Age | Commit message (Collapse) | Author |
|
We run AUH from the autobuilder and we don't need the cron script any more,
we'd point people at the scripts there instead.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We don't need the ability to change this per layer, drop the option
and simplify the code.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
- Layer options can be defined as command line arguments
- multiple layers can be defined in a single command
each layer is analyzed 1 by 1 to check if upgrades are available.
Signed-off-by: David Pierret <david.pierret@smile.fr>
Reviewed-by: Yoann Congal <yoann.congal@smile.fr>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Yoann Congal <yoann.congal@smile.fr>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This is to prepare for the futur loop on layers.
Signed-off-by: Yoann Congal <yoann.congal@smile.fr>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Yoann Congal <yoann.congal@smile.fr>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Both branches contained the same code
Signed-off-by: Yoann Congal <yoann.congal@smile.fr>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
and 32 bit arm with musl
Everything else isn't really necessary and greatly lengthens AUH jobs.
Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
|
|
attaching them as files
This utilizes the 'scissors' feature as described here
https://git-scm.com/docs/git-mailinfo#Documentation/git-mailinfo.txt---scissors
and allows easy handling of AUH emails with tools like b4, patchwork or
even 'git am' directly, eliminating the need to get the attachment first.
Signed-off-by: Alexander Kanavin <alex@linutronix.de>
|
|
emails
This is optional, enabled by default, and allows easy extraction of patches
and other information to one's local disk (no need to go over individual maintainer
emails).
Signed-off-by: Alexander Kanavin <alex@linutronix.de>
|
|
MIMEBase should not be used directly and results in
sending raw binary data without encoding it with base64.
Signed-off-by: Alexander Kanavin <alex@linutronix.de>
|
|
This avoids guarding them against NoneType everywhere
Signed-off-by: Alexander Kanavin <alex@linutronix.de>
|
|
The stderr is only valid in newer versions of bitbake since bitbake
commit 1ecc1d94 ("process: Do not mix stderr with stdout"), in older
versions of bitbake (e.g. oe-core/dunfell), the stderr is None type
and fails AUH updates as follows:
```
DEBUG:Running 'devtool upgrade libfoo -S 1234abcd'
ERROR: libfoo: Failed(unknown error)
Traceback (most recent call last):
File "/oe/auto-upgrade-helper/upgrade-helper.py", line 481, in run
step(self.devtool, self.bb, self.git, self.opts, pkg_ctx)
File "/oe/auto-upgrade-helper/modules/steps.py", line 85, in devtool_upgrade
devtool_output = devtool.upgrade(pkg_ctx['PN'], pkg_ctx['NPV'], pkg_ctx['NSRCREV'])
File "/oe/auto-upgrade-helper/modules/utils/devtool.py", line 30, in upgrade
return self._cmd(cmd)
File "/oe/auto-upgrade-helper/modules/utils/devtool.py", line 22, in _cmd
return stdout + stderr
TypeError: can only concatenate str (not "NoneType") to str
INFO: libfoo: Auto commit changes ...
```
Check whether stderr is None and if it is, only use stderr.
Signed-off-by: Marek Vasut <marex@denx.de>
|
|
At some point devtool started printing to stderr, which means
patch rebase issues weren't anymore correctly caught, and bogus
updates were being sent out.
Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
|
|
Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
|
|
Attempting to send to unassigned@yoctoproject.org bounces immediately,
which means the CC recipients (such as oe-core list) won't get the email
either.
Instead, send to cc directly.
Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
|
|
In some situation this may happen in initial builds and
should not be treated as an error.
Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
|
|
The module was doing a lot of things, none of which are necessary:
- setting up a separate buildhistory repo for each recipe
- running cleansstate before doing initial build
- messing about with buildhisory revisions
Instead just run buildhistory-diff on the defaults right
after building the first target. This should also address
the various buildhistory problems seen on the AB.
Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
|
|
Rename upgradehelper.py into upgrade-helper.py since the conf file and the git
repo are named with a dash to separate "upgrade" and "helper".
The documentation in the README and upgradehelper.py were also already using
upgrade-helper.py instead of upgradehelper.py.
For all those reasons, let's be consistent and rename the python script.
Signed-off-by: Quentin Schulz <foss@0leil.net>
|
|
Signed-off-by: Quentin Schulz <foss@0leil.net>
|
|
Signed-off-by: Quentin Schulz <foss@0leil.net>
|
|
A few comments were misleading or incomplete and some defaults were incorrect.
Signed-off-by: Quentin Schulz <foss@0leil.net>
|
|
Even though a recipe might not have a MAINTAINER set, one still might want to
upgrade it.
This makes it possible to upgrade said kind of recipes and still keep the stats
meaningful by displaying one for all maintainerless recipes.
Signed-off-by: Quentin Schulz <foss@0leil.net>
|
|
Previously it was same as status update recipients.
Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
|
|
Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
|
|
Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
|
|
Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
|
|
Signed-off-by: Alexander <alex.kanavin@gmail.com>
|
|
Signed-off-by: Yann CARDAILLAC <yann.cardaillac@smile.fr>
|
|
Otherwise, if bitbake fails, there will be tracebacks later on.
Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
|
|
Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
|
|
The code in distrodata.bbclass related to the do_checkpkg task is rather
dated, has holes in it (ignoring some recipes) and has horrible locking
and csv related issues.
We should use modern APIs such as tinfoil to make the calls we need directly
against bitbake, cutting out the middleman and clarifing the code.
This change uses a lib/oe/recipeutils function call to remove the need
to use checkpkg at all.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Previously it would continue, which eventually resulted in an unhelpful, generic
error message printed by 'devtool finish'.
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
|
|
This was undocumented and made a number of hardcoded assumptions.
The options were to make it configurable, or just remove the logic,
and I chose the latter.
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
|
|
This helps in particular when a package upgrade builds fine, but then
causes other packages to fail. Depending on how close to the root
of dependency tree the failing packages are, this can badly subvert
the unattended upgrade process. So for that case, the better approach
is to revert every upgrade commit, so that each upgrade is effectively
built against master, which makes it less likely to fail.
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
|
|
Get it from bitbake instead.
[YOCTO #12566]
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
|
|
As devtool does not always remove it.
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
|
|
Maintainer whitelisting was broken, in particular.
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
|
|
Specifically, add more information about various config settings,
add a few important points to README, and remove config file specifics
from README, so that there's only one source for what the settings are
(the sample config file).
Also comment out everything from the config file, so that those in a
hurry can take it into use immediately. Most settings are specific to
a maintainer, and so can't be provided in a 'generic' way.
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
|
|
Settings mentioned in the README have been ported over and now users
are able to copy and start modifying for personal use cases.
This file will served as a reference to minimize errors when users
trying to create a new conf file by following the README.
Signed-off-by: Chang Rebecca Swee Fun <rebecca.swee.fun.chang@intel.com>
|
|
Signed-off-by: Chang Rebecca Swee Fun <rebecca.swee.fun.chang@intel.com>
|
|
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
|
|
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
|
|
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
|
|
Particularly:
1) determine per-package if ptest is supported, and add to the list
of packages to install accordingly
2) remove broken code that finds image logs in the build directory;
this will be replaced in the following commit
3) remove code that parses said logs for specific package failures;
this is prone to mistmatches and is best left to a human.
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
|
|
This adds enormously to run time, and is not particularly necessary
when running locally. If something fails, it's best to let the maintainer sort
it out as soon as possible, rather than try to guess what needs to be excluded.
Also, do not build a separate ptest image; ptests will be added
to the standard integration testimage.
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
|
|
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
|
|
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
|
|
messages
There's some logic to ascribe failures to specific recipe upgrade commits; it's prone
to mismatches and is lacking context about what else was upgraded. Let's make testimage
support available for local maintainer use only - unattended mass-upgrades will not
try to run testimages.
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
|
|
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
|