?? ()
(gdb) x /1i $pc
=> 0x12843c: mcr 15, 0, r0, cr7, cr10, {5}
I'm not sure why pmb.parse.arch is being imported here given that it's
the module we already are in. Replace it with pmb.config which this
module actually needs.
Instead of assuming the architecture of the foreign device to be what
the user selected in pmbootstrap init, actually query the architecture
from the device and use that.
This does mean that one extra ssh connection is necessary, which does
slow down the procedure somewhat. However, I think that is worth the
user experience improvement this brings.
Also, the deduction process can be skipped by manually specifying
--arch in the sideload invocation should it fail, or if one really wants
to skip that extra ssh roundtrip.
Related: https://gitlab.com/postmarketOS/pmbootstrap/-/issues/2317
Sometimes it's useful to map an arbitrary machine type to an Alpine
architecture. alpine_native only lets you get the Alpine architecture
mapped to the machine type of the system pmbootstrap is running on. As
such, break out this functionality into a new function that takes the
machine type as a parameter.
The intent of 9a74848f19 was to not print
a stacktrace when this happened. However, I still opted to raise an
exception, so you still get a stacktrace, just with a more helpful
message. There's no need to give the user a big stacktrace here, so
raise a NonBugError instead of a RuntimeError.
Implement a function to verify the type of QCDT image by comparing the
first four bytes of the file. This is represented in a deviceinfo
variable, used by boot-deploy for dt.img generation.
Often I find myself confused and subsequently frustrated by the
behaviour of --fork-alpine as usually I apply whatever changes I want to
the aports package before forking. This is due to that in the proper
aports repo I can leverage the full power of git and apply patches or
revert commits, whereas after the package has been forked to pmaports
the context git needs to do this is lost.
To fix that, this commit introduces --fork-alpine-retain-branch which
works the same way as --fork-alpine except it doesn't change the aports
branch to match the current channel.
If there are multiple sections to a subpackage declaration, the middle one
(item 1 in the sequence in this case) is the custom function name which we
should use instead of the deduced one.
For example:
$pkgname-something-separate:something_separate:noarch
Here, the subpackage is called $pkgname-something-separate, but function names
cannot contain hyphens, so it's instead given the custom name
something_separate where the hyphen is replaced by an underscore. At the end,
the architecture of the subpackage is also overriden as noarch instead of
whatever architecture the main package has.
However, it is also possible to only override architecture of a subpackage and
not the function name.
In such cases, we get this:
$pkgname-something::noarch
We still have a "middle section" (item 1 in the sequence), but it will be
empty. As such, we not only need to check whether there are more than 1
subpackage part, but also whether that extra subpackage part is an empty string
or not.
This fixes an issue where pmbootstrap would not end up just setting the
properties of subpackages with a set arch but no custom function as
None instead of giving them proper depends, install, pkgdesc, et cetera
properties.
Due to that logging.verbose is "monkeypatched" into logging and not
present by default in the module, Mypy isn't able to pick up that it
exists. As such, just disable the check altogether in this file for
now.
Repurposing argcomplete as a boolean telling whether the module
argcomplete was found results in a variable with two distinct purposes
and types, which is bad practice and confuses Mypy. As such, don't
assign False to argcomplete if the module is not found and instead just
check for whether it exists in sys.modules to determine whether it was
imported successfully.
I zap chroots a lot, since I've found that it often "fixes" a lot of
weird issues that come about if you have stale chroots laying around.
So a common pattern I do is "pmb zap && pmb install ...". Having an
option to pmb install let's me simplify this.
Reviewed-by: Oliver Smith <ollieparanoid@postmarketos.org>
Link: https://lists.sr.ht/~postmarketos/pmbootstrap-devel/%3C20231214234051.4673-1-clayton@craftyguy.net%3E
