Analysis shows that "bitbake core-image-ptest-all" spends a lot of
time in scenequeue_updatecounters and much of it is rebuilding a set
which doens't change. Reorder the code to avoid that performance
glitch.
(Bitbake rev: 923c19b9713e398d8e66e6d4422dfd4c18a03486)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Rather than looping through things we looped through on the previous execution,
start looping where we left off for setscene processing. This gives speed
improvements depending on the kind of build being executed.
(Bitbake rev: 00f4d932e3af0eeb333339cbe942010fc76dee0f)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
There have been complaints about the performance of large multiconfig builds
for a while. The key missing data point was that the builds needed to have large
overlaps in sstate objects. This can be simulated by building the same things with
just different TMPDIRs. In runqueue/bitbake terms this equates to large numbers of
deferred tasks.
The issue is that the expensive checks in the setscene loop were hit every time
through runqueue's execute function before the check on deferred tasks. This leads
to task execution starvation as that only happens once per iteration.
Move the skip check earlier in the function which speeds things up enormously
and should improve performance of such builds for users.
(Bitbake rev: 9c6c506757f2b3e28c8b20513b45da6b4659c95f)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
do_unpack generates a 'Bitbake Fetcher Error: UnpackError' when
uncompressing a single 'file.zst', zstd command line returns the error
'unexpected end of file', and the input file 'file.zst' becomes empty.
The command line is 'zstd --decompress --stdout file.zst > file.zst'.
So the issue is that the input and output filenames are the same.
Fix the output filename when using zstd for a single file.
(Bitbake rev: 5881ca0e0d53937cd25a57ff682a3b7577b0b7f8)
Signed-off-by: Pascal Eberhard <pascal.eberhard@se.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Removes the code in bitbake to show custom backtrace formatting for
exceptions. In particular, the bitbake exception code prints function
arguments, which while helpful is a security problem when passwords and
other secrets can be passed as function arguments.
As it turns out, the handling of the custom serialized exception stack
frames was pretty much made obsolete by d7db75020ed ("event/msg: Pass
formatted exceptions"), which changed the events to pass a preformatted
stacktrack list of strings, but the passing of the serialized data was
never removed.
Change all the code to use the python traceback API to format exceptions
instead of the custom code; conveniently traceback.format_exception()
also returns a list of stack trace strings, so it can be used as a drop
in replacement for bb.exception.format_exception()
(Bitbake rev: 2cda75a185aaf8f657f072dac34f8cef9d75f63a)
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
There's no easy way to list all of the tasks in a recipe, you can either
look at __BBTASKS (internal variable, shouldn't be used) or iterate all
items in the datastore looking for variables with the 'task' flag set
(which is slow).
Solve this problem by adding a bb.build.listtasks() function that
returns an immutable copy of the __BBTASSK variable.
(Bitbake rev: 185c4b803962b20ba65a7d885dfe1a14e68736ef)
Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Testing shows the worst case CDN response time can be up to 100s. The wget fetcher
is used for accessing sstate from the CDN so increase our timeouts there to match
our worst case repsonse times.
(Bitbake rev: 7aca591529e115bc277f93811d8c586630acc8c3)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The default pipe size is 64kb on builds, which can be inefficient
for larger log files from workers. Increase the pipe size to 512kb
since build systems have decent amounts of memory and this is a more
efficient way of batching the data.
Tweak the default read sizes to match the pipe size for efficiency.
Since the contstant is only present in python 3.10 onwards, add
some compatibility code.
(Bitbake rev: 69c14e46600ba5ae9703f67704ab2548875ae6d7)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
DANGLINGAPPENDS_WARNONLY is a very poorly designed interface and
is commonly abused. The challenge is that once it is set, by any
layer, it applies everywhere.
Some layers rely on this to get notification they need to update
bbappend files and having the behaviour change from inclusion of an
antisocial layer is not good.
In addition, showing warnings as an accepted thing on the console
devalues them and trains the user to ignore them. I want to steer us
away from this mindset.
We could extend the functionality and make it apply only to certain
layers, or only to certain appends but then we've basically re-invented
BBMASK.
Given all the above, we should drop support for BB_DANGLINGAPPENDS_WARNONLY
and direct anyone with issues to BBMASK instead.
https://lists.openembedded.org/g/openembedded-architecture/message/2029
[YOCTO #14870]
(Bitbake rev: fca9c9e3cb6f8e9f99bf51dc5e8a8d83f4c84c69)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Create a new tinfoil API "finalizeData" which does key expantion and
runs the anonymous functions to allow bitbake-getvar to have the same
output as "bitbake -e".
(Bitbake rev: 63db64ca98564420a378f14c337074ae06da8c63)
Signed-off-by: Yoann Congal <yoann.congal@smile.fr>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fixes what looked like a type since parseConfiguration.needconfig is
already set just after parseConfiguration() definition.
(Bitbake rev: d371df029ece3b9e600a89d08337c437a8ddbf63)
Signed-off-by: Yoann Congal <yoann.congal@smile.fr>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The idle and main loops have socket select calls to know when to execute.
This means we can increase the normal timeout frequency since this is
just a fall back and have some small efficiency gains.
(Bitbake rev: 8d8e17af8619c976819170c9d5d9a686a666c317)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If there are no idle functions present, don't sent heartbeat events. These
are only meant to happen while builds are active.
(Bitbake rev: 9a2d5e63b07c3912838781776c61f0f1ac9640e1)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Keeping this code separate just makes the code harder to understand,
merge them.
(Bitbake rev: e5ac26a0e1779df1da3277bf48899c8f7642f1f8)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We're seeing some failures on hosts where slow "idle" loop iterations are
causing bitbake server timeouts. These seem to happen particularly in the
dump_signatures() function within runqueue.
That isn't entirely surprising since it creates a pool of threads to execute
work an at best can take around 10s to execture and return control backto the
main loop. On a slow system, it is understandable this can take longer,
particularly as these functions are creating large chunks of IO.
Since the work is being done in threads, we can launch them, return to idle
and check on the results periodically as they complete.
This should hopefully address some of the remaining timeout issues we see on
the autobuilder in oe-selftest sstate tests.
(Bitbake rev: e66f1b643b4b77404ba31f2704cda5af9bf00a57)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
lz4c has been considered deprecated by upstream since at least 2018 [1]
and has been disabled by default recently [2]. openSUSE Tumbleweed is
no longer packaging the deprecated version and others will likely
follow.
Luckily, we're not using any legacy-specific options and, going back as
far as Ubuntu 16.04/Fedora 25, both lz4 and lz4cli are installed as part
of the same package
1. https://github.com/lz4/lz4/pull/553
2. https://github.com/lz4/lz4/pull/1479
(Bitbake rev: 907472034b344e4eb73cfd43059a413469f52e1c)
Signed-off-by: Justin Bronder <jsbronder@cold-front.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
GNOME gitlab has occasional downtime which impacts bitbake-selftest
and causes autobuilder failures. Switch to our own mirror for test
purposes to avoid those issues.
(Bitbake rev: 91e268b11ed683bd197026f9b36001f6d54ee05c)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
GNOME gitlab has occasional downtime which impacts bitbake-selftest
and causes autobuilder failures. Switch to our own mirror for test
purposes to avoid those issues.
(Bitbake rev: 008808755ed6cfeb6c41273e69ce718f0833c26c)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The file may be a symlink when mirror from local disk, and the symlink will be
broken when the linked file is removed, use os.path.islink() to check and
remove the broken symlink.
(Bitbake rev: d0deb5fe6e7a84b07b49ef685b209729d2fdc780)
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fixed:
BB_GIT_SHALLOW = "1"
BB_GENERATE_SHALLOW_TARBALLS = "1"
$ bitbake utfcpp -cfetch && bitbake utfcpp -ccleanall
The downloads/gitsmshallow_github.com.nemtrif.*.tar.gz won't be cleaned without
this fix.
(Bitbake rev: bab7a8970a0237a9d24217685a595e76a1336c07)
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fixed:
$ bitbake utfcpp -cfetch && bitbake utfcpp -ccleanall
The downloads/git2/github.com.nemtrif.ftest won't be cleaned without this fix.
(Bitbake rev: 79f25fc5c1b8d0e08540f4aa07875309f5325f47)
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The tmpdir such as downloads/tmplp3cnemv won't be removed without this fix.
(Bitbake rev: 2ba8d3214759142afc11f0a88d80eb30a8bcde3a)
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
There are 14 lines can be removed, and can make it easy to maintain.
(Bitbake rev: 0ea2c1ac079d63349407a69172ff80cd9acc7252)
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We really don't want to be skipping a test like this, drop the exception handling
code as in my testing we thankfully no longer seem to be seeing it. If we do again
we need to fix the underlying issue.
(Bitbake rev: b67547f890dc7f347ec05a97cbfb22d3703998cc)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fix the exception handling in the cancel builds function.
This involved adding WebDriverException which sometimes occurs but
also correcting the other exception handlers to continue to increment
the timeout to avoid test hangs.
(Bitbake rev: e111a2bd4f7a8a4dc2c63e94e91ac6cacca95af8)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Most of the tests that click on buttons need the DOM to stablise, including
any running JQuery code before the test can proceed. Add calls to do this
whenever we're about to click on an element.
(Bitbake rev: 0eb206b355248e2a874a62baec30025652f2a5a8)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Switch the clickable() calls to use the new element_clickable() function
which accepts a finder labmda function. This means if the element doesn't
yet exist, the code can rebuild the query and try again once a small
amount of time has elapsed.
There were a ton of timing related races around these element interactions
and this seemed to be the most robust way to address the issues.
The change also makes some of the elements slightly more specific so
the code can work effectively.
(Bitbake rev: 38643aadbb5a960004b886cf7709beaf2fc96652)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If the element never exists, the timeout is never incremented and the test
hangs indefinitely. Fix the exception handling to avoid that and allow
the timeout to happen.
(Bitbake rev: 9eabe923d457bbce65227da4cd71c275c32108e6)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fixes for the browser tests which:
* Add/improve waiting for specific elements
* Waiting for elements to be interactable, not just visible
* Hiding notification windows
* Using more specific asserts to obtain improved error messages
(Bitbake rev: 0c3f9bc2b9516d2af391f2e081280eb7fa62a05e)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Add a variant of the clickable function that accepts an element parameter.
(Bitbake rev: a34b2451a0e3baaf3b2fd2eb351847ff6a5ee033)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Our own Wait() class allows exception handling which this form of wrapper
does not. Switch the code to use our Wait() class to allow retrying upon
encountering those exceptions (such as an element not being present yet).
The displayed and visible test is what Selenium would be doing internally,
there is no JS reprensetation of clickable directly.
(Bitbake rev: 8266a01b750b3758badeee8fb3a1acfa72c17a93)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The code modifying the build object requires direct database access
which is potentially problematic. It was being run on a database which
was being reset after changes anyway so the code effectively did nothing.
Remove it as it is unneeded.
(Bitbake rev: 2e3fa624eb3972aef91d877337479eeb0f3dd365)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Drop the remaining poll parameters from the helpers code along
with the remaining sleep call since the tests no longer depend
on this.
This has the nice benefit of significantly speeding up the toaster
test runs (45 minutes down to 12 minutes overall).
If a parameter is needed, it should be the timeout, not the polling
frequency.
(Bitbake rev: 6de912e4f278ffd694fb2258482081dc3bc61c7a)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Drop the poll parameters and make the waits much more specific for
the requirements of the tests. This includes looping waiting for
a list of layer elements as that code was previously particularly
fragile.
(Bitbake rev: cf6b8e8aa5484110a41377ba42b3fdd9d6efd877)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Switch this test module to use the common project creation code which contains
race fixes. That code requires the database access wrapper be dropped and
we no longer have ordering constraints.
There is one test that does require database access. Move this to a separate class
and allow database access there. Use ordering constraints to allow them to run
after the main code. They depend on the project creation from the other class which
isn't ideal but good enough for now.
(Bitbake rev: fa10ba2a8749415d8f06cfc15c228c6eb7df1bcf)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
In some cases we want to wait until some element is not visible.
Add such a function helper.
(Bitbake rev: cede6519d25112037264550126b109903220b58c)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
These changes:
* Drop the poll parameters
* Add waits where needed or make them more specific
* Add calls to close notification boxes since they can obscure
other page elements
* In one case wait for a "Loading" message to be replaced
(Bitbake rev: 2bde969486a4761fa0e6f9590c9fa08d385666b1)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The project creation test is well covered in another module so just
drop this one as there isn't any extra value added here.
(Bitbake rev: d26414f191c73fb5330d10d97e240150a3fa1ee8)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Switch this test module to use the common project creation code which contains
race fixes. That code requires the database access wrapper be dropped and
we no longer have ordering constraints.
There are two tests that do require database access. Move these to a separate class
and allow database access there. Use ordering constraints to allow them to run
after the main code. They depend on the project creation from the other class which
isn't ideal but good enough for now.
(Bitbake rev: e441bfe98ac41d48692ffbaeec90a9c780337fa4)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Drop poll calls within the module, adding wait calls for elements of the
page where testing showed it was needed.
(Bitbake rev: 26746eae31f126ff87e678da66dcc976e412c89b)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The project page selection code is race prone. Create a common function to resolve
the race issue and use it from all the call sites rather than duplicate code.
(Bitbake rev: f2bd615b97a6ff3944fa9c1d89a0ea996a12943d)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Drop the remaining poll values and the sleep within the main wait function
as the races this was working around now seem to be resolved with the
preceeding patches.
(Bitbake rev: 91e54cbacc57316a4f2ca766afac81e7f3a54718)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
In my testing I wasn't able to see issues needing the poll parameter in the
browser tests. Drop those.
(Bitbake rev: 3be5e7c78129abf059476416279718156fc666b5)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Rather than hardcoded poll and sleep values, we want an overall long timeout
and to drop all the other 'sleeping' code.
To facilitate this, increase the overall timeout so we can then start to
remove the other values.
(Bitbake rev: 69a479383f9a8cae6600ce51b12b62be3a15ea38)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The tests would race and potentially fail as the elements on the
page being waited for were not specific enough.
Add suitable elements to wait for and drop the remainder of the poll
parameters since these are no longer needed since we no longer need
to sleep for things to appear.
(Bitbake rev: 9db2fd8534610ad8db90a7a696fa7b748b497d15)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The 'fail' calls are useful for find() failures but not for wrapping existing asserts
where they hide more useful error messages. Rework the code flow to give more readable
error reporting.
(Bitbake rev: 1a7a9af418874d3112aa6f148b30e5a4d3d176f3)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Switch this test module to use the common project creation code which contains
race fixes. That code requires the database access wrapper be dropped and
we no longer have ordering constraints.
(Bitbake rev: 653233a1c798d71800b41d3321203a48aa0337d8)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Switch this test module to use the common project creation code which contains
race fixes. That code requires the database access wrapper be dropped and
we no longer have ordering constraints.
(Bitbake rev: f015d96e28d89fd0aef76ff127ebde8613125a66)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Tweak the function name and return the created project ID.
Allow the release_name test to be skipped with a value of None.
Add an extra wait call to ensure the create project button is ready
before clicking it.
(Bitbake rev: d20c6e3c67298c5db33b220a2e4a554e6f2c4418)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
